Fast GC×GC with Short Primary Columns - Analytical Chemistry (ACS

Shellie and Marriott investigated the use of fast GC elution conditions for chiral ... in the laboratory and SGE International for specially made colu...
0 downloads 0 Views 241KB Size
Anal. Chem. 2006, 78, 2028-2034

Fast GC×GC with Short Primary Columns James Harynuk and Philip J. Marriott*

Department of Applied Chemistry, Royal Melbourne Institute of Technology, GPO Box 2476V, Melbourne, Victoria 3001, Australia

A novel approach to comprehensive two-dimensional gas chromatography (GC×GC) separations is presented, which operates in a new region of the “GC×GC optimization pyramid”. The technique relies on the use of short primary columns to decrease elution temperatures (Te) of analytes from the primary column, with a Te reduction of up to 50 °C illustrated. This in turn has implications that will expand the areas where GC×GC can be used, as decreased elution temperatures will allow GC×GC to be applied to mixtures of less volatile compounds or permit the use of less thermally stable stationary phases in the column ensemble. As well, it will allow GC×GC to be applied to thermally labile compounds through a reduction in elution temperature. With short primary columns, resolution and efficiency in the first dimension is sacrificed, but speed is gained; however, the second column in GC×GC provides additional resolution and separation of compounds of differing chemical properties. Thus, it is possible to recover some of the analytical separation power of the system to provide resolution of target analytes from sample impurities. As an example, a case study using short primary columns for the separation of natural pyrethrins, which degrade above 200 °C, is described. Even with the sacrifices of overall separation power that are made, there is still sufficient resolution available to separate the six natural pyrethrins from each other and the complex chrysanthemum extract matrix. The use of cold-on-column injection, a short primary column, and a high carrier gas flow rate allow the pyrethrins to be eluted below 200 °C, with separation in 17 min and complete resolution from sample matrix. Comprehensive two-dimensional gas chromatography (GC×GC) has been successfully applied to perform separations in many fields of research that conventionally rely on one-dimensional GC (1D GC) for their separation needs. Petrochemical separations were the first to take advantage of the new-found power that the technique offers,1 but this was followed rapidly by applications in fields as diverse as GC×GC fragrances and flavors,2 environmental analysis, and health research,3 to name a few. Detailed recent * To whom correspondence should be addressed. E-mail: philip.marriott@ rmit.edu.au. Tel: + 61-3-99252632. Fax: + 61-3-96391321. (1) Venkatramani, C. J.; Phillips, J. B. J. Microcolumn Sep. 1993, 5, 511-516. (2) Marriott, P.; Shellie, R.; Fergeus, J.; Ong, R.; Morrison, P. Flavour Fragrance J. 2000, 15, 225-239. (3) Liu, Z.; Sirimanne, S. R.; Patterson, D. G., Jr.; Needham, L. L. Anal. Chem. 1994, 66, 3086-3092.

