- Email: [email protected]
Analysis of underivatised low volatility compounds by comprehensive two-dimensional gas chromatography with a short primary column
Accepted Manuscript Title: Analysis of underivatised low volatility compounds by comprehensive two dimensional gas chromatography with a short primary column Authors: F´abio Junior Moreira Novaes, Chadin Kulsing, Humberto Ribeiro Bizzo, Francisco Radler de Aquino Neto, Claudia Moraes de Rezende, Philip John Marriott PII: DOI: Reference:
S0021-9673(17)31268-2 http://dx.doi.org/10.1016/j.chroma.2017.08.069 CHROMA 358814
To appear in:
Journal of Chromatography A
Received date: Revised date: Accepted date:
13-1-2017 21-6-2017 24-8-2017
Please cite this article as: F´abio Junior Moreira Novaes, Chadin Kulsing, Humberto Ribeiro Bizzo, Francisco Radler de Aquino Neto, Claudia Moraes de Rezende, Philip John Marriott, Analysis of underivatised low volatility compounds by comprehensive two dimensional gas chromatography with a short primary column, Journal of Chromatography Ahttp://dx.doi.org/10.1016/j.chroma.2017.08.069 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 1
Analysis of underivatised low volatility compounds by comprehensive two dimensional gas chromatography with a short primary column Fábio Junior Moreira Novaes1-3,Chadin Kulsing3, Humberto Ribeiro Bizzo4, Francisco Radler de Aquino Neto2, Claudia Moraes de Rezende1, Philip John Marriott3* 1
Aroma Analysis Laboratory, Chemistry Institute, Federal University of Rio de Janeiro, 21941-909, Rio de Janeiro – RJ, Brazil 2 Laboratory of Capillary Column Preparation and Chromatography, Chemistry Institute, Federal University of Rio de Janeiro, 21941-598, Rio de Janeiro - RJ, Brazil 3 Australian Centre for Research on Separation Science, School of Chemistry, Monash University, Wellington Road, Clayton, Victoria, 3800, Australia 4 Embrapa Food Technology, 23020-470, Rio de Janeiro - RJ, Brazil
Submitted to Journal of Chromatography A: CHROMA-S-17-00080.R1
Running Head: Analysis of low volatility compounds in GC×GC
* Corresponding Author: E-mail: [email protected] Tel: +61-3-9905 9630 Fax: +61-3-9905 4597 HIGHLIGHTS Raw coffee bean extract lipids were analysed using GC×GC, without derivatisation
Classes included fatty acids, diterpenes, tocopherols, sterols, DAGs and TAGs
The latter late eluting analytes required high temperature GC analysis
Cafestol and kahweol ester series were tentatively identified
Normal hydrocarbons and DAGs were indicated as common coffee oil components
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 2
Abstract Comprehensive two dimensional gas chromatography (GCGC) approaches with cryogenic modulation were developed for the qualitative analysis of selected low volatility compounds in raw coffee bean extracts, without derivatisation. The approaches employed short first (1D) and second (2D) dimension columns, specifically a 1D 65% phenyl methyl siloxane column (11 m) and a 2D 5% phenyl methyl siloxane (1 m), which allowed elution of high molar mass compounds (e.g. > 600 Da). Solutes included hydrocarbons, fatty acids, diterpenes, tocopherols, sterols, diterpene esters, and di- and triacylglycerides. An oven temperature program up to 350 C was employed. The effects of experimental conditions were investigated, revealing that the GCGC results strongly depended on the cryogenic trap T, and oven T program. An appropriate condition was selected and further applied for group type analysis of low volatility compounds in green Arabica coffee beans. Retention indices were compiled for 1D analysis and were similar for the composite column data in GCGC. The elution of some compounds was confirmed by use of authentic standards. The approach allowed direct analysis of coffee extract in ethyl acetate solution, with improved analyte peak capacity (approximately 200 compounds were detected) without prior fractionation or pretreatment of the sample. This avoided potential hydrolysis of high molar mass conjugate esters as well as degradation of thermally labile compounds such as the derivatives of the diterpenes cafestol and kahweol.
Keywords: High temperature gas chromatography; comprehensive two-dimensional gas chromatography; DAGs; TAGs; coffee
1. Introduction Coffee is one of the most important traded commodities in the world, appreciated by more than 80% of the adult population [1]. After the roasting and brewing processes, the
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 3
released aroma results in more than 900 different volatile compounds [2]. The original coffee matrix contains about 800 low volatility substances, primarily lipids responsible for up to 17% w/w of the green bean contents [3]. These compounds have been reported by several researchers as fundamental ingredients in coffee aroma, in addition to giving favorable characteristics to the brew such as creaminess, wettability, and as an aid in formation and maintenance of foam especially for espresso preparation [4-7]. Most of the lipids are present in the endosperm of the beans, with a minor component of wax in the outer layer. Coffee oil was reported to consist of triacylglycerols (TAG (~75%), sterols (3.2% esterified and 2.2% in alcohol form), fatty acids (1%), phosphatides (0.3%), tocopherols (0.05%), tryptamine derivatives (≤1%), alcohols (0.4%) and esters from furankaurane pentacyclic diterpenes (up to 20%; n-C14:0-C24:0) [3,8]. This latter compound class differentiates coffee oil from other edible vegetable oils [9]. High temperature gas chromatography (HTGC) is an extension of GC where the oven temperature exceeds an arbitrary limit in conventional capillary (of up to 300-320 ºC), with a temperature ≥350 ºC [10,11]. This technique has proven to be effective for determination of high molar mass compounds (e.g. > 600 Da), long chain hydrocarbons (>C60) and wax esters [10-13], terpene esters such as α- and β-amyrin, simiarenols and lupeols [10-12,14], free and esterified tocopherols [14], sterols [15] and TAGs [10,13,16]. In a previous report, HTGC was used for the direct determination of major oil constituents in green Arabica coffee [17]. However, the technique was unable to resolve several compound types which co-eluted in the one dimensional (1D) separation. Comprehensive two-dimensional GC (GCGC) offers superior separation efficiency through enhanced resolution and increased peak capacity arising from the two-column separation approach, and improved limit of detection (LOD) as a result of refocusing when thermal modulation is employed [18]. Conventional GCGC applies a two dimensional separation to the whole sample in a single analysis, where long first dimension (1D; e.g. 30 m) and short second dimension (2D; e.g. 1-2 m) columns are employed in a manner that largely preserves the 1D separation according to the modulation ratio requirements [19]. Alternatively, a short 1D column with thin stationary phase film (0.1 µm) can be used in order to speed up the analysis [11,20]. This option exhibits less column bleed and higher detectability, and allows compounds to elute at lower T than otherwise would be found on long, lower phase ratio columns. This concept improves elution quality of thermolabile compounds and low volatility, polar, or intact conjugated compounds, without requirement of
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 4
clean-up, hydrolysis and derivatisation steps [10]. Application of this approach in GCGC is thus considered to be a key extension of the technique. In this study, a short 1D column approach was developed for the GCGC separation of low volatility compounds without derivatisation. Effects of separation conditions such as the trap temperature and oven temperature program, and flow rate were investigated. Application of the approach was demonstrated for group type analysis and identification of hydrocarbons, fatty acids, diterpene alcohols and esters, tocopherols, sterols, diglycerides (DAG) and TAG in a green Arabica extract sample.
