Food Research International 45 (2012) 103108

Contents lists available at SciVerse ScienceDirect

Food Research International

journal homepage: www.elsevier.com/locate/foodres

Chemical changes of thermoxidized virgin olive oil determined by excitation

emission uorescence spectroscopy (EEFS)
Noelia Tena, Ramn Aparicio, Diego L. Garca-Gonzlez
Instituto de la Grasa (CSIC), Padre Garca Tejero 4, E-41012 Sevilla, Spain

a r t i c l e

i n f o

Article history:
Received 5 August 2011
Accepted 12 October 2011
Keywords:
Excitationemission matrix (EEM)
PARAFAC
Thermoxidation
Virgin olive oil
Phenols
Oxidation products

a b s t r a c t
The uorophores present in virgin olive oil evolve during any thermoxidation process and thereby uorescence spectroscopy can be used as a straightforward approach to study the oxidative degree of oils.
Samples of heated virgin olive oils collected from a fryer every two hours up to 94 h were analyzed to
study their excitationemission matrix (EEM) and the evolution of the uorescent composition during
the thermoxidation process. The potential usefulness of EEM to study thermoxidized virgin olive oils has
been explored. For this purpose, parallel factor analysis (PARAFAC) was used to interpret the results. The
emission prole obtained from three-factor PARAFAC model calculated for 47 samples of thermoxidized
oil in the range of ex = 250298 nm and em = 300700 nm allows the decomposition of the uorescence
landscape in specic information about phenols, pigments and oxidation products respectively. The
chemical interpretation from the PARAFAC model was checked with chemical analysis of phenols, tocopherols and chlorophyll pigments.
2011 Elsevier Ltd. All rights reserved.

1. Introduction
The ability of virgin olive oil to emit uorescent radiation is due to
the presence of uorescent compounds as chlorophylls, pheophytins,
tocopherols, Vitamin E and oxidative compounds (Galano, Durn,
Correa, Roldn, & Rodriguez, 2003; Sikorska et al., 2004). Previous
studies based on uorescence spectroscopy of olive oil have associated several emission spectrum regions with these compounds present
in olive oil. The region between 300 and 400 nm is attributed to the
tocopherols and phenols (Sikorska, Grecki, Khmelinskii, Sokorski, &
Koziol, 2005; Zandomeneghi, Carbonaro, & Caffarata, 2005), the
region of 440445 nm shows good correlation with the K232 and
K270 (Kyriakidis & Skarkalis, 2000), the derivatives of vitamin E are
associated with the region 400600 nm (Dupuy et al., 2005), and
the bands in the region of 600700 nm are associated with the
chlorophyll and pheophytin pigments (Dupuy et al., 2005).
Different approaches based on uorescence spectroscopy have
been implemented for the characterization of edible oils. The differences between the uorescence emission spectra of olive oil and
other edible oils (Sikorska et al., 2005) and between different categories of olive oil (Nicoletti, 1990) have been previously described. The
uorescence spectra have also been used to detect adulterations, as
the presence of hazelnut oil in virgin olive oil (Sayago, Morales, &
Aparicio, 2004), and even to determine the degree of oil oxidation

Corresponding author. Tel.: + 34 954 61 15 50; fax: + 34 954 61 67 90.

E-mail address: dluisg@cica.es (D.L. Garca-Gonzlez).
0963-9969/$ see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodres.2011.10.015

