Anal. Chem. 2001, 73, 1646-1649
Limitations to the Use of Solid-Phase Microextraction for Quantitation of Mixtures of Volatile Organic Sulfur Compounds R. A. Murray
Horticulture Research InternationalsEast Malling, West Malling, Kent ME19 6BJ, U.K.
A study of the range of volatile organic sulfur compounds produced by brassica plants has highlighted limitations to the use of Carboxen/PDMS fibers for their analysis by solid-phase microextraction (SPME). These fibers are sometimes advocated for the analysis of sulfur gases, but a quantitative comparison of analytical data derived by SPME and by direct gas sampling of standard mixtures of volatile low molecular weight sulfur compounds at 0.01-10 mg/L has identified potential errors associated with their use. Higher molecular compounds displace lower molecular weight compounds as a consequence of competition for active sites on the fiber, and the relative proportions of the components adsorbed onto the fiber depend on their ratio in the headspace. As their relative concentrations change from sample to sample, the varying interactions result in irregular analytical responses, reflected in erratic calibration curves. Standards containing single components are not valid; only a standard containing all components found in the sample to be analyzed, and at the same relative concentrations, is appropriate. In practice, this may preclude the use of the fibers for quantitative analysis of multicomponent mixtures. Since its introduction to the scientific community in 1993, solidphase microextraction (SPME) has been adopted enthusiastically for a wide range of air and water monitoring applications. The technique offers a simple, quick alternative to the traditional methods of purge-and-trap, liquid-liquid extraction and other sample handling techniques, and is reported to give linear response over a wide range of analyte concentrations.1 There are other equally important factors, however, that must be recognized and considered when SPME is used for the analysis of mixtures of volatile compounds. For instance, as SPME fibers are not uniformly sensitive to all compounds,2 relative GC peak areas for an SPME sample do not properly reflect the true proportions of the components in the headspace. Similarly, the adsorption selectivity of the fiber and its discrimination between components in brassica headspace samples must be taken into account for quantitative use.3 The same limitation was recognized when (1) Supelco Bulletin 923, 1998. (2) Bartelt, R. J. Anal. Chem. 1997, 69, 364-372. (3) Ulrich, D.; Krumbein, A.; Schonhof, I.; Hoberg, E. Nahrung 1998, 42, 392394.
1646 Analytical Chemistry, Vol. 73, No. 7, April 1, 2001
quantifying volatile sulfur compounds produced by truffles,4 while SPME and direct gas sampling indicated significantly different ratios of the components in a mixture of 10 compounds frequently encountered in chemical ecology.5 A poly(dimethylsiloxane) (PDMS)-coated fiber was used in all these studies, which extracts volatile compounds by a noncompetitive absorption mechanism and is unlikely to be affected by sample composition. However, a theory of analyte extraction by competitive adsorption onto the limited number of active sites on selected porous polymer SPME predicted that coextractives can reduce the amount of an analyte extracted from a mixture.6 The authors concluded that great care should be exercised when performing quantitative analysis with porous polymer SPME fibers. The model is not applicable to fibers coated with Carboxen (activated carbon)/PDMS in which capillary condensation can occur, but as the general description of the extraction process is similar to that for porous polymer coatings, similar reservations are likely to be applicable. It has been suggested, however, that the unique pore structure of Carboxen1006 enables extraction of all analytes without displacement of the lighter analytes,7 and the corresponding fibers are recommended for the analysis of low molecular weight sulfur compounds. I have recently undertaken a study of the production from brassicas of volatile organic sulfur (VOS) compounds to include their quantitation in headspace samples. A review of the literature revealed that SPME is often the method of choice for this class of compounds, particularly to gain the benefit of its simplicity and sensitivity. In particular, the technique has been applied to the detection and quantitation of VOS compounds in air8 or evolved from samples as disparate as fungi,4 brassica tissue,9-11 beer,12,13 and wine.14-16 However, these various studies, some of which used Carboxen/PDMS fibers, were designed to measure a single (4) Pelusio, F.; Nilsson, T.; Montanarella, L.; Tilio, R.; Larsen, B.; Fachetti, S.; Madsen, J. J. Agric. Food Chem. 1995, 43, 2138-2143. (5) Agelopolous, N. G.; Pickett, J. A. J. Chem. Ecol. 1998, 24, 1161-1172. (6) Gorecki, T.; Yu, X.; Pawliszyn, J. Analyst 1999, 124, 643-649. (7) Supelco Application note 141, 1998. (8) Haberhauer-Troyer, C.; Rosenberg, E.; Grassbauer, M. J. Chromatogr., A 1999, 848, 305-315. (9) Vaughn, S. F.; Boydston, R. A. J. Chem. Ecol. 1997, 23, 2107-2116. (10) Bending, G. D.; Lincoln, S. D. Soil Biol. Biochem. 