Measurement 103 (2017) 227–234

Contents lists available at ScienceDirect

Measurement
journal homepage: www.elsevier.com/locate/measurement

Application of electronic nose with MOS sensors to prediction of

rapeseed quality
Marek Gancarz a, Jolanta Wawrzyniak b, Marzena Gawrysiak-Witulska b, Dariusz Wia˛cek a,
Agnieszka Nawrocka a, Marcin Tadla a, Robert Rusinek a,⇑
a
Institute of Agrophysics Polish Academy of Sciences, Doświadczalna 4, 20-290 Lublin, Poland
b
Institute of Food Technology of Plant Origin, Faculty of Food Science and Nutrition, Poznań University of Life Science, Wojska Polskiego 28, 60-637 Poznań, Poland

a r t i c l e i n f o a b s t r a c t

Article history: Rapeseed is one of the main raw materials regarded as a source of edible oil for humans. Stored in order to
Received 28 November 2016 maintain continuity of production can undergo deterioration of the quality. Therefore, there is a need for
Received in revised form 10 January 2017 rapid methods for the assessment of its quality. The quality of rapeseed during 31 days of storage was
Accepted 24 February 2017
studied. Colony Forming Unit (CFU), Ergosterol content (ERG), Fourier Transform Infrared Spectroscopy
Available online 27 February 2017
(FT-IR), and Volatile Organic Compounds (VOCs) were examined using an electronic nose (Agrinose).
The electronic nose was built of 8 metal-oxide semiconductor sensors (type MOS) and one representing
Keywords:
the Micro Electro Mechanical Systems technology (MEMS). Principal Component Analysis (PCA), as a
Metal-oxide semiconductor sensors
Rapeseed
method of data analysis, was applied to the visualisation rapeseed groups of different quality. An analysis
Electronic nose of sensorgrams (sensor drift) with a strong signal was performed. Six from the eight sensors gave a clear
Colony forming units response to spoiled rapeseed VOCs. The results have shown a correlation between microbiological and
Ergosterol content chemical methods for assessment of quality with responses of electrochemical sensors.
Fourier transform infrared spectroscopy Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction assessment of microbiological contamination of agricultural and

Rapeseed is one of the most important oilseed materials world-
wide [1]. Currently, rapeseed accounts for more than 10% of the
global production of oil-bearing plants. Material stored in silos
and warehouses is at constant risk of deterioration, therefore its
quality should be continuously monitored [2–4].
Generally, microbiological and chemical methods for assess-
ment of the level of contamination of rapeseed are widely used.
Microbiological methods, providing an answer to the question of
the amounts and types of fungal microflora, are laborious and time
consuming. Chemical (indirect) methods for detection and quan-
tification of fungal infection based on assessment of chemical
markers, i.e. ergosterol, chitin, and other cell wall components,
require time-consuming and tedious preparation of samples [5–
9]. Another way to detect spoiled seeds can be analysis of volatile
substances that are produced by fungi, e.g. Aspergillus, Penicillium
and Fusarium. These fungi produces 1-octen-3-ol, 2-octen-1-ol, 3-
octanol, 1-octanol and 3-octanone. These substances form about
67–97% of volatile fraction in wheat and corn. Therefore, there is
a need for development of new techniques and methods for
⇑ Corresponding author.
E-mail address: r.rusinek@ipan.lublin.pl (R. Rusinek).
food products. One of these is the analysis of volatile organic com-
pounds with the use of gas sensors in the form of a device known
as the electronic nose (enose) [10,11]. The term ‘‘electronic nose”
denoting an instrument composed of a sensor set and able to
recognise simple and complex odours was coined in 1988 by Gard-
ner and Bartlett, who defined it as ‘‘a device comprising a set of
electrical and chemical sensors selectively responsive to a variety
of volatile compounds and an appropriate system of recognition
of response forms capable of recognising simple or complex
aromas” [12]. Since then, the development of the production tech-
nology of electronic transducers and techniques for analysis of
large data sets allowed an increasing use of analysis of volatile
compounds in many areas of life, in particular in medical diagnos-
tics [13], food industry [14–16] and environmental protection [17].
Chemically sensitive sensors of volatile compounds are devices
that operate on the basis of changes in electrical values (e.g. elec-
tric current, resistance, impedance, etc.), which are caused by vola-
tile substances. The most common group of electrochemical
sensors includes resistive gas sensors, which are composed of a
receptor-transducer element, base, electrodes, heater, and optional
filters. The most important element in this type of transducers is
the receptor-transducer element. Most frequently, it is a multi-
phase system of solids, in which one phase is always a chemically

http://dx.doi.org/10.1016/j.measurement.2017.02.042
0263-2241/Ó 2017 Elsevier Ltd. All rights reserved.
228 M. Gancarz et al. / Measurement 103 (2017) 227–234

Fig. 1. Scheme of the experimental set-up.

