The autooxidation process in linoleic acid screened by Raman spectroscopy

Article abstract of DOI:10.1002/jrs.4121

The chemical changes associated to the autooxidation process of linoleic acid (LA) were detected by Raman spectroscopy and interpreted in the light of density functional theory (DFT) calculations performed for both the fatty acid and its main oxidation products. The present methodology, applied for a six-day period upon induction of oxidation (through heating), allowed to understand the chemical modifications occurring during the oxidation process. Raman spectroscopy was shown to be a suitable and reliable technique for assessing the oxidation degree of fatty acid samples, particularly pure fatty acids, mainly when computational methods are used alongside to predict the spectral features of the distinct chemical entities involved. Screening of the oxidation process was mostly based on the loss of intensity of the bands assigned to LA cis-double bonds. Copyright

Full text of DOI:10.1002/jrs.4121

Research article
Received: 9 March 2012
Revised: 17 April 2012
Accepted: 29 April 2012
Published online in Wiley Online Library: 8 November 2012
(wileyonlinelibrary.com) DOI 10.1002/jrs.4121
The autooxidation process in linoleic acid
screened by Raman spectroscopy
N. F. L. Machado,a L. A. E. Batista de Carvalho,a J. C. Oterob
and M. P. M. Marquesa,c*
The chemical changes associated to the autooxidation process of linoleic acid (LA) were detected by Raman spectroscopy and
interpreted in the light of density functional theory (DFT) calculations performed for both the fatty acid and its main oxidation
products. The present methodology, applied for a six-day period upon induction of oxidation (through heating), allowed to
understand the chemical modifications occurring during the oxidation process. Raman spectroscopy was shown to be a
suitable and reliable technique for assessing the oxidation degree of fatty acid samples, particularly pure fatty acids, mainly
when computational methods are used alongside to predict the spectral features of the distinct chemical entities involved.
Screening of the oxidation process was mostly based on the loss of intensity of the bands assigned to LA cis-double bonds.
Copyright © 2012 John Wiley & Sons, Ltd.
Supporting information may be found in the online version of this article.
Keywords: linoleic acid; oxidation; assay for antioxidant activity; Raman spectroscopy; DFT calculations
sensitivity of the technique as well as its readiness, making it suit-
able for the detection of low concentrations of oxidation products
Introduction
The oxidative deterioration of lipids has always been an issue
regarding food quality, causing a nutritional value loss[1] and
leading to the appearance of undesirable odours and flavours in
lipid containing foods – rancidity. Thus, several analytical methods
have been developed aiming at an accurate assessment of the
state of preservation of edible oils and other lipid-containing food
products.[2–4] The majority of these methods are based on
chromatographic or spectrophotometric techniques, which require
a pre-treatment of the samples – these may introduce errors due to
mishandling, or to the presence of contaminants and additives
that may interfere with the results.[2,5] Furthermore, the sample
preparation is sometimes complex and often involves handling of
hazardous chemicals. In fact, to this date, there is no simple and
reliable method able to perform a direct measurement of the
chemical content of edible oils, determining the presence of
oxidation products.[5]
In order to fill this gap, in particular regarding the screening of
fatty acids, attention has been paid to vibrational spectroscopy in
the few last years.[5–9] The assessment of either a fatty acid or a
mixture (e.g. edible oils) through vibrational spectroscopy methods
allows to gather information on the sample composition, as well as
on the unsaturation pattern of the compounds present[8] and on
the corresponding molecular conformation (e.g. cis or trans).[10,11]
Furthermore, the presence of oxidation products is easily detected
through analysis of the spectral pattern, which may be used to
measure the degree of oxidation in either a fatty acid or a blended
oil,[6,8,12] by interpretation of the change in relative intensity of
specific fatty acid bands.
in oils. A great deal of work has been developed in the last 15 years
regarding the evaluation of the chemical composition of edible
oils,[7,9] aiming at the determination of the oil’s geographic origin
(e.g. for virgin oils) and at an accurate food quality control.[9] In fact,
the amount of work available within this field with recourse to IR
has already lead IUPAC to establish standard procedures, based
on thoroughly characterised bands (such as the one at 967 cmꢀ1
,
assigned to non-conjugated trans-double bonds).[3] Furthermore,
FTIR has been applied to the determination of standard indices
used in the classification of edible oils, such as the free fatty acids’
content[10] or the iodine value and the saponification number.[13]
Regarding the oxidation status assessment, IR has not received
special attention until recently,[4,14] although encouraging results
have been reported within this area in the last few years.[6,9,15]
Raman spectroscopy, in turn, has been increasingly recognised
as a promising technique for screening the oxidation process in
lipids. Although already applied for the characterisation,[16]
authentication,[17] unsaturation profile assessment[18] and free
fatty acid content evaluation,[19] it has not yet been fully explored
as an accurate analysis method of oils and fats. Some work has
recently been published dealing with fatty acid oxidation, clearly
evidencing the suitability of the technique for this specific
*
Correspondence to: M. P. M. Marques, Department of Life Sciences, Faculty of
Science and Technology, University of Coimbra, Ap. 3046, 3001-401 Coimbra,
Portugal. E-mail: pmc@ci.uc.pt
a Research Unit ‘Molecular Physical Chemistry’, University of Coimbra, Portugal
Vibrational spectroscopy studies on fatty acids and their
oxidation products have been carried out mainly by infrared (IR).
Actually, the technical advances in this methodology, such as the
introduction of Fourier transform IR (FTIR) and the development
of improved detectors and sources, have greatly increased the
b Department of Physical-Chemistry, Faculty of Sciences, University of Málaga,
Spain
c
Department of Life Sciences, Faculty of Science and Technology, University of
Coimbra, Portugal
J. Raman Spectrosc. 2012, 43, 1991–2000
Copyright © 2012 John Wiley & Sons, Ltd.

