Use of a Solution Cathode Glow Discharge for Cold Vapor Generation

Aug 19, 2008 - A novel vapor-generation technique is described for mercury determination in aqueous solutions. Without need for a chemical reducing ag...
0 downloads 4 Views 621KB Size
Download PDF
Anal. Chem. 2008, 80, 7043–7050

Use of a Solution Cathode Glow Discharge for Cold Vapor Generation of Mercury with Determination by ICP-Atomic Emission Spectrometry Zhenli Zhu,†,‡ George C.-Y. Chan,‡ Steven J. Ray,‡ Xinrong Zhang,† and Gary M. Hieftje*,‡ Department of Chemistry, Key Laboratory for Atomic and Molecular Nanosciences of Education Ministry, Tsinghua University, Beijing 100084, P.R. China, and Department of Chemistry, Indiana University, 800 East Kirkwood Avenue, Bloomington, Indiana 47405 A novel vapor-generation technique is described for mercury determination in aqueous solutions. Without need for a chemical reducing agent, dissolved mercury species are converted to volatile Hg vapor in a solution cathode glow discharge. The generated Hg vapor is then transported to an inductively coupled plasma for determination by atomic emission spectrometry. Mercury vapor is readily generated from a background electrolyte containing 0.1 M HNO3. Vapor generation efficiency was found to be higher by a factor of 2-3 in the presence of low molecular weight organic acids (formic or acetic acids) or alcohols (ethanol). Optimal conditions for dischargeinduced vapor generation and reduced interference from concomitant inorganic ions were also identified. However, the presence of chloride ion reduces the efficiency of Hgvapor generation. In the continuous sample introduction mode, the detection limit was found to be 0.7 µg L-1, and repeatability was 1.2% RSD (n ) 11) for a 20 µg L-1 standard. In comparison with other vapor generation methods, it offers several advantages: First, it is applicable to both inorganic and organic Hg determination; organic mercury (thiomersal) can be directly transformed into volatile Hg species without the need for prior oxidation. Second, the vapor-generation efficiency is high; the efficiency (with formic acid as a promoter) is superior to that of conventional SnCl2-HCl reduction. Third, the vapor generation is extremely rapid and therefore is easy to couple with flow injection. The method is sensitive and simple in operation, requires no auxiliary reagents, and serves as a useful alternative to conventional vapor generation for ultratrace Hg determination.

been established throughout the world. Mercury has a high vapor pressure and is the only metal recognized as being able to form a monatomic vapor at room temperature. As a result, cold vapor generation (CVG)2 has been combined with atomic absorption spectrometry (AAS) or atomic fluorescence spectrometry (AFS) for mercury determination in many application areas.3 The CVG technique is particularly attractive because of its high sample introduction efficiency and its ability to separate the analyte (Hg) from complex matrixes. Conventional CVG makes use of chemical reducing agents such as sodium tetrahydroborate(III) or tin(II) chloride for the reduction of Hg(II) to Hg(0).4 This reduction is very efficient and can be accomplished in both batch and online systems. However, these reducing agents are expensive and unstable (i.e., freshly prepared reagents are required). In addition, traditional tetrahydroborate(III)-based CVG methods are prone to interferences by transition metals due to their interaction with the reducing reagent or catalytic decomposition of the reaction products on the surfaces of the reduced concomitant metals.5 Efforts continue in an attempt to develop new approaches for CVG of trace elements. Electrochemical vapor generation, which uses electrons as reductants,6-8 has been shown to be a suitable alternative to chemical reaction CVG. The most significant advantage of electrochemical generation is that it obviates the need for chemical reducing reagents. However, there are several shortcomings for electrochemical vapor generation.9 First, the cathode material must be carefully selected because it strongly influences the performance of the electrochemical process. Second, the cathode surface requires frequent conditioning, typically on a daily basis. Transition metal ions interfere with the vapor production by being reduced and deposited on the cathode

Because of its high toxicity and because it accumulates and persists in the environment and biota,1 there is increasing concern about the determination and monitoring of mercury. Mercury pollution is recognized as a major environmental issue and health hazard for humans. Strict regulations for the maximum allowable mercury concentration in environmental and food samples have

