Validation of mercury determination by solid sampling Zeeman atomic absorption spectrometry and a specially designed furnace

Talanta 70 (2006) 962–965

Validation of mercury determination by solid sampling Zeeman atomic absorption spectrometry and a specially designed furnace Karl-Heinz Grobecker a,∗ , Albena Detcheva b a

European Commission Joint Research Centre, Institute for Reference Materials and Measurements, Retieseweg 111, B-2440 Geel, Belgium b Bulgarian Academy of Sciences, Institute of General and Inorganic Chemistry, Acad. G. Bonchev Str., Bl. 11, 1113 Sofia, Bulgaria Received 15 September 2005; received in revised form 11 May 2006; accepted 18 May 2006 Available online 25 July 2006

Abstract Certified reference materials (CRMs) of different origin were used to validate the direct determination of total mercury by solid sampling Zeeman atomic absorption spectrometry (SS-ZAAS) and a specially designed furnace. The temperature program provides only for one step. Atomisation of mercury and pyrolysis of the matrix is performed at a constant temperature in the range of 900–1000 ◦ C. Calibration points achieved by CRMs and aqueous solutions are covered by one calibration line, indicating the absence of matrix effects. Relatively high amounts of chlorine, known for causing problems in mercury determination do not influence analytical results. The excellent accuracy of the method results in a very good agreement with the certified values. The precision of SS-ZAAS measurements in a range from 0.5 to 50 ng Hg does not exceed 3% R.S.D. A limit of quantification of 0.008 ␮g g−1 Hg was achieved. © 2006 Elsevier B.V. All rights reserved. Keywords: Solid sampling Zeeman atomic absorption spectrometry; Method validation; Mercury; Specially designed furnace; Matrix interferences

1. Introduction The Minamata disaster in 1956 showed the drastic effects of mercury on human health. Humans and animals exposed to high mercury concentrations have suffered disruption of the nervous system, brain damage, kidney malfunction, stomach disruption, DNA alteration and negative reproductive effects. Mercury occurs in the environment from natural sources and as a result of atmospheric deposition and pollution caused by man’s activity. It easily accumulates in the aquatic food chain, including fish and seafood, largely as methyl mercury, which is the form of toxicological concern. Codex guideline levels and European Commission Regulation amendment 221/2002 (OJEC 2002) maximum levels are set at 1 mg kg−1 for large predatory fish and 0.5 mg kg−1 for all other fish [1–4]. Fast, sensitive, accurate and precise analytical methods are required for the control of mercury in samples of environmental and biological origin. The detection of mercury at trace levels is a complex analytical task because of its specific physical and chemical properties.

∗

Corresponding author. Tel.: +32 14 571 715; fax: +32 14 571 787. E-mail address: [email protected] (K.-H. Grobecker).

0039-9140/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.talanta.2006.05.086

The most sensitive mercury resonance line lies in the vacuum UV region and is therefore not suitable for conventional spectrometers. The less sensitive wavelength at 253.7 nm is the only available alternative. Additionally chemical separation of mercury from the sample matrix is difficult due to its high volatility. Already at normal drying temperatures mercury losses have been observed [5]. A plenty of studies were published in recent decades dealing with a variety of acid mixtures for the digestion process while others were concentrating on different noble metals for trapping mercury vapours prior to determination by cold vapour atomic absorption spectrometry (CV-AAS) [6–9]. All of them were more or less prone to analyte losses and/or contamination. The inevitable dilution of the sample is another drawback. Direct determination of mercury by graphite furnace AAS was described by few authors and didn’t gain much attention because of the inferior sensitivity compared to CV-AAS [10,11]. To present all methodical attempts would go beyond the scope of this study, but the most rigorous approach to improve sensitivity and avoid contamination and losses, the direct mercury analysis in solid samples cannot be ignored. Numerous approaches using different types of atomisers including also the

K.-H. Grobecker, A. Detcheva / Talanta 70 (2006) 962–965

963

Fig. 1. Schematic cross-section of the specially designed furnace.

graphite furnace with the addition of various chemical modifiers have been reported [1,12–15]. In recent publications a differential atomic absorption spectrometer and a two-section catalytic pyrolyser is described. The solid sample in a container is directly placed in the first section of the pyrolyser and vapourised in an air stream at 800 ◦ C. All mercury compounds are decomposed and the organic compounds are catalytically oxidised in the second section. Mercury vapour and remaining smoke are transferred into a multi-path analytical cell (optical length 9.6 m). The efficient pyrolysis allows for sample weights of more than 100 mg achieving a detection limit of 0.5 ng l−1 , while Zeeman background correction guarantees interference-free measurements [16–18]. SS-ZAAS in combination with specially designed atomisers seems to be a very promising solution for the direct detection of mercury in environmental and biological samples [19–23]. In this study we used a modified Hadeishi furnace [24] in combination to a direct ZAAS. Total mercury was determined in reference materials of various origins to validate this method. Precision and accuracy were calculated from the results obtained. The linear range and the limit of quantification were determined as well.