N. F. L. Machado et al.
task.[5,7,8,11,12] Actually, Raman spectroscopy presents several
advantages as compared to FTIR, namely its non-invasiveness,
high sensitivity and good reproducibility, apart from the fact that
it needs virtually no sample preparation, enabling rapid and non-
destructive measurements and yielding unique fingerprint
spectra for each compound (in a mixture). In fact, the preparation
procedure required for IR spectra collection (using solvents such
as solid KBr) may interfere with the concentration of the
compound(s) to analyse and lead to reproducibility problems.
This difficulty may be overcome by attenuated total reflectance-
FTIR, which (as Raman) does not need any sample preparation,
but requires an elaborate cleaning procedure of the crystal in
order to avoid the so-called ‘memory effect’.[5] Furthermore, the
presently available portable Raman devices, enable, in a near
future, in situ analysis of a sample, that will be of undisputable
relevance for application in food industry.[7]
Raman spectroscopy
The Raman spectra were obtained at room temperature, in a triple
monochromator Jobin-Yvon T64000 Raman system (focal distance
0.640 m, aperture f/7.5) equipped with holographic gratings of
1800 grooves.mmꢀ1. The premonochromator stage was used in
the subtractive mode. The detection system was a liquid nitrogen
cooled non-intensified 1024ꢁ256 pixel (1") charge coupled
device chip.
The 514.5 nm line of an Ar+ laser (Coherent, model Innova 300-05)
was used as the excitation radiation, providing ca. 80 mW at the sam-
ple position. A 90º geometry between the incident radiation and the
collecting system was employed. The entrance slit was set to 200 mm,
as well as the slit between the premonochromator and the spectro-
graph. The samples were sealed in Kimax glass capillary tubes of
0.8 mm inner diameter. Under the above-mentioned conditions,
the error in wavenumbers was estimated to be within 1 cmꢀ1
.
In order to assess the potential antioxidant capacity of new
compounds – acting as free radical scavengers – several standard
methods for evaluation of antioxidant activity have been devel-
oped, either based on hydrogen atom transfer (HAT) reactions (1),
or on electron transfer (ET) processes (2). In the former (HAT), a
competition reaction scheme is generally followed, during which
the antioxidant and the substrate will compete for the scavenging
of thermally generated peroxyl radicals – oxygen radical absor-
bance capacity, and total radical trapping antioxidant parameter
tests being the most widely used.[20,21] The ET methods, in turn,
are colorimetric assays based on the reduction of a standard oxi-
dant (added to the analysed sample) that changes colour in the
process. Thus, they rely on the measurement of the sample’s absor-
bance at distinct wavelengths – the absorption maxima of the stan-
dard oxidant or radical and of its corresponding reduced species.
The absorbance change upon reduction, directly related to concen-
tration, yields the degree of sample reduction and consequently
the activity of the tested antioxidant. These kind of experiments
encompass the total phenol assay by Folin-Ciocalteau, the Trolox
equivalence antioxidant capacity test (also known as ABTS, from
2,20-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) which is the
reduced species), the ferric ion reducing antioxidant power method
and the cupric ion reducing antioxidant capacity assay.[20–22]
Since linoleic acid (LA) is a fatty acid widely used as an oxidation
target in many of these standard assays for antioxidant capacity,[20]
the screening of its oxidation status through Raman spectroscopy
will hopefully lay the grounds for the development of a simple,
reliable and accurate technique for the evaluation of the antioxi-
dant activity of a wide range of compounds. With a view to assess
the potential of Raman spectroscopy for the development of such a
standard antioxidant assessment technique, it was applied to the
autooxidation process of LA, followed for a six-day period upon
induction of oxidation (through heating). Computational simula-
tions, by quantum mechanical calculations at the density functional
theory (DFT) level, have been undertaken for the systems under
study – LA and its main oxidation products – with the purpose of
assisting the assignment of the corresponding spectral features,
and its changes throughout the experiment.
The spectra were always collected with an acquisition time of 10 s
for an average of ten accumulations, except for the region between
3200 and 3600 cmꢀ1 for which an integration time of 30 s was used.
Quantum mechanical calculations
The quantum mechanical calculations were performed using the
GAUSSIAN 03W program,[23] within the DFT approach in order to
properly account for the electron correlation effects (particularly
important in this kind of unsaturated systems). The widely
employed hybrid method denoted by B3LYP, which includes a
mixture of HF and DFT exchange terms and the gradient-
corrected correlation functional of Lee, Yang and Parr[24] as
proposed and parameterised by Becke[25] was used, along with
the double-zeta split valence basis set 6-31G**.[26]
Molecular geometries were fully optimised by the Berny
algorithm, using redundant internal coordinates:[27] the bond
lengths to within ca. 0.1 pm and the bond angles to within
ca. 0.1º. The final root-mean-square gradients were always less
than 3ꢁ10ꢀ4 hartree.bohrꢀ1 or hartree.radianꢀ1. No geometrical
constraints were imposed on the molecules under study.
Calculation of the harmonic vibrational wavenumbers was carried
out for all conformations (at the same level of theory) in order to
check for minima in the potential energy surface and to obtain the
predicted vibrational pattern (both Raman and IR) of the compounds.
Wavenumbers above 400 cmꢀ1 were scaled by a factor of 0.9614[28]
before comparing them with the experimental data, in order to
correct for the anharmonicity of the normal modes of vibration.
Oxidation experiments
After obtaining the Raman spectrum of freshly unfrozen LA,
about 1 g of fatty acid was placed in an air-circulating oven at a
temperature of ca. 60º C, inside a punctured eppendorf wrapped
in aluminium foil (in order to avoid photo-induced oxidation). For
a six-day period, samples were collected every day, and the
corresponding Raman spectra were obtained.