(2) Hatch, W. R.; Ott, W. L. Anal. Chem. 1968, 40, 2085–2087. (3) Cullen, M. Atomic Spectrometry in Elemental Analysis; Blackwell Publishing: Oxford, U.K., 2004. (4) Welz, B.; Sperling, M. Atomic Absorption Spectrometry; Wiley-VCH Verlag Gmbh: Weinheim, Germany, 1999. (5) He, Y. H.; Hou, X. D.; Zheng, C. B.; Sturgeon, R. E. Anal. Bioanal. Chem. 2007, 388, 769–774. (6) Arbab-Zavar, M. H.; Rounaghi, G. H.; Chamsaz, M.; Masrournia, M. Anal. Sci. 2003, 19, 743–746. (7) Cerveny, V.; Rychlovsky, P.; Netolicka, J.; Sima, J. Spectrochim. Acta, Part B 2007, 62, 317–323. (8) Li, X.; Wang, Z. H. Anal. Chim. Acta 2007, 588, 179–183. (9) Denkhaus, E.; Golloch, A.; Guo, X. M.; Huang, B. J. Anal. At. Spectrom. 2001, 16, 870–878.

* To whom correspondence should be addressed. Phone: +1 812 855 2189. Fax: +1 812 855 0985. E-mail: [email protected]. † Tsinghua University. ‡ Indiana University. (1) Clarkson, T. W.; Magos, L.; Myers, G. J. N. Engl. J. Med. 2003, 349 (18), 1731–1737. 10.1021/ac8011126 CCC: $40.75  2008 American Chemical Society Published on Web 08/19/2008

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

7043

Figure 1. Schematic diagram of the closed SCGD cells as a vapor generation unit for ICP emission measurement: (A) U-tube design and (B) improved version for reduced dead volume.

surface. Furthermore, adsorption of gaseous reaction products reduces vapor generation efficiency. Free radicals have been used as reductants and applied to vapor generation of many elements, including Hg. Photochemically induced (PI)-CVG, which utilizes a photoreaction process based on UV-vis irradiation,5,10-14 is a newly emerging vaporgeneration method. Hydrogen and carboxyl radicals that arise from photodissociation of low molecular weight organic acids (e.g., formic, acetic, and propionic) are employed to reduce mercury ions to the neutral vapor. This process obviates the need for expensive and high-purity reducing reagents (i.e., NaBH4 or SnCl2), is amenable to speciation analysis with or without chromatographic separation, and is less prone to interference from concomitant ions. Recently, a sono-induced (SI) Hg CVG technique was reported, which also eliminates the need for conventional reducing agents.15-18 Reduction of Hg(II) to Hg(0) was accomplished in a sonoreactor owing to the reducing gases and radicals formed upon sonolysis of formic acid added to the sample solution. A drawback of SI-CVG is that the presence of oxidizing substances in the sample precludes the reduction of Hg(II) to Hg(0) by ultrasound. In addition, it is used only in a batch system and is difficult to couple with flow injection. It is well-known that reactive free radicals and electrons can be generated by a glow discharge. The electrolyte cathode atmospheric-pressure glow discharge (ELCAD)19 is a relatively new optical emission source that uses an electrolyte solution as the cathode in a dc glow discharge, with a metal counter electrode positioned in the atmosphere above the solution. It operates in an environment saturated with water vapor and is reported to contain H and OH radicals and free electrons. Various ELCADlike discharge arrangements have been investigated in different laboratories, mostly for achieving miniaturized and/or simplified analytical methods.20-27 One particularly simple design, referred to as a solution-cathode glow discharge (SCGD),28-30 has recently been developed by the Hieftje group. This new version offers greatly reduced size and improved emission sensitivity. It exhibits detection limits and precision levels approaching or even better than those obtainable by inductively coupled plasma atomic emission spectrometry (ICP-AES) for a range of elements. However, its detection limits for Hg30 are still in the range of tens 7044

