Food Chemistry 246 (2018) 192–201

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Review

Electronic noses in classification and quality control of edible oils: A review T

a,⁎ a a b
Tomasz Majchrzak , Wojciech Wojnowski , Tomasz Dymerski , Jacek Gębicki ,
Jacek Namieśnika
a
Department of Analytical Chemistry, Faculty of Chemistry, Gdańsk University of Technology, Narutowicza 11/12, 80-233 Gdańsk, Poland
b
Department of Chemical and Process Engineering, Faculty of Chemistry, Gdańsk University of Technology, Narutowicza 11/12, 80-233 Gdańsk, Poland

A R T I C L E I N F O A B S T R A C T

Keywords: The growing demand for wholesome and nutritious food leads to intensification of production, which in turn can
Edible oils have a detrimental effect on quality and well-being of consumers. For that reason, it is important to develop
Electronic nose novel methods of food control which would be characterized by a short time of analysis, adequate sensitivity and
Food control relatively low cost. One such technique involves the use of multi-sensory devices called electronic noses. In
Quality assessment
recent years there has been a rapid development of this method, especially in the area of food control. Electronic
olfaction can be successfully used in the analysis of edible oils, in particular in the determination of the product’s
geographical origin, and in detection of adulteration as well as deterioration caused by external factors.

1. Introduction The tool most commonly used in industry for assessing the flavour
or aroma of products such as cosmetics, food or beverages is still the
Ensuring the quality of products is of outmost importance in food human olfactory sense. However, the cost of employing trained sensory
manufacturing. This is true not only in the case of final products, but experts is relatively high since they can only work for a short period of
also raw materials and intermediates. An example could be vegetable time due to sensory fatigue (Wojnowski, Majchrzak, Dymerski, Gębicki,
oils which can be consumed directly, as well as used as an ingredient or & Namieśnik, 2017). Other limitations of sensory evaluation include
in food processing, mostly for frying. For this reason, it is important to low reproducibility and repeatability of results as well as the inability to
be able to reliably monitor the quality of edible oils. perform quantitative analysis (Dymerski, Chmiel, & Wardencki, 2011).
Both the global production and consumption of vegetable oils is The physicochemical properties of these products can also be de-
steadily increasing. Experts at OECD and FAO estimate that per capita termined using instrumental techniques, e.g. gas chromatography or, as
production of vegetable oils shall increase by 1.5% p.a. relative to the is often the case with determination of oil samples, using classical
2013–2015 period and the global production of vegetable oils shall analytical techniques like titration. The use of these techniques is re-
therefor increase by over 40 Mt by 2025 (OECD-FAO, 2016). This latively time- and labour-consuming, and the relationship between the
growing demand might have an adverse effect on the overall quality of physicochemical properties of odorous substances and their hedonic
edible oils which, together with increased interest of consumers in qualities is yet to be fully understood. For these reasons, there is a need
proper nutrition presents new challenges to analysts determining the to develop a technique which could supplement the human olfactory
composition and quality of food products. sense and be used to obtain sensory data within a short period of time

Abbreviations: ANN, artificial neural networks; ANOVA, analysis of variance; AOCS, American oil chemists' society; AV, anisidine value; BAW, bulk acoustic wave; CA, cluster analysis;
CC, canonical correlation; CP, conducting polymer; DFA, discriminant function analysis; DHS, dynamic headspace; Duncan’s MRT, Duncan’s multiple range tests; EEC, European
economic community; EVOO, extra virgin olive oil; FAME, fatty acid methyl esters; FAO, food and agriculture organization; FGC, fast gas chromatography; FIA, flow injection analysis;
FID, flame ionization detector; FLT, fisher linear transformation; GC, gas chromatography; HCPC, hierarchical clustering on principal component; HPSEC, high performance size exclusion
chromatography; HRGC-O, high-resolution gas chromatography – olfactometry; ICA, independent component analysis; IUPAC, international union of pure and applied chemistry; IV,
iodine value; KNN, K-nearest neighbours; LDA, linear discriminant analysis; MDS, multidimensional scaling; MIR, mid-infrared spectroscopy; MOS, metal oxide semiconductor; MOSFET,
metal-oxide semiconductor field-effect transistor; MS, mass spectroscopy; NIR, near-infrared spectroscopy; NMR, nuclear magnetic resonance; NNMF, non-negative matrix factorization;
OECD, organisation for economic co-operation and development; OO, olive oil; PCA, principal component analysis; PCC, pearson correlation coefficient; PDO, protected designation of
origin; PLS, partial least squares; PLS-DA, partial least squares – discriminant analysis; PV, peroxide value; QDA, quadratic discriminant analysis; RBD, refined, bleached and deodorized;
RBF, radial basis function; RR, ridge regression; SAW, surface acoustic wave; SHS, static headspace; SIMCA, soft independent modelling by class analogy; SLDA, stepwise linear
discriminant analysis; SPME, solid phase micro-extraction; SVD, singular value decomposition; SVM, support vector machin; TD, thermal desorption; TGS, taguchi gas sensors; UV–Vis,
ultraviolet–visible spectroscopy; VA, voltamperometry
⁎
Corresponding author.
E-mail addresses: [email protected] (T. Majchrzak), [email protected] (W. Wojnowski), [email protected] (T. Dymerski),
[email protected] (J. Gębicki), [email protected] (J. Namieśnik).

https://doi.org/10.1016/j.foodchem.2017.11.013
Received 4 June 2017; Received in revised form 26 September 2017; Accepted 2 November 2017
Available online 03 November 2017
0308-8146/ © 2017 Elsevier Ltd. All rights reserved.
T. Majchrzak et al. Food Chemistry 246 (2018) 192–201