2. Experimental 2.1 Chemicals and Materials Ethyl acetate was obtained from Merck Co. (Merck KGaA, Darmstadt, Germany). Normal alkane standards (C8 to C22, C24, C28 and C30), saturated (14:0, 16:0, 18:0 and 20:0) and unsaturated (18:1c9, C18:1t9, 18:2c, 18:3c, 22:1 and 24:1) fatty acids were purchased from Supelco (Bellefonte, PA; all 99% purity). Sterols (campesterol, 65%; β-sitosterol, 40% and stigmasterol, 95%) and triacylglycerols (1,2,3-triheptadecanoylglycerol and 1,2distearoyl-3-palmitoyl-rac-glycerol, ~99%) were purchased from Sigma–Aldrich (St. Louis, MO). Furankaurane coffee diterpenes cafestol and kahweol were isolated from Brazilian green Arabica coffee as described in our previous work [17]. 2.2 Instrumentation An Agilent 6850 GC instrument (Santa Clara, CA) equipped with a flame ionisation detector (GCFID) was employed. Hydrogen was used as carrier gas, with different flow rates as required. Pulsed split injection was performed at 330 C with an inlet pressure of 25 psi for 0.5 min, prior to the establishing the pressure values for the required flow rate. The GC analysis employed an Rtx-65TG column (11 m 0.25 mm I.D. 0.1 µm film thickness; Restek, Bellefonte, PA) with various temperature programs as indicated in the Figures. The GCGC column set consisted of a mid-polar Rtx-65TG column (as above) as the 1D column and a non-polar MEGA-5 Fast column (1.0 m 0.1 mm I.D. 0.1 µm; MEGA s.n.c., Legnano, Italy) as the 2D column. A deactivated press-tight connector (Restek) was used to connect the two columns through a longitudinally modulated cryogenic system (LMCS, Chromatography Concepts Ltd, Doncaster, Australia) used as a modulator, with the trap set at different temperatures. Modulation periods (PM) of 6 and 9 s were applied. The injection
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 5
volume was set at 1μL for all samples. Data in both 1D GC and GCGC were collected at 100 Hz. 2.3 Sample preparation Green Arabica coffee beans obtained from São José do Vale do Rio Preto (Rio de Janeiro, Brazil) were finely ground using a mortar and pestle. Two hundred mg of the sample was transferred into a vial (4 mL) containing a magnetic bar and ethyl acetate (2 mL). The vial was sealed with an airtight screw cap and submitted to hot direct extraction at about ethyl acetate boiling point (77 °C) for 30 min under vigorous stirring. The resulting suspension was filtered using a syringe filter (13 mm × 0.45 µm pore size Teflon membrane, Labquip Technologies, Melbourne, Australia) and the filtered solution directly applied to GC analysis. 2.4 Data processing Data acquisition and processing were performed using ChemStation software (Agilent). Microsoft Excel 2010 and Fortner Transform 3.3 (Fortner, Inc., Savoy, IL) were used for data visualisation. Some compounds in the coffee sample were identified by comparison of their retention time and index (I) with authentic standards. For I determination, a mixture of alkanes was injected using the same experimental condition as that of the sample analysis, for both 1D GC analysis, or GC×GC. For the former, the modulator was turned off so solute passes unhindered through both the 1D and 2D columns; the very short retention time on the 2D column is expected to give negligible error to the I calculation. As analytes eluted from the 1D column in a linear T program [21], I values were calculated according to the van den Dool and Kratz relationship: I 100n 100(
t R (i ) t R ( n ) t R ( n 1) t R ( n )
)
(1)
where tR is measured as the total retention time on both the 1D and 2D columns. n and n+1 are the carbon numbers of alkane standards which bracket the analyte i.
3. Results and discussion 3.1 Separation of coffee sample in 1D GC HTGC analysis of coffee oil (Fig. 1) was performed as reported in our previous work [17]. It is possible to identify the elution order for each group of compounds in the chromatogram to partially resolve high molecular weight compounds with very similar chemical structures such as TAGs; their structural similarity makes their complete separation difficult (Fig. 1; a list of compound identities for available authentic standards is given in
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 6
Table S1). This result provides a general idea about the coffee lipid composition from saponifiable and unsaponifiable lipids. However, there is still lack of information on minor components corresponding to peaks with either low signal abundance or coelution with major compounds. Several overlapped peaks were observed for hydrocarbons and fatty acids, and for DAGs with diterpene esters as well as coelution within these groups (e.g. between cafestol and kahweol esters) and also with TAGs. Similar HTGC conditions were thus selected as the starting point for further analysis in GCGC.
3.2 Effect of trap and oven temperature programs on the GCGC results Initial GC×GC analysis of the coffee sample was performed by using the same temperature (T) program as that applied in Fig. 1. However, improvements in GC×GC parameters were necessary and then carefully investigated in order to obtain a simple elution pattern of the analytes, especially for DAGs, diterpene esters and TAGs, which are the most complex and major components of the sample. The studied conditions included trap T, oven T program, and modulation period (PM). The range of parameters which were varied are shown in Fig. 2, with further experimental comment provided in electronic Supplementary Information Section S1. Also illustrated is the correlation between modulation process and the LOD for HT-GC×GC (Fig. S1-S2). According to this study, a difference between oven and trap T (T) of >100 °C could not effectively release analytes with high molecular weights, and resulted in some peak tailing and broadening. Thus T of 60-80C was chosen throughout the subsequent analyses, improving the peak shape with minimised baseline noise level caused by the column bleed (Fig. S1C). With the chosen trap T program (Fig. 2 and Fig. S2C), good compound group separation (with the exception of TAGs) was obtained as shown in Fig. 3A. In order to improve separation within each group, a slower ramping rate can be applied. For example, decrease of the first (from 20 to 12 °C min-1) and second (from 5 to 3 °C min-1) ramping rates improved separation of diterpene and DAG groups. These groups are located within the 1tR and 2tR ranges of 20-30 min and 3-5 s, respectively (Fig. 3B). A suitable condition can be obtained using a single ramping rate of 12 °C min-1 where all the groups were adequately separated without wraparound (Fig. 3C). Even greater group separation was obtained using 10 °C min-1 (Fig. 3D). However, the application of slower ramping rate resulted in broader
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 7
analyte peaks with lower intensities which decreased the number of detected peaks. The flow rate was increased from 2 to 3 mL min-1 after 28 min in order to minimise wraparound for the TAG region, since analyte 2tR was longer than the PM of 6 s (Fig. 3D). An alternative approach may be to increase PM, e.g. from 6 to 9 s. However, this can lead to apparent reduction of 1D separation, resulting in sampling more peaks into the 2D separation giving broader peaks in the transformed 1D modulated chromatogram when using PM = 9 s . This does not appear to be the case for the TAGs (Supporting Information Fig. S3), but may be of concern for other poorly resolved components such as DAGs. As a result, 10 °C min-1 ramp rate and 6 s PM with increasing flow rate were selected as the experimental condition applied for further analysis below. 3.3 Identification of specific regions of coffee lipid components in GCGC chromatograms With the above ramp rate, injection of the standards mix demonstrated good separation of different compound types. Due to the mid-polar – nonpolar column configuration, highly polar fatty acids (FA; these are underivatised, so are relatively polar, compared with the less polar methyl ester FAME) eluted at earlier 2tR in the 2D separation plane. FA were followed by diterpenes, tocopherols, sterols, DAGs, diterpene esters and TAGs. Non-polar alkanes are also present in almost all analyses, so elute as the most retained components. The compound class groups are indicated in Fig. 4A-D. Retention time and I data are provided in Supporting Information Table S1 with 34 compounds identified according to injections of authentic standards. Under the same condition, GCGC results for the coffee sample are shown in Fig. 4C-D and in Table S1, showing that compound types are easily separated into specific regions in the contour plot, avoiding some of the co-elution problems present in 1D GC. DAGs have not been reported as common or abundant components in green coffee oil, but they can be found as TAG hydrolysis products from roasted coffee storage [3,23]. Furthermore, DAGs in green coffee oil can be produced from TAG degradation during green coffee post-harvest processing (drying and/or storage). However, they can be a component in the crude beans as reported by Nikolova-Damianova et al. [24]. Patui et al. [25] performed quantification of DAGs in coffee seeds during germination. TAG compositions are commonly inferred based on measurement of free fatty acid content in hydrolysed coffee oil samples, but identifying actual TAG components require measurement of individual TAG in the samples. This can be confused with fatty acid content obtained from DAGs, which are also hydrolysed under the same reaction conditions.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 8
Chromatographic analysis of non-hydrolysed green coffee oils can be problematic due to the co-elution of DAGs with other compound classes such as diterpene esters causing errors in group type analysis. In addition, hydrocarbons (which are rarely reported in coffee oil samples) are also present in green coffee oils e.g. as reported by Folstar using GC to quantify a homologue n-alkane series from 16 to 31 carbon atoms which corresponded to 0.5-1 % w/w of green coffee oil [5,26]. Pujol et al. also found n-alkanes (< 3%) in coffee waste from the soluble coffee industry [27]. These two findings were confirmed in this study, where hydrocarbons are observed as part of the coffee bean composition, as indicated by authentic standard injection (Table S1). In addition, according to GCGC analyses, it is not likely that sterol esters were present (or they are present in negligible amounts which could not be detected in GCGC), since there were no peaks observed within the expected region of these compounds. According to the molar mass range, sterols elute within the region of diterpenes and DAGs, and logically sterol esters are expected to elute between diterpene esters and TAGs, but they are not seen in results here (Figs. 1 and 4).