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

of micrograms per liter, which is still not adequate for direct Hg determination in many environmental or food samples. The present study was designed to test the feasibility and to develop a new Hg vapor generation technique based on the SCGD. The new system eliminates the use of chemical reducing agents. Several experimental parameters were optimized, and analytical figures of merit were determined. The effect of organic species on vapor generation efficiency and potential interference by concomitant elements were also investigated. EXPERIMENTAL SECTION Instrumentation. Two designs of the closed SCGD cell have been constructed and evaluated (Figure 1). One design (Figure 1A) is in the form of a closed U-tube and is similar to the one previously described.21 The tungsten anode, tapered to a rounded (10) Guo, X. M.; Sturgeon, R. E.; Mester, Z.; Gardner, G. J. Anal. Chem. 2003, 75, 2092–2099. (11) Guo, X. M.; Sturgeon, R. E.; Mester, Z.; Gardner, G. J. Anal. Chem. 2004, 76, 2401–2405. (12) Vieira, M. A.; Ribeiro, A. S.; Curtius, A. J.; Sturgeon, R. E. Anal. Bioanal. Chem. 2007, 388, 837–847. (13) Zheng, C. B.; Li, Y.; He, Y. H.; Ma, Q.; Hou, X. D. J. Anal. At. Spectrom. 2005, 20, 746–750. (14) Han, C. F.; Zheng, C. B.; Wang, J.; Cheng, G. L.; Lv, Y.; Hou, X. D. Anal. Bioanal. Chem. 2007, 388, 825–830. (15) Gil, S.; Lavilla, I.; Bendicho, C. Anal. Chem. 2006, 78, 6260–6264. (16) Gil, S.; Lavilla, I.; Bendicho, C. J. Anal. Atom. Spectrom. 2007, 22, 569– 572. (17) Gil, S.; Lavilla, I.; Bendicho, C. Spectrochim. Acta, Part B 2007, 62, 69–75. (18) Ribeiro, A. S.; Vieira, M. A.; Willie, S.; Sturgeon, R. E. Anal. Bioanal. Chem. 2007, 388, 849–857. (19) Mezei, P.; Cserfalvi, T. Appl. Spectrosc. Rev. 2007, 42, 573–604. (20) Mezei, P.; Cserfalvi, T. J. Phys. D: Appl. Phys. 2006, 39, 2534–2539. (21) Cserfalvi, T.; Mezei, P. J. Anal. At. Spectrom. 2005, 20, 939–944. (22) Cserfalvi, T.; Mezei, P. J. Anal. At. Spectrom. 2003, 18, 596–602. (23) Kim, H. J.; Lee, J. H.; Kim, M. Y.; Cserfalvi, T.; Mezei, P. Spectrochim. Acta, Part B 2000, 55, 823–831. (24) Mezei, P.; Cserfalvi, T.; Janossy, M. J. Phys. D: Appl. Phys. 1998, 31, L41– L42. (25) Davis, W. C.; Marcus, R. K. J. Anal. At. Chem. 2001, 16, 931–937. (26) Yagov, V. V.; Gentsina, M. L. J. Anal. Chem. 2004, 59, 64–70. (27) Jenkins, G.; Franzke, J.; Manz, A. Lab Chip 2005, 5, 711–718. (28) Webb, M. R.; Andrade, F. J.; Hieftje, G. M. J. Anal. At. Spectrom. 2007, 22, 766–774. (29) Webb, M. R.; Andrade, F. J.; Hieftje, G. M. Anal. Chem. 2007, 79, 7807– 7812. (30) Webb, M. R.; Andrade, F. J.; Hieftje, G. M. Anal. Chem. 2007, 79, 7899– 7905.

Figure 2. Schematic diagram of the experiment setup (P: peristaltic pump). Refer to Figure 1 for details of the solution-cathode glow discharge (SCGD). The insert shows the construction of the glass liquid separator (GLS).

point, was positioned several millimeters above the sample solution. A graphite electrode (cathode) in the reservoir grounds the solution. A 1-kΩ resistor inserted between the tungsten anode and the positive output of a high-voltage power supply (Kepco BHK 2000-0.1MG, Flushing, NY) is used to stabilize the discharge current. The discharge is operated in constant-current mode. The discharge is initiated by temporarily reducing the distance between the anode and the solution. The plasma is ignited between a 0.7 mm diameter area at the tip of the tungsten rod and the solution surface. The gap, or height of the discharge, was set to 2 mm in the present study. This design was used in the initial experiments to investigate its feasibility as a vapor-generation method for Hg. In experiments with this design, the sample was mixed with the blank solution when it was introduced into the cell. Therefore, the dominant source of dead volume was the solution volume (about 15 mL). In this case, the cell acts as an exponential dilution flask with a very long (at least 20 min) wash-out time. To overcome this limitation, a more compact cell (gas-phase internal volume of about 3 mL) was developed (cf. Figure 1B). In this design, the sample solution is delivered through a glass capillary, which is conveniently adapted from a 10-µL disposable micropipet (Fisher Scientific Co., Pittsburgh, PA) with a 1.1 mm outer diameter and 0.38 mm inner diameter. A portion of the solution that is not vaporized by the discharge overflows into a glass waste reservoir; this overflow provides an electrical connection between the discharge and the solution in the reservoir. In this case, the discharge is sustained between the anode and the outlet of the capillary tube. Samples were supplied to the cell with the aid of a Gilson (Middleton, WI) peristaltic pump at a flow rate of 2.5 mL min-1. To reduce fluctuations caused by pulsations of the peristaltic pump, a home-built pulse damper was included between the pump and the SCGD. The damper consisted of a knotted length of peristaltic pump tubing (1.52-mm inner diameter Tygon tube of 45-cm length before knotting). In the flow injection experiment, an electrically actuated pneumatic valve (Rheodyne 7010) was used; switching was accomplished by means of a Rheodyne 5710 actuator, which was driven by a square wave produced by a Beckman 9010 function generator. The injection channel was used to fill a sample loop (100 µL) attached to the valve. When Hg-containing samples were fed into the discharge, volatile Hg0 vapor was produced. All generated products issuing from the plasma were swept by an argon stream through a

