698
9+10, 2002
The importance of GC and GC-MS in perfume
analysis
Arian van Asten*
Perfume Competence Centre, Laundry Global Technology Centre, Unilever Research, P.O. Box 114,
NL-3130 AC Vlaardingen, The Netherlands
Gas chromatography (GC) and mass spectrometry
(MS) are by far the most important analytical tech-
niques in the perfume industry. Both perfume
houses (the suppliers) and the home and personal-
care companies (the clients) rely primarily on GC
and GC-MS for unraveling the composition of per-
fumes (i.e. perfume formulation), quality control,
competitor analysis and trace analysis on substrates
and in the headspace. State-of-the-art perfume for-
mulation is based on perfume-specific Kovats Index
(KI) and MS databases. By applying FID (flame-
ionization detector) response-correction factors, the
accuracy of the perfume-formulation process can be
further improved. Because of the complexity of per-
fumes, use is made of GC columns and conditions
that offer maximum resolution rather than minimum
analysis time. Mass-spectral deconvolution tools can
be very useful in identifying perfume ingredients
from GC-MS data in cases of co-elution or strong
matrix interference. By applying the MS detector in
the selected ion monitoring (SIM) mode, GC-MS is
very suitable for trace analysis of perfume ingre-
dients, thus enabling the study of perfume efficacy
during use of home and personal-care products.
Recent developments in the field of comprehensive
GC (i.e. GCGC) also seem very promising for per-
fume analysis. # 2002 Published by Elsevier Science
B.V. All rights reserved.
1. Introduction
The first evidence of the use of fragrances by
mankind dates back to the ancient Egyptian
civilizations around 2400 BC [1]. Throughout
*Tel.: +31-10-4606490; Fax: +31-10-4605166.
E-mail: arian-van.asten@unilever.com
0165-9936/02/$ - see front matter
PII: S0165-9936(02)00807-5
trends in analytical chemistry, vol. 21, nos.
the ages, perfumes have become increasingly
more important and accessible. Nowadays, the
perfume business is a billion-dollar industry [2].
The key stakeholders in this global industry are
the perfume houses and the fine-fragrance and
the home and personal-care companies. The
perfume houses are relatively unknown compa-
nies that create and supply perfumes according
to the wishes of their customers. The houses
also synthesize and trade the individual raw
materials used in perfumes.
A contemporary perfume for a home-care
product (e.g. as applied in household cleaners,
detergents and fabric conditioners) generally
costs around E15–20/kg and comprises 10–50
individual perfumery raw materials. The per-
fume level in these products is generally in the
range of 0.05–0.5 wt%. A typical personal-care
perfume (e.g. as applied in shampoos, deo-
dorants, shower gels, facial cremes and soap
bars) is more expensive (E20–50/kg) and can
comprise up to 80 individual raw materials. Per-
fume levels are usually in the range 0.5–2 wt%.
The fine fragrances are the most costly and
complex and can contain over 100 ingredients.
Considering the fact that perfume raw mate-
rials are themselves quite often complex mix-
tures of synthetic or natural (e.g. essential oils)
organic compounds, the determination of the
composition of an unknown perfume, the so-
called perfume-formulation process, is not an
easy task. Apart from the olfactive expertise of
the perfumers, it also requires advanced analysis
tools. Because most perfume ingredients are
apolar and (semi)-volatile in nature, for dec-
ades GC and GC-MS have been, and still
are, the most important analytical techniques
in the perfume industry. Perfume-formulation
# 2002 Published by Elsevier Science B.V. All rights reserved.
trends in analytical chemistry, vol. 21, nos. 9+10, 2002
work, perfume-stability studies, perfume-pro-
duct interaction studies, perfume-packaging
interaction studies, perfume-substrate interac-
tion studies (e.g. perfume deposition on fabric
or perfume longevity on skin) and perfume-
headspace analysis are generally all based on the
use of GC and GC-MS in combination
with suitable extraction or thermal desorption
methodologies.
Of the efforts of the many experienced ana-
lytical chemists that work in the perfume
industry, very little appears in the open literature
because that is usually not in the best interest of
the companies operating in this highly competi-
tive market. This contribution will hopefully
provide insight in the way GC and GC-MS
are used in the perfume industry. Additionally,
it will address recent developments in this
field that are of particular interest for
perfume analysis.