2028 Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

reviews of the technique and its applications can be found in the literature.4-7 The reason for the interest in GC×GC is clear. Through the serial connection of two GC columns with different selectivities via a GC×GC modulator, mixtures of analytes are separated by two (orthogonal) mechanisms and the resultant peaks are arranged in a two-dimensional plane with a much greater peak capacity than a 1D GC separation. Not only does this allow much more detailed separations of samples so complex that with any single-column analysis there would be multiple coelutions, but in performing the separation, the ordered retention patterns that exist for homologues and other related compounds (but are often obscured due to the lack of separation space in 1D) become much more obvious. This feature can be used to tentatively identify unknown peaks in a chromatogram based on their position relative to other known peaks, with a high degree of confidence. A third benefit to GC×GC separations is that because all of the material exiting the primary column is collected and focused by the modulator into narrow peaks that elute with peak widths typically in the range of 30-100 ms at half-height, peak intensities are greatly enhanced. These bands of analytes that are injected to the secondary column invariably include contaminants such as septum and primary column bleed, which are also focused by the modulator. However, the secondary column is usually capable of resolving these extraneous peaks from the analyte peaks, and in the cases where this is accomplished, the sensitivity of the technique over conventional GC analysis is greatly increased because the only sources of noise bleed from the secondary column and noise associated with the detector (contaminants in the detector gases, electronic noise, etc.) which combined are usually much less than the chemical noise associated with the injector and primary column, especially as the oven temperature increases. With the introduction of GC×GC, it has become apparent that bleed associated with GC columns does have a structure that is rather overlooked in single-column analysis. The need to employ a second column in GC×GC often means that analysts look to polar, less thermally stable phases for orthogonal separation, when usually they might use a low-polarity, more thermally stable phase in single-column analysis. This compromises the upper operating temperature of the analysis. Also, polar (e.g., poly(ethylene glycol)) phases are seen to have many bleed components in the (4) Go´recki, T.; Harynuk, J.; Paniæ, O. J, Sep. Sci. 2004, 27, 359-379. (5) Marriott, P. J.; Shellie, R. A. Trends Anal. Chem. 2002, 21, 573-583. (6) Beens, J.; Brinkman, U. A. Th. Analyst 2005, 130, 123-127. (7) Marriott, P. J.; Morrison, P. D.; Shellie, R. A.; Dunn, M. S.; Sari, E.; Ryan, D. R. LC-GC Eur. 2003, 16, 23-31. 10.1021/ac0519413 CCC: $33.50

© 2006 American Chemical Society Published on Web 02/04/2006

two-dimensional (2D) GC×GC separation space when used as the first column in GC×GC. This becomes of concern when the upper temperature of an analysis nears the recommended maximum phase temperature, and in this context, reductions in elution temperature can be advantageous for GC×GC separations. With all that is offered by GC×GC, it is surprising that it has not been more widely accepted and adopted as a standard technique by the GC community. There are several possible reasons for this, including the cost of the instrumentation, the relatively higher level of operator skill that is required for developing analytical procedures and analyzing data (when compared to 1D GC analyses), and preconceived notions about the complexity of the technique. Some of these ideas may, in turn, originate from previous experience with older generations of multidimensional GC equipment, which could be cumbersome to optimize. The newer generations of conventional multidimensional GC systems are much easier to operate, due to increased sophistication in machining and new tools such as electronic pressure control. From a point of view of physically setting up the instrumentation for GC×GC separations, the commercial GC×GC equipment that is currently available is easier to set up than the conventional multidimensional GC equipment. When examining the other possible reasons why GC users may be reluctant to adopt GC×GC separations, there are three that become readily apparent: analysis time, perceived lack of need, and applicability of the technique to their sample. Most of the GC×GC separations that have been reported in the literature to date require long analysis times, usually more than 30 min, and often 1 h or more. Many researchers and academics are willing to invest the time and wait for a high-resolution separation to complete, but in a field where the trend in recent years has been toward higher throughput and faster turnaround, these long analyses are unattractive for industry and commercial laboratories. Many analysts may also believe that their current separation may be good enough for their task. In many cases, this may be true as they may only have a handful of analytes that can be easily separated by 1D GC. However, this belief could cause difficulties in accurate quantitation, due to the coelution of contaminants in the standards with analyte peaks (as shown below) or other coelutions that were not observable until a GC×GC separation was performed that cast doubts on results from previous singlecolumn analyses.8 Finally, while GC×GC may be conceptually attractive for some analysts, they may feel that they cannot use the technique because they have compounds that either elute at higher temperatures than those permissible by the polar phases required for one of the GC×GC separation dimensions or because their analytes are thermally labile. The present research aims to dispel some of these ideas, possibly making GC×GC more attractive to some users and extending the range of applications to much more thermally labile compounds or less volatile compounds. The approach relies on the use of a short primary column to decrease the elution temperatures (Te) of compounds and to speed up the overall analysis time. Previously, short thick-film columns have been used in the first dimension of a GC×GC separation, but for different reasons than those here. A 3.8 m × 0.1 mm × 3.5 µm 100% poly(8) Xu, L.; Reddy, C. M.; Farrington, J. W.; Frysinger, G. S.; Gaines, R. B.; Johnson, C. G.; Nelson, R. K.; Eglinton, T. I. Org. Geochem. 2001, 32, 633645.