gas-liquid separator (GLS) and then to a commercial ICP-AES spectrometer (Horiba Jobin-Yvon ACTIVA ICP-AES spectrometer, Longjumeau, France) for detection (cf. Figure 2). The transfer tube between the GLS and the ICP was about 60 cm (i.d. 4 mm), with an internal volume calculated to be about 18 mL. The GLS (shown in Figure 2) is a condenser-based design consisting of two concentric tubes. The water-cooled condenser is 20 cm (length) × 1 cm (i.d.) × 2 cm (o.d.). Coolant water circulating through the condenser is held at approximately 5 °C. The operating conditions of the ICP-AES instrument are as follows: forward power 1200 W, outer and intermediate gas flows 14 L min-1 and 0.6 L min-1, respectively. The inner diameter of the injector tube of the ICP torch was 2 mm. The central-channel gas flow was optimized, as will be discussed later; the studied range was from 0.5 to 1.1 L min-1. The ACTIVA ICP-AES spectrometer employs a Czerny-Turner configuration for wavelength dispersion with a two-dimensional CCD detector for vertically resolved emission measurement. The instrument therefore has the capability to simultaneously measure the entire vertical emission profile of the plasma over a defined spectral range (the x-axis of the CCD detector corresponds to wavelength and the y-axis to vertical location in the plasma). In the present study, an extended region in the ICP was used for photon collection; the observation height ranged from 6 to 15 mm above the load coil. The spectral resolution of the spectrometer was 8 pm. Unless otherwise specified, integrated spectral intensity was used throughout the study; that is, the intensity of all CCD pixels over the entire spectral peak was integrated. Off-peak background correction was used, and background intensity was measured at CCD pixels adjacent to the spectral peaks. For continuous sample introduction, the integration time was set at 3 s with 10 replicates for each measurement. For flow injection measurements, a 0.2 s integration time was used. Reagents and Samples. All chemicals were of analyticalreagent grade. A Hg(II) stock solution (1000 mg L-1) was prepared from mercuric nitrate; working standards were made by suitable dilution with deionized water. Unless otherwise specified, all sample solutions were prepared in dilute nitric acid (Merck) with the final pH adjusted to 1.0. To enhance vapor generation efficiency, HCOOH (Merck), C2H5OH, or CH3COOH was added to each sample and standard solution. The following substances were employed for the interference study: NaCl, NaNO3, KNO3, Cu(NO3)2, AgNO3, Ni(NO3)2, Co(NO3)2, KMnO4, K2Cr2O7, and HCl. Thiomersal (98%) was purchased from Sigma. Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

7045

Figure 3. Comparison of the efficiency of Hg signal generation for three alternative methods: (A) ICP with Hg solution introduced by means of a pneumatic nebulizer; (B) SCGD plasma used to generate Hg0, which is then directed into the ICP; and (C) SCGD used in a cathodic electrolysis mode. Hg0 is then sent into the ICP. Hg concentration: 1 mg L-1. Notice the scale difference in the three spectra.