and thermal deterioration in virgin olive oils (Poulli, Chantzos,

Mousdis, & Georgiou, 2009; Tena, Garca-Gonzlez, & Aparicio,
2009). On the other hand, the uorescence excitation spectra with
emission at 330 and 450 nm have been used to control the evolution
of phenols, vitamin E and degradation of hydroperoxides during
heating (Cheikhousman et al., 2005). Furthermore, applications with
total synchronous uorescence spectroscopy and excitationemission
uorescence spectroscopy (EEFS) have been developed to monitor
changes in the olive oil during storage in different conditions
(Sikorska et al., 2008).
A technique selected to ensure the safety and quality of any food
product should be rapid and easy to use (Belton et al., 1995). Fluorescence spectroscopy, being a rapid technique, provides valuable information on thermolabile compounds (e.g. phenols, vitamin E,
pigments), and therefore, this technique has the advantage over
other spectroscopic methodologies of providing specic information
on oxidation. In consequence, it can be proposed as a suitable tool
for monitoring the deterioration degree of frying olive oils. Synchronous uorescence and excitationemission uorescence spectroscopies are preferable for the analysis of complex multicomponent
samples such as edible oils, compared with conventional uorescence
spectroscopy (Poulli, Chantzos, Mousdis, & Georgiou, 2009; Poulli,
Mousdis, & Georgiou, 2009). In particular the excitationemission
uorescence spectroscopy (EEFS) involves the simultaneous acquisition
of a range of excitation wavelengths (ex) and emission wavelengths
(em). As a result, a total uorescence prole of the sample over the
scanned range is obtained. This uorescence landscape is called the
excitationemission matrix (EEM) and it has the ability of obtaining

104

N. Tena et al. / Food Research International 45 (2012) 103108

simultaneous information of different compounds in the sample. This

procedure improves the selectivity of uorescence spectroscopy and it
has proved to be useful in discerning olive oils by slight differences in
their composition.
The statistical tools to interpret the EEFS results require dividing
the information provided by the three-dimensional matrix (EEM)
into two-dimensional arrays. The parallel factor analysis (PARAFAC)
has the advantage of working with three-dimensional arrays
providing a limited number of factors that represent the whole
matrix. The model obtained with PARAFAC simplies the interpretation
of the excitationemission proles and it provides a unique solution,
regardless of the type of rotation of the factors (Harshman, 1970).
The objective of this work is to propose the excitationemission
uorescence spectroscopy combined with PARAFAC algorithm as a
tool to monitor the changes in the uorescence composition of virgin
olive oil during thermoxidation. For this purpose, the information
obtained from the EEM was processed to determine when the oil
had to be discarded for human consumption by correlating the most
signicative spectral bands with the percentage of total polar
compounds. The chemical assignment of the spectral bands was done
according to the work reported by other authors and supported by the
chemical analysis of phenols, tocopherols, and chlorophyll pigments.
2. Materials and methods
2.1. Sample preparation
The study was carried out with forty-seven samples of thermoxidized virgin olive oils. These samples were obtained through a
thermoxidation process carried out in a 4 L domestic fryer with automatic temperature controller (Heidolph EKT 3001, Schwabach, Germany).
The fryer was set at 190 C for 94 h in cycles of 8 h per day. A sample
of 40 mL was collected every 2 h until the end of the heating process.
Samples were kept in brown glass vials at 4 C until further chemical
and spectroscopic analyses.
2.2. Analysis of total polar compounds
The percentage of total polar compounds (% PC) was determined
gravimetrically according to the IUPAC Standard Method No. 2.507.
Nonpolar and polar fractions were separated from 1 g of oil by silica
gel column chromatography (20 g silica adjusted to a water content
of 5%, w/w) hexane/diethyl ether (90:10, v/v) and 150 mL diethyl
ether as elution systems, respectively. The nonpolar fraction was
eluted with 150 mL of n-hexane/diethyl ether (90:10, v/v), while
the polar fraction was eluted with 150 mL of diethyl ether. Efciency
of the separation was conrmed by TLC using n-hexane/diethyl ether/
acetic acid (80:20:1, v/v/v) and visualized with iodine vapor. The
percentage of polar fraction was calculated by weighing both fractions
after the evaporation of the solvents.
2.3. Excitationemission uorescence spectroscopy (EEFS)
Spectrouorimetric measurements were performed with a Cary
Eclipse (Varian Ibrica, Madrid, Spain). This instrument was equipped
with a continuous xenon lamp, excitation and emission monochromators, and a photomultiplier. A 10 mm 10 mm quartz cell (3 mL) was
used for right-angle RA uorometry. The wavelength range for the
emission monochromator was 300700 nm, and for the excitation
monochromator was 250 nm at 298 nm, intervals for each dimension
being 1 and 4 nm respectively. The polarizer operated in automatic
mode. The excitation and emission slits were both of 5 nm. The photomultiplier voltage was set at 600 V to avoid saturation. The time taken
to acquire a spectrum was of 10 min (intermediate speed). All data
were exported to an Excel spreadsheet and were subsequently treated
with the softwares Matlab v7.8 (The MathWorks, Natick, MA) and

