See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318392776

Quality Assessment of Fried Palm Oils using Fourier Transform Infrared

Spectroscopy and Multivariate Approach

Article  in  Chemical Engineering Transactions · January 2017

DOI: 10.3303/CET1756139

CITATION READS

1 248

3 authors, including:

Zanariah Hashim Ida Idayu Muhamad

Universiti Teknologi Malaysia Universiti Teknologi Malaysia
13 PUBLICATIONS   49 CITATIONS    240 PUBLICATIONS   1,465 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

functional food and drug delivery system View project

Biomaterial for tissue engineering View project

All content following this page was uploaded by Zanariah Hashim on 13 July 2017.

The user has requested enhancement of the downloaded file.

 829

A publication of

CHEMICAL ENGINEERING TRANSACTIONS

VOL. 56, 2017 The Italian Association
of Chemical Engineering
Online at www.aidic.it/cet
Guest Editors: Jiří Jaromír Klemeš, Peng Yen Liew, Wai Shin Ho, Jeng Shiun Lim
Copyright © 2017, AIDIC Servizi S.r.l.,
ISBN 978-88-95608-47-1; ISSN 2283-9216 DOI: 10.3303/CET1756139

Quality Assessment of Fried Palm Oils using Fourier

Transform Infrared Spectroscopy and Multivariate Approach
Zanariah Hashim*, Siti Syatiqa Atiera Mad Zaki, Ida Idayu Muhamad
Department of Bioprocess & Polymer Engineering, Faculty of Chemical & Energy Engineering, Universiti Teknologi
Malaysia, 81310 Skudai, Johor, Malaysia
zanariahhashim@utm.my

Palm oil is one of the important commodities of Malaysia and is superior for frying due to its high oxidative
stability and high content of unsaturated fatty acids. The quality of frying oil will decrease with time and
repeated cycles of frying, and may subsequently affect the safety of the food. Several parameters such as
peroxide value (PV) and conjugated diene value (CDV) have been used conventionally to monitor the quality
of edible oils. In this study, Fourier Transform Infrared (FTIR) spectroscopy was used to evaluate the quality of
-1
palm oils after repeated cycles of frying. FTIR spectra were recorded from 4,000 to 400 cm , and analysed
using multivariate data analysis (MVA) to select important spectral regions and identify clustering pattern
-1
among samples. FTIR frequencies at 1,744 and 2,922 cm were found to be the most discriminating variables
influencing the separation between fresh and used frying oils. The correlations between peak areas at 1,744
-1
and 2,922 cm with PV and CDV values were also determined, yielding linear regression models with high
2
correlation coefficient, R . The results demonstrate that FTIR can be used as a rapid, reliable and accurate
analytical method for the quality assessment of frying oils.

1. Introduction
Palm oil is one of the most abundant oils used in the society and is a native of equatorial West Africa. Palm oil
is obtained directly from the flesh of the fruit which is making it easier to extract compared to other oils that are
obtained from the seeds. Evidence obtained from Egyptian tombs shows that palm oil was used in food for
over 5,000 years ago (Friedel, 1987). This already proves that palm oil has a long history of food use because
of its easy extraction and production. Palm oil also has the highest growth among vegetable oils especially in
Southeast Asia, because it has the highest productivity of all vegetable oils, at 5 t oil/ha, compared with 0.30
for soybean oil, 0.37 for rapeseed oil and 0.42 for sunflower oil (Gunstone, 2001).
Palm oil and its products belong to one of the most demanded oils used for the preparation of fried foods. This
is due to its relatively cheap price and it can be obtained in large amounts with high oxidative stability, and
hence results in high quality and tasty foods (Berger et al., 1991). Its high oxidative stability at high
temperatures is due to its natural antioxidants and its moderate content of polyunsaturated fatty acids
(Gunstone, 2001). The oil meets the requirements of the consumer and producer of fried food.
During frying process, the hot oil serves as a heat exchange medium at which the heat being transferred to
the food being fried. When the time passes, it leads to darkening of the oil. This is due to the oil degradation
which involves the heat, air and moisture to which the oil is exposed to, leading to oxidation, hydrolysis and
polymerisation (Gebhardt, 1996). It is very important to monitor the free fatty acids, peroxide value, iodine
value, conjugated diene and conjugated triene values which are linked to oil degradation process (Talpur et
al., 2014).
In this study, frying experiments of French fries in palm oils were conducted repeatedly and the oil parameters
were measured. Two of the oil parameters which are peroxide value and conjugated diene value were
selected as a representative conventional quality measurement in addition to spectral analysis using Fourier
Transform Infra-Red (FTIR) spectroscopy. FTIR has been proven to be fast, easy and environmental friendly
as the sample preparation is very easy and often no sample preparation is needed (van de Voort et al., 2001).

