Determination of Total Mercury in Sediments by Microwave-assisted Digestion–Flow Injection–Inductively Coupled Plasma Mass Spectrometry A� GNES WOLLER*†a, HERVE� GARRAUDa‡, FABIENNE MARTINa‡, OLIVIER F. X. DONARDa‡ AND PE� TER FODORb aL aboratoire de Photophysique et PhotochimieMole�culaire, Universite� de Bordeaux I, CNRS URA 348, 351, Cours de la L ibe� ration, 33405 T alence Cedex, France bDepartment of Chemistry and Biochemistry, University of Horticulture and Food Industry, V illa�nyi u. 29–34, Budapest, 1114, Hungary A method for the determination of total mercury in sediment cury by ICP-MS, on the other hand, is not very easy. The major problem is the extensive memory effect, which requires samples was developed. Extraction of mercury from a sample long washout times. It seems that mercury, even at relatively matrix was carried out in an open vessel microwave digestion low concentrations, can adhere to the walls of the spray system while maintaining mild conditions during digestion in chamber and the transfer tubing of the introduction system, order to avoid any loss of mercury due to volatilization.A causing contamination of subsequent samples and a continuous complexing agent (EDTA) and a surfactant (Triton X-100) decrease in sensitivity in aqueous calibration. A possible expla- were added to the samples in order to eliminate memory nation of this phenomenon could be that mercury vapour effects associated with mercury determinations and to obtain builds up slowly in the spray chamber and this is caused by reproducible linear calibration curves.Standard additions and the increased volatility of mercury due to the large pressure internal standardization were used for calibration and drop during high-pressure pneumatic nebulization.6,13 This correction in an FI–ICP-MS detection system. The method situation makes accurate and precise determinations of mercury was validated using the certified reference material PACS-1 in environmental samples impossible.Possible solutions to this and the reference materials IAEA-356 and S19 and gave problem include alternative sample introduction systems, e.g., results in good agreement with the certified and reference a direct injection nebulizer (DIN), which has been proposed values. Sediment samples from Arcachon Bay were also by several workers.14,15 The technique is promising, although analysed. Detection limits of 10 ng g-1 for solution and in sediment analysis the very small internal diameter of the 1 ng g-1 for dry sediment samples were obtained.DIN capillary would possibly cause problems due to clogging. Keywords: Mercury; microwave-assisted digestion; flow Flow injection (FI) techniques have been used with various injection ; inductively coupled plasma mass spectrometry; atomic spectroscopic measurements to facilitate calibration sediment samples; reference materials and dilution, to reduce sample consumption and to lower the risk of user-introduced contamination.16 With ICP-MS, a continuous aspiration solution analysis with large amounts of Lakes, rivers and coastal waters are important indicators of total dissolved solids may cause matrix deposition on the environmental pollution.The determination of contaminants cones. In consequence, reduced sensitivity, instrument drift, in aquatic systems helps in evaluating the present state of the degradation of measurement precision and cone orifice clog- environment.Mercury is of considerable interest as it is widely ging may be expected. When FI sample introduction is used, used in industry in the production of chemicals, pesticides, the attack of harsh matrices on the sampler and skimmer electrical apparatus, paints, dental mateials, etc. Concerning cones and instrument drift when analysing difficult samples the determination of mercury in aquatic systems, because this are reduced and there is essentially no degradation of the element tends to accumulate in bottom sediments,1 the estab- ICP-MS detection power.17 Combined with standard additions lishment of its concentration in sediments can play a key role and internal standardization, FI is an excellent tool to decrease in detecting sources of pollution.matrix effects with complex samples. Various methods for the determination of total mercury in Sample preparation is a critical factor in the analysis of natural samples have been reported, including AFS,2 AAS,3 environmental