Statistica v7.0 (Statsoft, Tulsa, OK). The experimental conditions allowed

getting spectra of satisfactory intensity, resolution and signal-to-noise
ratio. After each sample the cell was cleaned using detergent, followed
by Milli-Q hot water and acetone to dry it and eliminate possible rests
of samples. Each sample was analyzed in duplicated.
2.4. Determination of tocopherols
Tocopherols were analyzed by following the IUPAC Standard
Method No. 2.432. A solution of oil in hexane (10 mg/mL) was
analyzed by HPLC (Agilent Technologies 1200) on a silica gel column
(Merck, Superspher Si60, particle size 4 m, 250 mm 4 mm i.d.), eluting with hexane/2-propanol (99:1, v/v) at a ow rate of 1 mL/min. The
injection volume was of 20 L. A uorescence detector (Agilent Technologies 1100) with the excitation wavelength set at 290 nm and emission
wavelength at 330 nm was used. The quantication was carried out by
using an external calibration with a solution of -tocopherol/Hexane
(0.55.0 g/mL) as standard.
2.5. Determination of phenols
A standard solution (0.5 mL) made of p-hydroxyphenylacetic
(0.12 mg/mL) and o-coumaric acids (0.01 mg/mL) in methanol was
added to a sample of ltered virgin olive oil (2.5 g). A rotary evaporator at 40 C under vacuum was used to evaporate the solvent and the
oily residue was dissolved in 6 mL of hexane.
The diol-bonded phase cartridge was conditioned according to
Mateos et al. (2001). After the sample loading, the nal residue was
extracted with 500 L of methanolwater (1:1 v/v) at 40 C and a ltrated aliquot (20 L) of the nal colorless solution was injected onto
the HPLC system (LaChrom Elite Tokio, Japan), equipped with a diode
array detector. The column was a Lichrospher 100RP-18 column
(4.0 mm i.d. 250 mm; 5 m, particle size) (Darmstadt, Germany)
maintained at 30 C. The gradient elution, at a ow rate of 1.0 mL/min,
was achieved by using the following mobile phases: a mixture of
water/phosphoric acid (95.5:0.5 v/v) (solvent A) and methanol
acetonitrile (50:50 v/v) (solvent B). The change of solvent gradient
was programmed as follows: from 95% (A)-5% (B) to 70% (A)-30%
(B) in 25 min; 62% (A)-38% (B) in 10 min; 62% (A)-38% (B) in
5 min; 55% (A)-45% (B) in 5 min; 47.5% (A)-52.5%(B) in 5 min and
100% (B) in 5 min, followed by 5 min of maintenance. The chromatographic signals were obtained at 235, 280, and 335 nm.
Quantication of phenols cinnamic acid and lignans was carried out
at 280 nm by using p-hydroxyphenylacetic acid as internal standard.
Quantication of avones was done at 335 nm by using o-coumaric
acid as internal standard. The response factors and recoveries were
based on the procedure carried out by Mateos et al. (2001).
2.6. Determination of chlorophyll pigments
The determination of chlorophyll pigments was carried out by
implementing the method proposed by Gertz and Fiebig (2006) to
quantify the pirofeotina a. A sample of 300 mL was diluted in 1 mL
of hexane. Separation of the pigments (pheophytins, pyropheophytin
a and chlorophylls) was carried out by using a miniaturized column
chromatography on a silica gel column. After loading the sample,
the eluate of chlorophyll pigments were collected by passing 5 mL
of acetone and it was analyzed by HPLC (Agilent technologies 1100,
Santa Clara, California) equipped with a C18-RF Spherisorb ODS-2
(25 cm 4 mm id; 3 m particle size) (Waters, Saint QuentinYvelines, France) and a diode array detector. It worked in isocratic
regime, using H2O/MeOH/acetone as mobile phase (4:36:60 v/v/v)
with a ow of 1 mL/min. The chromatographic signals were obtained
at 410 nm. The compounds identied and quantied were: pheophorbide, chlorophyll, pheophytins a, a, b, and b, and pyropheophytin a.