Please cite this article as: Hashim Z., Zaki S.S.A.M., Muhamad I.I., 2017, Quality assessment of fried palm oils using fourier transform infrared
spectroscopy and multivariate approach, Chemical Engineering Transactions, 56, 829-834 DOI:10.3303/CET1756139
830

FTIR used in combination with multivariate data analysis have allowed for a rapid evaluation of a large volume
of spectral data for quality control.

2. Materials and Methods

2.1 Frying procedure
French fries and palm oil were purchased from the local market. The frying process was carried out in an
electrical deep fryer with a temperature controller using 2 kg palm oil. When the oil reached 170 °C, 100 g of
frozen French fries was placed into the hot oil for one frying cycle. Frying was done for 10 minutes for each
cycle, with two frying cycles conducted within 1 h. A total of 10 frying cycles were performed using the same
frying oil. After every frying process, 10 mL of oil was collected and cooled at room temperature. The collected
samples were kept in capped amber bottles and stored in a refrigerator to protect from deterioration until
further analysis. The volume of oil was not replenished and was reused for the entire frying process.
2.2 FTIR analysis
FT-IR spectra were recorded using the IRTracer-100 (Shimadzu, Japan) equipped with an attenuated total
reflectance (ATR) accessory. Drops of oil samples were placed on a measurement plate and measurements
−1
were taken at 40 scans and at a resolution of 2 cm with Happ-Genzel apodisation. After each measurement,
the plate was carefully cleaned from any previous residues by wiping using acetone and hexane and dried
with a soft tissue before filling in with the next sample. A proprietary Lab Solutions software (Shimadzu,
−1
Japan) was used for FTIR data collection and processing. All spectra were recorded from 4,000 to 400 cm
and read as absorbance in triplicate before taking the averaged value.
2.3 Multivariate data analysis
Multivariate data analysis of FTIR spectra was performed using MetaboAnalyst 3.0 (Xia et al., 2015). Peak
intensities were scaled to unit variance.
2.4 Determination of peroxide value (PV)
PV was determined according to the thiocyanate method. 0.01 g of oil sample was mixed in a disposable
glass tube with 9.8 mL chloroform-methanol (7 : 3 v/v) in a vortex mixer for 2 - 4 s. Ammonium thiocyanate
solution (50 μL, 30 % w/v) was added and the sample was then mixed in a vortex mixer for 2 - 4 s. Then, 50
μL of iron (II) chloride solution ([0.4 g barium chloride dihydrate dissolved in 50 mL H 2O] + [0.5 g FeSO4•7H2O
dissolved in 50 mL H2O] + 2 mL 10 M HCl, with the precipitate, barium sulphate, filtered off to produce a clear
solution]) was added, and the sample was mixed in a vortex mixer for 2 - 4 s. After 5 min incubation at room
temperature, the absorbance of the sample was read at 500 nm against a blank that contained all the
reagents except the sample using a spectrophotometer.
To make the stock solution, 1 mL was taken from a 1 mg/mL iron (III) solution and diluted to 100 mL with the
solvent. Next, six samples containing 5 to 30 µg iron (III) were made for calibration model. The peroxide value
of the models was then calculated as milligram equivalents oxygen per kilogram oil (mEq O2/kg-oil) using
Eq(1) (Farhoosh and Moosavi, 2009).
PV = (Corrected absorbance x m)/ (55.84 x W x 2) (1)
where, Corrected absorbance = absorbance of sample – absorbance of blank, m = slope of calibration curve,
W = mass in gram of the samples taken, 55.84 = atomic weight of iron
2.5 Determination of conjugated diene value (CDV)
CDV was measured spectrophotometrically at 234 nm and read against HPLC grade hexane as blank. The oil
-1
samples were diluted to 1 : 600 with hexane. An extinction coefficient of 29,000 L mol was utilised to quantify
the concentration of conjugated dienes formed during the frying process (Farhoosh and Moosavi, 2009).