3.4 Group type analysis of different compounds in the coffee oil sample Quantitative comparison between different compound types can be performed according to the peak area analysis as shown in Fig. 5. The GCGC result (Fig. 4C-D and Supporting Information Fig. S4) clearly indicates substantial co-elution between different compound types in a single dimension column (Fig. 1). As a result, the group type analysis result obtained from 1D GC showed significant errors compared to the more accurate values obtained from GCGC (Fig. 5), especially for the analysis of minor compounds such as hydrocarbons, diterpenes, sterols, diterpene esters and DAG, which were not well resolved in 1D GC. Quantification errors are caused by several overlaps among hydrocarbons and fatty acids, and both these groups with diterpenes, tocopherols and sterols. With the exception of the hydrocarbons and TAGs, all other compounds are normally derivatised prior to GC analysis, to enhance their signals and in some cases improve their peak shapes. Furthermore, cryogenic modulation permits refocusing of these compounds at the beginning of the 2D separation which increases their detectability. This translates into detection of about 200 peaks according to the 2D contour plot (Fig. S4). Compared with 1D GC, GCGC also provided quantification of group type classes of compounds considerably closer to those described in the literature (Fig. 4) [3,8]. The developed GCGC technique is especially useful
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 9
for improved analysis of diterpenes and sterols which are important chemical markers to distinguish coffee species, supporting authentication of coffee blends [7,28]. In addition, diterpenes have been correlated with coffee cupping tests as a potential quality marker [17], while DAGs and TAGs were found to be markers for coffee beans treated using different degrees of roasting, from medium to very dark [29].
3.5 Analysis of target regions of DAGs, diterpene esters and TAGs in GCGC Apart from the group type analysis, specific compounds within different groups in the GCGC result (Figure 3B) were further analysed. This is illustrated by the zoomed region in Fig. 6 where diterpene esters and DAGs are clearly separated, allowing detection of >40 individual peaks. Whilst analysis of these compounds was difficult in 1D GC due to the coelution of cafestol and kahweol esters and DAGs on the mid-polarity column phase as shown in Fig. 1 and reported in our previous study [17], GC×GC offers a facile quantitative measure of these species. It is even more difficult to separate positional isomers of cafestol and kahweol esters (e.g. cafestol linoleate and kahweol oleate) in 1D GC, whilst improved separation could be obtained in GCGC (Fig. 6). In addition, analysis of individual TAGs can also be performed with GCGC. Since TAGs are the dominant species present in coffee beans, their injected amounts were deliberately increased in the above experiments in order to allow detection of other minor compound classes. This resulted in the observation of TAG peak broadening and poor chromatographic resolution (Fig. 7A), which causes difficulty in analysis of other – trace – TAG compounds that may be present [30]. The wraparound phenomenon accompanied by some measure of ‘streaking’ of the component is also observed for TAGs (Fig. 3C, D). In this study, the problem was solved using a 10-fold dilution of the sample as well as an increase in the final oven T from 350 to 370 °C. As a result, the TAG concentration was decreased, whilst their elution time was reduced, which resulted in sharper peaks, thus keeping elution within the modulation period of 6 s as shown in Fig. 7B.
4. Conclusions GCGC with a short 1D column has been demonstrated for analysis of low volatility compounds, without derivatisation. The approach was applied for qualitative analysis of hydrocarbons, fatty acids, tocopherols, sterols, diterpene alcohols and esters, and di- and
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 10
triacylglycerols. Relative amounts of these compound types were also quantified based on total peak areas of the modulated peaks of compounds in different groups. The established approaches broadens compound ranges which can be analysed by GCGC using cryogenic modulation, allowing improved compound identification and group type analysis as an alternative to the application of liquid chromatography, and high resolution mass spectrometry, based on high temperature GC.
Acknowledgements This work was carried out with financial support of CNPq (Process Grant: 401044/2014-9), FAPERJ, EMBRAPA CAFÉ, and CAPES.
Supplementary data Supplementary data associated with this article can be found in the online version.
References
[1] V. Sridevi, P. Giridhar, G.A. Ravishankar, Evaluation of roasting and brewing effect on antinutritional diterpenes - cafestol and kahweol in coffee, Glob. J. Med. Res. 11 (2011) 1-7. [2] P.R.A.B. Toledo, L. Pezza, H.R. Pezza, A.T. Toci, Relationship between the different aspects related to coffee quality and their volatile compounds, Compr. Rev. Food Sci. Food Saf. 15 (2016) 705-719. [3] K. Speer, I. Kölling-Speer, The lipid fraction of the coffee bean, Braz. J. Plant Physiol. 18 (2006) 201-216 [4] I. Litman, S. Numrych, The role lipids play in the positive and negative flavors of food, ACS Symp. Ser. 75 (1978) 1-17. [5] P. Folstar, Lipids, in: R.J. Clarke, R. Macrae (Eds.), Coffee Volume 1: Chemistry, Springer Netherlands, London, 1985, pp. 203-222. [6] C.A.B. de Maria, R.F.A. Moreira, L.C. Trugo, Volatile Components in roasted coffee. Part. I: Heterocyclic Compounds, Química Nova 22 (1999) 209-217. [7] D. Pacetti, E. Boselli, M. Balzano, N.G. Frega, Authentication of Italian espresso coffee blends through the GC peak ratio between kahweol and 16- O-methylcafestol, Food Chem. 135 (2012) 1569-1574.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 11