Table 1
Technical data of Agrinose sensors.

Type Description Detecting Response

range (ppm)
TGS2600 – B00 General air contaminants, hydrogen and carbon monoxide 1–3 (H2) Yes
TGS2610 – C00 LP gas, butane 500–10,000 Yes
TGS2602 – B00 Ammonia, Hydrogen sulfide (high sensitivity to VOC and odorous gases) 1–30 (EtOH) Yes
TGS2611 – C00 Natural gas, methane 500–10,000 Yes
TGS2611 – E00 Natural gas, methane (carbon filter) 500–10,000 Low response
TGS2612 – D00 Propane, methane, solvent vapours, 1–25% LEL Low response
iso-butane (carbon filter)
TGS2620 – C00 Solvent vapours, volatile vapours, alcohol 50–5000 Yes
AS – MLV – P2 CO, butane, methane, ethanol, hydrogen. Specifically designed for 10–10,000 Yes
volatile organic compounds (VOCs)

sensitive material whose electrical conductivity depends on the
structure of analysed gases. The mechanism of action of resistive
sensors of volatile compounds has not been fully elucidated, but
it depends on the active sites of the receptor-transducer element,
diffusion, adsorption, phenomena occurring on the surface of the
semiconductor, electrical conductivity, chemical reactions, and
catalysis [18]. The response of sensors to volatile organic com-
pounds is a resultant of complex chemical processes. It is based
on the reaction of the receptor-transducer element to oxygen
chemisorption, which increases the resistance of this element. This
value changes, especially when the sensor is surrounded by air
containing volatile organic compounds, which react with the
adsorbed oxygen.
A series of preliminary tests in order to select the most promis-
ing sensors have been done. The most important features are: high
sensitivity to VOCs comparing to e.g. conducting polymer sensor,
simplicity of operation and measurement stability. The authors
have attempted to use chemically sensitive sensors such as the
metal oxide semiconductor (commonly called MOS, one in the
MEMS technology) to determine the changes in the quality of rape-
seed during 31 days of storage. One month period of storage should
demonstrate the usefulness of the electronic nose to the early
detection of changes in the quality of rapeseed [19].
2. Material and methods
2.1. An experimental set – electronic nose
The examined material was winter rapeseed variety Sensation
(Brassica napus L. (partim)) harvested in Poland and provided by
the Institute of Agrophysics PAS, Poland. Rapeseed samples with
moisture content of 12.5% (w.b.) were stored at a stable tempera-
ture (20 °C) and humidity (81% RH) in three special glass chambers
(Fig. 1). The three chambers provide a sufficient amount of volatile
substances from the headspace to be analysed.
Agrinose (Food Volatile Compound Analyser) has been designed
and built in the Institute of Agrophysics, PAS, Poland. The device
consists of eight metal-oxide semiconductor (MOS) sensors [20],
which were selected with the following criteria: lower power con-
sumption, a similar type and geometry of the sensors in the array,
low susceptibility to humidity and temperature, and common use
of the gas sensors tested in a similar measurement device. As
ensured by the producers, the sensors strongly respond to the pres-
ence of ketones, fatty acids, esters, and alcohols that can be
expected in fungal metabolites in rapeseed spoilage [7,21]. Table 1
presents the types and technical data of Agrinose sensors. Seven of
them (TGS type) were produced by Figaro Engineering (Japan) and
one by Ams (USA). A measurement cycle according to the sample
protocol consisted of 10 s baseline purge, 40 s sample draw-in,
5 s laboratory air purge and 140 s sample purge. Analog signals
were converted to digital signals by means of software DasyLab.
Obtained sensorgrams were converted to the ⁄.xls format and anal-
ysed using software Statistica (version 12.0, StatSoft Inc., USA).
2.2. Ergosterol analysis
The ergosterol analysis was performed according to the method
of Abramson and Smith [5] and Pronyk et al. [18]. Briefly, 5 g-
rapeseeds samples were milled, extracted in 50 ml of methanol,
cooled, filtered and saponificated for 30 min. After saponification,
the obtained samples were cooled and filtered once again. Next,
the mixture was extracted twice with 50 ml of n-hexane. 12.5 ml
of the n-hexane extract was taken to dryness under a nitrogen
stream at 40 °C. The obtained residue was dissolved in 2 ml of
cyclohexane and added to a 3-ml SPE (solid phase extraction) car-
tridge containing 500 mg silica gel (Strat Silica, Phenomenex) and
rinsed with cyclohexane. To obtain ergosterol acetate acetylation
by acetic anhydride was required. After acetylation the eluate
was taken to dryness. For HPLC analysis, the purified ergosterol
acetate was dissolved in 1 ml of acetonitrile. The 10 ml-aliquots
 M. Gancarz et al. / Measurement 103 (2017) 227–234 229