Results and discussion
Experimental
General spectral assignment
Chemicals
With a view to assist the interpretation of the Raman data
gathered along the autooxidation process of LA, DFT calculations
were performed yielding LA’s minimum energy conformation
(Fig. 1) as well as its predicted Raman spectrum. The same
LA (cis-9,cis-12-octadecadienoic acid) (≥ 99%) was purchased
from Sigma-Aldrich Química S.A. (Sintra, Portugal) and stored at
ꢀ20º C (to avoid oxidation).
wileyonlinelibrary.com/journal/jrs
Copyright © 2012 John Wiley & Sons, Ltd.
J. Raman Spectrosc. 2012, 43, 1991–2000

Autooxidation process in linoleic acid
theoretical procedure was undertaken for the acid’s major
autooxidation products, two hydroperoxide species (Fig. 1). This
combined spectroscopic (Raman) and theoretical approach
constitutes a relevant tool for the accurate interpretation of the
experimental vibrational data, allowing a full understanding of
the chemical changes due to the oxidation process over time.
The Raman spectrum obtained for LA (Fig. 2) comprises the
typical vibrational features of a cis-fatty acid, namely the band
at 3013 cmꢀ1 (assigned to the symmetric u(CH) mode associated
to the double bonds) and the signal at 1659 cmꢀ1, the most
intense of the spectrum (ascribed to the in-phase u(C = C)
stretching, Table 1).
former not being affected by oxidation which allows its use
for spectral normalisation in this region, enabling to detect
oxidation-induced intensity in the remaining bands. The feature
at 2855 cmꢀ1 arises mainly from the antisymmetric stretching
of the methylene group between the two chain double bonds,
which is annihilated by oxidation (Fig. 1).
The 1500–1800 cmꢀ1 spectral range also contains signals
which are prone to be affected by the oxidation process (Fig. 2).
The interval between 1600 and 1750 cmꢀ1 is particularly
important, since it reflects variations due to conformational
changes (namely cis–trans) and to the occurrence of conjugated
unsaturated systems (e.g. in hydroperoxides, Fig. 1) that arise
upon oxidation of the fatty acid. Additionally, the generation of
secondary oxidation products, such as conjugated aldehydes,
dienes and trienes (Fig. S1, Supplementary Material) may also
be monitored by the occurrence of distinct spectral features
within this interval.[8,15] The sample of pure LA displays a very
Some vibrational spectral bands have been used in previous
studies in order to monitor lipid oxidation, namely at 1266,
ca. 1660 and ca. 3010 cmꢀ1 [6,8]
The CH2 and CH3 symmetric
.
and antisymmetric stretching modes (aliphatic moiety of the
molecule) presently observed in the high wavenumber region
(2850 to 3000 cmꢀ1) are sensitive to the lateral packing effects
of the hydrocarbon chains[29] and can therefore be used as a
measure of the oxidation degree of the sample. These comprise
a shoulder at 2961 cmꢀ1, assigned to u(CH3)as, and two broad
bands at 2931 and 2904 cmꢀ1, corresponding to the symmetric
and antisymmetric (CH) stretchings of the aliphatic chains
(Table1). Two features due to antisymmetric stretching modes
of CH2 aliphatic groups appear at 2875 and 2855 cmꢀ1, the
weak signal at 1744 cmꢀ1 corresponding to the carboxylic uC=O
,
a strong band at 1659 cmꢀ1 (the most intense one in the
spectrum), and a shoulder at 1640 cmꢀ1 ascribed to the out-of-
phase stretching of the cis-C = C bonds (Table 1).
The LA band at 1441 cmꢀ1, ascribed to the CH2 scissoring
(aliphatic chains), is remarkably unchanged during the oxidation
experiment, thus being suitable for the normalisation of the
spectra collected after oxidation, at distinct time-points. In fact,
2
1
3
I
18
15
4
5
A
17
16
6
II
7
14
8
12
10
B
9
13
11
13
9
11
12
10
II
12
I
10
13
11
9
II
C
I
C
O
H
Figure 1. Calculated (B3LYP/6-31G**) lowest energy conformations for linoleic acid (A) and its two main autooxidation products (hydroperoxides)
(B and C). The highlighted area represents the major oxidation sites. I and II stand for the two different aliphatic moieties in these systems. This figure
is available in colour online at wileyonlinelibrary.com/journal/jrs
J. Raman Spectrosc. 2012, 43, 1991–2000
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jrs

N. F. L. Machado et al.
(CH3) + (CH2)
s(CH)cis
2600
2800
3000
3200
3400
3600
Wavenumber / cm-1
(C=C)
I
1640
(CC) + r(CH2)
sciss(CH2)
II
(CH) + t(CH2)
(CH2)
1744
+
I
II
200
400
600
800
1000
1200
1400
1600
1800
Wavenumber / cm-1
Figure 2. Experimental Raman spectrum (100–1800 cmꢀ1 and 2600–3600 cmꢀ1, at room temperature) for pure linoleic acid. (I and II correspond to
spectral expansions).
the ratio between the intensity of this band and that of the signal out-of-phase deformation of the CH olefinic moiety, is quite
at ca. 1660 cmꢀ1, ascribed to the u(C = C) modes, has been suitable for monitoring the oxidation of the fatty acid, involving
previously proposed as a measure of the oxidation degree of the formation of cis-trans conjugated systems. The band at 947
fatty acid samples.[18] The signal at 1303 cmꢀ1 corresponds cmꢀ1 is also related with to the oxidation site (Fig. 1), since it
to the CH2 twisting modes (aliphatic chains), with a minor involves the rocking of the CH2 group located between the two
contribution from the OH bending, and the one at 1266 cmꢀ1 double bonds in the LA molecule, coupled to the u(C–C) vibration
is assigned to the in-plane CH bending deformations from the of the bonds connecting the olefinic carbons with the aliphatic
cis-double bonds, coupled to the CH2 rocking modes from ones, with a minor contribution from the in-plane CH deforma-
the molecule’s moiety II (Fig. 1) from the methylene between tion of the olefinic groups (Table 1). The (C10C11C12) deformation
the two double bonds (Table 1). Therefore, this band is expected of the LA’s saturated moiety (Fig. 1) yields a feature at 910 cmꢀ1
to suffer major intensity changes upon oxidation.[8]
characteristic from this system and prone to undergo changes
,
When analysing the spectral interval between 1200 and during the oxidation experiment. On the contrary, the remaining
800 cmꢀ1, two intense features can be readily seen: an assembly two bands detected in this region, at 873 and 841 cmꢀ1 (Fig. 2),
of four superimposed bands between 1120 and 1060 cmꢀ1 and do not give any information on the oxidation process as
another group between 980 and 840 cmꢀ1 encompassing five they are associated to molecular groups far apart from the
bands, both sets displaying similar intensities (Fig. 2). The unsaturation sites (Table 1).