samples.With mercury, special care must be ICP-AES4 and ICP-MS,5–9 although the most commonly taken to avoid losses by volatilization and/or adsorption. employed and most sensitive method for the determination of Microwave digestion seems to be a method of choice over mercury in environmental samples is still cold vapour atomic conventional procedures for the digestion of mercury in sedi- spectrometric techniques (CVAAS, CVAFS).10–12 ment and soil samples.Amongst the reasons are the shorter ICP-MS has become a widely accepted technique applied to extraction time, the higher extraction efficiency, the easier environmental samples because of its unique capabilities. control of digestion parameters and the ease of ICP-MS, when compared with AFS, FAAS and ICP-AES, automation.3,6,10,12 offers exceptional sensitivity and excellent accuracy along with In this paper, we present a method for the determination of multi-element and isotope ratio measurement capabilities. The mercury in sediments by FI sample introduction followed by accuracy of isotope dilution techniques was shown to be ICP-MS detection after a microwave-assisted sample digestion.superior to other techniques when the optimum isotopic ratio The method was validated using two reference materials and for analysis was maintained in the case of mercury determi- a certified reference material. Real samples collected from nation in environmental samples.9 The determination of mer- Arcachon Bay (France) were also analysed.† Present address: The Department of Chemistry and Biochemistry, EXPERIMENTAL University of Horticulture and Food Industry, Villa�nyi u. 29–34, Instrumentation Budapest, 1114, Hungary. An ELAN 5000 ICP-MS instrument (Perkin-Elmer SCIEX, ‡ Present address: CNRS EP 132, He�lioparc, 2 Avenue du Pre�sident Angot, 64000 Pau, France. Thornhill, Ontario, Canada) equipped with a Scott-type Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12 (53–56) 53double-pass spray chamber was used throughout the analyses. 2 ml of H2O2 (Merck, Darmstadt, Germany) and digestion was continued at a power setting of 20W for a further 5 min. The standard ICP-MS operating conditions used in this study are presented in Table 1. Optimization was carried out daily After cooling, the samples were diluted with Milli-Q water according to the concentration range of the samples.Analysis with a normal verification solution (10 ng g-1, Rh, Mg, Pb, Ce, Ba). Raw data were collected by the ELAN software was carried out either on the same day as, or the day after, sample digestion. Samples were kept cool (4°C) in the dark through a personal computer (IBM PS/2 Model 70). A FIAS-200 FI system (Bodenseewerk Perkin-Elmer, overnight in case of delayed analysis. Stability tests for 3 d storage carried out with standard reference materials showed U� berlingen, Germany) was used, fully controlled by the ELAN software.The system includes two peristaltic pumps and a that the aforementioned conditions preserved the total amount of extracted mercury. flow injection valve fitted with a 0.2 ml injection loop. Peak areas were used to characterize the analytical response and Analytical Procedure concentrations were calculated after normalization of the data to the internal standard signal followed by appropriate blank The final solutions (5 ml) contained 0.1% Triton X-100 (Sigma, subtraction.St. Louis, MO, USA), 0.1% EDTA (added as 1 g l-1 solution) Sample digestion was carried out in a single-mode reflux (Aldrich–Chemie, Steinheim, Germany), 0.1% v/v ammonia focused microwave system (Microdigest A-301, Prolabo, Paris, solution (Merck) and an internal standard consisting of a Frawer setting of 200 W. The power 20 ng ml-1 solution of thallium (Merck). It was necessary to system used with a focused microwave apparatus provides a add Triton X-100 as a surfactant and EDTA as a complexing continuous microwave emission at each power level.During agent in order to obtain a linear calibration curve (see Results digestion of sediments, in all cases 10% (20W) of the maximum and Discussion). Mercury working standard solutions were power output was used. The construction of the system allows prepared from a 1000 mg ml-1 Hg stock standard solution a change of time setting in 1 min steps.Specially designed glass (Spex Industries, Edison, NJ, USA) by successive dilutions vessels and the construction of the system allow