analyses were not reported in MDGC

analyses were not reported in MDGC analyses, isomaltol was
described with a typical caramel odour. Once again, the divergence
between the odorant descriptions might have been linked to the
existence of a co-elution on the first dimension. Although isomaltol
is well-known for its sweet caramel odour (Pittet, Rittersbacher, &
Muralidhara, 1970), its odorant properties in a mixture with five
compounds cannot yet be predicted.
Fraction no 7 was characterised by roasted, caramel/lactone and
pharmaceutical odour notes by GC–O analyses (Table 1) but no
compound could be identified after GC–MS analysis due to several
co-elutions which rendered the mass spectra uninterpretable. The
2D MS chromatogram of this heart-cut fraction from caramel B
extract is given in Fig. 3. Thirteen compounds were detected, of
which seven were identified by MS and three showed odour activity.
Hexanoic acid was described as having a sharp and spicy odour.
4-Methyl-(5H)-furan-2-one (RI = 1048) and guaiacol (RI = 1093)
were respectively characterised by roasted and pharmaceutical
odour notes.
The 2D analysis of fraction no 8 led to the detection of six compounds,
two of which were identified: 2,4-dimethylcyclopentan-
1,3-dione and c-octalactone. In fraction no 9, previously associated
with a lactone and floral odour, only 5-acetyldihydro-(3H)-furan-
2-one (or solerone) was identified. In fraction no 10, described as
being floral, lactone, spicy and roasted after 1D GC–O analyses, only
nonanoic acid was described with a spicy and smoked odour by our
panellists following 2D GC–O analyses. c-Decalactone was identified
at RI = 1478 in the same fraction but with no detectable odour.
In these fractions no 8, 9 and 10, olfactometric detection did not
provide any pertinent information regarding the qualitative properties
of the compounds and particularly concerning lactone
odours. In this case, the small number of panellists (only two)
was probably the cause.
In fraction no 12, described as having chocolate vanilla/caramel/
tobacco odours, two new compounds were identified: dihydro-4-
hydroxy-(3H)-furan-2-one and vanillin. Neither was detected by
our panellists. We can therefore assume that vanillin was the main
contributor to the vanilla note, but further analyses are necessary
with a larger number of panellists to confirm this finding.
4. Discussion
Multidimensional gas chromatography appears to be an efficient
way to overcome identification issues in complex food samples
such as caramel. In particular, heart-cutting is a suitable
approach to investigate the composition of targeted fractions. This
study focused on 12 odorant fractions in which compounds potentially
making an important contribution to aroma could not be
identified by 1D GC–MS analyses. Olfactometric detection at the
end of 2D GC was performed by two experienced panellists in order
to obtain a qualitative description of the odorants. The results
revealed some divergences from the 1D GC–O results, probably
due to the smaller number of panellists in the MDGC study. Indeed,
one of the main disadvantages of characterisation technique such
as heart-cutting, is the large number of analyses required for just
one sample. Then, for practical reasons, it is practically impossible
to perform a quantitative GC–O study with a dozen panellists.
Comprehensive two-dimensional gas chromatography (GCGC)
has recently been the subject of intense development and has
now become one of the most powerful techniques in terms of resolution
(Seeley & Seeley, 2013). This technique enables the full
characterisation of a sample in two orthogonally stationary phases
with only one run. However, it also necessitates the use of ‘‘fast’’
detectors such as Time of Flight-MS (TOF-MS) or Flame Ionisation
Detector (FID) because of the very high separation speed on the second column. Thus combining GCGC with olfactometric detection
is not possible because of the slow breathing rate of humans