Food Chemistry 185 (2015) 355–361

Contents lists available at ScienceDirect

Food Chemistry journal homepage: www.elsevier.com/locate/foodchem

Analytical Methods

Application of integrated comprehensive/multidimensional gas chromatography with mass spectrometry and olfactometry for aroma analysis in wine and coffee Sung-Tong Chin a, Graham T. Eyres b, Philip J. Marriott a,⇑ a b

Australian Centre for Research on Separation Science, School of Chemistry, Monash University, Wellington Road, Clayton, VIC 3800, Australia Department of Food Science, University of Otago, PO Box 56, Dunedin 9056, New Zealand

a r t i c l e

i n f o

Article history: Received 21 November 2013 Received in revised form 9 December 2014 Accepted 3 April 2015 Available online 9 April 2015 Keywords: Cumulative SPME Olfactometry Multiple headspace sampling Odour assessment Shiraz wine aroma High resolution GC Cryogenic trapping

a b s t r a c t Component coelution in chromatographic analysis complicates identification and attribution of individual odour-active volatile molecules in complex multi-component samples. An integrated system incorporating comprehensive two-dimensional gas chromatography (GC  GC) and multidimensional gas chromatography (MDGC), with flame ionisation, olfactometry and mass spectrometry detection was developed to circumvent data correlation across different systems. Identification of potent odorants in Shiraz wine and the headspace of ground coffee are demonstrated as selected applications. Multiple solid-phase microextraction (SPME) sampling with GC-O located odour-active regions; GC  GC established the complexity of odour-active regions; MDGC provided high-resolution separation for each region; simultaneous ‘O’ and MS detection completed the analysis for target resolved peaks. Seven odour regions in Shiraz were analysed with MDGC-O/MS detection, revealing 11 odour volatiles through matching of mass spectrometry and retention indices from both separating dimensions, including acetic acid; octen-3-ol; ethyl octanoate; methyl-2-oxo-nonanoate; butanoic acid, 2-methylbutanoic acid, and 3methylbutanoic acid; 3-(methylthio)-1-propanol; hexanoic acid; b-damascenone; and ethyl-3-phenylpropanoate. A capsicum odour in ground coffee was identified as 2-methoxy-3-isobutylpyrazine with a 5-fold increase in S/N of the odorant when acquired using a 6-time cumulative SPME sampling approach. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Understanding the important compounds responsible for food aroma such as in wine and coffee is valuable to improve production processes, such as viticulture practice, vinification, and the roasting process, in order to achieve optimum product quality in these products (Chin, Eyres, & Marriott, 2011; Clarke & Bakker, 2004; Ebeler & Thorngate, 2009; Flament, 2002; Semmelroch & Grosch, 1996; Shimoda & Shibamoto, 1990). Volatile components in such products exhibit complex composition, comprising mainly alcohols, esters, acids, and other minor compounds including norisoprenoid ketones, and sulphur- and nitrogen-containing compounds in the wine, which impact greatly on the global aroma even at sub ppb level (Davis & Qian, 2011; Ebeler & Thorngate, 2009; Pons, Lavigne, Eric, Darriet, & Dubourdieu, 2008; Siebert, Wood, Elsey, & Pollnitz, 2008; Simpson, Capone, & Sefton, 2004). Analysis of the complex composition requires development and ⇑ Corresponding author. Tel.: +61 3 99059630; fax: +61 3 99058501. E-mail address: [email protected] (P.J. Marriott). http://dx.doi.org/10.1016/j.foodchem.2015.04.003 0308-8146/Ó 2015 Elsevier Ltd. All rights reserved.

exploitation of enhanced separation strategies for disclosing the odour-active volatiles that contribute to unique aroma characteristics. Often, pre-separation prior to GC–MS is preferred in characterisation of flavour, either by classic acid/base or solvent extraction, or other column chromatography approaches (Qian, Burbank, & Wang, 2007). Fractionation by offline high performance liquid chromatography (HPLC) followed by multidimensional gas chromatography (MDGC)-O/MS was utilised to reveal several impact odorants in wine, such as a blackberry flavour contributed by ethyl 2-hydroxy-4-methylpentanoate ester (Falcao, Lytra, Darriet, & Barbe, 2012), and a strong prune odour by c-nonalactone, b-damascenone, and 3-methyl-2,4-nonanedione ketones (Pons et al., 2008). Identification of rotundone as the peppery note in wine was performed by coupling solid phase extraction (SPE) with solid phase microextraction (SPME) to selectively concentrate the target component prior to GC–MS detection (Siebert et al., 2008). Schmarr et al. (2010) reported a solid phase extraction (SPE) clean-up procedure followed by online HPLC coupled to MDGC– MS for quantifying 3-alkyl-2-methoxypyrazines in wines and

