3D Front-Face Fluorescence Spectroscopy For Characterization of Olive Oil
3D Front-Face Fluorescence Spectroscopy For Characterization of Olive Oil
3D Front-Face Fluorescence Spectroscopy For Characterization of Olive Oil
Abstract: 3D front-face fluorescence spectroscopy and principal component analysis (PCA) were used to differentiate
between Extra Virgin Olive (EVOO) and Olive Oil (OO). The results showed that 3D front-face fluorescence could be an
effective tool to characterize EVOO and OO. Fourteen samples of olive oils were acquired directly from producers and
from retail markets and their 3D fluorescence spectra were measured and analyzed. The first principal component of the
PCA model retained the most important variability as a combination of freshness due to the presence of antioxidant
compounds and chlorophyll; and of oxidation stage due to the presence of oxidation products. Olive oil samples presented
different spectral patterns providing different scores for EVOO and OO in the PCA model. EVOO samples selected according
to the differences in their scores in the PCA were exposed to indirect light. The evolution of their emission spectra was
monitored with a right-angle set-up showing differences in concordance with their scores. This technique, the 3D front-face,
is useful for characterization of olive oil samples according to their fluorescent compounds. The combination with other
techniques (1H HR-NMR, HPLC, NIR spectrometry, etc.) may also be useful in protocols devoted to decision making
processes with regard to olive oil quality specifications.
Citation: Lleó, L., N. Hernández-Sánchez, F. Ammari, and J. M. Roger. 2016. 3D front-face fluorescence spectroscopy for
characterization of extra virgin olive oil. Agricultural Engineering International: CIGR Journal, 18(4):190-199.
poly- unsaturated fatty acids strongly affects the oxidative The front-face approach provided accurate
process (Poulli et al., 2009). measurements for edible oils in non-diluted samples
The guarantee of authentication of EVOO is one of (Ammari, 2012a; Ammari et al., 2012b, 2012c).
the most challenging issues of the olive oil sector. Among Despite the interpretation of fluorescence spectral
1
others, high resolution nuclear magnetic resonance H data is complex due to the presence of many fluorophores,
HR-NMR (Hernández-Sánchez et al., 2014), different multivariate analysis methods (Independent
high-performance liquid chromatography HPLC, Fourier Component Analysis –ICA-, Principal Component
Transform Infrared FT-IR, visible VIS, near infrared NIR, Analysis –PCA–, Partial Least Square –PLS– regression,
mid infrared MIR spectroscopy methods have been applied PLS Discriminant Analysis) could successfully facilitate
to adulteration detection, geographical origin determination the interpretation and of olive oils (Kassouf et al, 2014;
and oxidation status characterization (Frankel, 2010; Ammari et al., 2012b, 2012c; Valderrama et al., 2011;
Aparicio et al., 2013). Currently, there is a need to improve SádeckáJ., TóthováJ., 2007).
and harmonize the existing normative and regulations The objective of the present work is to study the
concerning olive oil. Several international institutions (e.g. characteristic spectral pattern of EVOO and OO on the basis
International Olive Council, IOC) are actively involved in of the differences between signal patterns and intensities
the developing of normative for olive oil products, labeling arising from different fluorescent compounds. 3D front-face
regulations, and rapid, easy and accurate instrumental fluorescence spectroscopy was used to identify the spectral
techniques and analytical methodologies. regions of interest for EVOO and OO samples. PCA was
Different types of fluorescence spectroscopy applied to facilitate the interpretation of the fluorescence
methods have been used for analysis of EVOO quality data and the characterization of the samples.
because they are less time-consuming and more Finally, two EVOO samples selected according to
cost-efficient compared to other analytical procedures the PCA model were exposed to indirect light during five
(Karoui and Blecker, 2011). months. Their emission spectra were monitored with a
In addition, fluorescence spectroscopy is a right-angle set-up in order to evaluate the relationship
noninvasive, highly selective and sensitive technique. between the PCA model characterization and the further
Other outstanding advantages are the absence of solvents evolution of the spectra.