3. Results and Discussion

3.1 FTIR spectra of palm oil samples
FTIR spectra show the existence of specific functional groups in the samples. Since all vegetable oils
including palm oil have similar composition, the observed peaks are the same. All spectra from fresh and used
palm oils demonstrated similar pattern with no distinguishable differences, except for the intensities or peak
areas (Figure 1). Table 1 summarises the major peaks observed from palm oil samples.
 831

Figure 1: Representative FTIR spectra of fresh, 1-time and 10-times fried palm oil samples

Table 1: Major peaks observed from the FTIR spectra of palm oil samples
-1
Peak no Wavelength (cm ) Functional group Reference
1 3,008 cis double bond of unsaturated fatty acids (Talpur et al., 2014)
2 2,922 C–H stretching band from fatty acid hydrocarbon (Silverstein et al., 1991)
chain
3 2,853 C–H stretching band from fatty acid hydrocarbon (Silverstein et al., 1991)
chain
4 1,744 C=O stretching (ester) band (Talpur et al., 2014)
5 1,461 bending vibrations of CH2 and CH3 aliphatic (Silverstein et al., 1991)
groups
6 1,160 stretching vibrations of the C–O ester group (Talpur et al., 2014)

3.2 Peroxide value

Peroxide value (PV) is commonly used to determine the rancidity of a sample containing fat or oil subject to
oxidation. PV shows the degree of oxidation in the substance and measures the amount of total peroxides as
a product of primary oil oxidation. This value is a commonly used indicator of the shelf life of a product
because an elevated peroxide value will accompany disagreeable odours. Fresh oils usually have peroxide
values well below 10 mEq O2/kg-oil (Egan et al., 1981), though it also depends on the type of oil and whether
or not the oil has been refined since PV is greatly reduced by the refining process. A rancid taste is often just
noticeable for peroxide values between 10 and 20. In this study, PV value for fresh palm oil was 0.20 mEq
O2/kg-oil, and increased to 1.48 - 5.78 mEq O2/kg-oil after frying. As shown in Figure 2(a), the PV values
reached a plateau after eight times of frying. This observation is consistent with previous reports indicating the
instability of peroxides, which are formed during the early stages of oxidation and highly susceptible to further
changes that results in the formation of secondary oxidation products. Previous study reported a decrease in
the PV of oil samples after an initial increase (Che Man et al., 1999). In another study, the PV values of
832

vegetable oils were reported to be lower than 4.42 mEq O2/kg-oil during the frying process and none of them
were above the limit of 10 mEq O2/kg-oil (Farhoosh and Moosavi, 2009).
3.3 Conjugated diene value
Conjugated diene (CDV) is formed as intermediates through a shift of double bond of polyunsaturated acids
(PUFA). It is a representative index that can be used during frying to characterise the status of oxidation of
edible oils (Houhoula et al., 2002). From Figure 2(b), it can be seen that CDV increased linearly with frying
times. While Farhoosh (2009) reported that the content founds for conjugated diene of vegetable oils during
-1
the frying process ranged from 5 to 42 mmol L , this study showed that the CDV values for fried palm oils are
-1
between 0.1 to 0.6 mmol L , possibly due to the difference in the types of oil.

(a) (b)

Figure 2: Changes in (a) peroxide values (PV) and (b) conjugated diene values (CDV) in palm oils during
frying process

3.4 Principal component analysis

Principal component analysis (PCA) is a widely used method in multivariate data analysis, often used with
spectral data that consist of thousand features that necessitate data reduction. In PCA, multi-dimensional data
are reduced to several principal components (PCs) that best represent the data in low dimensions. The PCs
are expressed in terms of percentage of variables explained. PC1 has the largest % variables explained,
followed by PC2, PC3 and so on. In this study, full FTIR spectra of fresh and used frying oils were subjected to
PCA to obtain important spectral regions that differentiate the samples and identify clustering pattern among
different frying times. As a result, separation between fresh and used palm oils can be obtained from principal
component 2 (PC2), which accounts for 9 % of the separation (Figure 3(a)). From the loading plot, two
wavelengths were found to contribute to the separation between the two groups (Figure 3(b) and 3(c)), which
-1 -1
are 2,922 and 1,744 cm . The absorbance values for fresh oil at 2,922 and 1,744 cm are higher compared to
used frying oils, with a decreasing trend in absorbance with increasing frying times (Figure 3(b) and 3(c)).
3.5 Comparison between conventional method and FTIR-MVA
To examine the feasibility of FTIR in the determination of palm oil quality during frying, the correlations
between conventional method using PV and CDV with FTIR were determined. Based on PCA result, the peak
-1
area at wavelength 2,922 and 1,744 cm were chosen to derive the linear regression with PV and CDV
values. The result is summarised in Table 2. Two types of data points were used, which are individual data
points and averaged data points (n = 3). Higher correlation values were obtained when averaged data were
-1 -1
used. In addition, the wavelength at 1,744 cm gave better linear regression compared to 2,922 cm and thus
2
higher prediction ability of PV and CDV values. The high correlation values (R > 0.9) indicates the potential of
using FTIR as an alternative to PV and CDV measurement for the quality monitoring of frying oils. FTIR is a
desirable method as it can measure directly without sample preparation, does not involve harmful chemicals,
fast result within a few minutes, easy operation and reproducible spectral data. FTIR spectra may also be
extended to derive the correlation with other oil parameters such as fatty acid content, iodine value and colour
index, and thus developed into a single platform for quality analysis.
 833