1999, 31, 695-703. (11) Charron, C. S.; Sams, C. E. J. Am. Soc. Hortic. Sci. 1999, 124, 462-467. (12) Scarlata, C. J.; Ebeler, S. E. J. Agric. Food Chem. 1999, 47, 2505-2508. (13) Hill, P. G.; Smith, R. M.. J. Chromatogr., A 2000, 872, 203-213. (14) Gandini, N.; Riguzzi. R. J. Agric. Food Chem. 1997, 45, 3092-3094. (15) Mestres, M.; Sala, C.; Marti, M.; Busto, O.; Guasch, J. J. Chromatogr, A 1999, 835,137-144. 10.1021/ac001176m CCC: $20.00
© 2001 American Chemical Society Published on Web 02/22/2001
volatile component in an otherwise relatively consistent substrate or did not consider possible interactions between multiple components and the effect on quantitation. To permit valid quantitation of the main VOS components anticipated to be present in the headspace of the brassica samples in this study, standard gas samples with a range of defined concentrations of the various components were prepared and samples were taken for GC analysis by SPME and by direct gas withdrawal with a syringe. Low molecular weight VOS compounds are known to be particularly susceptible to adsorption onto metal and glass surfaces, and special precautions must be taken to prepare and maintain accurate standards. To minimize such problems, gas standards were prepared in proprietary gas sampling bags fabricated from poly(vinyl fluoride), a chemically inert fluorocarbon polymer recommended for sampling of VOS compounds. EXPERIMENTAL SECTION Methanethiol (MeSH), dimethyl sulfide (Me2S), carbon disulfide (CS2), dimethyl disulfide (Me2S2) and 2-propenyl isothiocyanate (PITC) were obtained from Aldrich Chemical Co., Dorset, U.K. A Hewlett-Packard 5890 Series II GC equipped with electronic pressure control, a CP-Sil 5 CB column (30 m × 0.25 mm i.d., 1.0-µm film thickness, Chrompack), split/splitless injector fitted with a 0.75-mm-i.d. inlet liner, flame ionization detector (FID), and Hewlett-Packard 3396A integrator was used for all analyses. The hydrogen carrier gas was maintained at a constant flow of 1.29 mL/min, corresponding to an initial linear velocity of 35 cm/s at 30 °C. The initial oven temperature was held at 30 °C for 3 min, increased to 200 °C at 20 °C/min, and held at 200 °C for 5 min. The injector and detector temperatures were 240 and 275 °C, respectively. Preparation and Analysis of Standard Gas Samples. Standard mixtures of the VOS compounds at 1.0 and 10 mg/l were prepared by injecting liquid reagents (dispensed from a 10- or 100-µL microsyringe) or gaseous methanethiol (dispensed from a 5-mL gastight syringe) directly into a Tedlar gas sampling bag (Aldrich) fitted with a Teflon-lined silicone rubber septum and prefilled with 3 L of dry N2 to minimize hydrolysis and oxidation of the standards (particularly MeSH, which is oxidized rapidly to Me2S2 in air). Additional standard mixtures at 0.01 and 0.1 mg/L were prepared by serial dilution of appropriate higher concentration standards. After all components of a mixture had evaporated (typically 30 min), a 10-µL gastight syringe was equilibrated with the standard gas mixture by 20 repetitive fill-and-empty cycles prior to collection of a 10-µL gas sample, followed by immediate injection in splitless mode onto the GC column. A further sample was collected 30 min later by exposing a 75-µm Carboxen/PDMS SPME fiber (Supelco, Dorset, U.K.) to the standard gas mixture for 15 min. The fiber was immediately inserted into the GC injector, set for split injection (60:1 split ratio), and adsorbed volatile compounds were desorbed for 5 min. A preliminary comparison of split and splitless injections showed no discrimination between the components of the sample when operating in split mode. Decay of VOS Compounds with Time. The 1.0 mg/L gas standard was stored at 20 °C, and its contents were analyzed by both sampling methods after a further 1, 3, 6, and 9 days. (16) Mestres, M.; Marti, M.; Busto, O.; Guasch, J. J. Chromatogr., A 1999, 849, 293-297.