and reagents, and the requirement of small amounts of
2 Materials and methods
sample. Classic fluorescence spectroscopy, front-face
fluorescence spectroscopy, synchronous scanning 2.1 Materials
fluorescence spectroscopy have provided chemists with a Fourteen samples of olive oils were acquired directly
sensitive approach to determine the oil quality for from producers and from retail markets. These samples
soybean, corn, sunflower, and olive oils (Ammari et al., were grouped according to the indications of producers
2012a; Kyriakidis and Skarkalis, 2000; Sikorska et al., and of commercial labelling: seven EVOO samples (E1 to
2004; Sikorska et al., 2005; Sikorska et al., 2008). E7) and seven OO samples (O1 to O7).
Application of these methods to EVOO has a high Analytical analyses were carried out at an external
potential because most of the interesting components, laboratory according to European Regulations, EEC No
and α-tocopherol, and primary and secondary oxidation Table 1 summarizes the relevant characteristics of
products are fluorescent molecules. these samples. Note that K232 values are not available for
samples O1 to O3.
192 December, 2016 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 18, No.4
The two groups of oils, EVOO and OO, matrices of the different samples of olive oil were pooled
accomplished the limits for K232 and K270 according to in a [14 x 105 x 898] 3-way cubic array (14 spectra from
EEC No 2568/91 (K232 ≤ 2.5 and K270 ≤ 0.22 for EVOO; 14 oil samples, 105 excitation wavelengths and 898
K270 ≤ 0.9 for non-EVOO). However, the K232 value of emission wavelengths). Data were then unfolded to give a
E1 sample is higher than the limit; note that this sample [14 x 94290] matrix in order to apply the chemometric
has been obtained from a retail market, where it could analysis.
have undergone uncontrolled oxidation. The unfolded 3D spectra of the seven EVOO
2.2 3D front-face fluorescence spectroscopy and samples were mean centered (with respect to the mean of
chemometric analysis these seven spectra) and then submitted to PCA,
3D front-face fluorescence spectra were measured obtaining the corresponding loadings and scores. The
directly on the oil samples without prior preparation using loadings were afterwards refolded for further
a spectrofluorometer (LS45, Perkin-Elmer) equipped with interpretations.
a xenon lamp source, excitation and emission In addition, the unfolded 3D spectra of the seven OO
monochromators and a front-face sample-cell holder. samples were also mean centered (considering the mean
Measurements were carried out using quartz of the EVOO spectra) and then, the loadings previously
cuvettes (10 mm × 10 mm × 45 mm). The excitation computed were used to obtain their corresponding scores.
wavelengths ranged from 230 to 646 nm (step 4 nm) and 2.3 Right-angle fluorescence spectroscopy
emission wavelengths ranged from 250 to 698.5 nm (step A right-angle set-up prototype devoted to fast
0.5 nm). Excitation and emission monochromator slit inspection was designed and assembled by
widths were set at 10 nm. Emission monochromator scan LPF-TAGRALIA.
-1
speed was 800nms . A photomultiplier of voltage 650 V The fluorescence spectra were obtained using a
was used. photonic multi-channel spectrometer (Hamamatsu, Japan)
The data corresponding to each sample were with detection wavelengths ranging from 196.9 nm to
arranged in a [105 x 898] matrix. All the elementary 958.8 nm (step 0.75 nm, 1024 wavelengths). A UV-VIS
December, 2016 3D front-face fluorescence spectroscopy for characterization of olive oil Vol. 18, No. 4 193
light source (L10290, Hamamatsu, Japan) was used with throughout five months with this set-up.
a deuterium lamp, with spectral range from 200 nm to The emission spectra were normalized by computing
400 nm. Two optical filters were coupled to the the ratio of the signal intensity at each wavelength to the
right-angle set-up so as to constrain both the excitation sum of the signal along the whole spectrum.
and the emission wavelength ranges. An optical filter
3 Results and discussion
limited the incident light to wavelengths lower than 400
nm. The other optical filter limited detected light to 3.1 3D front-face fluorescence spectra
wavelengths higher than 400 nm. Measurements were The 3D fluorescence spectra exhibited several
carried out using quartz cuvettes (10 mm × 10 mm × 45 well-defined regions of intense fluorescence. Figure 1
The two EVOO samples showing the highest and the depicted with comparable color scale. The left column
lowest PCA scores in the front-face model were exposed corresponds to EVOO group and the right one to OO
to indirect light in transparent glass bottles. The evolution group. In a first approach, both types of olive oil samples
of the emission spectra was monthly monitored present different 3D fluorescence patterns.