RESULTS AND DISCUSSION Comparison of Hg Signal-Generation Efficiency. The efficiency of the SCGD was investigated with the U-tube design (cf. Figure 1A). A schematic diagram of the experiment is shown in Figure 2. The efficiencies of three sample introduction methods were compared. The first involved conventional solution sample introduction into the ICP-AES by a concentric nebulizer and a spray chamber. For the second approach, the sample was introduced into the SCGD and the generated vapor was then swept by an argon carrier into the ICP. In this case, the SCGD was operated at a discharge current of 50 mA. Since hydrogen (a product of electrolysis) is produced at the graphite cathode when the discharge is ignited, the feasibility of generating Hg vapor in this cathode zone was also investigated by switching the cathode and anode of the U-tube. This mode involves essentially electrolysis, since the graphite cathode is immersed in the sample solution whereas the SCGD is formed in the open atmosphere. Because analyte emission from the ICP depends not only on the amount of introduced sample (i.e., on the Hg introduction efficiency) but also on the excitation characteristics of the plasma, it is necessary to maintain constant excitation conditions in order to compare the three methods fairly. It is well-known that water loading in the plasma has a strong effect on its excitation characteristics and robustness,31 so maintaining a constant flux of water aerosol/ vapor into the plasma is a prerequisite for matched plasma excitation characteristics. Therefore, in the present study, both the nebulizer and the SCGD continued in operation regardless of which mode of sample introduction was used. If the Hg-containing sample was introduced into the SCGD, a blank solution (HNO3 at pH 1.0) was then fed into the nebulizer channel (and vice versa). (31) Novotny, I.; Farinas, J. C.; Wan, J. L.; Poussel, E.; Mermet, J. M. Spectrochim. Acta, Part B 1996, 51, 1517–1526.

7046

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

The Hg (253.7 nm) emission spectra obtained by conventional pneumatic nebulization, vapor generation from the SCGD plasma, and vapor generation from SCGD cathode electrolysis under matched ICP conditions are depicted in Figure 3. The greatest intensity was obtained from vapor generation by the SCGD plasma; the signal is about 16 times of that produced by pneumatic nebulization. Although electrolysis in the cathode zone also produces volatile Hg species, its efficiency is much lower than that of the SCGD plasma and is even slightly inferior to conventional nebulization. Figure 3 clearly demonstrates the feasibility of utilizing the SCGD as a vapor-generation method for Hg determination when the sputtered/evaporated water vapor is condensed and removed by a GLS. Neutral-atomic emission from several metallic elements has been observed by others in various ELCAD-like sources. The presence of such emission verifies the presence of free atoms (e.g., Na) in the plasma that are unlikely to exist in solution. Because the rotational temperature of the SCGD plasma is about 3000 K,30 thermal dissociation of sputtered species from the solution is a likely mechanism for the generation of free Na or Hg atoms. In addition, there are many similarities between the SCGD and flame absorption/emission chemistries.28 Therefore, a similar thermal atomization process might be responsible for Hg reduction here. In addition, hydroxyl radicals, atomic hydrogen, and free electrons are all present in the SCGD, as evidenced by experimental observation of very strong OH emission bands and the significantly Stark-broadened (due to collisions with free electrons) hydrogen atomic emission line in its Balmer series.19 Therefore, the radicals and free electrons in the SCGD might also play an important role in the reduction process. Because of its high vapor pressure and ability to exist as a monatomic vapor, the generated

Figure 4. Optimization of carrier gas flow rate (A) and discharge current (B) for the low-volume cell. Error bars in the figure represent standard deviations of the results.

Hg can be efficiently transferred to the ICP with less loss than most other elements. Use of the SCGD as a vapor-generation technique for other known volatile hydride-forming elements (As, Se, Pb, Sn) was also investigated. However, there was no measurable ICP emission from these elements even when they were introduced into the SCGD at a level of 10 mg L-1, a concentration that is orders of magnitude higher than the detection limits in the ICP by conventional solution nebulization. This observation is in agreement with the work of Cserfalvi et al.,21 in which the generated vapor from an ELCAD plasma was introduced directly into the ICP without a GLS. The absence of signals from hydride-forming elements might be the result of the high temperature of the ELCAD plasma, at which the hydrides either do not form or else dissociate and are lost in transport to the ICP. This possibility will be the subject of future investigations. Optimization of Carrier Gas Flow Rate and Discharge Current. The U-tube cell has a large dead volume and is not appropriate for coupling with flow injection. Therefore, a compact cell (Figure 1B) was designed in which the solution was introduced through a capillary tube. Experiments were undertaken to optimize the carrier gas flow rate and the discharge current on the basis of the signal from a 100 µg L-1 Hg(II) solution. Here, the effect of varying the carrier gas on ICP emission intensity is considered as a whole, and its effect on the transport efficiency of the volatile Hg vapor and plasma excitation characteristics are not separated. Moreover, from this discussion onward, the conventional nebulizer and spray chamber system described in the previous section was removed from the system and no longer used. As can be seen in Figure 4A, the maximum signal was obtained at an argon flow rate of 488 mL min-1. Higher flow rates reduced the signal, probably because of plasma cooling by the higher carrier gas flow.31 At lower flow rates, the intensity also dropped, likely because of reduced transport efficiency of the generated Hg vapor. On the basis of these results, a flow rate of 488 mL min-1 was selected for the remaining experiments. The influence of discharge current on vapor generation efficiency was also investigated. The discharge can be operated

