Anal. Chem. 2003, 75, 5532-5538
Targeted Multidimensional Gas Chromatography Using Microswitching and Cryogenic Modulation Philip Marriott,* Michael Dunn, Robert Shellie, and Paul Morrison
Australian Centre for Research on Separation Science, Department of Applied Chemistry, RMIT University, GPO Box 2476V, Melbourne 3001, Australia
A new method is described that allows fast target analysis in multidimensional gas chromatography by using a microswitching valve between two GC columns, with cryogenic trapping and rapid re-injection of trapped solutes in the second dimension. The essence of the procedure is that heart-cut fractions from the first column (1D) can be selectively transferred to column 2 (2D), where a moveable cryogenic trap first focuses the transferred solute(s) at the head of the second column and then permits their facile rapid analysis on 2D. Since 2D is a short narrow-bore column, which exhibits very fast analysis (on the order of a few seconds elution), peak responses (heights) are significantly enhanced (by up to 40-fold). Additionally, by using a 2D phase of a selectivity different from that used for 1D, it is possible to also separate components that are not resolved on the first column and to increase the resolution for other compounds. The heartcut valve isolates the section(s) of solutes of interest from the first column separation, and this provides a considerable simplification to the chromatogramsin addition to the separation and sensitivity advantages. By using this method, multidimensional gas chromatography with multiple heart-cuts can be completed within the same time as the primary column separation. Since the described method permits non-heart-cut fractions to be transferred to a monitor detector, normal detection of these fractions is still permitted. By modulation of the cryotrap, it is also possible to achieve comprehensive two-dimensional gas chromatography for the heart-cut fractions; however, only those compounds passed to the second, separation column, which passes through the cryotrap, will be subjected to GC×GC analysis. The technique and the various modes of operation are described in this paper.
such as the case when minor components are overlapped by major components (which either tail badly or are overloaded).4 Historically, the principal coupled-column GC method, providing a significant increase in separation power for regions of a complex sample in GC, involves heart-cutting from a primary column to a secondary column, and we shall refer to this as the familiar MDGC method. Techniques such as multichromatography5 or pressure tuning,5-8 which give other options for improving resolution and also employ two columns in series, column 1 (1D) and column 2 (2D), will not be further specifically described here. In conventional MDGC, the heart-cut fraction may be either focused at the start of the 2D, for instance, by cryogenic means, or may be simply passed to 2D and permitted to elute without any focusing. The first of these options may be used to collect together a number of different heart-cut fractions at the start of the second, conventional dimension, column, and then at some later stage analyze these collected fractions in a single analysis step. Thus, this method can be considered to incorporate two distinct analyses, with the total analysis time being about double that of a normal single-column analysis. Prior to the secondcolumn analysis commencing, the GC oven will be cooled to the start temperature, the cryotrap fluid turned off, and then the oven programmed for the second-column analysis. This was used for PCB specific congener analysis with five heart-cuts permitting seven target congeners to be quantified.9 Alternative, more convoluted, methods have been described. Thus, Wilkins et al.10,11 used a series of parallel traps to collect various fractions from a nonpolar column petroleum separation and then subsequently separately eluted each onto a second, polar column. Each resulting chromatogram must be considered an independent analysis, and so each will extend the analysis time by the duration that it takes to elute all the components for that heart-cut. Perhaps the ultimate MDGC procedure was that of Gordon et al.,12 who investigated a
Bertsch summarized the history and recent developments in multidimensional gas chromatography (MDGC) in a two-part review in 1999-2000.1,2 The use of coupled-column separations in GC is well established, and the driving force for such work derives from no other goal than to increase the separation power of the technique.3 Often, this also yields improved quantification,
(4) Schomburg, G. In Sample Introduction in Capillary Gas Chromatography; Sandra, P., Ed.; Dr. Alfred Huthig Verlag: Heidelberg, 1985; Vol. 1, pp 235261. (5) Hinshaw, J. V.; Ettre, L. S. Chromatographia 1986, 21, 561-572. (6) Kaiser, R. E.; Leming, L.; Blomberg, L.; Rieder, R. I. J. High Resolut. Chromatogr., Chromatogr. Commun. 1985, 8, 92-97. (7) Sandra, P.; David, F.; Proot, M.; Diricks, G.; Verstappe, M.; Verzele, M. J. High Resolut. Chromatogr., Chromatogr. Commun. 1985, 8. (8) Sacks, R.; Akard, M. Environ. Sci. Technol. 1994, 28, 428A-433A. (9) Kinghorn, R. M.; Marriott, P. J.; Cumbers, M. J. High Resolut. Chromatogr. 1996, 19, 622-626. (10) Wilkins, C. L. Anal. Chem. 1994, 66, 295A-301A. (11) Ragunathan, N.; Krock, K. A.; Klawun, C.; Sasaki, T. A.; Wilkins, C. L. J. Chromatogr., A 1999, 856, 349-397.