Figure 1 3D front-face fluorescence spectra, left column corresponds to the seven EVOO samples (E1 to E7) and
right column to the seven OO (O1 to O7). Y axes: wavelengths of emission from 250 to 698.5 nm. X axes:
wavelengths of excitation from 230 to 646 nm.
Two examples corresponding to an EVOO (E6) and comprising the characteristic spectral patterns, as
to an OO (O5) samples have been considered for the illustrated in Figure 2a and 2b respectively.
detailed description of the most important regions
194 December, 2016 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 18, No.4
(a) (b)
Figure 2 (a) Example of a 3D front-face fluorescence spectrum of an EVOO (E6); (b) Example of a 3D
front-face fluorescence spectrum of OO (O6). The regions of high intensity emission are labeled from R1 to R5. Note
the different color scale.
The spectral emission is the result of different for the main fluorophores of olive oils that are related to
signals emitted by the corresponding different freshness and oxidation conditions. Table 2 summarizes
fluorophores present in the sample. Previous researches these emission and excitation bands as well as the
provide ranges of excitation and emission wavelengths corresponding references.
Table 2 Excitation-emission bands corresponding to the main compounds related to freshness and
oxidation conditions in olive oil.
COMPOUNDS Excitation- emission bands
Polyphenols and tocopherols (Zandomeneghi et al., 2005) excitation 270-310nm; emission 300- 390 nm
excitation 320–420nm;
Oxidation products (Ammari, 2012)
emission 400–500nm
In the EVOO samples, five regions of interest are Table 2. Second area of emission (region 2, R2) ranges
revealed. The area of the highest intensity emission signal approximately from 615 to 680 nm corresponding to an
(region 1, R1) ranges approximately from 300 to 370 nm, excitation band from 270 to 310 nm. These excitation and
corresponding to an excitation band from 270 to 310 nm emission bands are not referenced in bibliography as
(Figure 2a). This emission area is ascribed to tocopherols produced by a specific compound of olive oil. However,
and polyphenols according to literature summarized in
December, 2016 3D front-face fluorescence spectroscopy for characterization of olive oil Vol. 18, No. 4 195
products. 12000
10000
Three rather less intense regions appear in
8000
PC1 score
fourth region appears with positive loading values, whose The PC1 scores are computed for both EVOO and
location is concordant with the fluorescence region R5 OO groups (Figure 3b). In the case of the EVOO samples
arising from oxidation products. (from E1 to E7), the scores present the lowest values
The different signs of loadings related to derived from their highest fluorescence of antioxidant
antioxidants (negative) and oxidation products (positive) compounds. The EVOO sample E3 presents the lowest
may enhance the identification of the highest quality score, whereas the sample E5 shows the highest value
EVOO through the computed score. Such EVOO would (Figure 3b), which reflects the differences in their 3D
be characterized by high antioxidant content with spectral patterns (Figure 4 a y b).
beneficial effect on human health and on potential olive
oil extended shelf-life.
(a) (b)
(c) (d)
Figure 4 (a) 3D front-face fluorescence spectra of the EVOO with the lowest PC1 score (E3) among EVOO
group; (b) 3D front-face fluorescence spectra of the EVOO with the highest PC1 score (E5) among EVOO group;
(c) Evolution of the emission spectra along five months (M0 to M5) of the E3 sample (excitation from 200 to 400
nm); (d) Evolution of the emission spectra along five months (M0 to M5) of the E5 sample (excitation from 200 to
400 nm).