2. State-of-the-art perfume formulation
with GC and GC-MS
To determine the composition of a pure
perfume sample, e.g. for cost analysis or quality
control, the perfume houses and home and
personal-care companies use a very similar
approach based on capillary GC with flame
ionization and mass spectrometric detection.
The perfume sample is, after appropriate dilu-
tion, analyzed with capillary GC on at least two
different columns. The PDMS (i.e. poly-
dimethyl siloxane; apolar phase) and PEG (i.e.
polyethylene glycol; polar phase) columns are
the most frequently used for this purpose.
Temperature-programmed Kovats Indices
(KIs) of the unknown peaks are determined
with the use of a series of n-alkanes or linear
esters. Additionally, GC-MS is used to obtain
electron-impact ionization (EI) mass spectra.
With the use of specific perfume KI and EI-
MS databases, the unknowns are identified.
The complexity of some of the perfumes
requires this ‘‘three-point’’ identification pro-
cess because peak overlap and mixed mass
spectra are quite common. Fig. 1 illustrates the
699
identification of a perfume ingredient following
this process. The ‘‘crude’’ perfume composition
is determined on the basis of a 100% measure-
ment of the FID peak areas in one of the
chromatograms (usually the chromatogram
obtained on the apolar column). Finally, the
composition undergoes further fine tuning by
adding together the ingredients that make up a
single perfume raw material and applying some
correction factors based on knowledge of the
raw materials used in the perfume industry.
The accuracy of the perfume-formulation
process is mainly determined by the quality of
the databases and the expertise of the for-
mulators. Therefore, perfume-analysis labora-
tories spend a lot of time ensuring that their
perfume KI and MS libraries are up to date (by
analyzing new perfume ingredients) and ‘‘clean’’
(i.e. not containing irrelevant or erroneous data).
Such perfume libraries contain on average
between 1000 and 3000 records. (Nowadays,
perfumers can use up to 5000 different raw
materials but only about 1500 ingredients are
used frequently).
Especially challenging is the identification of
essential oils in full formulations. These steam
distillates or cold pressings of natural products,
such as flowers and fruits, can comprise up to
50 individual ingredients. Additionally, the main
compounds are usually quite common, and
(cheaper) synthetic analogues can also be used
in the composition. However, for some oils the
presence of specific trace markers can indicate
their presence in the perfume.
Although generally not taken into account in
perfume analysis, the variation of the FID
response as function of molecular structure can
be a source of substantial systematic error.
However, the simple procedure, as described by
Scanlon and Willis et al. [3], can easily be
implemented in the perfume-formulation pro-
cess and it provides an accurate estimation of
the relative FID-response factor, as illustrated in
Fig. 2. The approach is based on the additional
effects of functional groups in a molecule on the
FID response, as established experimentally.
These corrections lead to a so-called Effective
Carbon Number (ECN) for each molecule [3].
700 trends in analytical chemistry, vol. 21, nos. 9+10, 2002

Although, on average, systematic errors by
neglecting these effects will not be huge, some
common perfume ingredients, such as DPG
(dipropylene glycol, an odorless solvent), helio-
tropine and helional, possess FID sensitivities
that are less than 50% of those for terpenes,
such as limonene, pinene and myrcene.
For competitor-perfume analysis or perfume-
stability studies, the perfume has to be extracted
from the home or personal-care product prior
to analysis. However, formulation of the per-
fume in the extract is still conducted according
to the process outlined above. Determination of
the perfume level in the product is usually based
on an internal standard method. However, it
should be noted that the perfume formulation is
less accurate from extracts because the co-
extraction of part of the product matrix (e.g.

Fig. 1. Illustration of peak identification in the perfume-formulation process. A: GC-FID chromatogram of a detergent perfume
on a 50 m0.2 mm0.33 mm HP1 capillary-GC column; B: the same GC-FID chromatogram but on a 60 m0.25 mm0.25
mm Supelcowax capillary-GC column. Identified compound (indicated by arrows): hexyl cinnamic aldehyde; EI-MS library fit of
99/100; KI-HP1=1721 (1721 in library); KI-Supelcowax=2373 (2377 in library).
trends in analytical chemistry, vol. 21, nos. 9+10, 2002
Fig. 2. Formulation of a test perfume with and without correction for the variation in FID-response factor.
surfactants) results in contaminated chromato-
grams and mass spectra.