356

S.-T. Chin et al. / Food Chemistry 185 (2015) 355–361

musts with complete elimination of the interfering matrix background. These tedious pre-fractionation procedures required longer processing time, with the possibility of artefact formation. Notwithstanding numerous reported techniques, successful aroma analysis demands achieving the greatest possible resolution, with ease of application, whilst also achieving an absence of artefacts and overlapping components generated during the measurement process. Comprehensive two-dimensional gas chromatography coupled to time-of-flight mass spectrometry (GC  GC–TOFMS) has been successfully utilised for non-targeted volatile analysis that permits simultaneous analysis of a significantly larger number of compounds found in the wine headspace (Chin et al., 2011; Robinson, Boss, Heymann, Solomon, & Trengove, 2011). The role of MDGC and GC  GC to deliver improved chromatographic resolution has been recently reviewed (Marriott, Chin, Maikhunthod, Schmarr, & Bieri, 2012). A recent study proposed an integrated GC  GC/MDGC system with simultaneous MS, FID and olfactometry detection, capable of implementing both GC  GC and heart-cut (H/C) MDGC modes within a single unit GC system (Chin, Eyres, & Marriott, 2012b). An analytical strategy using this proposed system for odorants analysis in samples with complex background such as wine was suggested. The present work extends the application of the integrated GC  GC/MDGC system for analysis in several complex coeluting odour-active regions of aroma extract from Shiraz wine and ground coffee. Cumulative SPME was incorporated for enhanced sensitivity analysis of odour-active compounds. Matching of retention indices (RI) obtained from 1D and 2D separation was used for compound identification together with acquired mass spectrum comparison with established MS database. 2. Materials and methods 2.1. Materials Shiraz wine (Lindeman’s, Victoria, Australia; 2010 Vintage) was purchased from a local wine retailer. Coffee powder (Harris coffee, NSW, Australia) was purchased from a local store. SPME fibres consisting of polar polyacrylate (PA), the blended porous particle phase carboxen and divinylbenzene with polydimethylsiloxane (PDC), and holders for manual sampling were gifts from Supelco (Sigma–Aldrich St. Louis, MO, USA). Hexanoic acid, octen-3-ol and b-damascenone (>98% purity) were gifts from Australian Botanical Product (Hallam, Australia). Standard compounds with purity >97% included butanoic acid, 2-methylbutanoic acid, 3methylthiopropanol, ethyl 3-phenylpropanoate, and a saturated alkane series from C9 to C22 were obtained from either Fluka or Aldrich (Sigma–Aldrich). GC grade hexane, analytical grade sodium chloride (NaCl), and acetic acid were obtained from Merck Chemical Co. (Merck KGaA, Darmstadt, Germany). 2.2. Cumulative SPME sampling Cumulative SPME sampling for wine was performed manually as previously described (Chin, Eyres, & Marriott, 2012a). Repeated sampling of 2  PA + 2  PDC with 30 min extraction time for each fibre (denoted as 2  PADC) were sequentially desorbed in the injector for 3 min each, and volatiles were accumulated in a cryotrap (CT) installed just after the injector for cryofocussing. The sampling protocol required separate fibres to be placed in the sample headspace at 3 min intervals, each for the 30 min sampling period, then individually desorbed, in order to minimise total analysis time. Up to six separate PDC fibres were used for evaluation of coffee aroma, allowing up to 6  PDC accumulations (40 min extraction time each) with desorption as

above, for trace analysis. Thus the total extraction/desorption time was 6  3 min + 40 min. The GC programme commenced 3 min after the final SPME desorption step into the GC injector, immediately followed by switching off the CT CO2 supply. 2.3. GC-O, GC  GC, Heart-cut-MDGC-O/MS analysis An integrated GC  GC/MDGC system as described previously (Chin et al., 2012b) was used for this study. The system (Supplementary information Fig. S1) comprises an Agilent 7980/ 5975C series GC–MS (Agilent Technologies, Nunawading, Australia) retrofitted with a liquid CO2 CT device (SGE Scientific, Ringwood, Australia), a SGE olfactory port (ODO II model, SGE Scientific, Ringwood, Australia), an Agilent G2855A Deans switch device (DS), an Agilent G3180B 2-way effluent splitter (ES), and an Everest model Longitudinally Modulated Cryogenic System (LMCS; Chromatography Concepts, Doncaster, Australia). For the Shiraz wine sample, a polar DB-FFAP column (30 m length  0.25 mm ID  0.25 lm film thickness (df); Agilent Technologies), a non-polar BPX5 (0.9 m  0.10 mm ID  0.10 lm df; SGE Analytical Science) and non-polar DB-5 ms column (30 m  0.25 mm ID  0.25 lm df; Agilent Technologies) were used for the 1D, short 2DS and long 2DL separation columns respectively. For the coffee sample, an ionic liquid phase SLB-IL59 column (30 m  0.25 mm ID  0.20 lm df; Supelco), a mid-polar VF200 ms (30.0 m  0.25 mm ID  0.25 lm df) and a microbore VF200 ms (1.0 m  0.10 mm ID  0.10 lm df; both from Agilent Technologies) was applied as 1D, long 2DL and short 2DS columns respectively. The SGE CT served for solute trapping of SPME desorbed analytes at the inlet end of the 1D column, as well as trapping of H/C fractions at the front end of the long 2DL column. The CT of the LMCS was positioned near the beginning of the short 2 DS column for GC  GC operation. The inlet of the 1D column was connected to a split/splitless injector whilst the 1D outlet and both 2D column inlets were connected to the DS. Effluent from the short 2Ds column outlet was split equally via a Y-union and 2 deactivated fused silica (DFS) capillaries (55 cm  0.10 mm ID) to a flame ionisation detector (FID) and an olfactory port respectively. Meanwhile, the effluent from the long 2DL column outlet was equally split by the ES device operated at 30.0 psi and directed to both the MS detector via a DFS (80 cm  0.10 mm ID) transfer line heated at 240 °C and another DFS (75 cm  0.10 mm ID) to the olfactory port respectively. The same olfactory port was used for each channel, and the two separate channel DFS transfer lines both terminated at the nose cone of the olfactory port. The inlet pressure was initially applied at 48.5 psi during SPME desorption, and then ramped to 51.0 psi immediately after commencing the GC programme giving a flow rate of 2.0 mL/min at 150 °C in 1D, with the DS held constant at 47.5 psi. Following 1D separation of the SPME extract, the DS diverted the solute to either O/FID via 2DS without operating the LMCS (for GC-O/FID), or to O/ MS dual detection via 2DL (for GC-O/MS). Two experienced panelists were used to evaluate the odour active regions during olfactometry analysis. The GC inlet was set at 250 °C, with helium carrier gas and split vent open after 2 min. The oven programme was set at 60 °C for 2 min, increased to 120 °C (20 °C/min), then to 250 °C (3 °C/min) with 5 min hold. GC  GC–FID analysis was conducted by operating the LMCS CT at 20 °C with a modulation period (PM) of 4 s. The FID was operated at 250 °C, with acquisition rate of 20 Hz and 100 Hz for 1D GC and GC  GC analysis respectively. To perform H/C MDGC operation, a 2-step sequence was programmed for the H/C event. The required switching time (0.4 min duration for each H/C zone) was entered into the events software in accordance with GC-O/FID analysis for the target