assignment of the first one is rather complex, as several
Below 800 cmꢀ1, only low intensity broad bands are observed
vibrational modes contribute to the same signal, thus being (Fig. 2), namely out-of-plane deformations of both the olefinic CH
virtually impossible to relate any change within this region with and OH groups, at 726 and 604 cmꢀ1, respectively. At lower
the oxidation process. In the 980–840 cmꢀ1 interval, in turn, wavenumbers, the Raman signals are generally originated by
the bands are well defined and can be assigned to specific vibrations involving the skeleton’s heavy atoms: the feature at
vibrational modes. The signal at 973 cmꢀ1, ascribed to the 403 cmꢀ1 corresponds to the CCC deformation of the aliphatic
wileyonlinelibrary.com/journal/jrs
Copyright © 2012 John Wiley & Sons, Ltd.
J. Raman Spectrosc. 2012, 43, 1991–2000

Autooxidation process in linoleic acid
Table 1. Experimental Raman wavenumbers for linoleic acid – at day 0 (LA) and at day 6 of the oxidation experiment
Experimental Raman Wavenumbers
aCalculated
bAssignment
(cmꢀ1
)
LA
sample at day 6
3370/3600
LA
HP 1
HP 2
3622
3609
3624
3608
u(OH)perox.
u(OH)carbox.
1640 + 1659
3415/3555
3290
3608
3292
3071
3013
2959
2928
2908
3068
FR ((1416 + 1659);3013)
ys(CH)cis
3013
2961
3029
2951
2936
2920
2918
2918
2895
2877
3034
2951
2936
2921
2920
2017
2895
3035
2946
2936
2920
2920
2918
2896
uas(CH3)met.
us(CH2)aliph.
us(CH3) + uas(CH2)aliph.
2931
2904
2875
2855
2793
2722
1744
2875
2856
2790
yas(CH2)aliph.
yas(C11H2)
combination mode
combination modes
u(C = O)carbox.
cu(C = O)aldehydes
y(C = C)iph.
1737
1695
1659
1776
1776
1776
1659
1676
1640
1670
u(C = C)ooph.
1635
1597
1457
1440
1418
1661
1622
1456
1442
1437
1426
1395
1662
1623
1456
1440
1435
1426
1397
u(C = C)iph. + cu(C = C)conj.
u(C = C)ooph.
1456
1441
1416
1457
1442
1437
1425
1388
1387
1374
1342
1336
1328
1310
1287
1279
1258
1254
1248
1238
1213
1211
1203
r(CH3)
sciss.(CH2)aliph.
sciss.(CH2)I
1402
1403
d(CH)olef.
1365
1346
1367
1346
1372
1333
1374
1343
ds(CH3)
o(CH2)aliph. + d(OH)carbox.
1320
1303
1266
1322
1303
1256
1323
1312
1285
1312
1307
1284
o(CH2)I
t(CH2)aliph.
1256
1250
1245
1261
1255
1246
d(CH)cis + t(CH2)I + v(CH2)II + dv(C11H2) + ed(CH)trans
1223
1219
1190
1234
1221
1193
1226
1212
1208
1207
1186
1179
1227
1214
1212
1202
1187
1179
o(CH2)II
t(CH2)I
t(CH2)I + fd(CH)olef. + d(OH)carbox.
1165
1142
1164
1179
1174
1157
d(OH)carbox. + o(CH2)I + t(CH2)II + eu(C–C)conj. + cu(C–C)
u(C–O)carbox. + d(OH)carbox. + t(CH2)I
u(C–C)olef. + u(C–O)perox.
d(CH)olef. + t(CH2)aliph. + r(CH3) + dt(C11H2) +ed(CH)perox.
t(CH2)II + r(CH3) + dt(C11H2) +fd(CH)olef.
t(CH2)I + u(C–O)carbox. + d(OH)carbox.
u(C–C)I + u(C–O)carbox. + eu(C–O)perox. + t(CH2)I
g1123
1111
1082
1077
1065
1140
1122
1088
1070
1052
1051
1028
1025
1025
1139
1115
1090
1072
1061
1045
1029
1029
1024
1112
1088
1077
1066
1091
1089
1071
1055
1044
1041
1025
1029
1028
1023
1002
995
u(C–C)aliph.
1021
1001
u(C–C)I
u(C–C)aliph. + u(C–C)olef.
(Continues)
J. Raman Spectrosc. 2012, 43, 1991–2000
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jrs

N. F. L. Machado et al.
Table 1. (Continued)
Experimental Raman Wavenumbers
aCalculated
HP 1
bAssignment
(cmꢀ1
)
LA
sample at day 6
973
LA
HP 2
973
980
978
974
933
894
867
863
846
801
756
726
695
597
995
976
993
985
g(CH = CH)ooph. +
du(C–C)olef. + u(C–C)aliph.
947
910
873
952
907
872
r(C11H2) + u(C10-C11) + u(C11-C12) + d(CH)olef.