N. Tena et al. / Food Research International 45 (2012) 103108

40
100

total phenols
alfa-tocopherol
PC
total chlorophyll pigments

90

30

70
60
20

50
40
30

Polar compounds (%)

Residual content (%)

80

10
20
10
0
0

10

15

20

25

30

35

0
40

Thermoxidation time (h)

Fig. 1. Evolution of residual percentage of total phenols, -tocopherol and total chlorophyll
pigments and the percentage of total polar compounds (% PC) with the thermoxidation
time. The dotted line indicates the legal limit to discard the frying oils (25% PC).

The quantication was carried out by using an external calibration line

with a solution of chlorophyll a/acetone (0.20.5 mg/kg).

2.7. Statistical analysis

Prior to examining the whole set of uorescence data, the EEMs
of the 47 samples were arranged in a three-dimensional structure
of size 47 400 13 (samples em ex). The array was decomposed by PARAFAC algorithm. The explained variance (93.991%)

105

calculated by residual analysis indicated that three was the optimal

number of factors. Nonnegative constraints were applied as they are
advisable when working with spectral data set (Bro, 1997). The
processing of spectroscopic data with PARAFAC algorithm has been
extensively described elsewhere (Andersen & Bro, 2003; Bro,
1997). The multiway model PARAFAC allows an easy interpretation
of three dimensional matrix because the model decomposes threeway uorescence data into spectral excitation and emission proles
of uorophores. Thus, the total matrix was decomposed into three
matrices (excitation, emission and samples), each one having
three factors. The relationship between factors and the chemical
variables (e.g. vitamin E, phenols) was studied by multiple linear
regression analysis (MLRA). In order to avoid hyperoptimistic results
MLRA was carried out under the strictest conditions (Fisher-distribution
table at p = 0.05). The arrangement of data and PARAFAC analysis was
carried out with Matlab v7.8 (The MathWorks, Natick, MA), while
Statistica 7.0 (StatSoft, Tulsa, OK) was used for carrying out the rest of
statistical analyses.

3. Results and discussion

The thermoxidation process of vegetable oils during frying
induces a series of changes in its chemical composition that affects
the quality and the safety of the frying oil and food (Matthus,
2006). The most established method to regulate the use of frying
oils is the determination of the percentage of the total polar compounds (% PC), 25% being the maximum limit to discard the frying
oils in most of the European countries (Stier, 2001). Thus, the % PC
was determined in the samples to know the alteration degree and
their suitability for consumption. Fig. 1 shows the evolution of % PC
of the samples heated up to 40 h. The oil reached 25% PC after 30 h

Table 1
Evolution of the concentration (mg/kg) of individual uorophores present in the olive oil samples during 40 h of thermoxidation.
Time
(h)

% PCa

0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
29
30
31
32
33
34
35
36
38

3.45
3.94
4.73
5.10
5.67
7.11
8.66
9.07
9.40
12.60
13.02
13.94
14.47
18.56
20.05
20.22
20.76
26.91
27.52
27.60
29.03
30.95
31.03
36.40

a
b
c
d
e
f
g
h
i

Chlorophyll pigments

-Tocopherol

Phenols

Pheophb

Pheoc

Pyrophd

HyTe

Ac-HyTf

Tyg

Pinoresinol

p-HPEA-EAh

2.66
1.06
0.36
0.15

2.07
0.23
0.14
0.12

1.62
4.11
2.32
2.12

0.06

0.06

0.57

0.04

tr

0.40

0.03

tri

0.13

tri

tri

tri

tri

tri

tri

17.71
10.60
8.30
7.03
4.83
2.25
1.22
0.84
0.22
0.23
0.11
0.12
0.08
0.08
0.07
0.09
0.07
0.08
0.06
0.06
0.07
0.07
tri
tri

7.17
6.31
4.12
4.53
3.09
1.97
1.14
0.94
0.93
0.35
0.18
0.12
0.11
0.10
0.08
0.09
0.07
0.06
0.05
0.05
0.04
0.03
0.02
tri