(dimethylsiloxane) primary column has been used with conventional GC×GC oven programming rates and flow rates in the analysis of analytes including the BTEX (benzene, toluene, ethylbenzene, xylenes) compounds in petrol.9 As these peaks elute early in the chromatogram, a thick-filmed column was required to provide peaks that were broad enough to be modulated often enough with the 4-s modulation period that was used. However, analytes also included heavier aromatic compounds, and a longer column with this phase ratio would likely have been too retentive for the larger compounds. In contrast, this technique is aimed at decreasing elution temperatures and increasing the speed of GC×GC separations. As a demonstration, the technique is applied to the separation of natural pyrethrin compounds in Chrysanthemum cineraefolium extract. The natural pyrethrins are a family of six compounds extracted from chrysanthemum flowers and are valuable insecticides, especially for organic farmers, as their use is permitted in organic agriculture.10,11 For the purpose of demonstration of this novel GC×GC approach, these analytes are useful due to the fact that they are thermally labile, degrading at temperatures over 200 °C.11,12 Additionally, when extracted from chrysanthemum flowers, they are present in a reasonably complex matrix that can have numerous coeluents and present a challenge to the analyst where resolution of all components is required. EXPERIMENTAL SECTION GC analyses were conducted on a 6890GC (Agilent Technologies, Burwood, Australia) equipped with a cold-on-column injector (COC). For GC×GC separations, modulation was performed using an LMCS II modulator (Chromatography Concepts, Doncaster, Australia). This modulator uses liquid CO2 as the cryogen. Detection was by FID at 20 and 100 Hz for 1D and GC×GC separations, respectively. Conventional 1D GC separations were performed using a 30 m × 0.25 mm × 0.25 µm BPX5 column (5% phenyl equivalent polysilphenylene siloxane phase; SGE International, Ringwood, Australia) with a 4.8 m × 0.25 mm deactivated fused-silica retention gap (SGE); the carrier gas was H2 at 1.5 mL/min constant flow. A 1-µL sample of solution was injected by COC injection. The oven was programmed from 50 (1-min hold) to 320 °C at 15 °C/min. To compare the conventional column with a short column, 1D GC separations were also performed using a 5 m × 0.25 mm × 0.25 µm BPX5 column with a 5 m × 0.25 mm deactivated fusedsilica retention gap (SGE) and COC injection with H2 carrier gas at 1.5 mL/min. The oven program was the same as that used in the conventional analyses. GC×GC separation of the pyrethrin standards and chrysanthemum extract was conducted using the short column with retention gap and COC injection for the first dimension (1D) and a 0.3 m × 0.15 mm × 0.15 µm BPX50 column (50% phenyl equivalent polysilphenylene siloxane phase; SGE) in the second dimension (2D). The modulator was operated with a modulation (9) Frysinger, G. S.; Gaines, R. B.; Ledford, E. B., Jr. J. High Resolut. Chromatogr. 1999, 22, 195-200. (10) Zang, X.; Fukuda, E. K.; Rosen, J. D. J. Agric. Food Chem. 1998, 46, 22062210. (11) Po´l, J.; Wenclawiak, B. W. Anal. Chem. 2003, 75, 1430-1435. (12) Bicchi, C.; Brunelli, C.; Galli, M.; Sironi, A. J. Chromatogr., A 2001, 129140.

