Anal. Chem. 1997, 69, 844-847
Pervaporation as an Alternative to Headspace D. W. Bryce, A. Izquierdo, and M. D. Luque de Castro*
Department of Analytical Chemistry, Faculty of Sciences, University of Co´ rdoba, E-14004 Co´ rdoba, Spain
The use of pervaporation as an alternative to headspace is proposed. The analytical system involves the speciation of organomercury compounds in solid samples using pervaporation, which has been coupled for the first time to gas chromatography. The speciation of mercury as Me2Hg, Et2Hg, and MeHgCl has been carried out without any derivatization of the analytes, which, after separation from the solid matrix, are preconcentrated on a Tenax minicolumn prior to desorption and chromatographic separation on a semicapillary column (HP-1) prior to atomic fluorescence detection. No column degradation was observed. Linear ranges and detection limits slightly better than those obtained by headspace GC were observed for mercury species in solid samples. Excellent recoveries (between 95 and 107%) for mercury species added to complex solid samples were obtained by this extremely simple and easily automated setup. Although the concentration of mercury in the environment is generally considered to be low, methylation of inorganic mercury occurs, forming predominantly methylmercury,1-3 while biogenic formation of volatile dimethylmercury, methylethylmercury, elemental mercury4,5 and ethylmercury chloride6,7 can also take place. As a result, these highly toxic organomercury compounds may be found in sediments, water, and fish due to long-distance transport in the atmosphere.8 The most common methods for the determination of organomercury species are GC-ECD9-11 and GC-MIP,12-14 with either packed or capillary columns. Several authors have reported problems with the chromatographic performance of organomercury species due to the polar nature of the Hg-Cl bond, resulting in peak tailing and degradation of the species due to thermal instability on the column, as well as column degradation.15,16 (1) Craig, P. J. Organometallic Compounds in the Environment; Longman: London, 1986. (2) Jensen, S.; Jernelo ¨v, A. Nature (London) 1969, 223, 753. (3) Kudo, A.; Mortimer, D. C. Environ. Pollut. 1979, 19, 293. (4) Ebinhaus, R.; Wilken, R. D. Appl. Organomet. Chem. 1993, 7, 127. (5) Aymot, M.; Mierle, G.; Lean, D. R. S.; McQueen, D. J. Environ. Sci. Technol. 1994, 28, 2366. (6) Alli, A.; Jaffe, R.; Jones, R. J. J. High Resolut. Chromatogr. 1994, 17, 745. (7) Horvart, M.; Mandic´, V.; Liang, L.; Bloom, N. S.; Padberg, S.; Lee, Y. H.; Hintlemann, H.; Benoit, J. Appl. Organomet. Chem. 1994, 8, 553. (8) Slmer, F.; Schuster, G.; Seiler, W. J. Atmos. Chem. 1985, 57, 2638. (9) Mahan, K.; Mahan, S. Anal. Chem. 1977, 49, 662. (10) Dogan, S.; Haerdi, W. Anal. Chim. Acta 1978, 101, 433. (11) Heiden, R.; Aikens, D. Anal. Chem. 1979, 51, 151. (12) Bache, C.; Lisk, D. Anal. Chem. 1971, 43, 950. (13) Talmi, Y. Anal. Chim. Acta 1975, 74, 107. (14) Decadt, G.; Baeyens, W.; Bradley, D.; Goeyens, L. Anal. Chem. 1985, 57, 2788. (15) Horvat, M.; Byrne, A. R.; May, K. Talanta 1990, 37, 207. (16) Mena, M. L.; McLeod, C. W.; Jones, P.; Withers, A.; Minganti, V.; Capelli, R.; Quevauviller, Ph. Fresenius’ J. Anal. Chem. 1995, 351, 456.