[8] A. Farah, Coffee constituents, in: Y.-F. Chu (Ed.), Coffee: Emerging health effects and disease prevention, John Wiley & Sons Inc., Iowa, 2012, pp. 21-58. [9] D. Pacetti, P. Lucci, N.G. Frega, Unsaponifiable matter of coffee, in: V.R. Preedy (Ed.), Coffee in health and disease prevention, Elsevier Inc., London, 2015, pp. 119-126. [10] A.S. Pereira, F.R. Aquino Neto, High-temperature high-resolution gas chromatography: breaching the barrier to the analysis of polar and high molecular weight compounds, Trends Anal. Chem., 18 (1999) 126-136. [11] A.S. Pereira, M.C. Padilha, F.R. Aquino Neto, Two decades of high-temperature gas chromatography (1983-2003): what’s next? Microchem. J., 77 (2004) 141-149 [12] P.A. Sutton, M.J. Wilde, S.J. Martin, J. Cvačka, V. Vrkoslav, S.J. Rowland, Studies of long chain lipids in insects by high temperature gas chromatography and high temperature gas chromatography-mass spectrometry, J. Chromatogr. A, 1297 (2013) 236-240. [13] P.A. Sutton, S.J. Rowland, High temperature gas chromatography-time-of-flight-mass spectrometry (HTGC-ToF-MS) for high-boiling compounds, J. Chromatogr. A, 1243 (2012) 69-80. [14] A.S. Pereira, E.A. Nascimento, F.R. Aquino Neto, Lupeol alkanoates in Brazilian propolis, Z. Naturforsch. 57c (2002) 721-726. [15] H.L.N. Lau, C.W. Puah, Y.M. Choo, A.N. Ma, C.H. Chuah, Simultaneous quantification of free fatty acids, free sterols, squalene, and acylglycerol molecular species in palm oil by high-temperature gas chromatography-flame ionization detection, Lipids, 40 (2005) 532-528. [16] C. Rui-Samblás, A. González-Casado, L. Cuadros-Rodrígues, F.P.R. Carcía, Application of selected ion monitoring to the analysis of triacylglycerols in olive oil by high temperaturegas chromatography/mass spectrometry, Talanta 82 (2010) 255-260. [17] F.J.M. Novaes, S.S. Oigman, R.O.M.A. Souza, C.M. Rezende, F.R. Aquino Neto, New approaches on the analyses of thermolabile coffee diterpenes by gas chromatography and its relationship with cup quality, Talanta 139 (2015) 159-166. [18] P.J. Marriott, S.-T. Chin, B. Maikhunthod, H.-G. Schmarr, S. Bieri, Multidimensional gas chromatography, TrAC, Trends Anal. Chem. 34 (2012) 1-21. [19] W. Khummueng, J. Harynuk, P.J. Marriott, Modulation ratio in comprehensive twodimensional gas chromatography, Anal. Chem. 78 (2006) 4578−4587. [20] J. Harynuk, P.J. Marriott, Fast GC×GC with short primary column, Anal. Chem. 78 (2006) 2028-2034.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 12
[21] M. Jiang, C. Kulsing, Y. Nolvachai, P. J. Marriott, Two-dimensional retention indices improve component identification in comprehensive two-dimensional gas chromatography of saffron, Anal. Chem. 87 (2015) 5753−5761. [22] Y. F. Wong, S.-T. Chin, P. Perlmutter, P. J. Marriott, Evaluation of comprehensive twodimensional gas chromatography with accurate mass time-of-flight mass spectrometry for the metabolic profiling of plant-fungus interaction in Aquilaria malaccensis, J. Chromatogr. A 1387 (2015) 104-115. [23] A.T. Toci, V.J.M.F. Neto, A.G. Torres, A. Farah, Changes in triacylglycerols and free fatty acids composition during storage of roasted coffee, LWT - Food Sci. Technol. 50 (2013) 581-590. [24] B. Nikolova-Damianova, R. Velikova, G. N. Jham, Lipid classes, fatty acid composition and triacylglycerol molecular species in crude coffee beans harvested in Brazil, Food Res. Int. 31 (1998) 479-486. [25] S. Patui, L. Clincon, C. Peresson, M. Zancani, L. Conte, L. D. Terra, L. Navarini, A. Vianello, E. Braidot, Lipase activity and antioxidant capacity in coffee (Coffea arabica L.) seeds during germination, Plant Sci. 219-220 (2014) 19-25. [26] P. Folstar, The composition of wax and oil in green coffee beans. Thesis, Wageningen University, The Netherlands, 1976. [27] D. Pujol, C. Liu, J. Gominho, M. À. Olivella, N. Fiol, I. Villaescusa, J. Pereira, The chemical composition of exhausted coffee waste, Ind. Crops Prod. 50 (2013) 423-429. [28] W. Kamm, F. Dionisi, L.-B. Fay, C. Hischenhuber, H.-G. Schmarr, K.-H. Engel, Rapid and simultaneous analysis of 16-O-methylcafestol and sterols as markers for assessment of green coffee bean authenticity by on-line LC–GC, JAOCS, 79 (2002) 1109-1113. [29] J.S. Rosa, O. Freitas-Silva, J.R.C. Rouws, I.G.S. Moreira, F.J.M. Novaes, D.A. Azevedo, N. Schwab, R.L.O. Godoy, M.N. Eberlin, C.M. Rezende, Mass spectrometry screening of Arabica coffee roasting: A non-target and non-volatile approach by EASI-MS and ESI-MS, Food Res. Int. 89 (2016) 967-975. [30] L. Ramos, J. Sanz, Annex Troubleshooting, Compr. Anal. Chem. 55 (2009) 283-298.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 13
Figure Legends Fig. 1. (A) 1D GC chromatogram for green Arabica coffee extract illustrating the regions of fatty acids, diterpenes (Dit), tocopherol (TC), sterols, diacylglycerols, diterpene esters and triacylglycerols. Oven T program: 70 °C (0.25 min) to 300 °C at 20 °C min−1 then 5 °C min-1 to 350 °C (hold 12.25 min).
Fig. 2. Schematic diagram of the optimisation strategy to obtain an acceptable 2D plot based on effects of (1) trapping T profiles, (2) oven T programs and (3) modulation period (PM).
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 14
Fig. 3. Effect of oven temperature programs: (A) 20 °C min-1 followed by 5 °C min-1 after 11.75 min, (B) 12 °C min-1 followed by 3 °C min-1 at 19.41 min, (C) 12 °C min-1, and (D) 10 °C min-1, for the GCGC results for green Arabica coffee extract. The arrows indicate where the oven T reaches 300 °C. The flow rate was increased from 2 to 3 mL min-1 after 28 min for (D). The same cluster of peaks is circled in each case; these are the DAGs and diterpene esters, and are also shown in Fig. 3C. The starting and final temperatures in all cases were 70 °C (0.25 min) and 350 C (15 min), respectively and profile of temperature difference between oven and trap T (ΔT) is 80 °C, but it is reduced to 60 °C for the TAGs region.
Fig. 4 GCGC chromatograms for: (A and B) standards mix (hydrocarbons, saturated and unsaturated fatty acids, diterpenes (Dit), tocopherol (TC) and sterols), (C) green Arabica coffee extract, and (D) the same chromatogram as C but obtained with an expanded response range focusing on the minor components and elution regions highlighting features for each substance group. Oven T program: 70 °C (hold 0.25 min) ramp at 10 C min-1 to 350 C (hold 15 min), and profile of temperature difference between oven and trap T (ΔT): 80 °C/4-28 min and 60 °C/28-40 min.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 15
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 16
Fig. 5. Quantitative group type analysis results obtained from 1D GC and GC×GC chromatograms (Fig. 1 and 4) compared with the literature values obtained from Speer & Kölling-Sper [3] and Farah [8]. The values obtained in this study were normalised based on the total group peak areas for each analysis.
Fig. 6. (A) Chemical structures of furankaurane pentacyclic diterpenes present in Arabica coffee beans. Free form: R = H and esterified: R = acyl chain length of n-C14:0, C16:0, C16:1, C17:0, C18:0, C18:1, C18:2, C18:3, C20:0, C20:1, C22:0, C24:0. (B) 2D plot expansion of the corresponding region in Figure 4B.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 17
Fig. 7. GC and GCGC results with the focus on triacylglycerols (TAGs) present in Arabica coffee beans: (A) 1D GC result with (B) the corresponding GCGC result (expansion of Figure 4C) using oven temperature program 70 °C (hold 0.25 min) ramp T 10C min-1 to 350 C (hold15 min) and profile of temperature difference between oven and trap T (ΔT) (80 °C/4-28 min and 60 °C/28-40 min), (C) 1D GC result obtained with the same condition as (A) except that with a final oven temperature of 370 °C and (D) the corresponding GCGC result obtained using 10-fold sample dilution.
