Journal of Food Engineering 210 (2017) 42e49

Contents lists available at ScienceDirect

Journal of Food Engineering

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

Fluorescence spectroscopy as a non destructive method to predict

rheological characteristics of Tilsit cheese
Zhyldyzai Ozbekova, Asylbek Kulmyrzaev*
Department of Food Engineering, Kyrgyz-Turkish Manas University, Prospekt Mira, 56, 720038 Bishkek, Kyrgyzstan

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

Article history: The objective of this study was to investigate the potential of fluorescence spectroscopy to predict
Received 6 October 2016 rheological characteristics of semi-hard cheeses as yield stress (tL), flow stress (tF), storage modulus (G0 )
Received in revised form and loss modulus (G00 ) measured at linear-viscoelastic, yield stress and flow stress oscillation regions.
13 April 2017
Melting temperatures and chemical composition of the semi-hard cheeses were also predicted using
Accepted 20 April 2017
fluorescence spectra. Principal component analysis (PCA) and partial least squares regression (PLSR)
Available online 22 April 2017
were applied to the fluorescence spectra to extract information on the rheological properties, chemical
composition, and melting temperatures. tL and tF were predicted with R2 ¼ 0.90 from the vitamin A
Keywords:
Cheese
emission and excitation spectra, respectively. Melting temperatures, moisture, protein and fat contents
Oscillation rheology were predicted with R2 ¼ 0.98 from the vitamin A emission spectra. This study demonstrates that
Fluorescence spectroscopy fluorescence spectroscopy has potential for the accurate, non-destructive and rapid prediction of cheese
Multivariate statistics rheology at linear-viscoelastic, yield stress and flow stress oscillation regions simultaneously.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction
Cheese is used as a final product in the human diet, as well as an
important ingredient in various foods to form desirable texture,
taste, flavor and nutritional value. The use of cheese as an ingre-
dient is affected by complex physical, thermal and mechanical
processes (John, 2008; Subramanian et al., 2006). The rheological
properties of cheese undergo significant changes in relation to
temperature, duration and intensity of mechanical stress, method
of transportation as well as changing shear rate (O’Callaghan and
Guinee, 2004). Since the cheese is exposed to these different fac-
tors, the study of rheological behavior of cheeses in a wide range of
the shear stress, i.e. in linear visco-elastic (LVE), yield stress, and
flow stress regions becomes a task of critical importance.
In the LVE range small sinusoidal stresses or strains are applied
to a cheese sample at levels that do not cause significant irrevers-
ible changes to the cheese internal structure. The LVE range is
characterized by viscous (G00 ) and elastic (G0 ) moduli the values of
which remain unchanged throughout the LVE region indicating
that the microstructure of the cheese is undisturbed. Therefore the
* Corresponding author.
E-mail addresses: zhyldyzaiozbekova@gmail.com (Z. Ozbekova), kulmyrzaev@
yandex.ru (A. Kulmyrzaev).
LVE region is suitable for probing cheese structure and structure
development during different processes. The Yield Stress region
shows a significant change in the structure of the material. It also
indicates the start of plastic deformation when a breakdown occurs,
consequently, modulus decreases. The Flow Stress region can be
used to determine the crossing point of two modulus (G’ and G”) at
which gelling time and beginning of sample flow are determined
(Mezger, 2011). The melting temperature of cheese is also an
important quality parameter that characterizes the readiness for
implementation and transition of the product from one processing
step to another (Karoui et al., 2003).
The most common methods to determine the rheological
properties of cheeses are dynamic and transient tests (Venugopal
and Muthukumarappan, 2003). Textural and rheological charac-
teristics of cheeses can also be studied using compression
(Kulmyrzaev et al., 2005; Campanella et al., 1987; Buffa et al., 2001)
and extensional (Ak et al., 1993; Muliawan and Hatzikiriakos, 2007)
tests. Rheological properties of cheeses can be measured using
small amplitude oscillatory shear experiments (Joshi et al., 2004)
and tube viscometry techniques (Leach et al., 2003). Despite the
usefulness of these techniques, a major drawback is that they are
tedious, destructive, invasive, time-consuming and require highly
skilled operators (Strasburg and Ludescher, 1995; Purna et al.,
2005).
Fluorescence spectroscopy is a highly sensitive, rapid, non-