Fig. 2. Dependence of colony forming units (CFU) - (a) and ergosterol content - (b) on the storage time.

were injected into an HPLC system (Acella, Thermo Scientific, USA)
with a C-18 reverse-phase column (Hypersil Gold, 50  2.1 mm,
1.9 mm particles, Thermo Scientifc, USA) at temperature 35 °C. A
diode-array spectrophotometric detector was set at 282 nm.
Methanol at 450 ml/min was used as the mobile phase. The ergos-
terol acetate was eluted at 2.15 min.
2.3. Colony Forming Units (CFU)
The level of mould contamination of the rapeseed samples was
determined as Colony Forming Units (CFU) of moulds per 1 g of
grain. Analysis of CFUs in the rapeseed was performed in accor-
dance with the procedure described in the standard
PN-ISO 21527-2:2009 [23].
2.4. FT-IR spectra measurements and data manipulation
To acquire Fourier transform infrared (FT-IR) spectra were used
a Nicolet 6700 FT-IR spectrometer (Thermo Scientific, USA)
equipped with a diamond ATR attachment. The FT-IR spectra were
recorded over the range of 4000 and 400 cm 1 with resolution of
4 cm 1. Each spectrum resulted from 128 scans to obtain an opti-
mal signal-to-noise ratio. The FT-IR spectrum of each sample was
an average of five registered spectra.
230 M. Gancarz et al. / Measurement 103 (2017) 227–234

Fig. 3. Difference spectra of rapeseed samples stored for 5, 15, and 25 days.

Fig. 4. The maximum response (DR/Rmax) to VOCs of sensors AS-MLV-P2 and TGS 2602 – (a), TGS 2600 and 2620 – (b), and TGS 2610 and TGS 2611 – (c) during 31 days of
storage.

