Purge-and-Trap Isothermal Multicapillary Gas Chromatographic

Purge-and-Trap Isothermal Multicapillary Gas Chromatographic ...https://pubs.acs.org/doi/pdf/10.1021/ac980361lSimilarby I Rodriguez Pereiro - ‎1998 ...
0 downloads 4 Views 94KB Size
Download PDF
Anal. Chem. 1998, 70, 4063-4069

Purge-and-Trap Isothermal Multicapillary Gas Chromatographic Sample Introduction Accessory for Speciation of Mercury by Microwave-Induced Plasma Atomic Emission Spectrometry Isaac Rodriguez Pereiro,† Andrzej Wasik,‡ and Ryszard Łobin´ski*

CNRS, EP132, Helioparc, 2, avenue Pr. Angot, F-64000 Pau, France

A compact device based on purge-and-trap multicapillary gas chromatography was developed for sensitive speciesselective analysis of methylmercury and Hg2+ by atomic spectrometry. The operating mode includes in situ conversion of the analyte species to MeEtHg and HgEt2 and cryotrapping of the derivatives formed in a 0.53-mm-i.d. capillary, followed by their flash (60 mL min-1) compatible with an MIP AES detector (no dilution with a makeup gas is required). Developments regarding each of the steps of the analytical procedure and effects of operational variables (sample volume, purge flow, trap temperature, separation conditions) are discussed. The device allows speciation of MeHg+ and Hg2+ down to 5 pg g-1 in urine and, after a rapid microwave-assisted hydrolysis, down to 0.1 ng g-1 in solid biological samples with a throughput of 6 samples/h. The analytical protocols developed were validated by the analysis of DORM-1 (dogfish muscle), TORT-1 (lobster hepatopancreas), and Seronorm urine certified reference materials. Determination of and discrimination between ionic mercury and methylmercury in environmental, clinical, and foodstuff samples is, in view of their (different) toxicity, an important area of trace element analytical chemistry.1 Whereas the determination of mercury down to low picogram per milliliter levels in environmental samples is considered to be a routine analytical task owing to cold vapor atomic absorption (CV AAS) or atomic fluorescence spectrometric (CV AFS) techniques, that of methylmercury and, especially, the simultaneous species-selective analysis for Hg2+ * To whom correspondence should be addressed. Tel.: + 33 5 59 80 68 84. Fax: + 33 5 59 80 12 92. E-mail: [email protected]. † On leave from Department of Analytical Chemistry, Universidad de Santiago de Compostela, 15706 Santiago, Spain. ‡ On leave from Chemical Faculty, Department of Analytical Chemistry, Politechnika Gdan´ska, ul. G. Narutowicza 11/12, 80-952 Gdan´sk, Poland. (1) Van Burg, R.; Greenwood, M. R. In Metals and their Compounds in the Environment; Merian, E., Ed.; VCH: Weinheim, 1991; p 1045. S0003-2700(98)00361-8 CCC: $15.00 Published on Web 08/22/1998