The scores of OO group are higher than those of the (Figure 3b), whereas O5 shows the farthest score, which
EVOO group, in agreement with their higher fluorescence also reflects the differences in their spectral patterns
of the oxidation compounds (Figure 1). Sample O2 (Figure 5).
presents a score value close to those of the EVOO group
December, 2016 3D front-face fluorescence spectroscopy for characterization of olive oil Vol. 18, No. 4 197
(a) (b)
Figure 5 (a) 3D front-face fluorescence spectra of the OO with the lowest PC1 score (O2) among OO group; (b) 3D
front-face fluorescence spectra of the OO with the highest PC1 score (O5) among OO group.
For the EVOO group, the value and the narrow features. As illustrated in Figure 3, generally those oils
dispersion of the scores highlight a predominant content with high polyphenol content present low value of PC1
of antioxidant compounds, and a slight rate of scores, such as E3, E6 and E7 samples.
degradation. In contrast, the scores within the OO group Within OO group, oil sample O2 presented the
show a much higher value and variability, reflecting highest polyphenol content, whereas those oils with much
different oxidation status. Such behavior could be lower polyphenol content, presented higher score value,
expected as OO samples are blends of refined and virgin such as O1, O3, O5 and O6 samples. The K270 value of
olive oils (either extra or non-extra). The refined olive O2 sample is the lowest of the OO group (close to the
oils had undergone a more intense oxidation (Velasco and upper limit of the normative, K270 ≤ 0.22) and the total
Dobarganes, 2002). content of polyphenols is the maximum value within its
According to Table 1, the two groups of samples group. In contrast, olive oil O6 appears as the most
EVOO and OO present differences in their analytical degraded sample and shows low polyphenol content and
measurements. The observed K270 values are lower for high K270value (Table 1).
EVOO samples (ranging from 0.12 to 0.14) than for OO 3.3 Eevolution of EVOO samples
samples (ranging 0.28 to 0.50), reflecting a higher status The EVOO samples showing the extreme PC1
of oxidation in the latter group, as expected. In addition scores were exposed to indirect light, and the evolution of
polyphenols content is usually higher for EVOO group the emission spectra was monitored for five months with
than for OO group, although some non-EVOO samples a right-angle set-up (Figure 4).
present high values. As for K232, the ranges of the values This set-up assures the detection of the chlorophyll
are overlapped. The same fact is observed in α-tocopherol fluorescence signal, which is a major indicator of the oil
content. However, for α-tocopherol, it could be not evolution under different storage conditions since it acts
necessarily related to the oxidation stage since its addition as antioxidant under dark and as pro-oxidant under light
to non-extra virgin olive is allowed by the European conditions. In Figure 4c and 4d the peak at 670 nm is
regulation. ascribed to chlorophyll fluorescence. The evolution of the
These analytical measurements summarized in Table products of oxidation, from both primary and secondary
1 are in agreement with the observed fluorescence processes, can be monitored as well. In Figure 4c and 4d
198 December, 2016 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 18, No.4
and Food Safety, 5(4):169-186. olive oil. Analytica Chimica Acta, 839(10):14–25.
http://onlinelibrary.wiley.com/doi/10.1111/j.1541-4337.200 doi:10.1016/j.aca.2014.06.035
6.00009.x/full 5 Kyriakidis, N. B., and P. Skarkalis. 2000. Fluorescence spectra
Frankel, E. N. 2010. Chemistry of Extra Virgin Olive Oil: measurement of olive oil and other vegetable oils. Journal
Adulteration, Oxidative Stability, and Antioxidants. of AOAC International, 83(6):1435-1439.
Journal of Agricultural and Food Chemistry, 58(10): Poulli, K. I., G. A. Mousdis, and C. A. Georgiou. 2009. Monitoring
5991-6006. http://pubs.acs.org/doi/pdf/10.1021/jf1007677 olive oil oxidation under thermal and UV stress through
García-González, D. L., R. Aparicio-Ruiz, and R. Aparicio. 2008. synchronous fluorescence spectroscopy and classical assays.
