25 Type Analysis of Citrus Essential Oils by Multidimensional Supercritical Fluid Chromatography/ Gas Chromatography Takashi YARITA, Akira NoMURA and Yoshiyuki HoRIMOTo Department of Analytical Chemistry, National Institute of Materials and Chemical Research, Higashi, Tsukuba, Ibaraki 305, Japan A multidimensional supercritical fluid chromatography (SFC)/ gas chromatography (GC) system was developed for the characterization of citrus essential oils. For on-line coupling of SFC with GC, the carbon dioxide mobile phase used in the SFC was split after a UV detector and then introduced into a GC injector. The elution order in the SFC on a silicagel column was based on the polarities of the solutes. First, citrus essential oils were separated into three fractions (hydrocarbons, aldehydes and esters, and alcohols) by SFC utilizing stepwise pressure programming. Then, each fraction was introduced into the GC and determined using a flame ionization detector for the quantification of three fractions as well as their components in the oils. The SFC/ GC system was applied to the characterization of citrus essential oils. All of the citrus essential oils used in this study mainly consisted of hydrocarbons. Keywords Multidimensional chromatography, supercritical fluid chromatography, gas chromatography, type analysis, citrus essential oil Citrus essential oils are used in the perfume and food industries. In general, these oils comprise a large number of components, most of which are terpene hydrocarbons and their derivatives, such as esters, aldehydes and alcohols. Capillary gas chromatography (GC) is useful for the determination of the oil components.'-3 However, it is necessary that a mass spectrometer, for example, be used as a detector of GC for obtaining structural information concerning the components. Multidimensional high-performance liquid chromatography (HPLC)/ GC, in which an HPLC is on-line coupled with a GC, has been recognized as being a separation technique whereby much information concerning the samples can be obtained in a single run. Therefore, multidimensional HPLC/ GC is useful for the analysis of complex samples, such as natural products. Various applications are reported4-', including a type of analysis for citrus essential oils One of the problems in the HPLC/ GC system is removing large volumes of HPLC eluent when introduced into the GC injector. Although various types of interfaces for HPLC/ GC have been proposed, they were complicated. Supercritical fluid chromatography (SFC) is a separation technique that is a bridge between GC and LC. Capillary SFC was applied to the separation of grapefruit oil by Richter and co-workers9, and the effect of the column temperature and CO2 density was described. However, both qualitative and quantitative analyses were not performed. A semi-preparative scale packed-column SFC was applied to the fractionation of lemon-peel oil by Yamauchi and Saito.10 In this study, lemon-peel oil was fractionated on a silica-gel column into several compound types, namely hydrocarbons, alcohols and aldehydes, esters and others, using a stepwise pressure gradient and modifier addition. Because supercritical C02, which is usually used as a mobile phase in SFC, becomes the gas phase under atmospheric pressure, the SFC is expected to be easily connected to other chromatographic equipment, such as a GC and an SFC. A few reports have appeared concerning the multidimensional SFC/ GC system as well as its application to fuel oil characterizations.ll-13 In this paper we report on a multidimensional SFC/ GC system in which two different chromatographic modes are combined and applied to the type analysis of citrus essential oils. Various kinds of citrus essential oils can be characterized by means of a quantification of three types of fractions (hydrocarbons, esters and aldehydes, and alcohols) as well as their components. Experimental Instrumentation A schematic diagram of the multidimensional SFC/ GC system employed in this study is shown in Fig. 1. The part of the SFC comprised a CO2 cylinder; an HPLC pump (Shimadzu LC-6A), a pump head of which was cooled so as to maintain a stable flow; a pump controller (Shimadzu SLC-6A); a precolumn; a sample injector (Rheodyne Model 7125) with a 20-µ1 sample
26 ANALYTICAL SCIENCES FEBRUARY 1994, VOL. 10 Table 1 system Operating conditions and columns of the SFC/GC Fig. 1 Schematic diagram of the SFC/GC system (A) and details of the interface (B) employed in this study. 