* To whom correspondence is to be addressed. E-mail: philip.marriott@ rmit.edu.au, Tel: + 61-3-99252632. Fax: + 61-3-96391321. (1) Bertsch, W. J. High Resolut. Chromatogr. 1999, 22, 647-665. (2) Bertsch, W. J. High Resolut. Chromatogr. 2000, 23, 167-181. (3) Giddings, J. C. Anal. Chem. 1984, 56, 1258A-1270A.
5532 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
10.1021/ac034492k CCC: $25.00
© 2003 American Chemical Society Published on Web 09/04/2003
very complex tobacco flu-cured essential oil sample and sampled the first-dimension separation into 23 separate fractions, with each individually sequentially analyzed as a second-dimension run. The total analysis time was 48 h; however, the total separation was impressive, many new compounds were identified, and a large number of components were resolved. The revelation of such complexity, although perhaps not surprising, was clearly tempered by the total analysis time, and since then the literature has been relatively silent on other groups adopting such a procedure. Perhaps the most active area for MDGC has been for essential oil analysis13-15 and, specifically, applications such as chiral separations with heart-cutting from an achiral primary column to an enantioselective column.16,17 In 1991, Liu and Phillips18 reported a “comprehensive twodimensional” GC technique (GC×GC) that for the first time could subject all sample components to the total separation potential of two sequential GC columns, with total analysis time corresponding to the elution time of the first column. While the literature describing GC×GC is not substantial, applications demonstrating the complexity of cigarette smoke using TOFMS detection,19 and high-resolution essential oil analysis20 including chiral analysis,21 are available and may be compared with the above traditional MDGC studies. These are sufficient to illustrate the promise and potential of GC×GC. The key to realizing what previously had been hypothetically proposed by Giddings22 was to use a modulation process that effectively collected sequential parts of the firstcolumn effluent and rapidly separately re-inject each to the second column, with the process continuing over the total elution period. The expectation is that the two-column analysis may resolve components that the single column by itself is unable to. The only viable way to implement this procedure is to ensure that the collection process is faster than the bandwidth of an individual peak eluted from 1D. Thus, about four samplings per 1D peak is considered acceptable.23 Each component is now split into a number of separate subpeaks, and to prevent each seconddimension “subchromatogram” from mixing with the subsequent re-injection, elution on 2D must be very fastsof the order of the modulation period. Since this must be about 1-5 s, then 2D must likewise provide very fast analysis. Thus, GC×GC uses short (∼1 m), narrow-bore (∼0.1-mm i.d.), thin-film (∼0.1-µm film thickness df) columns as the second-column separation medium. The only drawback of GC×GC, a technique embraced by this group24 and (12) Gordon, B. M.; Uhrig, M. S.; Borgerding, M. F.; Chung, H. L.; Coleman, W. M., III; Elder, J. F., Jr.; Giles, J. A.; Moore, D. S.; Rix, C. E.; White, E. L. J. Chromatogr. Sci. 1988, 26, 174-180. (13) David, F.; Sandra, P. In Capillary Gas Chromatography in Essential Oil Analysis; Sandra, P., Bicchi, C., Eds.; Dr. Alfred Huethig Verlag: Heidelberg, 1987; pp 387-428. (14) Marriott, P. J.; Shellie, R.; Cornwell, C. J. Chromatogr., A 2001, 936, 1-22. (15) Bicchi, C.; D’Amato, A.; Rubiolo, P. J. Chromatogr., A 1999, 843, 99-121. (16) Bicchi, C.; Pisciotta, A. J. Chromatogr. 1990, 508, 341-348. (17) Mondello, L.; Verzera, A.; Previti, P.; Crispo, F.; Dugo, G. J. Agric. Food Chem. 1998, 46, 4275-4282. (18) Liu, Z.; Phillips, J. B. J. Chromatogr. Sci. 1991, 29, 227-231. (19) Dallu ¨ ge, J.; Vreuls, R. J. J.; Beens, J.; Brinkman, U. A. Th. J. Chromatogr., A 2002, 974, 169-184. (20) Marriott, P.; Shellie, R.; Fergeus, J.; Ong, R.; Morrison, P. Flavour Fragrance J. 2000, 15, 225-239. (21) Shellie, R.; Marriott, P.; Cornwell, C. J. Sep. Sci. 2001, 24, 823-830. (22) Giddings, J. C. J. High Resolut. Chromatogr., Chromatogr. Commun. 1987, 10, 319-323. (23) Murphy, R. E.; Schure, M. R.; Foley, J. E. Anal. Chem. 1998, 70, 15851594.