844 Analytical Chemistry, Vol. 69, No. 5, March 1, 1997
Headspace GC has been used on only a handfull of occasions in an attempt to avoid column degradation, as only the volatile mercury species are injected onto the column.17,18 As with most methods of analysis for mercury species in solid samples, a derivatization step is necessary in order to liberate the mercury from the sample, followed by the common derivatization of the methyl- or ethylmercury chloride to dialkyl derivatives.17,18 Due to the low levels of the species involved, a preconcentration step is also required, normally on a resin containing dithiocarbamate groups,19 sulfhydryl groups,20 or dimethylsiloxane.21 Unfortunately, any derivatizing reactions prevent the speciation of mercury from being carried out, as there can be no discrimination between the di- and monoalkyl species originally present in the sample. Headspace is also difficult to automate and is a fairly complicated technique to carry out. In the last couple of years, our group has been applying pervaporation in conjunction with flow injection analysis (FI) for the separation of volatile analytes or reaction products from complex liquid and solid samples.22-25 In this technique, the sample is placed in the donor chamber of a laboratory-designed and -built pervaporation cell, which is then heated, either conventionally (in a water bath) or by microwaves. The volatile analytes or reaction products then evaporate to a gaseous gap above the sample, from where they diffuse through a membrane to a static or dynamic acceptor stream, where they are taken to the detector or an FI manifold for further derivatization prior to detection. The main advantage of pervaporation is that the sample never comes into contact with the membrane, and so the membrane pores never become blocked. In the present research, we propose the use of pervaporation coupled with semicapillary gas chromatography and atomic fluorescence detection. The mercury species in the solid samples pervaporate through the membrane, where they are collected in a stream of argon and taken to a minicolumn containing Tenax T.A. in order to preconcentrate the species. Once the species are desorbed from the Tenax minicolumnn, a GC method based on one developed by Alli et al.6 and later commercialized by Stockwell26 is used to separate and detect the organomercury species. This method, which does not show the frequently reported chromatographic problems for MeHgCl (peak tailing, column degradation), uses atomic fluorescence detection, which (17) Lansens, P.; Baeyens, W. Anal. Chim. Acta 1990, 228, 93. (18) Lansens, P.; Meuleman, C.; Baeyens, W. Anal. Chim. Acta 1990, 229, 281. (19) Emteborg, H.; Baxter, D. C.; Frech, W. Analyst 1993, 118, 1007 (20) Lee, Y. H.; Mowner, J. Anal. Chim. Acta 1989, 221, 259. (21) Cai, Y.; Bayona, J. M. J. Chromatogr. 1995, 696, 113 (22) Mattos, I. L.; Luque de Castro, M. D. Anal. Chim. Acta 1994, 298, 159. (23) Mattos, I. L.; Luque de Castro, M. D.; Valca´rcel, M. Talanta 1995, 42, 755. (24) Papaefstathoiu, I.; Luque de Castro, M. D. Anal. Chem. 1995, 67, 3916. (25) Papaefstathoiu, I.; Luque de Castro, M. D.; Valca´rcel, M. Fresenius’ J. Anal. Chem. 1996, 354, 442. (26) Stockwell, P. Personal communication. S0003-2700(96)00456-8 CCC: $14.00
© 1997 American Chemical Society
Figure 1. System for the speciation of mercury in solid samples using pervaporation/gas chromatography/atomic fluorescence detection. PM, pervaporation module; IV, injection valve; Tenax, Tenax minicolumn; I, injector; GC, gas chromatograph; P, pyrolysis unit; D, detector; He, helium; Ar, argon. The sections marked 1, 2, 3, and 4 correspond to the pervaporation, retention/desorption, chromatographic separation, and detection steps, respectively.
is surprisingly uncommon in the gas chromatographic separation of mercury species. Pervaporation has the advantages of being easily adapted for use in laboratories,27 and does not involve any derivatization of the mercury species, allowing speciation of Me2Hg, Et2Hg, and MeHgCl to be carried out. The fact that only the volatile mercury species are injected onto the column means that no degradation is observed. This is the first time that pervaporation, which appears to be an interesting technique for the separation of analyes from complex samples,24,28,29 has been used in conjunction with an adsorbent such as Tenax to preconcentrate the analytes, and this is the first time it has been coupled to GC. EXPERIMENTAL SECTION Apparatus. A Varian Star 3400X gas chromatograph connected to a pyrolysis unit and a Merlin atomic fluorescence detector (both P.S. Analytical, London, England) were used for the separation/detection of the organomercury compounds. A 15 m × 0.53 mm i.d. fused silica semicapillary column coated with a 2.6 µm film thickness of methylsilicone H.P.-1 (Hewlett Packard) was used to separate the compounds. The injector temperature was maintained at 50 °C. The temperature of the chromatographic oven was programmed at 40 °C (1 min), 20 °C/min to 175 °C, and held at 175 °C for 1 min. He (C-50, Carburos Meta´licos, Barcelona, Spain) at 40 mL/min was used as the carrier gas. A computer with Varian Star Chromatography Workstation software for data collection and treatment and muffle oven were also used. To separate the analytes from the solid matrix, a laboratorydesigned and -built pervaporation unit22-25,28 consisting of two chambers (donor and acceptor) with an inlet and outlet channel in the acceptor chamber, was used. The volume of both the donor and acceptor chambers could be altered by addition of spacers. The unit, made of methacrylate, was held together by means of four stainless steel screws and two aluminum supports. A Rheodyne 5041 injection valve, an adjustable gas flowmeter (P.S. (27) Prinzing, U.; Ogbomo, I.; Schmidt, H. L. Sens. Actuators 1990, B1, 542.29. (28) Papaefstathoiu, I.; Tena, M. T.; Luque de Castro, M. D. Anal. Chim. Acta 1995, 308, 246. (29) Bryce, D. W.; Izquierdo, A.; Luque de Castro, M. D. Anal. Chim. Acta, in press.