1) carbon dioxide cylinder; 2) HPLC pump; 3) pump controller; 4) precolumn; 5) sample injector; 6) HPLC separation column; 7) column oven; 8) UV detector; 9) switching valve; 10) restrictor; 11) split/splitless injector; 12) cryofocusing; 13) carbon dioxide cylinder; 14) GC capillary column; 15) flame ionization detector; 16) GC column oven; 17) helium gas inlet; 18) capillary tube; 19) pressure gauge; 20) septum purge vent; 21) split vent; 22) stop valve. loop; a separation column of 4.6 mm i.d.x50 mm packed with silica-gels; a column oven (Shimadzu LC-1); a UV detector (Shimadzu SPD-6A) and a restrictor made of a capillary tube of 50 µm i.d.x300 mm. A Shimadzu GC- 9A system with a split/ splitless injector (Shimadzu SPL- G9) and a flame ionization detector (FID) was used as the part of the GC. For interfacing of the SFC to the GC, the C02 mobile phase was split by a switching valve (Rheodyne Model 7000) placed between the UV detector and the restrictor, and then introduced into the GC injector via a capillary tube of 25 µm i.d.x200 mm having a small orifice at the end. The flow rate of the C02 introduced into the GC injector was about 20 ml/ min as the gas phase when the pressure of the C02 mobile phase of SFC was l l MPa. Carbon dioxide, together with helium, was introduced into the GC injector during the heartcutting. The introduced components were concentrated in the top of a GC column using another liquid C02 throughout the heartcutting. Materials The separation columns used in the SFC and the GC were a Chromatorex-Si (particle size, 5 µm; pore diameter, 300 A) from Fuji Silysia Chemical Ltd. and an SPB-5 (size, 0.32 µm i.d.x60 m; film thickness, 1.0 µm) from Supelco Inc., respectively. Lemon oil (named A), lime oil, grapefruit oil and orange oil were donated from a flavor company. Lemon oil (B) was donated from another flavor company. Limonene, citronellal, a- terpineol and geranyl acetate used as standard samples and lemon oil (C) were obtained from Nacalai Tesque Inc. HPLC-grade hexane was used as a solvent for standard samples and was obtained from Kanto Chemical Co., Inc. Analytical conditions A split mode was used for GC injection when the C02 mobile phase used in the SFC was introduced into the GC. The chromatographic conditions of the type analysis of the standard samples and various kinds of the citrus essential oils by the SFC/ GC system are listed in Table 1. Results and Discussion In this study, four standard samples, such as limonene, citronellal, a-terpineol and geranyl acetate, were used as typical compounds for hydrocarbons, aldehydes, alcohols and esters, respectively. The retention behavior of the standard samples on the silica-gel column in SFC was investigated by plotting k' of the standard samples vs. the density of the C02 mobile phase (Fig. 2). Hexane was used to obtain to for calculating k'. The compounds having a lower polarity, such as limonene, were eluted faster than those having a higher polarity, such as a-terpineol. It seems that the elution order was based on the polarities of the solutes under these conditions, and was similar to that in normal-phase LC. Therefore, the citrus essential oils were intended to be separated into three types of compounds according to the pressure of the C02 mobile phase in the SFC. Figure 3 is a supercritical fluid chromatogram of standard samples. The stepwise pressure programing
27 Fig. 2 Effect of the density of carbon dioxide on k' in SFC. Samples: a-terpineol (O); geranyl acetate (S); citronellal (p); limonene (/). Fig. 4 Gas chromatograms of standard samples and of fractions obtained from pre-separation by SFC. Peak identifications: 1) limonene; 2) citronella!; 3) a-terpineol; 4) geranyl acetate. Fig. 3 Supercritical fluid chromatogram samples. Peak identifications: 1) limonene; 3) geranyl acetate; 4) a-terpineol. conditions indicated in Fig. 3 were used for this analysis. The elutes were fractionated according to the pressure of the C02 mobile phase. That is, under these conditions, limonene was eluted in fraction I, citronellal and geranyl acetate were eluted in fraction II and a-terpineol was eluted in fraction III. of 2) standard citronella!; The GC system used in this study has a split/ splitless injector, as shown in Fig. 1. Both stop valves at the septum purge vent and the split vent were closed in the case of splitless injection. Therefore, the heartcuttings from SFC to GC could not be applied under the splitless mode. In addition, the flow rate of C02 gas introduced into the GC injector should be less than that of the split vent, even in the case of split injection. For this reason, the heartcuttings could not be applied under a low split ratio. It takes 10 min to heartcut each fraction obtained by the SFC, as shown in Fig. 3. When the heartcuttings were performed without a cryofocusing technique, the GC peak shapes of those components having a low boiling point, such as limonene, were poor. Figure 4 shows gas chromatograms of standard samples and of three fractions obtained from pre-separation by SFC. The GC peak shapes of the components transferred from SFC were considerably good. The proportion of the peak area of an individual sample was varied according to the C02 pressure in SFC, probably because of a variation of the split ratio in the GC injector. In GC, the standard samples were eluted according to their boiling points. The retention behavior of the standard samples in GC was different from that in SFC. The SFC/ GC system was applied to the type analysis of lemon oil (A), as shown in Figs. 5 and 6. Figure 5 shows pre-separation by SFC, in which the lemon oil (A) was previously separated into three fractions, as same as in Fig. 3. Some of the components of lemon oil (A) were eluted out of three fractions under these SFC
