Food Chemistry 142 (2014) 262–268

Contents lists available at ScienceDirect

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

Short communication

Elucidation of the volatile composition of Marsala wines by using

comprehensive two-dimensional gas chromatography
Giacomo Dugo a, Flavio A. Franchina b, Maria R. Scandinaro a, Ivana Bonaccorsi b, Nicola Cicero a,
Peter Q. Tranchida b,⇑, Luigi Mondello b,c
a
Dipartimento di Scienze dell’Ambiente, della Sicurezza, del Territorio, degli Alimenti e della Salute (S.A.S.T.A.S.), Università di Messina, Viale Ferdinando d’Alcontres 31, 98166
Messina, Italy
b
Dipartimento di Scienze del Farmaco e dei Prodotti per la Salute, Università di Messina, Viale Annunziata, 98168 Messina, Italy
c
Centro Integrato di Ricerca (C.I.R.), Università Campus-Biomedico, Via Álvaro del Portillo, 21, 00128 Roma, Italy

a r t i c l e i n f o a b s t r a c t

Article history: The present contribution describes a research work focused on the elucidation of the composition of the
Received 28 March 2013 headspace of Marsala wine. Four sample-types, of different ageing (‘‘fine’’, ‘‘superiore secco’’, ‘‘superiore
Received in revised form 9 July 2013 riserva’’, ‘‘vergine’’) were subjected to headspace solid-phase microextraction-comprehensive 2D GC
Accepted 12 July 2013
analysis. At the outlet of the second GC dimension, the eluting analytes were split between a flame
Available online 20 July 2013
ionisation detector (for relative quantification purposes) and a rapid-scanning quadrupole mass
spectrometer (for compound identification). Over 500 peaks were detected in each application, with a
Keywords:
total of 128 compounds tentatively-identified considering the four sample types (mainly esters, alcohols,
Marsala wine
Comprehensive two-dimensional gas
ketones, and aldehydes). The results attained open a door on the highly complex nature of the Marsala
chromatography headspace; furthermore, they also demonstrated that the use of one-dimensional GC technologies, for
SPME the untargeted analysis of complex aroma profiles (e.g., dessert wines), is often too much of an analytical
Mass spectrometry challenge.
Flame ionisation detector Ó 2013 Elsevier Ltd. All rights reserved.
Volatiles

1. Introduction In general, the consumption of foods and drinks is tightly related

Marsala wine, or simply ‘‘Marsala’’, is a well-known, highly
appreciated and economically important dessert wine, produced
exclusively in Sicily (Trapani province). Marsala is exported all over
the world and is considered one of the four most important dessert
wines together with Madeira, sherry and port. Marsala is charac-
terised by an average alcoholic v/v content of 18%, and comes in
three colours, namely ‘‘oro’’ (gold), ‘‘ambra’’ (amber) and ‘‘rubino’’
(ruby). The type of colour is dependent on the variety of grapes
employed (Inzolia, Damaschino, Nerello Mascalese, etc.). Apart
from colour, Marsala is classified by the degree of ageing and sugar
content. Considering the former characteristic, Marsala can be:
‘‘fine’’ (above 1 year), ‘‘superiore’’ (above 2 years), ‘‘superior riser-
va’’ (above 4 years), ‘‘vergine’’ (above 5 years), and ‘‘stravecchio’’
(above 10 years). With regards to the concentration of reducing
sugars, the groups are: ‘‘secco’’ (below 40 g/L), ‘‘semi-secco’’ (be-
tween 40 and 100 g/L), and ‘‘dolce’’ (above 100 g/L) (La Torre
et al., 2008).
⇑ Corresponding author. Tel.: +39 090 6766510; fax: +39 090 358220.
E-mail address: ptranchida@unime.it (P.Q. Tranchida).
to the stimulation of the human senses, odour and taste. Food fla-
vour, along with texture and appearance, has a fundamental impor-
tance in the attraction of the consumer towards a particular food.
With respect to odour, this sensation is generated by highly com-
plex mixtures of volatile molecules (defined generically as aroma),
in a variety of concentrations (Fisher & Scott, 1997). Currently, the
analytical technique of choice for the untargeted elucidation of ar-
oma profiles, in foods and drinks, is one-dimensional gas chroma-
tography (1D-GC), hyphenated to a mass spectrometer (MS). It is
obvious that GC–MS approaches can give qualitative (and possibly
quantitative) information, but nothing on the odour sensation gen-
erated by a specific analyte. Such important information can be at-
tained through GC-olfactometry (GC-O) (d’Acampora Zellner et al.,
2008). However, one of the main problems related to the 1D-GC
analysis of aromas, with either MS and/or olfactometric detection,
is that such samples are excessively complex for a single GC col-
umn. The main consequence of insufficient separation power is
that, often, compounds co-elute at the column outlet.
The most suitable GC technique for the untargeted analysis of
highly complex samples (>200 solutes) is comprehensive 2D-GC
(GC  GC), an approach introduced over twenty years ago (Liu &
Phillips, 1991). GC  GC analyses are usually carried out on a

