Journal of Chromatography A, 1365 (2014) 183–190
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Multiplexed dual first-dimension comprehensive two-dimensional gas chromatography–mass spectrometry with contra-directional thermal modulation夽 Benjamin Savareear, Matthew R. Jacobs, Robert A. Shellie ∗ Australian Centre for Research on Separation Science (ACROSS), University of Tasmania, Private Bag 75, Hobart 7001, Australia
a r t i c l e
i n f o
Article history: Received 30 May 2014 Received in revised form 3 September 2014 Accepted 5 September 2014 Available online 16 September 2014 Keywords: Gas chromatography Comprehensive two-dimensional Dual first-dimension columns Multiplexed Essential oil
a b s t r a c t A multiplexed dual-primary column comprehensive two-dimensional gas chromatography–mass spectrometry approach (2GC × GC–MS) is introduced. The approach splits injected samples into two first-dimension columns with different stationary phases, and recombines the two streams into one second-dimension column that terminates at a single detector. The approach produces two twodimensional chromatograms for each injection, and is made possible by using a dual-stage modulator operated in contra-directional modulation mode. The dual two-dimensional chromatograms produced by this single detector system provide complementary information due to selectivity differences between the three separation columns used in the column ensemble. An aged Australian tea tree (Melaleuca alternifolia) essential oil was analyzed to demonstrate the 2GC × GC–MS approach. The number of compounds separated by each of the GC × GC separations in the 2GC × GC experiment is comparable to conventional GC × GC experiments with matching column configurations. Robust peak assignment was possible for this sample based on the combination of MS library matches and multiple linear retention index searching. Forty-nine components (22 unique) were identified using a non-polar × mid-polar column combination and 34 components (7 unique) were positively identified using a polar × mid-polar column combination. Twenty-seven peak assignments were corroborated by positive identification in both of the multiplexed separations. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Comprehensive two-dimensional gas chromatography coupled with mass spectrometry (GC × GC–MS) is a highly suitable technology for essential oil characterization [1–6]. Since many essential oil constituents have similar mass spectral features [7] it is normal practice to obtain retention indices of compounds from one or more different stationary phases to facilitate correct identification of separated compounds [8]. To date many essential oil analysts have utilized dual channel GC [9] to obtain multiple linear retention indices (LRI) on different stationary phases in a single run [10]. The benefit of multiple LRI was very well-illustrated by Bicchi et al. who showed that the percentage of correct identifications obtainable through retention indices alone is approximately 65% with one stationary
夽 Presented at 38th International Symposium on Capillary Chromatography and 11th GC × GC Symposium, 18–23 May 2014, Riva del Garda, Italy. ∗ Corresponding author. Tel.: +61 3 6226 7656; fax: +61 3 6226 2858. E-mail address: [email protected] (R.A. Shellie). http://dx.doi.org/10.1016/j.chroma.2014.09.014 0021-9673/© 2014 Elsevier B.V. All rights reserved.
phase, 80% with two different-stationary phases and more than 90% if three different stationary phase columns are employed [8]. GC × GC–MS uses different stationary phases in the firstand second-dimension; however, use of the second-dimension retention time is not well established for compound identification. Although some studies have proposed different methods for calculating retention indices in the second-dimension [11–14], they have not been widely adopted due to their complexity. GC × GC–MS analysis typically depends on the first-dimension retention time and mass spectral data for compound identification [1–6]. Easy access to a second reliable LRI for separated compounds would be highly beneficial for GC × GC–MS separations. Drawing inspiration from previous GC × 2GC investigations [15–17], which use two different second-dimension separation columns and two detectors, we recently introduced a multiplexed, dual second-dimension column GC × 2GC approach that recombines effluent from the two second-dimension columns before analyte detection, eliminating the need for two detectors. The present investigation builds upon the foundation laid in our earlier work using the principle of contra-directional modulation [18]. A novel
184
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
Table 1 Advanced modulation parameters for contra-directional modulation. GC × 2GC First stage Second stage 2GC × GC First stage Second stage
Cool time till off 3000 ms Cool time till on 3000 ms
Cool off time 2500 ms Cool on time 1000 ms
Heat time till on 2500 ms Heat time till off 1000 ms
Heat on time 3500 ms Heat off time 5000 ms
Cool time till off 1500 ms Cool time till on 1500 ms
Cool off time 1200 ms Cool on time 1200 ms
Heat time till on 1200 ms Heat time till off 1200 ms
Heat on time 1800 ms Heat off time 1800 ms
approach for multiplexed dual first-dimension GC × GC (2GC × GC), providing dual LRI, coupled with MS is introduced. Qualitative analysis of Melaleuca alternifolia essential oil (Australian tea tree oil) is used to demonstrate the utility of the system developed herein. 2GC × GC–MS analysis with MS database matching and multiple LRI searching provides robust peak assignment based on three orthogonal parameters: MS, LRI1 , and LRI2 . Retention index based searches are performed on LRI data obtained from either of the first-dimension columns, or both of the firstdimension columns by cross searching. Additional supporting information upon which peak identity can be made is based on peak coordinates within the structured chromatograms obtained from the two GC × GC separations. 2. Materials and methods 2.1. Chemicals and reagents Australian tea tree (M. alternifolia) essential oil was obtained from a local supermarket (Sandy Bay, Tasmania, Australia) and stored at room temperature for an extended period (beyond the manufacturer’s recommended use-by date; SEPT/2011). A C7 –C30 n-paraffin hydrocarbon mixture (1000 g/mL; Sigma–Aldrich, Castle Hill, Australia) was used for determination of first dimension LRI. Both the Australian tea tree essential oil and alkane mixture were diluted (1:10, v/v) in dichloromethane (Sigma–Aldrich) prior to GC analyses. 2.2. Instrumentation and experimental conditions All analyses were performed using a Leco GC × GC-FID instrument with an LN2 Cooled Thermal Modulator (Leco Australia, Castle Hill, Australia). The chromatograph was equipped with split/splitless injector, operated with a 20:1 split ratio and inlet temperature of 230 ◦ C. Injected sample volume was 1 L. Hydrogen carrier gas was generated using a Parker Balston H2PEM-260 hydrogen generator. For convenience of contra-directional column installation and operation, the auxiliary (second-dimension) column oven was removed from the GC × GC system. All columns were heated using the main GC oven in all analysis. The modulator offset temperature was +15 ◦ C. Effluent from both secondary columns was monitored by a single FID operated at 100 Hz and 250 ◦ C. All GC × GC-FID results were collected and processed using Leco ChromaTOF software.
polyethylene glycol (Stabilwax; Restek); 2 D2 0.4 m × 100 m i.d. fused silica capillary coated with 0.1 m 50% phenyl, 50% dimethyl polysiloxane (Rxi-17SilMS; Restek). Total carrier gas flow rate was 2.0 mL min−1 (Hydrogen; constant flow). Each of the seconddimension columns operates at efficiency-optimized flow [19] using this flow regime, while the first-dimension column is between efficiency- and speed-optimized flow. Total modulation period was 6.0 s to maintain appropriate modulation ratio. Custom modulation parameters are provided in Table 1. The oven temperature program was 40 ◦ C (0.2 min hold) ramped at 6 ◦ C min−1 to 180 ◦ C. The first-dimension column was connected to the two second-dimension columns using a 3-port SilFlow connector (SGE Analytical Science, Ringwood, Australia). Flow from the second-dimension columns was recombined using a 3-port SilFlow connector into a 0.2 m × 220 m i.d. deactivated uncoated fused silica transfer line that terminated at the FID. The second dimension columns were installed contra-directionally in the GC × GC modulator. 2.4. 2GC × GC-FID Multiplexed 2GC × GC is achieved by using contra-directional modulation, where two parallel first-dimension columns are installed contra-directionally in the GC × GC modulator as shown in Fig. 1. 2GC × GC analyses were performed by splitting the flow from the inlet into two first-dimension columns by means of a twinhole graphite ferrule (SGE Analytical Science), all other column connections were made using press tight connectors (Restek). The first dimension columns were installed contra-directionally in the GC × GC modulator. Two first dimension columns were employed in this study; a 30 m × 250 m i.d. fused silica capillary coated with 0.25 m dimethyl polysiloxane (DB-1; Agilent Technologies, Mulgrave, Australia); and a 30 m × 250 m i.d. fused silica capillary coated with 0.25 m polyethylene glycol (DB-WAX; Agilent Technologies). The second-dimension column was a 0.45 m × 150 m i.d. fused silica capillary coated with 50% phenyl, 50% dimethyl polysiloxane (Rxi-17SilMS; Restek). Total carrier gas flow rate was
2.3. GC × 2GC-FID GC × 2GC analyses were performed by using the general instrument configuration described in reference [18]. The first dimension column 1 D was a 30 m × 250 m i.d. fused silica capillary coated with 0.25 m dimethyl polysiloxane (Rtx-1; Restek, Bellefonte, PA, USA). The two second-dimension columns were: 2 D 0.4 m × 100 m i.d. fused silica capillary coated with 0.1 m 1
Fig. 1. Illustration of the column configuration used for multiplexed 2GC × GC. (A) first-dimension Column 1: polyethylene glycol; (B) first-dimension Column 2: dimethyl polysiloxane; (C) second-dimension column: 50% phenyl, 50% dimethyl polysiloxane; (D) dual stage thermal modulator.