http://dx.doi.org/10.1016/j.jfoodeng.2017.04.023
0260-8774/© 2017 Elsevier Ltd. All rights reserved.
 Z. Ozbekova, A. Kulmyrzaev / Journal of Food Engineering 210 (2017) 42e49
destructive and easy to use an analytical technique that provides
information on the presence of fluorescent molecules (Lakowicz,
2006). Natural compounds that exhibit fluorescence in cheeses
are vitamin A and tryptophan residues, which provide specific in-
formation on the physical state of triglycerides, and protein
conformational changes (Lakowicz, 2006), respectively. For
instance, the melting temperature of fat in cheese has been deter-
mined from vitamin A fluorescence spectra recorded at different
temperatures (Karoui et al., 2003). Fluorescence spectroscopy has
been considered to monitor rheological parameters (Karoui and
Dufour, 2006), light-induced changes (Andersen et al., 2005),
quality and ripening of cheeses (Kulmyrzaev et al., 2005) and their
molecular structure and molecular changes throughout the
ripening (Kulmyrzaev et al., 2005; Karoui et al., 2007).
Studies conducted earlier assessed the potential of front face
fluorescence spectroscopy to predict rheological parameters of
cheeses exhibited in LVE region only, i.e. the storage modulus G0 ,
loss modulus G00 , and the loss factor tgd ¼ G”/G’ (Karoui et al., 2003).
Stress at fracture (sF), strain at fracture (εF), work to fracture (WF)
and modulus of deformability (E) of soft cheeses obtained con-
ducting uniaxial compression also well correlate with fluorescence
spectra (Kulmyrzaev et al., 2005). However, as mentioned earlier,
cheese as a functional ingredient can be subjected to mechanical
processing (mixing, extrusion, transportation in tubes etc.) at shear
stresses that can cause disruption of cheese structure and flow. LVE
range of viscoelastic materials (G’ ¼ const) is limited by the value of
strain (shear) gL (Mezger, 2011). gL is the strain value at which the
curve G0 begins to deviate from the LVE plateau value (G’ ¼ const). At
strain amplitudes higher than gL the structure of the sample
changes irreversibly. Thus, the limiting value gL and corresponding
to it the limiting value tL (yield stress) characterize the point at
which the structure of the sample is deformed irreversibly.
Therefore gL or tL is considered as a critical rheological parameter
that provides knowledge on the ability of viscoelastic materials to
resist applied external force (stress), i.e. the mechanical strength of
the internal structure of materials. Commonly yield stress tL is used
in industrial practice to characterize structure strength of visco-
elastic materials (Mezger, 2011). Another significant rheological
characteristic of viscoelastic materials determined by the oscilla-
tory rheology is the flow point gF or flow stress tF. The flow point
occurs as the crossover point G’ ¼ G” at which the internal structure
of the material is braking to such an extent causing the material to
flow. In industrial practice the flow stress tF is used in engineering
processing of viscoelastic materials at which high rate shear
deformation occurs (mixing, extrusion, pumping through chan-
nels). Therefore, tL (yield stress) and tF (flow stress) are significant
rheological characteristics of cheeses from both scientific and
practical point of view.
The objective of this study was to investigate the potential of
fluorescence spectroscopy to predict the rheological characteristics
of semi-hard cheeses such as tL (yield stress), tF (flow stress),
modulus G’, and loss modulus G” measured at three oscillation
(yield stress, flow stress and LVE) regions using amplitude sweep
oscillatory tests. Melting points measured using temperature
sweep tests and chemical compositions of the cheeses were also
predicted using fluorescence spectra of the cheeses. Principal
Component Analysis (PCA) and Partial Least Squares Regression