16.82
13.20
12.62
11.52
10.50
9.78
9.55
8.52
7.80
7.29
6.11
5.09
4.97
4.31
4.08
3.99
4.48
4.05
4.10
4.00
3.99
3.31
2.75
2.23

3.85
3.42
3.36
3.26
2.80
2.50
2.30
2.18
2.10
2.05
1.70
1.42
1.23
0.94
0.78
0.56
0.52
0.50
0.49
0.48
0.47
0.46
0.42
0.40

tri
tri
tri
tri
tri
3.67
3.76
4.49
4.90
5.20
5.42
4.82
4.63
4.32
4.24
4.08
3.81
3.49
3.47
3.43
3.37
3.33
2.90
2.50

Percentage of total polar compounds.

Pheophorbide.
Pheophytin b.
Pyropheophytin.
Hydroxytyrosol.
Hydroxytyrosol acetate.
Tyrosol.
Ligstroside aglycone, oxidized aldehyde and hydroxylic form.
Trace levels.

228.38
220.49
177.85
162.15
152.38
127.89
119.56
106.60
80.04
78.01
59.06
56.27
4.56

tri

tri

106

N. Tena et al. / Food Research International 45 (2012) 103108

of thermoxidation (dotted line), which points out the maximum time

that this oil can be used as frying oil for that temperature and frying
conditions.
Fig. 1 also shows the evolution of the residual contents (%) of total
phenols and chlorophyll pigments, both of them having uorescent
properties. The chlorophyll pigments were the uorophores whose
concentrations decrease faster during the thermoxidation process.
The concentration decay of these compounds was slower after reaching 7.1% PC and they reduced their concentration at trace levels when
the oil had 25% PC. The evolution of the concentration of the individual chlorophyll pigments (Table 1) shows that all of them decreased
during the rst hours of thermoxidation, except for pyropheophytin,
which increased its concentration fourfold in the rst two hours of
thermoxidation and it degraded up to trace levels after 30 h.
The concentration of total phenols and -tocopherol (Fig. 1) shrunk
at the same rate and they reached a plateau when the oil had 25% PC. The
individual study of the phenols shows that the simple o-diphenols,
hydroxytyrosol and hydroxytyrosol acetate (Table 1) were the phenols
with more variation in their concentrations during the thermoxidation
process. All these compounds reached trace levels after 20 h of thermoxidation, the oil containing 20% PC (Table 1). On the contrary tyrosol and
pinoresinol, together with some secoiridoid oxidized derivatives as pHPEA-EA (Table 1), remained in the oil after 38 thermoxidation hours.
These results partially match with those presented by Brenes, Garca,
Dobarganes, Velasco, and Romero (2002).
Once the chemical analyses revealed the chemical changes on the
main uorophores in the oil, the evolution of these compounds was
studied by excitationemission uorescence spectroscopy (EEFS).
The 47 excitationemission matrices were analyzed by PARAFAC