© 1998 American Chemical Society

and MeHg+ continues to be a difficult task without an available commercial system.2 Developments over the past decade in the speciation analyses of Hg in aqueous samples have enabled the cumbersome and lowyield solvent and solid-phase extraction procedures3 to be replaced by more reliable methods based on in situ formation and liquidgas extraction of volatile derivatives of Hg2+ and MeHg+, followed by their cryotrapping, thermal desorption, pyrolysis, and on-line detection of Hg0 by CV AAS4 or CV AFS.5,6 Similar approaches were reported for sediment extracts7-10 and biological tissue hydrolysates,11-13 slowly replacing, in the latter case, the traditional Westo¨o¨14 procedure. The use of NaBH4 as a derivatization reagent has remained predominant,4,7,12 despite the short half-life (2 h) of MeHgH,15 ambiguity about the formation of mercury dihydride, and limited reliability of conversion of Hg2+ to Hg0. The only alternative is the formation of ethyl derivatives with NaBEt4,16 the use of which has been reported to suffer from the dismutation of ethyl derivatives on the chromatographic column11 as well as its duration (up to 45 min).5 Typically, the derivatized mercury species are trapped in a silanized quartz glass U-trap which is packed with a chromatographic sorbent [e.g., Chromosorb (80-100 mesh)], coated with a nonpolar phase (e.g., 10% OV-101, SP-2100), and cooled with liquid nitrogen to -196 °C. The retained compounds are thermally desorbed from the trap by heating it electrically4,7,11,12 (2) Lobinski, R.; Marczenko, Z. Spectrochemical Analysis for Trace Metals and Metalloids; Elsevier: Amsterdam, 1996; pp 517-542. (3) References in Bloom, N. Can. J. Fish Aquat. Sci. 1989, 46, 1131-1140. (4) Puk, R.; Weber, J. H. Anal. Chim. Acta 1994, 292, 175-183. (5) Bloom, N. Can. J. Fish Aquat Sci. 1989, 46, 1131-1140. (6) Stockwell, P. B.; Corns, W. T. Int. Labmate 1994, 19, 33-35. (7) Tseng, C. M.; de Diego, A.; Martin, F. M.; Donard, O. F. X. J. Anal. At. Spectrom. 1997, 12, 629-635. (8) Bloom, N.; Fitzgerald, W. F. Anal. Chim. Acta 1988, 208, 151-161. (9) Liang, L.; Horvat, M.; Bloom, N. S. Talanta 1994, 41, 371-379. (10) Horvat, M.; Bloom, N. S.; Liang, L. Anal. Chim. Acta 1993, 281, 135-152. (11) Fischer, R.; Rapsomanikis, S.; and Andrea, M. Anal. Chem. 1993, 65, 763. (12) Tseng, C. M.; de Diego, A.; Martin, F. M.; Amouroux, D.; Donard, O. F. X. J. Anal. At. Spectrom. 1997, 12, 743-750. (13) Liang, L.; Horvat, M.; Cernichiari, E.; Gelein, B.; Balogh, S. Talanta 1996, 43, 1883-1888. (14) Westo¨o¨, G. Acta Chem. Scand. 1967, 21, 1790. (15) Filippelli, M.; Baldi, F.; Brinckman, F. E.; Olson, G. J. Environ. Sci. Technol. 1992, 26, 1457-1460. (16) Rapsomanikis, S. Analyst (London) 1994, 119, 1429-1439.

Analytical Chemistry, Vol. 70, No. 19, October 1, 1998 4063

Figure 1. Scheme of the purge-and-trap multicapillary chromatography interface for sample introduction for the determination of mercury species.

or in a chromatographic oven.5,8 A number of approaches require a two-step preconcentration, including a preconcentration step on a trap filled with a sorbent (Carbotrap or Tenax) at room temperature, followed by flash thermal desorption of the analytes onto a packed column.9,10,13 The advantages of this approach are its compatibility with AAS,4,7,11,12 AFS,5,6,8-10 and FAPES17 and its high tolerance to water in the purged gas. The drawbacks include high dead volume of the system (at common flow rates used, 80150 mL min-1, the peak half-width is ∼10 s), the need for silanization,7,12 and the likely dismutation reactions.11 The system is not compatible with an MIP, while other detectors (AAS, AFS) require an additional pyrolysis step. A thermal desorption temperature-programmed (-196 f 200 °C) chromatographic separation run following the purge-and-trap step takes typically 5-20 min. An alternative is the use of capillary (e0.53 mm i.d.) traps.15,18,19 The compact size of such injection systems allows them to be mounted on a conventional GC oven. The low internal volume of the liner makes an efficient injection on a capillary column possible and, consequently, enables a high-resolution analysis, isothermal on a 50-m column15 or by oven temperature programming when shorter columns are used.18,19 The disadvantage is the limited (up to 10 mL) volume of the solution purged, the need for a cryostat supplying the coolant for the water trap, the possibility of losses of analytes at low temperatures (-15 °C), and the need for a full-size chromatographic oven. The purpose of this work was to develop a compact accessory for sample introduction into an atomic spectrometer that would enable a simultaneous species-selective analysis for MeHg+ and Hg2+ and would combine the advantages of the approaches discussed above and eliminate their drawbacks. This was achieved by introducing a minimulticapillary column for isother(17) Jimenez, M. S.; Sturgeon, R. E. J. Anal. At. Spectrom. 1997, 12, 597-601. (18) Ceulemans, M.; Adams, F. C. J. Anal. At. Spectrom. 1996, 11, 201-206. (19) Gerbersmann, C.; Heisterkamp, M., Adams, F. C.; Broekaert, J. A. C. Anal. Chim. Acta 1997, 350, 273-285.

