JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1992 VOL. 7 15 Determination of Methylmercury Species by Capillary Column Gas Chromatography With Axially Viewed Inductively Coupled Plasma Atomic Emission Spectrometric Detection Takunori Kato Hokkaido Institute of Environmental Sciences N 19 W72 Kitaku Sapporo Hokkaido 060 Japan Takashi Uehiro Akio Yasuhara and Masatoshi Morita National Institute for Environmental Studies 16-2 Onoga wa Tukuba lbaraki 305 Japan Methylmercury species were determined by capillary column gas chromatography (GC) coupled with inductively coupled plasma (ICP) atomic emission spectrometry. Methylmercury species were converted into the iodide form and separated on a chemically bonded capillary column. An axially viewed ICP with an echelle monochromator was used as a highly selective and sensitive mercury detector for GC.The detection limit of methylmercury was calculated to be 3 pg as Hg at a signal-to-noise ratio of 2. The linear range was more than three orders of magnitude. The relative standard deviation of ten replicate measurements of 30 pg of methylmercury (as Hg) was 5%. Keywords Inductively coupled plasma atomic emission spectrometry; axial viewing; coupled gas chromato- graphy; mercury speciation; methylmercury determination Since a plasma emission detection method for gas chro- matography (GC) was first published,' this technique has been widely utilized in the speciation of volatile organo- metallics because of its high selectivity high sensitivity and wide linear range regardless of their molecular structure.Among the plasma detectors the microwave-induced plasma (MIP) has more often been used compared with the inductively coupled plasma (ICP) or direct current plasma (DCP) probably owing to its small size and good matching of gas flow rates. The MIP detector is very sensitive for volatile species containing mercury selenium arsenic and other element^,^-^ but its shortcoming is that the introduc- tion of solvent into the plasma results in quenching of the discharge. For this reason it is necessary to bypass the solvent vapour before it enters the cavity or to ignite the plasma after the solvent has passed through the cavity. With the ICP the plasma is maintained in the presence of an organic solvent and therefore such a procedure is not necessary.In this work the capability of ICP atomic emission spectrometry (AES) as a detector for GC was examined for the determination of alkylmercury compounds. As it has been reported that an axially viewed (AXV) ICP has a more intense analyte emission and lower background intensities than those in a conventional side-viewed (SDV) ICP,5-7 the AXV-ICP was employed in the experiment. For separation of alkylmercury compounds a short chemically bonded quartz capillary column was used because it was expected to have the least active sites to adsorb alkylmercury. Experimental Instrumentation The GC-ICP-AES system and connections are shown in Fig. 1 and the optimum operating conditions are summar- ized in Table 1. The column employed here is a chemically bonded fused-silica capillary column (3 x 0.35 mm i.d.) coated with methylsilicone (5 pm layer) (Gasukuro Kogyo Tokyo Japan).The splitless injection mode was employed. The end of the capillary column was directly introduced to the head of the inner tube of the plasma torch. To avoid Capillary 1 ICP (axial view) I I colum Heater P k ' Recorder Fig. 1 system (axially viewed) Schematic diagram of the capillary column GC-ICP-AES condensation part of the column between the GC outlet and the plasma torch was passed through a 2 mm i.d. poly(tetrafluorethy1ene) (PTFE) tube and heated with a tape heater to the maximum column temperature (1 50 "C). The capillary column in the torch was warmed indirectly by passing the argon sample gas over an electrically heated Nichrome wire. In order to determine the optimum conditions for ICP a metallic mercury reservoir was placed in the oven of the gas chromatograph to serve as a continuous and constant mercury source.In order to maintain the GC-ICP interface conditions the reservoir was inserted between the column and the transfer capillary (80 cm). For the AXV construction the torch box assembly was rotated so that the torch was in a horizontal attitude. Optical observations were made through the exhaust hole at the top of the torch box while the plasma gases were extracted through the normal side-viewing port. The monochromator employed was an echelle type (Spectra Span 111). A reduced half-sized image of the AXV plasma was focused on the entrance slit of the monochro- mator with a quartz lens (50 mm diameter f= 10 cm) or a reduced one tenth-size image of the SDV plasma was focused on the entrance slit with a quartz lens (10 mm diameter f= 5 cm).A Jeol DX300 mass spectrometer equipped with a Hewlett-Packard Model 57 1 OA gas chromatograph was used to identify methylmercury species. A methylsilicone capillary column 15 m x 0.35 mm i.d. 5 pm layer) (Gasukuro Kogyo) was used for the separation.16 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1992 VOL. 7 Table 1 GC-ICP-AES instrumentation and optimum conditions Gas chromatography- Gas chromatograph Column Carrier gas He Flow rate Injector temperature 140 "C Column temperature Purge time 0.5 min Hewlett-Packard Model 5890A with splitless injector Fused-silica capillary column (3 in x 0.3 mm i.d.) coated with methylsilicone film thickness 5 pm 7.5 ml min-l (head pressure 20 kPa at 30 "C) Programmed from 70 "C (hold time 0.5 min) to 150 "C at 30 "C min-l Inductively coupled plasma atomic emission spectrometry- ICP Daini Seikosha R.f.generator Matching box Torch Plasma-Therm (Fassel type) Plasma gas Ar Flow rates Jhhelle-type monochromator Spectrametrics Spectraspan I11 Wavelength 253.7 nm Slit (width x height) Photomultiplier Amplifier 40 MHz 0.5 kW with automatic power controller unit Automatic matching unit with a vacuum-type condenser Outer 18 1 min-I intermediate 8.5 1 min-I sample 0.3 1 min-l Entrance 200 x 500 pm Exit 200 x 500pm Hamamatsu R292 at 950 V Field effect transistor input op-a.mp current-voltage converter (10 V mA-') time constant 1 s ? A With an SDV plasma the highest SIB ( 100) was obtained a t 0.5 kW r.f.power 0.7 1 min-l sample gas flow rate and observation height 17 mm above the load coil. However the highest SIB did not correspond to the highest S/N. and close to the dark current (net background 0.15 nA dark current 0.1 nA). The highest SIN (about 1000) was observed at 0.9 kW r.f. power 0.6 1 min-l sample gas flow rate and observation height 17 mm above the load coil. Under these conditions the SIB was reduced to 25. With the AXV plasma the highest S/B (550) was flow rate (Fig. 2). The mercury emission intensity corre- sponding to the highest S/B condition at the respective r.f. power was not reduced significantly by reducing the r.f. power from 1.1 to 0.5 kW and the optimum conditions for the highest S/N (about 20 000) were almost the same as that 500 - .- 0 c 2 g 300 m * C cx) Under these conditions the background intensity was low c.-0 3 0 x - obtained at 0.5 kW r.f. power and 0.3 1 min-' sample gas z - iij 100 - 1 1 I I I for the highest S/B. The flow rates of the outer and 0.2 0.4 0.6 0.8 1.0 intermediate gas had little influence on the S/B and S/N. Sample gas fIow/I min-' If we compare the signal and background intensities from the AXV- and SDV-ICP the background intensities of the two observation systems were almost the same and the signal intensity of the AXV-ICP was 20 times greater than that of the SDV-ICP at the respective optimum S/N. This improve- ment in S/N was four times greater than the reported values and might be partially owing to the characteristics of the Fig.2 Relative intensity of Hg and SIB ratio by GC-ICP-AES. and Bt o.7 kW; c and c 0.9 kW; and D and D 1.1 kW Reagents power A and A' o.5 kw; All of the reagents were of analytical-reagent grade. A stock solution was prepared by dissolving 37 mg of methyl- mercury chloride (Wako Osaka Japan) in 100 ml of benzene. Working standard solutions were obtained by diluting the stock solution with benzene to the required concentration. Results and Discussion ICP Optimization The optimum conditions were determined by considering the mercury signal-to-background ratio (SIB) mercury emission intensity and mercury signal-to-noise ratio (S/N). The mercury emission intensity was measured by introduc- ing mercury-containing gas. The gas was prepared by passage through a metallic mercury reservoir kept at 40 "C.The exact mercury concentration in the gas was not known but the concentration was kept constant (about 5 ng s-l) during the experiments. Cchelle-type monochromator.