S.-T. Chin et al. / Food Chemistry 185 (2015) 355–361

olfactory zones. The SGE CT cooling was initiated 3 min prior to transfer of the target zone H/C, in order to cryogenically trap and refocus the H/C solutes at the head of 2DL. The SGE CT coolant supply was closed to remobilise the trapped H/C into 2DL. The inlet, DS and ES were the adjusted to 30 psi, 35 psi and 30 psi respectively, giving a lower carrier flow of 2.0 mL/min at 50 °C for 2DL separation together with simultaneously back-flushing the 1D column. During this 2D run, the oven was re-commenced from 50 °C for both wine and coffee samples (initial 0.5 min hold), to 260 °C (5 °C/min) and held for 10 min. The MS used electron ionisation at 70 eV and 230 °C with a mass scan range 40–350 m/z. MS data processing utilised the automated mass spectral deconvolution and identification system (AMDIS) programme version 2.66, and NIST MS library version 2.0f. RI data were obtained by injecting alkanes and transferring them to the CT at the head of 2DL; they were then eluted following the same method as for odour regions. 3. Results and discussion Performance of the proposed integrated GC  GC/MDGC-O/MS system has been demonstrated previously (Chin et al., 2012b), and provides consistency of compound analysis for volatile analytes across the different analysis modes. Heart-cut MDGC serves to deliver improved separation with respect to peak capacity whilst the complementary technique of GC  GC provides a simultaneous multidimensional separation for the entire sample in a single analysis (Marriott et al., 2012), with a range of benefits such as structurally ordered chromatograms in the 2D separation plane and increased detectability (Beens & Brinkman, 2005; Marriott, Eyres, & Dufour, 2009). Meanwhile, cumulative SPME sampling of 2  PADC was reported to enhance peak response and improves extract representativeness (Chin et al., 2012a, 2012b), and compensates for the losses of sensitivity caused by effluent splitting to 2 simultaneous detectors, i.e. olfactory port and FID or MS. This system was re-assembled a year since the earlier publication and further applied here to odour analysis of two sample matrices, Shiraz wine and ground coffee. 3.1. Shiraz wine analysis Fig. 1 illustrates the GC–FID chromatogram and corresponding GC  GC contour plot of volatile compounds from Shiraz wine

Fig. 1. GC analysis of Shiraz wine with SPME sampling of volatile using 2  PADC (refer to text) showing (A) GC  GC–FID contour plot, and (B) 1D GC–FID result. Dotted boxes (1–7) indicate odour regions which were H/C to 2DL (see later).

357

sampled by 2  PADC. Olfactory assessment was conducted in conjunction with GC–FID/O analysis at this stage, followed by subsequent GC  GC analysis in order to examine the complexity of the volatile profile. GC  GC also provides an overall 2D view of the total volatile composition of the sample, independent of just assignment of odour regions. Such a plot becomes useful if in subsequent work standards of the odour compounds can locate appropriate peak contours, and allows rapid profiling of many samples without requiring the H/C MDGC approach. The basic method can further be modified to permit both GC  GC–MS and MDGC– MS to be used. The GC-O result was obtained and correlated with GC-SNIF data reported previously (Chin et al., 2012a) for locating significant odour regions. Selection of several odour-active peaks were indicated in Fig. 1A (dotted boxes), where overlapping peaks were observed in the GC  GC result (Fig. 1A). Notably, either trace level response or only FID noise signal was observed in region 7, but a fruity/smoke/licorice perception was noted. A number of these selected regions were further subjected to H/C-MDGC-O/ MS analysis for odorant identification. Table 1 lists the odorants identified in the selected odour regions with their corresponding 1D and 2DL RI data. The estimated RI result matches with previous findings in the online databases (National Institute of Standards, 2013; The Pherobase: Database of Pheromones, 2013), providing an additional level of confirmation for all the tentative peak assignments for identified compounds, except methyl 2-oxononanoate for which RI data evidently have yet to be reported. Compound identification is more reliable with three confirmative data points, namely mass spectrometry, 1 D and 2D RI. Notably, once the retention time of saturated aliphatic alkane series have been projected in both 1D and 2D separation, the molecular information i.e. mass spectrometry, 1D and 2D RI are readily available in a single analysis using the integrated system described. Several odour regions in this work, with potent aroma characteristics exhibited detection frequency of 3 or more out of 6, as revealed in previous GC-O work (Chin et al., 2012a). Odour regions 1 and 6 in Fig. 1A were discussed previously (Chin et al., 2012b) as a proof of concept. Odour region 1 comprised of acetic acid, octen-3-ol, and ethyl octanoate as odour contributors, whereas hexanoic acid with b-damascenone were identified in odour region 6. Fig. 2A –E illustrate the TIC obtained with MS of the expanded separation accomplished by the long 2DL column, with reduced carrier flow, for the H/C regions 2, 3, 4, 5 and 7. Olfactory assessment was conducted throughout every 2D GC–MS analysis, so that the odorant compounds responsible for the perceived aroma could be identified for each H/C zone. H/C-MDGC clearly reveals that all the odorants detected in 1D GC analysis co-eluted with numerous interfering volatiles diverted from 1D during the same heart-cut time. Separation of odour region 2 (1tR 16.8 min; refer to Fig. 1) by H/C MDGC is demonstrated in Fig. 2A. 1D GC–MS indicated only the abundant peak 2,3-butanediol which does not appear to match the spicy capsicum-like odour note that was perceived in this region. The 2DL chromatogram of H/C region 2 (Fig. 2A) reveals the complexity of the transferred effluent with a multitude of extra peaks, beside the 2,3-butanediol compound (2tR  9 min; refer to Fig. 2A). Note that the MS response scale (range 20,000 counts here) has been significantly expanded; so many components are of low abundance. A similar odour response to that for zone 2 was obtained under 2DL MDGC analysis at the asterisked peak located at 2tR 22.0 min (Fig. 2A) after removing other background volatiles via the 2D separation. The mass spectrum of this unknown peak matched to methyl 2-oxononanoate with 80% similarity; no authentic standard was available to confirm this assignment. Fig. 2B corresponds to H/C zone 3 which exhibited a strong sweaty odour in 1D-GC-O/FID evaluation. The same sweaty odour was perceived at the asterisked peak eluting at 2tR 9.32 min which was