2. Experimental

If direct ZAAS (magnetic field at the light source) is applied, the dimensions of the furnace can vary in a wide range. The absorption cell (22.5 cm length) used in this study is about 10 times longer than recent graphite furnace atomisers and results in a corresponding increase of the sensitivity according to LambertBeer-law. The tube of the furnace made of nickel–chromium alloy is heated at a constant temperature in the range of 900–1000 ◦ C to achieve highest sensitivity and a rapid release of Hg from the matrix [24]. These tubes showed a superior durability (>2000 h) compared to pure nickel tubes. The special design of the furnace makes mercury sorption on cold ducts impossible and improves analysis reproducibility and correctness. Water cooling at the sample introduction zone prevents losses of mercury and at the ends of the inner tube it protects optical and electronic devices of the spectrometer against excessive heat. In the constantly heated furnace the atomisation runs under isothermal conditions. The nickel–chromium surface of the tube including the chippings which are pressed tightly into the outer tube is large and together with the applied temperature it completes the catalytic reduction of mercury and matrix compounds. A constant magnetic field is applied to the light source—a mercury gas discharge lamp. The orientation of this magnetic field of 1.0 T is transverse to the optical axis. The experimental conditions for the SS-ZAAS determination of mercury are listed in Table 1.

2.1. Instrumentation

2.2. Reference samples and solutions

A Gr¨un SM20 ZAAS (Gr¨un Analytische Meß-Systeme GmbH, Germany) equipped with a modified two-chamber Hadeishi furnace (Fig. 1) was used for the determination of mercury. This furnace prototype is not commercially available and was developed at the Institute for Reference Materials and Measurements. Samples were weighed on a micro-balance (Sartorius 4503 MP6, Germany) into platinum transport boats (250 ␮l volume). Platinum boats are commercially available and have all the necessary physical and chemical properties. One boat at a time is transferred to an automatic introduction system which inserts the sample boat in the furnace and takes it back when the pyrolysis is complete.

The solid CRMs sewage sludge BCR CRM-146 (9.49 ± 0.76 ␮g g−1 Hg); Tuna fish BCR CRM-464 (5.24 ± 0.10 ␮g g−1 Table 1 Experimental conditions for the SSZAAS determination of Hg Wavelength Spectral bandpass Lamp type Lamp current Measurement mode Measurement time Magnetic field conditions Argon flow rate

253.7 nm 0.25 nm ENL 4.0 mA Peak height 4.0 s Permanent, 1.0 T 30 l h−1

964

K.-H. Grobecker, A. Detcheva / Talanta 70 (2006) 962–965

Hg) and City Waste BCR CRM-176 (31.4 ± 1.1 ␮g g−1 Hg) were used for calibration. The different points on the calibration curve were achieved by variation of the CRMs weight. Aqueous calibration solutions were prepared from a 1000 ␮g ml−1 Hg stock standard solution (Merck, Germany). Polypropylene ware was used throughout.

Fig. 2. Mercury signals obtained with the specially designed furnace.

3. Results and discussion 3.1. Matrix effects Three experimental signal shapes of a blank, a liquid standard solution (0.2 ng Hg) and a certified reference material (human hair BCR 397 12.3 ± 0.5 ␮g g−1 Hg) are shown in Fig. 2. The signal shape of the human hair sample is representative for all organic materials measured and the background is not affecting the signal if the sample weight does not exceed 10 mg of BCR 397. All signals were of the ideal single peak form, no double peaks were observed, which is a prerequisite for using peak height evaluation. Peak height measurements of the Gr¨un SM 20 showed sufficient precision to be used for data evaluation. Despite different matrices of three solid CRMs Sewage Sludge BCR-146, Tuna fish BCR-464, City Waste BCR-176 and 10 ␮l aliquots of calibration solutions which correspond to 0.5; 1.0; 2.5; 5.0; 10.0; 25.0 and 50.0 ng Hg, all points are covered by the same curve (Fig. 3), indicating the absence of matrix effects. Some of the certified reference materials used in this study are of marine origin and contain various amounts of chlorine and iodine. Findings presented in Table 2 show that relatively high amounts of chlorine do not affect the analytical results. The linear range surmounts 50 ng Hg and a quantification limit of 0.008 ␮g g−1 Hg was achieved by measuring Salmon BCR-725 (up to 35 mg sample weight).