Δ(C10C11C12
)
859
847
r(CH3) + u(C–C)II + r(CH2)I
841
805
777
726
680
604
576
844
844
796
756
744
841
803
u(C–C)I + r(CH2)I
u(C1C2) + r(CH2)I
Δ(OCO)carbox. + r(CH2)I
g(CH = CH)cis
779
722
766
598
g(C9H = C10H)
601
584
596
g(OH)carbox.
Δ(C8C9 = C10) + Δ(C11C12 = C13
)
g490
g456
481
466
440
406
480
474
460
398
Δ(C–C = C)both + Δ(CCO)perox. + g (OH)carbox.
Δ(COO)perox. + Δ(CCC)aliph. + g(OH)carbox.
403
373
241
402
372
217
406
368
248
Δ(CCC)aliph. + Γ (CC = C)
Δ(CCC)II
235
222
234
skeletal mode
The most relevant bands undergoing changes upon oxidation are represented in bold. aAt the B3LYP/6-31G** level; wavenumbers above 400 cmꢀ1
were scaled by 0.9614.[28] bAtoms are numbered according to Fig. 1(A); the subscripts I and II refer to the distinct aliphatic moieties of the molecule
(Fig. 1). d – in-plane deformation, t – twisting, r – rocking, o – wagging, sciss. – scissoring, g – out-of-plane deformation, Δ – in-plane deformation of
c
skeleton atoms, iph – in-phase, ooph – out-of-phase; as and s refer to antisymmetric and symmetric modes, respectively. Assigned to secondary
oxidation products, according to.[8] dFor LA. eFor the hydroperoxides. fFor both LA and the hydroperoxides. gIntermediate Raman signals
appearing during the oxidation process, not visible at day 6.
chains coupled to C = C–C out-of-plane deformations of the
olefinic groups. The band at 373 cmꢀ1 is related to the CCC
deformations of the aliphatic chain II and the signal at 241 cmꢀ1
arises from a skeletal mode (Table 1). It should be reminded
that oxidation leads, in a first stage, to the transformation of
the cis-cis molecule into a cis-trans conjugated system (Fig. 1),
therefore strongly altering the molecule skeleton’s architecture
which leads to changes in this region of the Raman spectrum.
(Figs. 1 and S1). Even though this is a good indication of the
oxidation process, several distinct chemical systems contribute
to the increase in this band’s intensity which renders it quantita-
tively meaningless.
Below 3100 cmꢀ1, important changes can be observed in the
Raman profile upon oxidation (Fig. S3, Supplementary Material):
the intensity of the us(CH) of cis-double bonds, at 3013 cmꢀ1
,
decreases continuously (Fig. S3). This behaviour is to be expected,
since the oxidation process leads to the disruption of the isolated
cis-double bonds, and to the formation of a cis-trans conjugated
system. On the other hand, the bands at 2931, 2904 and 2855
cmꢀ1 (Fig. S3), also display well-defined intensity trends. The signal
at 2875 cmꢀ1, in turn, does not to vary significantly upon oxidation,
thus being used to normalise this spectral region. The superim-
posed bands detected between 2900 and 2950 cmꢀ1 increase
upon oxidation (Fig. S3) thanks to the contribution of the
symmetric and antisymmetric (CH2) stretching modes from the
lateral chains I and II (Fig. 1) of the newly formed peroxides (Table 1).
Furthermore, the signal at 2855 cmꢀ1 decreases during oxidation
(Fig. S3), as it is related to the CH2 located between the two double
bonds, specific from non-oxidised LA (Table 1, Fig. 1).
Additionally, aiming at an accurate assessment of the bands
due to the oxidation process and of their intensity variation
tendency, the spectrum of pure LA (at day 0) was subtracted from
each spectrum collected during the experiment (one per day, for
6 days). Discarding the negative profiles resulting from the
decrease in some LA bands (Fig. 3), four new visible signals
Influence of the oxidation process in the Raman pattern
3600–2700 cmꢀ1 spectral region
The variations observed in this spectral range, upon oxidation,
are mainly due to the formation of hydroperoxides and small-
chain alcohols (Fig. S1), as well as to changes in the aliphatic
chain architecture and corresponding saturation degree
(detected below 3100 cmꢀ1).
Although being often undetectable by Raman spectroscopy,
the u(OH) bands were observed upon a thorough scan of the
3600–3200 cmꢀ1 spectral interval (Fig. S2, Supplementary
Material). A very weak broad band between ca. 3555 and
3415 cmꢀ1 is detected at day zero, being clearly more intense
at the end of the experiment (six days after) (Table 1, Fig S2). This
is due to the presence of an increasing number of OH groups in
the sample, contributing to this feature, belonging either to
the primary hydroperoxides or to secondary oxidation products
wileyonlinelibrary.com/journal/jrs
Copyright © 2012 John Wiley & Sons, Ltd.
J. Raman Spectrosc. 2012, 43, 1991–2000

Autooxidation process in linoleic acid
deserve attention: (1) at 2983 cmꢀ1, assigned to the u(CH) of the
cis-double bond from the pure fatty acid (in agreement with the
calculations); (2) at 2944, 2926 and 2907 cmꢀ1 (Fig. 3) ascribed to
CH2 stretching modes (according to the calculations undertaken
for the peroxides, Table 2).
the spectral interval between 1760 and 1660 cmꢀ1 can be
observed, thanks to the contribution of distinct oxidation
products containing a carbonyl group. As to the band at
1659 cmꢀ1, ascribed to the (C = C) stretching of the cis-double
bonds of the fatty acid, a variation with no specific trend can
be perceived when the spectra are normalised relative to the
1441 cmꢀ1 signal (Fig. S4-AI). This can be understood in the light
of the chemical changes taking place during the autooxidation
process: thermally stressed samples are known to undergo
conformational rearrangements (yielding trans-systems),[5,8,15,30]
and the oxidation process leads to an additional formation of
trans-conjugated species (peroxides and secondary oxidation
products). This diversity of unsaturated systems contributing to
the Raman profile around 1700–1600 cmꢀ1 justifies the uneven
behaviour of the 1659 cmꢀ1 band, which does not allow to
relate the corresponding intensity variation to the degree of LA
oxidation. In fact, when this signal is used to normalise the
spectrum (Fig. S4-BI), the neighbouring Raman bands maintain the
same trend as the non-normalised ones, evidencing the irrelevance
of this feature for evaluating the oxidation state of the sample.