358

S.-T. Chin et al. / Food Chemistry 185 (2015) 355–361

Table 1 Identification of odorants in wine from the heart-cut regions with first and second column retention details. Odour regiona 1 1 1 2 3 4 4 5 6 6 7

Identified compoundb f

Acetic acid Octen-3-olf Ethyl octanoatef Methyl 2-oxononanoate Butanoic acidf 2-Methylbutanoic acidf 3-Methylbutanoic acidf 3-(Methylthio)-1-propanolf Hexanoic acidf b-Damascenonef Ethyl 3-phenylpropanoatef

tR (min)d

tR (min)d

Icale

Odour descriptionc

1

2

1

Vinegar Earthy Fruity Spicy/capsicum Sweaty/cheese Sweaty/cheese Sweaty/cheese Brothy Sweaty Floral Fruity/smoke/licorice

14.10 14.10 14.40 16.80 19.66 21.0 21.0 22.64 26.56 26.56 28.05

6.30 15.35 22.83 22.84 9.32 11.21 11.58 15.12 16.18 28.56 27.49

1421 1421 1425 1502 1591 1636 1636 1676 1819 1819 1873

2

Icale

1

658 961 1212 1212 759 822 835 953 989 1404 1368

1404–1477 1423–1465 1422–1446 – 1610–1650 1652–1667 1652–1691 1705–1745 1829–1872 1811–1836 1837–1905

Iref

*

2

Iref

**

625–663 960–1012 1180–1208 – 775–826 838–875 839–888 962–987 977–1036 1360–1404 1348–1370

a

Odour region corresponding to Fig. 2A. Compound identified by MS with library matching over 80% similarity. Odour perceived in 1D-GC-O/FID and 2D-GC-O/MS. d Retention time observed in the first dimension (1tr) and second dimension (2tr) of MDGC–MS analysis. 2tr is determined from the commencement of the second oven programme run. e 1 I cal and 2Ical estimated by linear temperature programmed alkane reference standards according to van den Dool and Kratz. f Compound identity was confirmed using reference compound. * RI reference values for polar DB-FFAP phase according to Pherobase database (http://www.pherobase.com/database/kovats/kovats-index.php). ** RI reference values for apolar DB-5 ms phase according to NIST database (http://webbook.nist.gov/chemistry/). b

c

Fig. 2. GC analysis of (A) the 2DL result obtained from the spicy capsicum odour region H/C from 1D, shown in Fig. 1 zone 2; asterisked peak identity matches to methyl 2-oxo-nonanoate. (B) The 2DL result obtained from the sweaty/cheesy odour region H/C from 1D, shown in Fig. 1 zone 3; asterisked peak identity matches to butanoic acid. (C) The 2DL result obtained from the sweaty/cheesy odour region H/C from 1D, shown in Fig. 1 zone 4; asterisked peaks identities (from right to left) match to 2-methylbutanoic acid and 3-methylbutanoic acid. (D) the 2DL result obtained from the brothy odour region H/C from 1D, shown in Fig. 1 zone 5; asterisked peak identity matches to 3-(methylthio)-propanol (3MTP). (E) The 2DL result obtained from the fruity/smoke/licorice odour region H/C from 1D, shown in Fig. 1 zone 7; asterisked peak identity matches to ethyl 3-phenylpropanoate (E3PP).