digestion with Milli-Q water. Mercury standard spikes were added to under atmospheric pressure without any microwave leakage. the sample prior to determination by FI–ICP-MS to give As the system is an open-vessel microwave system, an Aspivap additions to the final concentrations of 0, 10, 20 and fume treatment system (Prolabo) is used, which effectively 40 ng ml-1.Once the solutions had been prepared they were neutralizes the acid fumes generated. analysed within 2 h. The stability of these solutions does not seem to exceed 24 h, which could be attributed to the effect of Sample Digestion the surfactant. The rinsing solution for the FI system was 5% v/v hydro- High-purity de-ionized water purified with a Milli-Q analytical- chloric acid (Romil). reagent grade water-purification system (Millipore, Chester, Cheshire, UK) was used throughout.Acids used were all super- Reference Materials and Real Samples purity acids and other reagents were of analytical-reagent grade. Samples, standards and final volumes were measured A certified reference material (PACS-1, National Research by mass in all cases. Vessels were rigorously cleaned, soaked Council of Canada) and two reference materials (IAEA-356, for 24 h in 10% nitric acid and thoroughly rinsed with Milli-Q S19), the latter produced for European intercomparison studies water before use.(BCR, EC),18 were used for validation of the method. Digestion parameters used throughout this work are given Sediment samples from Arcachon Bay were collected at in Table 2. In the procedure, in all cases 0.25 g of dry sediment several points and were placed in clean polyethylene bottles. sample was digested with 8 ml of HNO3 (Romil, Cambridge, Surface sediments were refrigerated at -20 °C immediately. UK) at a power setting of 20W (10%) for 5 min.The sample As the extraction procedure has been developed for dry was left to cool for about 5 min, followed by the addition of sediments, it was necessary to freeze-dry the wet sediments before analysis. The freeze-dryer used was an RP 2V (Department CIRP Lyophilisation, Argenteuil, France). Table 1 Operating conditions of the ELAN 5000 ICP-MS instrument Amounts of 10–30 g of wet sediment were introduced into the used for mercury determination stainless steel vessels provided by the manufacturer.In order ICP-MS conditions— to prevent losses of mercury due to volatilization, mild con- Forward rf power 1100W ditions were maintained throughout lyophilization. Samples Plasma gas flow rate 15 l min-1 were left to dry for 52 h before being transferred into polyethyl- Auxiliary gas flow rate 0.8 l min-1 ene bottles and kept in a refrigerator at 4°C. Similarly to the Nebulizer gas flow rate 0.98 l min-1 reference materials, samples were homogenized by shaking for Sampler and skimmer cones Nickel about 5 min before addition of acids and microwave digestion.Data acquisition— Scan mode Peak hop transient Measurements Dwell time 100 ms The performance of the FI–ICP-MS system was verified regu- Sweeps per reading 5 larly according to the daily procedure. Two isotopes of mercury Readings per replicate 60 No. of replicates 1 (masses 200 and 202) were always measured and calculations Signal processing Spectral peak integrated were normally based on the mercury-202 isotope.Mercury-200 Isotope measured 202Hg, 200Hg was used to check and confirm results in case of doubt. Internal standard 205Tl Analyses were carried out in duplicate and appropriate blank subtraction was used. To check for contamination of the digestion procedure and sample manipulation, a blank solution Table 2 Digestion parameters in an open-vessel (Prolabo) microwave was prepared and carried through each set of analyses.system for determination of total mercury in sediment samples by FI–ICP-MS. Amount digested: 0.25 g of sediment sample RESULTS AND DISCUSSION Step Reagent added Power setting/W Time/min Minimizing the Memory Effect of Mercury 1 8 ml conc. HNO3 20 5 Mercury is a difficult element to determine in real samples by 2 2 ml 30% H2O2 20 5 ICP-MS. Problems arising from its strong memory effect are 54 Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12Table 3 Summary of results for the determination of mercury in mostly manifested through unacceptable correlation of cali- reference sediment samples by microwave digestion and FI–ICP-MS bration curves or a continuous decrease in sensitivity.Our