Fig. 3. Calibration curve for Hg using solid CRMs and liquid standards—sewage sludge BCR-146 (9.49 ± 0.76 ␮g g−1 Hg); Tuna fish BCR-464 (5.24 ± 0.10 ␮g g−1 Hg); city waste BCR-176 (31.4 ± 1.1 ␮g g−1 Hg); aqueous standard solutions.

Table 2 Mercury determination of chlorine containing samples Sample

Concentration (␮g g−1 Hg) Cert. value

Aquatic plant BCR-60 Aquatic plant BCR-61 Olive leaves BCR-62 Mussel tissue BCR-278 Dogfish muscle NRCC DORM-1 Lobster Hepat. NRCC TORT-1 Lobster Hepat. NRCC TORT-2 Indicative values: in parenthesis.

0.343 0.230 0.284 0.188 0.798 0.33 0.27

± ± ± ± ± ± ±

0.04 0.02 0.02 0.007 0.074 0.05 0.06

Exp. value 0.32 0.25 0.29 0.19 0.779 0.34 0.28

± ± ± ± ± ± ±

0.06 0.03 0.04 0.04 0.044 0.04 0.04

Cl (%) (1.0) (0.23) (0.07) (3.2) 1.13 5.58 Not cert.

K.-H. Grobecker, A. Detcheva / Talanta 70 (2006) 962–965 Table 3 Calculation of precision Hg (ng)

Absorbance

50 25 10 5 2 1 0.5

0.0421 0.0215 0.1342 0.0673 0.0325 0.0173 0.0168

± ± ± ± ± ± ±

0.0008 0.0004 0.0015 0.0010 0.0009 0.0004 0.0004

n

Precision (%)

Gain

50 20 20 20 20 20 20

1.84 1.86 1.15 1.49 2.77 2.37 2.38

1 1 5 5 5 5 6

Table 4 Calculation of accuracy Sample

Tuna fish BCR-464 Light sandy soil BCR-142 Light sandy soil BCR-142R Aquatic plant BCR-60 Aquatic plant BCR-61 Olive leaves BCR-62 Milk powder BCR-151 Pig kidney BCR-186 Mussel tissue BCR-278 Spinach NIST-1570 Pepperbush NIES-1 Human hair NIES-5 Fly ash BCR-38 Bovine liver NIST-1577a Orchard leaves NIST-1571 Plankton BCR-414 Bovine liver BCR-185 Human hair BCR-397 Dogfish muscle NRCC DORM-1

Concentration (␮g g−1 Hg) Cert. value

Exp. value

5.24 0.1040 0.067 0.343 0.230 0.284 0.101 1.970 0.188 0.031 0.056 4.41 2.10 0.016 0.155 0.276 0.044 12.3 0.798

5.34 0.1043 0.063 0.344 0.240 0.289 0.103 1.962 0.190 0.032 0.053 4.43 2.12 0.017 0.158 0.265 0.042 12.2 0.779

Accuracy (%)

965

Experimental Hg signal shapes derived from liquid and solid samples were of the ideal single peak form, which allows peak height evaluation. A calibration curve was achieved by using aqueous solutions and solid CRMs, all points were covered by one line, indicating that SS-ZAAS mercury measurements were matrix independent. Even relatively high amounts of chlorine were not affecting analytical results. The high accuracy of the method is evident when comparing certified and experimental values. The precision of the method does not exceed 3% R.S.D. over a range from 0.5 to 50 ng Hg. The limit of quantification was calculated to be 0.008 ␮g g−1 Hg and allowed to measure very low Hg contents without extraction or accumulation procedures. References

1.91 2.88 5.97 0.29 4.35 1.76 1.98 0.41 1.06 3.23 5.36 0.45 0.95 6.25 1.93 3.99 4.54 0.81 2.38

