Anal. Chem. 2003, 75, 98-103
Headspace Liquid-Phase Microextraction of Chlorobenzenes in Soil with Gas Chromatography-Electron Capture Detection Gang Shen and Hian Kee Lee*
Department of Chemistry, National University of Singapore, 3 Science Drive 3, Republic of Singapore 117543
The organic solvent film formed in a microsyringe barrel was used as an extraction interface in headspace liquidphase microextraction (HS-LPME) of chlorobenzenes. Some common organic solvents with different vapor pressures (9.33-12 918.9 Pa) were studied as extractants. The results indicated that even the solvent with the highest vapor pressure (cyclohexane) can be used to carry out the extraction successfully. In general, the reasons for successful extraction are the very small space (5 mm3) within the microsyringe barrel and the fast equilibrium between gaseous analytes and organic solvent film. Both of these factors significantly reduced the risk of solvent loss during extraction. Thus, the choice of extraction solvent for the present method was very flexible. From the viewpoint of extraction efficiency, toluene (which has relatively low vapor pressure) was found to provide the best extraction efficiency. The effects of sampling volume, organic solvent volume, syringe withdrawal rate, and number of extraction cycles were also investigated. The procedure with respect to repeatability and limits of detection was evaluated by soil spiked with chlorobenzenes. Repeatability was between 5.7 and 17.7%, and the limits of detection were 6-14 ng/g. HS-LPME was shown to be an inexpensive, fast, and simple sample preparation method for volatile compounds. Sample preparation is a critical step in an analytical procedure. The main aim of sample preparation is to transfer the analyte into a form that is prepurified, concentrated, and compatible with the analytical system.1 Conventionally, this can be accomplished by liquid-liquid extraction (LLE) and solid-phase extraction (SPE). But as is well known, these methods are time-consuming, tedious, may require too much organic solvent, and can be relatively expensive. Although SPE uses low amounts of organic solvent and can be automated, it is limited to semivolatile or nonvolatile compounds with boiling points higher than the desorption solvent temperature.2 Moreover, LLE and SPE may result in analyte losses, contamination, and generally poorer precision. Therefore, recent work on sample preparation has focused on the development of simpler (preferably one-step), solvent-saving (even solventfree), miniaturized, and automatic or semiautomatic methods. * Corresponding author. E-mail:
[email protected]. (1) Ulrich S. J. Chromatogr., A 2000, 902, 167-194. (2) Masque´, N.; Marce´, R. M.; Borrull, F. Trends Anal. Chem. 1998, 17, 384394.
98 Analytical Chemistry, Vol. 75, No. 1, January 1, 2003
Solid-phase microextraction (SPME), solvent microextraction (SME), and membrane-based liquid microextraction are recent examples of such development. SPME is a solvent-free process developed by Pawliszyn and co-workers3 that includes simultaneous extraction and preconcentration of analytes from aqueous samples. Its principle is based on the partitioning of analytes between sample matrixes and the polymer-coated stationary phase on a silica fiber. At least three modes of SPME have been developed: direct-immersion,4,5 headspace,6,7 and membraneprotected SPME.8 The latter two modes are to eliminate the effect of a dirty matrix on the fiber in the direct-immersion mode. SPME has achieved tremendous success and has been widely used for drugs, food, and environmental pollutants9-12 and is regarded as a rugged, sensitive, and accurate method. The disadvantages are that it is still relatively expensive and the polymer coating is fragile and easily broken. Furthermore, sample carryover is sometimes difficult or impossible to be eliminated.1 In recent years, SME has been shown to be a viable alternative sample preparation method to conventional LLE.13,14 It requires smaller volumes (e.g., 200 µL or less) of organic solvent to extract analytes from moderate amounts of aqueous matrixes. Jeannot and Cantwell15 developed a liquid-liquid microextraction system by which extraction was achieved into a single drop. One disadvantage, however, was that extraction and injection were performed separately in two different devices. Subsequently, liquid-liquid-liquid microextraction (LLLME) based on the principle of supported liquid membrane (SLM) was developed. Porous polypropylene hollow fiber with impregnated organic solvent was used as an interface between acceptor phase and donor.16 The fiber provided reasonable selectivity so that (3) Arthur, C. L.; Pawliszyn, J. Anal. Chem. 1990, 62, 2145-2148. (4) Zambonin, C. G.; Palmisano, F. J. Chromatogr., A 2000, 874, 247-255. (5) Eisert, R.; Levsen, K. J. Am. Soc. Mass Spectrom. 