with standard additions. Analytical figures of merit of the system are first aim before performing analyses of sediment samples was also shown therefore to find a procedure for eliminating these effects.We first optimized the analytical conditions with standards Reference material Reference value/mg g-1* Measured value/mg g-1† n‡ in simple solutions before carrying out analyses of real samples. To ensure the stability of mercury during transport and S19 91.07 95±3 6 nebulization, the pH of the solutions was kept alkaline by the IAEA-356 7.62 7.3±0.1 4 PACS-1§ 4.57±0.16 4.7±0.3 4 addition of ammonia solution, and the mercury was complexed with 0.1% EDTA. Also, 0.1% Triton X-100 was added to Correlation coefficient of calibration 0.995–1.000 improve the sample transport efficiency through its action as Limits of detection 10 ng g-1 (for solution) a surfactant.The method proposed by Campbell et al.6 resulted 1.0 ng g-1 (for dry sediment in linear calibration curves with a correlation coefficient samples) between 0.995–1.000. The fact that the sensitivity did not Recovery 95–105% change when repeated calibrations were executed indicates * Concentration values expressed on dry-mass basis.that the memory effect was completely eliminated in this way. † Results given with 95% confidence interval. It is likely that complexation transforms mercury into a more ‡ Number of replicates used for calculation. uniform and inert form, so no losses from volatilization and § Certified reference material. no sensitivity changes due to vapour build-up in the nebulization spray chamber occur. Triton as a surfactant might also additions separately also did not gave acceptable results, help to ensure uniform conditions and nebulization seems to whereas the combination of the two techniques proved to be be more stable, because losses during sample transport are excellent when determining mercury in sediment matrices. minimized.The behaviour of mercury becomes more predict- It has been shown that using an internal standard with a able and controllable. With these conditions, we were able to mass number close to that of the analyte element improves obtain a calibration curve with an excellent correlation the precision.19 Thallium, having a mass number of 205, is coefficient and without any memory effect. In Fig. 1, a transient close enough to the mercury-202 isotope, although its calcu- signal of the FI–ICP-MS system can be seen. Comparing the lated extent of ionization in argon ICP is 100% whereas for shape of peaks obtained for thallium (used as internal standard, mercury it is only about 38%.18 The internal standard chosen, see later) with those obtained for the two mercury isotopes, combined with standard additions, corrected the instrument we can see that minimum tailing and asymmetry resulted in instabilities and the interferences to a reasonable extent.This the case of mercury. This indicates that problems emerging seems to reinforce the observations of Vanhaecke et al.19 which from the memory effect of mercury have been overcome. indicated that the ionization energy of the internal standard has no or only secondary importance with regard to signal Determination of Mercury in Sediments precision and accuracy. Results for the total mercury content obtained for the analysis of reference materials are given in Table 3.Analytical figures Digestion Procedure of merit of the system are also indicated. For the determination of mercury in solid environmental The sample matrix can significantly change the instrument samples by FI–ICP-MS, a pre-treatment is necessary.This was sensitivity, preventing the use of external calibration methods. achieved by a microwave-assisted digestion procedure that has The technique of standard additions and the use of internal basically two objectives with regard to mercury determination: standards have traditionally been used to overcome the matrix first, to oxidize the organic matter in the sample and thus to effects of an analyte, since these quantification methods take liberate the mercury species from the sample matrix; and into account the instrument sensitivity for each sample matrix.second, to oxidize fully the liberated mercury to HgII in order In the given system, the analysed sediment samples contributed to prevent losses. These two objectives must also be kept in a complex matrix effect to the determination