4064 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

mal high-flow-rate separations compatible with the MIP atomic emission detector, optimizing purge-and-trap injection onto such a column, and replacing the cryostat with an efficient 30-cm-long tubular Nafion membrane to remove water from the purged stream of analytes. EXPERIMENTAL SECTION Apparatus. The interface developed is shown in Figure 1. The device consists of four integrated custom-made components: (1) a purge vessel, (2) a membrane purifier to eliminate water from the gas phase, (3) a capillary cryotrap, and (4) an isothermal isolated compartment housing a multicapillary column. The purge vessel from which analytes were purged with nitrogen was a glass tube of 15 cm × 3 cm i.d. with a volume of ∼70 mL, or one of 9 cm × 3 cm i.d. with a volume of 40 mL. The purge gas enters the sample solution through a glass tube (3 mm i.d., 6 mm o.d.), of which a section of the wall of length 1 cm just close to the bottom was fritted. The bottom itself was not fritted. The purge vessel was equipped with a head made of Teflon with three openings: one for the purge gas introduction tube, one enabling the on-line introduction of NaBEt4 solution, and one enabling the evacuation of the gas phase containing the derivatized species. The plugging of the capillary cold trap was avoided by removing water vapor from the purge gas stream. The dryer consisted of a Nafion desiccant tube (1.2 mm i.d.) enclosed in a polypropylene outer tube (1/4 in. o.d.). Nitrogen was forced to flow between both tubes at 1 L min-1 in the direction opposite to the purge gas stream to sweep any water that diffused through the Nafion tube walls. The dryer used was made of a commercial 24-in. Nafion dryer, 1/16 in. o.d. (Permapure, Toms River, NJ), by cutting the tube and the polypropylene housing in order to reduce the length of the tubular membrane to 30 cm. The Nafion dryer was mounted directly on one of the openings of the purge vessel. The derivatized species were trapped on a fused silica capillary (11 cm × 0.53 mm i.d.) coated with a 5-µm CP-Sil 8 CB layer

(Chrompack, Middelburg, The Netherlands). The capillary was housed in a 1-mm-i.d. stainless steel tube which was housed in a 16-mm-i.d. brass tube. Electrical contacts on the stainless steel tube enable connection to a transformer 100 VA/5 V and passing through it a current of ∼20 A. A heating rate of 300 °C s-1 could be achieved in this way. Nitrogen, precooled to a desired temperature in liquid nitrogen, was passed between the stainless steel and the brass tube, which enabled the trap to be cooled to -150 °C. Dedicated electronics enabled the regulation of the trapping and desorption temperatures. A six-way GC injection valve (Valco Europe, Schenkon, Switzerland) made it possible to divert the carrier stream off the column (cryotrapping phase) or on the column (desorption phase). The trap, the valve, and the connected capillaries were housed in an Al body, of which the temperature could be controlled between the ambient temperature and 200 °C. Signal acquisition of the MIP AED was triggered at the moment when the trap heating was started. Analyte species were separated on a custom-made minimulticapillary column (22 cm × 1200 capillaries × 0.038 mm i.d. × 0.25 µm SE-54). Alternatively, a multicapillary column consisting of 919 1-m long, 40-µm-i.d. capillaries coated with 0.2 µm of SE30 (Alltech Associates, Inc., Deerfield, IL) was used. The column was connected at the injector end to a deactivated alumina tube (∼0.3 m × 0.53 mm i.d.). At the detector site, a deactivated silica tube (0.32 mm i.d.) served as a transfer line to the detector. The column was housed in a custom-made compact oven. Its dimensions were 220 mm × 10 mm o.d. in the case of the minimulticapillary and 200 × 200 × 100 mm in the case of the 1-m capillary. An HP model G2350A microwave-induced plasma atomic emission detector (Hewlett-Packard, Wilmington, DE) was used. Data were handled using HP model D3398A ChemStation software. An HP model 6890 gas chromatograph with a split injector and an autosampler was used to optimize the separations on the multicapillary columns. Biological materials were dissolved in a vessel similar to that used for the purge-and-trap process, fitted with a 10-cm condenser using a Synthewave model S402 (2.45 GHz, maximum power 300 W) microwave digester (Prolabo, Fontenay-sous-Bois, France). Reagents, Standards, and Solutions. HPLC grade solvents and analytical grade chemicals obtained from Aldrich (Milwaukee, WI) and Milli-Q water (Millipore, Milford, MA) were used throughout unless otherwise stated. Glassware was cleaned using a common detergent, thoroughly rinsed with tap and Milli-Q water, soaked for 12 h in a 10% nitric acid solution, and finally rinsed with Milli-Q water just before use. Sodium tetraethylborate (NaBEt4) was obtained from Strem Chemicals (Bisscheim, France). The reagent was manipulated under dry nitrogen to prevent its degradation, and fresh 0.1% (w/ v) aqueous solutions were prepared every 8 h. The acetate buffer was prepared by dissolving 0.1 M of sodium acetate in water and adjusting the pH to 4 with acetic acid. Tetramethylammonium hydroxide (TMAH), a 25% aqueous solution, was obtained from Fluka (St. Quentin-Fallavier, France). Mercury chloride (HgCl2) and methylmercury chloride (MeHgCl) were obtained from Aldrich. Individual stock solutions (∼1 mg mL-1 as Hg) were prepared by dissolving each compound in a 1% (v/v) HNO3 solution. Diluted standards and mixtures of both compounds were prepared in the same solvent.