others over the past 5 years, is the need to present the data in two-dimensional format and the consequent questions surrounding automated data processing. However, the experience acquired over recent years has suggested that alternative opportunities exist for advanced separation analysis, but which may obviate the need for sophisticated data treatment or integration. Thus, the present work revisits MDGC approaches, but with rapid microvalve switching, while maintaining the very effective re-injection capability achieved with the cryogenic modulation process (referred to as the longitudinally modulated cryogenic system, LMCS) developed in this laboratory.25,26 The above strategies for fast analysis on 2D are employed. One of the modes of operation of the LMCS has been termed targeted multidimensional GC,27 without a switching heart-cut valve. This offers a capability similar to the method described below, but the entire sample has to pass through both columns, and due to the implementation procedure, it sacrifices the possibility of acquiring chromatographic data for the entire sample. It should be noted that the previously described targeted analysis procedure may still be employed in the present system. This paper describes initial experiments and observations using this new approach. EXPERIMENTAL SECTION Gas Chromatography System. All analyses were performed using an Agilent Technologies 6890 model gas chromatograph (Little Falls, DE) equipped with two flame ionization detectors (FID 1 was operated at 5-Hz data acquisition and FID 2 at 100-Hz data acquisition throughout), 7683 series autosampler, two injection modules, and Chemstation software. The GC was retrofitted with an Everest model longitudinally modulated cryogenic system (Chromatography Concepts, Doncaster, Australia), and a 10-port microswitching valve (model EH6C10WT, VICI Valco Instruments, Houston, TX). The GC was equipped with a split/splitless injector, operated at 200 °C; an injection volume of 1.0 µL was employed, using a split ratio of ∼100:1 unless otherwise stated. The carrier gas was hydrogen, and the column head pressure was 13.61 psi. Figure 1A is a schematic diagram of the instrumental arrangement, while Figure 1B shows the details of the connections to the switching valve. All valve-switching operations, cryogenic fluid supply, and targeted cryogenic trap movement are controlled through the Chemstation events control. A separate controller is used for the continual modulation used for GC×GC. The second carrier gas supply is provided through injection port 2 to the switching valve (V). From the switching valve, primary column effluent flows either to FID 1 through a segment of uncoated capillary transfer line (TL) or to FID 2 through the 2D column, which is fed through the LMCS cryotrap (CT). Both short columns between the valve and the detectors are approximately the same lengths. The arrangement ensures that hydrogen gas flows through both of these columns at all times. Separation Columns. The column set consisted of a fusedsilica capillary column of 95% methyl-5% phenyl polysilphen(24) Marriott, P.; Ong, R.; Shellie, R.; Western, R.; Shao, Y.; Perera, R.; Xie, L.; Kueh, A.; Morrison, P. D. Aust. J. Chem. 2003, 56, 187-191. (25) Marriott, P. J.; Kinghorn, R. M. Anal. Chem. 1997, 69, 2582-2588. (26) Kinghorn, R. M.; Marriott, P. J.; Dawes, P. A. J. High Resolut. Chromatogr. 2000, 23, 245-252. (27) Marriott, P. J.; Ong, R. C. Y.; Kinghorn, R. M.; Morrison, P. D. J. Chromatogr., A 2000, 892, 15-28.
Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
5533
Figure 1. (A) Schematic diagram of the switching - modulation gas chromatographic instrument. (B) Details of the valve connections and operation. TL, uncoated capillary transfer line; CT, cryotrap; V, 10-port valve. 1D and 2D are first- and second-separation columns, respectively.
ylene-siloxane (BPX5) phase (0.25-µm film thickness, df) with dimensions 30 m × 0.25 mm i.d., as the first column, terminated at the switching valve. Two columns lead from the valve to the detectors. The first comprised a segment of uncoated, deactivated fused-silica capillary (1.0 m × 0.1 mm i.d.) connected to FID 1. The second is the second-dimension separation column, with poly(ethylene glycol) (BP20) phase (0.10-µm df) of dimensions 1.0 m × 0.10 mm i.d. and is connected to FID 2. The distance from the cryotrap to the detector is slightly shorter than the column length, at ∼0.85 m. All columns were from SGE International (Ringwood, Australia). GC Conditions. For method development, the following conditions were employed: initial and final oven temperatures of 60 and 240 °C and program rate of 10 °C/min with total analysis time of 18 min. For lime oil/pesticide analysis, the same initial and final temperatures were used as above, with a program rate of 5 °C/min, and a sufficient hold time at the final temperature to ensure all components were eluted. Samples. The standard sample chosen for method development was prepared from a selection of components of different polarities and chemical types. Table 1 lists the sample components. As a more complex application sample, a cold-pressed lime oil spiked with selected organochlorine (OC) pesticides was used, where the pesticides were the target components and were in smaller abundances than most of the major essential oil components. The OC standard was obtained from Ultra Scientific (Catalog No. US-112B; North Kingstown, RI) and consisted of a 5534 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
solution of ∼2000 µg/mL in acetone of each of the following compounds (numbered in order of elution in Figure 5): 1, R-BHC; 2, β-BHC; 3, γ-BHC; 4, δ-BHC; 5, heptachlor; 6, aldrin; 7, heptachlor epoxide; 8, endosulfan I; 9, 4,4′-DDE; 10, dieldrin; 11, endrin; 12, endosulfan II; 13, 4,4′-DDD; 14, endrin aldehyde; 15, endosulfan sulfate; 16, 4,4′-DDT; 17, methoxychlor. This sample was diluted as required for GC analysis. The pesticide spiked lime oil sample was made by adding 200 µL of lime oil to 800 µL of ∼200 mg/L pesticide in hexane, giving a nominal pesticide concentration of 160 mg/L per component. This sample was injected by split mode (20:1 split ratio) into the GC column. Description of Operation. Performance of the switching valve, heart-cutting process, cyrofocusing efficiency, and targeted multidimensional and comprehensive separations was evaluated by the studies described below. A series of test mixture injections were made with the following operations: (1) Separation on 1D followed by transfer (a) through the deactivated column to FID 1 or (b) through the 2D column to FID 2. This permits reproducibility of each arrangement to be checked and also the FID response magnitudes to be contrasted for each component with each detector. (2) Separation on 1D with most sample components transferred through the deactivated column to FID 1 but with heart-cutting of selected fractions to 2D column and recorded on FID 2. This will permit the effectiveness of heart-cutting to be evaluated and also to study whether there are any changes to retention times of components arising from the action of valve switching by comparing times of components on FID 1 with those obtained in operation 1a above and on FID 2 for operation 1b. (3) Experiment 2 is repeated, but with cryotrapping of heartcut components at the head of the 2D column. Each of the heartcut fractions will be separately remobilized or re-injected into 2D and eluted before the next heart-cut is sent to 2D. In this instance, resolution of specific components on 2D may be