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

2029

Figure 1. Improved model of the GC×GC optimization pyramid proposed by Dimandja.13 Each face of the pyramid is a parameter that can be optimized. Here sensitivity is shown as gray (back face), structure is shown as white (bottom face), resolution is shown with diagonal lines sloping up (right-front face), and speed is shown with diagonal lines sloping down (left-front face). Most GC×GC separations so far in the literature have operated in region A. In the present research, GC×GC is applied in region B. See text for detailed explanation of the figure and refinements to the model.

period of 3.0 s at +20 °C. Modulation was performed on the 2D column, with injections occurring 9.5 cm from the start of this column; thus, the effective length of the secondary column was 20.5 cm. Cold-on-column injection was used with a carrier gas flow rate of 8.0 mL/min. The oven program that was used for the GC×GC separations was 50 (1-min hold)-170 °C at 50 °C/min, then to 210 °C at 5 °C/min, and finally to 320 °C at 40 °C/min (3-min hold). The samples that were used for testing the technique were a series of linear alkanes from C8 to C30 in hexane (Supelco, Bellefonte, PA), a pyrethrin standard (Supelco) diluted to 10 mg/L total pyrethrin in 2-propanol, and a sample of chrysanthemum extract, spiked with 10 mg/L total pyrethrin. All connections between columns were made using glass press-fit unions. The chrysanthemum extract was prepared by chopping chrysanthemum flowers (0.2 g) and subjecting them to pressurized liquid extraction using a Dionex 200 ASE instrument. The extraction was conducted using 2-propanol at 130 °C and 1500 psi, without any further sample cleanup or preparation. The extracts were spiked with pyrethrin because the flowers were found to have almost no type II pyrethrin content. Whether this was due to an extended storage period prior to extraction or the fact that these particular flowers contained essentially no type II pyrethrins to begin with is unknown and irrelevant for this particular study. RESULTS AND DISCUSSION Recently, Dimandja has described the GC×GC optimization pyramid,13 based on the well-known GC optimization triangle of speed, resolution, and sensitivity. In his conceptual model, the parameter of structure is added at the fourth vertex of the pyramid. This model shows the interrelation of the four parameters and how, depending on how close to a given vertex one operates, they can choose to optimize for one parameter at the expense of the others; however, we propose a slight modification to the model. We propose that rather than placing each of the parameters on a vertex of the pyramid, each is placed on a face, as shown in Figure 1. The reason for this is that, with the parameters on the vertices, it is implied that, in the optimization of one parameter, all others (13) Dimandja, J.-M. Mediterranean Separation Science Symposium, University of Messina, Italy, July 2005.

2030 Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

must be compromised. GC×GC separations do not appear to be so restricted and perhaps a more accurate picture is obtained using the model in Figure 1. A set of separation conditions is viewed as a point within the volume of the pyramid; the closer the point is to a given face, the more optimal the conditions for that particular parameter. If the point resides on a given face of the pyramid, that parameter is fully optimized, and there is a compromise between the other three parameters. However, it is possible to optimize for two parameters, for example, resolution and sensitivity, by choosing conditions on an edge where two faces intersect and allow a compromise between the remaining two parameters. It is even possible to optimize for three of the parameters at the expense of the last by choosing conditions that are at a vertex where three faces intersect. As most GC×GC chromatographers have, to date, been interested in demonstrating the power of the technique and applying it to highly complex separations that would be impossible by conventional GC, they have operated within the region marked A in Figure 1. In this region, the separation is aimed at obtaining a high degree of resolution, preserving the 2D chromatographic structure, and achieving good sensitivity. This comes at the expense of analysis time. In the present research, the region B in Figure 1 is explored. For many separations, the high peak capacity offered by GC×GC may be excessive, and this excess resolution may be sacrificed for speed, while still maintaining sensitivity and chromatographic structure. It is well known that, in conventional GC separations, speed can be gained by utilizing shorter columns, higher linear velocities, or both. One of the consequences of applying these principles is that the elution temperature (Te) of compounds is decreased, all other parameters being held constant. This was previously demonstrated in 1D GC for a host of compounds, including the natural pyrethrins where a shorter conventional i.d. capillary column was employed that necessarily resulted in loss of efficiency.12 This is demonstrated by the separation of alkanes in Figure 2. Panel A in this figure shows the separation of alkanes on the ubiquitous 30 m × 0.25 mm × 0.25 µm 5% phenyl 95% methylsiloxane column with a 4.8-m deactivated fused-silica retention gap, using a typical flow rate (1.5 mL/min) and oven program (15 °C/min) with COC injection. In panel B, a 5-m length of this column (with a 5-m-long retention gap) is used to perform the separation of the alkane sample with the same flow rate and oven program. These two separations serve to demonstrate the reduction in elution temperature that can be achieved with a short column; on the 30-m column, the Te for n-C30 is 313 °C, and on the 5-m column, the Te has decreased by 50 °C to 263 °C. The implications of this are important not only for the analysis of thermally labile compounds, as will be shown here, but also for general GC×GC analysis. In many applications, the upper temperature of the separation is limited by whichever of the two columns has the lower thermal stability (and may be more critical where the less stable column is in the first dimension). This means that for GC×GC separations that have peaks eluting from the primary column at temperatures in excess of ∼260-280 °C, the choices for stationary-phase combinations are very much restricted. However, if the Te of a compound could be reduced from 300 to ∼250 °C, the choices of stationary phases would be greatly increased.