A linear baseline was used to correct each spectrum in the
OMNIC software (v.8.2, Thermo Fischer Scientific Inc., USA). The
averaged spectra were normalized against the highest band located
at 1743 cm 1 using ORIGIN (v.9.0 PRO, OriginLab Corp., USA).This
band is connected with stretching of the ester C@O group in oil
triglycerides [24] and is not observed in the spectra of ergosterol
(see Fig. S1 in the Supplementary Material) and cell wall polysac-
charides (chitin [25] and glucans [26]). To determine changes in
the FTIR spectrum connected with the fungal development during
the rapeseed storage, the difference spectra were calculated.
Briefly, normalized spectra of the control sample (see Fig. S1 in
the Supplementary Material) were subtracted from the spectra of
 M. Gancarz et al. / Measurement 103 (2017) 227–234
the stored rapeseed. The bands observed in the difference spectra
can be assigned to the infrared bands of ergosterol and/or fungal
polysaccharides (glucans, chitin, chitosan).
2.5. Statistical analysis
Analysis of variance, simple correlations, and Principal Compo-
nent Analysis (PCA) were performed at a significance level of
a = 0.05 using software Statistica (version 12.0, StatSoft Inc.,
USA). The PCA analysis was used to determine the relationships
between CFU, ERG, and sensor response DR/Rmax for six sensors
used in this study. To the statistical analysis were taken standard-
ized response DR/Rmax from six replicates. The optimum number of
principal components obtained in the analysis was established on
the basis of Cattel’s criterion. A data matrix with 32 rows and 8 col-
umns was built for overcoming the ability of the Agrinose to distin-
guish of the seed degradation. The input matrix was auto-scaled.
3. Results and discussion
3.1. Microbiological and chemical analysis
Fig. 2 present respectively (a) colony forming units and (b)
changes in the ergosterol content during 31 days of storage. Given
the type of fungi, three phases of the storage were observed: I –
fungi characteristic for field microflora: Fusarium, Alternaria, Cla-
dosporium, Acremonium, and Scopulariopsis, II – fungi mainly char-
acteristic for field microflora: Fusarium, Alternaria, Cladosporium,
Acremonium, Scopulariopsis, and the beginning of growth of storage
microflora: Penicillium and Aspergillus, and III – storage microflora:
Penicillium, Eurotium ssp., and Aspergillus ssp. (candidis, ochraceus,
fumigatus, niger). During first stage of storage over the first ten
days, there was a slight decrease in the CFU of moulds associated
with their adaptation to new environmental conditions [27]. In
phase II – the CFU of fungi started to increase substantially and,
next, during the last phase III – systematically increased up to
the end of experiment. The levels of ERG increased during all the
days of the experiment. The ergosterol content increased more
than 3 lg/g on the fourth day of storage and more than 6 lg/g
on the ninth day. Ergosterol content over 3 lg/g is a concentration
of ergosterol that should not be found in cereal grains intended for
human consumption. The ergosterol concentration of 6 lg/g is
indicated as the upper limit of ergosterol concentration in good
quality seeds [28]. Detailed description of the fungal species for
the whole storage period is provided in Table S1 in the Supplemen-
tary Materials.
Ergosterol, which is a component of the fungal cell membrane,
can also be detected using FT-IR spectroscopy. This technique can
also be used to detect polysaccharides of the fungal cell walls e.g.
glucans, chitin, and chitosan [29]. It should be noted that the anal-
ysis of the content of the fungal structural compounds from the FT-
IR spectra is qualitative rather than quantitative. Specific IR bands
connected with these structural compounds are used to determine
their presence/absence in the stored rapeseed samples.
The FT-IR spectrum of the control sample is presented in Fig. S1
in the Supplementary Materials. A few characteristic bands for
rapeseed were observed. The bands at 1155, 1377, 1459, 1743,
2854, and 2923 cm 1 could be assigned to CAO stretching, CH2
bending, ester C@O stretching, and aliphatic CH2 stretching in
lipids, respectively. These bands were also observed in the FT-IR
spectra of edible oils [30]. Bands corresponding to the oscillations
of the peptide bond were located at 1237 (amide III), 1545 (amide
II), and 1648 cm 1 (amide I). The spectral region 3000–3600 cm 1
was connected with OAH and NAH stretching, while the band at
231
1052 cm 1 can be assigned to CAOAC ring vibrations in carbohy-
drates [31].
The difference spectra were calculated to determine changes in
the FT-IR spectrum connected with the fungal development during
the rapeseed storage. In accordance with the microbiological data,
the storage time was divided into three periods differing in the
type of fungal species (see Fig. 2 and Table S1 in the Supplementary