Virgin olive oil - Chemical implications on quality and Food Chemistry, 117(3):499-503.
health. Eurpean Journal of Lipid Science & Technology, http://www.sciencedirect.com/science/article/pii/S0308814
110(7): 602–607. 609004841
http://onlinelibrary.wiley.com/doi/10.1002/ejlt.200700262/ Sádecká, J., and J. Tóthová. 2007. Fluorescence spectroscopy and
pdf chemometrics in the food classification − a review. Czech
Guimet, F., J. Ferré, R. Boqué, and F. X. Rius. 2004. Application Journal of Food Sciences, 25(4):159–173.
of unfold principal component analysis and parallel factor Sikorska, E., A. Romaniuk, I. V. Khmelinskii, R. Herance, J. L.
analysis to the exploratory analysis of olive oils by means Bourdelande, M. Sikorski, and J. Koziol. 2004.
of excitation-emission matrix fluorescence spectroscopy. Characterization of Edible Oils Using Total Luminescence
Analytica Chimica Acta, 515(1):75-85. Spectroscopy. Journal of Fluoresc., 14(1):25-35.
doi:10.1016/j.aca.2004.01.008 http://link.springer.com/article/10.1023%2FB%3AJOFL.00
Guoping, W., D. Quingzhu, and H. Zhenyu. 2008. Independent 00014656.75245.62
Component Analysis and its applications in signal Sikorska, E., T. Górecki, I. V. Khmelinskii, M. Sikorski, and J.
processing for analytical chemistry. Trends Anal. Chem., Koziol. 2005. Classification of edible oils using
27(4):368–376. doi:10.1016/j.trac.2008.01.009 synchronous scanning fluorescence spectroscopy. Food
Guzmán, E., V. Baeten, J. A. F. Pierna, and J. A. García-Mesa. Chemistry, 89(2):217-225.
2015. Evaluation of the overall quality of olive oil using doi:10.1016/j.foodchem.2004.02.028
fluorescence spectroscopy. Food Chemistry, 173:927-34. Sikorska, E., I. V. Khmelinskii, M. Sikorski, F. Caponio, M. T.
doi:10.1016/j.foodchem.2014.10.041 Bilancia, A. Pasqualone, and T. Gomes. 2008.
Hernández-Sánchez, N., L. Lleó, T. R. Cuadrado, M. D. Molero, Fluorescence spectroscopy in monitoring of extra virgin
and M. Ruiz-Altisent. 2014. HR Nuclear Magnetic olive oil during storage. International Journal of Food
Resonance Spectroscopy for authentication of olive oil Science and Technology, 43(43):52-61.
quality. In Proceedings of International Conference of Valderrama, P., P. H. Março, N. Locquet, F. Ammari, and D. N.
Agricultural Engineering, pp 1-6 Zurich, Switzerland. Rutledge. 2011. A procedure to facilitate the choice of the
http://oa.upm.es/35330/ number of factors in multi-way data analysis applied to the
International Olive Council (IOC). 2013. Trade standards applying natural samples: Application to monitoring the thermal
to olive oils and olive-pomace oils.COI/T.15/NC No 3/Rev. degradation of oils using front-face fluorescence
7. spectroscopy. Chemometrics and Intelligent Laboratory
Karoui, R., and C. Blecker. 2011. Fluorescence spectroscopy Systems, 106(2):166–172.
measurement for quality assessment of food systems - a doi:10.1016/j.chemolab.2010.05.011
review. Food and Bioprocess Technology, 4(3):364–386 Velasco, J., and C. Dobarganes. 2002. Oxidative stability of virgin
http://link.springer.com/article/10.1007%2Fs11947-010-03 olive oil. European Journal of Lipid Science and
70-0#page-1 Technology, 104(9-10):661-676.
Kassouf, A., M. El Rakwe, H. Chebib, V. Ducruet, D.N. Rutledge, Zandomeneghi, M., L. Carbonaro, and C. Caffarata. 2005.
and J. Maalouly. 2014. Independent components analysis Fluorescence of vegetable oils: Olive oils. Journal of
coupled with 3D-front-face fluorescence spectroscopy to Agricultural and Food Chemistry, 53(3):759-766.
study the interaction between plastic food packaging and