1995, 6, 1119-1130. (6) Lee, X. P.; Kumazawa, T.; Sato, K.; Suzuki, O. J. Chromatogr. Sci. 1997, 35, 302-308. (7) Karaisz, K. G.; Snow, N. H. J. Microcolumn Sep. 2001, 13, 1-7. (8) Zhang, Z.; Poerschmann, J.; Pawliszyn, J. Anal. Commun. 1996, 33, 219221. (9) Jinno, K.; Kawazoe, M.; Saito, Y.; Takeichi, T.; Hayashida, M. Electrophoresis 2001, 22, 3785-3790. (10) Marsili, R T. J. Agric. Food Chem. 2000, 48, 3470-3475. (11) Ramos, E. U.; Meijer, S. N.; Vaes, W. H. J.; Verhaar, H. J. M.; Hermens, J. L. M. Environ. Sci. Technol. 1998, 32, 3430-3435. (12) Yu, X.; Pawliszyn, J. Anal. Chem. 2000, 72, 1788-1792. (13) Cacho, J.; Ferreira, V.; Fernandez, P. Anal. Chim. Acta 1992, 264, 311317. (14) Guidotti, M. J. High Resolut. Chromatogr. 1996, 19, 469-471. (15) Jeannot, M. A.; Cantwell, F. F. Anal. Chem. 1996, 68, 2236-2240. 10.1021/ac020428b CCC: $25.00
© 2003 American Chemical Society Published on Web 11/23/2002
LLLME could be effectively used in the extraction of drugs from plasma. In our laboratory, three new types of microextraction techniques have been developed. He and Lee17 developed static and dynamic liquid-phase microextraction (LPME), in which the extraction was performed into an organic drop of solvent (static LPME) or within the microsyringe barrel (dynamic LPME). After extraction, the extract can be directly injected into a gas chromatograph (GC) for analysis. They are shown to be fast, economical, and simple one-step microextraction techniques. Recently, hollow fiber-protected liquid-phase microextraction was developed.18 A small length of a porous hollow fiber was placed at the syringe tip. The fiber was filled with organic solvent which could also impregnate the pores. Because of the small size of the pores, the detrimental effect of particles and unwanted large molecules in the matrix could be obviated. Therefore, it is an ideal extraction technique for dirty samples, such as soil slurry and biological matrixes. Unlike SPME, it is difficult to use microdrop LPME for headspace extraction because almost all the widely used organic solvents in GC have high vapor pressures, which result in them evaporating too quickly in air. Theis et al.19 reported droplet headspace SME where octanol, which has a very low vapor pressure (9.33 Pa), was employed. Obviously the selection of suitable organic solvents is limited since the vapor pressure of the solvent rather than its extraction efficiency may be a prime consideration. In the present work, we studied a new approach to LPME, that is, headspace LPME (HS-LPME). Chlorobenzenes were used as model compounds. They are hazardous to health and have been listed as priority pollutants by the United States Environmental Protection Agency (USEPA).20 They are introduced into our environmental matrixes, such as water, soil, and sediments, because of their wide usage.21-23 The results indicated that HS-LPME is an efficient extraction technique to analyze chlorobenzenes in complicated matrixes such as soil. It was demonstrated to be extraordinarily inexpensive, fast, and convenient. In comparison to droplet SME, the selection of extractant solvents is more flexible. This work provided an alternative extraction method to solvent microextraction. EXPERIMENTAL SECTION Chemicals. 1,3,5-Trichlorobenzene (TCB), 1,2,3,4-tetrachlorobenzene (1,2,3,4-TeCB), 1,2,4,5-tetrachlorobenzene (1,2,4,5TeCB), pentachlorobenzene (PCB), hexachlorobenzene (HCB), and 1,4-dibromobenzene (used as internal standard) were purchased from Aldrich (Milwaukee, WI). Stock standard solutions were prepared in acetone, with concentration levels of ∼500 µg/ (16) Pedersen-Bjergaard, S.; Rasmussen, K. E. Anal. Chem. 1999, 71, 26502656. (17) He, Y.; Lee, H. K. Anal. Chem. 1997, 69, 4634-4640. (18) Shen, G.; Lee, H. K. Anal. Chem. 2002, 74, 648-654. (19) Theis, A. L.; Waldack, A. J.; Hansen, S. M.; Jeannot, M. A. Anal. Chem. 2001, 73, 5651-5654. (20) Karnofsky, B. Hazardous Waste Management Compliance Handbook, 2nd ed.; Van Nostrand Reinhold: New York, 1997. (21) Meharg, A. A.; Wright, J.; Osborn, D. Sci. Total Environ. 2000, 251/252, 243-253. (22) Llompart, M.; Li, K.; Fingas, M. Talanta 1999, 48, 451-459. (23) Zoumis, T., Schmidt, A.; Grigorova, L.; Calmano, W. Sci. Total Environ. 2001, 266, 195-202.