Figure 2. Separation of linear alkanes on a BPX5 phase using a conventional 30-m column with a 4.8-m retention gap and COC injection (A) and using a 5-m segment of the same column with a 5-m retention gap and COC injection (B). Carrier gas flow rates and oven programs were identical for the two analyses. The Te of C30 is decreased from 313 °C in (A) to 263 °C in (B).

The advantages of this approach for the analysis of thermally labile compounds is demonstrated by the separation of a pyrethrin

standard (Figure 3) under the same conditions as those used for the 1D alkane separations. When the injection is carried out under conventional GC conditions (Figure 3A), there are numerous degradation products that are formed due to degradation on the column as the elution temperatures of all the pyrethrins are above 250 °C. Due to the extent of degradation, it is impossible to identify the pyrethrin compounds in the sample based solely on their retention time. The peaks eluting in the first portion of the chromatogram (2-7 min) are from other compounds in the standard, likely resulting from the chrysanthemum sample from which the standard was derived. When the same standard is analyzed with the short GC column (Figure 3B), it becomes possible to identify the pyrethrin compounds based on their retention times, although there is likely some degradation that occurs during the separation as evidenced by the extra peaks in the pyrethrin elution region (when compared to Figure 4) and the fact that the pyrethrin isomers elute at temperatures that range from 207 to 243 °C, still high enough to cause their degradation. In this separation, there are also some compounds that coelute or nearly coelute with the pyrethrins and that may be part of the sample. While it can be argued that choosing a different stationary phase for the column may in fact be more suitable for this particular separation, the present column serves for demonstrating the principle of using shorter columns in GC×GC. Additionally, when compounds from the extraction of the crude pyrethrins from the chrysanthemum sample are added to the chromatogram, they will increase its complexity and quite possibly cause coelutions regardless of the phase used. Thus, any attempts to elute the pyrethrins from this column at lower temperatures would require further sacrifices in resolution that

Figure 3. Separation of pyrethrin standard. In (A), the conditions of Figure 2A are used, and there is extensive degradation of analytes as shown in the expanded region where the pyrethrins all elute above 250 °C. In (B), the conditions of Figure 2B are used, and there is less degradation due to decreased Tes for the pyrethrins (between 207 and 243 °C).