Analytical), PTFE tubing (0.5 mm i.d.), and a semicapillary fused silica column (0.53 mm i.d.) were used to connect the pervaporation unit to the gas chromatograph and control the flow rate through the pervaporation system. Reagents. Standard solutions of mercury compounds were prepared in dichloromethane (Romil Chemicals, England). Helium and argon (both C-50, Carburos Meta´licos) were used as the carrier and makeup gases, respectively. Methylmercury chloride was provided by P.S. Analytical. Dimethylmercury and diethylmercury were obtained from Strem Chemicals Ltd. (Bischheim, France). Tenax T.A. (60-80 mesh) from Phase Sep (Queensferry, U.K.) was used to retain the analytes. Procedure. The proposed method can be considered as four steps, as detailed below, which are shown in Figure 1. (1) Pervaporation of Organomercury Species. A 0.5 g portion of sample was placed in the donor chamber of the pervaporation cell. The cell was closed, connected to the system, and placed in the water bath at 95 °C for 15 min while a flow of argon passed through the upper acceptor chamber of the module. As the mercury compounds evaporated into the gas layer above the sample and then through the membrane, they were collected in the argon acceptor stream. (2) Preconcentration/Desorption of the Organomercury Species. A minicolumn of Tenax, positioned in the loop of an injection valve, placed in an ice bath, was used to preconcentrate the mercury species that had passed through the membrane of the pervaporation unit. To desorb the retained species and inject them onto the chromatographic column, the flow of Ar from the pervaporation unit was diverted so that the loop of the valve was shut off with no gas flow through it. The Tenax minicolumn was taken out of the ice bath and placed in a muffle oven at 350 °C for 30 s, after which the valve was switched to the “inject” position, allowing a flow of He to pass through the column and flush the desorbed species from the loop of the injection valve, through the connection between the injection valve and the chromatograph, and onto the chromatographic column. (3) Chromatographic Separation of the Organomercury Species. The He stream containing the desorbed organomercury compounds entered the injector in a continuous manner via a Analytical Chemistry, Vol. 69, No. 5, March 1, 1997
845
hypodermic needle fixed in the injector inlet, with a helium flow of 20 mL/min. This then mixed with an additional flow of He, also at 20 mL/min, which flowed through the normal carrier gas inlet to the injector. The chromatograph oven was programmed at 40 °C (1 min), 20 °C/min to 175 °C, and held at 175 °C for 1 min. This provided good separation of the three compounds and was quicker than the method described by Alli et al.6 (10 vs 18 min). (4) Detection of the Separated Species. As the individual species emerged fom the column, they passed through a pyrolysis unit set at 800 °C in order to break the compounds down to elemental mercury. This stream was then mixed with a makeup gas (Ar, 70 mL/min) before continuing to the detector. A second flow stream of at 250 mL/min was used as sheath gas, which resulted in greater reproducibility29 by maintaining the He/Ar/ Hg flow in the beam of the detector. RESULTS AND DISCUSSION Optimization. The variables of the system that had to be optimized were split into four areas: first, those involved in the chromatographic separation; second, those involved in detection; then the retention/desorption variables; and finally those involved in the pervaporation step. The variables were all studied using the univariate method. A 15 m × 0.53 mm i.d. fused silica semicapillary column coated with a 2.6 µm film thickness of methylsilicone HP-1 was used to separate the compounds. The injector temperature was maintained at 50 °C throught. As the helium flow rate was increased from 5 to 40 mL/min, better peak height and definition for all three species, along with greater sampling frequencies, was found; however, above 40 mL/min, no real improvement was observed, and slight peak overlap for the Et2Hg and MeHgCl peaks occurred. Thus, 40 mL/min was chosen as optimum. Similarly, for the ramp rate, increased peak height and reduced retention times were found upon increasing the ramp rate from 5 to 20 °C/min. Above 20 °C/min, no improvement was observed, and so 20 °C/min was used for all further experiments. The detection step had been studied previously by the authors29 in a different setup, and so very little optimization was necessary. Only the flow rate of the makeup gas (argon) had to be studied, showing that 70 mL/min was optimum. A literature search showed that both Chromosorb and Tenax were candidates for the retention of organomercury