mL for each compound, and were stored in a freezer at ∼-20 °C. Working solutions were prepared by dilution of stock standards with ultrapure water (NANOpure, Barnstead). These solutions were stored in the dark at 4 °C. The internal standard (at 15 µg/ L) level was prepared in the organic solvent used as acceptor. Toluene and cyclohexane from J.T. Baker (Phillipsburg, NJ) were of HPLC grade. Octane (>99%) was from Fluka (Buchs, Switzerland). Acetone (pesticide grade) was purchased from Fisher Scientific (Fair Lawn, NJ). Chromatographic Analysis. Analysis of chlorobenzenes was performed on a Hewlett-Packard 5890 Series II gas chromatograph equipped with an electron capture detector (GC/ECD). The GC was fitted with ZB-5 capillary column (30 m × 0.32 mm i.d., 0.25µm phase thickness) from Zebron (Torrance, CA). The following temperature program was employed: 85 °C for 2 min; 8 °C/min to 110 °C; then 5 °C/min to 180 °C; finally 25 °C/min to 280 °C, held for 5 min. The injector temperature was 250 °C, and all injections were made in the splitless mode. The detector temperature was set at 260 °C. Preparation of Soil Sample. The chlorobenzene-free soil samples were air-dried, pulverized, and sieved to a grain size of 2 mm. A total of 50 g of soil was mixed with acetone until the soil sample was completely covered by the solvent and formed a slurry. Then, an appropriate volume of the standard solution was spiked into the soaked soil. The bulk of the solvent was slowly evaporated at room temperature by thorough manual shaking. This sample was kept overnight in the fume hood to dry it completely. The prepared soil sample was stored in the refrigerator at 4 °C prior to analysis. RESULTS AND DISCUSSION Headspace Liquid-Phase Microextraction. A conventional 10-µL microsyringe (SGE, Sydney, Australia) designed for GC was adopted. The extraction consists of the following steps: (1) Withdraw 2 µL of organic solvent into the microsyringe. (2) Pass the microsyringe needle through the headspace sample vial septum and keep the needle suspended over the liquid sample. (3) Withdraw 5 µL of gaseous sample at 1.0 µL/s, then depress the plunger back to the original mark immediately, and hold for 5 s. The same process was repeated 25 times. Finally, remove the syringe needle from the vial and inject the chlorobenzenesenriched organic solvent into the GC for analysis. Figure 1 shows graphically the extraction process. When the syringe plunger was withdrawn, a very thin organic solvent film (OSF) was generated on the inner syringe wall as the gaseous sample was drawn in. The chlorobenzenes in the vapor partitioned between the OSF and the gaseous sample. When the syringe plunger was depressed, the chlorobenzenes-enriched OSF was transferred into the bulk organic solvent. This operation could significantly increase the surface area of the interface by using a very small amount of organic solvent, which is helpful to increase the extraction efficiency. Equation 115 describes the relation among extraction speed and volumes of the organic and aqueous solution, where k is rate
k)
(
Ai V0 β0 κ +1 V0 Vaq
)
(1)
constant, V0 and Vaq are the volumes of the organic and aqueous Analytical Chemistry, Vol. 75, No. 1, January 1, 2003
99
Table 1. Physicochemical Properties of Chlorobenzenes and Method Extraction Data compounds
Log KOWa
Hb
Log KOAc
1,3,5-TCB 1,2,3,4-TeCB 1,2,4,5-TeCB PCB HCB
4.19 4.60 4.64 5.17 5.73
2.19 × 10-3 6.90 × 10-4 1.00 × 10-3 7.10 × 10-4 1.70 × 10-3
5.24 6.14 6.02 6.69 6.88
a Partition coefficient of octanol-water. See ref 27. b Henry’s law constant (atm m3/mol). See ref 27. c Partition coefficient of octanolair. Calculated from KOA ) KOW/H′, where H′ is the dimensionless air-water partition coefficient deduced from H/RT (R is a physical constant; T is temperature). See ref 28.