-As the vertical slit-height was limited to 500 pm which was the maximum in the monochromator used it was necessary to reduce the ICP image size to one tenth in the SDV arrangement in order to observe an emission zone of a few millimetres. On the other hand the limited slit-height was suitable for the AXV construction because the analyte emission which was restricted in the narrow centre zone of the plasma image could be selected from the surrounding bright torus-shaped b<ackground emission by the narrow slit-height. G.as Chromatograph Optimization The effect of the injection port temperature on methylmer- cury was investigated using GC-ICP-AES over the range 90-200 "C.The injection volume used was 2 pl of 3.18 pg ml-' methylmercury solution in benzene. The peak intensity of the methylmercury increased with increasing irljection port temperature up to 130 "C and remained almost constant in the range 130-150 "C. The peakJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1992 VOL. 7 17 3 0 2 4 Ti me/m i n Fig. 3 Resolution of mercury species of 1 Hg; 2 benzene; 3 CH,HgX; and 4 C2H5HgX by GC-ICP-AES intensity gradually decreased above 160 "C and leading and tailing of the peak appeared. These effects might be due to the decomposition of the analyte. An alternative insert made of PTFE was checked but no significant change was observed.The injection port temperature was therefore set at 140 "C. In order to separate the methylmercury from the solvent (benzene) and to concentrate the methylmercury on the column head an initial temperature of 70 "C and a hold time of 0.5 min were chosen. Column temperature programmes covering the range from 70 "C (hold time 0.5 min) up to 150 "C with ramp rates of 10 20 30 and 40 "C min-l were studied. When the normal capillary column (25-50 m) was used at a linear velocity of 40 cm-I high sensitivity was not obtained because of peak broadening probably owing to the adsorp- tion or decomposition of the methylmercury on the column. In general the peak intensity of the analyte increases gradually with an increase in the carrier gas flow rate and with shortening of the column length.Therefore the use of a short column at a high carrier gas flow rate is useful for the rapid and sensitive determination of mercury compounds. In this study a 3 m column (2.5 m in the GC oven and 0.5 m at the GC-ICP interface) and a 7.5 ml min-l flow rate at 30 "C (head pressure 20 kPa linear velocity 1.3 m s-l) were chosen because of reduced adsorption effects short retention time (2.2 min) and high sensitivity. The theoretical plate value of the column was around 2500. Although these conditions (short column length and high flow rate) are unusual in capillary column GC they might be suitable for labile or thermally unstable compounds such as methylmercury. Separation of Mercury Species Separation was examined by using GC-ICP-AES for mer- cury vapour methylmercury chloride and ethylmercury chloride. Under the GC conditions mentioned above these species were adequately separated. The peak shape for methylmercury chloride was symmetrical and narrow (Fig. 3).However when a standard sample of methylmercury chloride and an environmental sample were injected repeat- edly a problem occurred the single peak of methylmercury was split into three peaks and the ratio of each peak was dependent on the amount of methylmercury chloride 1501 0 0 2 4 6 Retention tirne/min 1000 1 1 x10 150 200 250 300 350 mlz Fig. 4 (a) Chromatogram obtained by total ion monitoring and (6) mass spectra of CH,HgX (X=Cl Br I) obtained by GC-MS. 1 CH3HgCI; 2 CH,HgBr; and 3 CH,HgI injected. When very small amounts of methylmercury chloride (below 100 pg) were injected only one peak appeared.However when amounts of more than 200 pg were injected two or three peaks appeared. In order to identify the compounds giving rise to these peaks 3 ng of methylmercury were injected 20 times and the fractions corresponding to methylmercury peaks were collected in cold benzene (6 "C). The benzene solution was concentrated to a small volume by flushing with nitrogen gas and analysed by GC-MS. A total ion monitoring chromatogram and the mass spectrum of each peak are shown in Fig. 4. The peaks were identified as methylmercury chloride bromide and iodide. The single peak that appeared when small amounts of methylmercury (less than 100 pg) were injected was confirmed as methylmercury iodide.It was noteworthy that methylmercury iodide was identified in spite of injecting an authentic sample of methylmercury chloride. Conversion from chloride into iodide seemed to occur at the injection port and on the column. The stability constants of the methylmercury halides are known to decrease in the order iodide > bromide > c h l ~ r i d e . ~ ~ ~ Therefore the presence of even trace amounts of iodide may have given rise to the appearance of methylmercury iodide. The source of bromine and iodine was not identi- fied. This reaction also occurred when a real (atmospheric) sample was injected. It can be assumed that bromide and iodide in the sample were adsorbed or remained on the injection port or column and caused the formation of the bromide and iodide species.The splitting of the peak resulted in a deterioration of the detection limit and reproducibility. Therefore it was decided to convert all methylmercury species into the iodide. This conversion was completed by treatment with iodine dissolved in benzene. Benzene was shaken with hydriodic acid and several microlitres of the benzene layer were injected. By this treatment all methylmercury species including those which were originally due to the chloride and bromide analogues appeared at the position of the iodide species. Once this treatment had been carried out no peaks due to chlorideJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1992 VOL. 7 18 t - m C a m .