(PLSR) were applied to fluorescence spectra to extract information
on rheological, chemical, and melting point of semi-hard cheeses.
2. Materials and methods
2.1. Cheese samples
Three kinds of pre-packed semi-hard cheese Tilsit (n ¼ 3) with
43
20% (low-fat (LF), 45% (medium-fat (MF)) and 55% (high-fat (HF))
fat in dry matter (FDM) were obtained from a local supermarket.
Each cheese piece was cut into three sample portions and one was
used for rheological measurements, while second and third por-
tions were used in measuring chemical composition and fluores-
cence spectroscopy, respectively. The samples were identified,
vacuum packed and stored at 4  C before conducting experiments.
2.2. Chemical analysis
The moisture, fat and protein content of cheese samples were
measured. Moisture in cheese was determined by the oven drying
method at 130  C (AOAC International 948.12, 2000). Fat content
was determined using an extraction procedure with petroleum
ether in a Soxhlet apparatus (Distillation System Vapodest 20,
Germany) according to AOAC 920.125 (AOAC International, 2000).
Kjeldahl method (AOAC International 991.20, 2000) was applied to
measure protein content using an Extraction Unit EV6 All/16
(Gerhardt, Germany). All chemical analyses were carried out on
three replicates of each cheese sample and average values were
taken.
2.3. Rheological measurements
The cheese samples were sliced into thin disks (2 mm thick and
25 mm diameter) and stored at 4  C until analysis in plastic bags to
prevent dehydration. The rheological measurements were con-
ducted using an MCR 302 rheometer (Anton Paar, Graz, Austria)
with a parallel plate measuring system PP25 (Ø 25 mm), and the
Peltier temperature control unit (P-PTD 200/56). The results of the
rheological measurements were analyzed using Rheoplus/32 V3.61
(Anton Paar Germany GmbH, D-73760 Ostfildern) software.
Сheese exhibits both elastic (G0 ) and viscous (G00 ) behavior
depending on deformation conditions (shear rate) (Guinee, 2011).
Therefore, oscillatory tests were conducted to obtain additional
information on the elastic behavior of the cheese samples. Ampli-
tude sweep (AS) tests were performed at fixed angular frequency
u ¼ 10 rad/s and strain values varied as 0.01%-100%. Temperature
was maintained at 25, 30, 35, 40, 45, 50, 55, 60, 65 and 70  C.
Oscillation tests were conducted in yield stress, flow stress, and LVE
ranges and G’, G” and t (yield stress) were measured. The oscillatory
tests were performed in triplicate at a given temperature and
average values of the measured characteristics were obtained.
Melting temperatures of the cheese samples were determined
using temperature sweep (TS) test. The angular frequency (u) and
strain (g) were set constant as 10 rad/s and 1%, respectively. The
temperature ramp was 20e100  C, in a heating rate of 1  C per
3 min. Tests were performed in triplicate for each sample and the
average values were taken.
2.4. Fluorescence spectroscopy
Fluorescence spectra were recorded using a Fluoromax-4 spec-
trofluorometer (Horiba Jobin Yvon, USA) provided with a single-
position (56 ) thermostatically controlled cell holder dedicated to
front-face fluorescence. The temperature of the cell holder was
controlled by a digital temperature controller (VWR, Model 1136D,
USA) set at 25, 30, 35, 40, 45, 50, 55, 60, 65 and 70  C. The cheese
samples were cut into rectangular bar shaped specimens with
1 cm  1 cm cross-section and 4.2 cm length in order to fit into a 5-
ml quartz cuvette. The cheese specimens were placed in the quartz
cuvette, transferred into the cell holder of the spectrofluorometer
and upon reaching the desirable temperature fluorescence spectra
were collected. Emission spectra of tryptophan residues
(305e480 nm, excitation: 290 nm) and vitamin A (340e620 nm,
44 Z. Ozbekova, A. Kulmyrzaev / Journal of Food Engineering 210 (2017) 42e49