Table 1. Optimum Purge-and-Trap, GC, and AED Parameters for Speciation of Methylmercury and Inorganic Mercury Purge-and-Trap Conditions purge vessel temperature 20 °C purge time 6 min purge flow 75 mL min-1 Nafion dryer flow 1 L min-1 trapping temperature -100 °C desorption temperature 150 °C GC Parameters 1-m column column flow oven program 22-cm column column flow oven program AES Parameters transfer line temperature cavity block temperature wavelength helium makeup flow spectrometer purge flow solvent vent time 1-m column 22 cm column H2 pressure O2 pressure

65 mL min-1 isothermal at 70 °C 120 mL min-1 isothermal at 45 °C 250 °C 250 °C 253.65 nm 140 mL min-1 2 L min-1 0.19 min 0.10 min 50 psi 10 psi

The reference materials, DORM-1 (dogfish muscle) and TORT-1 (lobster hepatopancreas), with certified contents of methylmercury and total mercury were obtained from the National Research Council of Canada (NRCC). The reference material of urine (Seronorm trace elements in urine) was purchased from Nycomed Pharma (Oslo, Norway). The lyophilized urine material was first reconstituted with 5 mL of aqueous 1% HNO3 solution. The reconstituted samples were stored in the dark at 4 °C and analyzed within 5 days. Procedures. Operating Conditions. Optimum purge-and-trap, GC separation, and AED conditions for the determination of organomercury species are summarized in Table 1. Analysis of Water. A sample of 10-50 mL of water [spiked deionized water or filtered (0.45 µm) seawater] was placed in the purge vessel. The pH was adjusted to 4.0 with 1 mL of the buffer solution, and 100 µL of the NaBEt4 solution was added. The mixture was purged at 75 mL min-1 for 6 min. Once the purgeand-trap step has finished, the position of the six-way valve was changed, and the trap was electrically heated to desorb analytes onto the chromatographic column. Analysis of Urine. A sample (100 µL) was spiked on 10 mL of water, followed by the addition of 1 mL of the buffer solution and 500 µL of the NaBEt4 solution. The analysis was further carried out as in the case of water samples. Microwave-Assisted Purge-and-Trap Analysis of Organomercury in Biological Tissues. A freeze-dried sample of 0.1-0.2 g and 5 mL of TMAH solution were placed in a reaction vessel and exposed to the microwave field at 45 W for 2.5 min. An aliquot (25-100 µL) of the cool TMAH extract was placed in the purge vessel and made up with 10 mL of water. Volumes of 1 mL of buffer and 800 µL of the NaBEt4 solution were added. The procedure was continued as described above. Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