and at a low cost. A promising development in this area is the use of analysis is used (Dymerski et al., 2011).
multisensorial systems, the so-called ‘electronic noses’. These are de- Electronic noses resemble their mammalian equivalents mainly in
vices equipped with an array of chemical sensors the response signal of function. Volatile compounds are introduced through the sampling
which is processed using pattern recognition algorithms. mechanism and transferred to the sensors chamber. The response signal
The increased interest in application of electronic olfaction in the of these sensors is then processed using pattern recognition algorithms.
evaluation of the quality of vegetable oils can be linked with the in- The type of sensors that is most commonly used in e-noses is metal
troduction of certain regulations by the European Commission. Some oxide semiconductors (MOS), but metal oxide semiconductor-field ef-
examples are Regulation 2081/1992 which introduced the ‘Protected fect transistors (MOSFET), conducting polymer sensors (CP) and pie-
Denomination of Origin’ (PDO) (Council of the European Union, 1992) zoelectric sensors are also used. The current trend in electronic nose
or Regulations 1019/2002 (The Commission of the European development is to minimize the impact of changes in temperature and
Communities, 2002) and 182/2009 (The Commission of the European relative humidity on the sensor’s output signal, while increasing the
Communities, 2009) which concern the marketing standards of vege- sensitivity. To this end in the late 1990’s electronic noses based on mass
table oils. Yet another Regulation (1335/2013) mandates the inclusion spectrometry (MS) were introduced. These devices are characterized by
of information concerning the geographical origin of edible oils on the their superior stability, sensitivity and by a wider range of potential
product’s label (The European Commission, 2013). applications compared to their sensor-based equivalents (Mielle,
Consumption of poor quality oil can have a detrimental effect on the Marquis, & Latrasse, 2000). This offers a solution to some persistent
consumer’s health, as evidenced by the so-called ‘toxic oil syndrome’ – problems such as sensor poisoning or non-linear response signal. Cur-
the poisoning of over 20.000 people after ingesting edible oil con- rently, the MS-based e-nose technology is being rapidly advanced – a
taminated with aniline. This single event resulted in over 1200 deaths trend that can in the future culminate in the development of a solution
(Posada de la Paz, Philen, & Borda, 2001). Assuring proper food safety in which analytes are introduced directly into the mass spectrometer’s
is another reason to develop a sensitive method which would enable ionization chamber without the need for prior separation on a chro-
fast, routine evaluation of vegetable oils. In the past decades, a tech- matographic column (Martı́, Busto, Guasch, & Boqué, 2005). In order to
nique has been developed which shows promise in this regard – the increase the versatility of electronic nose devices, the most appropriate
electronic nose. solution would be to pre-separate the gas mixture before introducing it
In this article, an emphasis will be placed on the application of into the detection system. Therefore, rapid gas chromatography (Fast,
electronic olfaction in investigations of edible oil’s classification, de- Flash-GC) is used (Dymerski et al., 2011). These systems are char-
tection of adulterations and quality evaluation, as well as on a critical acterized by limited resolution, but providing them with non-selective
discussion of data analysis and sampling. sensors, such as FID (Wiśniewska, Dymerski, Wardencki, & Namieśnik,
2015) or PID (Nölscher, Sinha, Bockisch, Kï Upfel, & Williams, 2012),
increases the application potential of the device. Both MS and GC based
2. Electronic nose systems
e-noses have many advantages, and the most important of these are
high sensitivity and the ability to selectively separate the mixture. In
The idea of an electronic nose (e-nose) as an ‘intelligent’ system
addition, the familiar operation principle of these devices enables re-
equipped with an array of chemical sensors was first brought up in the
ference of results to advanced instrumental analysis techniques such as
early 1980’s (Persaud & Dodd, 1982), and first commercially available
GC–MS. It should be noted that such e-noses are usually stationary,
devices were introduced in 1993. The milestones in the development of
require appropriate measurement conditions and are expensive to
electronic olfaction for analysis of edible oils are presented in Fig. 1.
manufacture and maintain. However, it should be pointed out that the
The construction of electronic noses is relatively straightforward,
use of such devices may involve generation of a large number of vari-
especially compared to advanced analytical techniques such as gas
ables, such as fragmentation patterns for MS analysis and GC chroma-
chromatography. They are usually comprised of an incubated sensor
tography peaks. Using pattern recognition in situations where there is a
chamber, a pneumatic system, detection chamber with an array of
small number of independent measurements can lead to a so-called
sensors and a data acquisition and processing system. Nowadays, the
'voodoo correlation', that is, showing a false correlation between results
term electronic nose pertains to devices that also employ well-known
(Amann et al., 2014). Therefore, the number of variables should not
analytical techniques such as mass spectrometry and gas chromato-
exceed the number of objects in the data matrix.
graphy, mostly fast gas chromatography. It is, therefore, necessary to
The improvement of existing electronic noses and introduction of
update the definition proposed by Persaud & Dodd and to consider
new ones results in steadily increasing interest in potential applications
electronic olfaction as devices used for rapid identification and classi-
of these devices. When using early e-nose models equipped with
fication of mixtures of volatile compounds, for which multivariate

Fig. 1. Certain milestones in the development of electronic noses for analysis of edible oils.

193
T. Majchrzak et al. Food Chemistry 246 (2018) 192–201

Table 1
Advantages and disadvantages of different e-nose systems.

Detection type Advantages Disadvantages Possible application

Semiconductor sensors (MOS) Variety of commercially available models Sensor drift Cheapest solutions
Relatively inexpensive Susceptible to poisoning Stationary devices
High power consumption Most of the current applications
Humidity-dependent signal Most common in prototype solutions

Electrochemical sensors (EC) Low power consumption Little choice of commercially available Few commercially available solutions
High durability sensors Portable devices
High sensitivity Bulky Dedicated solutions
Minimal impact of humidity on the signal

Piezoelectric sensors (SAW, BAW) Small-sized Little choice of commercially available Portable devices
Short response time sensors Miniaturized systems
High sensitivity and selectivity Low signal-to-noise ratio Electronic noses dedicated to specific solutions
High elaboration cost
Low reproducibility of measurement

MS-based High sensitivity Large-sized Laboratory screening

Universal application High power consumption For complicated matrices
Large quantity of obtained variables High cost Searching for correlation with other mass
Quantitative and qualitative information Complex construction spectrometry techniques
Universal detector

GC-based Numerous application possibilities Large-sized Laboratory screening

Large amount of data collected Complex construction In solutions where one or a few sensors are used
Possibility of simple quantitative and Need for carrier gas Control of technological processes in industrial
qualitative analysis Relatively long analysis time plants
Analysis of selected part of volatile fraction

‘classical’ chemical sensors it was difficult to analyse complex matrices, define the outliers. For example, principal component analysis is com-
e.g. food or drugs due to insufficient measurement stability, the need monly used to classify and find connections between objects. It is one of
for frequent calibration, sensor poisoning and high energy consumption the most commonly used techniques (see Supplement).
(MOS sensors operate at 300–500 °C). The novel solutions, e.g. elec- Unsupervised methods are used to classify vegetable oils by bota-
tronic noses based on ultra-fast gas chromatography (Xiao et al., 2014) nical origin (Gan, Che Man, Tan, NorAini, & Nazimah, 2005b), quality
might alleviate some of these issues. (González Martín et al., 2001) or geographic origin (Haddi et al., 2013).
The most common taxonomy of electronic noses is based on the The advantage of unsupervised methods is that due to the lack of a
detection method of volatiles. The devices are classified into the ones learning stage, a relatively small amount of input data is needed to
that are equipped with an array of chemical sensors, e.g. MOS, obtain meaningful information. However, unsupervised methods such
MOSFET, CP, surface acoustic wave sensors (SAW), bulk acoustic wave as PCA and CA cannot be used in automatic classification algorithms,
sensors (BAW), and the ones in which other means of detection are apart from pre-processing. Each decision about classifying an object as
utilized (e.g. MS). Another taxonomy is based on the sensors’ operating belonging to a group is made by the person taking the measurement. In
temperature, i.e. ‘hot’ and ‘cold’. Both have advantages and dis- order to automate the classification into predetermined groups, or to
advantages, e.g. using e-noses equipped with the former allows to compare the quality of the oil, its freshness, degree of oxidation or
minimize the impact of relative humidity on the device’s output signal. packaging compliance, it would be more appropriate to use supervised
However, high operating temperature leads to increased power con- methods. Particularly useful supervised methods are linear discriminant
sumption, which limits possible field applications. The advantages and analysis (LDA), support vector machines (SVM) and artificial neural
limitations of various types of e-noses and their potential uses are networks (ANNs) which are discussed later in this section. It should be
presented in Table 1. The decision which device type to use should be emphasized that the most relevant data analysis methods used to
dictated by the conditions in which it is going to be deployed. compare the obtained results with the results of the reference analyses
are, in particular, those result of which is numerical (ANN, LDA, DA).
The examples of statistical data analysis methods are described below.
3. Data analysis in electronic olfaction

The output of a multisensorial system presents a very complex set of 3.1. Supervised methods
data. Each sensor’s response is a different analytical signal and with
slight differences between particular samples the change in its ampli- In supervised methods, in order to train the model, both the input
tude can be minuscule. For that reason it is necessary to use data and the target results are provided. After the training stage inputs are
analysis methods to properly interpret the output signal (Dymerski assigned to particular classes. Techniques of this type are commonly
et al., 2011). The main purpose of data analysis methods is to empha- used for classification and discrimination between samples when there
size the similarities and differences between the components of the is a reference sample with desirable characteristics, which is then
dataset through reduction of dimensionality. This is usually done by compared with other samples. Artificial neural networks (ANN), linear
transforming the output signal in the following stages: pre-processing, discriminant analysis (LDA) and support vector machines (SVM) are
feature extraction, classification and decision making. Data analysis among the methods most commonly used in connection with electronic
methods are often classified based on the way in which the model is noses.
‘taught’ into supervised and unsupervised methods. The choice of an Linear discriminant analysis is a probabilistic classification tech-
appropriate statistical analysis method depends on the number of ob- nique in which the data space is rotated in such a way as to maximize
jects and their variables, the complexity of the problem, and the com- the variance between classes and at the same time minimize the var-
putational capabilities of the software. Unsupervised methods can be iance within classes (Hai & Wang, 2006a). This is achieved by trans-
used to get initial information, initial recognition of dependencies, or to posing the data points to space of reduced dimensionality. In LDA