0308-8146/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.foodchem.2013.07.061
 G. Dugo et al. / Food Chemistry 142 (2014) 262–268
conventional (i.e., 30 m  0.25 mm ID) non-polar column, followed
by a more polar micro-bore capillary column (i.e., 1–2 m  0.10 mm
ID). The transfer device (‘‘modulator’’) operates continuously
throughout the analysis and enables the trapping, and then re-injec-
tion of millisecond-wide chromatography bands. GC  GC experi-
ments are visualised in a 2D manner, by using dedicated software.
The main benefits of cryogenic GC  GC approaches, over one-
dimensional GC methods, are enhanced sensitivity (peak focusing),
selectivity (use of two different stationary phases), and peak capac-
ity; an additional advantage is represented by the organised elution
patterns of homologous series of compounds (i.e., linear alkanes,
fatty acid methyl esters, etc.), a feature which enables reliable ana-
lyte identification, even in the absence of a mass spectrometer. For
more detailed information on the technique, the reader is directed
to the literature (Cortes, Winniford, Luong, & Pursch, 2009).
In general, the use of GC  GC has enabled a much deeper in-
sight on the true composition in volatiles of real-world samples.
The aroma of alcoholic beverages, in particular wine, has been
often analysed by using GC  GC, using headspace solid-phase
microextraction (HS-SPME), and time-of-flight MS detection (Rob-
inson, Boss, Heymann, Solomon, & Trengove, 2011; Weldegergis,
Crouch, Górecki, & de Villiers, 2011a; Weldegergis et al., 2011b;
Welke, Manfroi, Zanus, Lazarotto, & Alcaraz Zini, 2012). In such
studies, the complex nature of the headspace of wines was
highlighted.
In the present research, an HS SPME-GC  GC-FID/MS method
was developed for the analysis of Marsala. The FID trace was
exploited for quantification purposes, while the MS data was used
for identification. To the best of the present authors’ knowledge, no
GC  GC research has been performed previously on Marsala
wines, and only a single paper published (‘‘vergine’’ Marsala) using
GC–MS (Di Stefano, 1985).
2. Experimental
2.1. Samples
Four types of Marsala, namely ‘‘fine’’, ‘‘superiore secco’’, ‘‘supe-
riore riserva dolce’’, and ‘‘vergine’’, were kindly donated by a pro-
ducer located in Marsala (Sicily, Italy). The bottles were stored at
ambient temperature, in the dark, prior to analysis.
A C7–C30 alkane mixture, for linear retention index (LRI) calcu-
lations, was kindly supplied by Supelco (Milan, Italy).
2.2. HS SPME process
The SPME fibres (Supelco) evaluated in the present study were:
triple phase divinylbenzene (DVB)/Carboxen (CAR)/poly-
dimethylsiloxane (PDMS) (50/30 lm), PDMS (100 lm), CAR/PDMS
(75 lm). A Shimadzu AOC-5000 autosampler (Kyoto, Japan) was
used for the HS-SPME operations.
Briefly, 4 mL of Marsala were introduced in a 20-mL vial. The fi-
bre was exposed in the headspace for 30 min, at ambient labora-
tory temperature (25–26 °C). During the extraction the vial was
agitated (clockwise-anticlockwise alternate rotation) at 500 rpm.
After this process, the fibre was thermally desorbed in the GC
injection port for 1.0 min at 270 °C, in splitless mode (after
1 min, a 10:1 split ratio was applied).
2.3. GC  GC-FID/MS conditions
The GC  GC–MS/FID applications were carried out on a Shima-
dzu GC  GC system, consisting of two independent GC2010 gas
chromatographs and a QP2010 Ultra quadrupole mass spectrome-
ter (qMS) (Kyoto, Japan). The primary GC was equipped with an
263
AOC-20i auto-injector, a split-splitless injector (270 °C), and a
cable extension for the MS connection (due to the presence of
the second oven). The first column was connected by using a SGE
SilTite mini-union (Ringwood, Victoria, Australia) to an uncoated
tubing (1 m  0.25 mm ID), that was passed through a heated
transfer line (280 °C) into the second oven, where a dual-stage
loop-type modulator (under licence from Zoex Corporation, Hous-
ton, TX) system was installed, and was finally connected to the sec-
ondary column by using a fixed outlet capillary column splitter
(SGE); the latter was then linked to a 1 m  0.10 mm ID  0.10 lm
df (for FID analysis) and to a 1.5 m  0.10 mm ID  0.10 lm df col-
umn (for MS analysis). In both cases, a Supelcowax-10 (100% poly-
ethylene glycol) stationary phase was employed (Supelco).
Cryogenic modulation was applied every 5 s, with a hot jet
(340 °C) duration of 375 ms. The first column was an SLB-5 ms
30 m  0.25 mm ID  0.25 lm df column (silphenylene polymer
virtually equivalent in polarity to poly (5% diphenyl/95% meth-
ylsiloxane); Supelco). Carrier (He) pressure: 150 kPa (constant lin-
ear velocity). Temperature program (equivalent in both ovens):
from 50 °C (1 min) to 280 °C, at 3 °C/min.
MS parameters: the sample was analysed in full scan mode,
with a scan speed of 10,000 amu/s, a mass range of m/z 40–360,
and a sampling frequency of 25 spectra/s; interface and ion source
temperatures were 250 and 200 °C, respectively. MS ionisation
mode was electron impact ionisation. Data were acquired by using
the GCMSsolution software ver. 4.0, while the MS database was the
FFNSC 2.0 (Shimadzu). Bidimensional chromatograms were gener-
ated by using the ChromSquare software ver. 1.6 (Shimadzu Eur-
ope, Duisberg, Germany).
FID parameters: temperature was 280 °C; acquisition fre-
quency: 125 Hz; gases: make-up (He): 40 mL/min; H2: 40 mL/
min; air: 400 mL/min.
3. Results and discussion
3.1. SPME method optimisation
Initially, the SPME operation conditions were optimised consid-
ering fibre stationary phase, extraction time, and desorption time.
The three extraction phases were evaluated at two times (15 and
45 min), and using a fibre desorption time of 5 min (270 °C). With
regards to the extraction temperature, no additional heating was
used because the objective of the research was to provide a profile
of Marsala volatiles, which is best performed at ambient tempera-
ture. Heating could potentially accelerate the extraction period,
but would also alter the ‘‘normal’’ composition of Marsala head-
space (Marsala is consumed at ambient temperature), and, conse-
quently, the HS SPME-GC  GC fingerprint would not faithfully