4065

RESULTS AND DISCUSSION Removal of Water from the Analytes Carrying Stream. The main problem associated with the use of an open-tubular capillary (0.53 mm i.d) cryotrap is the immediate plugging by water vapor present in the gas stream containing the extracted analytes.15,18,19 Traditionally, excessive H2O is removed from the purge gas by a condenser kept at about -15 °C, mounted upstream of the trap to retain water from the gas stream carrying analytes. The obvious disadvantage of such a system is the use of a cumbersome and expensive cryostat and analyte losses at subzero temperatures. Another drawback is that purge flow rates commonly applied in such systems to prevent blocking the trap are relatively low (1030 mL min-1), which results in long analysis times (10-20 min).15,18,19 Therefore, the use of tubular Nafion H2O-permeable membranes that have become popular for drying volatile hydroand halohydrocarbons20,21 and organosulfur compounds22,23 was considered. Nafion is an anionic polymer with a tetrafluoroethylene backbone and perfluorinated ether side chains, terminating in hydrophilic sulfonic acid sites.24 The actual dryer consisted of a Nafion tube either embedded with 5-Å molecular sieves21 or enclosed in a polypropylene outer tube. A dry gas flows in countercurrent between the Nafion and the polypropylene tubes to sweep water that diffused through the Nafion tube walls.20 The latter approach was chosen because of its higher efficiency. Attempts were made to minimize the length of Nafion tubing required to remove water from the gas stream carrying the analytes. When working with a countercurrent gas stream flow 10 times larger than the purge flow, a 30-cm piece of Nafion was found sufficient to avoid plugging of the trap, even after 30 min of purging at 100 mL min-1. The use of this Nafion dryer in place of a cryostat allows for a considerable reduction in the size of the device and avoids the risk of losses of less volatile analytes (e.g., HgEt2). Optimization of Working Conditions. The large number of parameters that affect derivatization reaction, purge-and-trap, separation, and detection conditions required a systematic optimization approach. Detection conditions were set first according to Rodriguez Pereiro et al.25 Conditions for the separation of MeHgEt and Et2Hg on a minimulticapillary column were optimized using a split injection of these standards in hexane. Then the purge-and-trap conditions were optimized. The principal variables that can affect the yield of a purge-and-trap process are purge flow, purge time, and trap temperature. These parameters are interdependent and were optimized by the total factorial experiment at literature conditions16 for the formation of mercury and methylmercury ethyl derivatives (pH 4.0, cNaBEt4 ) 0.01% w/v). Standards of MeHgCl and HgCl2 (∼200 pg of each as Hg) were spiked over 10 mL of water. The values used were as follow: trap temperature, -30, -40, -60, -80, -100, and -120 °C; purge flow, 35, 55, 75, and 100 mL; purge time, 0-12 min every 2 min. Once the optimum conditions were found, effects of the volume and the temperature of the analyzed solution were investigated. (20) Cochran, J. W. J. High Resolut. Chromatogr. 1987, 10, 573-575. (21) Zygmunt, B. J. Chromatogr. 1996, 725, 157-163. (22) Ridgeway, R. G. J.; Bandy, A. R.; Thornton, D. C., Mar. Chem. 1991, 33, 321-334. (23) Wardencki, W. J. Microcolumn Sep. 1995, 7, 51-57. (24) Leckrone, K. J.; Hayes, J. M. Anal. Chem. 1997, 69, 911-918. (25) Rodriguez Pereiro, I.; Wasik, A.; Lobinski, R. J. Chromatogr. 1998, 795, 359-370.

4066 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

Figure 2. Effect of temperature of the cryogenic trap on the trapping efficiency of the ethylated MeHg+ and Hg2+ species. Purge time, 8 min.

Cryotrapping Temperature. The minimum allowable trap temperature is limited by the efficiency of the removal of water from the analyte carrying gas prior to trapping (trap blockage is faster at lower temperatures) and by the necessity to achieve a reasonably fast desorption rate, especially important when a minimulticapillary column is used (the lower the temperature, the slower the desorption and the broader the peak). Indeed, the use of a multicapillary column at a constant temperature does not allow the refocusing of the analyte band at the top of the column. These limitations set the lowest useful value of the trapping temperature for mercury species at -120 °C. Figure 2 shows that, in the case of MeHgEt and Et2Hg, it is actually sufficient to work at -100 °C, even at high flow rates, to achieve the quantitative trapping yield. The trapping temperature can be increased at reduced flow rate, at the expense of time necessary for the quantitative recovery of analyte compounds from the analyzed solution. Purge Flow and Time. The efficiency of liquid-gas extraction of MeHgEt and HgEt2 is controlled by two factors: (1) kinetics of the ethylation reaction and (2) kinetics of the transfer of the derivatized species from the reaction liquid into the gas phase. The two processes are simultaneous and interdependent. The effect of purge time at different purge flow rates is shown in Figure 3. It shows that a time of 6 min is necessary at a flow rate of 75 mL min-1 and cannot be reduced further by increasing the flow rate. This suggests that the kinetics of the derivatization and not that of purge is controlling the process. A purge flow of 75 mL min-1 and a time of 6 min were selected as working conditions. The time necessary for the quantitative recovery of MeEtHg and Et2Hg at 55 mL min-1 is almost twice as long, while no quantitative recovery of HgEt2 is possible at all at 30 mL min-1. Figure 3 also demonstrates the high efficiency of the capillary trap. After 10 min of purge at 100 mL min-1, no loss of MeHgEt or HgEt2 from the trap was noticed at -100 °C. Purge Vessel Temperature. The high efficiency of water removal by the Nafion dryer in comparison with a cryostat allowed us to attempt to accelerate the ethylation reaction and thus the purge step by heating the purged solution, either conventionally or in a microwave field. Indeed, it turned out that, even when the purged solution was heated to 80 °C, the 30-cm Nafion tube dryer showed sufficient efficiency to avoid the blockage of the capillary trap during 5 min with a purge gas flow rate of 100 mL min-1.