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
185
1.5 mL min−1 (Hydrogen; constant flow), which provides speedoptimized flow in the second column. Total modulation period was 3.0 s to maintain appropriate modulation ratio, and the oven temperature program was 40 ◦ C (0.2 min hold) ramped at 7 ◦ C min−1 to 220 ◦ C (3 min hold). Custom modulation parameters are provided in Table 1.
than 100,000 entries, containing LRI information for a wide range of aroma compounds from many literature references [20,21].
2.5. GC × GC-FID
In the present investigation, two long primary columns are installed as parallel first-dimension columns. This column configuration mirrors a previous GC × 2GC study that used two short second-dimension columns. While column installation is straightforward, a key practical consideration for contra-directional modulation approach relates to column alignment within the modulator. Proper column alignment is critical to ensure that neither of the two column segments is allowed to block the jet of cool gas onto the other column. Fig. 2 compares the multiplexed comprehensive two-dimensional chromatograms obtained from the analysis of M. alternifolia essential oil using GC × 2GC and 2GC × GC. There are advantages and disadvantages of each of these multiplexed approaches. The first advantage of GC × 2GC is that peaks in both chromatograms have matching first-dimension retention times. This makes the distribution of peaks in GC × 2GC more intuitive than the distribution observed in 2GC × GC, but this is also disadvantageous because GC × 2GC only provides a single firstdimension LRI. Having two dissimilar second-dimension columns also creates optimization challenges to avoid wrap-around. A single second-dimension column greatly alleviates this challenge. Next, it is important to use a non-polar stationary phase column in the first-dimension for GC × 2GC experiments, since retention indices are more rugged on non-polar stationary phases. However the requirement to use a non-polar stationary phase first-dimension column means that second-dimension columns can only be more polar than the first-dimension column. As the purpose of the dual second-dimension separations is to resolve any co-eluted compounds, this restriction limits peak spreading throughout the two-dimensional separation space. Indeed the individual GC × GC chromatograms produced by the GC × 2GC approach, shown in Fig. 2A exhibit more similarities than differences. A column ensemble comprising one long first-dimension column and two short second-dimension columns in GC × 2GC is not optimal for fully harnessing the complementary selectivity of the three stationary phases. Conversely, using two long first-dimension columns and one short second-dimension column provides a demonstrable
GC × GC analyses were performed using the columns as 2GC × GC experiments, except they were installed in the regular configuration through the GC × GC modulator and all column connections were made using press-tight connectors (Restek). The flow rate was 1.0 mL min−1 for all experiments. A modulation period of 3.0 s (with default modulation parameters) was employed and the temperature program was 40 ◦ C (0.2 min hold) ramped at 5.8 ◦ C min−1 to 220 ◦ C (3 min hold). 2.6. 2GC × GC–MS 2GC × GC–MS analysis was performed using the Leco GC × GC system described above, which was connected via an externally controlled heated transfer line, to the heated MS transfer line of an Agilent 5975C VL MSD (Agilent Technologies). The external heated transfer line was a 155 mm Agilent LTM transfer line, which was controlled with an LTM Controller (Agilent Technologies). The external heated transfer line and heated MS transfer line temperatures were 200 ◦ C. The MS detector was operated with a source temperature of 150 ◦ C, quadrupole temperature 230 ◦ C, with a detector voltage of 70 eV. Chemstation software (Agilent Technologies) was used to acquire MS data in the fast scanning mode with a reduced mass scan range of 40–220 m/z, yielding a detector sampling rate of 25.95 Hz. Data acquisition was triggered by a start-out signal from a remote port provided by the Leco GC × GC modulator. GC columns and other operating conditions were identical to those described above for the 2GC × GC-FID experiments. MS data was visualized using Transform (Fortner Research, Boulder, CO, USA). Mass spectra were matched to the Terpene Library (containing mass spectra of essential oil components compiled by Robert P. Adams, Baylor University Plant Biotechnology Center). Further analysis of 2GC × GC–MS results using interactive LRI filters was performed using Aroma Office 2D Software (Gerstel K.K., Tokyo, Japan). This software is a searchable database with more
3. Results and discussion 3.1. Implementation and benefits of 2GC × GC system
Fig. 2. Multiplexed two-dimensional separation space for the separation of M. alternifolia essential oil. (A) Separations obtained using GC × 2GC, (B) separations were obtained using 2GC × GC.
186
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
Fig. 3. Two-dimensional separation space for GC × GC and 2GC × GC analyses of M. alternifolia essential oil. (A) Separation obtained using GC × GC (non-polar × mid-polar), (B) separations obtained using 2GC × GC, and (C) separation obtained using GC × GC (polar × mid-polar).
improvement over GC × 2GC in utilizing the available separation space. Retention of any given solute in the second-dimension of temperature-programmed GC × GC analysis is dependent upon the solute’s interaction with the first- and second-dimension stationary phases [22]. Having two first-dimension columns in the 2GC × GC experiment is valuable since they make different contributions to retention in the second-dimension column. In the top half of Fig. 2B (dimethyl polysiloxane × 50% phenyl, 50% dimethyl polysiloxane column combination; hereafter referred to as non-polar × mid-polar) analytes are strongly retained in the second-dimension column due to strong solute-stationary phase interactions with the second-dimension column stationary phase. The separation in the bottom half of Fig. 2B, which represents a column combination of polyethylene glycol × 50% phenyl, 50% dimethyl polysiloxane (hereafter referred to polar × mid-polar), demonstrates how a stronger interaction with the first-dimension column stationary phase decreases retention in the second-dimension. The interaction of ␦elemene (peak 32) and cyclosativene (peak 37) illustrates this effect in the two chromatograms shown in Fig. 2B. Although GC × 2GC experiments previously described in the literature demonstrate beneficial class discrimination for homologous series [15–18], 2GC × GC is more appropriate for analysis of genuinely heterogeneous multicomponent samples. One may postulate that greater utilization of the two-dimensional separation space (compared to Fig. 2A) would be achieved by using a mid-polar × non-polar plus mid-polar × high-polar GC × 2GC combination, but the benefit of recording reliable LRIs would be diminished. Fig. 3 compares the chromatograms obtained from 2GC × GC and conventional GC × GC analyses of M. alternifolia essential oil. The two GC × GC analyses were performed using the same columns as the 2GC × GC experiment. The columns were not multiplexed in these experiments; they were installed in a conventional GC × GC configuration. It is evident that GC × GC chromatogram integrity is preserved in the 2GC × GC experiment. The corresponding chromatograms exhibit striking resemblance and the number of peaks separated in each GC × GC analysis is similar to the corresponding portion of the 2GC × GC analysis. Importantly, the 2GC × GC experiment maintains an appropriate number of modulation slices in line with accepted guidelines. This ensures sufficient peak sampling to minimize modulation-induced loss of
first-dimension resolution [23]. By employing a total modulation period of 3 s the modulation ratio is at least 2 in the 2GC × GC experiments, which is suitable for semi-quantitative screening purposes according to the recommendations of Khummueng et al. [24]. By nature of the contra-directional modulation approach, the second-dimension separation needs to be twice as fast the normal requirement for GC × GC. A fast second-dimension separation is especially important in 2GC × GC to avoid wrap-around, which would cause the two separation windows in the 2GC × GC chromatogram to overlap. Wrap-around is quite easily avoided by using short second-dimension columns and applying speed-optimized flow (for the second-dimension column). Fine-tuning the separation space can be achieved by manipulating the initial temperature and temperature ramp(s). A 1.5 s separation window is sufficient for the sample considered in the present investigation. Unconverted GC × GC chromatograms of 1,8-cineole (peak 7) obtained from the 2GC × GC experiment are shown in Fig. 4. The figure shows that each of the multiplexed GC × GC separations produces three modulated pulses above baseline for 1,8-cineole. Measured peak
Fig. 4. 3 s modulated peak pulses in phase obtained from 2GC × GC for 1,8-cineole (peak 7). (A) non-polar × mid-polar, (B) polar × mid-polar.
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
187
Fig. 5. Peak apex plots showing coordinates of peak maxima from 2GC × GC–MS analysis of M. alternifolia essential oil. (A) non-polar × mid-polar, (B) polar × mid-polar. Peak numbers correspond to those in Tables 2 and 3.