of mercury such mind when the digestion reagents are chosen. A microwave- that external calibration was found to be inadequate.Even assisted digestion procedure developed by Lamble and Hill10 though linear calibration curves with excellent correlation was used as a starting point, after having completed some coefficients were obtained when standard solutions (with added preliminary experiments on the digestion system. A modifi- EDTA and Triton X-100) were analysed, the results were not cation was made, however, since the procedure needed to be in agreement with the reference values when the S19 material adapted to ICP-MS detection. The sulfuric acid used as a was analysed.The use of internal standard or standard reagent in the first step of the digestion procedure was replaced with nitric acid in order to protect the sampler and skimmer cones from the strong corrosive effect of sulfuric acid. Conditions of the digestion procedure are summarized in Table 2. Elemental mercury (Hg0), being a highly volatile element, can easily be lost during microwave digestion in an open system.Care has to be taken to avoid the formation of Hg0 by overheating (too high power settings or too long a heating time) or by aspirating the generated fumes too strongly. This is the reason why the lowest possible power setting (20W, 10%) was used throughout the heating and why care was taken when adding reagents to avoid venting off acid fumes which may have resulted in losses of mercury. The fume aspiration head was set to the highest position in order to reduce aspiration of the fumes.Another important factor when Fig. 1 Transient signal of spiked sediment sample from Arcachon performing a microwave-assisted digestion procedure is the Bay. ICP-MS parameters are given in Table 1. A, 202Hg; B, 200Hg; and C, 205Tl. amount of the digested sample. Preliminary experiments Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12 55Table 4 Quantitative analysis of sediment samples from Arcachon showed that recoveries were low (30–60%) and not repetitive Bay.The total mercury content was determined by microwave diges- when 0.5 g of sediment sample was digested with 10 ml of acid tion followed by FI–ICP-MS reagent. However, 0.25 g of dry sediment proved to be sufficient to obtain exact and reproducible results when performing the Sample Total Hg/ng g-1 Water content of wet sample (%) digestion procedure using the optimum conditions. All these A 190 66.5 precautions together provided conditions mild enough to retain B 260 62.7 all of the mercury.On the other hand, the digestion procedure C 180 63.7 was rigorous and long enough to recover 100% of the mercury. D 240 64.6 The digestion procedure is simple, and addition of the second Mean 218 acid reagent causes no problems since the system is open. The complete digestion procedure for one sample takes only 15 min, which can be further shortened when analysing samples in series. There is a considerable risk of contamination with open- wave-assisted digestion and FI–ICP-MS. A simple and fast vessel microwave systems, but throughout our experiments no digestion procedure was used and, with FI-ICP-MS detection, contamination was observed.resulted in an average 100% recovery for total mercury in the mercury concentration range 4–100 mg g-1. The application Reference Materials and Real Samples of FI with standard additions, internal standardization and The certified reference material PACS-1 and the reference complexation of mercury eliminated the classical problems materials IAEA-356 and S19 were used to validate this method.experienced when using direct calibration with conventional Results obtained for the determination of total mercury in ICP-MS methods. Disadvantages of this method are the small these three sediment reference materials are presented in sample throughput, which is a direct consequence of the Table 3. The values obtained for the reference materials (which standard additions method.The possibility of on-line standard have different concentration ranges and different matrix com- additions is offered by the FI system. The complete automation positions) are in good agreement with the given values. The of the detection system retaining the same limits of detection calibration curves for standard additions gave correlation may yet prove to be a successful method for routine analysis. coefficients in the