Elimination efficiency of different reagents for the memory effect of mercury using ICP-MS

Abstract

The present study was carried out to solve the problems of long washout time and non-linear calibration curves encountered in mercury analysis using inductively coupled plasma mass spectrometry. Comparisons of the washing efficiency for different reagents to eliminate the mercury memory effect were made. Of all the selected washing reagents, mercapto reagents such as 2-mercaptoethanol and L-cysteine could efficiently clean the instrument. Taking the toxicity and odor of mercaptoalkanol into account, L-cysteine was preferred as the most suitable washing reagent and was added into the standard and sample solutions. A good linear calibration curve was obtained with the correlation coefficient of 0.9999. The detetion limit for addition of L-cysteine was 0.024 μg l⁻¹. Addition of 0.18% L-cysteine into the sample solution also facilitates the washout of mercury even using deionized water. The recoveries of mercury in certified reference materials, i.e. human hair and dogfish muscle, were 97.2% and 98.3%, respectively, when using 0.18% L-cysteine in the sample solutions.

---

Introduction

Inductively coupled plasma mass spectrometry (ICP-MS) with a peristaltic pump is widely used for the determination of trace metals. It offers exceptional sensitivity and excellent accuracy along with multi-element and isotope ratio measurment capabilities.¹ However, in the analysis of mercury, a series of problems were encountered when using the conventional 2% HNO₃ solution, such as long washout time, non-linear calibration curves, decreasing sensitivity with time, and signal counts dependent on the sample matrix.² This is because mercury, even at relatively low concentrations, can adhere to the walls of the spray chamber and the transfer tubing of the sample introduction system,^3,4 the so-called mercury memory effect. This makes accurate and precise determinations of mercury difficult or nearly impossible using ICP-MS.

A number of methods have been proposed to solve these problems, e.g. gold addition,⁵ hydrobromic acid washing,⁶ Triton X-100/ammonia/ethylenediaminetetraacetic acid (EDTA) addition,¹ and 2-mercaptoethanol (ME) addition in carrier solution.² However, no comparisons among all those washing methods have been made.

In this Technical Note, different washing reagents such as EDTA and its disodium salt, KBr, Na₂S, Na₂S₂O₃, ME, (R,R)-1,4-dimercapto-2,3-butanediol (dithiothreitol, DTT), and L-cysteine (Cys) were tested to compare their washing efficiency. Gold was not tested owing to its possible deposition on nickel cones, which may cause clogging in the ICP-MS system. The best washing reagent was added into both standard and sample solutions to get linear calibration curves and to eliminate the memory effect of mercury. The detection limit and accuracy were also evaluated.

Experimental

Instrument

A Thermo Elemental X7 ICP-MS was used throughout the analysis. The standard ICP-MS operating conditions used in this study are presented in Table 1. Optimization was carried out daily with a normal tuning solution (1 ng ml⁻¹, Be, Co, In, U). Raw data were collected by the PlasmaLab software through a personal computer.

Table 1 Operating conditions of the Thermo Elemental X7 ICP-MS used for mercury analysis

ICP-MS conditions
Forward power	1200 W
Cool gas flow rate	13.0 l min^−1
Auxiliary gas flow rate	0.75 l min^−1
Nebulizer gas flow rate	0.84 l min^−1
Sample uptake rate	0.6 ml min^−1
Nebulizer	Glass concentric
Spray chamber	Quartz impact bead
Interface	HPI
Sampling depth	80 mm
Data acquisition
Scan mode	Main peak jump
Dwell time	10 ms
Sweeps per reading	300
Isotope measured	^202Hg
Internal standard	^115In (not used for washout study)

---

Reagents and standards

Deionized water (>18 MΩ) was obtained from a Milli-Q water-purification system (Millipore, UK). EDTA, EDTA-2Na, KBr, Na₂S, Na₂S₂O₃ (AR, Beihua, China), ME (>99%, Amersco, USA), Cys (>99%, merck, USA), and DTT (merck, USA) were used as washing reagents. The Chinese certified reference material (CRM) human hair (GBW 07601, National Research Centre for CRMs, China) and Canadian CRM dogfish muscle (Dorm-2, National Research Council, Canada) were used for quality controls. Mercury chloride stock standard solution was obtained as 1000 mg l⁻¹ (as Hg) solution (GBW 08617) and diluted as required using deionized water daily. The Hg dilution processes using water were mainly used to compare the washing efficiency of different reagents.