188
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
Table 2 Peak assignments of 27 components for which all search criteria were satisfied. Numbers in the table correspond to those in Fig. 5. MS library match quality refers to comparison with the Terpene Library. Peak no
4 6 7 17 18 23 32 35 37 39 41 43 44 47 49 50 51 52 53 54 55 56 57 58 59 60 61
Peak assignment
Myrcene p-Cymene 1,8-Cineole Terpinen-4-ol ␣-Terpineol cis-Carveol ␦-Elemene ␣-Copaene Cyclosativene ␣-Gurjunene trans-␣-Bergamotene -Caryophyllene Aromadendrene Allo-aromadendrene ␥-Muurolene Valencene Bicyclogermacrene ␣-Muurolene trans-Calamenene Spathulenol trans-Nerolidol Caryophyllene oxide Globulol Cedrol Guaiol Elemol -Cadinol
Column Set 1
Column Set 2
% match MS1
LRI1obs
2
tR(1)
LRI1lit
(LRIobs − LRIlit )(1)
% match MS2
LRI2obs
2
95 96 94 82 93 82 86 85 93 93 84 95 88 92 84 89 82 82 95 84 82 88 93 90 86 90 80
983 1014 1021 1165 1178 1200 1333 1353 1366 1406 1423 1434 1443 1454 1476 1489 1494 1499 1511 1541 1544 1551 1563 1572 1573 1591 1613
1.35 1.39 1.38 1.51 1.51 1.5 1.62 1.77 1.31 1.38 1.66 1.42 1.66 1.42 1.38 1.42 1.66 1.46 1.38 1.46 1.73 1.42 1.51 1.51 1.46 1.47 1.46
983 1014 1021 1164 1177 1202 1338 1357 1372 1411 1428 1438 1448 1458 1480 1493 1498 1504 1518 1547 1549 1557 1568 1577 1575 1595 1617
0 0 0 1 1 −2 −5 −4 −6 −5 −5 −4 −5 −4 −4 −4 −4 −5 −7 −6 −5 −6 −5 −5 −2 −4 −4
93 96 95 88 95 81 91 87 89 90 86 93 87 91 85 88 84 89 90 86 82 89 88 84 87 84 82
1164 1278 1225 1638 1731 1813 1494 1544 1485 1522 1594 1606 1609 1650 1691 1717 1755 1721 1840 2165 1974 2027 2038 2004 2115 2088 2134
0.15 0.31 0.34 0.34 0.19 0.07 0.19 0.15 0.61 0.65 0.12 0.57 0.1 0.46 0.54 0.46 0.57 0.5 0.38 0.19 0.12 0.27 0.19 0.12 0.19 0.19 0.23
areas are 718 ± 8 and 581 ± 7 (n = 4) for 1,8-cineole peaks in the non-polar × mid-polar and polar × mid-polar 2GC × GC separations respectively. Although these peak area values are statistically significantly different (p < 10−6 ) due to a slightly uneven split between the two first-dimension columns, no further effort was made to equalize the split ratio in the present study, since the aim was to demonstrate qualitative performance benefits. Previous work discussed use of normalized peak areas for semi-quantitative analysis in GC × 2GC experiments [18] and this is also likely to be beneficial for the current configuration. 3.2. Coupling of 2GC × GC system with mass spectrometry (2GC × GC–MS) Although time of flight mass spectrometry (TOF-MS) is the mass selective detector of choice for GC × GC–MS, quadrupole MS instruments have been successfully used for this purpose [2,4,25,26], usually by reducing the scan mass range to obtain a reasonable data acquisition rate [2]. The present investigation applied a reduced mass scan range from 40 to 220 m/z with fast scanning mode providing a data acquisition rate of ca. 26 Hz. This data acquisition rate is adequate for proof-of-principle qualitative analysis, although a faster instrument would be required to realize the full potential of 2GC × GC–MS analysis for quali-quantitative analysis (identification and amount). A 2GC × GC–MS peak apex plot of M. alternifolia essential oil along with peak numbers in both GC × GC separations is provided in Fig. 5. The 2GC × GC–MS chromatogram is provided in the supporting information (see Fig. S1). Supplementary Fig. S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma.2014.09.014. Primary identification of separated compounds was carried out by MS1 library searching peaks in one of the multiplexed GC × GC separations, using a minimum similarity index threshold of ≥80%. Secondary confirmation of MS1 based peak identification was achieved by comparing the first-dimension LRI1 of each peak with literature LRI using AROMA Office 2D software. The threshold
tR(2)
LRI2lit
(LRIobs − LRIlit )(2)
1160 1271 1212 1628 1729 1834 1473 1523 1481 1531 1576 1601 1605 1650 1691 1726 1740 1725 1844 2147 1959 2009 2060 2046 2096 2083 2167
4 7 13 10 2 −21 21 21 4 −9 18 5 4 0 0 −9 15 −4 −4 18 15 18 −22 −42 19 5 −33
for permitted LRI1 shift was ±10 for the dimethyl polysiloxane stationary phase column. The next level of confirmation was achieved by using LRI2 to locate a peak in the alternate multiplexed GC × GC separation. To this end, any peak name that remained after MS1 library search and LRI1 filter was input (name-search) in Aroma Office 2D. Name-search responses were limited to the polyethylene glycol stationary phase. Here, a full list of LRI2 from the database is reported for entries matching that compound name. The mean LRI2 reported by Aroma Office 2D is used to perform a reverseLRI2 search on the alternate first-dimension column, using LRI2 of ±50. This reverse LRI2 search narrows the search area in the twodimensional separation space for MS2 library searching. Additional information, such as relative peak intensity and second-dimension relative retention within the structured two-dimensional separation space is used to assist peak selection for MS2 library searching. Final confirmation is performed by MS2 library searching using the same similarity threshold as the MS1 library search. A peak name is only added to the list of identified components if the thresholds for MS1 search, LRI1 filter, LRI2 search, and MS2 search are satisfied. Table 2 provides details of 27 components identified in M. alternifolia essential oil, using the four peak assignment criteria. The identified components are also shown in a peak-apex plot (Fig. 5) to permit cross-referencing with the 2GC × GC chromatogram (Fig. 3). Table 3 provides a list of 35 M. alternifolia components that failed at least one identification criterion above. Despite failing one or more criteria, many of these components were identified using alternative strategies. These components are grouped according to the strategy used for peak assignment. Group I comprises 19 components whose identity was indicated by MS1 library searching and confirmed by LRI1 . Similarly, Group II contains four components that were identified by MS2 library searching and confirmed by LRI2 . The components assigned to Groups I and II were not found in the alternate GC × GC separation within the corresponding match thresholds. Broadening the LRI threshold did not lead to further confirmatory results. Limited further confirmation of peak identity of these 35 components is largely due to peak co-elution, which
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
189
Table 3 Peak assignments of 35 components for which at least one search criterion was not satisfied. Numbers in the table correspond to those in Fig. 5. MS library match quality refers to comparison with the Terpene Library. Peak no.
Group I 1 2 3 8 9 10 11 12 13 14 15 22 31 33 36 38 40 48 62 Group II 5 16 21 46 Group III 26 28 29 Group IV 19 20 24 25 27 30 34 42 45
Peak assignment
␣-Thujene ␣-Pinene Sabinene trans--Ocimene ␥-Terpinene Terpinolene Linalool cis-Sabinene hydrate Allo-ocimene Neo-allo-ocimene cis-Sabinol Myrtenol trans-Pinocarvyl acetate ␣-Cubebene ␣-Ylangene -Cubebene -Gurjunene Virdiflorene ␣-Eudesmol Limonene p-Cymen-8-ol trans-Piperitol ␣-Humulene cis-Verbenyl acetate iso-Bornyl acetate Myrtenyl acetate 19 20 24 25 27 30 34 42 45
Column Set 1
Column Set 2
% match MS1
LRI1obs
2
tR(1)
LRI1lit
(LRIobs − LRIlit )(1)
96 99 98 83 83 89 91 87 89 86 86 86 81 87 89 93 85 86 81
923 930 970 1032 1052 1071 1083 1083 1105 1118 1148 1190 1293 1342 1358 1387 1420 1468 1634
1.27 1.31 1.35 1.58 1.38 1.38 1.89 1.43 1.42 1.43 1.43 1.43 1.63 1.66 1.69 1.73 1.42 1.38 1.54
923 930 970 1032 1051 1070 1082 1082 1106 1117 1147 1191 1296 1348 1363 1392 1424 1473 1636
0 0 0 0 1 1 1 1 −1 1 1 −1 −3 −6 −5 −5 −4 −5 −2
– – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – –
– – – –
– – – –
– – – –
1014 1163 1204 1449
– – – –
96 93 82 85
84 87 83
1262 1275 1285
1.58 1.58 1.54
1264 1278 1290
2 3 5
– – – – – – – – –
1181 1181 1233 1244 1264 1291 1345 1431 1446
1.58 1.46 1.58 1.69 1.69 1.54 1.58 1.42 1.93
– – – – – – – – –
– – – – – – – – –
leads to poor MS library search results. For instance, limonene (peak 5) was positively identified using the polyethylene glycol stationary phase first-dimension separation using MS2 and LRI2 . The Aroma Office 2D software indicates a LRI1 of 1014 for the dimethyl polysiloxane column. Thus this peak must be eluted between pcymene (peak 6 LRI1 1014) and 1,8-cineole (peak 7 LRI1 1021) in the chromatogram, but it is obscured in the separation. This problematic peak triplet has been previously reported in the same essential oil using one-dimensional GC–MS analysis [27]. Like Group I components, identity of the three components assigned to Group III was indicated by MS1 library searching and confirmed by LRI1 . None of these components were located within the allowable LRI2 windows in the polyethylene glycol stationary phase first-dimension separation space. However, in this case the widely discussed class-separation leads to further corroboration of peak assignment. Class/structure association is often not readily apparent in essential oil analyses due to the heterogeneity of the samples, but polar first-dimension columns lead to some useful class information [28]. For example, the terpene esters that are shown as crosses in Fig. 5 were not found within expected LRI2 in the polyethylene glycol separation space. Utilizing the information provided by the structured separation space to indicate the compound class of unknown peaks, and using their relative peak intensities as a further indicator, narrows the candidate list of unassigned peaks for further MS2 library searching. For example, peak assignment of cis-verbenyl acetate (peak 26) was achieved
% match MS2
LRI2obs
2
tR(2)
LRI2lit
(LRIobs − LRIlit )(2)
– – – – – – – – – – – – – – – – – – –
1021 1022 1119 1250 1253 1278 1546 1515 1380 1421 1731 1794 1764 1466 1483 1533 1697 2191
– – – – – – – – – – – – – – – – – – –
1200 1887 1779 1663
0.27 0.07 0.07 0.07
1199 1857 1747 1663
−1 −30 −32 0
82 84 81
2228 2248 2131
0.07 0.07 0.07
– 1578 1690
– 670 441
– – – – – – – – –
1813 1759 2240 2162 2392 2177 2350 1900 2795
0.15 0.23 0.07 0.07 0.07 0.07 0.07 0.39 0.12
– – – – – – – – –
– – – – – – – – –
using this strategy. Absolute peak areas of these peaks are 4.3 × 108 and 4.2 × 108 arbitrary units respectively (2GC × GC-FID data). Peak assignment is justified by MS1 match = 84% and LRI1 = 1262, and MS2 library search upon the remaining unassigned candidate peaks giving an MS match = 82% for the peak with LRI2 = 2228. Finally Group IV represents a subset of the remaining unassigned components in M. alternifolia essential oil. While the identity of these components remains unknown it is still possible to locate the respective peak in the alternate GC × GC separation using a strategy similar to that described for Group III. These unknowns were cross-referenced in the two GC × GC chromatograms by using relative peak intensities as the primary criterion for narrowing the candidate list and matching the respective MS spectra from each of the multiplexed separations.