calculated to compare with resolution obtained on the total column set. Peak area recovery measurements will indicate the effectiveness of the cyrofocusing step. (4) Experiment 3 is then repeated but the cryotrap is operated in “comprehensive” mode by using a modulation period selected for the particular study conditions. In this instance, rather than apply the GC×GC separation to all components in a sample, only those delivered to 2D via the heart-cut process will exhibit the classic modulation effects of GC×GC on the fast second-dimension column. FID 1 will continue to provide normal GC analysis of all other compounds in the sample. By using continuous modulation with method 1b, the full sample can still be subjected to GC×GC analysis. RESULTS AND DISCUSSION Comparison of Elution through the Two Sequential Column Arrangements. Since one of the second columns (TL) is uncoated, and hence nonretaining, then the difference in retention times in operations 1a and 1b should be close to the degree to which the retaining column 2D holds up the individual solutes. It will also illustrate how comparable the second columns are in their overall restriction effect on carrier gas flow. If the two columns are exactly matched, then they should have very close overall retentions. This is important when various heart-cut events are
Table 1. List of Compounds Studied and Their Retention Times for Operations 1 and 2 (n ) 5) tR (% RSD)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
compound
FID 1 operation 1a
FID 2 operation 1b
FID 1 operation 2
hexan-1-ol nonane R-pinene propylbenzene heptan-1-ol decane γ-terpinene octan-1-ol undecane linalool nonan-1-ol terpinen-4-ol naphthalene dodecane decan-1-ol bornyl acetate tridecane
4.563 (0.027) 4.852 (0.031) 5.465 (0.030) 5.773 (0.034) 5.871 (0.033) 6.184 (0.034) 7.222 (0.036) 7.317 (0.034) 7.624 (0.038) 7.764 (0.038) 8.785 (0.038) 9.069 (0.037) 9.145 (0.035) 9.423 (0.037) 10.217 (0.033) 10.469 (0.032) 10.543 (0.038)
4.770 (0.000) 5.053 (0.009) 5.679 (0.010) 5.991 (0.000) 6.092 (0.000) 6.399 (0.009) 7.448 (0.006) 7.546 (0.006) 7.846 (0.007) 7.995 (0.007) 9.019 (0.006) 9.296 (0.005) 9.385 (0.005) 9.675 (0.005) 10.454 (0.005) 10.699 (0.004) 10.783 (0.004)
4.562 (0.010) 4.850 (0.015) 5.464 (0.010)
Figure 2. Chromatograms of the test sample for operations 1 and 2, without cryotrapping. (A) FID 1 result for operation 1a. (B) FID 1 result for operation 2. (C) FID 2 result for operation 2 (without cryotrapping)
to be implemented, since these events should not significantly affect the retentions of later solutes. Five repeat injections were made in each valve configuration. Figure 2A illustrates the chromatogram of the test solutes (Table 1) on FID 1. The respective chromatogram of FID 2 for operation 1b is not significantly different from Figure 2A and so will not be shown; Table 1 summarizes data for retention time (tR) values and relative standard deviations (RSDs) for all components under these two operations. Table 2 compares area data from the two operations for the standard solution. It is noted that FID 1 responds slightly greater than FID 2 (refer to operations 1a and 1b data, Table 2), and the heart-cutting operation does not affect the quantitative measurement of relative peak responses for those compounds recorded at FID 1 and FID 2. Minor variations in peak areas may be attributed to small changes (increases) expected in concentrations as solvent evaporates from the injection vial. In summary, the process of using a valve between 1D, TL, and 2D, and valve switching to deliver components to 2D does not affect quantitative recovery of components at either FID 1 or FID 2. The small
FID 2 operation 2
5.784 (0.008) 5.896 (0.008) 6.205 (0.009) 7.261 (0.010) 7.368 (0.007) 7.665 (0.007) 7.804 (0.011) 8.816 (0.005) 9.103 (0.011) 9.190 (0.011) 9.460 (0.009) 10.246 (0.008) 10.501 (0.005) 10.584 (0.005)