Analytical procedure

Different mercury chloride solutions (5 and 50 ng ml⁻¹) were continuously introduced into the ICP-MS for about 1 min or 3 min (3 or 10 runs, the time interval of each run may alternate between 19 s to 23 s for different operations) and no rinse was performed between individual runs. Then the system was cleaned using different washing reagents. The data of Hg counts via runs were used to compare the cleaning efficiency of different washing reagents (the internal standard ¹¹⁵In was not used for the washing experiment).

Results and discussion

Preliminary experiment on washing efficiency of different reagents

In the preliminary study, 5 and 50 ng ml⁻¹ Hg solutions were introduced into the instrument for about 1 min (3 runs), then it was washed using different washing reagents for an additional 3.8 min (12 runs). The results for the 50 ng ml⁻¹ Hg solution introduction and washing are shown in Fig. 1.

Fig. 1 Washing curves using different washing reagents for introduction of 50 ng ml⁻¹ Hg²⁺.

As it can be seen in Fig. 1, for a lower concentration of ME (0.05%), the washing efficiency is not as good as that of 0.1% ME. For the ICP-MS system, high concentrations of organic materials such as Cys and ME have higher efficiency to reduce the possiblity of Hg adhering to the walls of the spray chamber and the transfer tubing of the sample introduction system, but, which may cause clogging owing to more deposition of carbon on nickel cones at higher concentrations.

According to the preliminary experiment, although the Hg concentration varies from 5 to 50 ng ml⁻¹, nearly all the reagents chosen and even the commonly used 2% nitric acid can effectively wash out mercury to the blank level within 1 min. This could be attributed to the relatively short introduction time (only about 1 min) of mercury and a complete washing out process each time before the next sample. Therefore, in the next experiment, a longer continuous introduction time of 3.8 min (10 runs) into the ICP-MS system and more washing steps were taken. In addition, salts such as EDTA-2Na, KBr, Na₂S, and Na₂S₂O₃ were not applied because of their possible deposition on nickel cones which may cause clogging in the ICP-MS.

Further experiments on Hg washout using selected washing reagents

In the next study, the 50 ng ml⁻¹ Hg solutions were introduced into the instrument for 3.8 min (10 runs), then washed using ME (0.1%, v/v, in 2% HNO₃ or deionized water), DTT (0.05%, w/v, in 2% HNO₃ or deionized water), Cys (0.18%, w/v, in 2% HNO₃), EDTA (0.05%, w/v, in 2% HNO₃ or deionized water) and HNO₃ (2%) for an additional 10 runs. Finally, ME (0.1%, v/v, in 2% HNO₃) or Cys (0.18%, w/v, in 2% HNO₃) was used for further washing. Comparisons of further washing using ME and Cys were also studied. The results are shown in Fig. 2.

Fig. 2 Washing curves using different washing reagents and further washing for the introduction of 50 ng ml⁻¹ Hg²⁺. For legends with two washing modes listed (see vertical dash dot lines), the first one is for washing after 3.8 min (10 runs) of introduction using different reagents, and the second one for further washing after 7.6 min (20 runs) using 0.1% ME or 0.18% Cys in 2% HNO₃ solution.

According to Fig. 2, with a longer introduction time, all the washing reagents can still wash mercury out to the baseline within 2 min. However, a difference can be seen for different washing reagents with further washing. For example, the washout using ME (both in H₂O and HNO₃), DTT (in HNO₃) and HNO₃ shows a jump at 4.2 min (the 11th run), while no obvious jump can be seen for DTT (in H₂O), Cys and EDTA (both in water and HNO₃). The jump can be attributed to the washout of Hg adhered to walls. Therefore, ME (both in H₂O and HNO₃), owing to its high affinity of mercapto groups (–SH) to mercury, gives out a larger jump than other reagents such as HNO₃ and EDTA solutions. Although EDTA is a good chelator for many metal ions, it seems that it is not an ideal coordinate reagent for mercury.