Figure 1. (a) GC detector response from 10-µL gas samples collected from standard mixtures containing all components at equal concentration within any one mixture. (b) GC detector response from a 15-min Carboxen/PDMS SPME sample. Samples were collected from standard mixtures containing all components at equal concentration within any one mixture.
Analysis of Standard Gas Samples of Different Composition. A series of gas standards was prepared by sequential addition of the five VOS components to a single sampling bag. The five components, each at 1.0 mg/L, were added in the order MeSH, Me2S, Me2S2, CS2, and PITC, and the gas mixture was analyzed after each addition by both sampling methods. Influence of SPME Sampling Time on Selectivity. A gas standard containing MeSH, Me2S, Me2S2, CS2, and PITC at 1.0 mg/L was sampled and analyzed by SPME for periods of 0.25, 0.5, 1, 2, 5, 10, and 15 min. RESULTS AND DISCUSSION Comparison of Direct Sampled and SPME Sampled Standard Gas Mixtures. To accommodate the presentation of data reflecting large differences in the values observed for detector response to the components of the gas mixtures, particularly when sampled by SPME, the responses for each component were normalized by comparison with the value at 1.0 mg/L (Figure 1), at 0 days (Figure 2), or after 15 min (Figure 3). The five components of the VOS mixture were well resolved, with retention times of 3.05, 4.54, 4.91, 7.95, and 9.31 min for MeSH, Me2S, CS2, Me2S2, and PITC, respectively. The response factor (peak area/ Analytical Chemistry, Vol. 73, No. 7, April 1, 2001
1647
Figure 3. Influence of sampling time on selectivity of adsorption by a Carboxen/PDMS SPME fiber. Samples were collected from a standard mixture containing all components at a concentration of 1.0 mg/L. Table 1. Analysis of Mixtures of Volatile Organic Sulfur (VOS) Compounds Following Sequential Addition of Components at 1.0 mg/La VOS MeSH Me2S Me2S2 CS2 PITC
integrator peak area 11 553
11 608 20 738
11 834 20 304 14 303
11 377 19 791 13 731 trace
11 307 20 013 13 745 ndb 14 105
a Each sample comprised a 10 µL gas sample injected in splitless mode, equivalent to 10 ng of each component. b ND, not detected,
Figure 2. (a) Decay with time of VOS compounds in a mixture containing all components at an initial concentration of 1.0 mg/L and measured in a 10-µL gas sample. (b) GC detector response for samples collected on a Carboxen/PDMS SPME fiber to monitor the decay with time of VOS compounds in a mixture containing all components at an initial concentration of 1.0 mg/L.