difference in tRs (of ∼0.1-0.2 min) and responses on the two detectors were not significant between operations 1a and 1b, and no adjustment in flame conditions or column lengths to correct for the differences was deemed necessary. Effect of Heart-Cutting Events. With the system set up to transfer effluent to the uncoated transfer line TL, the switching valve was programmed to pass four selected heart-cut fractions to the coated 2D column. Without cryotrapping, it was expected that (i) the heart-cut component tRs on 2D would be equal to those obtained in operation 1b and (ii) the components that travel directly to the uncoated column and are recorded at FID 1 will have the same tRs as they had under operation 1a. Figure 2B shows a typical chromatogram of the FID 1 result, and the FID 2 result for the heart-cut fractions is given in Figure 2C. Table 1 lists component retention times for each of these results and an indication of their comparison with the respective tRs where no heart-cuts are used can be gleaned. The data indicate that the system is quite well balanced with the respective retention times within small absolute margins for the different operations. This supports the robustness of the system to perform multiple heartcuts. In a further test, a single valve switch was conducted, but using a range of switching times (e.g., 0.5-3.5 min), which showed that again the retentions of components eluting later than the time of the heart-cut event were not affected by more than a few seconds, and efficiency on 1D was retained. Cryotrapping of Heart-Cut Fractions. Targeted Analysis. The heart-cutting events described above do not permit an improvement in solute resolution under the conditions employed (short 2D column, no cryotrapping). Since most of the travel of the solutes is on 1D, then FID 1 and FID 2 results essentially reflect the separation obtained at the end of 1D. However, in GC×GC, it is known that the second-column performance requires very narrow bands or slices to be passed to the second column. This then permits high 2D efficiency to be achieved, and the effect of this column’s selectivity upon separation of co-trapped solutes can be realized. In the present case, this is achieved by cryofocusing the heart-cut bands. Provided that these are very narrow, and are rapidly remobilized, then it is expected that 2D might provide a Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
5535
Table 2. Peak Areasa for Standard Mixture Components under Various Operations (n ) 5) area (% RSD) compound
FID 1 operation 1a
FID 2 operation 1b
FID 1/2 operation 2
FID 1/2 operation 3
hexan-1-ol nonane R-pinene propylbenzenec heptan-1-olc decane γ-terpinenec octan-1-olc undecane linalool nonan-1-ol terpinen-4-olc naphthalenec dodecane decan-1-ol bornyl acetatec tridecanec
23.9 (1.5) 32.6 (1.1) 21.4 (1.0) 39.1 (0.8) 25.7 (1.2) 33.7 (1.0) 36.9 (0.9) 25.5 (1.2) 36.0 (1.1) 30.8 (1.0) 27.6 (0.8) 36.8 (0.9) 30.9 (1.0) 43.4 (0.7) 28.8 (1.3) 37.5 (0.8) 37.5 (0.9)
19.6 (0.8) 30.6 (0.4) 20.1 (0.3) 36.0 (0.4) 21.8 (1.1) 31.6 (0.4) 34.7 (0.3) 21.6 (0.6) 33.7 (0.3) 28.7 (0.1) 23.7 (0.8) 34.4 (0.6) 29.1 (0.8) 40.0 (0.4) 25.2 (0.8) 34.4 (0.6) 33.0 (0.4)
23.8 (1.2)b 33.1 (1.1) 21.7 (0.7) 37.0 (0.9) 22.1 (1.1) 45.9 (1.4) 36.8 (0.9) 22.7 (1.2) 44.0 (1.7) 31.1 (0.8) 27.6 (0.9) 36.4 (0.9) 30.9 (1.1) 45.0 (0.9) 28.6 (0.8) 38.3 (3.0) 35.8 (0.9)
24.0 (1.5)b 33.5 (1.2) 22.0 (0.7) 31.9 (1.6) 25.4 (0.6) 45.3 (3.1) 42.1 (0.7) 25.3 (2.5) 36.6 (0.6) 31.2 (0.6) 27.6 (1.6) 41.7 (1.4) 36.5 (1.1) 45.9 (0.7) 29.1 (0.8) 43.5 (0.7) 43.0 (0.6)
a Peak areas in pA‚s. b The components in italics are those recorded at FID 1 (i.e., the non-heart-cut components). c These components are heart-cut to 2D in the valve-switching operation.
Figure 3. Comparison of chromatograms for the test sample obtained on FID 2 using four heart-cuts (operation 2). (A) No cryotrapping. (B) With cryotrapping (insets are expanded presentations of each heart-cut); remobilizations into the 2D column are at times 6.30, 7.70, 9.50, and 11.00 min respectively. The four heart-cuts are shown as 1-4, with the stepped baseline rise corresponding to the duration of each heart-cut event. The step rise is equivalent to the observed baseline response in Figure 2A.