represent that of the Marsala volatile composition.
With regards to the fibre selectivity, the heterogenous composi-
tion (in terms of polarity) of Marsala headspace must be consid-
ered. In fact, Marsala volatiles range from acids, to alcohols, onto
esters, aldehydes and ketones, down to hydrocarbons. The PDMS
liquid polymer showed a poor coverage for the more polar volatiles
(i.e., alcohols, acids, aldehydes); on the other hand, the mixed DVB/
CAR/PDMS and CAR/PDMS fibres, which consist of a porous solid
and liquid polymer, did not discriminate between low and high-
polarity volatiles. However, because CAR (particle size: 2–10 lm)
is characterised by smaller pores with respect to DVB, the CAR/
PDMS fibre gave an excellent performance towards the lower
MW volatiles, and was much less selective towards the higher
MW ones (>200 amu). Overall, it was the DVB/CAR/PDMS phase
that provided the best coverage in terms of polarity and analyte
MW.
After defining the most appropriate fibre, four different extrac-
tion periods were evaluated, namely 15, 30, 45, 60 and 75 min. It
264 G. Dugo et al. / Food Chemistry 142 (2014) 262–268
was found that the lower MW analytes reached an equilibrium at
30 min, while 60 min were required for the more heavier volatiles
(responses did not substantially increase at 75 min). However,
complete equilibrium was not necessary due to the excellent
sensitivity enhancement of GC  GC. Consequently, as a good com-
promise, an extraction period of 30 min was chosen.
Desorption periods of 1, 2 and 5 min were evaluated, using the
previously optimised SPME conditions. It was found that signifi-
cant differences did not occur using the three different periods,
and so a 1-min desorption time was employed in all applications.
3.2. GC  GC–MS/FID analyses
With regards to the GC  GC–MS/FID method, a ‘‘classical’’ non-
polar/polar column combination was employed. The column seg-
ments (see Section 2.3) connected to the detectors guaranteed a
50:50 split. The MS system was operated at 25 Hz and generated
a sufficient number of data points/peak for quantification (at least
10), if desired. However, MS analyte response factors can vary
greatly and so such instruments are generally not used to derive
peak area% information. In this respect, FID response factors vary
less, and hence such a detector can be exploited to attain such
semi-quantitative data. The FID was operated at 125 Hz, and re-
constructed each GC  GC peak (normally) with at least 50 data
points.
An HS SPME-GC  GC–MS fingerprint, relative to a Marsala
‘‘superiore secco’’, is shown in Figs. 1–3 (initial, middle and final
part of the chromatogram). The complex nature of the Marsala
headspace is evident, with over 500 compounds detected. Tenta-
tive analyte identification (Table 1) was performed through MS
database searching, with the application of two filters: one based
on a minimum spectral similarity value (75%), and the other on
experimental/database linear retention indices (Table 1). In the lat-
ter case, MS database matches with an LRI value in excess of ±15
units, compared to the experimental LRI value of the unknown,
were automatically deleted. LRI values were calculated in a one-
dimensional manner, as previously reported (Purcaro et al.,
2009). The application of a wide LRI window was necessary be-
cause the contribution of the secondary (‘‘wax’’) column, towards
the retention of the more polar compounds, must be considered.
Fig. 1. Full-scan HS SPME-GC  GC–MS chromatogram expansion (initial part) of Marsala ‘‘superiore secco’’. For peak identification see Table 1.
The relatively low similarity filter value was used to avoid the
omission of low-concentration analytes, which notably are charac-
terised by reduced similarity matches. Overall, 100 compounds
were tentatively-identified in the ‘‘superiore secco’’ sample (Ta-
ble 1), which could appear to be a low number considering the high
number of analytes detected. Such an occurrence can be related to
two reasons: (I) the low signal-to-noise ratios of many peaks; (II)
the absence of the target spectra in the MS database. The three
chromatogram expansions, reported in Figs. 1–3, illustrate the
positions of the components identified in the ‘‘superiore secco’’
sample. Considering all four samples, 128 different compounds
were tentatively identified. Ethanol (apart from water, the most
abundant Marsala constituent) was not included as it obviously
eluted within the MS solvent cut-off time. With regards to the
other three samples, 87 compounds were identified in ‘‘fine’’, 91
in ‘‘superiore riserva’’, and 89 in ‘‘vergine’’. An abundant low-
molecular weight constituent, namely ethyl propanoate (peak 1),
was poorly entrapped by the cryogenic modulator (see Fig. 1).
The poor entrapment caused difficulties in reliable quantification
(using dedicated GC  GC software), and so the FID% area was
not derived for the constituent. Considering the 128 compounds,
the most abundant were the esters (fifty-seven – 44.5%), followed
by alcohols (twenty – 15.6%), ketones (thirteen – 10.2%), and alde-
hydes (twelve – 9.4%).
Prior to discussion on quantification values, it must be noted
that HS SPME-extracted analytes will give a good, but not an en-
tirely faithful representation of the true headspace composition.
In this respect, the FID% areas relate to the SPME fibre uptake.
However, the data herein reported does give a good idea of the
headspace composition. True% values could be potentially attained
via calibration, a highly complicated procedure considering the
complexity of the ‘‘Marsala’’ matrix.
Table 2 lists a series of Marsala constituents, with an FID% area
of over 1% in at least one of the samples. The value of 100% refers to
the sum of the peak areas of the compounds identified (e.g., 87
compounds were considered in Marsala fine). Intra-day precision
(n = 3), for the analytes listed in Table 2 and for all four samples,
was satisfactory with relative standard deviation values always
lower than 20%. The most abundant ester (and compound in gen-
eral) was found to be ethyl octanoate (peak 85), and decreased
 G. Dugo et al. / Food Chemistry 142 (2014) 262–268 265