16

Factor 1

12

Oxidation
compounds

0
300

350

400

450

500

550

600

650

700

20

Chlorophyll
pigments

16

Factor 2

12
8

algorithm. Fig. 2 shows the emission proles obtained from the

three-factor PARAFAC model. The three plots one for each factor
showed a band with a maximum at 675 nm (em), which is associated
with the presence of chlorophyll pigments in the samples (Dupuy
et al., 2005). The rest of the emission prole depended on the individual factor. Thus, the emission prole of factor 1 showed a band with a
maximum at 525 nm, assigned to oxidation compounds (Dupuy et al.,
2005). This band (em = 450650 nm) is absent in the emission prole of factor 2, and it appears at less intensity in the emission prole of
factor 3. The emission prole of factor 2 was characterized with the absence of bands except for that assigned to the chlorophyll pigments in
the range em = 650700 nm. Finally, the emission prole of factor 3
showed a characteristic band with a maximum at 350 nm (em),
which was absent in the proles of the other two factors and it was associated with the presence of tocopherols and phenols (Sikorska et al.,
2005; Zandomeneghi et al., 2005). In general, taking into account the
range 300650 nm (omitting the em range of pigments), the emission
prole of factor 3 was similar to the spectra of the samples at the
beginning of the thermoxidation process when they were acquired at
a single excitation wavelength (270 nm) (Tena et al., 2009). On the
contrary, the emission prole of factor 1 was related to the spectra of
samples at the end of the process, which include an intensive band in
the range 450650 mm (em) assigned to oxidation products (Tena
et al., 2009). These results showed that the data modelling carried out
by PARAFAC allows obtaining uorescence proles that can be used as
ngerprints of specic compounds.
Fig. 3 shows the score plot for the excitation wavelengths
(ex = 250298 nm) by representing the factors 2 and 3. The plot of
these two factors pointed out a clear division of the excitation wavelengths into two groups, establishing the boundary between both at
280 nm. The excitation wavelengths above this value (280300 nm)
include the maximum for phenols and tocopherols (Sikorska et al.,
2004; Zandomeneghi et al., 2005), which partly justies the distinction between these two groups. By contrast, an excitation wavelength
of 260270 nm (value 0.3 on the factor 2) generates information on
oxidation compounds (COI/T.15/NC no. 3/Rev. 3, 2009; Kyriakidis &
Skarkalis, 2000). This results show that the optimum excitation
wavelength to study the thermoxidized olive oils by conventional
uorescence spectroscopy would be next to 280 nm since it would
allow acquiring information from both phenols and oxidation products.
Regarding the sample scores, Fig. 4 shows the projection plot of
the PARAFAC model using the factors 1 and 2. This information was
used to carry out a variance analysis of thermoxidized oils. Fig. 4
shows that the initial sample of olive oil, nonheated (3.5%PC), clearly
differs from the other samples. According to the percentage of total
polar compounds (Fig. 1), the rest of the samples can be divided

0.9

0.8

298
-4
300

350

400

450

500

550

600

650

0.7

700

0.6
8

Factor 3

Tocopherols and phenols

Factor 3

0.5
0.4
0.3

0.2
2
0.1
0
300

294

350

400

450

500

550

600

650

700

Wavelenght (nm)
Fig. 2. Spectral proles obtained from the three-factor calculated for the emission
wavelengths from the PARAFAC model.

0.0

290
286

278
266
274 270
262 258 250

282
0.1

0.2

0.3

0.4

254
0.5

Factor 2
Fig. 3. Projection plot of factors 2 and 3 calculated for the excitation wavelengths from
the PARAFAC model. The numbers on the gure correspond to the ex in nm.

N. Tena et al. / Food Research International 45 (2012) 103108

0.5

107

4. Conclusions

3.5
3.9

0.4

Excitationemission uorescence spectroscopy combined with

the PARAFAC algorithm allows getting individualized information
about series of uorescence compounds. The information obtained
by this procedure may serve for a better interpretation of the spectra
collected with conventional uorescence spectroscopy (Tena et al.,
2009). On the other hand, the emission proles obtained from the
three-factor PARAFAC model minimize the intensity of chlorophyll
pigment, which means an advantage in the study of other series of
uorophores that show lower absorption intensity. As a result, the
spectra obtained for each factor permit an individualized study of
oxidation products, tocopherols and phenols.

4.7

< 8% PC

0.3

5.1
5.7

Factor2

7.1

0.2
8.7

0.1

0.0

8-36% PC

67.9
67.766.0

> 36% PC
65.7
65.3

64.8 64.0
63.8
63.0 62.4

61.9
61.3
55.2
60.5
54.5

-0.1

-0.2

0.06

0.08

0.10

0.12

9.1
9.4

12.6
13.0
13.9
14.5
18.6
20.1
20.2
20.8
26.9
27.6
27.5
30.929.0
31.0
53.9
52.2
46.2
47.8
50.6
36.4
36.0
42.9
37.1
45.636.8
37.9

0.14

0.16

0.18

0.20

Factor1
Fig. 4. Sample projection plot of the rst two factors from the PARAFAC model. The
numbers on the gure correspond to the percentage of total polar compounds (% PC).