greater opportunity for resolution. Figure 3 compares the chromatogram of direct heart-cut (no cryotrapping) elution on 2D (Figure 3A) with that obtained when cryotrapping and fast remobilization is used (Figure 3B) for each heart-cut fractions. The cryofocusing effect of the solutes as a sharply focused band, and re-injection effected by simply moving the cryotrap toward the column inlet, produce sharp bands. Expanded inset diagrams indicate the separation and enhanced resolution of the component pairs on the short 2D column. The cryotrap is moved at the following times: 6.3, 7.7, 9.5, and 11.0 min for heart-cut fractions 1-4, respectively. Note that the trapped bands are remobilized simply by exposure to the oven temperature. Table 2 results 5536 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
indicate that peak areas are preserved in this process, so no solute is lost in the switching step. Tables 3 and 4 list peak widths and resolutions of the heart-cut components for this analysis recorded at FID 2 and compare these with data on FID 1. Generally, the heart-cutting process does not significantly affect peak widths or resolutions, unless cryotrapping is implemented on 2D. Slightly larger values of wh at FID 1 may be due to the length of tubing connecting ports 5 and 10 in the 10-port valve (Figure 1B). When cryotrapping is used, even though peak pairs were trapped in the single focusing step, Rs values were uniformly greater on the short 2D BP20 column than those obtained at the end of the full (1D + 2D) column set. This is despite the effective length of 2D being
Table 3. Peak Half-Widths for Components on FID 2 for Heart-Cut Components, without and with Cryotrapping, and Their Respective Widths on FID 1
peak
component
FID 1 wh operation 1a
4 5 7 8 12 13 16 17
propylbenzene heptan-1-ol decane γ-terpinene octan-1-ol undecane linalool nonan-1-ol
0.0253 0.0260 0.0259 0.0263 0.0245 0.0276 0.0244 0.0260
FID 2 wh operation 2
FID 2 wh (cryotrap) operation 3
0.0231 0.0228 0.0239 0.0236 0.0237 0.0255 0.0233 0.0250
0.002 05 0.001 56 0.001 69 0.001 53 0.001 97 0.001 47 0.001 87 0.001 27
Table 4. Comparative Resolution of Heart-Cut Components on FID 2 without and with Cryotrapping, and Resolutions on FID 1 component pairs
FID 1 Rs (Figure 2A)
FID 2 Rs (Figure 2C)
FID 2 Rs (cryotrap; Figure 3B)
4/5 7/8 12/13 16/17
2.3 2.2 2.0 2.1
2.5 2.3 1.8 1.8
4.2 4.9 3.7 2.3
only 0.85 m. Peak heights recorded by FID 2 are, of course, very much larger when cryotrapping is used, due to zone compression effects (height data are not shown in these tables but may be compared from Figure 3 responses). Height increases of 25-40 times were noted. Comprehensive Analysis. In the above section, the cryotrap is operated in discrete modulation event modeswith one re-injection per heart-cut. By operating the cryotrap at a repetitive modulation period of 2 s, each heart-cut will be analyzed in a comprehensive GC×GC format. The cryotrap longitudinal oscillation is kept on for the total analysis, but only when the valve passes heart-cut solutes to 2D will compounds be modulated. FID 2 will, therefore, record the pulsed peaks typical of GC×GC operation, and FID 1 will simply display the normal GC result found in operation 2 above. Figure 4A is the FID 2 result for all solutes transferred to 2D, presented in contour plot style, and Figure 4B is the FID 2 contour plot result for the heart-cut process. This latter result compares well in terms of peak positions with the full comprehensive two-dimensional gas chromatography of the sample (Figure 4A). The relative peak positions in 2D space agrees with the observed resolution of peak pairs reported in Table 4, as noted by the selected circled pairs. Analysis of a Pesticide-Spiked Lime Oil Sample. The above results suggest that the switching system can be used for a more complex sample, with heart-cutting, cryofocusing, and rapid analysis of target components. This is termed fast targeted multidimensional gas chromatography here, because the heartcut fractions are analyzed under fast GC conditions on the second column, simultaneous with development of the total separation on 1D. While this example is chosen as a demonstration of implementation of the method, the analysis of pesticides in foodstuffs and other consumer goods is a common application. It is often achieved by use of specific detection to provide specificity of analysis, which cannot be achieved by use of a nonselective
Figure 4. Comprehensive two-dimensional gas chromatography result obtained while using the cryotrap with a modulation period of 2 s (operation 4). (A) 2D contour presentation of FID 2 results as per operation 1b (all components transferred to 2D). (B) 2D contour presentation of FID 2 result with only heart-cut components transferred to 2D as per operation 2. Circled components are examples of corresponding compound positions using the two procedures.