Fig. 2. Full-scan HS SPME-GC  GC–MS chromatogram expansion (middle part) of Marsala ‘‘superiore secco’’. For peak identification see Table 1.

Fig. 3. Full-scan HS SPME-GC  GC–MS chromatogram expansion (final part) of Marsala ‘‘superiore secco’’. For peak identification see Table 1. Inset refers to boxed area, and
was extracted at a lower signal intensity.

from the youngest to the oldest Marsala: 40.9%, 37.5%, 31.2%, and
19.4%. The presence of ethyl octanoate has been reported in Ma-
deira wine [Campo, Ferreira, Escudero, Marqués, & Cacho, 2006
(purge-and-trap extraction], in red wine [Feirrera, López, & Cacho,
2000 (solvent extraction)], and in papaya wine [Pino & Queris,
2012 (solvent extraction)]. In the last two investigations, ethyl
octanoate was reported to have a fruity odour and a (low) thresh-
old of 5 lg/L. Another ester, namely diethyl succinate (peak 81),
presented an opposite behaviour, inasmuch that the concentration
increased from the youngest to the oldest Marsala: 2.7%, 4.9%, 5.7%,
and 9.2%. Diethyl succinate has been described in Madeira wine
(Campo et al., 2006), although considered as a constituent with a
low odour activity. Ethyl butyrate (peak 7) and isoamyl acetate
(peak 15), which were also found in by far the highest amounts
in the ‘‘vergine’’ sample, have also been reported in Madeira wine
(Campo et al., 2006), the first with a fruity odour and a threshold of
20 lg/L, and the second with a banana odour and a threshold of
30 lg/L. Another ester found in Madeira wine (Campo et al.,
2006), namely ethyl hexanoate (peak 36; fruity odour/threshold:
14 lg/L), was found in the lowest amounts in the ‘‘vergine’’ sample
(9.7%).
The most abundant alcohol (apart from ethanol) was found to
be isopentyl alcohol (peak 2), measured in comparable amounts
in all sample-types (17.9–21.0% range). In HS SPME-GC-O applica-
tions on Spanish red wines (Martí, Mestres, Sala, Busto, & Guach,
2003), isopentyl alcohol was found to give a fundamental
266 G. Dugo et al. / Food Chemistry 142 (2014) 262–268

Table 1
Peak identification and experimental LRI values relative to the four Marsala samples (in parenthesis database LRI values).