The shoulder observed at 1640 cmꢀ1 was found to be blurred by
the widening of the 1659 cmꢀ1 feature (reflecting the presence of
1800–1550 cmꢀ1 spectral region
For LA, containing only isolated double bonds, three main bands are
observed in this interval: at 1744 cmꢀ1, due to u(C = O), and at 1659
cmꢀ1 (the most intense feature) with a shoulder at 1640 cmꢀ1
(Fig. S4, Supplementary Material, day 0), corresponding to the
simultaneous u(C = C) from the isolated cis-double bonds (Table 1).
Throughout the oxidation process, three new Raman bands
arise at 1695, 1635 and 1597 cmꢀ1 (Fig. S4). The signal at 1695
cmꢀ1, in particular, can be assigned to the u(C = O) mode of
conjugated aldheydes,[8,15] which are red-shifted due to conjuga-
tion between the C = O and C = C double bonds. Additionally,
the Raman bands characteristic of LA in this spectral interval
undergoes changes during the experiment (Fig. S4). The very
weak signal at 1744 cmꢀ1 suffers an intensity enhancement
due to the contribution of carbonyl groups from chemical entities
produced by oxidation. Also, an intensity enhancement in
0.15
0.10
0.05
0.00
day 2
day 3
day 4
day 5
day 6
I
II
III
-0.05
-0.10
-0.15
-0.20
2600
2800
3000
3200
Wavenumber / cm-1
I
II
III
Figure 3. Difference Raman spectra (2600–3200 cmꢀ1) for linoleic acid, at each day of the oxidation experiment, relative to the spectrum at day 0.
(The arrows reflect the trend of the signal intensity changes upon oxidation. I, II and III correspond to spectral expansions). This figure is available in
colour online at wileyonlinelibrary.com/journal/jrs
J. Raman Spectrosc. 2012, 43, 1991–2000
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jrs

N. F. L. Machado et al.
Table 2. Trends detected in the difference Raman spectra of linoleic acid upon oxidation – spectrum of oxidised LA at day 6 minus spectrum at day 0
aTrends in intensity change: + refers to an increase; - refers to a decrease. bAtoms are numbered according to Fig. 1(A). White cells – literature and present
work; dotted cells – present work, for LA; light grey cells – present work, for hydroperoxides; dark grey cells – literature. d – in-plane deformation, t – twisting,
r – rocking, o – wagging, sciss. – scissoring, g – out-of-plane deformation, Δ – in-plane deformation of skeleton atoms, Γ – out-of-plane deformation of
skeleton atoms; ooph – out-of-phase; as and s refer to antisymmetric and symmetric modes, respectively. cSecondary oxidation products, according to[8]
.
dAssigned according to[7]
.
several oxidation species), two new bands arising at 1635 and
1597 cmꢀ1. The former can be tentatively assigned to u(C= C) of
conjugated trienes (Table 2), in agreement with to Muik et al.[8]
However, its intensity increases faster than that of the neighbouring
signals assigned to oxidation products (Fig. 4), thus supporting the
possibility of this band being assigned to different chemical species.
In turn, the feature at 1597 cmꢀ1, not previously reported in spec-
troscopic studies of oxidation processes of edible oils, is presently
assigned (in the light of the DFT results) to the out-of-phase u(C= C)
mode of the conjugated peroxides (Table 1).
Analysing the difference Raman spectra collected at different
days during the experiment (6 days) relative to the spectrum at
day 0 (free LA) allows to clearly distinguish the main features
that arise from the oxidation process (Fig. 4). A new band can
be found at 1672 cmꢀ1, due to the u(C = C) mode of isolated
trans-double bonds,[7] while at 1663 and 1651 cmꢀ1, two signals
with undefined trends are visible (Fig. 4-III). While the latter can
be assigned to the u(C = C) of the isolated cis-double bonds, the
former seems to have contributions from distinct systems, which
renders its assignment quite ambiguous.
Two weak bands evidencing a clear intensity enhancement
throughout the oxidation process (Fig. S5) are found at 1190 and
1165 cmꢀ1. For the oxidised samples, the former has contributions
from d(CH) of the conjugated moiety of the peroxides, as well as
from their CH2 wagging modes, (Fig. 4-II, Table 2). In turn, the band
at 1165 cmꢀ1, clearly visible after subtracting the spectrum at day 0
from those of the oxidised species (Fig. 4-II), may be related to the
presence of 2,4-decadienal, which is expected to have a strong
u(CC) feature at this wavenumber value.[8] Furthermore, this
signal seems to be enhanced by the contribution of a vibrational
mode from the peroxides, mainly due to u(CC) of the conjugated
moiety (Table 2) (which is predicted to be rather intense by the
DFT calculations).