Before 7.6 min (the 20th run), all the washing reagents resulted in mercury counts to its blank level. However, it is just a superficial phenomenon. If it is further washed using ME (in 2% HNO₃ solution), another jump comes out quickly. Of all these reagents, washing with HNO₃ gives the biggest jump, which can be attributed to the fact that it has no affinity to mercury. There is also a jump for EDTA (both in H₂O and HNO₃) in the curves, owing to its weak chelation ability to mercury. The washing efficiency of EDTA in HNO₃ is higher than in H₂O. Other reagents, such as DTT and Cys, give almost no jump when washed with additional ME because of the existence of –SH in DTT and Cys.

The washing effeciency of ME and Cys was also compared. No matter which one was used first, followed by the other one, no jump appears in the curves after 20 runs. Therefore, both of them can serve as suitable washing reagents. Yet, if the odor and central nervous system (CNS) toxicity of mercaptoalkanol are considered, it is preferable to choose Cys as a washing reagent.

L-Cysteine addition in standard and sample mercury solution

Since Cys is a good affinitive reagent to mercury and could wash out mercury efficiently, it was added to the standard and sample solutions before they were introduced into the ICP-MS system to eliminate or decrease the memory effect of mercury.

A good linear calibration curve could be obtained after addition of Cys with the correlation coefficient of 0.9999 (n = 6) (for the commonly used 2% HNO₃, r = 0.9864, n = 6). Although the detection limit (defined as three times the standard deviation of the blank solution) for addition of Cys was 0.024 μg l⁻¹, which was higher than nitric acid (0.018 μg l⁻¹), it was satisfactory for routine applications. Furthermore, the addition of Cys into sample solutions simplifies the washout process. After each sample analysis, the commonly used 2% HNO₃ or even deionized water can clean the system to the blank level (data not shown here).

For QC/QA, the Chinese CRM human hair (GBW 07601) and Canadian CRM dogfish muscle Dorm-2 were used to evaluate the effect of Cys addition after routine CH₃OH/NaOH digestion.¹¹ The Hg concentration in human hair obtained using this method was 0.35 ± 0.06 μg g⁻¹ (n = 10), which was 97.2% of the certified value (0.36 ± 0.05 μg g⁻¹). The recovery of mercury in dogfish muscle was 98.3% (certified value 4.64 ± 0.26 vs. determinated value 4.56 ± 0.30 μg g⁻¹, n = 6).

However, long-time and continuous introduction of Cys into the instrument may cause clogging owing to deposition of carbon on cones. To solve this problem, other techniques such as flow injection (FI) techniques^2,7–10 can be applied. On the other hand, for the relatively short-time analysis with timely manual cone cleaning and no FI system available, it is still recommendable to use the method of Cys addition.

Acknowledgements

The authors acknowledge financial support from the Natural Science Foundation of China (10490180) and the International Atomic Energy Agency (Contract No. 13246).

References

1. A. Woller, H. Garraud, F. Martin, O. F. X. Donard and P. Fodor, J. Anal. At. Spectrom., 1997, 12, 53.
2. C. F. Harrington, S. A. Merson and T. M. D’Silva, Anal. Chim. Acta, 2004, 505, 247.
3. M. J. Campbell, G. Vermeir, R. Dams and P. Quevauviller, J. Anal. At. Spectrom., 1992, 7, 617.
4. D. S. Bushee, Analyst, 1988, 113, 1167.
5. J. Allibone, E. Fatemian and P. J. Walker, J. Anal. At. Spectrom., 1999, 14, 235.
6. J. Yoshinaga and M. Morita, J. Anal. At. Spectrom., 1997, 12, 417.
7. C. F. Harrington, S. A. Merson and T. M. D’Silva, Anal. Chim. Acta, 2004, 505, 247.
8. D. Beauchemin, K. W. M. Siu and S. S. Berman, Anal. Chem., 1988, 60, 2587.
9. J. Tyson, Fresenius’ J. Anal. Chem., 1988, 329, 663.
10. K. E. Jarvis, A. L. Gray, and R. S. Houk, Handbook of Inductively Coupled Plasma Mass Spectrometry, Blackie, New York, 1992, p. 119.
11. M. Logar, M. Horvat, I. Falnoga and V. Stibilj, Fresenius’ J. Anal. Chem., 2000, 366, 453.

---

Footnote

† This work was presented in part at the 2nd China International Symposium on Persistent Toxic Substances, Beijing, May 15–18, 2005.

---