Peak Compound Fine Superiore secco Superiore riserva Vergine

1 Ethyl propanoate 710 710(708) 708 709
2 Isopentyl alcohol 736 735(729) 732 733
3 sec-Butyl carbinol 737 742(741) 739 737
4 Ethyl isobutyrate 753 753(752) 753 952
5 Isobutyl acetate 770 770(768) – 768
6 Isoamyl formate 794 791(791) – –
7 Ethyl butyrate 802 802(803) 801 802
8 Ethyl lactate 816 816(814) 812 815
9 Ethyl pyruvate 807(807) – – –
10 Furfural 840 837(845) 836 836
11 Ethyl (E)-crotonate 842 842(839) – –
12 Ethyl 2-methylbutanoate 846 845(842) 844 845
13 Ethyl isovalerate 847 850(850) 848 850
14 n-Hexanol 868 872(867) 868 870
15 Isoamyl acetate 875 875(871) 872 875
16 2-Methylbutyl acetate – 877(873) 874 –
17 2-Heptanone 892 893(898) 894 892
18 2-Butylfuran 899 899(890) – –
19 Ethyl pentanoate 901 901(899) 900 900
20 2-Heptanol – 905(913) – 905
21 n-Heptanal 904 904(906) 902 904
22 2-Acetylfuran 913 913(913) – –
23 Methyl hexanoate 923 923(922) – –
24 2-Methyl-3-heptanone – – 927(934) 932
25 Hexyl formate – – 928 (929) –
26 Hois de rose oxide – – 969 (968) –
27 1,1-Diethoxy-3-methylbutane 948 948(946) 948 948
28 Benzaldehyde 964 965(960) 964 965
29 Isoamyl propanoate – – – 967(965)
30 Vinyl amyl carbinol 979 980(978) 979 979
31 3-Octanone 984 985(986) 985 –
32 2-Octanone – – – 989(989)
33 trans-5-Isopropenyl-2-methyl-2-vinyl-tetrahydrofuran 990 990(989) 990 990
34 Mesitylene – – – 994 (994)
35 Delta-2-carene – 996(1000) 997 997
36 Ethyl hexanoate 998 998(1003) 999 999
37 Ethyl (3Z)-hexenoate – – 1002 (1003) –
38 n-Octanal 1004 1003(1006) – 1004
39 cis-Dehydrolinalool oxide – 1005(1006) 1005 1005
40 Isoamyl isobutyrate 1011 1010(1014) – 1011
41 Hexyl acetate 1012 1011(1012) 1011 1012
42 1,4-Cineole – 1016(1017) – –
43 Pseudocumene 1023 1023(1020) – 1023
44 p-Cymene 1026 1025(1025) 1025 1025
45 Eucalyptol – – 1033 (1032) –
46 2,2,6-Trimethylcyclohexanone – – 1035 (1035) 1035
47 2-Ethyl-1-hexanol 1029 1029(1030) 1029 1030
48 Ethyl (2E)-hexenoate 1042 1042(1042) 1042 1042
49 2,2-Dimethyl-5-(1-methylpropenyl)tetrahydrofuran 1044 1044(1045) 1044 –
50 Phenylacetaldehyde 1046 1046(1045) 1045 1045
51 Ethyl 2-furoate 1052 1052(1053) 1050 1053
52 Isoamyl butyrate 1054 1054(1057) 1054 1054
53 Ethyl 3-acetylpropionate – 1061(1065) – 1061
54 Diethyl malonate – 1067(1071) 1067 –
55 Butanoic acid,1-methyl-2-oxopropyl ester 1072 1071(1066) 1069 1071
56 cis-Linalool oxide – 1072(1069) 1071 1070
57 3-Nonanone 1084 1084(1088) 1084 1084
58 cis-Hex-3-enal diethylacetal – 1088(1088) 1088 1088
59 Terpinolene 1089 1089(1086) – –
60 tr-Linalool oxide 1087(1086) – – –
61 2-Nonanone 1090 1090(1090) 1090 1090
62 1,3-Butanediol diacetate – 1090(1087) – –
63 Ethyl heptanoate 1097 1097(1101) 1095 1097
64 Ethyl sorbate – 1099(1102) 1099 1099
65 Linalool 1100 1100(1101) 1100 1101
66 3-Methylbutyl 2-methylbutyrate 1105 1105(1104) – 1102
67 n-Nonanal 1106 1106(1107) 1105 1104
68 a-Thujone – – 1108 (1100) –
69 n-Heptanoic acid – – 1113 (1116) 1114
70 Sotolone – – 1114 (1115) –
71 5-Methyl-5-nonanol – 1112(1107) – –
72 1,2-Heptanediol – 1114(1114) – –
73 Phenethyl alcohol 1119 1117(1113) 1117 1118
74 methyl octanoate 1123 1123(1125) 1123 1123
 G. Dugo et al. / Food Chemistry 142 (2014) 262–268 267