194
T. Majchrzak et al. Food Chemistry 246 (2018) 192–201

classes are characterized by normal distribution and equal dispersion components’ are aligned in such a way as to explain the greatest
(González Martín et al., 2001). Classes are determined based on a amount of variance and at the same time retain as much input data as
training data set (Melucci et al., 2016). There are numerous variations possible (Mildner-Szkudlarz, Jeleń, & Zawirska-Wojtasiak, 2008). The
of this method, e.g. stepwise linear discriminant analysis (SLDA) first principal component explains the greatest amount of variance
(Aparicio, Rocha, Delgadillo, & Morales, 2000). within the data set and the subsequent principal components, perpen-
Artificial neural networks can be defined as basic, interconnected dicular to each other, contain increasingly less information regarding
computational units (nodes) which transform the input data into sets of the remaining variance. Principal component analysis can be used for
output values. Nodes work in parallel and can ‘communicate’ with each analysis, classification and reduction of dimensionality of a numerical
other. In chemometrics neural networks are used to perform both su- data set as well as for elimination of data redundancy. Additionally PCA
pervised and unsupervised pattern recognition. In order to classify the is also often used during the pre-processing stage of statistical analysis
data points the network builds a model based on the training data set and for detection of outliers (Aparicio et al., 2000).
with known outputs. At this stage weight is assigned to each connection In cluster analysis (CA) a taxonomy of the data set is built based on
between two nodes in such a way as to best match the output with the the similarities or distances between points in the data set. The result of
target (Hai & Wang, 2006a). In the subsequent stage data without as- cluster analysis is often depicted as a dendrogram (Haddi et al., 2013).
signed targets is introduced to the network to be processed based on the The distances between individual data points and the centre of each
previously built model. cluster is calculated using the pooled within-group covariance matrix
Support vector machines is a supervised learning technique that is (Xu, Yu, Liu, & Zhang, 2016) with each object classified as belonging to
used to develop classifiers (Bougrini et al., 2014). It was first presented the cluster closest to it (Huang, Guo, Qiu, & Chen, 2007). At first, each
at the Conference on Computational Learning Theory in 1992 (Boser, object is assigned to its own cluster and then, in order to determine
Guyon, & Vapnik, 1992) which makes it one of the more recent tech- when two or more objects belong to a single, higher cluster, the
niques for classification and prediction. It is based on algorithms known threshold is gradually lowered. This way aggregates of clusters begin to
as ‘kernel methods’ (Cristianini & Scholkopf, 2002). In SVM data is form with the differences between individual components of each
embedded in a multi-dimensional space and separated into classes by a cluster becoming increasingly pronounced. Eventually, all the objects
hyperplane. The generalization power with which the hyperplane se- merge into a single cluster. Cluster analysis is often used to identify
parates the classes depends on its margin, defined as the distance be- possible sub-classes in groups of objects (Melucci et al., 2016). The
tween the hyperplane and the data points closest to it - support vectors taxonomy of selected methods of statistical data processing is depicted
(El Barbri, Llobet, El Bari, Correig, & Bouchikhi, 2008). An example of in Fig. 3.
separation of a dataset into two distinct classes by a hyperplane is given
in Fig. 2. Since the margin is determined based on the closest training 4. Sampling in electronic nose technique
instances, the classification based on this maximum margin can be in-
sufficiently accurate, which is why often an additional safety margin is Proper sampling of the volatile fraction and conveying it to the
determined (Cristianini & Scholkopf, 2002). sensor array represents a significant challenge during the design of
analytical methodology involving the use of an electronic nose. Most
commonly the sampling system works in either static or dynamic
3.2. Unsupervised methods headspace mode. The analytes can also be isolated and pre-con-
centrated using solid phase microextraction (SPME). Which technique
Data analysis techniques in which target outputs are not supplied is used depends mostly on the sensitivity of sensors in the detection
during the learning stage are called unsupervised methods. They system, differences in composition of discriminated oil samples and the
usually work by building clusters or plots based on the statistical state of matter that they are in. It is a common practice to use longer
properties of input data and are used for reducing the dimensionality of period of sampling, e.g. by extension of the incubation time in order to
data sets, classification and detection of distinct sub-classes. Cluster enrich the headspace, retention of the sampled gas in the sensors
analysis and principal component analysis are among the most com- chamber or the exposure time of SPME fibre to analytes. For that
monly used unsupervised methods. reason, a single analysis often takes up to an hour. According to its
Principal component analysis (PCA) is a powerful pattern recogni- definition, the use of an electronic nose should produce results within
tion technique. It can be used to obtain a graphical representation of several minutes (Dymerski et al., 2011). Therefor it is necessary to find
analysis results with the differences and similarities between data sets the optimum between the device’s sensitivity and the time of a single
emphasized (Hai & Wang, 2006a). The original data space is trans- measurement. The methods of sampling of gaseous substances that are
formed into a new coordinate system in which the vectors, or ‘principal most commonly used in electronic noses during the analysis of edible
oils are described below and depicted in Fig. 4.

4.1. Static headspace (SHS)

Static headspace is one of the earliest and most basic methods of

sampling. The fact that only a relatively small fraction of the headspace
can be sampled is the main limitation of this method. However it is still
in use, especially in devices equipped with sensitive detectors and in
some electronic noses, in particular models based on fast GC (Ghosh,
Chatterjee, Bhattacharjee, & Bhattacharyya, 2016) and MS (Casale,
Casolino, Oliveri, & Forina, 2010; Park et al., 2011). Static headspace
sampling in e-noses is usually performed either by drawing the sample
with a syringe (Haddi et al., 2011; Mildner-Szkudlarz & Jeleń, 2010) or
with a sampling loop (Guadarrama, Rodrıguez-Mendez, De Saja, Rıos, &
Olıas, 2000). Sampled headspace can be then introduced into the sen-
sors chamber mixed with a stream of carrier gas or adsorbed on a
sorbent material and then thermally desorbed (Ghosh et al., 2016;
Fig. 2. Separation of a dataset by a plane in the SVM method.
Melucci et al., 2016). Analysis is usually preceded by incubation in

195
T. Majchrzak et al. Food Chemistry 246 (2018) 192–201

Fig. 3. The taxonomy of the most commonly used data

analysis techniques.

4.2. Dynamic headspace (DHS)

Due to relative simplicity of the pneumatic system DHS is often used

as a sampling method in electronic noses. An advantage of this method
when used in e-noses is the relatively short response time and faster
recovery as the analytes are flushed from the system. In electronic noses
two types of DHS sampling are commonly used. In the most popular
solution a continuous flow of carrier gas is directed through the sam-
ple’s headspace into the sensor chamber or directly into it bypassing the
sample in order to flush the sensors and remove the residue from pre-
vious analysis (Cimato et al., 2006; González Martín et al., 2001). The
second type resembles dynamic headspace sampling systems used in gas
chromatography (the so-called ‘purge and trap’). This type of sampling
is utilized in the commercially available zNose™ (Gan, Tan, Che Man,
NorAini, & Nazimah, 2005a). When using DHS it is not necessary to
place oil samples in special vials or other vessels (Sonia et al., 2009).
There is also no need for prior incubation of the sample in order to
facilitate the transfer of analytes to the volatile fraction (Bougrini et al.,
2014; Cimato et al., 2006). Even though it is not necessary to generate
headspace prior to analysis, doing so might improve the sensitivity of
the measurement. Because of that incubation periods of several seconds
(Ghosh et al., 2016), minutes (González Martín, Pérez Pavón, Moreno
Cordero, & García Pinto, 1999) and up to an hour (Aparicio et al., 2000;
James, Scott, O’Hare, Ali, & Rowell, 2004) are sometimes introduced.
Longer incubation periods not only extend the time of a single analysis,
but can also facilitate the generation of the products of thermal de-
gradation of oils, and so shorter incubation is generally desirable.