4. Conclusion A multiplexed dual-primary column 2GC × GC–MS system was developed that splits the primary flow into two first-dimension columns and recombines the two streams into a single seconddimension column, which ultimately ends at a single detector. The unique feature of this approach is that it is able to generate two LRIs and two second-dimension retention ordinates from a single injection, while using a single detector. The analysis of M. alternifolia essential oil shows the complementary selectivity of each column
190
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
with two different first-dimension column LRIs obtained for each analyte. Furthermore, the contra-directional modulation regime in 2GC × GC maintains information content of two individual GC × GC analyses, without the need for dual injection. Forty-nine components were identified using the non-polar × mid-polar column combination and 34 components were positively identified using the polar × mid-polar column combination. There were 22 unique compounds identified using the non-polar × mid-polar column combination and 7 unique compounds identified using the polar × mid-polar column combination. It is understandable that the former combination provides superior identification since non-polar retention indices are more reliable. Of these identified components, 27 peak assignments were corroborated by positive identification in both of the multiplexed separations. This aspect is the most important benefit of the multiplexed GC × GC–MS approach. This 2GC × GC–MS system with Aroma office 2D LRI library has proven to be very effective for reliable identification of M. alternifolia essential oil constituents. Acknowledgements This research was supported under Australian Research Council’s Discovery Projects funding scheme (project number DP110104923). Robert Shellie is the recipient of an Australian Research Council Australian Research Fellowship (project number DP110104923). Support from the University of Tasmania Central Science Laboratory is gratefully acknowledged and the authors thank Dr Jack Cochran (Restek) for valuable discussions. References [1] R. Shellie, P. Marriott, C. Cornwell, Characterization and comparison of tea tree and lavender oils by using comprehensive gas chromatography, J. High Resolut. Chromatogr. 23 (2000) 554–560. [2] R.A. Shellie, P.J. Marriott, Comprehensive two-dimensional gas chromatography-mass spectrometry analysis of Pelargonium graveolens essential oil using rapid scanning quadrupole mass spectrometry, Analyst 128 (2003) 879–883. [3] Z.L. Cardeal, M.D.R. Gomes da Silva, P.J. Marriott, Comprehensive twodimensional gas chromatography/mass spectrometric analysis of pepper volatiles, Rapid Commun. Mass Spectrom. 20 (2006) 2823–2836. [4] P.Q. Tranchida, R.A. Shellie, G. Purcaro, L.S. Conte, P. Dugo, G. Dugo, L. Mondello, Analysis of fresh and aged tea tree essential oils by using GC × GC-qMS, J. Chromatogr. Sci. 48 (2010) 262–266. [5] J. Omar, B. Alonso, M. Olivares, A. Vallejo, N. Etxebarria, Optimization of comprehensive two-dimensional gas chromatography (GC × GC) mass spectrometry for the determination of essential oils, Talanta 88 (2012) 145–151. [6] P. Marriott, R. Shellie, J. Fergeus, R. Ong, P. Morrison, High resolution essential oil analysis by using comprehensive gas chromatographic methodology, Flavour Fragr. J. 15 (2000) 225–239. [7] W.A. König, N. Bülow, Y. Saritas, Identification of sesquiterpene hydrocarbons by gas phase analytical methods, Flavour Fragr. J. 14 (1999) 367–378.
[8] C. Bicchi, A. Binello, A. D’Amato, P. Rubiolo, Reliability of Van den Dool retention indices in the analysis of essential oils, J. Chromatogr. Sci. 37 (1999) 288–294. [9] P. Sandra, C. Bicchi (Eds.), Chromatographic Methods. Capillary Gas Chromatography in Essential Oil Analysis, Huethig, Heidelberg/Basel/New York, 1987, pp. 275–286. [10] H. van den Dool, P.D. Kratz, A generalization of the retention index system including linear temperature programmed gas-liquid partition chromatography, J. Chromatogr. 11 (1963) 463–471. [11] R.J. Western, P.J. Marriott, Retention correlation maps in comprehensive twodimensional gas chromatography, J. Sep. Sci. 25 (2002) 832–838. [12] R.J. Western, P.J. Marriott, Methods for generating second dimension retention index data in comprehensive two-dimensional gas chromatography, J. Chromatogr. A 1019 (2003) 3–14. [13] J. Beens, R. Tijssen, J. Blomberg, Prediction of comprehensive two-dimensional gas chromatographic separations a theoretical and practical exercise, J. Chromatogr. A 822 (1998) 233–251. [14] J. Beens, R. Tijssen, J. Blomberg, Comprehensive two-dimensional gas chromatography (GC × GC) as a diagnostic tool, J. High Resolut. Chromatogr. 21 (1998) 63–64. [15] J.V. Seeley, F.J. Kramp, K.S. Sharpe, A dual-secondary column comprehensive two-dimensional gas chromatograph for the analysis of volatile organic compound mixtures, J. Sep. Sci. 24 (2001) 444–450. [16] J.V. Seeley, F.J. Kramp, K.S. Sharpe, S.K. Seeley, Characterization of gaseous mixtures of organic compounds with dual-secondary column comprehensive two-dimensional gas chromatography (GC × 2GC), J. Sep. Sci. 25 (2002) 53–59. [17] S. Bieri, P.J. Marriott, Generating multiple independent retention index data in dual-secondary column comprehensive two-dimensional gas chromatography, Anal. Chem. 78 (2006) 8089–8097. [18] B. Savareear, R.A. Shellie, Multiplexed dual second-dimension column comprehensive two-dimensional gas chromatography (GC × 2GC) using thermal modulation and contra-directional second-dimension columns, Anal. Chim. Acta 803 (2013) 160–165. [19] L. Blumberg, Theory of fast capillary gas chromatography: Part 3. Column performance vs. gas flow rate, J. High Resolut. Chromatogr. 22 (1999) 403–413. [20] N. Ochiai, K. Sasamoto, K. MacNamara, Characterization of sulphur compounds in whiskey by full evaporation dynamic headspace and selectable onedimensional/two-dimensional retention time locked gas chromatographymass spectrometry with simultaneous element-specific detection, J. Chromatogr. A 1270 (2012) 296–304. [21] Advanced Aroma Office 2D software the GERSTEL Selectable 1D/2D-GC/MS system, http://www.gerstelus.com/products subcat.php?id=55 [22] J.B. Phillips, J. Beens, Comprehensive two-dimensional gas chromatography: a hyphenated method with strong coupling between the two dimensions, J. Chromatogr. A 856 (1999) 331–347. [23] R.E. Murphy, M.R. Schure, J.P. Foley, Effect of sampling rate on resolution in comprehensive two-dimensional liquid chromatography, Anal. Chem. 70 (1998) 1585–1594. [24] W. Khummueng, J. Harynuk, P.J. Marriott, Modulation ration in comprehensive two-dimensional gas chromatography, Anal. Chem. 78 (2006) 4578–4587. [25] G.S. Frysinger, R.B. Gaines, Comprehensive two-dimensional gas chromatography with mass spectrometric detection (GC × GC/MS) applied to the analysis of petroleum, J. High Resolut. Chromatogr. 22 (1999) 251–255. [26] G.S. Frysinger, R.B. Gaines, C.S. Reddy, GC. × GC: a new analytical tool for environmental forensics, Environ. Forensics 3 (2002) 27–34. [27] R. Shellie, P. Marriott, G. Zappia, L. Mondello, G. Dugo, Interactive use of linear retention indices on polar and apolar columns with an MS-Library for reliable characterisation of Australian tea tree and other Melaleuca sp. oils, J. Essent. Oil Res. 15 (2003) 305–312. [28] L. Mondello, A. Casilli, P.Q. Tranchida, P. Dugo, G. Dugo, Comprehensive twodimensional GC for the analysis of citrus essential oil, Flavour Fragr. J. 20 (2005) 136–140.