detector such as the FID. Thus, the analysis of malathion in rosemary oil28 by using pulsed flame photometric detection was contrasted with MS-TIC analysis, and multiresidue pesticide analysis of a lanolin sample using dual detection with ECD and NPD29 provided suitable analysis specificity. The results in Figure 5 show the series of comparative chromatograms obtained for this analysis. Figure 5A is the singlecolumn analysis of pesticide-spiked lime oil. The pesticide sample is shown as a separate trace in Figure 5B so that the regions of overlap of lime oil and spiked pesticide components are obvious. Component 17 eluted later than the 47-min range displayed in this chromatogram. By selecting the desired regions to heart-cut (here, those regions that show most complete pesticideslime oil component overlapsare chosen), it is possible to selectively pass those target regions to 2D that require improved separation. Thus, Figure 5C is the analysis of the lime oil spiked with pesticide, under the switching/cryotrapping/fast GC arrangement. Presented as expanded regions at the upper panel of Figure 5C is first the analysis of the pesticide sample alone (allowing accurate retention times of the pesticides to be found), and the middle panel is expanded regions R-δ of the spiked sample. Figure 5A indicates approximately the sections R-δ that are heart-cut to 2D, and the respective heart-cut durations are 28.50-28.90, 30.00-30.55, (28) Amirav, A.; Jing, H.-W. J. Chromatogr., A 1998, 814, 133-150. (29) Jover, E.; Bayona, J. M. J. Chromatogr., A 2002, 950, 213-220.
Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
5537
Figure 5. Analysis of organochlorine pesticide-spiked lime oil. All results are recorded on FID 2. (A) Chromatographic (noncryotrapped) analysis of pesticide-spiked lime oil. (B) Analysis of the pesticide sample under the same conditions as (A). (C) Main chromatogram shows analysis on FID 2 of the pesticide-spiked lime oil sample, with switching, cryotrapping, and fast GC analysis on 2D of components 1, 2 and 3, 6, and 11 and 12 using the switching event times for the four heart-cuts R-δ, respectively, shown in (A) (note that all other solutes are directed to the uncoated column and analyzed on FID 1). The upper series of expanded insets correspond to the pesticide standard sample analyzed under the same conditions, and the lower series of expanded insets correspond to the peaks for R-δ seen in the main chromatogram. Lime oil components transferred to 2D are indicated by LO.
34.66-35.00, and 41.00-41.80 min. Pesticide components 1, 2 and 3, 6, and 11 and 12 are selected in the four heart-cuts, respectively. Comparison of parts B and C of Figure 5 illustrates that component 1 response increases from about 6 to 185 pA, or 30 times, due to the cryotraping process, while component 6 increases from 3 to 110 pA, or 37 times. Absolute increases and improved detection limits depend primarily on detection acquisition rate, and retention time on the second column. Figure 5C generally supports the improved analysis of the pesticides in the heart-cut/cryotrapping/fast separation process, with the lime oil components (marked LO in Figure 5C) in most cases much better resolved from the pesticide components than they were in the original analysis (Figure 5A). CONCLUSIONS This study has demonstrated the use of a microswitching valve to effect selected heart-cutting in a multidimensional gas chromatography system, where the second column operates under fast GC conditions. The operation of the MDGC system is enhanced by use of cryotrapping focusing of heart-cut solutes on the second column. While cryofocusing is commonly used in previous MDGC methods, for this study, the use of a longitudinally modulated (moving) cryotrap allows very rapid remobilization of each of the heart-cut regions at the start of the second column to
5538 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
maintain the performance required of fast GC on 2D. Therefore, this allows on-the-fly analysis of each heart-cut on 2D during development of the chromatographic separation on 1D. Zone compression leads to the now well-recognized response increase on FID 2, while a change of stationary phase from 1D to 2D gives enhanced separation. The described system may also be operated in comprehensive two-dimensional gas chromatography mode, either for the total injected sample or for selected heart-cut regions. In the present example, 2D provides elution of heart-cut fractions within 5-10 s since a column length of only ∼1 m was used. Further improved resolution for even more complex heartcuts may be achieved by choosing a longer 2D column, at the expense of sensitivity gain. Note that, in GC×GC, 2tR values that greatly exceed the modulation period are generally not desirable since modulated peaks from successive modulations may interfere. Thus, longer 2D retention is undesirable in GC×GC. The discrete operation of the present heart-cut system does not have this constraint, although as described here, it also does not apply the 2D separation power of the coupled column system (as achieved in GC×GC) to the whole sample. Received for review May 9, 2003. Accepted July 18, 2003. AC034492K