S0021-9673(17)31268-2 http://dx.doi.org/10.1016/j.chroma.2017.08.069 CHROMA 358814
To appear in:
Journal of Chromatography A
Received date: Revised date: Accepted date:
13-1-2017 21-6-2017 24-8-2017
Please cite this article as: F´abio Junior Moreira Novaes, Chadin Kulsing, Humberto Ribeiro Bizzo, Francisco Radler de Aquino Neto, Claudia Moraes de Rezende, Philip John Marriott, Analysis of underivatised low volatility compounds by comprehensive two dimensional gas chromatography with a short primary column, Journal of Chromatography Ahttp://dx.doi.org/10.1016/j.chroma.2017.08.069 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 1
Analysis of underivatised low volatility compounds by comprehensive two dimensional gas chromatography with a short primary column Fábio Junior Moreira Novaes1-3,Chadin Kulsing3, Humberto Ribeiro Bizzo4, Francisco Radler de Aquino Neto2, Claudia Moraes de Rezende1, Philip John Marriott3* 1
Aroma Analysis Laboratory, Chemistry Institute, Federal University of Rio de Janeiro, 21941-909, Rio de Janeiro – RJ, Brazil 2 Laboratory of Capillary Column Preparation and Chromatography, Chemistry Institute, Federal University of Rio de Janeiro, 21941-598, Rio de Janeiro - RJ, Brazil 3 Australian Centre for Research on Separation Science, School of Chemistry, Monash University, Wellington Road, Clayton, Victoria, 3800, Australia 4 Embrapa Food Technology, 23020-470, Rio de Janeiro - RJ, Brazil
Submitted to Journal of Chromatography A: CHROMA-S-17-00080.R1
Running Head: Analysis of low volatility compounds in GC×GC
* Corresponding Author: E-mail: [email protected] Tel: +61-3-9905 9630 Fax: +61-3-9905 4597 HIGHLIGHTS Raw coffee bean extract lipids were analysed using GC×GC, without derivatisation
Classes included fatty acids, diterpenes, tocopherols, sterols, DAGs and TAGs
The latter late eluting analytes required high temperature GC analysis
Cafestol and kahweol ester series were tentatively identified
Normal hydrocarbons and DAGs were indicated as common coffee oil components
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 2
Abstract Comprehensive two dimensional gas chromatography (GCGC) approaches with cryogenic modulation were developed for the qualitative analysis of selected low volatility compounds in raw coffee bean extracts, without derivatisation. The approaches employed short first (1D) and second (2D) dimension columns, specifically a 1D 65% phenyl methyl siloxane column (11 m) and a 2D 5% phenyl methyl siloxane (1 m), which allowed elution of high molar mass compounds (e.g. > 600 Da). Solutes included hydrocarbons, fatty acids, diterpenes, tocopherols, sterols, diterpene esters, and di- and triacylglycerides. An oven temperature program up to 350 C was employed. The effects of experimental conditions were investigated, revealing that the GCGC results strongly depended on the cryogenic trap T, and oven T program. An appropriate condition was selected and further applied for group type analysis of low volatility compounds in green Arabica coffee beans. Retention indices were compiled for 1D analysis and were similar for the composite column data in GCGC. The elution of some compounds was confirmed by use of authentic standards. The approach allowed direct analysis of coffee extract in ethyl acetate solution, with improved analyte peak capacity (approximately 200 compounds were detected) without prior fractionation or pretreatment of the sample. This avoided potential hydrolysis of high molar mass conjugate esters as well as degradation of thermally labile compounds such as the derivatives of the diterpenes cafestol and kahweol.
Keywords: High temperature gas chromatography; comprehensive two-dimensional gas chromatography; DAGs; TAGs; coffee
1. Introduction Coffee is one of the most important traded commodities in the world, appreciated by more than 80% of the adult population [1]. After the roasting and brewing processes, the
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 3
released aroma results in more than 900 different volatile compounds [2]. The original coffee matrix contains about 800 low volatility substances, primarily lipids responsible for up to 17% w/w of the green bean contents [3]. These compounds have been reported by several researchers as fundamental ingredients in coffee aroma, in addition to giving favorable characteristics to the brew such as creaminess, wettability, and as an aid in formation and maintenance of foam especially for espresso preparation [4-7]. Most of the lipids are present in the endosperm of the beans, with a minor component of wax in the outer layer. Coffee oil was reported to consist of triacylglycerols (TAG (~75%), sterols (3.2% esterified and 2.2% in alcohol form), fatty acids (1%), phosphatides (0.3%), tocopherols (0.05%), tryptamine derivatives (≤1%), alcohols (0.4%) and esters from furankaurane pentacyclic diterpenes (up to 20%; n-C14:0-C24:0) [3,8]. This latter compound class differentiates coffee oil from other edible vegetable oils [9]. High temperature gas chromatography (HTGC) is an extension of GC where the oven temperature exceeds an arbitrary limit in conventional capillary (of up to 300-320 ºC), with a temperature ≥350 ºC [10,11]. This technique has proven to be effective for determination of high molar mass compounds (e.g. > 600 Da), long chain hydrocarbons (>C60) and wax esters [10-13], terpene esters such as α- and β-amyrin, simiarenols and lupeols [10-12,14], free and esterified tocopherols [14], sterols [15] and TAGs [10,13,16]. In a previous report, HTGC was used for the direct determination of major oil constituents in green Arabica coffee [17]. However, the technique was unable to resolve several compound types which co-eluted in the one dimensional (1D) separation. Comprehensive two-dimensional GC (GCGC) offers superior separation efficiency through enhanced resolution and increased peak capacity arising from the two-column separation approach, and improved limit of detection (LOD) as a result of refocusing when thermal modulation is employed [18]. Conventional GCGC applies a two dimensional separation to the whole sample in a single analysis, where long first dimension (1D; e.g. 30 m) and short second dimension (2D; e.g. 1-2 m) columns are employed in a manner that largely preserves the 1D separation according to the modulation ratio requirements [19]. Alternatively, a short 1D column with thin stationary phase film (0.1 µm) can be used in order to speed up the analysis [11,20]. This option exhibits less column bleed and higher detectability, and allows compounds to elute at lower T than otherwise would be found on long, lower phase ratio columns. This concept improves elution quality of thermolabile compounds and low volatility, polar, or intact conjugated compounds, without requirement of
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 4
clean-up, hydrolysis and derivatisation steps [10]. Application of this approach in GCGC is thus considered to be a key extension of the technique. In this study, a short 1D column approach was developed for the GCGC separation of low volatility compounds without derivatisation. Effects of separation conditions such as the trap temperature and oven temperature program, and flow rate were investigated. Application of the approach was demonstrated for group type analysis and identification of hydrocarbons, fatty acids, diterpene alcohols and esters, tocopherols, sterols, diglycerides (DAG) and TAG in a green Arabica extract sample.