now clearly resolved during 2DL-GC-O/MS analysis. This was identified as butanoic acid through its mass spectrum and retention index match. For H/C zone 4, two isomeric volatile peaks were found to contribute to a sweaty/cheese odour, shown as the asterisked peaks at 11.21 min and 11.59 min in Fig. 2C. The compounds responsible were identified as 2-methylbutanoic acid and 3-methylbutanoic acid. These isomeric peaks were partially

separated in the 2D-GC analysis, but baseline resolution of the volatile isomers could be achieved using a 2D column with different selectivity such as cyanopropyl or ionic liquid phases (Ando & Sasaki, 2011; Mjøs, 2005). Nevertheless, the resolution of these peaks was sufficient to ascribe odour activity and a good mass spectrum to each of the peaks. A sulphur-containing odorant, 3-methylthio-propanol (3MTP), corresponding to the asterisked peak at 2tR 15.12 min in Fig. 2D, was responsible for contribution of a brothy odour in zone 5 (Fig. 1B). Whilst the response to the FID indicates a relatively minor peak in Fig. 1B, identification is supported by the clear resolution in the 2DL column, and the use of an authentic reference sample giving the same retention index, odour quality and mass spectrum. After confirmation of the compound by the classical heart-cut MDGC-O/MS approach, the proposed system configuration also enables direct correlation of such data to the GC  GC result of odour region 5. The peak contour with 2tR = 1.1 s is readily distinguished in the GC  GC contour plot as 3MTP which is the first high abundance compound eluted from the 2DS column in this case, and is similarly found to be located in the result acquired by the 2DL column. This alcohol compound, having a polar –OH attribute, locates as the early-eluting band cluster in the GC  GC plot as anticipated for the retention order on a polar–apolar column combination (Marriott et al., 2012). Contribution of sulphur volatiles to wine aroma is important, with some compounds used as possible off-flavour markers in the product (Davis & Qian, 2011). Being the most abundant volatile sulphur compound in wine (in the mg/L range), 3MTP (or methionol) is generated during yeast fermentation of grape must in the presence of the corresponding amino acid, methionine (Moreira, de Pinho, Santos, & Vasconcelos, 2011). Fedrizzi et al. revealed that significant incremental increase for 3MTP was observed during the wine ageing process, where an enhanced rate was likely dependent on the ageing temperatures of the wine (Fedrizzi, Magno, Finato, & Versini, 2010). 3MTP was also found to diminish in wine during oxidative storage (Fedrizzi et al., 2011). Identification of the low abundance ethyl 3-phenylpropanoate (E3PP) compound at 2tR 27.49 min is evident in 2D-GC–MS analysis (Fig. 2E), giving a fruity/smoke/licorice odour contribution to the Shiraz wine headspace aroma. Previous work with GC–MS in selected ion mode and sensory evaluation demonstrated that E3PP has a very low odour threshold; estimated at 1.6 lg/L in alcoholic solution (Ferreira, López, & Cacho, 2000). Aromatic ester

S.-T. Chin et al. / Food Chemistry 185 (2015) 355–361

359

volatiles generally accumulate during primary and malolactic fermentation as well as during storage, through enzymatic synthesis and hydrolysis reaction (Sumby, Grbin, & Jiranek, 2010). Fig. 3A illustrates the corresponding peak contour of E3PP in the GC  GC plot present amongst multiple background components in this odour region, whilst Fig. 3B shows the retention of a E3PP reference compound in GC  GC analysis, confirming the location of E3PP in the GC  GC display. The compound E3PP has been previously observed in Merlot wine (Welke, Manfroi, Zanus, Lazarotto, & Alcaraz Zini, 2012), and Pinotage wine (Weldegergis, Crouch, Górecki, & de Villiers, 2011) using GC  GC–TOFMS approaches.

3.2. Coffee analysis An alternative column set comprising an ionic liquid phase SLBIL59 and a mid-polar phase VF-200 ms as 1D and 2D respectively was tested for headspace analysis of ground coffee volatiles. Fig. 4 illustrates the 1D gas chromatogram and the corresponding GC  GC contour plot acquired from ground coffee headspace using single SPME sampling. A majority of the extracted volatile compounds elute after 10 min on the 1D IL59 column, which allows cryogenic modulation by liquid CO2 for most of the compounds. Application of VF-200 ms as the 2D column in GC  GC analysis results in all components eluting within one modulation period in the contour plot shown in Fig. 4B. Later elution of coffee volatiles leads to a higher elution temperature (Te) which results in shorter 2 D retention times (2tR). This reduces the possibility of compound wrap-around during GC  GC analysis, and is beneficial compared with the use of a 15 m FFAP column as 1D (Chin et al., 2011). The GC  GC separation occupies much of the 2D space in this work, and is comparable to results reported elsewhere (Tranchida et al., 2009), which employed a 50 lm ID column as 2D in order to achieve enhanced GC  GC resolution. Fig. 5A and 5B illustrate the 2D GC–MS analysis for odour region 1 (nutty character) and region 3 (floral character) respectively in Fig. 4A using the developed MDGC approach. Peak i in Fig. 5A corresponds to the nutty odour, tentatively identified either as 2-ethyl-6-methyl pyrazine (CAS No. 13925-03-6), or 2-ethyl-5methyl pyrazine (CAS No. 13360-64-0) by its MS spectrum. Peak ii in Fig. 5B, responsible for the floral character, suffers from low detectability and some co-elution, causing identification

Fig. 3. GC  GC contour plot of (A) enlarged view of odour region 7 indicated in Fig. 1, (B) authentic E3PP standard.

Fig. 4. GC analysis of ground coffee volatiles using SPME sampling (1  PDC; refer to text) showing (A) 1D GC–FID result, and (B) GC  GC–FID contour plot. Dotted box indicates odour region which was H/C to 2DL (see Figs. 5 and 6 below).