Table 1 (continued)

Peak Compound Fine Superiore secco Superiore riserva Vergine

75 Ethyl 3,5,5-trimethylhexanoate – – 1147 (1147) 1147
76 2-Ethylhexyl acetate 1148 1148(1150) – -
77 Nerol oxide 1153 1153(1152) 1153 1153
78 3,3-Dimethylcyclohexyl methyl ketone – – 1146 (1142) –
79 Linalool ethyl ether 1167 1167(1166) 1167 1167
80 Ethyl benzoate 1174 1174(1170) 1172 1172
81 Diethyl succinate 1179 1179(1183) 1179 1183
82 1-Decen-3-ol – – 1186 (1190) –
83 Pelargol – – 1194 (1200) –
84 Ethyl (4E)-octenoate 1188 1188(1189) – 1188
85 Ethyl octanoate 1198 1198(1202) 1199 1200
86 a-Terpineol 1201 1200(1195) 1199 –
87 n-Decanal 1207 1208(1208) 1208 1208
88 Carvacryl methyl ether – – 1233 (1239) –
89 Ethyl phenylacetate 1245 1246(1246) 1246 1246
90 Ethyl (2E)-octenoate – 1247(1251) 1247 –
91 Isoamyl hexanoate 1250 1251(1252) 1251 1250
92 2-Phenylethyl acetate 1256 1258(1257) – 1258
93 trans-myrtanol 1261 1262(1270) 1262 1262
94 2-Phenyl crotonaldehyde – – – 1273
95 4-Ethylguaiacol 1278 1278(1275) – 1278
96 Tetrahydrolavandulyl acetate – 1279(1272) 1279 1280
97 d-Octalactone – – 1289 (1280) 1289
98 Neryl formate – 1282(1276) – –
99 Ethyl nonanoate 1295 1297(1297) 1295 1297
100 Nonanoic acid 1291(1289) – – –
101 n-Undecanal 1308 1309(1309) – 1312
102 trans-Tetrahydrojasmone 1313 1316(1310) 1316 1316
103 Whisky lactone – 1326(1325) 1324 1326
104 Methyl decanoate 1327 1327(1327) 1327 –
105 (E)-b-damascenone 1383 1384(1379) 1383 1383
106 Ethyl trans-4-decenoate 1387 1387(1380) 1387 1387
107 n-Undecanol – – 1388 (1379) –
108 Ethyl decanoate 1394 1395(1399) 1398 1395
109 n-Dodecanal 1410 1410(1410) 1410 1410
110 Isoamyl octanoate 1445 1445(1449) 1445 1445
111 n-Tridecanal 1511 1511(1516) 1511 1511
112 Ethyl dodecanoate 1594 1594(1598) 1594 1594
113 Dodecyl acetate – – – 1611
114 n-Tetradecanal 1613 1613(1614) 1613 1613
115 Methyl dihydrojasmonate 1651 1650(1642) 1649 1651
116 n-Tetradecanol 1681 (1680) – – –
117 n-Pentadecanol 1782 (1784) – – –
118 Tridecyl methyl ketone – 1698(1697) – –
119 Ethyl tetradecanoate 1793 1793(1794) 1793 1793
120 1-Octadecene 1800(1793) – – –
121 Isopropyl tetradecanoate 1824 1824(1826) 1824 1824
122 Farnesyl acetate 1834 1834(1832) – –
123 Phytone 1842 1842(1841) 1841 1842
124 Methyl hexadecanoate – 1926(1925) 1926 1926
125 n-Hexadecanoic acid 1977 1977(1977) 1977 1977
126 Ethyl palmitate – 1992(1993) 1992 1992
127 Isopropyl hexadecanoate 2023 2022(2023) 2022 2022
128 Octadecyl acetate 2209(2212) – – –