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

2031

Figure 4. 1D separation of pyrethrins using COC injection, a short column GC×GC column set, a fast temperature program, and a flow rate of 8 mL/min. (A) Pyrethrin standard; (B) chrysanthemum extract spiked with pyrethrins. The elution temperature of pyrethrin II is 198 °C, below the degradation threshold of 200 °C.

cannot be afforded in a 1D mode. To gain the required resolution for the sample, one must turn to a multidimensional solutions here, GC×GC. To perform a GC×GC separation of the pyrethrin isomers, with the goal of having all isomers elute below 200 °C and in the shortest possible time, a 0.3 m × 0.15 mm × 0.15 µm BPX50 column was connected to the short column ensemble, through the LMCS module, as the secondary dimension. The oven program was changed to provide a rapid initial ramp to elute the early matrix peaks of no interest quickly, a region with a 5 °C/ min ramp where the pyrethrins elute, and a final fast ramp after the pyrethrins to bake out other matrix components. The flow rate through the column set was also increased to 8 mL/min. The 1D separations of the pyrethrin standard and the spiked chrysanthemum extracts are shown in Figure 4. In these separations, the elution temperature of pyrethrin II is 198 °C, meeting the temperature constraint for the method. There are also some possible coeluants that can be observed in the standard (Figure 4A) and some definite coelutants that are observed in the extract sample (Figure 4B). Figure 5 shows the GC×GC separations of the pyrethrin sample and chrysanthemum extracts using these conditions and a modulation period of 3 s (second dimension shifted by 967 ms for presentation purposes). As can be seen in the figure, the six compounds are clearly separated from each other and the matrix. Closer inspection of the standard chromatogram (Figure 5A) shows two compounds (marked a and b) that coelute in the first dimension with pyrethrin compounds. This shows how the separation in 1D on this particular column would likely be inadequate for quantitative analyses because these two compounds 2032

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

would make a contribution (albeit minor) to the signal being measured for jasmolin I and pyrethrin II even without the coelutions that result from the chrysanthemum matrix components. However, in GC×GC mode, the peaks are separated from each other and from the numerous matrix peaks that coelute in the primary dimension. Even with the high flow rates and the short primary column, the peak capacity of GC×GC is such that separation power can be sacrificed for speed, and baseline separation of the pyrethrin compounds from each other and the chrysanthemum extract can be maintained. It should be noted that, by shifting the GC×GC plot in the second dimension, the chromatogram becomes easier to interpret as the time window that the second dimension spans is aligned with the time window in which the compounds elute; however, when this is performed, the magnitude of the shift must be considered when comparing the 2D plot to the raw GC×GC signal. Another important feature of the method that should not be ignored is that the chrysanthemum extract has undergone no further sample cleanup or preparation than that provided by the PLE extraction. The reason for the broad vertical bands that appear at the beginning and end of the chromatogram was that, in order to conserve cryogen, the modulator was not operational during these portions of the separation. The experienced GC×GC chromatographer would almost certainly find the experimental conditions used for this separation to be surprising. The combination of a short, relatively thin-film primary column with a carrier gas flow rate that is almost 1 order of magnitude higher than what is usually used suggests that the peaks eluting from the primary column would be much too narrow to be modulated a sufficient number of times to preserve the primary separation and meet the requirements of a comprehensive multidimensional separation. However, Figure 6, which is a portion of the raw GC×GC data for the type I pyrethrins, shows that the conditions clearly allow the primary peaks to be modulated sufficiently to meet the definition of a comprehensive multidimensional separation. Obviously, the type II pyrethrins would also be sufficiently modulated as they were broader in the first dimension than the type I pyrethrins as seen in Figure 4. Pyrethrin peak base widths on the 1D column range from 9 to 15 s (see Figure 4B) so the 3-s modulation period chosen for this experiment is adequate. The relatively low Te value means that peaks might have inordinately long 2D retention (as a rule of thumb, retention doubles for every decrease of 15 °C) and so the 2D column used here was rather short (20.5-cm effective length). Considering the retention in 2D of the peaks, one must consider the absolute retention of the peaks and account for the shift in the GC×GC plot (967 ms). Thus, the absolute second-dimenson retention time (2tR) of pyrethrin I is 2.55 s and not 1.58 s as it appears in the chromatogram. Considering this peak, its half-height width is 245 ms, which is broad for GC×GC. Usually GC×GC peaks with this width are only observed when the compounds are very wellretained in the second dimension (and usually wrap around). However; in this separation the Te of pyrethrin I is 180 °C, which is at least 70 °C lower than on the 30-m column and almost 40 °C lower than on the short primary column with a conventional carrier gas flow rate. Thus, the compound should be very wellretained by the secondary column and give a broad peak. The