3. GC-column requirements for
perfume analysis: speed, plates or
selectivity?
The formulation of a complex perfume on the
basis of GC and GC-MS data can be performed
by an expert in one to two days. This indicates
that, for such a task, a perfume laboratory does
not require high-speed GC analysis. It is data
interpretation that is the limiting step in the
process. Therefore, the interest in short, nar-
row-bore GC columns (e.g. 10 m100 mm
columns) has been limited in the perfume
industry. Of course, not all perfume-analysis
tasks are related to perfume formulation. The
perfume houses have, for instance, comprehen-
sive programs in place to search for new odor
701
molecules with novel characteristics, olfactively,
physically or economically. As these programs
are more and more based on high-throughput-
synthesis and screening techniques, fast analysis
tools are also required. Consequently, there is an
interest in high-speed GC and GC-MS analysis
within the perfume houses, but not to speed up
the perfume-formulation process. To speed up
and improve the quality of perfume formula-
tion, one needs to simplify data interpretation.
The choice of the capillary-GC columns is very
important in that respect, not in terms of speed,
but rather in terms of resolution, sensitivity,
mass loadability, reproducibility, robustness and
selectivity.
The formulation process starts with the iden-
tification of the individual perfume ingredients
on the basis of KI values and mass spectra. The
more the ingredients (partially) overlap in the
chromatograms, the more difficult this identifi-
cation becomes. Automation of the formulation
702
process (e.g. automated MS library searches,
automated peak integration) is better achievable
when the chromatographic conditions are reso-
lution-optimized instead of speed-optimized.
Therefore, the perfume industry tends to prefer
long, medium-to-narrow-bore GC columns that
generate many plates for optimal resolution. If
the inner diameter is too small, pressure drops
become excessive, limiting the column length.
For that reason, typical columns used for per-
fume formulation are 30–60 m in length and
have an internal diameter of 0.25–0.32 mm.
In a perfume composition, numerous ingre-
dients can be present at a low level (i.e. < 1%)
and still have an important impact either olfac-
tively or economically. Therefore, sensitivity is
also an important aspect in GC and GC-MS per-
fume analysis. In this case, sensitivity is directly
linked to mass loadability, because perfumes can
also contain ingredients that make up to as much
as 50% of the total composition. Therefore, it is
important that capillary-GC columns provide
maximum sensitivity in combination with max-
imum dynamic range. Note that, by using high-
speed GC columns, both sensitivity and dynamic
range are affected negatively. Mass loadability
improves with the film thickness of the stationary
phase in the column. However, some perfume
ingredients have very high boiling points and
conditions have to be chosen such that the elu-
tion of all perfume ingredients, including the less
volatile, is ensured. For that reason, the film
thickness is usually in the range of 0.25–0.5 mm
leading to phase ratios of roughly 100–200.
An important part of identification is based
on the KI values usually measured on two
capillary-GC columns with different stationary
phases. The usefulness of the KI approach is
completely determined by the reproducibility
and robustness of the capillary-GC columns.
GC columns with fragile stationary phases that
are easily contaminated and/or degraded are
not popular in the perfume industry because
this results in KI ‘‘drifting’’ over the lifespan of
a capillary column. Therefore, KI libraries based
on the popular PDMS capillary columns (e.g.
HP1, DB1, OV1) are preferred in the perfume
industry. GC columns with this type of
trends in analytical chemistry, vol. 21, nos. 9+10, 2002
stationary phase are very robust (columns that
handle only pure perfume samples can be used
for years without significant changes in column
performance and selectivity) and can be pro-
duced with minimal column-to-column varia-
bility. Even KI values for perfume ingredients
on PDMS columns from different suppliers are
usually very similar. Unfortunately, this cannot
be said for the PEG columns. Although the
column-to-column variation for PEG columns
is usually acceptable for most manufacturers, the
corresponding KI libraries are supplier specific.
In general (and for the PEG columns in par-
ticular), a regular KI ‘health check’ of the
operational columns is highly recommended.