- 30 P9 1 rnin H Time - Fig. 5 Chromatogram of CH3HgI at a low concentration level obtained by GC-ICP-AES and bromide species appeared even though more than 200 samples were injected thereafter.Calibration and Detection Limit The calibration graph for methylmercury was linear in the range 0.006-6 ng of mercury (6 ng was the highest concentration examined). Fig. 5 demonstrates the detection of methylmercury at low concentration levels. The detection limit for methylmercury (as Hg) was 3 pg (injection volume 2 pl) at S/N=2. The relative standard deviation of ten replicate measurements was about 5% at the trace level (30 pg as Hg). Selectivity Because of the poor selectivity of the electron-capture detector for organomercury compounds many substances give false selectivities. The ICP detection system provides much better selectivity. At the 253.7 nm atomic emission line the selectivity for mercury when comparing CH3HgX with decane or undecane was well above 1 x 1 06 1.Hence environmental or biological samples can be injected with- out previous clean-up procedures. Because of the high- temperature characteristics of the ICP source atomization in the plasma is considered to be almost complete. For this reason the sensitivity is virtually independent of the chemical form (methyl- or ethylmercury). Atmospheric Samples Three air samples were collected in a residential area in Sapporo in September 1989 according to the method by Bzezinska et al.,*O using a tube (140 x 8 mm i.d.) packed with Tenax GC. The pumping speed was 2 1 min-' and the sampling time was 24 h. After the air sample had been collected the tube was connected to a nitrogen supply then the mercury compounds were thermally eluted into 0.5 ml of cold benzene in a micro-impinger by heating the tube at about 200 "C for 30 min with a nitrogen flow rate of 10 ml min-I.The benzene solution was concentrated to 50 pl by nitrogen gas flushing and then appropriate aliquots (2- 10 pl) of the benzene solution were injected into the GC-ICP- AES system. The concentrations of methylmercury in three air samples were calculated to be 17 28 and 54 pg M - ~ 1 1 I 0 2 4 Tim e/m i n Fig. 6 Chromatogram of an air sample obtained by GC-ICP-AES. Peak 1 benzene; and 2 54 pg m-3 of methylmercury (CH,HgX) respectively. No other alkylmercury compounds were de- tected. A typical chromatogram is shown in Fig. 6. Conclusions A gas chromatograph coupled with an ICP-AES detector was applied to the determination of trace amounts of methylmercury species.Methylmercury species were first converted into the iodide form by treatment with iodine dissolved in benzene. The iodide was readily separated on a short capillary column. The AXV-ICP detector offered advantages in analytical performance over the conventional SDV-ICP in optical configuration. In particular the AXV-ICP permitted a 20- fold more sensitive detection of mercury than the SDV- ICP. The method can be applied to the speciation of mercury at very low concentrations such as alkylmercury compounds in the atmosphere. The authors thank Dr. H. Ito for GC-MS analyses Mr. S. Sakai for air sampling and Dr. J. Edmunds for language correction. References 1 McCormack A. J. Tong S. C. and Cooke W. D. Anal. Chem. 1965,37 1470. 2 Talmi Y. CRC Crit. Rev. Anal. Chem. 1983 14 231. 3 Keliher P. N. Boyko W. J. Clifford R. H. Snyder J. L. and Zhu S. F. Anal. Chem. 1986 58 335R. 4 Broekaert J. A. C. Anal. Chim. Acta 1987 196 1. 5 Demers D. R. Appl. Spectrosc. 1979 33 584. 6 Faires L. M. Bieniewski T. M. Apel C. T. and Niemczyk T. M. Appl. Spectrosc. I 985 39 5. 7 Kawaguchi H. Tanaka T. and Mizuike A. Bunseki Kagaku 1984 33 129. 8 Talmi Y. Anal. Chim. Acta 1975 74 107. 9 Morita H. Sakurai H. and Shimomura S. Bunseki Kagaku 1982 31 314. 10 Bzezinska A. Van Loon J. Williams D. Oguma K. Fuwa K. and Haraguchi H. Spectrochim. Acta Part R 1983 38 1339. Paper 0/05823H Received December 31 I990 Accepted August 22 I991