Materials). The difference spectra of rapeseed samples stored for 5,
15, and 25 days are presented in Fig. 3. A few characteristic bands
located at 1035, 1240, 1545, 1639, 2932, and 3281 cm 1 were
observed in all the difference spectra. Only the band at
1545 cm 1 was not observed in the spectrum of a 5-day stored
sample. Comparison of the difference spectra with the spectrum
of the control sample (see Fig. S1 in the Supplementary Material)
showed that all spectra had three common bands located at ca.
1240, 1545, and 3281 cm 1. Therefore, it can be assumed that
the other bands were related to the fungal development. The inten-
sity of the bands at 1035, 1639, and 2932 cm 1 increased with the
storage time. The band at 1035 cm 1 can be connected with chitin
or glucan. The simultaneous presence of the bands at 1639 and
2932 cm 1 suggested that the band at 1035 cm 1 is connected
with glucan. These three bands were observed in the spectrum of
(1–3)-a-D-glucan extracted from Penicillium chrysongenum mycelia
by Wang et al. [26]. The intensity of the band at 1035 cm 1 is rel-
atively high, compared to the other bands in the difference spec-
trum; thus, this band can be also assigned to chitin. Moreover,
the presence of the band at 1639 cm 1 could confirm this state-
ment. Krishnaveni & Ragunathan [25] observed such bands in the
FT-IR spectrum of chitin extracted from Aspergillus terreus. The
band at 1037 cm 1 was also observed in the spectrum of ergos-
terol, which is presented in Fig. S2 in the Supplementary Materials.
As seen, the ergostrol spectrum had a few characteristic high-
intensity bands at 965, 1037, 1067, 1368, 1457, 2868, and
2952 cm 1. The band at 1037 cm 1 is rather connected with the
fungal polysaccharides than with ergosterol as suggested by the
medium intensity of this band in the ergosterol spectrum, lack of
the highest band at 2952 cm 1 in the difference spectra, and the
small amount of ergosterol (1–30 mg/g) determined in the rapeseed
samples. Furthermore, the FT-IR spectrum of pure ergosterol could
differ from that embedded in the cell membrane. Gagoś & Arc-
zewska [32] studied the FTIR spectra of DPPC-ergosterol (1:1)
monolayers and no bands characteristic for pure ergosterol were
observed. All observed bands in the monolayers were characteris-
tic for DPPC. Thus, it can be said that the IR bands characteristic for
ergosterol may be covered by more intensive bands of other chem-
ical compounds (e.g. lipids, polysaccharides). The results obtained
from the FT-IR spectroscopy showed that this technique can be
appropriate for fungal detection in stored agricultural commodities
by using IR bands connected with fungal polysaccharides located at
1035, 1639 and 2923 cm 1.
3.2. Electronic nose results
3.2.1. Analysis of the sensor response (Rmax of sensorgram)
Sensor response is a curve called a sensor drift or a sensorgram.
Generally, it consists of two phases, i.e. adsorption and desorption.
In this work, the authors analysed the sensorgrams of several best-
responsive MOS sensors of VOCs from spoiled rapeseed. Several
factors can be analysed: the baseline (with drift or stable) [33],
association curve, steady state (horizontal curve, if obtained), or
the dissociation [34]. For this study, the authors analysed the
steady-state or, if existed, the maximum response (DR/Rmax) of this
phase. The results of the investigation are shown in Fig. 4a–c.
TGS2611-E00 and TGS 2612-D00 did not respond to VOCs.
TGS2611-E00 was built for detection of methane and natural gases
and is equipped with a carbon filter; TGS 2612-D00 was built for
232 M. Gancarz et al. / Measurement 103 (2017) 227–234
Fig. 5. A plot of two principal component scores (a) and loadings (b) obtained for rapeseed during 31 days of storage.
detection of methane, propane, iso-butane, and solvent vapours.
The other six sensors gave a clear response to VOCs. This response
was grouped into three similar graphs which are indicated in
graphs. Day 4, when the ergosterol content increased more than
3 lg/g (good sample - ergosterol  3 lg/g and bad sample - ergos-
terol  3 lg/g) is highlighted in Fig. 4 as a dashed line.
It should be added that an ergosterol concentration of 3 lg/g is
the maximum concentration of ergosterol that can be found in cer-
eal grains intended for human consumption [35]. For rapeseed,
Pronyk et al. [22] suggested that levels of ergosterol content more
than 2 ppm is characteristic for spoiled material. The remaining
three periods connected with the types of the microflora are also
shown in Fig. 4. Higher responses of the sensors were obtained
for the first 11 days of the period characteristic for field microflora.
Sensors AS-MLV-P2, TGS 2602, TGS 2600, and TGS 2620 strongly
responded to the changes in CFU and ERG, whereas the response
of sensors TGS 2610 and TGS 2611 gradually increased. According
to the producers of the sensors (see Table 1), AS-MLV-P2 and TGS
2602 were designed for detection of VOCs, TGS 2600, and TGS 2620