contribution to the aroma, giving chemical notes. With regards to
another alcohol, phenethyl alcohol (peak 73), this compound was
found in the highest amounts in the ‘‘vergine’’ headspace (3.4%).
Phenethyl alcohol was found in Spanish red wines (Martí et al.,
2003), and was characterised by a rose odour. The same alcohol
has also been found as a major compound in papaya wine (Pino
& Queris, 2012), and Madeira wine (Campo et al., 2006).
The aldehyde furfural (peak 10), described as ‘‘paper, green
fruits’’, was found in much higher amounts in the ‘‘superior riser-
va’’ (2.6%) and ‘‘vergine’’ (2.9%) samples, and has been related to
ageing in Madeira wines [Sousa Câmara, Marques, Alves, & Silva
Ferreira, 2004 (solvent extraction)].
It is worthy of note that all the compounds listed in Table 2,
except for 2-ethyl-1-hexanol (peak 47), were also identified in
HS SPME-GC  GC–MS (a time-of-flight system was used)
experiments, performed on 9 Pinotage wines (Weldegergis et al.,
2011b). Moreover, the analytes increased to two (furfural, sec-bu-
tyl carbinol) if the previously-described GC–MS analysis of
‘‘vergine’’ Marsala is considered (Di Stefano, 1985). However, a
direct comparison with that work is difficult, because the sample
preparation method was entirely different and laborious. Specifi-
cally, three solvent mixtures were used for extraction, with each
fraction (divided on the basis of polarity) then subjected to
GC–MS analysis (a 20 m  0.2 mm ID polyethylene glycol capillary
was employed).
With regards to constituents quantified at levels <1%, some are
characterised by a potent odour activity. For example, sotolon
(peak 70) [3-hydroxy-4,5-dimethyl-2(5H)-furanone] was found at
the 0.02% level in the ‘‘superior riserva’’ and has been found in
Madeira (Campo et al., 2006; Sousa Câmara et al., 2004) and port
[Silva Ferreira, Barbe, & Bertrand, 2003 (solvent extraction)] wines,
and described as ‘‘nutty and spice-like’’, with a sensory threshold
of 19 lg/L. Beta-damascenone (peak 105) [(E)-1-(2,6,6-trimethyl-
1-cyclohexa-1,3-dienyl)but-2-en-1-one] was found in all four
268 G. Dugo et al. / Food Chemistry 142 (2014) 262–268