excitation: 322 nm) in the cheese samples were recorded. Excita- 3. Results and discussion
tion spectra of vitamin A were also recorded in the wavelength
range of 250e350 nm with the emission wavelength set at 410 nm. 3.1. Chemical composition and melting temperature
Each spectrum was recorded in triplicate on different aliquots. In
total 270 spectra (3 types of spectra, 3 types of cheese, 10 tem- The differences in chemical composition lead to differences in
peratures, and 3 repetitions) were recorded. texture, rheological properties, and fluorescence spectra among the
cheese samples.
The mean results and standard deviations for moisture, fat,
2.5. Multivariate statistical analysis protein contents and melting temperatures of the cheese samples
are presented in Table 1. Chemical composition, especially fat and
The objective of the statistical processing was to derive relevant protein content, considerably affect the rheology/texture of cheeses
information from the fluorescence spectral data allowing predic- (Hort et al., 1997). It is assumed that milk proteins contribute to
tion of rheological characteristics, chemical composition, and firmness and milk fats influence smoothness of the cheese. The fat
melting point of semi-hard cheeses. content of the LF cheese was measured as 9.8% and its protein
In order to reduce scattering effects, the fluorescence spectra content was 29.1%, while the MF cheese contained 24.1% and 24.9%
were normalized by reducing the area under each spectrum to a of fat and protein, respectively (Table 1). The HF cheese contained
value of 1 according to Bertrand and Scotter (1992): 31.1% of fat and 24.3% of protein, while its moisture content was
40.3%.
ci ¼ Fi =norm (1) Melting points of the cheeses measured using temperature
sweep (TS) test are presented in Table 1. Temperature-sweep dy-
and namic shear profiles obtained with the LF, MF and HF cheeses are
shown in Fig. 1. The melting points of the MF cheese and HF cheese
vffiffiffiffiffiffiffiffiffiffiffiffiffi
uX were 52.18  C and 59.57  C, respectively (Table 1). The temperature
u n 2
norm ¼ t Fj (2) sweeps of the MF and HF cheese samples had a crossover of G0 and
j¼1 G00 at which solid-like behavior changes into liquid-like behavior
(Fig. 1). Increasing temperature induces melting of the cheese fat
where ci is the corrected value at wavelength i, Fi is the fluorescence entrapped in the protein network, and, in agreement with earlier
intensity at emission wavelength i, Fj is the fluorescence at wave- studies (Reparet and Noe €l, 2003; Tunick, 2010), G0 and G00 of the MF
length j, and n is the number of data points for each spectrum. Thus, and HF cheeses gradually decreased, G0 remaining over G00 until
mainly the shifts of the maximum emission and the width changes their cross point. Domination of elastic behavior over viscous one
of the spectra, retaining most of the cheese structural information, before crossover G0 and G” (Fig. 1) can be explained by the me-
were considered. Normalization of the fluorescence spectra was chanical resistance of the protein network while the cheese fat
conducted using a custom-designed algorithm written in MatLab melts progressively (Reparet and Noe €l, 2003). At the temperatures
(The MathWorks Inc., MA, USA). above the melting point the cheese fat becomes liquid and struc-
Rheological and chemical data tables were standardized to tural support of the fat globules to the protein network reduces. The
assure zero mean and unit variance of each column. The experi- melting point of the LF cheese was not derived due to the shapes of
mental data were arranged in six data tables: (1) emission spectra the TS curves (Fig. 1). The LF cheese was the only variety with the
of tryptophan, (2) emission spectra of vitamin A, (3) excitation elastic component G0 always higher than the viscous component
spectra of vitamin A, (4) rheological properties, (5) chemical G00 , resulting in values without any crossing-point over the entire
composition, and (6) melting point. temperature range (25e70  C) (Fig. 1). Moreover, the values of G0
Principal component analysis (PCA) was applied to the and G00 were the highest among all the studied cheeses. This
normalized fluorescence spectral data to obtain a map describing observation could be explained by the fact that the low fat content
physical and chemical variations between the samples studied. PCA of the LF cheese might reduce the melting of the cheese (Reparet
finds combinations of variables that describe major trends in the and Noe €l, 2003; Tunick et al., 1993).
data. Mathematically, PCA relies upon an eigenvector decomposi-
tion of the covariance or correlation matrix of the process variables.
PCA reduces the dimensionality of the data, having lost the least 3.2. Rheological characteristics
amount of information, and describes objects with the few prin-
cipal components (PC) that explain as much as possible of the total Table 2 presents the values of the storage modulus (G0 ) and the
variance contained in the original data without altering their native
structure (Monfreda, 2012). Table 1
Partial least squares regression (PLSR) was applied in order to Chemical compositions and melting temperatures of the experimental cheese.
predict rheological characteristics, chemical composition and
Cheese
melting point of the cheese samples from fluorescence data. PLSR
a b c
LF MF HF
searches the relationship and interdependence of two (X, Y) or
d
more (X, Y, Z etc.) random variables and describes their common Mean SD Mean SD Mean SD
structure. The accuracy of the regression is expressed with a cor- Moisture (%) 54.11 0.185 41.7 0.173 40.33 0.073
relation coefficient (R2). It allows establishing the extent of the Protein (%) 29.12 0.045 24.89 0.095 24.27 0.010
relationship, on which predict the value of a random variable from a Fat (%) 9.76 0.086 24.11 0.110 31.08 0.710
Melting point ( C) a a 52.18 0.11 59.57 0.09
large set of independent values (Dijkstra, 2010). In PLSR ‘‘leave-one-
out’’ cross-validation process was used for validation, that is, a: melting temperature cannot be derived due to the shape of the temperature
leaving one sample of the calibration set at a time for prediction. sweep curves.
a
LF: low fat.
The custom-designed versions of PCA and PLSR programmed in b
MF: medium fat.
MatLab (The MathWorks Inc., MA, USA) were utilized in the sta- c
HF: high fat.
d
tistical data treatment. SD: standard deviation.
 Z. Ozbekova, A. Kulmyrzaev / Journal of Food Engineering 210 (2017) 42e49 45