Fig. 5. GC analysis of the 2DL result obtained from (A) the odour region 1, and (B) odour region 3 in coffee H/C from 1D (refer to Fig. 4 dotted region). Peak i and ii indicate the responsible compounds corresponding to the perceived odour character. Refer to text.

uncertainty for MS library matching (MS result not shown). Increasing the number of SPME samplings would improve compound detectability, but the current 2D GC strategy has insufficient resolution power for complete separation of the odour-active peak from the background in both cases. Therefore, a targeted approach to re-adjust separation conditions, such as a narrower H/C window, or alternative column phases or dimensions could be used in order to achieve resolution and solve this problem. Improving the separation power such as upgrading to 3D GC may be considered for resolving the complexity of coffee volatiles. An odour-active region with pronounced spicy, capsicum character (detected within a 0.3 min window, shown as the odour region 2 in Fig. 4A) in the 1D GC analysis was heart-cut and transferred to a longer length (i.e. 30 m) 2D VF-200 ms phase column for odorant verification by parallel MS and olfactory detection. Fig. 6A and B illustrate the 2D GC traces of the spicy, capsicum odour region H/C from 1D acquired using 1  PDC and 6  PDC cumulative sampling approaches respectively. Again, although the FID indicated only minor components, the better resolution of the H/C region on the 2D column allows many components to

360

S.-T. Chin et al. / Food Chemistry 185 (2015) 355–361

3.3. Utility of the integrated system

Fig. 6. GC analysis of the 2DL result obtained from the odour region 2 in coffee H/C from 1D (refer to Fig. 4 dotted region) using (A) 1  PDC sampling, (B) 6  PDC sampling. Asterisked peak identities matched to 2-methoxy-3-isobutylpyrazine: (C) mass spectrum of the asterisked peak in 6A, (D) mass spectrum of the asterisked peak in 6B.

be separated, and a much more sensitive response scale. For sample analysis using 1  PDC SPME sampling, the compound corresponding to the spicy, capsicum odour (asterisked peak at 2 tR 14.51 min in Fig. 5A) is well resolved from the background matrix, and could be tentatively identified as 2-methoxy-3-isobutylpyrazine (CAS 24683-00-9). However, the low abundance of this peak (signal-to-noise ratio, S/N = 9) provided only a 77% matching quality to the NIST MS database. The library matching quality was improved to >91% similarity (refer Fig. 6D) after cumulative sampling with 6-repeat samplings, due to higher ion intensity, and 5fold increase in S/N of the odorant (S/N = 60). The good correspondence of this approach is demonstrated by the set of 3 peaks at about 2tR 12.7 min, again showing about a 6+ fold increase in response. Despite the differences in ion intensity, both mass spectrum data sets obtained from 1  PDC (Fig. 6C) and 6  PDC (Fig. 6D) show little variation. It can be assumed that for even less abundant components, the multiple sampling approach will be of increasing advantage. By applying the cryogenic trapping approach for estimating retention indices in both separating dimensions, the 1 I and 2I data are 1461 (on SLB-IL59 phase) and 1243 (on VF-200 ms phase) respectively. At present, a library of I data on less common phases is lacking; RI information will be useful for compound identification as further work is conducted on similar column phases. The presence of 2-methoxy-3-isobutylpyrazine was reported previously in brewed coffee (Blank, Sen, & Grosch, 1992; Chin et al., 2011; Flament, 2002). However, fractionation of the volatile extract from brewed coffee was necessary prior to GC analysis, in order to discover this compound (Blank et al., 1992). Chin et al. (2011) distinguished this compound from volatile background matrix such as dihydro-2-methyl-3(2H)-thiophenone, 1-(2-furyl)2-propanone in 1D GC, by conducting GC  GC–TOFMS analysis. Flament (2002) discussed that 2-methoxy-3-isobutylpyrazine has a major contribution to the aroma impression of roasted coffee with an odour threshold below the ppb level.

Implementation of advanced separation tools for analysis of complex volatile samples is highly informative, but data correlation across 1D, MDGC and GC  GC systems can be difficult. Since the 1D GC-O/FID trace was generated from the same system, results obtained from alternative modes i.e. GC  GC and heart-cut 2DGC analysis can be precisely matched. Thus, the integrated MDGC system presented provides a convenient solution to overcome uncertainty of data correlation across multiple hyphenated systems that was applied in previous work (Chin et al., 2011; Eyres, Marriott, & Dufour, 2007; Rochat, Egger, & Chaintreau, 2009). A first column zone giving a detected olfactory response can be directly sampled into a 2D MDGC column for further separation, with olfactory assessment using the same olfactory detector. Since only a limited number of regions with components of interest can be analysed in a single analysis using heart-cut MDGC technique, multiple analyses may be required in order to fully characterise or determine the multiple aroma components in a complex sample mixture. The employment of GC  GC can be used for an overview and profiling of total sample complexity, which is valuable in its own right due to its simplicity, fast analytical process after the olfactometry confirmation of odour-active compounds, and information-rich compound-class composition. Tissot, Rochat, Debonneville, and Chaintreau (2012) extended the qualitative power of GC  GC resolution by demonstrating the advantage of GC  GC–FID for rapid quantification of volatile compounds in a mixture without requiring the use of authentic standards, through predicted response factors computed from known molecular formulas, which may be approached from partial response factors according to molecular functional groups. In addition, application of the cumulative SPME approach in combination with the system improves sensitivity and reduces discrimination in volatile extraction, as validated previously (Chin et al., 2012a). Nevertheless, headspace SPME showed limited applicability for some influential high-boiling aroma compounds such as lactones, whilst volatile polar compounds may result in tailing peaks (Weldegergis et al., 2011). Increasing extraction temperature during SPME extraction may facilitate the diffusion of organic compounds into the headspace, but heating the wine may create artefacts. Ultrasonication, or dynamic headspace with mild stress conditions e.g. lower extraction temperature, may be suggested for recovery of high-boiling and polar volatile analytes (Pawliszyn, 2009; San-Juan, Pet’ka, Cacho, Ferreira, & Escudero, 2010). Detection of 3-isobutyl-2-methoxypyrazine in Sauvignon Blanc wine was studied by Schmarr et al. using an online coupling of LC to MDGC analysis (Schmarr et al., 2010), who postulated that appropriate SPE volatile extraction should be conducted to remove interfering acid and neutral components prior to separation. By integrating MDGC and GC  GC techniques, improved online separation and signal amplification can be achieved, which greatly reduces losses of odour-impact compounds and avoids formation of artefacts that may occur during a fractionation process. In order to extend the successful analysis of the aroma-active compounds in highly complex matrices, novel analytical tools such as the hybrid GC  GC–MDGC concept (Mitrevski & Marriott, 2012) with three or more separation dimensions could be beneficial to identify compounds co-eluting in very complex regions. Identification of odorants is supported by a novel approach to attain RI directly on the 2D column by transferring alkane solutes to a cryotrap located at the inlet end of the 2D column.