2. Experimental 2.1 Chemicals and Materials Ethyl acetate was obtained from Merck Co. (Merck KGaA, Darmstadt, Germany). Normal alkane standards (C8 to C22, C24, C28 and C30), saturated (14:0, 16:0, 18:0 and 20:0) and unsaturated (18:1c9, C18:1t9, 18:2c, 18:3c, 22:1 and 24:1) fatty acids were purchased from Supelco (Bellefonte, PA; all 99% purity). Sterols (campesterol, 65%; β-sitosterol, 40% and stigmasterol, 95%) and triacylglycerols (1,2,3-triheptadecanoylglycerol and 1,2distearoyl-3-palmitoyl-rac-glycerol, ~99%) were purchased from Sigma–Aldrich (St. Louis, MO). Furankaurane coffee diterpenes cafestol and kahweol were isolated from Brazilian green Arabica coffee as described in our previous work [17]. 2.2 Instrumentation An Agilent 6850 GC instrument (Santa Clara, CA) equipped with a flame ionisation detector (GCFID) was employed. Hydrogen was used as carrier gas, with different flow rates as required. Pulsed split injection was performed at 330 C with an inlet pressure of 25 psi for 0.5 min, prior to the establishing the pressure values for the required flow rate. The GC analysis employed an Rtx-65TG column (11 m 0.25 mm I.D. 0.1 µm film thickness; Restek, Bellefonte, PA) with various temperature programs as indicated in the Figures. The GCGC column set consisted of a mid-polar Rtx-65TG column (as above) as the 1D column and a non-polar MEGA-5 Fast column (1.0 m 0.1 mm I.D. 0.1 µm; MEGA s.n.c., Legnano, Italy) as the 2D column. A deactivated press-tight connector (Restek) was used to connect the two columns through a longitudinally modulated cryogenic system (LMCS, Chromatography Concepts Ltd, Doncaster, Australia) used as a modulator, with the trap set at different temperatures. Modulation periods (PM) of 6 and 9 s were applied. The injection
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 5
volume was set at 1μL for all samples. Data in both 1D GC and GCGC were collected at 100 Hz. 2.3 Sample preparation Green Arabica coffee beans obtained from São José do Vale do Rio Preto (Rio de Janeiro, Brazil) were finely ground using a mortar and pestle. Two hundred mg of the sample was transferred into a vial (4 mL) containing a magnetic bar and ethyl acetate (2 mL). The vial was sealed with an airtight screw cap and submitted to hot direct extraction at about ethyl acetate boiling point (77 °C) for 30 min under vigorous stirring. The resulting suspension was filtered using a syringe filter (13 mm × 0.45 µm pore size Teflon membrane, Labquip Technologies, Melbourne, Australia) and the filtered solution directly applied to GC analysis. 2.4 Data processing Data acquisition and processing were performed using ChemStation software (Agilent). Microsoft Excel 2010 and Fortner Transform 3.3 (Fortner, Inc., Savoy, IL) were used for data visualisation. Some compounds in the coffee sample were identified by comparison of their retention time and index (I) with authentic standards. For I determination, a mixture of alkanes was injected using the same experimental condition as that of the sample analysis, for both 1D GC analysis, or GC×GC. For the former, the modulator was turned off so solute passes unhindered through both the 1D and 2D columns; the very short retention time on the 2D column is expected to give negligible error to the I calculation. As analytes eluted from the 1D column in a linear T program [21], I values were calculated according to the van den Dool and Kratz relationship: I 100n 100(
t R (i ) t R ( n ) t R ( n 1) t R ( n )
)
(1)
where tR is measured as the total retention time on both the 1D and 2D columns. n and n+1 are the carbon numbers of alkane standards which bracket the analyte i.
3. Results and discussion 3.1 Separation of coffee sample in 1D GC HTGC analysis of coffee oil (Fig. 1) was performed as reported in our previous work [17]. It is possible to identify the elution order for each group of compounds in the chromatogram to partially resolve high molecular weight compounds with very similar chemical structures such as TAGs; their structural similarity makes their complete separation difficult (Fig. 1; a list of compound identities for available authentic standards is given in
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 6
Table S1). This result provides a general idea about the coffee lipid composition from saponifiable and unsaponifiable lipids. However, there is still lack of information on minor components corresponding to peaks with either low signal abundance or coelution with major compounds. Several overlapped peaks were observed for hydrocarbons and fatty acids, and for DAGs with diterpene esters as well as coelution within these groups (e.g. between cafestol and kahweol esters) and also with TAGs. Similar HTGC conditions were thus selected as the starting point for further analysis in GCGC.
3.2 Effect of trap and oven temperature programs on the GCGC results Initial GC×GC analysis of the coffee sample was performed by using the same temperature (T) program as that applied in Fig. 1. However, improvements in GC×GC parameters were necessary and then carefully investigated in order to obtain a simple elution pattern of the analytes, especially for DAGs, diterpene esters and TAGs, which are the most complex and major components of the sample. The studied conditions included trap T, oven T program, and modulation period (PM). The range of parameters which were varied are shown in Fig. 2, with further experimental comment provided in electronic Supplementary Information Section S1. Also illustrated is the correlation between modulation process and the LOD for HT-GC×GC (Fig. S1-S2). According to this study, a difference between oven and trap T (T) of >100 °C could not effectively release analytes with high molecular weights, and resulted in some peak tailing and broadening. Thus T of 60-80C was chosen throughout the subsequent analyses, improving the peak shape with minimised baseline noise level caused by the column bleed (Fig. S1C). With the chosen trap T program (Fig. 2 and Fig. S2C), good compound group separation (with the exception of TAGs) was obtained as shown in Fig. 3A. In order to improve separation within each group, a slower ramping rate can be applied. For example, decrease of the first (from 20 to 12 °C min-1) and second (from 5 to 3 °C min-1) ramping rates improved separation of diterpene and DAG groups. These groups are located within the 1tR and 2tR ranges of 20-30 min and 3-5 s, respectively (Fig. 3B). A suitable condition can be obtained using a single ramping rate of 12 °C min-1 where all the groups were adequately separated without wraparound (Fig. 3C). Even greater group separation was obtained using 10 °C min-1 (Fig. 3D). However, the application of slower ramping rate resulted in broader
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 7
analyte peaks with lower intensities which decreased the number of detected peaks. The flow rate was increased from 2 to 3 mL min-1 after 28 min in order to minimise wraparound for the TAG region, since analyte 2tR was longer than the PM of 6 s (Fig. 3D). An alternative approach may be to increase PM, e.g. from 6 to 9 s. However, this can lead to apparent reduction of 1D separation, resulting in sampling more peaks into the 2D separation giving broader peaks in the transformed 1D modulated chromatogram when using PM = 9 s . This does not appear to be the case for the TAGs (Supporting Information Fig. S3), but may be of concern for other poorly resolved components such as DAGs. As a result, 10 °C min-1 ramp rate and 6 s PM with increasing flow rate were selected as the experimental condition applied for further analysis below. 3.3 Identification of specific regions of coffee lipid components in GCGC chromatograms With the above ramp rate, injection of the standards mix demonstrated good separation of different compound types. Due to the mid-polar – nonpolar column configuration, highly polar fatty acids (FA; these are underivatised, so are relatively polar, compared with the less polar methyl ester FAME) eluted at earlier 2tR in the 2D separation plane. FA were followed by diterpenes, tocopherols, sterols, DAGs, diterpene esters and TAGs. Non-polar alkanes are also present in almost all analyses, so elute as the most retained components. The compound class groups are indicated in Fig. 4A-D. Retention time and I data are provided in Supporting Information Table S1 with 34 compounds identified according to injections of authentic standards. Under the same condition, GCGC results for the coffee sample are shown in Fig. 4C-D and in Table S1, showing that compound types are easily separated into specific regions in the contour plot, avoiding some of the co-elution problems present in 1D GC. DAGs have not been reported as common or abundant components in green coffee oil, but they can be found as TAG hydrolysis products from roasted coffee storage [3,23]. Furthermore, DAGs in green coffee oil can be produced from TAG degradation during green coffee post-harvest processing (drying and/or storage). However, they can be a component in the crude beans as reported by Nikolova-Damianova et al. [24]. Patui et al. [25] performed quantification of DAGs in coffee seeds during germination. TAG compositions are commonly inferred based on measurement of free fatty acid content in hydrolysed coffee oil samples, but identifying actual TAG components require measurement of individual TAG in the samples. This can be confused with fatty acid content obtained from DAGs, which are also hydrolysed under the same reaction conditions.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 8
Chromatographic analysis of non-hydrolysed green coffee oils can be problematic due to the co-elution of DAGs with other compound classes such as diterpene esters causing errors in group type analysis. In addition, hydrocarbons (which are rarely reported in coffee oil samples) are also present in green coffee oils e.g. as reported by Folstar using GC to quantify a homologue n-alkane series from 16 to 31 carbon atoms which corresponded to 0.5-1 % w/w of green coffee oil [5,26]. Pujol et al. also found n-alkanes (< 3%) in coffee waste from the soluble coffee industry [27]. These two findings were confirmed in this study, where hydrocarbons are observed as part of the coffee bean composition, as indicated by authentic standard injection (Table S1). In addition, according to GCGC analyses, it is not likely that sterol esters were present (or they are present in negligible amounts which could not be detected in GCGC), since there were no peaks observed within the expected region of these compounds. According to the molar mass range, sterols elute within the region of diterpenes and DAGs, and logically sterol esters are expected to elute between diterpene esters and TAGs, but they are not seen in results here (Figs. 1 and 4).