into three different groups as it is shown in Fig. 4: oils with less than
8% PC, this moment of the experiment matching with the decrease of the
degradation rate of the chlorophyll pigments (Fig. 1); oils in the range
from 8% to 36% of total polar compounds; and oils with more than 36%
of total polar compounds. The latter percentage (36% PC) was achieved
at approximately 38 h of thermoxidation when the minimum concentration of phenols and trace levels of pigments were quantied
(Fig. 1). The low concentrations of these compounds at that moment
and thereafter provide a chemical justication for the separation of samples with more and less than 36% PC.
The three factors that explain the variance of the samples were used
as variables in a regression model relating these factors to the percentage of total polar compounds and the concentration of phenols, tocopherols and chlorophyll pigments. The objective was to evaluate the
relationship between the spectral information processed by PARAFAC
and the chemical data from chromatographic analyses. Table 2 summarizes the results obtained with multiple linear regression analysis
(MLRA). In all cases the adjusted quadratic regression coefcients
(R2adj) are higher than 0.90. The standardized coefcients (beta)
(Table 2) allow comparing the relative contribution of each factor to
the prediction of the concentration of those compounds. Considering
these factors, factor 1 is the largest contributor to predict the percentage
of total polar compounds as its standardized coefcient corresponded to
the higher absolute value (Table 2). Factor 2, in particular, has a higher
relative contribution in the regression model to predict the concentration of phenols and tocopherols, while factor 3 further explains the
concentration of chlorophyll pigments. These results demonstrate the
power of PARAFAC to analyze complex mixtures of compounds, such
as virgin olive oil, and allow studying individual chemical series by
means of the decomposition of spectral data. Furthermore, these
regression results explain the distribution of the samples in Fig. 4,
since the percentage of total polar compounds and the concentration
of phenols are explained by factors 1 and 2 respectively.

Table 2
R2-adjusted (R2-adj) and standardized coefcients of multiple linear regression analysis (MLRA) between the three PARAFAC factors calculated for the samples and several
series of uorophores.
Parameters

R2adj

Standardized coefcient (beta)