S.-T. Chin et al. / Food Chemistry 185 (2015) 355–361

4. Conclusion Complex volatile analytes in food products causes analytical issues by using a single dimensional GC approach. For better investigation of the odour active compounds, multiple GC systems (GC  GC, heart-cut 2DGC) can be applied in conjunction to GC-O, but reliable data correlation across different platforms is difficult. An integrated GC  GC/MDGC system with FID/O/MS detection was applied in this work in order to re-investigate the headspace volatiles from wine and brewed coffee odour. By using cumulative SPME sampling coupled to such a system, tedious fractionation procedures may be omitted, whilst 11 odorants in selected co-eluting odour regions of Shiraz wine were identified tentatively, as well as 3 potent odour regions detected in the ground coffee headspace. Olfactometric data from GC-O was readily matched to MDGC-O/MS identification, utilising the developed integrated system with complementary GCGC–FID analysis giving an overview of the sample’s complexity. This on-line hyphenation of multiple GC approaches reduces the need for data correlation across various individual GC methods for aroma analysis. Higher detection sensitivity due to cryo-modulation, and greater separation efficiency was achieved on the secondary column, for selected heartcut odour-active regions. Furthermore, compound identification was demonstrated through verification by simultaneous matching of acquired mass spectrometry information, as well as 1I and 2I data. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foodchem.2015. 04.003. References Ando, Y., & Sasaki, T. (2011). GC separation of cis-eicosenoic acid positional isomers on an ionic liquid SLB-IL100 stationary phase. Journal of the American Oil Chemists’ Society, 88(6), 743–748. Beens, J., & Brinkman, U. A. T. (2005). Comprehensive two-dimensional gas chromatography – A powerful and versatile technique. Analyst, 130(2), 123–127. Blank, I., Sen, A., & Grosch, W. (1992). Potent odorants of the roasted powder and brew of Arabica coffee. Zeitschrift für Lebensmittel-Untersuchung und Forschung, 195(3), 239–245. Chin, S.-T., Eyres, G. T., & Marriott, P. J. (2011). Identification of potent odourants in wine and brewed coffee using gas chromatography-olfactometry and comprehensive two-dimensional gas chromatography. Journal of Chromatography A, 1218(42), 7487–7498. Chin, S.-T., Eyres, G. T., & Marriott, P. J. (2012a). Cumulative solid phase microextraction sampling for gas chromatography-olfactometry of Shiraz wine. Journal of Chromatography A, 1255, 121–127. Chin, S.-T., Eyres, G. T., & Marriott, P. J. (2012b). System design for integrated comprehensive and multidimensional gas chromatography with mass spectrometry and olfactometry. Analytical Chemistry, 84(21), 9154–9162. Clarke, R. J., & Bakker, J. (2004). Wine flavour chemistry. Oxford: Blackwell Publishing Ltd.. Davis, P. M., & Qian, M. C. (2011). Progress on volatile sulfur compound analysis in wine. Volatile sulfur compounds in food (Vol. 1068, pp. 93–115). American Chemical Society. Ebeler, S. E., & Thorngate, J. H. (2009). Wine chemistry and flavor: Looking into the crystal glass. Journal of Agricultural and Food Chemistry, 57(18), 8098–8108. Eyres, G. T., Marriott, P. J., & Dufour, J.-P. (2007). Comparison of odor-active compounds in the spicy fraction of hop (Humulus lupulus L.) essential oil from four different varieties. Journal of Agricultural and Food Chemistry, 55(15), 6252–6261. Falcao, L. D., Lytra, G., Darriet, P., & Barbe, J. C. (2012). Identification of ethyl 2hydroxy-4-methylpentanoate in red wines, a compound involved in blackberry aroma. Food Chemistry, 132(1), 230–236. Fedrizzi, B., Magno, F., Finato, F., & Versini, G. (2010). Variation of some fermentative sulfur compounds in Italian ‘‘Millesimè’’ classic sparkling wines during aging and storage on lees. Journal of Agricultural and Food Chemistry, 58(17), 9716–9722.