This is especially true for set-ups that are used
to analyze perfume extracts. These extracts of
home and personal-care products also contain
significant amounts of surfactants. Part of this
surfactant matrix can irreversibly precipitate on
the stationary phase, resulting in adsorption
phenomena of polar perfume ingredients and
deviant KI values.
Even if the GC and GC-MS conditions are
carefully optimized in line with the above
guidelines, co-elution of some perfume ingre-
dients is observed for almost every perfume and
perfume extract. The number and the variety of
ingredients in perfumes are simply too large to
ensure complete separation on the basis of
volatility (PDMS column) and polarity (PEG
column) alone. Therefore, there is ongoing
interest from the perfume-analysis community
in the development of new stationary phases for
capillary GC. If a robust capillary-GC column
could be produced that would, for example,
offer ‘‘double-bond selectivity’’, it would defi-
nitely be applicable in perfume analysis. The
enantioselectivity offered by the commercially
available cyclodextrin capillary columns has
been exploited for perfume analysis for a long
time. These so-called chiral columns are parti-
cularly useful for identification and authenticity
control of essential oils used in perfumes [e.g.
4,5]. Synthetic analogues are usually racemic
and, in some cases, ratios of enantiomers can be
indicative of not only the natural source but also
the area of cultivation. The chiral composition
trends in analytical chemistry, vol. 21, nos. 9+10, 2002
of a mixture can, in some cases, also affect the
odor quality as the olfactive receptors in the
human nose are protein based and thus enantio-
specific [6]. Chiral columns are especially
employed for the formulation of oral-care
flavors used, for example, in toothpastes and
chewing gums. Oral-care flavors generally con-
tain a large amount of natural mint oils. The
quality and the price of these oils vary strongly
and depend on the type of crop (e.g. spearmint,
peppermint and cornmint), the origin (e.g.
Canada, USA and China) and the year of har-
vest. Identification is difficult and, in this case,
the use of chiral columns can be very informa-
Fig. 3. Chiral GC chromatograms of toothpaste oral-care flavors based on synthetic (A) and natural (B) ingredients
(Column: Alpha Dex 120 ex Supelco, 30 m0.25 mm).
703
tive. Fig. 3 is an example of an ‘‘enantioselective
chromatogram’’ of an oral-care flavor.
4. MS requirements for perfume analysis
As described above, perfume-analysis labora-
tories rely on in-house-recorded EI-MS libraries
to identify fragrance raw materials in perfume
submissions and product extracts. These librar-
ies are constantly updated by analyzing any new
raw material that appears in the market. The
relatively cheap bench-top quadrupole and ion-
trap MS systems are most commonly used in
704
the perfume industry. No extensive MS exper-
tise is required to maintain and use these spec-
trometers effectively, provided powerful
perfume MS libraries are employed.
The bench-top quadrupole MS has been
around longer than the ion-trap MS, which was
introduced in the late 1980s. As the quadrupole
is also a bit cheaper than the ion trap, it is the
most popular in the perfume laboratory. How-
ever, ion-trap spectrometers offer a number of
very positive features, when compared to the
quadrupole MS [7]:
. higher sensitivity in the full-scan mode;
. switching between EI and Chemical Ionization (CI)
without conversion; and,
. ability to perform MS/MS and even MSn experiments.
Occasionally, extracts or submissions contain
perfume ingredients that are so novel that they
are not yet commercially available and used by
perfume houses only to provide a competitive
edge to their submissions. No samples of these
so-called captives can, therefore, be obtained.
Consequently, structure elucidation is required
to obtain more insight into the new compound.
In this task, MS also plays a crucial role, as often
the quantities are too low to enable straightfor-
ward identification via NMR. The EI-MS spec-
trum can be very informative but it is often not
sufficient by itself to provide a positive identifi-
cation. Chemical ionization can be used to
determine the molecular weight of the
unknown, especially for molecules that exhibit
extensive fragmentation during EI. In addi-
tional, accurate mass measurements with sector
or TOF-MS instruments can be employed to
obtain the molecular formula [7].
Two commercially available libraries are pop-
ular in the perfume industry. The NIST 98
library serves as a spectrum-search ‘‘back-up’’.