3.4 Group type analysis of different compounds in the coffee oil sample Quantitative comparison between different compound types can be performed according to the peak area analysis as shown in Fig. 5. The GCGC result (Fig. 4C-D and Supporting Information Fig. S4) clearly indicates substantial co-elution between different compound types in a single dimension column (Fig. 1). As a result, the group type analysis result obtained from 1D GC showed significant errors compared to the more accurate values obtained from GCGC (Fig. 5), especially for the analysis of minor compounds such as hydrocarbons, diterpenes, sterols, diterpene esters and DAG, which were not well resolved in 1D GC. Quantification errors are caused by several overlaps among hydrocarbons and fatty acids, and both these groups with diterpenes, tocopherols and sterols. With the exception of the hydrocarbons and TAGs, all other compounds are normally derivatised prior to GC analysis, to enhance their signals and in some cases improve their peak shapes. Furthermore, cryogenic modulation permits refocusing of these compounds at the beginning of the 2D separation which increases their detectability. This translates into detection of about 200 peaks according to the 2D contour plot (Fig. S4). Compared with 1D GC, GCGC also provided quantification of group type classes of compounds considerably closer to those described in the literature (Fig. 4) [3,8]. The developed GCGC technique is especially useful
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 9
for improved analysis of diterpenes and sterols which are important chemical markers to distinguish coffee species, supporting authentication of coffee blends [7,28]. In addition, diterpenes have been correlated with coffee cupping tests as a potential quality marker [17], while DAGs and TAGs were found to be markers for coffee beans treated using different degrees of roasting, from medium to very dark [29].
3.5 Analysis of target regions of DAGs, diterpene esters and TAGs in GCGC Apart from the group type analysis, specific compounds within different groups in the GCGC result (Figure 3B) were further analysed. This is illustrated by the zoomed region in Fig. 6 where diterpene esters and DAGs are clearly separated, allowing detection of >40 individual peaks. Whilst analysis of these compounds was difficult in 1D GC due to the coelution of cafestol and kahweol esters and DAGs on the mid-polarity column phase as shown in Fig. 1 and reported in our previous study [17], GC×GC offers a facile quantitative measure of these species. It is even more difficult to separate positional isomers of cafestol and kahweol esters (e.g. cafestol linoleate and kahweol oleate) in 1D GC, whilst improved separation could be obtained in GCGC (Fig. 6). In addition, analysis of individual TAGs can also be performed with GCGC. Since TAGs are the dominant species present in coffee beans, their injected amounts were deliberately increased in the above experiments in order to allow detection of other minor compound classes. This resulted in the observation of TAG peak broadening and poor chromatographic resolution (Fig. 7A), which causes difficulty in analysis of other – trace – TAG compounds that may be present [30]. The wraparound phenomenon accompanied by some measure of ‘streaking’ of the component is also observed for TAGs (Fig. 3C, D). In this study, the problem was solved using a 10-fold dilution of the sample as well as an increase in the final oven T from 350 to 370 °C. As a result, the TAG concentration was decreased, whilst their elution time was reduced, which resulted in sharper peaks, thus keeping elution within the modulation period of 6 s as shown in Fig. 7B.
4. Conclusions GCGC with a short 1D column has been demonstrated for analysis of low volatility compounds, without derivatisation. The approach was applied for qualitative analysis of hydrocarbons, fatty acids, tocopherols, sterols, diterpene alcohols and esters, and di- and
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 10
triacylglycerols. Relative amounts of these compound types were also quantified based on total peak areas of the modulated peaks of compounds in different groups. The established approaches broadens compound ranges which can be analysed by GCGC using cryogenic modulation, allowing improved compound identification and group type analysis as an alternative to the application of liquid chromatography, and high resolution mass spectrometry, based on high temperature GC.
Acknowledgements This work was carried out with financial support of CNPq (Process Grant: 401044/2014-9), FAPERJ, EMBRAPA CAFÉ, and CAPES.
Supplementary data Supplementary data associated with this article can be found in the online version.
References
[1] V. Sridevi, P. Giridhar, G.A. Ravishankar, Evaluation of roasting and brewing effect on antinutritional diterpenes - cafestol and kahweol in coffee, Glob. J. Med. Res. 11 (2011) 1-7. [2] P.R.A.B. Toledo, L. Pezza, H.R. Pezza, A.T. Toci, Relationship between the different aspects related to coffee quality and their volatile compounds, Compr. Rev. Food Sci. Food Saf. 15 (2016) 705-719. [3] K. Speer, I. Kölling-Speer, The lipid fraction of the coffee bean, Braz. J. Plant Physiol. 18 (2006) 201-216 [4] I. Litman, S. Numrych, The role lipids play in the positive and negative flavors of food, ACS Symp. Ser. 75 (1978) 1-17. [5] P. Folstar, Lipids, in: R.J. Clarke, R. Macrae (Eds.), Coffee Volume 1: Chemistry, Springer Netherlands, London, 1985, pp. 203-222. [6] C.A.B. de Maria, R.F.A. Moreira, L.C. Trugo, Volatile Components in roasted coffee. Part. I: Heterocyclic Compounds, Química Nova 22 (1999) 209-217. [7] D. Pacetti, E. Boselli, M. Balzano, N.G. Frega, Authentication of Italian espresso coffee blends through the GC peak ratio between kahweol and 16- O-methylcafestol, Food Chem. 135 (2012) 1569-1574.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 11