Figure 3. Effect of purge time and purge flow on the liquid-gas extraction of the ethylated MeHg+ and Hg2+ species. Cryotrap temperature, -100 °C.

No significant difference between the yield of purge performed at room temperature (20 °C) and that at 80 °C was observed, so it was decided to work at room temperature. Similar results were obtained when microwave heating was used. The possibility to work at higher temperatures offers the option of analyzing solid samples by microwave-assisted digestion-liquid-gas extraction in the developed system. Effect of the NaBEt4 Concentration. Different volumes (from 50 µL to 1 mL) of a freshly prepared NaBEt4 solution (0.1%) were used for the simultaneous derivatization and purge-and-trap of MeHg+ and Hg2+ spiked over 10 mL of water. No differences in peak area or height were noticed. A default volume of 100 µL of 0.1% (w/v) borate solution was used. Effect of Sample Volume. Concentration detection limits achieved with the device developed are controlled by the volume of the analyzed sample, provided that the yield of the purge process is kept constant. At the experimental conditions given in Table 1, a 5-fold increase in the sample volume resulted in a 10-20% decrease of the purge yield (similar for MeHgEt and Et2Hg), which gives the possibility to increase the detection limit by a factor of 4 by increasing the volume of the analyzed solution. Optimization of the Chromatographic Separation Conditions. The objective was to find conditions which would enable the isothermal separation of the ethylmercury species in order to simplify the hardware by replacing the regular temperatureprogrammed oven by a much smaller oven kept at a constant temperature. An attractive way to achieve this was to use a multicapillary column that combines the large number of theoretical plates of a small diameter (40 µm i.d.) capillary with a highflow cross section owing to the use of a bundle of ∼1000 such capillaries.26-30 The consequence of the maximum efficiency of this column at high flow rates is the reduced separation time. We showed recently that the use of a 1-m multicapillary column for the separations of organotin and organolead species offered (26) Malakhov, V. V.; Sidelnikov, V. N.; Utkin, V. A. Dokl. Ross. Akad. Nauk (in Russian) 1993, 329, 749-751. (27) Cooke, W. S. Today’s Chemist at Work 1996, January, 16-20. (28) Schmitt, V.; Rodriguez Pereiro, I.; Lobinski, R. Anal. Commun. 1997, 34, 141-143. (29) Rodriguez Pereiro, I.; Schmitt, V.; Lobinski, R. Anal. Chem. 1997, 69, 47994807. (30) Rodriguez Pereiro, I.; Lobinski, R. J. Anal. At. Spectrom. 1997, 12, 13811385.

Figure 4. Chromatograms obtained with the developed purge-andtrap multicapillary GC device for a mixture of MeHg+ and Hg2+ species, ∼60 pg mL-1 each. 1, MeHgEt; 2, Et2Hg. (A) Column, 1 m × 919 capillaries × 43 µm (each); isothermal at 70 °C; flow rate, 65 mL min-1. (B) Minicolumn, 22 cm × 1200 capillaries × 38 µm (each); isothermal at 45 °C; flow rate, 120 mL min-1.

the possibility of reducing the time necessary for analysis at flow rates compatible with those required by the AED detector.28-30 To date, however, no sample introduction by thermal desorption onto a multicapillary column was attempted, apparently because of the need for rapid injection. Indeed, the combination of split injection with an autosampler allows one to achieve injection times under 0.1 s, which does not produce extra peak broadening. Nevertheless, the absence of a solvent and the relatively big difference between the volatilities of MeEtHg and HgEt2 makes it possible to accept the loss of a number of theoretical plates without losing the baseline resolution. Figure 4A shows a chromatogram at a constant temperature with the device developed using a 1-m multicapillary column. Baseline resolution is evident, whereas peak broadening is observed in comparison with split injection. The peak half-widths are 0.012 min in comparison with 0.007 min for MeHgEt and 0.029 min in comparison with 0.016 min for HgEt2. This results in a loss of the number of theoretical plates down to 2500, which is still sufficient to achieve the baseline separation. Because the separation is isothermal, the GC oven can be eliminated and replaced by a more compact insulated housing thermostated at a given temperature, which reduces the size (20 × 20 × 10 cm) and the cost of the system. An attractive alternative is the replacement of a still relatively cumbersome 1-m column by a custom-made short, straight minimulticapillary column of length of ∼20 cm. A chromatogram obtained for such a column (Figure 4B) shows practically no difference compared to that obtained with 1-m column. In terms of conditions, however, the separation temperature is considerably lower (45 instead of 70 °C) and the flow rate is higher (120 mL min-1). The unquestionable advantage is the reduction of the size of the separation unit which replaces the regular chromatographic oven. It should be noted that the 5-fold decrease of the column length results in only 2-fold reduction of the number of chromatographic plates which can be explained in terms of theory of multicapillary columns.31 Analytical Characteristics. Detection Limits. Absolute detection limits, defined as 3 times the signal-to-noise ratio, of the MIP AED were ∼0.5 and 2.0 pg for MeHg+ and Hg2+, respectively. (31) Sidelnikov, V. Personal communication.

Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

4067

Table 2. Figures of Merit of the Calibration Curve Obtained in the Peak Height Mode for the Analysis of a Mixture of MeHg+ and Hg2+ compound

calibration interval (total amount, ng as Hg)

MeHgEt HgEt2

0.05-10 0.1-10

intercept

slope

R2

3.4 ( 4.3 355 ( 4 0.9995 12.1 ( 4.7 98.5 ( 0.9 0.9996

The higher detection limit for Hg2+ is the consequence of the wider chromatographic peak. When a sample of 10 mL is purged, the above values correspond to concentration detection limits of 0.05 pg mL-1 for MeHg+ and 2 pg mL-1 for Hg2+. The relatively high value for Hg2+ is the consequence of the existence of a blank signal for this species. As discussed above, detection limits of 0.012 pg mL-1 can be obtained by increasing the volume of the purged solution. Linearity and Reproducibility. Linearity of the purge-and-trap injection under the conditions described above (see Procedures) was evaluated. The results obtained are shown in Table 2. Good linearity was observed for both compounds in the concentration interval studied. The intercept value for Hg2+ is significatively different from zero because of a blank signal. Reproducibility of injection was typically around 5% for amounts of MeHg+ and Hg2+ injected of ∼0.2 ng (which corresponds to a concentration of 20 pg mL-1 for 10 mL of water placed in the purge vessel). Blank Problems. Whereas no blank signal was observed for methylmercury down to a concentration of 10 fg mL-1 (detection limit), a HgEt2 peak was systematically observed when a sample of deionized water was analyzed under the conditions given in Table 1. A positive correlation between the buffer volume and the HgEt2 was observed. Purification of the buffer by reaction with 0.1% NaBEt4 and purging with nitrogen gas for 20 min, prior to its use, was done to reduce this contamination; nevertheless, a residual signal of ∼10% of the initial one still remained in the system. This blank did not increase when higher NaBEt4 concentrations were used. Hence, it is concluded that the origin of this contamination was laboratory air and the deposition of atmospheric particles on the labware. This can be removed only by working in clean room conditions. Carryover and Memory Effects. In initial experiments (purge flow 100 mL min-1), a considerable HgEt2 peak appeared in the blank analysis (10 mL of water with optimum concentrations of reagents) following the analysis of a real-world sample. The origin of this carryover was the Hg2+ species and not Et2Hg, since no signal was observed if a blank solution was analyzed without an NaBEt4 addition. The carryover peak persisted even after cleaning the Teflon head of the purge vessel and the glass frit with HNO3 (10% in water). It was thus presumed that the carrier gas carries aerosol particles of the sample solution that arrive at the dryer and react with aerosol particles carrying the borate in the next run. The problem was minimized by limiting the purge flow to 75 mL min-1 and placing a drop-catcher between the top of the vessel and the Nafion tube. Nevertheless, when the solution purged is intensely heated (e.g., in the case of microwave-assisted analysis of biological samples), it is necessary to install a gas/liquid separator at the top of the purge vessel to prevent large aerosol particles from 4068 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

Figure 5. Chromatograms for the analysis of real samples using the device developed. Conditions are as described in Procedures. 1, MeHgEt; 2, HgEt2. (A) Seawater (Arcachon bay, France). (B) TORT-1 reference material.