species from gases. Tenax was chosen as it appeared to provide quantitative desorption of the analytes with no peak tailing. Once the Tenax had been activated (280 °C, 2 h, 20 mL/min He), a minicolumn was prepared in Teflon tubing and placed in the loop of the injection valve in such a way that, in the “load” position, the stream of argon from the acceptor chamber of the pervaporation cell passed through the Tenax and then to waste in a clockwise direction, while the He carrier stream went directly to the chromatographic column. During the desorption step, the flow of argon through the Tenax was stopped by closing off both the entrance and exit to the loop of the injection valve, thus preventing any loss of desorbed analytes prior to injection. When the valve was switched to the “inject” position, the He carrier stream was diverted through the Tenax minicolumn in a counterclockwise direction before continuing to the chromatographic column, thus allowing retention and desorption to be carried out in opposite directions. When the retention temperature was studied, it became apparent that, by keeping the Tenax cool, the retention was greatly 846 Analytical Chemistry, Vol. 69, No. 5, March 1, 1997
Table 1. Optimization of Variables variable
range studied
optimum value
Pervaporation no. of spacers temperature (°C) time (min)
0-2 30-95 0-20
Retention/Desorption of Analytes retention temperature (°C) 0-25 desorption temperature (°C) 100-400 desorption time (s) 0-120 Injection time (s) 0-60
0 95 15 0 350 30 10
Chromatographic Separation ramp rate (°C/min) 5-40 He flow rate (mL/min) 5-50
20 40
Detection makeup gas flow rate (mL/min) 0-150 sheath gas flow rate (mL/min) 0-400
70 250
improved. Thus, 0 °C (i.e., in an ice bath) was chosen as optimum. At first, microwave-assisted desorption was considered; however, this gave poor recoveries for MeHgCl and no recoveries whatsoever for Me2Hg and Et2Hg, due to the fact that they are nonpolar. As a result, a conventional heat source had to be used to study the effect of the desorption temperature. The Tenax minicolumn was placed for a preset time in a muffle oven set at temperatures between 100 and 400 °C prior to switching the injection valve to inject the desorbed species onto the chromatographic column. As the temperature was increased, better recoveries were observed for all three species: 100% recoveries were obtained at 250 °C for both dialkyl species, whereas 350 °C was necessary for complete recovery of MeHgCl, which was confirmed by repeating the desorption process to make sure that the second run provided a blank signal. At 400 °C, a certain amount of breakdown of MeHgCl to elemental mercury was observed with the formation of an additional peak at t ) 0.2 min. As a result, 350 °C was chosen for all further work. When the time of desorption was studied, it was again apparent that the nonpolar dialkyl species were much easier to desorb than MeHgCl; Me2Hg and Et2Hg could both be completely desorbed in 15 s at 350 °C, whereas 30 s was necessary for MeHgCl. Times in excess of 30 s did not yield better results, but only the breakdown of MeHgCl to elemental mercury, as had been seen before. Therefore, 30 s was chosen as optimum. A brief study of the time of injection showed that 10 s was sufficient to allow all the desorbed species to be flushed from the loop to the chromatographic column. When the temperature of the pervaporation was studied, it was not surprising to find that, as the temperature was increased, so was the efficiency of the pervaporation, as has been seen in other papers on pervaporation.22-25,28,29 The pervaporation cell was placed in a water bath, which was filled with water to the level of the lower donor chamber, and the temperature was studied from 30 to 95 °C. Here, 95 °C was chosen as optimum, as it was the highest temperature that could be maintained with the water bath. The final variable to be studied was the time of the pervaporation process. Again, not surprisingly, longer times were found to yield better results, due to the dynamic character of the acceptor gas, which means that analyte-free gas is continually being placed in contact with the membrane, permitting an efficient and continuous displacement of the analytes from the air gap above the sample and diffusion through the membrane. The argon was allowed to
Table 2. Characteristics of the Method analyte
retention time (min)
linear range (ng/g)a
equation
r2
LOD (ng/g)
%RSDb
Me2Hg Et2Hg MeHgCl
0.6 2.9 3.4
0.595-5.95 1.61-50.52 518-25910
y ) 102039x - 27240 y ) 10879x + 1955 y ) 15.16x - 109.2
0.995 0.994 0.994
0.25 0.41 255
5.9 (1.12) 7.6 (3.22) 6.4 (1296)
a
All concentrations as Hg. b n ) 10 at the concentration (ng/g) given in parentheses.