[8] A. Farah, Coffee constituents, in: Y.-F. Chu (Ed.), Coffee: Emerging health effects and disease prevention, John Wiley & Sons Inc., Iowa, 2012, pp. 21-58. [9] D. Pacetti, P. Lucci, N.G. Frega, Unsaponifiable matter of coffee, in: V.R. Preedy (Ed.), Coffee in health and disease prevention, Elsevier Inc., London, 2015, pp. 119-126. [10] A.S. Pereira, F.R. Aquino Neto, High-temperature high-resolution gas chromatography: breaching the barrier to the analysis of polar and high molecular weight compounds, Trends Anal. Chem., 18 (1999) 126-136. [11] A.S. Pereira, M.C. Padilha, F.R. Aquino Neto, Two decades of high-temperature gas chromatography (1983-2003): what’s next? Microchem. J., 77 (2004) 141-149 [12] P.A. Sutton, M.J. Wilde, S.J. Martin, J. Cvačka, V. Vrkoslav, S.J. Rowland, Studies of long chain lipids in insects by high temperature gas chromatography and high temperature gas chromatography-mass spectrometry, J. Chromatogr. A, 1297 (2013) 236-240. [13] P.A. Sutton, S.J. Rowland, High temperature gas chromatography-time-of-flight-mass spectrometry (HTGC-ToF-MS) for high-boiling compounds, J. Chromatogr. A, 1243 (2012) 69-80. [14] A.S. Pereira, E.A. Nascimento, F.R. Aquino Neto, Lupeol alkanoates in Brazilian propolis, Z. Naturforsch. 57c (2002) 721-726. [15] H.L.N. Lau, C.W. Puah, Y.M. Choo, A.N. Ma, C.H. Chuah, Simultaneous quantification of free fatty acids, free sterols, squalene, and acylglycerol molecular species in palm oil by high-temperature gas chromatography-flame ionization detection, Lipids, 40 (2005) 532-528. [16] C. Rui-Samblás, A. González-Casado, L. Cuadros-Rodrígues, F.P.R. Carcía, Application of selected ion monitoring to the analysis of triacylglycerols in olive oil by high temperaturegas chromatography/mass spectrometry, Talanta 82 (2010) 255-260. [17] F.J.M. Novaes, S.S. Oigman, R.O.M.A. Souza, C.M. Rezende, F.R. Aquino Neto, New approaches on the analyses of thermolabile coffee diterpenes by gas chromatography and its relationship with cup quality, Talanta 139 (2015) 159-166. [18] P.J. Marriott, S.-T. Chin, B. Maikhunthod, H.-G. Schmarr, S. Bieri, Multidimensional gas chromatography, TrAC, Trends Anal. Chem. 34 (2012) 1-21. [19] W. Khummueng, J. Harynuk, P.J. Marriott, Modulation ratio in comprehensive twodimensional gas chromatography, Anal. Chem. 78 (2006) 4578−4587. [20] J. Harynuk, P.J. Marriott, Fast GC×GC with short primary column, Anal. Chem. 78 (2006) 2028-2034.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 12
[21] M. Jiang, C. Kulsing, Y. Nolvachai, P. J. Marriott, Two-dimensional retention indices improve component identification in comprehensive two-dimensional gas chromatography of saffron, Anal. Chem. 87 (2015) 5753−5761. [22] Y. F. Wong, S.-T. Chin, P. Perlmutter, P. J. Marriott, Evaluation of comprehensive twodimensional gas chromatography with accurate mass time-of-flight mass spectrometry for the metabolic profiling of plant-fungus interaction in Aquilaria malaccensis, J. Chromatogr. A 1387 (2015) 104-115. [23] A.T. Toci, V.J.M.F. Neto, A.G. Torres, A. Farah, Changes in triacylglycerols and free fatty acids composition during storage of roasted coffee, LWT - Food Sci. Technol. 50 (2013) 581-590. [24] B. Nikolova-Damianova, R. Velikova, G. N. Jham, Lipid classes, fatty acid composition and triacylglycerol molecular species in crude coffee beans harvested in Brazil, Food Res. Int. 31 (1998) 479-486. [25] S. Patui, L. Clincon, C. Peresson, M. Zancani, L. Conte, L. D. Terra, L. Navarini, A. Vianello, E. Braidot, Lipase activity and antioxidant capacity in coffee (Coffea arabica L.) seeds during germination, Plant Sci. 219-220 (2014) 19-25. [26] P. Folstar, The composition of wax and oil in green coffee beans. Thesis, Wageningen University, The Netherlands, 1976. [27] D. Pujol, C. Liu, J. Gominho, M. À. Olivella, N. Fiol, I. Villaescusa, J. Pereira, The chemical composition of exhausted coffee waste, Ind. Crops Prod. 50 (2013) 423-429. [28] W. Kamm, F. Dionisi, L.-B. Fay, C. Hischenhuber, H.-G. Schmarr, K.-H. Engel, Rapid and simultaneous analysis of 16-O-methylcafestol and sterols as markers for assessment of green coffee bean authenticity by on-line LC–GC, JAOCS, 79 (2002) 1109-1113. [29] J.S. Rosa, O. Freitas-Silva, J.R.C. Rouws, I.G.S. Moreira, F.J.M. Novaes, D.A. Azevedo, N. Schwab, R.L.O. Godoy, M.N. Eberlin, C.M. Rezende, Mass spectrometry screening of Arabica coffee roasting: A non-target and non-volatile approach by EASI-MS and ESI-MS, Food Res. Int. 89 (2016) 967-975. [30] L. Ramos, J. Sanz, Annex Troubleshooting, Compr. Anal. Chem. 55 (2009) 283-298.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 13
Figure Legends Fig. 1. (A) 1D GC chromatogram for green Arabica coffee extract illustrating the regions of fatty acids, diterpenes (Dit), tocopherol (TC), sterols, diacylglycerols, diterpene esters and triacylglycerols. Oven T program: 70 °C (0.25 min) to 300 °C at 20 °C min−1 then 5 °C min-1 to 350 °C (hold 12.25 min).
Fig. 2. Schematic diagram of the optimisation strategy to obtain an acceptable 2D plot based on effects of (1) trapping T profiles, (2) oven T programs and (3) modulation period (PM).
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 14
Fig. 3. Effect of oven temperature programs: (A) 20 °C min-1 followed by 5 °C min-1 after 11.75 min, (B) 12 °C min-1 followed by 3 °C min-1 at 19.41 min, (C) 12 °C min-1, and (D) 10 °C min-1, for the GCGC results for green Arabica coffee extract. The arrows indicate where the oven T reaches 300 °C. The flow rate was increased from 2 to 3 mL min-1 after 28 min for (D). The same cluster of peaks is circled in each case; these are the DAGs and diterpene esters, and are also shown in Fig. 3C. The starting and final temperatures in all cases were 70 °C (0.25 min) and 350 C (15 min), respectively and profile of temperature difference between oven and trap T (ΔT) is 80 °C, but it is reduced to 60 °C for the TAGs region.
Fig. 4 GCGC chromatograms for: (A and B) standards mix (hydrocarbons, saturated and unsaturated fatty acids, diterpenes (Dit), tocopherol (TC) and sterols), (C) green Arabica coffee extract, and (D) the same chromatogram as C but obtained with an expanded response range focusing on the minor components and elution regions highlighting features for each substance group. Oven T program: 70 °C (hold 0.25 min) ramp at 10 C min-1 to 350 C (hold 15 min), and profile of temperature difference between oven and trap T (ΔT): 80 °C/4-28 min and 60 °C/28-40 min.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 15
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 16
Fig. 5. Quantitative group type analysis results obtained from 1D GC and GC×GC chromatograms (Fig. 1 and 4) compared with the literature values obtained from Speer & Kölling-Sper [3] and Farah [8]. The values obtained in this study were normalised based on the total group peak areas for each analysis.
Fig. 6. (A) Chemical structures of furankaurane pentacyclic diterpenes present in Arabica coffee beans. Free form: R = H and esterified: R = acyl chain length of n-C14:0, C16:0, C16:1, C17:0, C18:0, C18:1, C18:2, C18:3, C20:0, C20:1, C22:0, C24:0. (B) 2D plot expansion of the corresponding region in Figure 4B.
Novaes et al.
Analysis of low volatility compounds in GC×GC
page 17
Fig. 7. GC and GCGC results with the focus on triacylglycerols (TAGs) present in Arabica coffee beans: (A) 1D GC result with (B) the corresponding GCGC result (expansion of Figure 4C) using oven temperature program 70 °C (hold 0.25 min) ramp T 10C min-1 to 350 C (hold15 min) and profile of temperature difference between oven and trap T (ΔT) (80 °C/4-28 min and 60 °C/28-40 min), (C) 1D GC result obtained with the same condition as (A) except that with a final oven temperature of 370 °C and (D) the corresponding GCGC result obtained using 10-fold sample dilution.