The collection is of high quality and is very
useful in the identification of non-perfume
materials in product extracts and in revealing
structural features of captives. The second
library worth mentioning was set up by Adams
[8] and contains a comprehensive set of spectra
of compounds found in essential oils. This
library can be applied to the identification of
trends in analytical chemistry, vol. 21, nos. 9+10, 2002
naturals in perfumes. Very recently, an updated
version has become available.
5. Mass spectral deconvolution
As stated above, the great complexity of per-
fume samples implies that MS spectra are rarely
pure-compound spectra. In the case of mixed
spectra, mass spectral deconvolution methods
can be of significant assistance. Various pack-
ages for deconvoluting mixed spectra are avail-
able. The AMDIS (Automated Mass spectral
Deconvolution and Identification System) soft-
ware, was initially developed by NIST for the
automated identification of chemical-warfare
agents from high-speed GC-MS data [9]. This
freeware package is able to identify ingredients
from raw GC-MS chromatograms, even in
cases of co-elution or severe matrix contamina-
tion. The algorithm employs four basic steps:
noise analysis; compound perception; spectral
‘‘deconvolution’’; and, finally, compound iden-
tification. The spectral ‘‘deconvolution’’ is based
on the model peak method, i.e. the construction
of purified mass spectra by adding extract ion
profiles of similar ‘‘chromatographic’’ peak
shape. In the final step of the process, these
purified mass spectra are matched against mass
spectra in home-made or standard mass spectral
libraries, such as the NIST98 library, for com-
pound identification.
The transfer of specific MS libraries of almost
every format to the AMDIS format is straight-
forward [9], making this tool a very useful
instrument for every laboratory involved in
compound identification using GC-MS. A per-
fume-specific AMDIS library is easily created
from dedicated MS libraries, and, in this
manner, the AMDIS package can be employed
in perfume-formulation work. With the pro-
gram, co-elution of perfume ingredients in
chromatograms of pure perfume samples is
easily spotted. Alternatively, when dealing with
perfume extracts of home and personal-care
products, AMDIS assists in tracking down traces
of perfume ingredients in matrix-interfered GC-
MS signals.
trends in analytical chemistry, vol. 21, nos. 9+10, 2002
As the processing of GC-MS data can be fully
automated by AMDIS, and the package calcu-
lates relative amounts and also offers the possi-
bility of implementing and using KI data,
AMDIS could, in principle, perform a fully
automated perfume formulation on the basis of
GC-MS data only. However, in our experience,
the algorithm is far from flawless and the effi-
ciency of the spectral purification and com-
pound identification seems to be highly mass-
spectrum dependent. Although it is worthwhile
pursuing the goal of automated perfume for-
mulation, the AMDIS package is currently used
by perfume formulators ‘‘only’’ as an additional
tool to check whether any ingredient or co-elu-
tion has been overlooked. The program is
especially useful in finding traces of perfume
ingredients that have co-eluted with the main
perfume ingredients or matrix compounds.
Quite often, conventional mass spectral
Fig. 4. Details of the AMDIS analysis of a GC-MS HP1 chromatogram of a detergent perfume. A: AMDIS detecting a trace of
Boisambrene Forte co-eluting in a large excess of hexyl cinnamic aldehyde; B: AMDIS not detecting the clear presence of nerol in
a citronellol peak.
705
searches and the human eye do not pick up the
presence of these traces. The strengths and
the weaknesses of AMDIS in the perfume-
formulation process are illustrated in Fig. 4.
6. Quantitative trace analysis of perfume
ingredients with GC-MS and SIM
The human nose is an extremely sensitive and
selective ‘‘detector’’. For many compounds that
invoke a response in our organoleptic receptors,
when it comes to detection limits, the nose is no
match for any conventional GC detector. How-
ever, the study and the understanding of human
perception in relation to the physical character-
istics of perfumes is necessary to optimize their
use in home and personal-care products. To
that end, there needs to be analysis of trace
amounts of perfume on substrates (e.g. tiles,
706
plates, fabric, skin, hair and teeth) and in the
headspace of these substrates. By combining the
proper sampling techniques (e.g. dynamic head-
space, SPME (solid-phase microextraction),
SPE (solid-phase extraction), solvent extrac-
tion), GC methodologies (e.g. large volume
injection, thermal desorption) and detection
strategies, overall detection limits can be
reduced to an extent that allows accurate analysis
of perfume ingredients at the ppb level.