4.3. Solid phase microextraction (SPME)

Solid phase microextraction is seldom coupled with e-nose tech-

nique, partly because it would involve an addition of complicated and
expensive peripheral devices, and also because it would entail an ex-
Fig. 4. Sampling techniques used in electronic olfaction: a- static headspace, b- dynamic tension of the time of a single analysis. Nevertheless, SMPE was used to
headspace, c- solid phase micro-extraction. enrich analytes from the headspace of Spanish olive oil (García-
González, Barie, Rapp, & Aparicio, 2004). Different SMPE fibres, ex-
order to enrich the headspace with analytes, although this is not always traction times, carrier gas flows and incubation temperatures were
the case. In one instance the incubation step was omitted because the compared. Best results were obtained using a fibre coated with Car-
electronic nose that was used for the analysis (Smart Nose 300, Marin- boxen and polydimethylsiloxane (CAR/PDMS). The highest enrichment
Epagnier, Switzerland) was based on mass spectrometry and was thus of analytes was achieved after the longest time of extraction, that is
very sensitive (Park et al., 2011). However, in majority of reviewed 40 min. Thus, the enrichment and trapping stage were by far the most
works the samples of edible oil were incubated for several tenths of time-consuming of the whole measurement since the analysis alone
minutes at 40–50 °C. higher temperatures can lead to thermal de- took app. one minute.
gradation of the sample, which would in turn introduce artefacts into
the analysis. Nevertheless it has been suggested that e.g. samples of 5. Analytical techniques in quality evaluation of edible oils
palm oil should be incubated at 160 °C for 15 min (Hong, Park, Choi, &
Noh, 2011). The methods that are most commonly used for the evaluation of
edible oils’ quality, composition and properties include sensory ana-
lysis, techniques in which titration is utilized, spectrophotometry and
gas chromatography. These methods are often references in various

196
T. Majchrzak et al. Food Chemistry 246 (2018) 192–201

Table 2
A comparison of techniques used in quality evaluation of edible oils.

Method of quality assessment of edible oils Advantages Disadvantages

Sensory analysis – direct evaluation of the oil’s quality – risk of sensory fatigue
– well-documented procedures and official standards – need to employ highly qualified personnel
– ease of use – limited sensitivity

Chemical analysis – straightforward procedure – use of organic solvents

– well-documented procedures and official standards – hard to fully automatize
– relatively low cost of a single analysis – limited sensitivity
– lack of qualitative information
– relatively time-consuming

Gas chromatography – qualitative and quantitative analysis – relatively time-consuming

– high sensitivity – often entails a sample preparation stage
– well-researched methodology – complex equipment
– guidelines present in official standards – no possibility to employ in-situ
– relatively high cost of equipment and use

Electronic nose – rapid analysis – sensor drift

– no sample preparation stage – no commercially available solutions for analysis of edible oils
– relatively low cost of a single analysis – moderate sensitivity
– can be used in portable devices – lack of qualitative information
– ease of use – risk of sensor poisoning
– still relatively expensive equipment

regulations and standards. However, the use of neither of them enables due to the need for large volumes of chemical reagents and manual
a rapid, simple in-situ analysis, without the need for sampling and control of the measuring apparatus these techniques do not find ap-
sample preparation. The advantages and disadvantages of currently plication in routine measurements.
used methods are listed in Table 2. Gas chromatography is usually used to determine the content of
Sensory analysis is based on the evaluation of the quality of a pro- fatty acids present in an oil sample. In order to do so the sample has to
duct by a team of trained panellists. It is among the oldest and com- first be derivatized so that fatty acids are transformed into methyl esters
monly used techniques for food quality control. The procedure of sen- (FAME). The procedure for determination of FAME is described in
sory analysis is clearly outlined and normalized (IOC, 2005). According several documents and standards (AOCS, 1996; European Union, 1991).
to the EU regulation, basic sensory attributes of olive oil are classified For determination of FAME gas chromatography coupled with a flame-
as either desirable (fruity, bitter, pungent) or undesirable, also called ionization detector (GC-FID) is most commonly used. Presently there is
defects (fusty, musty, winey-vinegary, muddy sediment, metallic and a significant number of studies into the use of GC–MS for the analysis of
rancid) (The Commission of the European Communities, 2002). How- the volatile fraction of edible oils. The use of chromatographic techni-
ever, it is important to note that the occurrence of olfactory sensation ques enables to determine the oils’ composition and thus the direct
depends on concentration and individual detection threshold of these cause of undesirable traits, particular flavours or nutrients profile.
substances (Morales, Luna, & Aparicio, 2005). Sensory analysis has There is a possibility of automatization of the analysis, which makes
numerous advantages: it is relatively simple, the results are easily in- these methods less labour-intensive compared to sensory and chemical
terpretable, it does not require expensive hardware and the results re- analysis. Still, it is necessary to collect a sample of oil and deliver it to a
late directly to the consumers’ experience. However, panellists are laboratory, and sometimes to perform derivatization, which sig-
professionals whose employment entails high expenses and a long nificantly increases the cost of analysis.
period of training. In some cases, such as routine in-situ measurements Electronic noses at this stage of technological advancement cannot
or on-line analysis in industrial plants these techniques are difficult to realistically replace the currently used methods. Most often research is
implement. conducted to replace or assist a group of trained panellists. This is
Chemical methods constitute a separate group of oil quality as- primarily due to the fact that, as in the sensory analysis, electronic
sessment methods. These include the peroxide value (PV), p-anisidine noses analyse the aroma of edible oils. In the literature there are nu-
value (AV), iodine value (IV), spectrophotometric measurement of merous examples of studies which try to combine these two approaches.
colour or determination of free fatty acids using gas chromatography. In In other cases, when trying to associate information obtained from an e-
the PV method peroxides and hydroperoxides formed in the oil during nose with reference methods, finding correlations is more difficult.
storage due to exposure to light and oxidation are determined by ti- However, it should be emphasized that an electronic nose can be used
tration with sodium thiosulphate. It is a reference method according to on its own for routine testing and portable solutions could in future be
ISO 3960–1977(E) (ISO 3960:1977, 1977), AOCS official method Cd used to assess the quality of oils in processing plants, restaurants or
8–53 (AOCS, 1996) and Regulation 2568/91 (European Union, 1991). homes. Therefore, the authors believe that while electronic noses will
Anisidine value is the number of secondary products of lipid com- not replace current methods, they can be used for preliminary analysis
pounds oxidation which are formed due to breakdown of peroxides and of oil quality. Thus, the potential use of these devices can lead to a
hydroperxides. It can be determined spectrophotometrically, in ac- reduction of the measuring cost by eliminating the need to collect and
cordance with ISO 6885:1988 (ISO 6885:1988, 1988) or AOCS Official transport a sample of suspect oil, the need to use reagents as in chemical
Method Cd 18–90 (AOCS, 1996). The determination of IV is described methods, or employ panellists, and have a laboratory of specified
in AOCS Official Method Cd 1b-87 (AOCS, 1996). In order to evaluate parameters. When the suspicion of an insufficient oil quality is deemed
the quality of oils measurement of absorbance in the visible light can likely, the oil could be subjected to the remaining tests. Examples of
also be used (European Union, 1991). Due to the relative simplicity of studies comparing the reference methods with the result of the elec-
measurement chemical methods are commonly used in analysis of oils. tronic olfaction are described in the next section.
Particularly popular are methods in which titration or measurement of
absorbance are utilized. For that reason, they are used in analytical
laboratories of sanitary agencies and manufacturing plants. However,