for general air contaminants, alcohols, and solvent vapours. TGS
2610 and TGS 2611 were designed for detection of a few kinds of
gases such as: natural gas, methane, butane, and LP gas. Generally,
the responses of all sensors had a similar tendency. In the first per-
iod (I), the DR/Rmax values achieved the maximum, and then
decreased in the next two stages. A similar relationship between
VOCs (microbial volatile organic compounds) for five days of incu-
bation was observed by Stoppacher et al. [36] for Trichoderma atro-
viride and by Weikl et al. [32] for Alternaria alternata.
3.2.2. Principal component analysis of sensor response
PCA was used to describe the relationship between the ergos-
terol content, colony forming unit levels, and sensor response
[37]. The particular components PC1 and PC2 accounted for
72.04%, 18.01% of the variation, respectively (approximately 90%
of the variance). The PC score plot of PC1 against PC2 is shown in
Fig. 5.
The PCA plot presented in Fig. 5 shows four different groups of
data: control sample, I – field microflora (1–11 days of storage), II –
 M. Gancarz et al. / Measurement 103 (2017) 227–234
mainly fungi characteristic for field microflora and the beginning of
growth of storage microflora (12–19 days), and III – mainly storage
microflora. Fig. 5 shows that the control samples are situated on
the negative side of the PC1 axis with clear distance to other sam-
ples with an increased level of ERG and CFU. The responses of sen-
sors (DR/Rmax) obtained from 1 to 11 days (group I) and 12–19 days
(group II) form a group located on the positive side of the PC1 axis,
whereas sensor responses (DR/Rmax) obtained from 20 to 31 days
(group III) are on the negative side. Groups I and II are clustered
together, which could imply that this is mainly field microflora.
This result suggested that this type of microflora produce charac-
teristic volatile organic compounds and has different intensity than
that in group III characteristic for the storage microflora [34]. The
scores plot of PC 1 against PC 2 has shown in Fig. 5b reflected. Sen-
sor response to both parameters ERG and CFU are correlated neg-
atively to each other.
This is related to the fact that microorganisms colonising the
seed ecosystem form a complex network of interspecific and envi-
ronmental interactions. The products of microbial secondary meta-
bolism, i.e. VOCs, exert an effect on the interactions taking place in
a colonised ecological niche. Weikl et al. [38] observed that the
level of compounds emitted by Alternaria alternata and Fusarium
oxysporum fungi in a mixed culture was lower than that in the cul-
ture of single strains. Hung et al. [39] reported that volatile com-
pounds can serve as signalling molecules facilitating microbial
‘‘communication” with the ambient environment. Furthermore,
they can serve a function of attracting or repelling signals for other
organisms, thereby playing an important role in establishment and
regulation of symbiotic relationships. Volatile compounds are also
formed for e.g. adaptation purposes and as developmental inducers
during the reproduction process.
Both in the adaptation phase and in the intensive growth phase,
there was an increase in the concentration of volatile compounds,
the production of which is probably associated with activation of
the metabolic apparatus. In the analysed seed samples, fungi rep-
resenting the so-called storage microflora were increasingly
detected after storage day 12. The storage fungi dominated in the
final stage of seed storage in the analysed moisture-temperature
conditions. A decrease in the concentration of volatile compounds
was observed in this period, which may be associated with lower
interspecies competition and microflora aging, which is accompa-
nied by a slowdown of metabolic transformations. In their investi-
gations of volatile compounds produced by Alternaria alternata and
Fusarium oxysporum fungi, Weikl et al. [37] noted that the concen-
tration of volatile compounds increased in the initial stage of
growth of the analysed strains until it reached a certain maximum
level, which was followed by a decline in the intensity of produc-
tion of these compounds. The authors demonstrated that the aging
fungal colonies produced smaller amounts of these compounds. In
their study, the highest levels of emission of volatile compounds
produced by these culture strains grown as a pure culture were
noted after 7 days expressed per plate and on day 3 expressed
per colony surface area. In the case of the culture of both strains,
the maximum level of these compounds was detected at a later
time, i.e. after 14 incubation days, expressed per colony surface
area. A similar trend was reported by Stoppacher et al. [36] during
the growth of a Trichoderma atroviride strain incubated for 5 days
on PDA agar medium. In their study, the maximum level of the
analysed volatile compounds was noted after 30–60 h