at currents ranging from 40 to 90 mA. It has been observed previously that atomic emission from the SCGD becomes stronger with increasing current.32 Figure 4B shows that this trend holds also for its effect on the Hg-emission intensity from the ICP, which increases with discharge current. Not surprisingly, the bulk solution vaporization rate also rises with discharge current. Presumably, the higher discharge power enhances both reduction and volatilization efficiency. However, at very high currents, excessive heating of the anode took place and the discharge became unstable. Therefore, a discharge current of 80 mA was employed in the remaining parts of the study. Effect of Organic Substances. It was reported for both PICVG5 and SI-CVG15 that the sensitivity for vapor generation can be enhanced by the presence of low-molecular weight organic substances. Accordingly, the effect of ethanol, formic acid, and acetic acid on the Hg generation efficiency was investigated also in the present study. These additives were spiked into the sample (prepared in 0.1 M HNO3) at a final concentration span from 0.1% to 1% (v/v). Ethanol was studied only up to a concentration of 0.7% because higher ethanol levels extinguished the ICP. Figure 5 shows the effect of these additives. It is clear that even a small amount (0.1%) of a low-molecular weight organic substance can enhance the Hg signal appreciably, and higher concentrations of the organic additives were even more effective. Although the reason for this enhancement was not investigated, it may be that the organic compounds yield additional reducing reagents in the discharge. In addition, the organic substances might alter the boiling point or surface tension of the sample solution and consequently change its vaporization rate, thereby contributing to the intensity enhancement. Among the three studied additives, formic acid is the most effective; in the presence of 1% formic acid, Hg emission was enhanced by a factor of 3. On the basis of these results, a 1% formic acid solution in 0.1 M HNO3 was used as the supporting electrolyte in later research. (32) Cserfalvi, T.; Mezei, P.; Apai, P. J. Phys. D: Appl. Phys. 1993, 26, 2184– 2188.

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

7047

Figure 5. Enhancement effects of organic additives on the efficiency of Hg-vapor generation. Error bars in the figure represent standard deviations of the results.

Figure 6. Comparison of signal generation efficiency of the SCGD with traditional chemical reduction. (A, traditional CVG method (SnCl2, 5%; HCl, 5%); B, SCGD without HCOOH; C, SCGD with 1% HCOOH).

Comparison of Cold Vapor Generation Efficiency with Conventional SnCl2-HCl System. The efficiency of the SCGD system was estimated roughly from a comparison of the AES signals it yields with those arising from the traditional approach based on SnCl2-HCl reduction.12,18 Identical sample introduction flow rates were used for the conventional SnCl2-HCl chemical vapor generation system and the SCGD. The results are shown in Figure 6. In a plain HNO3 medium without the addition of formic acid, the efficiency of the SCGD was estimated to be about 45%, assuming unity efficiency for the SnCl2-HCl system. The efficiency of the SCGD rose to 124% in the presence of 1% HCOOH. It can be concluded that the SCGD system is a high-efficiency vapor generation technique for Hg. The effect of the small amount of hydrogen that is produced from the discharge has not been considered in this case. However, several studies33,34 have (33) Goldwasser, A.; Mermet, J. M. Spectrochim. Acta, Part B 1986, 41, 725– 739. (34) Murillo, M.; Mermet, J. M. Spectrochim. Acta, Part B 1989, 44, 359–366.