28 ANALYTICAL SCIENCES FEBRUARY 1994, VOL. 10 of their high boiling points. After SFC pre-separation, each fraction was supplied to the GC injector to be separated (Fig. 6). Figure 6 also shows a gas chromatogram of lemon oil (A) compared with those of the three fractions obtained by SFC. In the gas chromatogram of fraction I, it is considered that the peaks detected at between 10 and 20 min were monoterpene hydrocarbons, and those detected after 30 min were sesquiterpene hydrocarbons based on a mass spectrometric analysis. In this study, the quantification of three fractions and their components of citrus essential oils were carried out by measuring the peak areas obtained by GC-FID based on the following assumptions. The compounds which belong to the same chemical class have the same sensitivity to FID. Therefore, the peak areas of the oil components obtained by FID are proportional to their amounts. In addition, aldehydes and esters, which were fractionated into fraction II, have the same sensitivity to FID if they have the same oxygen number in the molecules. The response of oxygen-containing organic compounds to FID were compared by Perkins and coworkers.l4 In practice, the sensitivity (peak area/ weight) of citronellal was equal to that of geranyl acetate. Therefore, three different calibration curves were used for the quantification of three fractions, respectively, transferred from the SFC. These curves were obtained from limonene (for fraction I), citronellal and geranyl acetate (for fraction II) and a-terpineol (for fraction III) used as standard samples. The ratios (wt%) of the three fractions and some components of lemon oil (A) were calculated (Table 2). Six kinds of components, the peaks of which are numbered in Fig. 6, were identified by mass spectrometry. As shown in Table 2, the lemon oil (A) almost comprised terpene hydrocarbons, in which limonene was comprised about 70% of the oil. To determine the reproducibility of the type analysis by the SFC/ GC system, the relative standard deviations (RSD) of some oil components were calculated, as shown in Table 3 (n=5). It is considered that the results of RSD obtained by SFC/ GC were poorer than those obtained by GC Table 2 Contents of three fractions and some components of lemon oil (A) Fig. 6 Gas chromatograms of lemon oil (A) and of fractions obtained from pre-separation by SFC. Peak identifications: 1) a-pinene; 2) limonene; 3) a-terpineol; 4) geranial; 5) geranyl acetate; 6) a-bisabolene. conditions. These components could not be eluted by GC when they were transferred from SFC to GC because a. Contents of the oil components. b. Contents of three fractions. Hydrocarbons, aldehydes and esters, and alcohols were eluted into fractions I, II, and III, respectively.
Table 3 Reproducibilities of the type analysis of lemon oil (A) by the SFC/GC 29 The authors wish to thank Mr. K. Sugiyama and Dr. Y. Abe for their support concerning the citrus essential oils and Mr. K. Nobuhara for contributing the silica-gel column. References Table 4 Composition (wt%) of various citrus oils alone. The SFC/ GC system was applied to the type analysis of various kinds of citrus essential oils (Table 4). All of the citrus essential oils used in this study mainly consisted of hydrocarbons. Lime oil included esters and/ or aldehydes in a high ratio compared with other oils. In conclusion, SFC could be easily connected to GC for the type analysis of citrus essential oils without any complicated interface. Various kinds of citrus essential oils were characterized by comparisons of their components separated by using two different chromatographic modes. The present multidimensional SFC/ GC was found to be very useful for the characterization of complex samples, such as natural oil products. 1. J. A. Staroscik and A. A. Wilson, J. Agric. Food Chem., 30, 507 (1982). 2. C. W. Wilson, III and P. E. Shaw, J. Agric. Food Chem., 32, 399 (1984). 3. M. G. Moshonas and P. E. Shaw, J. Agric. Food Chem., 32, 526 (1984). 4. I. L. Davies, M. W. Raynor, J. P. Kithinji, K. C. Bartle, P. T. Williams and G. E. Andrews, Anal. Chem., 60, 683A (1988). 5. M.-L. Riekkola, J. Chromatogr., 473, 315 (1989). 6. K. C. Bartle, I. L. Davies, M. W. Raynor and A. A. Clifford, J. Microcol. Sep.,1, 63 (1989). 7. K. Grob, "On-Line Coupled LC-GC", HUthig, Heidelberg, 1991. 8. F. Munari, G. Dugo and A. Cotroneo, J. High Resolut. Chromatogr.,13, 56 (1990). 9. B. E. Richter, M. R. Andersen, D. E. Knowles, E. R. Campbell, N. L. Porter, L. Nixon and D. W. Later, ACS Symp. Ser. 1987 (Pub. 1988), 366, p. 179. 10. Y. Yamauchi and M. Saito, J. Chromatogr., 505, 237 (1990). 11. J. M. Levy, J. P. Guzowski and W. E. Huhak, J. High Resolut. Chromatogr. Chromatogr. Commun., 10, 337 (1987). 12. J. M. Levy and J. P. Guzowski, Fresenius' Z. Anal. Chem., 330, 207 (1988). 13. J. M. Levy, R. A. Cavalier, T. N. Bosch, A. F. Rynaski and W. E. Huhak, J. Chromatogr. Sci., 27, 341 (1989). 14. G. Perkins, Jr., G. M. Rouayheb and L. D. Lively, "Gas Chromatography", ed. N. Brenner, J. E. Callen and M. Weiss, Chap. 19, p. 269, Academic press, New York, 1962. (Received August 30, 1993) (Accepted December 3, 1993)