entering the Nafion tubing. In the case of severe contamination of the Nafion, the latter can be cleaned by purging a 10% HNO3 solution for 15 min. Acid vapors were found to be effective enough to remove inorganic compounds from the Nafion surface. Analysis of Real-World Samples. Seawater. A chromatogram for a seawater sample filtered through a 0.45-µm filter is shown in Figure 5A. Generally, only inorganic mercury was found; some samples contained methylmercury levels below 0.5 ng L-1. The figures of merit were similar to those shown in Table 2. Urine Samples. The analysis of undiluted urine samples leads to loss of recovery because of the apparently insufficient amount of NaBEt4 and of foaming. An intake of 100 µL of urine sample required a 5-fold increase in the concentration of NaBEt4. At this sample intake, foaming was not considerable. The sample intake can be increased at least to 0.5 mL, but the addition of an antifoaming agent and an increase in the NaBEt4 concentration are then necessary. With a solution intake of 100 µL, the improved detection limits were 5 and 20 pg mL-1, respectively. Results of the quantification using an external calibration graph agreed with those obtained by the method of standard additions. The values obtained for Hg2+ (no MeHg+ signal was present) (50.4 ( 2.3 and 50.0 ( 4.7 ng mL-1 for five determinations on two consecutive days) were in good agreement with the certified value of 48.0 ng mL-1. Biological Tissues. Prior to a purge-and-trap analysis, the biological tissue sample must be solubilized. It was verified that

the microwave-accelerated solubilization of biological tissues with tetramethylammonium hydroxide (TMAH) developed initially by Szpunar et al.32,33 leads to no loss of Hg2+ and MeHg+ species. Microwave-assisted digestion was applied recently, in view of the mercury speciation, by Tseng et al.12 and Gerbersmann et al.,19 leading, however, in the latter case to a relatively long (20 min) procedure. A reoptimization of the conditions using a microwave digester with an IR sensor and computer-controlled reaction temperature led to a decrease of the solubilization time down to 2.5 min at 45 W (15% of the maximum power) using a 25% TMAH solution. The high sensitivity of the system required dilution of the sample hydrolysate. A chromatogram obtained for a lobster hepatopancreas sample is shown in Figure 5B. The recovery tests showed, however, that, at the conditions optimized for water samples, the ethylation yield was low (∼20%). It turned out to be necessary to increase the NaBEt4 concentration by a factor of 8 to achieve an ethylation yield over 90%. With a solution intake of 100 µL of a sample of 200 mg solubilized in 5 mL of TMAH, the detection limits achieved were 0.1 and 0.5 ng g-1 for methylmercury and inorganic mercury, respectively, in biological tissues. At this level, foaming was not a problem. The procedure developed was validated by the analysis of two certified reference materials: DORM-1 and TORT-1. The results agree with the certified values, as shown in Table 3. CONCLUSIONS The application of a multicapillary GC minicolumn for the isothermal separation of ethylated mercury species offers an attractive possibility to produce a compact accessory for an atomic spectrometer enabling speciation analysis of mercury in environmental samples. Besides the compact size of the device, it offers, (32) Szpunar, J.; Schmitt, V. O.; Lobinski, R.; Monod, J. L. J. Anal. At. Spectrom. 1996, 11, 193-199. (33) Szpunar, J.; Ceulemans, M.; Schmitt, V. O.; Adams, F. C.; Lobinski, R. Anal. Chim. Acta 1996, 332, 225-232.

Table 3. Results for the Determination of Mercury in Biological Certified Reference Materials found (µg/g as Hg) material

MeHg+

Hg2+

total

certified (µg/g as Hg) Hga

MeHg+

total Hg

DORM-1 0.71 ( 0.01 0.18 ( 0.04 0.89 ( 0.05 0.73 ( 0.06 0.80 ( 0.07 TORT-1 0.13 ( 0.01 0.24 ( 0.06 0.37 ( 0.07 0.13 ( 0.01 0.33 ( 0.06 a Determined as the sum of the concentrations found of MeHg+ (as Hg) and Hg2+.

in comparison with the packed column systems, a 10-fold narrower focused injection band at similar flow rates and, consequently, a much higher sensitivity with low-volume detectors such as, e.g., MIP AED. The analytical cycle is shorter because of the 10-fold reduction of the time required for the chromatographic separation. The use of a tubular Nafion membrane eliminates the need for a cryostat limiting analyte losses, allowing for higher sample volumes to be purged at higher purge gas flow rates and, consequently, enabling faster and more sensitive analysis. ACKNOWLEDGMENT This study was financially supported by the EC (Contract SMT4-CT96-2044 Automated Speciation Analyzer) and the French Government (Ministe`re de l’EconomiesD.G.C.C.R.F.). I.R.P acknowledges a postdoctoral fellowship from the Spanish government.

Received for review March 30, 1998. Accepted July 7, 1998. AC980361L

Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

4069