Factor 1

Factor 2

Factor 3

Polar compounds
Chlorophyll pigments
Total phenols
-Tocopherol

0.90
0.92
0.96
0.95

0.72
3.66
0.02
0.20

0.32
1.80
0.97
1.18

0.19
4.51
0.01
0.32

References
Andersen, C. M., & Bro, R. (2003). Practical aspects of PARAFAC modelling of excitation
emission data. Journal of Chemometrics, 17, 200215.
Belton, P. S., Kemsley, E. K., McCannc, M. C., Ttos, S., Wilson, R. H., & Delgadillo, I.
(1995). The identication of vegetable matter using Fourier Transform Infrared
Spectroscopy. Food Chemistry, 54(4), 437441.
Brenes, M., Garca, A. M., Dobarganes, M. C., Velasco, J., & Romero, C. (2002). Inuence
of thermal treatment simulating cooking processes on the polyphenol content in
virgin olive oil. Journal of Agricultural and Food Chemistry, 50, 59625967.
Bro, R. (1997). PARAFAC. Tutorial and applications. Chemometrics and Intelligent
Laboratory Systems, 38, 149171.
Cheikhousman, R., Zude, M., Bouveresse, D. J. R., Leger, C. L., Rutledge, D. N., & BirlouezAragon, I. (2005). Fluorescence spectroscopy for monitoring deterioration of extra
virgin olive oil during heating. Analytical and Bioanalytical Chemistry, 382,
14381443.
COI/T.15/NC no. 3/Rev. 3 (2009). International Olive Council (IOC). Trade Standard
applying to Olive Oil and Olive-Pomace Oil. Madrid, Spain.
Dupuy, N., Le Drau, Y., Ollivier, D., Artaud, J., Pinatel, C., & Kister, J. (2005). Origin of
french virgin olive oil registered designation of origins predicted by chemometric
analysis of synchronous excitationemission uorescence spectra. Journal of
Agricultural and Food Chemistry, 53, 93619368.
Galano, T., Durn, I., Correa, C. A., Roldn, B., & Rodriguez, M. I. (2003). Simultaneous
uorometric determination of chlorophylls a and b and pheophytins a and b in
olive oil by partial least squares calibration. Journal of Agricultural and Food
Chemistry, 51, 69346940.
Gertz, C., & Fiebig, H. J. (2006). Pyropheophytin aDetermination of thermal degradation products of chlorophyll a in virgin olive oil. European Journal of Lipid Science
and Technology, 108, 10621065.
Harshman, R. A. (1970). Foundations of the PARAFAC procedure: Models and
conditions for an explanatory multimodal factor analysis. WPP, 16, 184.
IUPAC 2.432 (1987). International Union of Pure and Applied Chemistry (IUPAC).
Determination of tocopherol and tocotrienols in vegetable oils and fats by HPLC.
Standard Methods for the Analysis of Oils, Fats and Derivatives (7th ed.). Oxford,
U.K: Blackwell.
IUPAC 2.507 (1987). International Union of Pure and Applied Chemistry (IUPAC).
Determination of polar compounds in frying fats. Standard Methods for the Analysis
of Oils, Fats and Derivatives (7th ed.). Oxford, U.K: Blackwell.
Kyriakidis, N. B., & Skarkalis, P. (2000). Fluorescence spectra measurement of olive oil
and other vegetable oils. Journal of AOAC International, 83, 14351438.
Mateos, R., Espartero, J. L., Trujillo, M., Ros, J. J., Len-Camacho, M., Alcudia, F., et al.
(2001). Determination of phenols, avones, and lignans in virgin olive oils by
solid-phase extraction and high-performance liquid chromatography with diode
array ultraviolet detection. Journal of Agricultural and Food Chemistry, 49,
21852192.
Matthus, B. (2006). Utilization of high-oleic rapeseed oil for deep-fat frying of French
fries compared to other commonly used edible oils. European Journal of Lipid
Science and Technology, 108, 200211.
Nicoletti, G. M. (1990). La uorescenza degli oli di oliva. Rivista Italiana delle Sostanze
Grasse, 67, 389396.
Poulli, K. I., Chantzos, N. V., Mousdis, G. A., & Georgiou, C. A. (2009). Synchronous
uorescence spectroscopy: Tool for monitoring thermally stressed edible oils.
Journal of Agricultural and Food Chemistry, 57, 81948201.
Poulli, K. I., Mousdis, G. A., & Georgiou, C. A. (2009). Monitoring olive oil oxidation
under thermal and UV stress through synchronous uorescence spectroscopy
and classical assays. Food Chemistry, 117, 499503.
Sayago, A., Morales, M. T., & Aparicio, R. (2004). Detection of hazelnut oil in virgin olive
oil by a spectrouorimetric method. European Food Research and Technology, 218,
480483.
Sikorska, E., Grecki, T., Khmelinskii, I. V., Sokorski, M., & Koziol, J. (2005). Classication
of edible oils using synchronous scanning uorescence spectroscopy. Food Chemistry, 89, 217225.
Sikorska, E., Khmelinskii, I. V., Sikorski, M., Caponio, F., Bilancia, M. T., Pasqualone,
A., et al. (2008). Fluorescence spectroscopy in monitoring of extra virgin olive
oil during storage. International Journal of Food Science and Technology, 43,
5261.

108

N. Tena et al. / Food Research International 45 (2012) 103108

Sikorska, E., Romaniuk, A., Khmelinskii, I. V., Herance, R., Bourdelande, J. L., Sokorski, M.,
et al. (2004). Characterization of edible oils using total luminescence spectroscopy.
Journal of Fluorescence, 14, 2535.
Stier, R. F. (2001). The measurement of frying oil quality and authenticity. In J. B. Rosell
(Ed.), Frying: Improving quality (pp. 165192). Boca Raton, FL: CRC Press.

Tena, N., Garca-Gonzlez, D. L., & Aparicio, R. (2009). Evaluation of virgin olive oil
thermal deterioration by uorescence spectroscopy. Journal of Agricultural and
Food Chemistry, 57, 1050510511.
Zandomeneghi, M., Carbonaro, L., & Caffarata, C. (2005). Fluorescence of vegetable oils:
Olive oils. Journal of Agricultural and Food Chemistry, 53, 759766.