(a-terpinene, a-phellandrene, b-phellandrene, a-bergamotene,
b-sesquiphellandrene, b-bisabolene) and after 78–102 h (3-
octanone, 1-octen-3-ol, g-pentyl-a-pyrone). In the present paper,
the maximum level of volatile compounds was obtained after ca.
10 days of seed storage, and a distinct decrease in the level was
noted from day 18–21. In this period, there was a decline in the
growth rate of the analysed fungal microflora population, which
233
would have achieved a stationary growth phase during the succes-
sive stage of growth [29].
4. Conclusions
During 31 days of storage ergosterol content increased, colony
forming unit decrease slightly over the first ten days, next system-
atically increased up to the end of experiment. The results obtained
from the FT-IR spectroscopy showed that this technique can be
supportive and appropriate for fungal detection in stored agricul-
tural commodities.
Agrinose built of metal oxide sensors was shown to be adequate
as a rapid, non-destructive, and inexpensive device for monitoring
the microbiological quality of rapeseed during the first over a
dozen days of storage. The results of PCA analysis have shown a
correlation between sensor response and the type of microflora.
Two sensors has got filter material in its housing which eliminates
the influence of alcohol in gaseous state. In this research these sen-
sors gave no significant response.
During storage, three stages of development microflora were
noted: (I) only field microflora, (II) mainly fungi characteristic for
field microflora and the beginning of growth of storage microflora,
and (III) storage microflora. In the first period (I), the values of the
DR/Rmax achieved the maximum for all the sensors used, and then
decreased in the next two stages.
The results of research encourage authors to continue work
with the electronic nose. The future work involves the use gas
chromatography to detect VOCs.
Acknowledgments
The research was supported by The National Centre for
Research and Development (NCBR), Poland, Grant No. PBS2/
A8/22/2013.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.measurement.
2017.02.042.
References
[1] A. Re˛kas, M. Wroniak, R. Rusinek, Influence of roasting pretreatment on high-
oleic rapeseed oil quality evaluated by analytical and sensory approaches, Int.
J. Food Sci. Technol. 50 (2015) 2208–2214.
[2] R. Rusinek, R. Kobyłka, Experimental study and discrete element method
modeling of temperature distributions in rapeseed stored in model bin, J.
Stored Prod. Res. 59 (2014) 254–259.
[3] M. Gawrysiak-Witulska, A. Siger, J. Wawrzyniak, M. Nogala-Kalucka, Changes
in tocochromanol content in seeds of Brassica napus L. during adverse
conditions of storage, J. Am. Oil Chem. Soc. 88 (2011) 1379–1385.
[4] M. Gawrysiak-Witulska, M. Rudzińska, J. Wawrzyniak, A. Siger, The effect of
temperature and moisture content of stored rapeseed on the phytosterol
degradation rate, J. Am. Oil Chem. Soc. (2012) 1673–1679.
[5] D. Abramson, D.M. Smith, Determination of ergosterol in canola (Brassica napus
L.) by liquid chromatograph, J. Stored Prod. Res. 39 (2003) 185–191.
[6] E. Magan, P. Evans, Volatiles as an indicator of fungal activity and
differentiation between species, and the potential use of electronic nose
technology for early detection of Grain spoilage, J. Stored Prod. Res. 36 (2000)
319–340.
[7] M. Nicolaisen, S. Supronien, L.K. Nielsen, I. Lazzaro, N.H. Spliid, A.F. Justesen,
Real-time PCR for quantification of eleven individual Fusarium species in
cereals, J. Microbiol. Methods. 76 (2009) 234–240.
[8] J. Wawrzyniak, A. Waśkiewicz, Ochratoxin A and citrinin production by
Penicillium verrucosum on cereal solid substrates, Food Addit. Contam. Part A
31 (2014) 139–148.
[9] M. Gawrysiak-Witulska, J. Wawrzyniak, A. Ryniecki, J. Perkowski, Relationship
of ergosterol content and fungal contamination and assessment of
technological quality of malting barley preserved in a metal silo using the
near-ambient method, J. Stored Prod. Res. 44 (2008) 360–365.
234 M. Gancarz et al. / Measurement 103 (2017) 227–234
[10] M. Rodriguez Mendez, Electronic Noses and Tongues in Food Science, in: A.
Kubiak (Ed.), Rapeseed Analysis by an Electronic Nose, Elsevier, 2016, pp. 93–
101 (Chapter 10).
[11] A. Kubiak, T. Wenzl, F. Ulberth, Evaluation of the quality of postharvest
rapeseed by means of an electronic nose, J. Sci. Food Agric. (2012) 2200–2206.
[12] A.D. Wilson, M. Bauietto, Applications and Advances in Electronic-Nose
Technologies, Sensors (Basel) 9 (7) (2009) 5099–5148.
[13] I.A. Casalinuovo, D. Di Pierro, M. Coletta, P. Di Fracesco, Application of
electronic nose for disease diagnosis and food spoilage detection, Sensors
(Basel) 6 (2006) 1428–1439.
[14] A. Loutfi, S. Coradeschi, G. Kumar Mani, P. Shankar, J. Bosco Balaguru
Rayappan, Electronic noses for food quality, A review, J. Food Eng. 144
(2015) 103–111.
[15] M. Peris, L. Escuder-Gilabert, A 21st century for food control: electronic noses,