Table 2 deal of precious information, even though GC-olfactometry data is

Relative % peak areas for the main compounds found in Marsala. equally important. In this respect, a heart-cutting multidimen-
Compound Fine Superiore secco Superiore riserva Vergine sional GC-O approach would be advisable.
Isopentyl alcohol 18.8 19.5 17.9 20.9
sec-Butyl carbinol 5.7 5.5 1.9 11.2 References
Ethyl butyrate 0.8 1.5 0.6 4.4
Ethyl lactate 0.4 0.2 0.1 2.0 Campo, E., Ferreira, V., Escudero, A., Marqués, J. C., & Cacho, J. (2006). Quantitative
Furfural 0.8 0.5 2.6 2.9 gas chromatography–olfactometry and chemical quantitative study of the
Ethyl isovalerate 0.5 0.8 1.2 1.1 aroma of four Madeira wines. Analytica Chimica Acta, 563, 180–187.
n-Hexanol 0.6 0.6 2.3 0.9 Cortes, H. J., Winniford, B., Luong, J., & Pursch, M. (2009). Comprehensive two-
Isoamyl acetate 1.7 2.2 0.4 4.2 dimensional gas chromatography review. Journal of Separation Science, 32,
Benzaldehyde 0.5 1.3 3.1 2.3 883–904.
d’Acampora Zellner, B., Dugo, P., Dugo, G., & Mondello, L. (2008). Gas
Ethyl hexanoate 12.4 12.7 13.9 9.7
chromatography-olfactometry in food flavour analysis. Journal of
2-Ethyl-1-hexanol 1.2 0.1 0.2 0.2
Chromatography A, 1186, 123–143.
Phenethyl alcohol 1.7 1.9 1.1 3.4 Di Stefano, R. (1985). Studio della composizione chimica e dei composti volatile del
Diethyl succinate 2.7 4.9 5.7 9.2 Marsala vergine. Vignevini, n, 12, 39–46.
ethyl octanoate 40.9 37.5 31.2 19.4 Feirrera, V., López, R., & Cacho, J. F. (2000). Quantitative determination of the
Ethyl decanoate 6.6 5.8 8.1 3.3 odorants of young red wines from different grape varieties. Journal of the Science
of Food and Agriculture, 80, 1659–1667.
Fisher, C., & Scott, T. R. (1997). Food flavours. Cambridge: The Royal Society of
Chemistry.
samples; it has been reported in papaya (Pino & Queris, 2012), Ma- La Torre, G. L., La Pera, L., Rando, R., Lo Turco, V., Di Bella, G., Saitta, M., et al. (2008).
Classification of Marsala wines according to their polyphenol, carbohydrate and
deira (Campo et al., 2006) and red wine (Martí et al., 2003). In the
heavy metal levels using canonical discriminant analysis. Food Chemistry, 110,
latter investigation, b-damascenone was described as being ‘‘peach 729–734.
jam, sweet’’, while in papaya a sensory threshold of 0.05 lg/L was Liu, Z., & Phillips, J. B. (1991). Comprehensive two-dimensional gas chromatography
using an on-column thermal modulator interface. Journal of Chromatographic
reported. Furthermore, b-damascenone was reported in the
Science, 29, 227–231.
low-polarity fraction, in the GC–MS ‘‘vergine’’ Marsala study (Di Martí, M. P., Mestres, M., Sala, C., Busto, O., & Guach, J. (2003). Solid-phase
Stefano, 1985). microextraction and gas chromatography olfactometry analysis of successively
Increasing values of whisky lactone (peak 103) were found from diluted samples. A new approach of the aroma extract dilution analysis applied
to the characterization of wine aroma. Journal of Agricultural and Food Chemistry,
‘‘superiore secco’’ (traces), to ‘‘superiore riserva’’ (0.03%), and ‘‘ver- 51, 7861–7865.
gine’’ (0.13%). The presence of whisky lactone, a compound charac- Pino, J. A., & Queris, O. (2012). Characterisation of odour-active compounds in
terised by a coconut odour and with a low threshold (67 lg/L), has papaya (Carica papaya L.) wine. International Journal of Food Science and
Technology, 47, 262–268.
been reported in Madeira wine (Campo et al., 2006) and in Purcaro, G., Tranchida, P. Q., Assis Jacques, R., Bastos Caramão, E., Moret, S., et al.
‘‘vergine’’ Marsala [Di Stefano, 1985 (high-polarity fraction)]. (2009). Characterization of the yerba mate (Ilex paraguariensis) volatile fraction
The discussion could continue on practically each of the using solid-phase microextraction-comprehensive 2-D GC-MS. Journal of
Separation Science, 32, 3755–3763.
compounds reported in Table 1, but would be outside the scope Robinson, A. L., Boss, P. K., Heymann, H., Solomon, P. S., & Trengove, R. D. (2011).
of the present paper, which is to unveil the highly complex nature Development of a sensitive non-targeted method for characterizing the wine
of the Marsala headspace. Considering the number of analytes de- volatile profile using headspace solid-phase microextraction comprehensive
two-dimensional gas chromatography time-of-flight mass spectrometry.
tected, such an untargeted application could probably not have
Journal of Chromatography A, 1218, 504–517.
been achieved adequately by using a one-dimensional GC process. Silva Ferreira, A. C., Barbe, J.-C., & Bertrand, A. (2003). A key odorant of the typical
Such a statement is fortified by observing the inset reported in aroma of oxidative aged port wine. Journal of Agricultural and Food Chemistry, 51,
4356–4363.
Fig. 3, which refers to the boxed area of the chromatogram, and
Sousa Câmara, J., Marques, J. C., Alves, M. A., & Silva Ferreira, A. C. (2004). 3-
has been extracted at a lower maximum signal intensity. The num- Hydroxy-4,5-dimethyl-2(5H)-furanone levels in fortified Madeira wines:
ber of compounds visible in the inset are much higher compared to relationship to sugar content. Journal of Agricultural and Food Chemistry, 52,
the boxed area, and gives a real picture of the complexity of the 6765–6769.
Weldegergis, B. T., Crouch, A. M., Górecki, T., & de Villiers, A. (2011a). Solid phase
sample’s volatile composition. extraction in combination with comprehensive two-dimensional gas
In conclusion, the HS SPME GC  GC–MS/FID method developed chromatography coupled to time-of-flight mass spectrometry for the detailed
has opened a door on Marsala wine. In general, the use of a GC–MS investigation of volatiles in South African red wines. Analytica Chimica Acta, 701,
98–111.
approach for the untargeted analysis of the volatile fraction of des- Weldegergis, B. T., de Villiers, A., McNeish, C., Seethapathy, S., Mostafa, A., Górecki,
sert wines (and wines in general) could be considered an excessive T., et al. (2011b). Characterisation of volatile components of Pinotage wines
analytical challenge. In principle, time-of-flight MS can spectrally using comprehensive two-dimensional gas chromatography coupled to time-
of-flight mass spectrometry (GC  GC–TOFMS). Food Chemistry, 129, 188–199.
‘‘deconvolute’’ overlapping analytes, at the GC outlet. However, Welke, J. E., Manfroi, V., Zanus, M., Lazarotto, M., & Alcaraz Zini, C. (2012).
the reliability of MS database identification does decrease, as the Characterization of the volatile profile of Brazilian Merlot wines through
extent of co-elution increases. Consequently, GC  GC appears as comprehensive two dimensional gas chromatography time-of-flight mass
spectrometric detection. Journal of Chromatography A, 1226, 124–139.
the most suitable pre-separation method in such applications. It
must be added, however, that GC  GC experiments provide a great