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Multiplexed dual first-dimension comprehensive two-dimensional gas chromatography–mass spectrometry with contra-directional thermal modulation夽 Benjamin Savareear, Matthew R. Jacobs, Robert A. Shellie ∗ Australian Centre for Research on Separation Science (ACROSS), University of Tasmania, Private Bag 75, Hobart 7001, Australia
a r t i c l e
i n f o
Article history: Received 30 May 2014 Received in revised form 3 September 2014 Accepted 5 September 2014 Available online 16 September 2014 Keywords: Gas chromatography Comprehensive two-dimensional Dual first-dimension columns Multiplexed Essential oil
a b s t r a c t A multiplexed dual-primary column comprehensive two-dimensional gas chromatography–mass spectrometry approach (2GC × GC–MS) is introduced. The approach splits injected samples into two first-dimension columns with different stationary phases, and recombines the two streams into one second-dimension column that terminates at a single detector. The approach produces two twodimensional chromatograms for each injection, and is made possible by using a dual-stage modulator operated in contra-directional modulation mode. The dual two-dimensional chromatograms produced by this single detector system provide complementary information due to selectivity differences between the three separation columns used in the column ensemble. An aged Australian tea tree (Melaleuca alternifolia) essential oil was analyzed to demonstrate the 2GC × GC–MS approach. The number of compounds separated by each of the GC × GC separations in the 2GC × GC experiment is comparable to conventional GC × GC experiments with matching column configurations. Robust peak assignment was possible for this sample based on the combination of MS library matches and multiple linear retention index searching. Forty-nine components (22 unique) were identified using a non-polar × mid-polar column combination and 34 components (7 unique) were positively identified using a polar × mid-polar column combination. Twenty-seven peak assignments were corroborated by positive identification in both of the multiplexed separations. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Comprehensive two-dimensional gas chromatography coupled with mass spectrometry (GC × GC–MS) is a highly suitable technology for essential oil characterization [1–6]. Since many essential oil constituents have similar mass spectral features [7] it is normal practice to obtain retention indices of compounds from one or more different stationary phases to facilitate correct identification of separated compounds [8]. To date many essential oil analysts have utilized dual channel GC [9] to obtain multiple linear retention indices (LRI) on different stationary phases in a single run [10]. The benefit of multiple LRI was very well-illustrated by Bicchi et al. who showed that the percentage of correct identifications obtainable through retention indices alone is approximately 65% with one stationary
夽 Presented at 38th International Symposium on Capillary Chromatography and 11th GC × GC Symposium, 18–23 May 2014, Riva del Garda, Italy. ∗ Corresponding author. Tel.: +61 3 6226 7656; fax: +61 3 6226 2858. E-mail address: [email protected] (R.A. Shellie). http://dx.doi.org/10.1016/j.chroma.2014.09.014 0021-9673/© 2014 Elsevier B.V. All rights reserved.
phase, 80% with two different-stationary phases and more than 90% if three different stationary phase columns are employed [8]. GC × GC–MS uses different stationary phases in the firstand second-dimension; however, use of the second-dimension retention time is not well established for compound identification. Although some studies have proposed different methods for calculating retention indices in the second-dimension [11–14], they have not been widely adopted due to their complexity. GC × GC–MS analysis typically depends on the first-dimension retention time and mass spectral data for compound identification [1–6]. Easy access to a second reliable LRI for separated compounds would be highly beneficial for GC × GC–MS separations. Drawing inspiration from previous GC × 2GC investigations [15–17], which use two different second-dimension separation columns and two detectors, we recently introduced a multiplexed, dual second-dimension column GC × 2GC approach that recombines effluent from the two second-dimension columns before analyte detection, eliminating the need for two detectors. The present investigation builds upon the foundation laid in our earlier work using the principle of contra-directional modulation [18]. A novel
184
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
Table 1 Advanced modulation parameters for contra-directional modulation. GC × 2GC First stage Second stage 2GC × GC First stage Second stage
Cool time till off 3000 ms Cool time till on 3000 ms
Cool off time 2500 ms Cool on time 1000 ms
Heat time till on 2500 ms Heat time till off 1000 ms
Heat on time 3500 ms Heat off time 5000 ms
Cool time till off 1500 ms Cool time till on 1500 ms
Cool off time 1200 ms Cool on time 1200 ms
Heat time till on 1200 ms Heat time till off 1200 ms
Heat on time 1800 ms Heat off time 1800 ms
approach for multiplexed dual first-dimension GC × GC (2GC × GC), providing dual LRI, coupled with MS is introduced. Qualitative analysis of Melaleuca alternifolia essential oil (Australian tea tree oil) is used to demonstrate the utility of the system developed herein. 2GC × GC–MS analysis with MS database matching and multiple LRI searching provides robust peak assignment based on three orthogonal parameters: MS, LRI1 , and LRI2 . Retention index based searches are performed on LRI data obtained from either of the first-dimension columns, or both of the firstdimension columns by cross searching. Additional supporting information upon which peak identity can be made is based on peak coordinates within the structured chromatograms obtained from the two GC × GC separations. 2. Materials and methods 2.1. Chemicals and reagents Australian tea tree (M. alternifolia) essential oil was obtained from a local supermarket (Sandy Bay, Tasmania, Australia) and stored at room temperature for an extended period (beyond the manufacturer’s recommended use-by date; SEPT/2011). A C7 –C30 n-paraffin hydrocarbon mixture (1000 g/mL; Sigma–Aldrich, Castle Hill, Australia) was used for determination of first dimension LRI. Both the Australian tea tree essential oil and alkane mixture were diluted (1:10, v/v) in dichloromethane (Sigma–Aldrich) prior to GC analyses. 2.2. Instrumentation and experimental conditions All analyses were performed using a Leco GC × GC-FID instrument with an LN2 Cooled Thermal Modulator (Leco Australia, Castle Hill, Australia). The chromatograph was equipped with split/splitless injector, operated with a 20:1 split ratio and inlet temperature of 230 ◦ C. Injected sample volume was 1 L. Hydrogen carrier gas was generated using a Parker Balston H2PEM-260 hydrogen generator. For convenience of contra-directional column installation and operation, the auxiliary (second-dimension) column oven was removed from the GC × GC system. All columns were heated using the main GC oven in all analysis. The modulator offset temperature was +15 ◦ C. Effluent from both secondary columns was monitored by a single FID operated at 100 Hz and 250 ◦ C. All GC × GC-FID results were collected and processed using Leco ChromaTOF software.
polyethylene glycol (Stabilwax; Restek); 2 D2 0.4 m × 100 m i.d. fused silica capillary coated with 0.1 m 50% phenyl, 50% dimethyl polysiloxane (Rxi-17SilMS; Restek). Total carrier gas flow rate was 2.0 mL min−1 (Hydrogen; constant flow). Each of the seconddimension columns operates at efficiency-optimized flow [19] using this flow regime, while the first-dimension column is between efficiency- and speed-optimized flow. Total modulation period was 6.0 s to maintain appropriate modulation ratio. Custom modulation parameters are provided in Table 1. The oven temperature program was 40 ◦ C (0.2 min hold) ramped at 6 ◦ C min−1 to 180 ◦ C. The first-dimension column was connected to the two second-dimension columns using a 3-port SilFlow connector (SGE Analytical Science, Ringwood, Australia). Flow from the second-dimension columns was recombined using a 3-port SilFlow connector into a 0.2 m × 220 m i.d. deactivated uncoated fused silica transfer line that terminated at the FID. The second dimension columns were installed contra-directionally in the GC × GC modulator. 2.4. 2GC × GC-FID Multiplexed 2GC × GC is achieved by using contra-directional modulation, where two parallel first-dimension columns are installed contra-directionally in the GC × GC modulator as shown in Fig. 1. 2GC × GC analyses were performed by splitting the flow from the inlet into two first-dimension columns by means of a twinhole graphite ferrule (SGE Analytical Science), all other column connections were made using press tight connectors (Restek). The first dimension columns were installed contra-directionally in the GC × GC modulator. Two first dimension columns were employed in this study; a 30 m × 250 m i.d. fused silica capillary coated with 0.25 m dimethyl polysiloxane (DB-1; Agilent Technologies, Mulgrave, Australia); and a 30 m × 250 m i.d. fused silica capillary coated with 0.25 m polyethylene glycol (DB-WAX; Agilent Technologies). The second-dimension column was a 0.45 m × 150 m i.d. fused silica capillary coated with 50% phenyl, 50% dimethyl polysiloxane (Rxi-17SilMS; Restek). Total carrier gas flow rate was
2.3. GC × 2GC-FID GC × 2GC analyses were performed by using the general instrument configuration described in reference [18]. The first dimension column 1 D was a 30 m × 250 m i.d. fused silica capillary coated with 0.25 m dimethyl polysiloxane (Rtx-1; Restek, Bellefonte, PA, USA). The two second-dimension columns were: 2 D 0.4 m × 100 m i.d. fused silica capillary coated with 0.1 m 1
Fig. 1. Illustration of the column configuration used for multiplexed 2GC × GC. (A) first-dimension Column 1: polyethylene glycol; (B) first-dimension Column 2: dimethyl polysiloxane; (C) second-dimension column: 50% phenyl, 50% dimethyl polysiloxane; (D) dual stage thermal modulator.