ng carbon) for each component was 4523, 5128, 32, 5600, and 2938, respectively. The very poor response of an FID detector to CS2 is well-known and precluded detection or quantitation of this component in most samples analyzed in this study. Direct gas sample analysis of standard mixtures containing all components at equal concentrations within any one mixture, and ranging from 0.01 to 10 mg/L, showed detector response (relative to that for 1 mg/L) closely proportional to the intended concentration (Figure 1a). Conversely, SPME gave erratic and distorted responses for the lower molecular weight components in the same mixtures (Figure 1b). Additionally, the slope of the response curves was small (∼0.0013 for PITC), demonstrating that SPME exhibited poor discrimination between samples of similar concentration. A comparison of the magnitude of the detector signal following direct gas sampling and SPME sampling of the 1.0 mg/L gas standard demonstrated that SPME increased the effective sample size by a factor of ∼1270 and ∼175 for PITC and MeSH, respectively. Panels a and b of Figure 2 compare the quantity of each component taken from a stored mixed gas sample by direct gas sampling or trapped by SPME and subsequently analyzed by GC. Direct gas sampling demonstrated that the concentration of all 1648 Analytical Chemistry, Vol. 73, No. 7, April 1, 2001
components in a standard mixture stored at 25 °C declined progressively, but at different rates, over a 9-day period. The decline displayed by Me2S after 3 days followed by a slight increase in concentration is presumably a consequence of production of this compound by decomposition of other components of the mixture. Conversely, following SPME of the same gas mixture only PITC showed a similar (but much smaller) decline in GC response: MeSH and Me2S showed a decline after 3 days, similar to that observed for the direct gas sample, while Me2S2 showed a continuously increasing amount detected by GC. The results reported in Table 1 demonstrate the influence of the presence of additional components on the quantitative measurement of each analyte in a mixture of VOS compounds. As expected, the concentration of each component in a 10-µL gas sample taken from the standard mixture was constant and independent of the presence of any other component. In stark contrast, the results reported in Table 2 demonstrate the profound interaction between the components when the same mixture was sampled for 15 min with a Carboxen/PDMS SPME fiber. The quantity of MeSH trapped by the fiber diminished following the consecutive addition of each component and was eventually reduced to 2.3% of that trapped when the compound was the sole component at the same concentration. Similar sequential reductions were observed for the other components of the mixture. The results in Figure 3 indicate that the initial rate of adsorption by the SPME fiber of the components in the gas mixture was inversely proportional to their molecular weight. Thus, the
Table 2. Analysis of Mixtures of Volatile Organic Sulfur (VOS) Compounds Following Sequential Addition of Components at 1.0 mg/La VOS MeSH Me2S Me2S2 CS2 PITC
integrator peak area 33658
9951 112840
5902 22027 139912
2145 13780 129903 277
770 2008 24677 ndb 298878
a Each sample was adsorbed onto a 75-µm Carboxen/PDMS SPME fiber for 15 min and desorbed for 5 min in the GC injector operating in split mode (60:1). b nd, not detected.
adsorption of MeSH and Me2S was greatest in the 0.5-min sample but declined thereafter, whereas the maximum for Me2S2 was observed in the 2-min sample. At least 10 min was required for PITC adsorption to stabilize at its maximum value, by which time the adsorption of MeSH, Me2S, and Me2S2 had declined to 14, 12, and 54% of their respective maximum values. The results of this and the preceding experiment accord with classical adsorption mechanism theory that, due to limited adsorption sites, the analytes with low affinity for these sites can be displaced by analytes with higher affinity, with the eventual establishment of a dynamic equilibrium (after 10 min in this study). Displacement of MeSH and Me2S is evident in this study at a concentration of 0.01-0.1 mg/L (Figure 1b). It is difficult to determine whether this range is typical for samples studied by other workers, as some reported values derived from SPME may be in error due to the unrecognized effect of competition. In our studies, however, headspace samples collected from brassica often fall within this range.
CONCLUSIONS These results demonstrate that it is not valid to use singlecomponent gas standards to analyze multicomponent mixtures of low molecular weight VOS compounds by SPME using Carboxen/PDMS fibers. The magnitude of the effect in this study is such that even small differences in the composition of standard and sample can result in large errors in quantitation. Although using a standard of appropriate composition would circumvent this limitation, in practice the correct standard cannot be identified until the composition of the sample is known. In principle, a series of iterative steps could be taken to match the standard and sample, but this would have to be done for each sample of different composition and is not practically feasible. This principle also presupposes that the identity and concentration of every component in the sample that might displace low molecular weight compounds from the fiber is known, so that an identical standard can be prepared. A good example of this limitation is the observed suppression of MeSH and Me2S (and to a lesser extent, Me2S2) adsorption by the presence of CS2, itself barely detectable. ACKNOWLEDGMENT This work was funded under the U.K. Horticulture LINK program by MAFF (Project Hort216/HL0136LSF), the Horticultural Development Council (Project CP6) and by the consortium members: Ambiox Ltd., Berry World Ltd., East Malling Trust, E S Black Ltd., Langmead Farms Ltd., The MicroBio Group Ltd., Standen Reflex, Tesco Stores Ltd., and TIO Ltd., project coordinator C. Mullins, Aberdeen University. Received for review October 3, 2000. Accepted January 23, 2001. AC001176M
Analytical Chemistry, Vol. 73, No. 7, April 1, 2001
1649