Figure 1. Diagram of headspace liquid-phase microextraction. (1) Before plunger moves up (the start point of one extraction cycle). (2) Sample withdrawal or sample discharge (when the direction is reversed). (3) The pause state after sample discharge (the end point of one extraction cycle).
phases, respectively, Ai is the interface area, β0 is the overall masstransfer coefficient with respect to the organic phase, and κ is the distribution coefficient. It is obvious that fast extraction requires small volumes of organic solvent and aqueous sample and a high surface area. In our previous work,18 we described that the extraction efficiency increased several times (4-5-fold) when the spherical organic droplet was configured into a rodlike shape. In other work on droplet extraction of gas samples,24,25 it was also found that a spherical or ovoid shape was not favorable in terms of extraction speed because of the small surface/volume ratio. Thus, U-shaped wire loops were employed to generate filmlike droplets for fast extraction. In the present work, the rodshaped organic solvent inside the syringe channel was used to form a film on the syringe channel wall. Besides the OSF, extraction may also occur between the gaseous sample and the main organic plug. Since the radius of the syringe barrel was very small, the cross-sectional surface area of the organic solvent profile was negligible compared to that of the OSF (the ratio was >200 in this experiment). Therefore, mass transfer of the chlorobenzenes was assumed to occur only between the gaseous sample and the OSF. The aqueous sample with chlorobenzenes spiked at 10 µg/L was used to optimize the extraction parameters. The sample was preheated at 40 °C for 15 min before extraction. Heating was maintained during the extraction. Selection of Organic Solvent. Careful attention should be paid to the selection of the extraction solvent. First, it should have a high boiling point and low vapor pressure in order to reduce the risk of evaporation. Second, the solvent should have good chromatographic behavior. Third, the partitioning coefficient should be high. Finally, high-purity organic solvents are easily available. According to these considerations, four solvents, toluene, cyclohexane, octane, and octanol, were considered. Among them, octanol is often employed in solvent microextraction.19 However, in the present work, octanol did not appear to be suitable chromatographically. The solvent peak interfered with those of the target compounds. Of the other solvents, although octane had (24) Liu, S.; Dasgupta, P. K. Anal. Chem. 1995, 67, 2042-2049. (25) Liu, S.; Dasgupta, P. K. Anal. Chem. 1995, 67, 4221-4228.
100
Analytical Chemistry, Vol. 75, No. 1, January 1, 2003
much lower vapor pressure and good chromatographic behavior, its extraction efficiency was lower than those of toluene and cyclohexane. This may be due to the lack of compatibility between solvent and analytes (principle of like attracts like). Finally, toluene was employed since it not only provided the highest extraction efficiency but also had lower vapor pressure (3786.3 Pa) than cyclohexane (12 918.9 Pa). Furthermore, droplet solvent microextraction was performed under similar conditions to test their stability in air. It was observed that hexane and cyclohexane (2 µL) evaporated within 1 min when they were exposed to air (in our laboratory that was air-conditioned at ∼25 °C). Toluene was found to be much more resistant and especially so with HS-LPME. This result indicated that the newly developed HS-LPME effectively limited the evaporation of the organic solvent. Therefore, the selection of organic solvent is more flexible for the present approach than microdrop headspace SME. Organic Solvent Volume and Sampling Volume. The sampling volume refers to the volume of the gaseous sample withdrawn into the syringe barrel. As shown in Figure 1, HS-LPME consists of four steps: withdrawal and discharge of the gaseous sample and two pauses between. Because the extraction occurred in the OSF, the key step is the formation of this film, i.e., its thickness and homogenization. The mass transfer between gas and liquid is assumed to be very fast since diffusion coefficients in the gas phase are typically ∼104 times greater than corresponding diffusion coefficients in condensed phases.26 This can also be proven by the diffusion coefficient of octanol-air (see Table 1). The large KOA values indicate that the analytes in the vapor can partition into the organic solvent film rapidly. Therefore, the pause time after withdrawal of the syringe plunger was set to zero, and 5 s after discharge. In the extraction system, there should be two partitioning equilibria, aqueous sample versus headspace, and headspace versus OSF. Therefore,
C0Va ) Ca,eqVa + Chs,eqVhs + Cosf,eqVosf
(2)
where C0 is the original concentration of analytes in the aqueous (26) Cussler, E. L. Diffusion, Mass Transfer in Fluid Systems; Cambridge University Press: Cambridge, 1984; Chapters 4 and 5. (27) Howard, P. H.; Meylan, W. M. Handbook of Physical Properties of Organic Chemicals; Lewis Publishers: Boca Raton, FL, 1997. (28) Boethling, R. S.; Mackay, D. Handbook of Property Estimation Methods for Environmental Chemicals: Environmental and Health Sciences; Lewis Publishers: Boca Raton, FL, 2000.