7048

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

reported that the presence of hydrogen in the ICP can increase emission line intensities. Therefore, the efficiency cited here is only an estimation. Measurement of Thiomersal. The toxicity of mercury is related not only to its concentration but also to its chemical form. Mercury can be present in environmental samples as inorganic mercury or organomercurials; in general, the latter are more toxic. Although CVG is favored for the determination of Hg at the ultratrace level, it is not effective for organomercury species and requires a prior oxidation of the organic mercury to yield inorganic mercury.13,35 In the present study, thiomersal was chosen as a representative organomercury compound to test whether the SCGD can be employed as a direct method for vapor generation from organic mercury species. Thiomersal (sodium ethylmercurithiosalicylate,C9H9HgNaO2S) is commonly used as both an antiseptic and an antimicrobial preservative in pharmaceutical formulations. It has been added to vaccines, cosmetics, eye drops, and contact lens solutions as a preservative to prevent bacterial contamination.16 It has been shown that traditional cold vapor methods cannot generate volatile Hg vapor directly from thiomersal without prior oxidation. For example, Bendicho and coworkers16 utilized UV/H2O2 to oxidize the thiomersal in a sample before its conversion to Hg vapor using SI-CVG. In the present study, thiomersal was converted directly to volatile Hg by using the SCGD. An initial experiment was carried out in the presence of 1% formic acid. When compared to the emission from a sample containing only inorganic mercury at an identical concentration, the intensity of the thiomersal-containing sample showed a 30% enhancement. It is possible that the enhancement is caused by the organic ligands, which serve a function similar to that of low-molecular weight organic substances discussed above (cf. Figure 5). The production of Hg vapor from thiomersal in the absence of HCOOH was also demonstrated. In agreement with the earlier discussion involving inorganic Hg, the process was then less effective; the Hg emission signal was about 40% lower than in the presence of HCOOH. It was not clear what product was formed when thiomersal was introduced into the SCGD; further experiments are needed to identify the volatile Hg species. Regardless, it can be concluded that the SCGD is a very efficient device for generating Hg0 vapor from thiomersal; its success suggests its potential as a promising technique for the determination of organic mercury. Interference from Inorganic Ions. To evaluate the susceptibility of this new cold vapor generation approach to matrixinduced errors, the effect of several alkali, transition, and noble metals and anions was studied (cf. Table 1). All samples contained inorganic Hg(II) at a concentration of 100 µg L-1 in 0.1 M HNO3, with 1% HCOOH added to promote vapor-generation efficiency. For samples that contain simple alkali-metal ions (i.e., NaCl, NaNO3, and KNO3) at 10 mg L-1, the effect was negligible; however, signal depressions of 12-21% were observed at concentrations of 100 mg L-1 of the same interferences. The effect of oxidizing agents was also investigated because it has been reported that even small amounts can preclude the reduction of Hg(II) in SI-CVG.15 No interference from K2Cr2O7 was observed even at 100 mg L-1, although KMnO4 caused a 40% (35) Fernandez, C.; Conceicao, A. C. L.; Rial-Otero, R.; Vaz, C.; Capelo, J. L. Anal. Chem. 2006, 78, 2494–2499.

Table 1. Effect of Inorganic Species on SCGD-Induced Vapor Generationa interference NaCl NaCl NaNO3 NaNO3 KNO3 KNO3 K2Cr2O7 K2Cr2O7 KMnO4 KMnO4 AgNO3 Cu(NO3)2 Cu(NO3)2 Co(NO3)2 Ni(NO3)2 AuCl3 Cl- (HCl) Cl- (HCl) Cl- (HCl) a

concentration (mg L-1)

recovery (%)

10 100 10 100 10 100 10 100 10 100 10 10 100 10 10 10 0.1 M 0.01 M 0.001 M

98 79 94 85 94 88 99 102 101 69 110 109 111 104 110 107 7 69 89

Hg concentration: 100 µg L-1.