The MS detector is frequently employed in
perfume-analysis laboratories when trace-level
analysis is required. If the perfume compounds
that have to be analyzed and their GC-elution
characteristics are known, the MS detector can
be operated in the SIM mode to enhance selec-
tivity and sensitivity. Normally, three ions are
taken for a single ingredient to allow for a purity
check by comparing the measured abundance
ratios of the ions to the ratios obtained from the
reference mass spectrum.
In general, a 10 to 100-fold sensitivity gain
can be obtained with SIM compared to the full-
scan mode of the mass spectrometer, as long as
the target ions possess a high abundance in the
mass spectrum. However, for complex samples,
care should be taken not to select any ions that
are also appearing in the matrix background.
Otherwise, the gain in sensitivity is often
accompanied with an increase in chemical noise
resulting in relatively high detection limits and
cumbersome identification.
A recent external publication by Procter and
Gamble illustrates the use of the SIM mode for
the trace analysis of perfume ingredients with
GC-MS [10]. This work describes the study of
the biodegradability of several classes of per-
fume ingredients in municipal wastewater-treat-
ment facilities. The trace concentrations of the
ingredients in the influent and the effluent
streams could be monitored with satisfactory
accuracy. The use of SIM trace analysis for the
accurate determination of perfume deposition
on fabric during washing and perfume retention
during the subsequent drying process is shown
in Fig. 5. A concentrated extract from a fabric
monitor is analyzed by GC-MS with dedicated
SIM windows.
trends in analytical chemistry, vol. 21, nos. 9+10, 2002
For most common perfume ingredients, sui-
table target ions can be identified enabling the
use of the methodology for almost every deter-
gent perfume. Because these concentrated
extracts still contain significant amounts of sur-
factants originating from the detergent, care has
to be taken to select ions for which the matrix
interference is minimal. Calibration is best per-
formed via standard addition with the perfume
under investigation. Standard addition is pre-
ferred over external calibration because it
accounts for matrix effects on the detector sen-
sitivity. Additionally, it allows for a regular
determination of the response factor during a
sample sequence. Although the MS detector can
be extremely sensitive, it does not have the
robustness of the FID, so significant sensitivity
drift is often observed. With this methodology,
the recovery of individual perfume ingredients
from washed and dried monitors can be accu-
rately determined down to the 0.5% level (cor-
responding to a concentration of approximately
50 ppb in the extract).
7. Comprehensive GCGC and
GCGC-MS for perfume analysis
The most promising development in GC
research in the last decade has without a doubt
been the introduction of two-dimensional com-
prehensive GC, often indicated as GCGC.
The technique offers unsurpassed resolution
and detection sensitivity by coupling two GC
columns and allowing continuous heart-cuts
from the first column to be analyzed on the
second column. To achieve this, either a cryo-
genic or a thermal modulator is used at the
point were both columns are coupled [11]. The
initial column is usually a regular apolar capil-
lary-GC column, whereas the second column
has a polar phase and ‘‘high-speed’’ dimensions
(e.g. short length, and small internal diameter
and film thickness). This allows all the fractions
eluting from the first column to be separated in
real-time by the second column. The resulting
datasets are huge, but, when displayed in con-
tour plots, they reveal the enormous separation
trends in analytical chemistry, vol. 21, nos. 9+10, 2002 707

capability of a well-designed GCGC method. regular GC analysis. The GCGC-contour plot

GCGC is especially suited when dealing with
very complex mixtures, such as petrochemical
products [13] or essential oils [11,12]. However,
further advancements in instrument robustness,
ease of use, data handling and detection are
necessary to make GCGC into a routine-ana-
lysis tool. So far, detection is mainly performed
with the FID, but coupling to TOF-MS detec-
tors has also been demonstrated recently [11].
The GCGC analyses of vetiver [12] and
lavender [11] essential oils have revealed that
these natural perfume ingredients are even more
complex than had been assumed on the basis of
of vetiver oil showed over 150 individual ingre-
dients. Because of the increase in sensitivity by
zone compression in the modulator, numerous
trace ingredients could be detected. Even with-
out identification of these ingredients, GCGC
could be very useful in the perfume industry to
determine the origin of an essential oil (place of
cultivation, year and even season of harvest)
and to detect adulterations.