197
T. Majchrzak et al. Food Chemistry 246 (2018) 192–201

6. Electronic noses in edible oil analysis required the development of a dedicated method of data preparation
and a suitable chemometric method. The presented strategy approaches
The electronic nose technique has numerous potential applications the food authenticity analysis in a novel way and its utility should be
in routine tests of edible oils due to the lack of sample preparation stage examined for other food samples.
and short time of a single analysis. Similarly to sensory methods used A commercially available electronic nose was similarly used to
for aroma assessment the sample’s headspace is analysed, however ar- analyse olive oil originating from various regions of Italy and from
omas can be detected below their odour threshold. Moreover, the issue Spain (Cosio, Ballabio, Benedetti, & Gigliotti, 2006). The results were
of sensory fatigue is not present. Unlike in many alternative methods, compared with free fatty acids content, PV and UV absorbance. The use
there is no need to use organic solvents such as hexane, reagents (eg. of the electronic nose and the artificial neural network algorithm pro-
titrating solution) or to perform derivatization which is commonly used vided satisfactory results in less time than when using other methods. In
in GC analysis. Electronic olfaction conforms with the guidelines of addition, in order to obtain a better classification model, some of the 32
green analytical chemistry, and its use leads to a significant decrease of available sensors have been selected. This type of procedure should
the cost of a single analysis. For that reason in recent years there has always be considered if there are a large number of variables, e.g. a
been a growing interest in the use of electronic noses for classification, multitude of sensors in the sensor array or data from MS based e-nose.
detection of adulterations and mislabelling and for the evaluation of Some other studies in that area include the determination of geo-
quality of edible oils. Tables in which selected applications of electronic graphical origin of various olive oils using electronic noses based on CP
noses in the analysis of edible oils’ volatile fraction can be found in the sensors (Portuguese and Spanish, Italian) (Guadarrama, Rodríguez-
Supplement. Méndez, Sanz, Ríos, & De Saja, 2001; Stella, Barisci, Serra, Wallace, &
De Rossi, 2000), SAW sensors (Spanish) (García-González et al., 2004),
6.1. Classification of oils MOS sensors (Moroccan) (Haddi et al., 2011, 2013), or fast gas chro-
matography (Italian) (Melucci et al., 2016).
The flavour, nutritional value and physicochemical properties of Classification can also be performed within a single geographical
edible oils depend not only on the raw material, but also on the method region, e.g. in the case of single-cultivar extra virgin olive oils from
of production. Electronic noses can be used for rapid classification and Tuscany (Cimato et al., 2006).
discrimination between various types of oils.
It was for example possible to classify 16 different edible oils, in- 6.3. Oil adulterations
cluding rapeseed, sunflower and grape seed oils based on the compo-
sition of their headspace (Gan et al., 2005b). Parameters which are used The consumers’ preference for a particular edible oil might depend
to evaluate the quality of oils, i.e. AV, IV, PV and fatty acid content on its flavour as well as nutritional value. Due to concerns regarding
were also determined. In a different study over 140 samples of olive oil potential health hazards the determination of authenticity of edible oils
of three different kinds, namely virgin, non-virgin and seed oil were is an important challenge for scientists specializing in food analysis.
analysed (González Martín et al., 1999) using an electronic nose Some manufacturers adulterate their products with oils of inferior
equipped with MOS. Best discrimination was achieved using ANN and quality. This practice concerns mostly the suppliers of olive oil
LDA (over 90%). An e-nose based on piezoelectric sensors was used to (Monfreda, Gobbi, & Grippa, 2012). Virgin olive oil is often adulterated
discriminate between samples of extra virgin olive oil, non-virgin olive by mixing with olive oil of poor quality or other, less expensive edible
oil and rapeseed oil (James et al., 2004), while a device equipped with oils. In order to detect such an adulteration the FOX 4000 electronic
CP sensors was used to distinguish between olive oils of superior and nose was used (Concepción et al., 2002). In total 333 samples of pure
inferior quality (Guadarrama et al., 2000). virgin olive oil and olive with and admixture of olive–pomace oil or
The properties of sesame seed oil can depend on how the seeds were sunflower oil. The oils were combined in five different ratios. A 95%
roasted. Different kinds of sesame seed oils were classified using gas successful classification was obtained using discriminant analysis. Si-
chromatography and two types of electronic noses, one of which was milar results were also obtained in the case of a blend of olive oil with
equipped with an array of semiconductor sensors and the other with a rapeseed oil (Mildner-Szkudlarz et al., 2008). In the former study the
mass spectrometer (Park et al., 2011). Unlike the MOS based e-nose, the same e-nose model was used, but obtained results were compared with
MS based electronic nose could be used to obtain initial information on the outcome of SPME-GC–MS and SPME-MS analysis. It was concluded
the composition of the sesame seed oils. The combination of these three that both electronic olfaction and SPME-GC can successfully be used as
devices was a useful tool that could be used to discriminate samples, an alternative for the more time-consuming methods. The adulteration
while at the same time indicating which chemical compounds are re- of olive oil with hazelnut oil were also investigated (Mildner-Szkudlarz
sponsible for this distinction. The study has shown that it is possible to & Jeleń, 2008), which is a difficult task due to similarities in the content
classify these oils based on the roasting temperature. of fatty acids, sterols and triacylglycerols (Mariani, Bellan, Lestini, &
Research aiming to discriminate samples of different botanical Aparicio, 2006; Zabaras & Gordon, 2004). These studies compared
origin rarely finds application in the oil processing industry. It is more SPME-Fast-GC, SPME-MS and HS-Electronic nose. Each of these devices
important to determine the quality of vegetable oils, e.g. manufactured was able to detect the 5% addition of the hazelnut oil, however, the last
in a production plant, or oils used in food production, and electronic two had a significantly shorter analysis time.
noses can be used for this purpose. It should be stressed, however, that Some edible oils which, considering their high price, can be con-
the presented research is largely scientific in nature and cannot be sidered luxury products may also be adulterated with cheaper oils. For
considered as viable for future industry use. example, argan oil can be mixed with much cheaper and readily
available sunflower oil. A prototype electronic nose equipped with an
6.2. Determination of geographical origin array of MOS sensors was used to evaluate the possibility of detecting
such an adulteration (Bougrini et al., 2014). Additionally, samples were
The evaluation of food products authenticity is connected with the also analysed using an electronic tongue. The use of both devices led to
determination of their geographical origin. The electronic nose tech- good results, however a better discrimination (100%) was obtained
nique finds application in this area. For example a HS-MS e-nose was using the electronic tongue.
used to discriminate extra virgin olive oil from the Liguria region Palm stearin oil contains more saturated fatty acids than palm olein
against other commercially available olive oils (Casale et al., 2010). In which is why it is considered to be of inferior quality. Blending of this
order to classify the samples SIMCA classifier was used, the input data two oils has a detrimental effect on the latter, which is commonly used
for which came from the e-nose, UV–Vis and NIR. This type of approach for frying. It was shown that a 3% admixture of palm stearin oil can be