signal suppression at a concentration of 100 mg L-1; no effect from KMnO4 was detected at the 10 mg L-1 level. The KMnO4 interference may be attributable to the oxidation of the formic acid promoter; discoloration of KMnO4 was readily observed at a concentration of 100 mg L-1. In addition, many reports36,37 have noted that the use of KMnO4 can cause several problems, including precipitation of MnO2 onto which Hg can adsorb. It was found that noble (Au, Ag) and transition (Cu, Co, and Ni) metals caused slight enhancements in the Hg signal. This is somewhat surprising, since signal suppression from noble metals (i.e., Au, Ag, Pt, and Pd) has been reported38 in conventional CVG. The reason for this difference is currently unknown but might be related to differences in the vapor generation mechanism. The effect of the chloride anion was also investigated. Serious suppression caused by chloride has been reported in both PI-CVG13 and SI-CVG.15 The suggested interference mechanism involves stabilization of Hg(II) by chloride owing to the formation of strong complexes.15 In the present study, the suppressive effect of chloride was observed even at the 0.001 M level; the interference became more severe with increasing chloride ion concentrations. Analytical Performance and Figures of Merit. The analytical characteristics of SCGD-induced cold vapor generation were evaluated under optimal operating conditions. With the integrated area under the spectral peak as a quantitative parameter, calibration curves were linear up to an Hg concentration of 200 µg L-1 (R ) 0.9997). Repeatability, expressed as the relative standard deviation of the spectral-peak area, was 1.2% (n ) 11) for the 20 µg L-1 Hg standard. The limit of detection (LOD), using the definition 3σ/m (σ is the standard deviation corresponding to 10 blank measurements and m is the slope of the calibration graph), was 0.7 µg L-1. The developed SCGD(36) Caroli, S.; Forte, G.; Iamiceli, A. L.; Lusi, A. Microchem. J. 1996, 54, 418– 428. (37) Christmann, D. R.; Ingle, J. D., Jr. Anal. Chim. Acta 1976, 86, 53–62. (38) Morita, H.; Tanaka, H.; Shimomura, S. Spectrochim. Acta, Part B 1995, 50, 69–84.

Figure 7. Temporal pattern of 100 µL injections of Hg(II) solution at a concentration of 100 µg L-1.

induced vapor generation method was used for the determination of thiomersal in synthetic water samples (spiked with 20 µg L-1 thiomersal); good recovery (97.8%) was observed. These results indicate that the new procedure has acceptable sensitivity and repeatability. It is attractive to apply the developed SCGD-induced vapor generation technique to transient analysis in order to reduce sample consumption and boost sample throughput. When used as an atomic emission source, the SCGD has already been applied to transient analysis.29 In the present study, it was found that the reaction time for Hg-vapor generation is virtually instantaneous; the volatile Hg vapor appears as soon as the sample is introduced into the SCGD, in contrast to the long reaction time in other CVG methods. For example, it was reported that the optimal length of the photoreactor tubing for PI-CVG is 5 m, which corresponds to an irradiation time of 40s.13 For SI-CVG, the optimal irradiation time was reported to be 10 min.15 In addition to the fast chemical kinetics, the narrow capillary (0.38-mm inner diameter) used in the compact SCGD cell to deliver sample solutions reduces dead volume and limits dispersion, which further favors transient analysis. Figure 7 shows the temporal profiles of repetitive flow injection peaks from 100 µL injections of a Hg(II) solution. The LOD was calculated to be 1.2 µg L-1 in the flow injection mode. The relative standard deviation of the peak areas was only 2.2% (n ) 8). CONCLUSION A novel cold vapor generation technique for Hg analysis was developed based on the solution cathode glow discharge (SCGD). It eliminates the need for chemical reducing reagents such as sodium tetrahydroborate or stannous chloride. The reduction of Hg(II) can be readily accomplished in a plain HNO3 medium (about 0.1 M). The presence of organic substances, such as formic acid, greatly enhances the vapor-generation efficiency. It is Hgselective and simple in operation and can be used as an alternative to conventional CVG. SCGD-induced vapor generation offers a number of other advantages. First, not only inorganic mercury (Hg(II)) but also organic mercury (thiomersal) can be directly transformed to volatile Hg vapor; no prior oxidation is needed. Second, the vapor generation efficiency appears to be high. The Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

7049

efficiency is even better than that of the conventional SnCl2-HCl method in the presence of HCOOH promoter. Third, the chemical kinetics are fast and the technique can be readily coupled with flow injection. Vapor generation is virtually instantaneous and occurs as soon as the sample is introduced into the SCGD. Furthermore, compared to other CVG methods, interferences from concomitant ions are mild. Oxidants do not suppress the signal even at the 10 mg L-1 level. SCGD-induced vapor generation is also potentially useful for field analysis since it can be readily adapted to portable CV-AAS/AFS systems. Future directions include investigating the underlying mechanism of the vaporgeneration process and assessing its application to a range of real samples.

7050

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

ACKNOWLEDGMENT Zhenli Zhu acknowledges the Chinese Scholarship Council for award of a fellowship. The authors are grateful to Horiba Jobin-Yvon (Longjumeau, France) for loan of the ACTIVA ICP spectrometer used in this work. This research was supported by the U.S. Department of Energy through Grant DE-FG0298ER14890.

Received for review June 2, 2008. Accepted July 23, 2008. AC8011126