106 3260 Pa in case of the HF cheese with 31.08% of fat at the same
temperature (Tables 1 and 2). G0 , G00 , and tF also decreased when
the fat content of the cheese samples increased. In agreement with
earlier studies, at low amplitude values, in the LVE range, both the
G0 and G00 curves display constant plateau values on different levels
105 (data not shown). The overall magnitudes of G0 , G00 , tL, and tF
G', G" (Pa)

decreased from the LF, MF cheese to the HF cheese (Table 2). Such
evolution of the rheology of cheeses is believed to be caused,
particularly, by the amounts of fat and moisture in cheese, that is
both G0 and G00 decrease with increasing moisture and fat (Tunick
104 G' (LF)
G" (LF) et al., 1993). Our results showed that tL and tF follow the same
G' (MF) pattern as well (Table 2). However, in the case of the cheese
G" (MF) samples tested in the present study, the contribution of the fat
G' (HF) content to the reduction of G0 , G00 , tL, and tF seems to be much
G" (HF) greater than that of the moisture content. The LF cheese with the
103
largest moisture content among the cheese samples (Table 1) is
30 40 50 60 70
supposed to be the softest one, since elevating the water content
Temperature (oC) in cheese results in greater hydration of the casein network to
cause a decrease in hardness and G’ (Tunick et al., 1993). However,
Fig. 1. Temperature-sweep dynamic shear profiles of the experimental cheeses. the magnitudes of G0 , G00 , tL, and tF obtained with the LF cheese

Table 2
Rheological properties of the cheeses at linear viscoelastic (LVE), yield stress, and flow stress regions at 25e70  C.

T ( C) LVE Yield stress Flow stress

0
G (kPa) 00
G (kPa) 0
G (kPa) 00
G (kPa) tL (kPa) G0 (kPa) G00 (kPa) tF (kPa)
LF
25 115.76 37.38 119.40 47.68 8.94 42.00 42.00 41.60
30 110.09 33.28 104.80 38.97 5.91 35.50 35.50 30.60
35 76.70 25.57 99.30 37.65 3.87 24.49 24.49 24.35
40 63.32 22.91 86.65 36.18 1.36 19.12 19.12 13.97
45 48.84 20.89 71.02 35.50 1.14 16.79 16.79 8.02
50 41.47 18.91 45.38 29.77 0.41 12.97 12.97 3.48
55 37.07 18.51 20.70 34.98 0.55 10.18 10.18 2.22
60 22.04 15.84 35.22 15.24 0.28 8.51 8.51 0.92
65 21.89 12.50 56.42 47.90 0.22 4.73 4.73 0.56
70 17.41 10.75 39.82 24.53 0.37 1.58 1.58 0.25
MF
25 107.39 35.07 88.79 37.02 4.09 24.78 24.78 15.95
30 65.45 23.13 55.76 23.38 2.90 20.77 20.77 8.22
35 39.26 15.74 35.73 15.45 1.74 13.26 13.26 6.19
40 29.76 12.85 27.55 12.43 1.08 9.19 9.19 2.98
45 21.62 10.87 18.78 10.87 0.86 7.80 7.80 1.71
50 14.35 8.57 13.82 8.47 0.34 7.13 7.13 0.93
55 8.93 6.86 8.70 8.23 0.25 6.97 6.97 0.34
60 5.82 5.57 6.50 6.84 0.13 5.62 5.62 0.14
65 3.50 3.62 3.43 5.69 0.12 4.22 4.22 0.09
70 1.87 2.80 1.76 3.56 0.08 1.89 1.89 0.05
HF
25 97 31.28 79.80 33.40 3.26 20.66 20.66 14.18
30 55.84 19.10 46.02 20.94 2.08 15.05 15.05 8.52
35 33.15 11.83 27.77 12.45 1.23 9.84 9.84 6.66
40 22.53 9.46 20.40 9.14 1.12 7.12 7.12 2.58
45 15.84 7.84 14.54 8.18 0.67 5.42 5.42 1.36
50 11.28 6.49 11.12 5.99 0.45 4.57 4.57 1.08
55 7.55 5.74 7.34 4.56 0.29 3.98 3.98 0.51
60 3.20 3.20 2.62 3.08 0.20 3.16 3.16 0.16
65 1.85 2.74 1.53 2.57 0.13 0.93 0.93 0.19
70 0.86 1.42 0.93 1.39 0.04 0.41 0.41 0.09