(a) (b)

(c)

Figure 3: (a) Score plot of principal component analysis of full FTIR spectra from fresh and used palm oils;
-1 -1
Loading values of different oil samples at (b) 2,922 cm and (c) 1,744 cm

Table 2: Linear regression between peak areas at featured wavelength (x) and PV or CDV values (y)
Featured PV CDV
wavelength Individual Averaged Individual Averaged
-1
(cm ) Linear R
2
Linear R
2
Linear R
2
Linear R
2

equation equation equation equation

2,922 y = -2.419x + 0.420 y = -5.968x 0.897 y = -2.419x 0.420 y = -0.511x 0.902
21.08 + 47.65 + 21.08 + 4.08
1,744 y = -4.907x + 0.679 y = -9.250x 0.974 y = -0.41x 0.670 y = -0.784x 0.963
26.91 + 48.14 + 2.25 + 4.09

4. Conclusions
FTIR spectroscopy combined with multivariate data analysis was proven to be a fast and reliable alternative
evaluation method for the stability and quality monitoring of frying oils. The effects of using palm oil for up to
ten times were studied by comparing the peroxide and conjugated diene values with spectral analysis using
FTIR. Through PCA modelling, the parameters and clustering patterns with identifiable features were
-1
obtained. FTIR frequencies at 2,922 and 1,744 cm were found to be the most discriminating variables
influencing the separation between fresh and used frying oils. The absorbance values for fresh oil at 2,922
-1
and 1,744 cm are higher compared to those of used frying oils, with a decreasing trend in absorbance with
-1
increasing frying times. Linear regression between peak areas at 2,922 and 1,744 cm with PV and CDV
showed good correlations, implying the potential use of FTIR as a single platform quality assessment method.

Acknowledgments

This study was financially supported by the Ministry of Higher Education Malaysia/ Universiti Teknologi
Malaysia (vot no.11J34).
 834

Reference
Berger K.G., Andaner W.T., Applewhite T.H., 1991, The Lauric oils medium chain fatty acid source, World
Conference on Oleochemicals Into the 21st Century, 8-12 October 1990, Kuala Lumpur, Malaysia, 88-93.
Che Man Y.B.C., Liu J.L., Jamilah B., Abdul Rahman R., 1999, Quality changes of refined-bleached-
deodorized (RBD) palm olein, soybean oil and their blends during deep-fat frying, J. Food Lipids 6, 181–
193.
Egan H., Kirk S.R., Sawyer R., Pearson D., 1981, Pearson’s Chemical Analysis of Foods, 8th ed., Churchill
Livingstone, New York, United States.
Farhoosh R., Moosavi S.M.R., 2009, Evaluating the Performance of Peroxide and Conjugated Diene Values in
Monitoring Quality of Used Frying Oils, J. Agric. Sci. Technol. 11, 173–179.
Friedel M.C., 1987, On fatty materials found in an Egyptian tomb at Abydos, Comptes Rendus 24, 648–651.
Gebhardt, B., 1996. Oils and Fats in Snack Foods, Ed. Hui Y.H., Bailey’s Industrial Oil & Fat Products, John
Wiley & Sons Inc., New York, US, 407–427.
Gunstone F.D., 2001, Structured and Modified Lipids, CRC Press, Florida, United States.
Houhoula D.P., Oreopoulou V., Tzia C., 2002, A kinetic study of oil deterioration during frying and a
comparison with heating, J. Am. Oil Chem. Soc. 79, 133–137.
Silverstein R.M., Bassler G.C., Morrill T.C., 1991, Spectrometric Identification of Organic Compounds, 5th ed.,
Wiley, New York, United States.
Talpur M.Y., Kara H., Sherazi S.T.H., Ayyildiz H.F., Topkafa M., Arslan F.N., Naz S., Durmaz F., Sirajuddin
2014, Application of multivariate chemometric techniques for simultaneous determination of five
parameters of cottonseed oil by single bounce attenuated total reflectance Fourier transform infrared
spectroscopy, Talanta 129, 473–80.
van de Voort F.R., Sedman J., Russin T., 2001, Lipid analysis by vibrational spectroscopy, Eur. J. Lipid Sci.
Technol. 103, 815–826.
Xia J., Sinelnikov I.V., Han B., Wishart D.S., 2015, MetaboAnalyst 3.0—making metabolomics more
meaningful, Nucleic Acids Res. 43, W251–W257.

View publication stats