198
T. Majchrzak et al. Food Chemistry 246 (2018) 192–201

detected using an electronic nose (Hong et al., 2011). There is also a increase. It was demonstrated that the PEN-3 electronic nose can be
good correlation (R2 = 0.99) between the results obtained using an MS- successfully used to investigate the stability of olive, peanut, soybean,
based electronic nose and gas chromatography when analysing samples rapeseed, camellia, corn, sunflower, linseed, and walnut oils incubated
of palm olein oil adulterated with palm stearin oil. at 105 °C for 12–36 h (Xu et al., 2016). Using cluster analysis, PCA and
Edible oils can also be adulterated with animal fat. Traditionally the LDA it was possible to differentiate between oxidised and non-oxidised
latter have been used for deep frying but in recent years the consumers oils regardless of their botanical origin. The compact size and port-
increasingly prefer using vegetable oils for this purpose (Stevenson, ability of this device in particular and of electronic noses in general are
Vaisey-Genser, & Eskin, 1984). The admixture of animal fats can be considerable advantages which can potentially facilitate the introduc-
undesirable both because of their potential detrimental impact on tion of this technique in routine analysis of the degree of oxidation of
human health (Cawley, Meyerhoefer, Gillingham, Kris-Etherton, & edible oils, supplementing the currently established methods.
Jones, 2017) and certain cultural and religious taboos. In a study aimed The greatest changes of the overall quality of oil caused by oxida-
at detecting such an adulteration a refined, bleached and deodorized tion occur during frying. In order to mitigate the effects of this process it
palm olein (RBD palm olein) was mixed with pig lard (Man, Gan, is a common practice to introduce antioxidants which increase the oil’s
NorAini, Nazimah, & Tan, 2005). The results obtained using a com- stability. In a study of the increase of frying disposal time of sunflower
mercially available electronic nose equipped with SAW sensor were oil by admixture of oleoresin rosemary and ascorbyl palmitate a com-
compared with results from official AOCS methods, namely IV, AV and mercially available electronic nose equipped with an array of MOS
free fatty acids content measurements. The e-nose technique was the sensors and coupled with fuzzy logic analysis was employed
most sensitive since its use allowed to detect the admixture of 1% v/v (Upadhyay, Sehwag, & Mishra, 2017a,b). It was shown that the e-nose
pig lard. The same device was also used to detect adulterations of virgin can be successfully used for predicting the frying disposal time of oils
coconut oil with RBD palm olein (Marina, Che Man, & Amin, 2010). used for frying which suggests that a similar device can in the future be
Some other applications of electronic noses in this area include the utilized in on-line monitoring of the process of frying. In other studies,
analysis of sesame seed oil (Hai & Wang, 2006a,b) and camellia oil (Hai the aim was to determine the effect of adding olive oil on the thermal
& Wang, 2006a) with the admixture of corn oil. A comparison of che- stability of rapeseed oil (Majchrzak et al., 2017). To this end, the
mometric methods has shown that using PCA it is impossible to detect thermal stability of refined rapeseed oil, olive oil and the commercially
adulteration. The linear discriminant analysis is significantly better for available mixture of both oils was compared. It has been shown that the
this purpose, although a complete discrimination between the samples addition of 5% olive oil increases the thermal stability of rapeseed oil.
was not achieved. Having sufficient amount of data and using the right For this purpose, the Rancimat test was used and the results were
feature extraction methods it is possible to use artificial neural net- supplemented with sensory evaluation and an analysis using Fast-GC
works to precisely predict the percentage of adulteration. based electronic nose. In the article, there is no indication of a direct
correlation between the methods used.
6.4. Assessment of oxidation
6.5. Other applications of electronic olfaction in the analysis of edible oils
The degree of oxidation can be influenced by storage conditions –
the rate of oxidation increases when the oil is exposed to light and high Edible oils of superior quality have certain characteristic, desirable
temperature. One study compared the progress of oxidation in olive oil flavours, whilst the presence of chemical compounds which have a
stored for up to 2 years in various lighting conditions (Cosio, Ballabio, detrimental effect on the product’s flavour leads to the occurrence of
Benedetti, & Gigliotti, 2007). The results obtained using electronic nose the so-called defects. It is particularly important to be able to detect
and tongue were juxtaposed with those obtained using reference these defects at low concentration levels and so a method the use of
methods recommended by the European Commission (European Union, which would allow for the detection of these compounds and undesir-
1991). It was shown that it is possible to discriminate between oil able aromas below their odour threshold could prove useful. An ex-
samples based on the conditions in which they were stored. In a dif- ample of application of electronic noses in this area is the use of a
ferent study, a commercially available electronic nose was used to as- device equipped with MOS sensors for the detection of defects in olive
sess the oxidation degree of olive oil samples incubated at 40 °C without oil (Lerma-García et al., 2010). The main defects in olive oil are char-
the access of light (Buratti, Benedetti, & Cosio, 2005). Peroxide value acterized as fusty, mouldy, muddy, rancid and winey. The use of elec-
determination was used as reference. Samples with different storage tronic olfaction and artificial neural networks enabled to determine the
times can be classified based on their degree of oxidation into two value of the defect. In a different study a device based on an array of 32
distinct categories, namely ‘fresh’ and ‘in oxidation’. CP sensors was used to determine the rancid defect in extra virgin olive
An e-nose device equipped with semiconductor sensors was also oil (Aparicio et al., 2000). Electronic noses can also find application in
used to evaluate the autooxidation degree of rapeseed oil stored at monitoring headspace composition of olive oil during malaxation
room temperature in original packaging over the period of 6 months (Sonia et al., 2009) or determining the intensity of the rancid-acidic
(Mildner-Szkudlarz et al., 2008). Sensory analysis and determination of odour of coconut oil (Ghosh et al., 2016).
the peroxide value were used as reference methods. The composition of
the oil’s volatile fraction was determined using GC–MS. There was a 7. Conclusion
clear correlation between the results of electronic nose analysis, re-
ference methods and concentration of several volatile compounds. A The growing awareness of importance of proper nutrition has made
different device based on an array of CP sensors was used to investigate the need to develop adequate tools for the analysis of edible oils more
the oxidation of rapeseed, corn and soy oils incubated at 60 °C for acute. Their desirable traits in addition to wholesomeness and pleasant
12 days without the access of light (Shen et al., 2001). In yet another flavour include physicochemical properties which make them suitable
study an electronic nose equipped with a SAW sensor was employed in for use in thermal processing. Application of the emerging technique of
the evaluation of oxidative stability of RBD palm olein (Gan et al., electronic olfaction facilitates rapid and relatively inexpensive assess-
2005a). The authors of the cited work emphasize that electronic noses ment of the quality, origin and type of vegetable oils without the need
are characterized by the fastest analysis time among the methods used for sample preparation. It can be used to supplement the currently used
for this purpose. However, the limitations arising from the use of SAW techniques, especially the ones which due to the extensive use of or-
sensors, such as high sensor production costs and limited commercial ganic solvents cannot be included among green sample preparation
availability, should be borne in mind (Wojnowski et al., 2017). methods or are relatively expensive.
The degree of degradation of edible oils is linked to temperature Based on the review of the subject literature it can be concluded that