Regarding the set of five bands observed between 980 and
830 cmꢀ1, intensity variations are readily detected, the first three
ones, at 973, 947 and 910 cmꢀ1, undergoing an intensity decrease
(Fig. S5). The feature at 973 cmꢀ1, assigned to the cis-olefinic moiety
(Table 1), was used by other authors for evaluating the oxidation of
edible oils,[8] in good agreement with the experimental and DFT
results presently shown for LA (Figure S5, Table 2). The two bands
at 947 and 910 cmꢀ1 correspond to vibrational modes exclusive
of the non-oxidised LA, the former to C11H2 rocking and the latter
to the C10C11C12 deformation mode (Table 1). Both undergo an
intensity decrease (Fig. S5), reflecting the rearrangement from a
non-conjugated (C= C–C–C = C) to a conjugated (C= C–C = C)
system upon oxidation. Vibrational modes specific from the
peroxides are to be found within this region, namely those involv-
ing the out-of-plane (CH) bending exclusively found in the perox-
ides’ conjugated moiety (Table 2). These features are probably
the reason for the irregular behaviour of the aforementioned
band. Finally, the intensity of the signal at 841 cmꢀ1, ascribed to
the simultaneous u(CC) vibrations of the whole aliphatic chain,
1550–650 cmꢀ1 spectral region
Apart from the band at 1441 cmꢀ1, which is unaffected by
oxidation (Fig. S5, Supplementary Material) and therefore useful
as a reference for the evaluation of the intensity changes during
this process, other important signals occur in this wavenumber
interval. The intensity decrease of the band at 1266 cmꢀ1
(Fig. S5) appears as an important feature in this kind of
experiment, since it is mainly due to the CH deformations of
the two cis-double bonds (Table 1) and may thus be used as a
suitable probe of the annihilation of these cis-double bonds in
the fatty acid, occurring during oxidation.
wileyonlinelibrary.com/journal/jrs
Copyright © 2012 John Wiley & Sons, Ltd.
J. Raman Spectrosc. 2012, 43, 1991–2000

Autooxidation process in linoleic acid
III
day 2
day 3
day 4
day 5
day 6
0.6
0.4
0.2
II
I
0.0
-0.2
200
400
600
800
1000
1200
1400
1600
1800
Wavenumber / cm-1
I
II
III
Figure 4. Difference Raman spectra (100–1800 cmꢀ1) for linoleic acid, at each day of the oxidation experiment, relative to the spectrum at day 0.
(The arrows reflect the trend of the signal intensity changes upon oxidation. I, II and III correspond to spectral expansions). This figure is available in
colour online at wileyonlinelibrary.com/journal/jrs
decreases markedly during oxidation. This is probably a conse-
quence of the structural rearrangement of the aliphatic chain from
the acid to the peroxides, lower intensities having been presently
predicted for the vibrational modes of the latter. Additionally, the
band at 726 cmꢀ1, assigned to the (CH) out-of-plane bending of
the cis-double bonds, loses intensity (Fig. S5) during the experi-
ment, as expected, reflecting the decrease in the ratio (LA:oxidation
products).
bending from the conjugated structure of the peroxides,
gradually formed as a function of time (Table 2).
Finally, all bands detected bellow 650 cmꢀ1 display quite weak
intensities (Fig. 2), which renders spectral subtraction unreliable
in this interval, as baseline subtraction may interfere with the
results. Hence, and even though some changes were observed upon
oxidation (Table 1) within this spectral region, it was considered
useless for evaluating the sample’s oxidation degree.
The difference spectra in the 1000–700 cmꢀ1 interval com-
prises four negative signals, at 973 910, 841 and 726 cmꢀ1
(Fig. 4-I, Table 2), corresponding to an intensity loss of the
bands resulting from the non-oxidised LA (Fig. S5). Mean-
while, in the 900–850 cmꢀ1 wavenumber range, although
some intensity enhancement seems to occur, the trend is
not constant for the entire time period of the experiment
(Fig. 4-I). This is probably due to contributions from peroxides
and, eventually, to the presence of volatile secondary oxida-
tion products (Fig. S1) that gradually degrade or evaporate,
thus contributing to the complex behaviour observed in this
spectral range. The difference spectra relative to day 0
(unoxidised sample) also unveil a new band at 819 cmꢀ1
(Fig. 4-I), not detectable by simple analysis of the Raman pro-
files registered at the distinct time-points. In the light of the
calculations, this feature is assigned to the CH out-of-phase
Conclusions
The spectral features presently gathered for LA, along with the clear
changes detected throughout the autooxidation experiment,
allowed to conclude that Raman spectroscopy is a suitable and
reliable technique for monitoring the oxidation degree in fatty
acids. Furthermore, the use of molecular orbital calculations in
combination with the spectroscopic measurements enable to
achieve an accurate vibrational assignment and to ascribe specific
spectral features to particular groups, both for the fatty acid and
for its oxidation products, in the light of the structural changes
occurring during the oxidation process. The intensity decrease
of the bands at 3013 cmꢀ1 and 1266 cmꢀ1 – associated to the
LA’s cis-double bonds – can be straightforwardly related to the
J. Raman Spectrosc. 2012, 43, 1991–2000
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jrs

N. F. L. Machado et al.
[4] J. C. Allen, R. J. Hamilton, Rancidity in Foods, Elsevier Applied Science,
London, New York, 1989.
[5] H. Yang, J. Irudayaraj, J. Am. Oil Chem. Soc. 2001, 78, 889.
[6] M. D. Guillen, N. Cabo, J. Agric. Food Chem. 1999, 47, 709.
[7] J. M. Chalmers, P. R. Griffiths, Handbook of Vibrational Spectroscopy,
J. Wiley, New York, 2002.
oxidation status of the sample, as well as the smooth decline of the
feature at 2855 cmꢀ1. A previously proposed approach,[31] based
on the ratio between the area of the bands located in the
1680–1620 cmꢀ1 interval and those at ca. 1660 cmꢀ1, was found
to be unsuitable for the present analysis (involving thermal stress)
due to the complexity of these features.[8]
[8] B. Muik, B. Lendl, A. Molina-Diaz, M. J. Ayora-Canada, Chem. Phys.
Lipids 2005, 134, 173.
In conclusion, the intensity decrease of the Raman bands related
to the cis-double bonds of an unsaturated fatty acid emerges as
an unequivocal sign of degradation through autooxidation and
hence as an important tool for probing the acid’s oxidation degree.
Actually, the fatty acid contains only isolated cis-double bonds
which constitute the oxidation sites within the molecule, being
steadily and gradually suppressed throughout this process, leading
to a marked weakening of the corresponding vibrational bands.