Figure 2. Effect of sampling volume on extraction. Analytes concentration is 10 µg/L. The solid lines represent experimental results. The dotted lines represent theoretical results. This figure only depicts the results for two chlorobenzenes; the others show similar trends.
sample. Ca,eq, Chs,eq, and Cosf,eq are analytes concentrations in the aqueous sample, headspace, and OSF, respectively. Va, Vhs, and Vosf are their corresponding volumes. In the present case, however, the partitioning between aqueous sample and headspace was negligible according to the following assumptions. This is because the volume of the gaseous sample is substantially smaller compared with that of the headspace in the vial, and the radius of the syringe needle is so small that the extraction system in the syringe barrel can be regarded as an independent one. Moreover, the amount of chlorobenzenes extracted by the OSF is much less than that in the headspace. Therefore, in this case, only the equilibrium inside the syringe barrel needed to be considered. Accordingly
Chs0Vhs ) Chs,eqVhs + Cosf,eqVosf
(3)
Figure 3. Relationship between organic solvent volume and extraction efficiency (peak area ratio of sample to internal standard). Each analyte concentration is 10 µg/L. Sampling volume, 5 µL.
equation, Vosf and Vhs can be calculated (see Figure 2), and Vorg is a constant.
Vosf ) πR2L - Vhs
(7)
Vhs/Vosf ) r2/(R2 - r2)
(8)
Thus,
where R is the radius of the syringe inner barrel and r is the radius of the cylindrical gaseous sample plug (see Figure 1). The results of above equation are constant when the plunger movement behavior is fixed. According to eqs 5 and 7,
Corg ) where Chs0 is the original concentration of the analyte in the headspace of the syringe barrel. For every extraction cycle, fresh gaseous sample is introduced. Therefore, Chs0 can be assumed to be constant.
Cosf,eq ) Kosf-hsChs,eq
(4)
where Kosf-hs is the partitioning coefficient between the OSF and the headspace. Combining eqs 3 and 4 gives
Cosf,eq )
Kosf-hsChs0Vhs Vhs + Kosf-hsVosf
(5)
After one extraction cycle, the concentration in the organic solvent (Corg) can be expressed as
Corg )
Kosf-hsChs0VhsVosf (Vhs + Kosf-hsVosf)Vorg
(6)
where Vorg is the volume of the organic solvent used. In the above
Kosf-hsChs0
Vhs V Kosf-hs + r /(R - r ) org 2
2
2
(9)
The equation shows that the amount of analytes extracted is proportional to the sampling volume (Vhs), and inversely proportional to the solvent volume (Vorg), which can be identified experimentally (Figures 2 and 3). The experimental (solid) lines in Figure 2 indicate that as the sampling volume increases to some point (>5 µL), the Corg would not be proportional to Vhs. It will be lower than the theoretical results (dotted line). This could be caused by the evaporation of the OSF since as sampling volume increased, the plunger movement period would also be longer. Moreover, although Corg increased linearly in the sampling volume of 2-5 µL, the intercepts were not zero. There are two possible reasons for the phenomenon. One is that when the sampling volume was