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
185
1.5 mL min−1 (Hydrogen; constant flow), which provides speedoptimized flow in the second column. Total modulation period was 3.0 s to maintain appropriate modulation ratio, and the oven temperature program was 40 ◦ C (0.2 min hold) ramped at 7 ◦ C min−1 to 220 ◦ C (3 min hold). Custom modulation parameters are provided in Table 1.
than 100,000 entries, containing LRI information for a wide range of aroma compounds from many literature references [20,21].
2.5. GC × GC-FID
In the present investigation, two long primary columns are installed as parallel first-dimension columns. This column configuration mirrors a previous GC × 2GC study that used two short second-dimension columns. While column installation is straightforward, a key practical consideration for contra-directional modulation approach relates to column alignment within the modulator. Proper column alignment is critical to ensure that neither of the two column segments is allowed to block the jet of cool gas onto the other column. Fig. 2 compares the multiplexed comprehensive two-dimensional chromatograms obtained from the analysis of M. alternifolia essential oil using GC × 2GC and 2GC × GC. There are advantages and disadvantages of each of these multiplexed approaches. The first advantage of GC × 2GC is that peaks in both chromatograms have matching first-dimension retention times. This makes the distribution of peaks in GC × 2GC more intuitive than the distribution observed in 2GC × GC, but this is also disadvantageous because GC × 2GC only provides a single firstdimension LRI. Having two dissimilar second-dimension columns also creates optimization challenges to avoid wrap-around. A single second-dimension column greatly alleviates this challenge. Next, it is important to use a non-polar stationary phase column in the first-dimension for GC × 2GC experiments, since retention indices are more rugged on non-polar stationary phases. However the requirement to use a non-polar stationary phase first-dimension column means that second-dimension columns can only be more polar than the first-dimension column. As the purpose of the dual second-dimension separations is to resolve any co-eluted compounds, this restriction limits peak spreading throughout the two-dimensional separation space. Indeed the individual GC × GC chromatograms produced by the GC × 2GC approach, shown in Fig. 2A exhibit more similarities than differences. A column ensemble comprising one long first-dimension column and two short second-dimension columns in GC × 2GC is not optimal for fully harnessing the complementary selectivity of the three stationary phases. Conversely, using two long first-dimension columns and one short second-dimension column provides a demonstrable
GC × GC analyses were performed using the columns as 2GC × GC experiments, except they were installed in the regular configuration through the GC × GC modulator and all column connections were made using press-tight connectors (Restek). The flow rate was 1.0 mL min−1 for all experiments. A modulation period of 3.0 s (with default modulation parameters) was employed and the temperature program was 40 ◦ C (0.2 min hold) ramped at 5.8 ◦ C min−1 to 220 ◦ C (3 min hold). 2.6. 2GC × GC–MS 2GC × GC–MS analysis was performed using the Leco GC × GC system described above, which was connected via an externally controlled heated transfer line, to the heated MS transfer line of an Agilent 5975C VL MSD (Agilent Technologies). The external heated transfer line was a 155 mm Agilent LTM transfer line, which was controlled with an LTM Controller (Agilent Technologies). The external heated transfer line and heated MS transfer line temperatures were 200 ◦ C. The MS detector was operated with a source temperature of 150 ◦ C, quadrupole temperature 230 ◦ C, with a detector voltage of 70 eV. Chemstation software (Agilent Technologies) was used to acquire MS data in the fast scanning mode with a reduced mass scan range of 40–220 m/z, yielding a detector sampling rate of 25.95 Hz. Data acquisition was triggered by a start-out signal from a remote port provided by the Leco GC × GC modulator. GC columns and other operating conditions were identical to those described above for the 2GC × GC-FID experiments. MS data was visualized using Transform (Fortner Research, Boulder, CO, USA). Mass spectra were matched to the Terpene Library (containing mass spectra of essential oil components compiled by Robert P. Adams, Baylor University Plant Biotechnology Center). Further analysis of 2GC × GC–MS results using interactive LRI filters was performed using Aroma Office 2D Software (Gerstel K.K., Tokyo, Japan). This software is a searchable database with more
3. Results and discussion 3.1. Implementation and benefits of 2GC × GC system
Fig. 2. Multiplexed two-dimensional separation space for the separation of M. alternifolia essential oil. (A) Separations obtained using GC × 2GC, (B) separations were obtained using 2GC × GC.
186
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
Fig. 3. Two-dimensional separation space for GC × GC and 2GC × GC analyses of M. alternifolia essential oil. (A) Separation obtained using GC × GC (non-polar × mid-polar), (B) separations obtained using 2GC × GC, and (C) separation obtained using GC × GC (polar × mid-polar).
improvement over GC × 2GC in utilizing the available separation space. Retention of any given solute in the second-dimension of temperature-programmed GC × GC analysis is dependent upon the solute’s interaction with the first- and second-dimension stationary phases [22]. Having two first-dimension columns in the 2GC × GC experiment is valuable since they make different contributions to retention in the second-dimension column. In the top half of Fig. 2B (dimethyl polysiloxane × 50% phenyl, 50% dimethyl polysiloxane column combination; hereafter referred to as non-polar × mid-polar) analytes are strongly retained in the second-dimension column due to strong solute-stationary phase interactions with the second-dimension column stationary phase. The separation in the bottom half of Fig. 2B, which represents a column combination of polyethylene glycol × 50% phenyl, 50% dimethyl polysiloxane (hereafter referred to polar × mid-polar), demonstrates how a stronger interaction with the first-dimension column stationary phase decreases retention in the second-dimension. The interaction of ␦elemene (peak 32) and cyclosativene (peak 37) illustrates this effect in the two chromatograms shown in Fig. 2B. Although GC × 2GC experiments previously described in the literature demonstrate beneficial class discrimination for homologous series [15–18], 2GC × GC is more appropriate for analysis of genuinely heterogeneous multicomponent samples. One may postulate that greater utilization of the two-dimensional separation space (compared to Fig. 2A) would be achieved by using a mid-polar × non-polar plus mid-polar × high-polar GC × 2GC combination, but the benefit of recording reliable LRIs would be diminished. Fig. 3 compares the chromatograms obtained from 2GC × GC and conventional GC × GC analyses of M. alternifolia essential oil. The two GC × GC analyses were performed using the same columns as the 2GC × GC experiment. The columns were not multiplexed in these experiments; they were installed in a conventional GC × GC configuration. It is evident that GC × GC chromatogram integrity is preserved in the 2GC × GC experiment. The corresponding chromatograms exhibit striking resemblance and the number of peaks separated in each GC × GC analysis is similar to the corresponding portion of the 2GC × GC analysis. Importantly, the 2GC × GC experiment maintains an appropriate number of modulation slices in line with accepted guidelines. This ensures sufficient peak sampling to minimize modulation-induced loss of
first-dimension resolution [23]. By employing a total modulation period of 3 s the modulation ratio is at least 2 in the 2GC × GC experiments, which is suitable for semi-quantitative screening purposes according to the recommendations of Khummueng et al. [24]. By nature of the contra-directional modulation approach, the second-dimension separation needs to be twice as fast the normal requirement for GC × GC. A fast second-dimension separation is especially important in 2GC × GC to avoid wrap-around, which would cause the two separation windows in the 2GC × GC chromatogram to overlap. Wrap-around is quite easily avoided by using short second-dimension columns and applying speed-optimized flow (for the second-dimension column). Fine-tuning the separation space can be achieved by manipulating the initial temperature and temperature ramp(s). A 1.5 s separation window is sufficient for the sample considered in the present investigation. Unconverted GC × GC chromatograms of 1,8-cineole (peak 7) obtained from the 2GC × GC experiment are shown in Fig. 4. The figure shows that each of the multiplexed GC × GC separations produces three modulated pulses above baseline for 1,8-cineole. Measured peak
Fig. 4. 3 s modulated peak pulses in phase obtained from 2GC × GC for 1,8-cineole (peak 7). (A) non-polar × mid-polar, (B) polar × mid-polar.
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
187
Fig. 5. Peak apex plots showing coordinates of peak maxima from 2GC × GC–MS analysis of M. alternifolia essential oil. (A) non-polar × mid-polar, (B) polar × mid-polar. Peak numbers correspond to those in Tables 2 and 3.