In addition, bearing in mind the difficult
challenge for contemporary GC techniques to
detect essential oils in perfume formulations,
GCGC analyses of perfumes could possibly

Fig. 5. GC-MS analysis of a perfume extract from a freshly-washed terry-towel-cotton monitor. A: with the MS operated in full-
scan mode, showing the excess of surfactants in the extract; B: with a selective SIM method to monitor the perfume ingredients
deposited on the cotton surface.
708
generate very specific patterns that can be linked
to the presence of these natural ingredients.
Taking this a step further, GCGC could even
offer sufficient peak capacity to enable fully
automated perfume formulation without com-
pound identification (e.g. without MS detec-
tion). In the two-dimensional separation space,
peak overlap would be rare and each ingredient
could be identified on the basis of a KI
‘‘couple’’. This KI ‘‘couple’’ would represent the
X,Y coordinates of the ingredient in the contour
plot. As the separation is temperature pro-
grammed in the first dimension and almost iso-
thermal in the second dimension, one
coordinate should be a true KI and the other
coordinate a linear programmed temperature
retention index [13]. Quantification could be
achieved by integrating the contour peaks and
determining the contribution of the individual
peak areas to the total area (i.e. the total peak
area of all second dimension chromatograms).
To realize this vision of automated perfume
formulation with GCGC, the following pro-
gress would have to be made :
1. Design of a suitable procedure and test mixture
with known KI ‘‘couples’’ to calibrate the system.
(The alkane or methylester homologues currently
used would not sample the entire separation
space).
2. Construction of a comprehensive perfume KI
‘‘couples’’ database by analyzing the individual
fragrance raw materials with the GCGC system.
3. Development of a data-treatment procedure to
identify peak contours, determine their KI ‘‘couple’’,
identify the perfume ingredients with the use of the
database, integrate entire contours and establish the
relative area versus the total peak area.
trends in analytical chemistry, vol. 21, nos. 9+10, 2002
Although this may seem a difficult task, work
by Beens [13] has already indicated that KI data
can be used to predict GCGC contour plots
of test mixtures. Therefore, the opposite
approach should also be viable. However, the
biggest challenge will be the development of the
data-treatment software.
Acknowledgements
The skills of all the people that work in Uni-
lever’s Perfume Competence Centre made this
contribution possible.
References
[1] D.H. Pybus, C.S. Sell, The Chemistry of Fragrances, Royal
Soc. Chem, Cambridge, United Kingdom, 1999.
[2] IFF’s (the largest perfume house) net fragrance sales were
$865m in 2000 (www.iff.com).
[3] J.T. Scanlon, D.E. Willis, J. Chromatogr. Sci. 23 (1985) 333.
[4] F. Tateo, M. Bononi, E. Dominicus, V. Fumagalli, Anal.
Commun. 36 (1999) 149.
[5] W.A. Konig, A. Kruger, D. Icheln, T. Runge, J. High Res.
Chromatogr 15 (1992) 184.
[6] B. Koppenhoefer, R. Behnisch, U. Epperlein,
H. Holzschuh, A. Bernreuther, P. Piras, C. Roussel, Per-
fum. Flavor 19 (1994) 1.
[7] H.-J. Hubschmann, Handbook of GC/MS, Wiley-VCH,
Weinheim, Germany, 2001.
[8] R.P. Adams, Identification of Essential Oil components by
Gas Chromatography/Mass Spectroscopy, Allured Pub-
lishing Corp., Carol Stream, Illinois, USA, 1995.
[9] http://chemdata.nist.gov/mass-spc/amdis/explain.html.
[10] S.L. Simonich, W.M. Begley, G. Debaere, W.S. Eckhoff,
Environ. Sci. Technol 34 (2000) 959.
[11] R. Shellie, P. Marriott, P. Morrison, Anal. Chem. 73 (2001)
1336.
[12] P. Marriott, R. Shellie, J. Fergeus, P. Morrison, Flavour
Fragr. J. 15 (2000) 225.
[13] J. Beens, Chromatographic Couplings for Unraveling Oil
Fractions, PhD thesis, Amsterdam, The Netherlands,
ISBN 90–9012133–1, 1998.