361

Fedrizzi, B., Zapparoli, G., Finato, F., Tosi, E., Turri, A., Azzolini, M., et al. (2011). Model aging and oxidation effects on varietal, fermentative, and sulfur compounds in a dry botrytized red wine. Journal of Agricultural and Food Chemistry, 59(5), 1804–1813. Ferreira, V., López, R., & Cacho, J. F. (2000). Quantitative determination of the odorants of young red wines from different grape varieties. Journal of the Science of Food and Agriculture, 80(11), 1659–1667. Flament, I. (2002). Coffee flavor chemistry. West Sussex: John Wiley & Sons. Marriott, P. J., Chin, S.-T., Maikhunthod, B., Schmarr, H.-G., & Bieri, S. (2012). Multidimensional gas chromatography. TrAC Trends in Analytical Chemistry, 34, 1–21. Marriott, P. J., Eyres, G. T., & Dufour, J.-P. (2009). Emerging opportunities for flavor analysis through hyphenated gas chromatography. Journal of Agricultural and Food Chemistry, 57(21), 9962–9971. Mitrevski, B., & Marriott, P. J. (2012). Novel hybrid comprehensive 2D – Multidimensional gas chromatography for precise, high-resolution characterization of multicomponent samples. Analytical Chemistry, 84(11), 4837–4843. Mjøs, S. A. (2005). Properties of trans isomers of eicosapentaenoic acid and docosahexaenoic acid methyl esters on cyanopropyl stationary phases. Journal of Chromatography A, 1100(2), 185–192. Moreira, N., de Pinho, P. G., Santos, C., & Vasconcelos, I. (2011). Relationship between nitrogen content in grapes and volatiles, namely heavy sulphur compounds, in wines. Food Chemistry, 126(4), 1599–1607. National Institute of Standards and Technology. (2013). Pawliszyn, J. (2009). Development of SPME devices and coatings. In J. Pawliszyn (Ed.), Handbook of solid phase microextraction (pp. 55–85). Beijing: Chemical Industry Press. Pons, A., Lavigne, V. R., Eric, F. R., Darriet, P., & Dubourdieu, D. (2008). Identification of volatile compounds responsible for prune aroma in prematurely aged red wines. Journal of Agricultural and Food Chemistry, 56(13), 5285–5290. Qian, M. C., Burbank, H. M., & Wang, Y. (2007). Preseparation techniques in aroma analysis. In R. Marsili (Ed.), Sensory directed flavor analysis (pp. 111–154). Boca Raton: CRC Press. Robinson, A. L., Boss, P. K., Heymann, H., Solomon, P. S., & Trengove, R. D. (2011). Development of a sensitive non-targeted method for characterizing the wine volatile profile using headspace solid-phase microextraction comprehensive two-dimensional gas chromatography time-of-flight mass spectrometry. Journal of Chromatography A, 1218(3), 504–517. Rochat, S., Egger, J., & Chaintreau, A. (2009). Strategy for the identification of key odorants: Application to shrimp aroma. Journal of Chromatography A, 1216, 6424–6432. San-Juan, F., Pet’ka, J., Cacho, J., Ferreira, V., & Escudero, A. (2010). Producing headspace extracts for the gas chromatography-olfactometric evaluation of wine aroma. Food Chemistry, 123(1), 188–195. Schmarr, H.-G., Ganß, S., Koschinski, S., Fischer, U., Riehle, C., Kinnart, J., et al. (2010). Pitfalls encountered during quantitative determination of 3-alkyl-2methoxypyrazines in grape must and wine using gas chromatography–mass spectrometry with stable isotope dilution analysis. Comprehensive twodimensional gas chromatography–mass spectrometry and on-line liquid chromatography–multidimensional gas chromatography–mass spectrometry as potential loopholes. Journal of Chromatography A, 1217(43), 6769–6777. Semmelroch, P., & Grosch, W. (1996). Studies on character impact odorants of coffee brews. Journal of Agricultural and Food Chemistry, 44, 537–543. Shimoda, M., & Shibamoto, T. (1990). Isolation and identification of headspace volatiles from brewed coffee with an on-column GC/MS method. Journal of Agricultural and Food Chemistry, 38(3), 802–804. Siebert, T. E., Wood, C., Elsey, G. M., & Pollnitz, A. P. (2008). Determination of rotundone, the pepper aroma impact compound, in grapes and wine. Journal of Agricultural and Food Chemistry, 56(10), 3745–3748. Simpson, R. F., Capone, D. L., & Sefton, M. A. (2004). Isolation and identification of 2Methoxy-3,5-dimethylpyrazine, a potent musty compound from wine corks. Journal of Agricultural and Food Chemistry, 52(17), 5425–5430. Sumby, K. M., Grbin, P. R., & Jiranek, V. (2010). Microbial modulation of aromatic esters in wine: Current knowledge and future prospects. Food Chemistry, 121(1), 1–16. The pherobase: Database of pheromones and semiochemicals. (2013). Available at: http://www.pherobase.com. Tissot, E., Rochat, S., Debonneville, C., & Chaintreau, A. (2012). Rapid GC–FID quantification technique without authentic samples using predicted response factors. Flavour and Fragrance Journal, 27, 290–296. Tranchida, P. Q., Purcaro, G., Conte, L., Dugo, P., Dugo, G., & Mondello, L. (2009). Enhanced resolution comprehensive two-dimensional gas chromatography applied to the analysis of roasted coffee volatiles. Journal of Chromatography A, 1216(43), 7301–7306. Weldegergis, B. T., Crouch, A. M., Górecki, T., & de Villiers, A. (2011). Solid phase extraction in combination with comprehensive two-dimensional gas chromatography coupled to time-of-flight mass spectrometry for the detailed investigation of volatiles in South African red wines. Analytica Chimica Acta, 701(1), 98–111. Welke, J. E., Manfroi, V., Zanus, M., Lazarotto, M., & Alcaraz Zini, C. (2012). Characterization of the volatile profile of brazilian merlot wines through comprehensive two dimensional gas chromatography time-of-flight mass spectrometric detection. Journal of Chromatography A, 1226, 124–139.