Figure 5. GC×GC separation of pyrethrin standard (A) and chrysanthemum extract (B) using the conditions of Figure 4 and a modulation period of 3 s. (a) and (b) in frame A show compounds that coelute in the first dimension with the standard compounds and are included in the standard.

Figure 6. Raw GC×GC signal of the type I pyrethrins from Figure 5A showing that even with the high oven programming rate and high flow rate sufficient modulations of the primary dimension are taken to preserve the separation.

efficiency was relatively low when compared to some other GC×GC separations (600 theoretical plates or 2927 plates/m for pyrethrin I). The highly retentive nature of the phase under these conditions is also illustrated by calculating the value for 2k′ ) 2tR′/ 2t for pyrethrin I, which in this separation was 161. m This technique of using short 1D columns (under more conventional flow rates) could also prove useful for chiral GC×GC. Shellie and Marriott investigated the use of fast GC elution conditions for chiral separations using an enantioselective 2D column phase.15 In that study, the goal was to achieve enantioresolution of chiral essential oil components on the second GC×GC column. It was found that the Te of analytes (and hence the conditions under which they were subjected to the 2D separation) greatly affected the separation power of the chiral 2D column.

reason for the short 2tR for this compound is attributed to the very high linear velocities under which the separation is conducted. When the flows in the columns were modeled using a flow model based on one for coupled-column systems,14 the velocity in the secondary column was found to be in the range of 1300 cm/s when pyrethrin I eluted. This also explains why the observed 2D

CONCLUSIONS The concepts explored and presented by this work are potentially wide-ranging in their scope. By using a short primary column, Te values of compounds are greatly reduced. This causes compounds to encounter the secondary column at a much lower temperature than they would normally, which in turn should increase the power of the secondary column, as the secondary separations are now conducted at lower temperatures, allowing

(14) Harynuk, J.; Go´recki, T. J. Chromatogr., A 2005, 1086, 135-140.

(15) Shellie, R. A.; Marriott, P. J. Anal. Chem. 2002, 74, 5426-5430.

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

2033

for shorter secondary columns to be used as well (strong retention may dictate that a shorter column must be employed). Using shorter 2D columns reduces the column efficiency, so the net effect is that total efficiency may be less than conventionally seen on 1-1.5-m 2D columns; however, in applying this technique, one is sacrificing separation power for speed. By eluting compounds at lower temperatures, the applicability of GC×GC to a number of conventionally difficult applications is increased, either for the analysis of thermally labile compounds, as shown here, or for the analysis of high-boiling compounds. Finally, a GC× GC separation has been demonstrated that for the first time to our knowledge utilizes the concepts of fast GC, short columns, COC injection, and higher than normal carrier gas flow rates to venture into an as yet unexplored region of the GC×GC optimization pyramid model, which has also been refined in this work. The GC×GC separation of the six natural pyrethrins

2034

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

in chrysanthemum extract was performed in 17 min. With the constraint of elution temperature, and when the ease of the sample preparationsgrinding, PLE, and direct injection of the PLE extractssis considered, an equivalent separation using a 1D GC technique would be difficult if not impossible. ACKNOWLEDGMENT The authors thank Mr. Paul Morrison for his ongoing help in the laboratory and SGE International for specially made columns used in this research. This research was also conducted with the support of the Victorian Institute for Chemical Sciences (Australia).

Received for review December 27, 2005. AC0519413

October

30,

2005.

Accepted