Table 3. Applications of the Method sample sandy soil sewadge sludge CRM a
analytes
amount added (ng/g)a (n ) 3)
recovery (%) ( RSD (n ) 3)
Me2Hg Et2Hg Me2Hg Et2Hg
0.562 1.613 0.468 1.344
100.1 ( 7.2 107.8 ( 1.3 106.2 ( 1.7 94.7 ( 6.0
All concentrations as Hg.
pass through the acceptor chamber continuously before passing to the Tenax minicolumn for times ranging from 2 to 30 min, while the pervaporation cell was placed in the water bath at 95 °C. As the times were increased from 2 to 15 min, the recoveries improved dramatically; however, for 20-30 min, no real improvement was observed, and so 15 min was chosen as optimum in terms of recovery and analysis time. A summary of the optimization process, showing all the variables, ranges studied, and the optimum values found, is shown in Table 1. Characteristics of the Method. Once the optimum values for all the variables had been established, the characteristics of the method such as calibration curve, precision, and limit of detection were studied for the three species. Samples of 0.5 g of diatomaceous earth, spiked with various concentrations of Me2Hg, Et2Hg, and MeHgCl, were prepared in triplicate and analyzed as described in the Procedure section. The results showed that the method provided good linearity for the three species (r2 ) 0.995, 0.994, and 0.994 for Me2Hg, Et2Hg, and MeHgCl, respectively). The precision, expressed as the %RSD of 10 replicates, was 5.9, 7.6, and 6.4 for Me2Hg, Et2Hg, and MeHgCl. A summary of all the results obtained is shown in Table 2. The lower sensitivity observed for the monomethyl mercury is probably due to its polarity, making it harder to separate from the sample matrix. Applications of the Method. The method was validated by studying recoveries from spiked samples. Two sample matrices
were chosen as representative of environmental samples: one a sandy soil, the other a sewadge sludge certified reference material (with no certified organomercury concentration). Recoveries of Me2Hg and Et2Hg at different concentrations for each of the two different samples, each carried out in triplicate, gave excellent results, confirming the possibilities for pervaporation as an easily automated alternative to headspace analysis. The recoveries are shown in Table 3. CONCLUSIONS A novel method for the separation of volatile organomercury species from solid samples prior to preconcentration and separation by GC/AFS has been developed and is proposed as an altenative to headspace GC. The continuous removal of the volatilized analytes through the membrane from the air gap above the sample provides displacement of the mass-transfer equilibrium and results in a higher efficiency of the separation process. Pervaporation has been used for the separation of the analytes from the solids, coupled for the first time to gas chromatography and atomic fluorescence detection for the speciation of Me2Hg, Et2Hg, and MeHgCl in solid samples without any sample pretreatment or derivatization. The process, which is easily automated and simple to use, can be applied to the separation and speciation of any volatile analytes or reaction products. ACKNOWLEDGMENT The authors would like to express their gratitude to P. S. Analytical for the loan of the pyrolysis unit and flowmeter and for providing MeHgCl, and CICyT for financial support (Grant No. PB93-0827). Received for review May 9, 1996. Accepted September 1, 1996.X AC960456S X
Abstract published in Advance ACS Abstracts, February 1, 1997.
Analytical Chemistry, Vol. 69, No. 5, March 1, 1997
847