In turn, the new Raman features arising upon oxidation, due to
derivatives formed along the process, were found to change
continuously throughout the experiment since these are often
transient oxidation products.
The success of Raman spectroscopy in probing the oxidation
state of a fatty acid sample, presently verified, coupled to a simpler
setup and virtually no sample preparation, renders this technique
a very useful and promising tool in food industry, namely for
assessing the quality of edible oils in situ (e.g. rancidity, unsatura-
tion pattern and presence of additives). Aiming at such an
application, the procedure presently described should be extended
to other isolated fatty acids in order to build a database of
characteristic vibrational patterns, related to distinct structural
features, allowing their unequivocal identification and the assess-
ment of structural modifications within their molecule (e.g. by
oxidation). Furthermore, the work presently carried out paves the
way for developing a standard method for screening antioxidant
capacity through Raman spectroscopy, using LA as the oxidised
substrate. The antioxidant activity of the analysed samples
would be measured by their ability to avoid autooxidation of this
fatty acid.
[9] S. Armenta, J. Moros, S. Garrigues, M. De La Guardia, Crit. Rev. Food
Sci. 2010, 50, 567.
[10] F. R. Vandevoort, A. A. Ismail, J. Sedman, J. Am. Oil Chem. Soc. 1995,
72, 873.
[11] G. L. Johnson, R. M. Machado, K. G. Freidl, M. L. Achenbach, P. J. Clark,
S. K. Reidy, Org. Process Res. Dev. 2002, 6, 637.
[12] J. K. Agbenyega, M. Claybourn, G. Ellis, Spectrochim. Acta A 1991, 47, 1375.
[13] F. R. Vandevoort, J. Sedman, G. Emo, A. A. Ismail, J. Am. Oil Chem. Soc.
1992, 69, 1118.
[14] J. Gray, J. Am. Oil Chem. Soc. 1978, 55, 539.
[15] J. Dubois, F. van de Voort, J. Sedman, A. Ismail, H. Ramaswamy, J. Am.
Oil Chem. Soc. 1996, 73, 787.
[16] V. Baeten, P. Hourant, M. T. Morales, R. Aparicio, J. Agr. Food. Chem.
1998, 46, 2638.
[17] N. A. Marigheto, E. K. Kemsley, M. Defernez, R. H. Wilson, J. Am. Oil
Chem. Soc. 1998, 75, 987.
[18] H. Sadeghi-Jorabchi, R. H. Wilson, P. S. Belton, J. D. Edwardswebb, D.
T. Coxon, Spectrochim. Acta A 1991, 47, 1449.
[19] B. Muik, B. Lendl, A. Molina-Diaz, M. J. Ayora-Canada, Anal. Chim. Acta
2003, 487, 211.
[20] D. Huang, B. Ou, R. L. Prior, J. Agr. Food. Chem. 2005, 53, 1841.
[21] R. Apak, K. Guclu, B. Demirata, M. Ozyurek, S. E. Celik, B. Bektasoglu,
K. I. Berker, D. Ozyurt, Molecules 2007, 12, 1496.
[22] C. Sánchez-Moreno, Food Sci. Technol. Int. 2002, 8, 121.
[23] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R.
Cheeseman, J. J. A. Montgomery, T. Vreven, K. N. Kudin, J. C. Burant, J.
M. I. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi,
G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara,
K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O.
Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, V.
Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K.
Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski,
S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D.
Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul,
S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz,
I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng,
A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen,
M. W. Wong, J. Gonzalez, J. A. Pople, in Gaussian 03, Revision D.01,
Gaussian, Inc, Wallingford CT, 2004.
Acknowledgements
The authors acknowledge financial support from the Portuguese
Foundation for Science and Technology – PEst-OE/QUI/UI0070/
2011. NFLM acknowledges PhD grant (SFRH/BD/40235/2007).
[24] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B Condens. Matter 1988, 37, 785;
B. Miehlich, A. Savin, H. Stoll, H. Preuss, Chem. Phys. Lett. 1989,
157, 200.
[25] A. D. Becke, Phys. Rev. A 1988, 38, 3098;
Supporting information
A. D. Becke, J. Chem. Phys. 1993, 98, 5648.
[26] G. A. Petersson, A. Bennett, T. G. Tensfeldt, M. A. Allaham, W. A. Shirley,
Supporting information may be found in the online version of
this article.
J. Mantzaris, J. Chem. Phys. 1988, 89, 2193.
[27] C. Peng, P. Y. Ayala, H. B. Schlegel, M. J. Frisch, J. Comput. Chem.
1996, 17, 49.
[28] J. P. Merrick, D. Moran, L. Radom, J. Phys. Chem. A 2007, 111, 11683.
[29] A. M. Amorim da Costa, M. P. M. Marques, L. A. E. Batista de Carvalho,
J. Raman Spectrosc. 2003, 34, 357;
References
[1] E. N. Frankel, Prog. Lipid Res. 1980, 19, 1.
[2] K. Warner, N. A. M. Eskin, Methods to Assess Quality and Stability of
Oils and Fat-containing Foods, AOCS Press, Champaign, Ill., 1995.
[3] IUPAC, Standard Methods for the Analysis of Oils, Fats and Derivatives
(7th edn), Union of Pure and Applied Chemistry, Blackwell Scientific
Publications, Oxford, U.K., 1987.
A. M. Amorim da Costa, M. P. M. Marques, L. A. E. Batista de Carvalho,
Vib. Spec. 2004, 35, 165.
[30] A. A. Christy, Open Spec. J. 2010, 4, 41.
[31] H. Sadeghi-Jorabchi, P. J. Hendra, R. H. Wilson, P. S. Belton, J. Am. Oil
Chem. Soc. 1990, 67, 483.
wileyonlinelibrary.com/journal/jrs
Copyright © 2012 John Wiley & Sons, Ltd.
J. Raman Spectrosc. 2012, 43, 1991–2000