Anal. Chim. Acta 638 (2009) 1–15.
[16] A. Sanaeifar, S.S. Mohtasebi, M. Ghasemi-Varnamkhasti, H. Ahmandi,
Application of MOS based electronic nose for the prediction of banana
quality properties, Measurement 82 (2016) 105–114.
[17] Ł. Guz, G. Łagód, K. Jaromin-Gleń, Z. Suchorab, H. Sobczuk, A. Bieganowski,
Application of gas sensor arrays in assessment of wastewater purification
effects, Sensors (Basel) 15 (2015) 1–21.
[18] A. Szczurek, B. Krawczyk, M. Maciejewska, VOCs classification based on the
committee of classifiers coupled with single sensor signals, Chemometr. Intell.
Lab. Syst. 125 (2013) 1–10.
[19] M. Gawrysiak-Witulska, A. Siger, R. Rusinek, Degradation of tocopherols
during rapeseed storage in simulated conditio of industrial silos, Int. Agrophys.
30 (2016) 39–45.
[20] V. Lippolis, M. Pascale, S. Cervellieri, A. Damascelli, A. Visconti, Screening of
deoxynivalenol contamination in durum wheat by MOS-based electronic nose
and identification of the relevant pattern of volatile compounds, Food Control
37 (2014) 263–271.
[21] H. Jeleń, E. Wa˛sowicz, Volatile fungal metabolites and their relation to the
spoilage of agricultural commodities, Food Rev. Int. 14 (4) (1998) 391–426.
[22] C. Pronyk, D. Abramson, W.E. Muir, N.D.G. White, Correlation of total
ergosterol levels in stored canola with fungal deterioration, J. Stored Prod.
Res. 42 (2006) 162–172.
[23] PN-ISO 21527-2:2009 Microbiology of food and feeding stuffs - Horizontal
method for the enumeration of yeasts and moulds - Part 2: Colony count
technique in products with water activity less than or equal to 0.95.
[24] M.D. Guillen, N. Cabo, Infrared spectroscopy in the study of edible oils and fats,
J. Sci. Food Agric. 75 (1997) 1–11.
[25] B. Krishnaveni, R. Ragunathan, Extraction and characterization of chitin and
chitosan from Aspergillus terreus sps, synthesis of their bionanocomposites and
study of their productive applications, J. Chem. Pharm. Res. 7 (2015) 115–132.
[26] T. Wang, T. Li, M. Wang, T. Tan, Integrative extraction of ergosterol, (1–3)-a-D-
glucan and chitosan from Penicillium chrysongenum mycelia, Chin. J. Chem.
Eng. 15 (2007) 725–729.
[27] J. Wawrzyniak, A. Ryniecki, M. Gawrysiak-Witulska, Kinetics of mould growth
in the stored barley ecosystem contaminated with Aspergillus westerdijkiae,
Penicillium viridicatum and Fusarium poae at 23–30 °C, J. Sci. Food Agric. 93
(2013) 895–901.
[28] P.F. Maupetit, B. Gatel, G. Cahagnier, M. Bothorel, B. Charlier, P. Collet, J.
Dauvillier, Laffiteau, G. Roux, Quantitative estimation of fungal infestation of
feedstuffs by determining ergosterol content, in: 44th Annual Meeting of
EAAP, 16–19 August, Aarhus, Denmark. (Abstr.), 1993.
[29] C. Levasseur-Garcia, Infrared spectroscopy applied to identification and
detection of microorganisms and their metabolites on cereals (corn, wheat,
and barley), Agricultural Science, Dr Godwin Aflakpui (Ed.), ISBN: 978-953-51-
0567-1, InTech, Available from: <http://www.intechopen.com/books/
agricultural-science/infrared-spectroscopy-applied-to-identification-and-
detection-of-microorganisms-and-their-metabolite>, 2012.
[30] N. Vlachos, Y. Skopelitis, M. Psaroudaki, V. Konstantinidou, A. Chatzilazarou, E.
Tegou, Applications of Fourier-transform-infrared spectroscopy to edible oils,
Anal. Chim. Acta 573–574 (2006) 459–465.
[31] D. Naumann, Infrared spectroscopy in microbiology, Encyclopedia of
Analytical Chemistry, 2006.
[32] M. Gagoś, M. Arczewska, FTIR spectroscopic study of molecular organization of
the antibiotic amphotericin B in aqueous solution and in DPPC lipid
monolayers containing the sterols cholesterol and ergosterol, Eur. Biophys. J.
41 (2012) 663–673.
[33] L. Zhang, Domain adaptation extreme learning machines for drift compensation
in E-nose systems, IEEE Trans. Instrum. Meas. 64 (2015) 1790–1801.
[34] SPRpages home. <http://www.sprpages.nl>, 2006-2016.
[35] S. Balasubramanian, S. Panigrahi, B. Kottapalli, C.E. Wolf-Hall, Evaluation of an
artificial olfactory system for grain quality discrimination, Food Sci. Technol.
40 (2007) 1815–1825.
[36] N. Stoppacher, B. Kluger, S. Zeilinger, R. Krska, R. Schuhmacher, Identification
and profiling of volatile metabolites of the biocontrol fungus Trichoderma
atroviride by HS SPME-MS, J. Microbiol. Methods 81 (2010) 187–193.
[37] L. Zhang, H. Nie, L. Dang, G. Li, Q. Ye, Ch. Kadri, Classification of multiple indoor
air contaminants by an electronic nose and a hybrid support vector machines,
Sensors Actuat. B: Chem. 174 (2012) 114–125.
[38] F. Weikl, A. Ghirardo, J.P. Schnitzler, K. Pritsch, Sesquiterpene emissions from
Alternaria alternate and Fusarium oxysporum: Effects of age, nutrient
availability, and co-cultivation, <www.nature.com/scientificreports>, 2016,
http://dx.doi.org/10.1038/srep22152.
[39] R. Hung, S. Lee, J.W. Bennett, Fungal organic compounds and their role in
ecosystem, Appl. Microbiol. Biotechnol. 99 (2015) 3395–3405.