range between 0.995–1.000.Recoveries between 95 and 105% were obtained, which represents an The authors thank the French Government for support (Bourse average maximum error of 10%. Limits of detection (LODs) du Gouvernement Franc�ais). were determined as three times the standard deviation of the blank divided by the slope of the calibration curve. For REFERENCES solutions an LOD of 10 ng g-1 and for dry sediment samples 1 Fo� rstner, U., and Wittmann, G. T. W., Metal Pollution in the about 1.0 ng g-1 (depending on the sample matrix) were Aquatic Environment, Springer, Berlin, 1983.obtained. The average precision of the whole procedure was 2 Godden, R. G., and Stockwell, P. B., J. Anal. At. Spectrom., 1989, 4, 301. better than 90% for the mercury concentration range 3 Bulska, E., Kandler, W., Paslawski, P., and Hulanicki, A., 4–100 mg g-1. Mikrochim. Acta., 1995, 119, 137. When the optimization and validation of both the extraction 4 Can�ada Rudner, P., Garcý�a de Torres, A., and Cano Pavo�n, J.M., procedure and the analytical determination with sediment J. Anal. At. Spectrom., 1993, 8, 705. reference materials had been completed, we determined the 5 Stroh, A., and Vo� llkopf, U., J. Anal. At. Spectrom., 1993, 8, 35. mercury content in real samples. The results obtained for 6 Campbell, M. J., Vermeir, G., Dams, R., and Quevauviller, P., J. Anal. At. Spectrom., 1992, 7, 617. sediment samples from Arcachon Bay are given in Table 4. 7 McLaren, J.W., Beauchemin, D., and Berman, S. S., Spectrochim. The mercury content was determined for samples from four Acta, Part B, 1988, 43, 413. different sampling areas. The water contents of these samples 8 Kalamegham, R., and Owen Ash, K., J. Clin. L ab. Anal., 1992, were not identical. The second column of Table 4 gives data 6, 190. concerning the water content of wet samples. Standard 9 Smith, G. R., Anal. Chem., 1993, 65, 2485. additions gave excellent calibration and the results seem to 10 Lamble, K.J., and Hill, S. J., J. Anal. At. Spectrom., 1996, 11, 1099. indicate an equal distribution of mercury of around 218 ng 11 Hall, A., Duarte, A., Caldeira, M. T. M., and Lucas, M. F. B., Sci T otal Environ., 1987, 64, 75. g-1 in the sediment f Since these samples have 12 Scifres, J., Cheema, V., Wasko, M., and McDaniel, W., Am. lower mercury concentrations than the reference standards, Environ. L ab., 1995, 6, 6. another analytical technique was used to confirm that the 13 Bushee, D.S., Analyst, 1988, 113, 1167. FI–ICP-MS results were accurate. Determination of total and 14 Wiederin, D. R., Smyczek, R. E., and Houk, R. S., Anal. Chem., methylmercury in this case was carried out by aqueous phase 1991, 63, 1626. ethylation followed by cryogenic GC–quartz furnace AAS after 15 Powell, M. J., Quan, E. S. K., and Boomer, D. W., Anal. Chem., 1992, 64, 2253, a microwave-assisted acid leaching.21 A two-sample paired 16 Tyson, J., Fresenius’ Z. Anal. Chem., 1988, 329, 663. Student’s t-test was used to calculate the significance of differ- 17 Jarvis, K. E., Gray, A. L., and Houk, R. S., Handbook of ences between mean values of the two different methods. At Inductively Coupled Plasma Mass Spectrometry, Blackie, New the 95% confidence level the results showed no significant York, 1992, pp. 119–124. difference. 18 Quevauviller, Ph., Fortunati, G. U., Filippelli, M., Baldi, F., According to the environmental regulations for soil in Bianchi, M., and Muntau, H., Appl. Organomet. Chem., 1996, 10 in the press. Europe, the sediment from Arcachon Bay is not polluted and 19 Vanhaecke, F., Vanhoe, H., and Dams, R., T alanta, 1992, 39, 737. further action is not required (limit in Germany 250 ng g-1 20 Houk, R. S., Anal. Chem., 1986, 58, 97A. and in the UK 1000 ng g-1). However, considering the assess- 21 de Diego, A., Tseng, C. M., Stoichev, T., Martin, F., and Donard, ment criteria of the Interim Canadian Environmental Quality O. F. X., unpublished data. Criteria for Contaminated Sites, where the limit is 100 ng g-1, 22 Visser, W. J. F., Contaminated L and Policies in Some Industrialized action should be considered.22 Countries, TCB, The Hague, Netherlands, 1994. CONCLUSION Paper 6/06044G Received September 2, 1996 This work has demonstrated the ability to determine total mercury with good precision in sediment samples by micro- Accepted October 29, 1996 56 Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12