199
T. Majchrzak et al. Food Chemistry 246 (2018) 192–201

the use of electronic noses might resolve several issues pertaining to Cosio, M. S., Ballabio, D., Benedetti, S., & Gigliotti, C. (2006). Geographical origin and
classification, determination of botanical origin, detection of adultera- authentication of extra virgin olive oils by an electronic nose in combination with
artificial neural networks. Analytica Chimica Acta, 567(2), 202–210.
tions and evaluation of the degree of oxidation of edible oils. In the Cosio, M. S., Ballabio, D., Benedetti, S., & Gigliotti, C. (2007). Evaluation of different
majority of cases the chemical sensors used in their construction were storage conditions of extra virgin olive oils with an innovative recognition tool built
able to detect volatile compounds below their odour threshold which is by means of electronic nose and electronic tongue. Food Chemistry, 101(2), 485–491.
Council of the European Union (1992). Council Regulation (EEC) No 2081/92, (L), 1–8.
a significant advantage over sensory analysis. Moreover, the time of a Cristianini, N., & Scholkopf, B. (2002). Support vector machines and kernel methods: The
single analysis is significantly shorter than in other instrumental new generation of learning machines. AI Magazine, 23(3), 31.
methods of analysis of oil samples. Dymerski, T. M., Chmiel, T. M., & Wardencki, W. (2011). Invited review article: An odor-
sensing system—powerful technique for foodstuff studies. Review of Scientific
Unfortunately, in some cases it was necessary to extend the head- Instruments, 82(11), 111101.
space enrichment stage in order to increase the sensitivity of detection. El Barbri, N., Llobet, E., El Bari, N., Correig, X., & Bouchikhi, B. (2008). Application of a
As the ability to rapidly analyse multiple samples is the most important portable electronic nose system to assess the freshness of Moroccan sardines.
Materials Science and Engineering: C, 28(5), 666–670.
advantage of electronic olfaction a balance should be found between
European Union (1991). Commission Regulation (Eec) No 2568/91. Official Journal of
this parameter and sensitivity. It should be noted however that at the the European Communities, (October), 1970.
current stage of development electronic noses have certain limitations Gan, H. L., Che Man, Y. B., Tan, C. P. P., NorAini, I., & Nazimah, S. A. H. (2005b).
which are in part due to the capabilities of data analysis methods and of Characterisation of vegetable oils by surface acoustic wave sensing electronic nose.
Food Chemistry, 89(4), 507–518.
commercially available sensors. The main weak point of the former is Gan, H. L., Tan, C. P., Che Man, Y. B., NorAini, I., & Nazimah, S. A. H. (2005a).
sensor drift, and also in certain cases limited longevity and sensitivity. Monitoring the storage stability of RBD palm olein using the electronic nose. Food
Concentration of particular volatile organic compounds such as alde- Chemistry, 89(2), 271–282.
García-González, D. L., Barie, N., Rapp, M., & Aparicio, R. (2004). Analysis of virgin olive
hydes, ketones or alcohols in the oil sample’s headspace is often in the oil volatiles by a novel electronic nose based on a miniaturized SAW sensor array
range of single ppb v/v. Several methods allow for quantitative analysis coupled with SPME enhanced headspace enrichment. Journal of Agricultural and Food
of the oils’ composition, which is difficult to achieve using multi-sen- Chemistry, 52, 7475–7479.
Ghosh, P. K., Chatterjee, S., Bhattacharjee, P., & Bhattacharyya, N. (2016). Removal of
sory devices. For that reason electronic noses might be used to sup- rancid-acid odor of expeller-pressed virgin coconut oil by gamma irradiation:
plement rather than replace the currently used standards. Commercially Evaluation by sensory and electronic nose technology. Food and Bioprocess
available electronic noses remain relatively expensive and have not yet Technology, 9(10), 1724–1734.
González Martín, Y., Concepción, M., Oliveros, C., Luis, J., Pavón, P., Pinto, C. G., &
seen widespread use. One solution might lay in miniaturization and Cordero, B. M. (2001). Electronic nose based on metal oxide semiconductor sensors
aiming towards the lab-on-a-chip concept, which might significantly and pattern recognition techniques: Characterisation of vegetable oils. Analytica
reduce the unit cost. Chimica Acta, 449, 69–80.
González Martín, Y., Pérez Pavón, L. J., Moreno Cordero, B., & García Pinto, C. (1999).
It can be concluded that the electronic nose technique shows great
Classification of vegetable oils by linear discriminant analysis of electronic nose data.
promise in food analysis, especially when developed towards particular Analytica Chimica Acta, 384(1), 83–94.
applications such as evaluation of the quality of edible oils. It can po- Guadarrama, A., Rodrıguez-Mendez, M. L., De Saja, J. A., Rıos, J. L., & Olıas, J. M. (2000).
tentially be commonly used by institutions tasked with food quality Array of sensors based on conducting polymers for the quality control of the aroma of
the virgin olive oil. Sensors and Actuators B, 69, 276–282.
control, manufacturers and even consumers for whom the quality and Guadarrama, A., Rodríguez-Méndez, M. L., Sanz, C., Ríos, J. L., & De Saja, J. A. (2001).
wholesomeness of consumed food is of paramount importance. Electronic nose based on conducting polymers for the quality control of the olive oil
aroma – Discrimination of quality, variety of olive and geographic origin. Analytica
Chimica Acta, 432, 283–292.
Appendix A. Supplementary data Haddi, Z., Alami, H., Bari, N. El, Tounsi, M., Barhoumi, H., Maaref, A., ... Bouchikhi, B.
(2013). Electronic nose and tongue combination for improved classification of
Supplementary data associated with this article can be found, in the Moroccan virgin olive oil profiles. Food Research International, 54, 1488–1498.
Haddi, Z., Amari, A., Ali, A. O., Bari, N. El, Barhoumi, H., Maaref, A., ... Bouchikhi, B.
online version, at http://dx.doi.org/10.1016/j.foodchem.2017.11.013. (2011). Discrimination and identification of geographical origin virgin olive oil by an
e-nose based on MOS sensors and pattern recognition techniques. Procedia
References Engineering, 25, 1137–1140.
Hai, Z., & Wang, J. (2006a). Detection of adulteration in camellia seed oil and sesame oil
using an electronic nose. European Journal of Lipid Science and Technology, 108(2),
Amann, A., Costello, B. de L., Miekisch, W., Schubert, J., Buszewski, B., Pleil, J., ... Risby, 116–124.
T. (2014). The human volatilome: Volatile organic compounds (VOCs) in exhaled Hai, Z., & Wang, J. (2006b). Electronic nose and data analysis for detection of maize oil
breath, skin emanations, urine, feces and saliva. Journal of Breath Research, 8(3), adulteration in sesame oil. Sensors and Actuators B: Chemical, 119(2), 449–455.
34001. Hong, E. J., Park, S. J., Choi, J. Y., & Noh, B. S. (2011). Discrimination of palm olein oil
AOCS (1996). Official and Tentative Methods of the American Oil Chemists’ Society. and palm stearin oil mixtures using a mass spectrometry based electronic nose. Food
Method Ca 5a–40 (5th ed.). Champaign. Science and Biotechnology, 20(3), 809–816.
Aparicio, R., Rocha, S. M., Delgadillo, I., & Morales, M. T. (2000). Detection of rancid Huang, J. Y., Guo, X. P., Qiu, Y. B., & Chen, Z. Y. (2007). Cluster and discriminant analysis
defect in virgin olive oil by the electronic nose. Journal of Agricultural and Food of electrochemical noise data. Electrochimica Acta, 53(2), 680–687.
Chemistry, 48, 853–860. IOC (2005). Method for the Organoleptic Assessment of Extra Virgin Olive Oil Applying
Boser, B. E., Guyon, I. M., & Vapnik, V. N. (1992). A training algorithm for optimal margin To Use a Designation of Origin, (22), 1–29.
classifiers. In Proceedings of the fifth annual workshop on Computational learning ISO 3960:1977 (1977). Animal and vegetable oils and fats - Determination of peroxide
theory – COLT ’92 (pp. 144–152). New York, New York, USA: ACM Press. value.
Bougrini, M., Tahri, K., Haddi, Z., Saidi, T., El Bari, N., Bouchikhi, B., ... Bouchikhi, B. James, D., Scott, S. M., O’Hare, W. T., Ali, Z., & Rowell, F. J. (2004). Classification of fresh
(2014). Detection of adulteration in argan oil by using an electronic nose and a edible oils using a coated piezoelectric sensor array-based electronic nose with soft
voltammetric electronic tongue. Journal of Sensors, 2014, 1–10. computing approach for pattern recognition. Transactions of the Institute of
Buratti, S., Benedetti, S., & Cosio, M. S. (2005). An electronic nose to evaluate olive oil Measurement and Control, 26(1), 3–18.
oxidation during storage. Italian Journal of Food Science, 17(2), 203–210. Lerma-García, M. J., Cerretani, L., Cevoli, C., Simó-Alfonso, E. F., Bendini, A., & Toschi, T.
Casale, M., Casolino, C., Oliveri, P., & Forina, M. (2010). The potential of coupling in- G. (2010). Use of electronic nose to determine defect percentage in oils. Comparison
formation using three analytical techniques for identifying the geographical origin of with sensory panel results. Sensors and Actuators B: Chemical, 147(1), 283–289.
Liguria extra virgin olive oil. Food Chemistry, 118(1), 163–170. Majchrzak, T., Lubinska, M., Różańska, A., Dymerski, T., Gębicki, J., & Namieśnik, J.
Cawley, J., Meyerhoefer, C., Gillingham, L. G., Kris-Etherton, P., & Jones, P. J. H. (2017). (2017). Thermal degradation assessment of canola and olive oil using ultra-fast gas
Estimates of the direct and indirect cost savings associated with heart disease that chromatography coupled with chemometrics. Monatshefte Fur Chemie, 148,
could be avoided through dietary change in the United States. Journal of Medical 1625–1630.
Economics, 20(2), 182–192. Man, Y. B. C., Gan, H. L., NorAini, I., Nazimah, S. A. H., & Tan, C. P. (2005). Detection of
Cimato, A., Dello Monaco, D., Distante, C., Epifani, M., Siciliano, P., Taurino, A. M., ... lard adulteration in RBD palm olein using an electronic nose. Food Chemistry, 90(4),
Sani, G. (2006). Analysis of single-cultivar extra virgin olive oils by means of an 829–835.
electronic nose and HS-SPME/GC/MS methods. Sensors and Actuators B: Chemical, Mariani, C., Bellan, G., Lestini, E., & Aparicio, R. (2006). The detection of the presence of
114(2), 674–680. hazelnut oil in olive oil by free and esterified sterols. European Food Research and
Concepción, O. C., Pérez Pavón, L. J., García Pinto, C., Fernández Laespada, E., Moreno Technology, 223(5), 655–661.
Cordero, B., & Forina, M. (2002). Electronic nose based on metal oxide semi- Marina, A. M., Che Man, Y. B., & Amin, I. (2010). Use of the SAW sensor electronic nose
conductor sensors as a fast alternative for the detection of adulteration of virgin olive for detecting the adulteration of virgin coconut oil with RBD Palm Kernel Olein.
oils. Analytica Chimica Acta, 459, 219–228. Journal of the American Oil Chemists’ Society, 87(3), 263–270.