3.2. Precision and accuracy The precision of SS-ZAAS mercury measurements was calculated over a range from 0.5 to 50 ng Hg by analysing aqueous solutions. The relative standard deviation in all cases does not exceed 3% R.S.D. The results are presented in Table 3. The accuracy of SS-ZAAS mercury measurements was calculated by analysing CRMs of different origin according to the formula: |exp. value − cert.value| accuracy (%) = × 100 cert. value where exp. value is mean experimental value of minimum five replicates and cert. value is value from certificate. An excellent agreement between the experimental and certified values was obtained, which is a proof for the good accuracy of the method. The results are presented in Table 4. 4. Conclusions The validation of total mercury determination by direct SSZAAS and a specially designed furnace without any chemical sample preparation was successfully performed by analysing environmental and biological CRMs.

[1] T.G. Knowles, D. Farrington, S.C. Kestin, Food Addit. Contam. 20 (2003) 813. [2] R.W. Dabeka, A.D. McKenzie, P. Bradley, Food Addit. Contam. 20 (2003) 629. [3] Joint FAO/WHO Food Standards Programme, Codex Committee on Food Addit. Contam., 37-th session, The Hague, The Netherlands, April 25–29, 2005, Discussion paper on guideline levels for methyl mercury in fish. [4] J.L. Love, G.M. Rush, H. McGrath, Food Addit. Contam. 20 (2003) 37. [5] U. Kurf¨urst, B. Rues, Labor Praxis 10 (1982) (1982) 1094 (in German). [6] B. Welz, M. Sperling, Atomic Absorption Spectrometry, 3rd ed., WileyVCH, Weinheim, New York, 1999, pp. 675–676. [7] R. Puk, J.H. Weber, Appl. Organomet. Chem. 8 (1994) 293. [8] W.L. Clevenger, B.V. Smith, J.D. Winefordner, Crit. Rev. Anal. Chem. 29 (1997) 1. [9] S. McIntosh, J. Baasner, Z. Grosser, C. Hanna, At. Spectrosc. 15 (1994) 161. [10] Z. Grobenski, W. Erler, U. Voellkopf, At. Spectrosc. 6 (1985) 91. [11] S.M. Maia, B. Welz, E. Ganzarolli, A.J. Curtius, Spectrochim. Acta Part B 57 (2002) 473. [12] A.F. da Silva, B. Welz, A.J. Curtius, Spectrochim. Acta Part B 57 (2002) 2031. [13] A. Sahuquillo, G. Rauret, M. Bianchi, A. Rehnert, H. Muntau, Anal. Bioanal. Chem. 375 (2003) 578. [14] B. Welz, G. Schlemmer, J.R. Mudakavi, J. Anal. At. Spectrom. 7 (1992) 499. [15] E. Bulska, W. Kandler, A. Hulanicki, Spectrochim. Acta Part B 51 (1996) 1263. [16] S. Sholupov, S. Pogarev, V. Ryzhov, N. Mashyanov, A. Stroganov, Fuel Process. Technol. 84 (2004) 473. [17] S.E. Pogarev, V. Ryzhov, N. Mashyanov, S. Sholupov, V. Zharskaya, Anal. Bioanal. Chem. 374 (2002) 1039. [18] E.A. Sondreal, S.A. Benson, J.H. Pavlish, N.V.C. Ralston, Fuel Process. Technol. 85 (2004) 425. [19] L. Steubing, K.H. Grobecker, J. Haneke, Zeitschrift f¨ur Umweltchemie und ¨ Okotoxikologie 3 (1991) 332 (in German). [20] K.H. Grobecker, B. Kl¨ußendorf, Fresenius Z. Anal. Chem. 322 (1985) 673 (in German). [21] A. Rosopulo, K.H. Grobecker, U. Kurf¨urst, Fresenius Z. Anal. Chem. 319 (1984) 540 (in German). [22] K.H. Grobecker, U. Kurf¨urst, Solid sampling by Zeeman graphite-furnaceAAS, a suitable tool for environmental analysis, in: H. Lieth, B. Markert (Eds.), Element Concentration Cadasters, VCH, 1990, pp. 121–137. [23] J. Pauwels, G.N. Kramer, L. De Angelis, K.H. Grobecker, Fresenius J. Anal. Chem. 338 (1990) 515. [24] D.A. Church, T. Hadeishi, L. Leong, R.D. McLaughlin, B.D. Zak, Anal. Chem. 46 (1974) 1352.