188
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
Table 2 Peak assignments of 27 components for which all search criteria were satisfied. Numbers in the table correspond to those in Fig. 5. MS library match quality refers to comparison with the Terpene Library. Peak no
4 6 7 17 18 23 32 35 37 39 41 43 44 47 49 50 51 52 53 54 55 56 57 58 59 60 61
Peak assignment
Myrcene p-Cymene 1,8-Cineole Terpinen-4-ol ␣-Terpineol cis-Carveol ␦-Elemene ␣-Copaene Cyclosativene ␣-Gurjunene trans-␣-Bergamotene -Caryophyllene Aromadendrene Allo-aromadendrene ␥-Muurolene Valencene Bicyclogermacrene ␣-Muurolene trans-Calamenene Spathulenol trans-Nerolidol Caryophyllene oxide Globulol Cedrol Guaiol Elemol -Cadinol
Column Set 1
Column Set 2
% match MS1
LRI1obs
2
tR(1)
LRI1lit
(LRIobs − LRIlit )(1)
% match MS2
LRI2obs
2
95 96 94 82 93 82 86 85 93 93 84 95 88 92 84 89 82 82 95 84 82 88 93 90 86 90 80
983 1014 1021 1165 1178 1200 1333 1353 1366 1406 1423 1434 1443 1454 1476 1489 1494 1499 1511 1541 1544 1551 1563 1572 1573 1591 1613
1.35 1.39 1.38 1.51 1.51 1.5 1.62 1.77 1.31 1.38 1.66 1.42 1.66 1.42 1.38 1.42 1.66 1.46 1.38 1.46 1.73 1.42 1.51 1.51 1.46 1.47 1.46
983 1014 1021 1164 1177 1202 1338 1357 1372 1411 1428 1438 1448 1458 1480 1493 1498 1504 1518 1547 1549 1557 1568 1577 1575 1595 1617
0 0 0 1 1 −2 −5 −4 −6 −5 −5 −4 −5 −4 −4 −4 −4 −5 −7 −6 −5 −6 −5 −5 −2 −4 −4
93 96 95 88 95 81 91 87 89 90 86 93 87 91 85 88 84 89 90 86 82 89 88 84 87 84 82
1164 1278 1225 1638 1731 1813 1494 1544 1485 1522 1594 1606 1609 1650 1691 1717 1755 1721 1840 2165 1974 2027 2038 2004 2115 2088 2134
0.15 0.31 0.34 0.34 0.19 0.07 0.19 0.15 0.61 0.65 0.12 0.57 0.1 0.46 0.54 0.46 0.57 0.5 0.38 0.19 0.12 0.27 0.19 0.12 0.19 0.19 0.23
areas are 718 ± 8 and 581 ± 7 (n = 4) for 1,8-cineole peaks in the non-polar × mid-polar and polar × mid-polar 2GC × GC separations respectively. Although these peak area values are statistically significantly different (p < 10−6 ) due to a slightly uneven split between the two first-dimension columns, no further effort was made to equalize the split ratio in the present study, since the aim was to demonstrate qualitative performance benefits. Previous work discussed use of normalized peak areas for semi-quantitative analysis in GC × 2GC experiments [18] and this is also likely to be beneficial for the current configuration. 3.2. Coupling of 2GC × GC system with mass spectrometry (2GC × GC–MS) Although time of flight mass spectrometry (TOF-MS) is the mass selective detector of choice for GC × GC–MS, quadrupole MS instruments have been successfully used for this purpose [2,4,25,26], usually by reducing the scan mass range to obtain a reasonable data acquisition rate [2]. The present investigation applied a reduced mass scan range from 40 to 220 m/z with fast scanning mode providing a data acquisition rate of ca. 26 Hz. This data acquisition rate is adequate for proof-of-principle qualitative analysis, although a faster instrument would be required to realize the full potential of 2GC × GC–MS analysis for quali-quantitative analysis (identification and amount). A 2GC × GC–MS peak apex plot of M. alternifolia essential oil along with peak numbers in both GC × GC separations is provided in Fig. 5. The 2GC × GC–MS chromatogram is provided in the supporting information (see Fig. S1). Supplementary Fig. S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma.2014.09.014. Primary identification of separated compounds was carried out by MS1 library searching peaks in one of the multiplexed GC × GC separations, using a minimum similarity index threshold of ≥80%. Secondary confirmation of MS1 based peak identification was achieved by comparing the first-dimension LRI1 of each peak with literature LRI using AROMA Office 2D software. The threshold
tR(2)
LRI2lit
(LRIobs − LRIlit )(2)
1160 1271 1212 1628 1729 1834 1473 1523 1481 1531 1576 1601 1605 1650 1691 1726 1740 1725 1844 2147 1959 2009 2060 2046 2096 2083 2167
4 7 13 10 2 −21 21 21 4 −9 18 5 4 0 0 −9 15 −4 −4 18 15 18 −22 −42 19 5 −33
for permitted LRI1 shift was ±10 for the dimethyl polysiloxane stationary phase column. The next level of confirmation was achieved by using LRI2 to locate a peak in the alternate multiplexed GC × GC separation. To this end, any peak name that remained after MS1 library search and LRI1 filter was input (name-search) in Aroma Office 2D. Name-search responses were limited to the polyethylene glycol stationary phase. Here, a full list of LRI2 from the database is reported for entries matching that compound name. The mean LRI2 reported by Aroma Office 2D is used to perform a reverseLRI2 search on the alternate first-dimension column, using LRI2 of ±50. This reverse LRI2 search narrows the search area in the twodimensional separation space for MS2 library searching. Additional information, such as relative peak intensity and second-dimension relative retention within the structured two-dimensional separation space is used to assist peak selection for MS2 library searching. Final confirmation is performed by MS2 library searching using the same similarity threshold as the MS1 library search. A peak name is only added to the list of identified components if the thresholds for MS1 search, LRI1 filter, LRI2 search, and MS2 search are satisfied. Table 2 provides details of 27 components identified in M. alternifolia essential oil, using the four peak assignment criteria. The identified components are also shown in a peak-apex plot (Fig. 5) to permit cross-referencing with the 2GC × GC chromatogram (Fig. 3). Table 3 provides a list of 35 M. alternifolia components that failed at least one identification criterion above. Despite failing one or more criteria, many of these components were identified using alternative strategies. These components are grouped according to the strategy used for peak assignment. Group I comprises 19 components whose identity was indicated by MS1 library searching and confirmed by LRI1 . Similarly, Group II contains four components that were identified by MS2 library searching and confirmed by LRI2 . The components assigned to Groups I and II were not found in the alternate GC × GC separation within the corresponding match thresholds. Broadening the LRI threshold did not lead to further confirmatory results. Limited further confirmation of peak identity of these 35 components is largely due to peak co-elution, which
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
189
Table 3 Peak assignments of 35 components for which at least one search criterion was not satisfied. Numbers in the table correspond to those in Fig. 5. MS library match quality refers to comparison with the Terpene Library. Peak no.
Group I 1 2 3 8 9 10 11 12 13 14 15 22 31 33 36 38 40 48 62 Group II 5 16 21 46 Group III 26 28 29 Group IV 19 20 24 25 27 30 34 42 45
Peak assignment
␣-Thujene ␣-Pinene Sabinene trans--Ocimene ␥-Terpinene Terpinolene Linalool cis-Sabinene hydrate Allo-ocimene Neo-allo-ocimene cis-Sabinol Myrtenol trans-Pinocarvyl acetate ␣-Cubebene ␣-Ylangene -Cubebene -Gurjunene Virdiflorene ␣-Eudesmol Limonene p-Cymen-8-ol trans-Piperitol ␣-Humulene cis-Verbenyl acetate iso-Bornyl acetate Myrtenyl acetate 19 20 24 25 27 30 34 42 45
Column Set 1
Column Set 2
% match MS1
LRI1obs
2
tR(1)
LRI1lit
(LRIobs − LRIlit )(1)
96 99 98 83 83 89 91 87 89 86 86 86 81 87 89 93 85 86 81
923 930 970 1032 1052 1071 1083 1083 1105 1118 1148 1190 1293 1342 1358 1387 1420 1468 1634
1.27 1.31 1.35 1.58 1.38 1.38 1.89 1.43 1.42 1.43 1.43 1.43 1.63 1.66 1.69 1.73 1.42 1.38 1.54
923 930 970 1032 1051 1070 1082 1082 1106 1117 1147 1191 1296 1348 1363 1392 1424 1473 1636
0 0 0 0 1 1 1 1 −1 1 1 −1 −3 −6 −5 −5 −4 −5 −2
– – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – –
– – – –
– – – –
– – – –
1014 1163 1204 1449
– – – –
96 93 82 85
84 87 83
1262 1275 1285
1.58 1.58 1.54
1264 1278 1290
2 3 5
– – – – – – – – –
1181 1181 1233 1244 1264 1291 1345 1431 1446
1.58 1.46 1.58 1.69 1.69 1.54 1.58 1.42 1.93
– – – – – – – – –
– – – – – – – – –
leads to poor MS library search results. For instance, limonene (peak 5) was positively identified using the polyethylene glycol stationary phase first-dimension separation using MS2 and LRI2 . The Aroma Office 2D software indicates a LRI1 of 1014 for the dimethyl polysiloxane column. Thus this peak must be eluted between pcymene (peak 6 LRI1 1014) and 1,8-cineole (peak 7 LRI1 1021) in the chromatogram, but it is obscured in the separation. This problematic peak triplet has been previously reported in the same essential oil using one-dimensional GC–MS analysis [27]. Like Group I components, identity of the three components assigned to Group III was indicated by MS1 library searching and confirmed by LRI1 . None of these components were located within the allowable LRI2 windows in the polyethylene glycol stationary phase first-dimension separation space. However, in this case the widely discussed class-separation leads to further corroboration of peak assignment. Class/structure association is often not readily apparent in essential oil analyses due to the heterogeneity of the samples, but polar first-dimension columns lead to some useful class information [28]. For example, the terpene esters that are shown as crosses in Fig. 5 were not found within expected LRI2 in the polyethylene glycol separation space. Utilizing the information provided by the structured separation space to indicate the compound class of unknown peaks, and using their relative peak intensities as a further indicator, narrows the candidate list of unassigned peaks for further MS2 library searching. For example, peak assignment of cis-verbenyl acetate (peak 26) was achieved
% match MS2
LRI2obs
2
tR(2)
LRI2lit
(LRIobs − LRIlit )(2)
– – – – – – – – – – – – – – – – – – –
1021 1022 1119 1250 1253 1278 1546 1515 1380 1421 1731 1794 1764 1466 1483 1533 1697 2191
– – – – – – – – – – – – – – – – – – –
1200 1887 1779 1663
0.27 0.07 0.07 0.07
1199 1857 1747 1663
−1 −30 −32 0
82 84 81
2228 2248 2131
0.07 0.07 0.07
– 1578 1690
– 670 441
– – – – – – – – –
1813 1759 2240 2162 2392 2177 2350 1900 2795
0.15 0.23 0.07 0.07 0.07 0.07 0.07 0.39 0.12
– – – – – – – – –
– – – – – – – – –
using this strategy. Absolute peak areas of these peaks are 4.3 × 108 and 4.2 × 108 arbitrary units respectively (2GC × GC-FID data). Peak assignment is justified by MS1 match = 84% and LRI1 = 1262, and MS2 library search upon the remaining unassigned candidate peaks giving an MS match = 82% for the peak with LRI2 = 2228. Finally Group IV represents a subset of the remaining unassigned components in M. alternifolia essential oil. While the identity of these components remains unknown it is still possible to locate the respective peak in the alternate GC × GC separation using a strategy similar to that described for Group III. These unknowns were cross-referenced in the two GC × GC chromatograms by using relative peak intensities as the primary criterion for narrowing the candidate list and matching the respective MS spectra from each of the multiplexed separations.