200
T. Majchrzak et al. Food Chemistry 246 (2018) 192–201

Martı́, M. P., Busto, O., Guasch, J., & Boqué, R. (2005). Electronic noses in the quality Stella, R., Barisci, J. N., Serra, G., Wallace, G. G., & De Rossi, D. (2000). Characterisation
control of alcoholic beverages. TrAC Trends in Analytical Chemistry, 24(1), 57–66. of olive oil by an electronic nose based on conducting polymer sensors. Sensors and
Melucci, D., Bendini, A., Tesini, F., Barbieri, S., Zappi, A., Vichi, S., ... Gallina Toschi, T. Actuators B: Chemical, 63(1), 1–9.
(2016). Rapid direct analysis to discriminate geographic origin of extra virgin olive Stevenson, S. G., Vaisey-Genser, M., & Eskin, N. A. M. (1984). Quality control in the use of
oils by flash gas chromatography electronic nose and chemometrics. Food Chemistry, deep frying oils. Journal of the American Oil Chemists Society, 61(6), 1102–1108.
204, 263–273. The Commission of the European Communities (2002). Commission Regulation (EC) No
Mielle, P., Marquis, F., & Latrasse, C. (2000). Electronic noses: Specify or disappear. 1019/2002.
Sensors and Actuators B: Chemical, 69(3), 287–294. The Commission of the European Communities (2009). Commission Regulation (EC) No
Mildner-Szkudlarz, S., & Jeleń, H. H. (2008). The potential of different techniques for 182/2009.
volatile compounds analysis coupled with PCA for the detection of the adulteration of The European Commission (2013). Commission Implementing Regulation (EU) No 1335/
olive oil with hazelnut oil. Food Chemistry, 110(3), 751–761. 2013.
Mildner-Szkudlarz, S., & Jeleń, H. H. (2010). Detection of olive oil adulteration with Upadhyay, R., Sehwag, S., & Mishra, H. N. (2017a). Electronic nose guided determination
rapeseed and sunflower oils using MOS electronic nose and SPME-MS. Journal of Food of frying disposal time of sunflower oil using fuzzy logic analysis. Food Chemistry,
Quality, 33(1), 21–41. 221, 379–385.
Mildner-Szkudlarz, S., Jeleń, H. H., & Zawirska-Wojtasiak, R. (2008). The use of elec- Upadhyay, R., Sehwag, S., & Mishra, H. N. (2017b). Frying disposal time of sunflower oil
tronic and human nose for monitoring rapeseed oil autoxidation. European Journal of using hybrid electronic nose-fuzzy logic approach. LWT – Food Science and
Lipid Science and Technology, 110(1), 61–72. Technology, 78, 332–339.
Monfreda, M., Gobbi, L., & Grippa, A. (2012). Blends of olive oil and sunflower oil: Wiśniewska, P., Dymerski, T., Wardencki, W., & Namieśnik, J. (2015). Chemical com-
Characterisation and olive oil quantification using fatty acid composition and che- position analysis and authentication of whisky. Journal of the Science of Food and
mometric tools. Food Chemistry, 134(4), 2283–2290. Agriculture, 95(11), 2159–2166.
Morales, M. T. T., Luna, G., & Aparicio, R. (2005). Comparative study of virgin olive oil Wojnowski, W., Majchrzak, T., Dymerski, T., Gębicki, J., & Namieśnik, J. (2017).
sensory defects. Food Chemistry, 91(2), 293–301. Electronic noses: Powerful tools in meat quality assessment. Meat Science, 131,
Nölscher, A. C., Sinha, V., Bockisch, S., Kï Upfel, T., & Williams, J. (2012). Total OH 119–131.
reactivity measurements using a new fast gas chromatographic photo-ionization de- Xiao, Z., Yu, D., Niu, Y., Chen, F., Song, S., Zhu, J., & Zhu, G. (2014). Characterization of
tector (GC-PID). Atmospheric Measurement Techniques, 5, 2981–2992. aroma compounds of Chinese famous liquors by gas chromatography–mass spectro-
OECD-FAO (2016). OECD-FAO Agricultural Outlook 2016–2025: Special Focus: Sub- metry and flash GC electronic-nose. Journal of Chromatography B, 945, 92–100.
Saharan Africa. Xu, L., Yu, X., Liu, L., & Zhang, R. (2016). A novel method for qualitative analysis of
Park, M. H., Jeong, M. K., Yeo, J., Son, H.-J., Lim, C.-L., Hong, E. J., ... Lee, J. (2011). edible oil oxidation using an electronic nose. Food Chemistry, 202, 229–235.
Application of solid phase-microextraction (SPME) and electronic nose techniques to Zabaras, D., & Gordon, M. (2004). Detection of pressed hazelnut oil in virgin olive oil by
differentiate volatiles of sesame oils prepared with diverse roasting conditions. analysis of polar components: Improvement and validation of the method. Food
Journal of Food Science, 76(1), C80–C88. Chemistry, 84(3), 475–483.
Persaud, K., & Dodd, G. (1982). Analysis of discrimination mechanisms in the mammalian
olfactory system using a model nose. Nature, 299(5881), 352–355. Tomasz Majchrzak was born in 1992 in Elbląg, Poland. He is currently pursuing a PhD in
Posada de la Paz, M., Philen, R. M., & Borda, A. I. (2001). Toxic oil syndrome: The chemical technology at the Department of Analytical Chemistry, Faculty of Chemistry,
perspective after 20 years. Epidemiologic Reviews, 23(2), 231–247. Gdańsk University of Technology. His research is focused on the use of new analytical
Shen, N., Moizuddin, S., Wilson, L., Duvicka, S., White, P., Pollak, L., ... Pollak, L. (2001). techniques, in particular two-dimensional gas chromatography, proton transfer reaction
Relationship of electronic nose analyses and sensory evaluation of vegetable oils mass spectrometry and electronic noses in the assessment of thermal degradation of ed-
during storage. JAOCS, 78(9), 937–940. ible oils. His scientific interests include sensor systems, gas chromatography, chemometry
Sonia, E., GianFrancesco, M., Roberto, S., Ibanez, R., Agnese, T., Stefania, U., & Maurizio, and mass spectrometry.
S. (2009). Monitoring of virgin olive oil volatile compounds evolution during olive
malaxation by an array of metal oxide sensors. Food Chemistry, 113(1), 345–350.

201