4. Conclusion A multiplexed dual-primary column 2GC × GC–MS system was developed that splits the primary flow into two first-dimension columns and recombines the two streams into a single seconddimension column, which ultimately ends at a single detector. The unique feature of this approach is that it is able to generate two LRIs and two second-dimension retention ordinates from a single injection, while using a single detector. The analysis of M. alternifolia essential oil shows the complementary selectivity of each column
190
B. Savareear et al. / J. Chromatogr. A 1365 (2014) 183–190
with two different first-dimension column LRIs obtained for each analyte. Furthermore, the contra-directional modulation regime in 2GC × GC maintains information content of two individual GC × GC analyses, without the need for dual injection. Forty-nine components were identified using the non-polar × mid-polar column combination and 34 components were positively identified using the polar × mid-polar column combination. There were 22 unique compounds identified using the non-polar × mid-polar column combination and 7 unique compounds identified using the polar × mid-polar column combination. It is understandable that the former combination provides superior identification since non-polar retention indices are more reliable. Of these identified components, 27 peak assignments were corroborated by positive identification in both of the multiplexed separations. This aspect is the most important benefit of the multiplexed GC × GC–MS approach. This 2GC × GC–MS system with Aroma office 2D LRI library has proven to be very effective for reliable identification of M. alternifolia essential oil constituents. Acknowledgements This research was supported under Australian Research Council’s Discovery Projects funding scheme (project number DP110104923). Robert Shellie is the recipient of an Australian Research Council Australian Research Fellowship (project number DP110104923). Support from the University of Tasmania Central Science Laboratory is gratefully acknowledged and the authors thank Dr Jack Cochran (Restek) for valuable discussions. References [1] R. Shellie, P. Marriott, C. Cornwell, Characterization and comparison of tea tree and lavender oils by using comprehensive gas chromatography, J. High Resolut. Chromatogr. 23 (2000) 554–560. [2] R.A. Shellie, P.J. Marriott, Comprehensive two-dimensional gas chromatography-mass spectrometry analysis of Pelargonium graveolens essential oil using rapid scanning quadrupole mass spectrometry, Analyst 128 (2003) 879–883. [3] Z.L. Cardeal, M.D.R. Gomes da Silva, P.J. Marriott, Comprehensive twodimensional gas chromatography/mass spectrometric analysis of pepper volatiles, Rapid Commun. Mass Spectrom. 20 (2006) 2823–2836. [4] P.Q. Tranchida, R.A. Shellie, G. Purcaro, L.S. Conte, P. Dugo, G. Dugo, L. Mondello, Analysis of fresh and aged tea tree essential oils by using GC × GC-qMS, J. Chromatogr. Sci. 48 (2010) 262–266. [5] J. Omar, B. Alonso, M. Olivares, A. Vallejo, N. Etxebarria, Optimization of comprehensive two-dimensional gas chromatography (GC × GC) mass spectrometry for the determination of essential oils, Talanta 88 (2012) 145–151. [6] P. Marriott, R. Shellie, J. Fergeus, R. Ong, P. Morrison, High resolution essential oil analysis by using comprehensive gas chromatographic methodology, Flavour Fragr. J. 15 (2000) 225–239. [7] W.A. König, N. Bülow, Y. Saritas, Identification of sesquiterpene hydrocarbons by gas phase analytical methods, Flavour Fragr. J. 14 (1999) 367–378.
[8] C. Bicchi, A. Binello, A. D’Amato, P. Rubiolo, Reliability of Van den Dool retention indices in the analysis of essential oils, J. Chromatogr. Sci. 37 (1999) 288–294. [9] P. Sandra, C. Bicchi (Eds.), Chromatographic Methods. Capillary Gas Chromatography in Essential Oil Analysis, Huethig, Heidelberg/Basel/New York, 1987, pp. 275–286. [10] H. van den Dool, P.D. Kratz, A generalization of the retention index system including linear temperature programmed gas-liquid partition chromatography, J. Chromatogr. 11 (1963) 463–471. [11] R.J. Western, P.J. Marriott, Retention correlation maps in comprehensive twodimensional gas chromatography, J. Sep. Sci. 25 (2002) 832–838. [12] R.J. Western, P.J. Marriott, Methods for generating second dimension retention index data in comprehensive two-dimensional gas chromatography, J. Chromatogr. A 1019 (2003) 3–14. [13] J. Beens, R. Tijssen, J. Blomberg, Prediction of comprehensive two-dimensional gas chromatographic separations a theoretical and practical exercise, J. Chromatogr. A 822 (1998) 233–251. [14] J. Beens, R. Tijssen, J. Blomberg, Comprehensive two-dimensional gas chromatography (GC × GC) as a diagnostic tool, J. High Resolut. Chromatogr. 21 (1998) 63–64. [15] J.V. Seeley, F.J. Kramp, K.S. Sharpe, A dual-secondary column comprehensive two-dimensional gas chromatograph for the analysis of volatile organic compound mixtures, J. Sep. Sci. 24 (2001) 444–450. [16] J.V. Seeley, F.J. Kramp, K.S. Sharpe, S.K. Seeley, Characterization of gaseous mixtures of organic compounds with dual-secondary column comprehensive two-dimensional gas chromatography (GC × 2GC), J. Sep. Sci. 25 (2002) 53–59. [17] S. Bieri, P.J. Marriott, Generating multiple independent retention index data in dual-secondary column comprehensive two-dimensional gas chromatography, Anal. Chem. 78 (2006) 8089–8097. [18] B. Savareear, R.A. Shellie, Multiplexed dual second-dimension column comprehensive two-dimensional gas chromatography (GC × 2GC) using thermal modulation and contra-directional second-dimension columns, Anal. Chim. Acta 803 (2013) 160–165. [19] L. Blumberg, Theory of fast capillary gas chromatography: Part 3. Column performance vs. gas flow rate, J. High Resolut. Chromatogr. 22 (1999) 403–413. [20] N. Ochiai, K. Sasamoto, K. MacNamara, Characterization of sulphur compounds in whiskey by full evaporation dynamic headspace and selectable onedimensional/two-dimensional retention time locked gas chromatographymass spectrometry with simultaneous element-specific detection, J. Chromatogr. A 1270 (2012) 296–304. [21] Advanced Aroma Office 2D software the GERSTEL Selectable 1D/2D-GC/MS system, http://www.gerstelus.com/products subcat.php?id=55 [22] J.B. Phillips, J. Beens, Comprehensive two-dimensional gas chromatography: a hyphenated method with strong coupling between the two dimensions, J. Chromatogr. A 856 (1999) 331–347. [23] R.E. Murphy, M.R. Schure, J.P. Foley, Effect of sampling rate on resolution in comprehensive two-dimensional liquid chromatography, Anal. Chem. 70 (1998) 1585–1594. [24] W. Khummueng, J. Harynuk, P.J. Marriott, Modulation ration in comprehensive two-dimensional gas chromatography, Anal. Chem. 78 (2006) 4578–4587. [25] G.S. Frysinger, R.B. Gaines, Comprehensive two-dimensional gas chromatography with mass spectrometric detection (GC × GC/MS) applied to the analysis of petroleum, J. High Resolut. Chromatogr. 22 (1999) 251–255. [26] G.S. Frysinger, R.B. Gaines, C.S. Reddy, GC. × GC: a new analytical tool for environmental forensics, Environ. Forensics 3 (2002) 27–34. [27] R. Shellie, P. Marriott, G. Zappia, L. Mondello, G. Dugo, Interactive use of linear retention indices on polar and apolar columns with an MS-Library for reliable characterisation of Australian tea tree and other Melaleuca sp. oils, J. Essent. Oil Res. 15 (2003) 305–312. [28] L. Mondello, A. Casilli, P.Q. Tranchida, P. Dugo, G. Dugo, Comprehensive twodimensional GC for the analysis of citrus essential oil, Flavour Fragr. J. 20 (2005) 136–140.