loss modulus (G00 ) measured at LVE, yield stress and flow stress
regions as well as the values of the yield stress (tL) and flow stress
(tF) measured at 25e70  C. In agreement with the studies reported
earlier, an increase in temperature results in a decrease of all the
measured rheological characteristics of the cheese samples (G0 , G00 ,
tL, tF) (Karoui et al., 2003). In addition, as fat content of the cheese
samples increase, the magnitudes of its rheological characteristics
decrease. For example, at 25  C the yield stress (tL) of the LF
cheese containing 9.76% of fat was 8942 Pa, which decreased to
remain always larger comparing to those of the MF and HF cheeses
at the same temperature (Table 2). Such a consequence could be
elucidated by the fact that, at the same time, the amount of fat in
cheese as well as the size and distribution of the lipid particles
influences the interactions within protein matrix and determine
hardness (Van Hekken et al., 2007). In the LF cheese, the quantity
of fat dispersed within the protein matrix was relatively low to
disrupt the matrix and resulted in a hard cheese matrix, i.e. in the
larger magnitudes of G’, G”, tL, and tF.
46 Z. Ozbekova, A. Kulmyrzaev / Journal of Food Engineering 210 (2017) 42e49

3.3. Fluorescence properties
At a constant temperature (25  C), the shapes of the normalized
fluorescence spectra of the cheese samples varied depending on
their chemical composition.
The tryptophan emission spectra exhibited a peak at 382, 392
and 387 nm recorded with the LF, MF and HF cheese sample,
respectively (data not shown).
The emission spectra of vitamin A of the cheese samples showed
the highest normalized emission intensity at 526 nm and two
shoulders located at 400e450 nm and 560 nm (data not shown).
The excitation spectra of vitamin A of the LF, MF and HF cheeses had
the highest absorption at about 325 nm and a shoulder at 312 nm
(data not shown). Additionally, the width and intensities of the
fluorescence spectra differed from one cheese sample to the other.
As it has been concluded in the previous section, G’, G”, tL, and tF
were mainly influenced by the fat content of the cheeses. Tem-
perature alters physical state of fat in the cheese samples, thus, it
also modifies rheological properties of the cheeses. The shapes of
vitamin A spectra are correlated with the physical state of the tri-
glycerides in the fat globules of an emulsion (Dufour et al., 2000)
and the protein/fat globule interactions (Herbert et al., 2000). In
Fig. 2 the normalized emission spectra of tryptophan (Fig. 2A),
emission spectra (Fig. 2B) and excitation spectra (Fig. 2C) of vitamin
A for selected temperatures (25, 45 and 70  C) obtained with the
MF cheese are presented. The shapes of the spectra distinctly
changed depending on the temperature. As the temperature
increased, the highest intensities of all obtained spectra decreased
(Fig. 2 A, B, C). The variation of the environment polarity affects the
maximum intensity of the emission of tryptophan residues
(Lakowicz, 2006; Ladokhin, 2000). As the temperature increased,
the maximum intensity of the emission of tryptophan residues in
the cheese samples decreased and the maximum intensity wave-
length shifted from 394 nm at 25  C to 379 nm and 378 nm at 45  C
and 70  C (Fig. 2A). The decrease of the tryptophan emission could
be explained by the increase in the hydrophobicity of the trypto-
phan environment (Rampon et al., 2003). The degree of tryptophan
environment hydrophobicity is also estimated by the value of the
shift of emission spectra (Ladokhin, 2000). The blue-shift
(decrease) of the maximum emission wavelength to shorter
wavelength range with increasing temperature suggested that the
tryptophan residues of milk proteins moved into more hydrophobic
environment (Fig. 2A). The reason of this could be conformational
changes in proteins due to heating. Similar spectral patterns were
also found in the tryptophan emission spectra of the LF and HF
cheeses (data not shown).
Modifications of the maximum intensities and shapes of the
emission spectra (Fig. 2B) and excitation spectra (Fig. 2C) of vitamin
A obtained with the MF cheese at different temperatures were
explained by the decrease in the viscosity of triglycerides of cheese

0,016 0,035
25 oC A 25 oC B
0,014 0,030
45 oC o
45 C
0,012 70 oC 0,025 70 oC
Intensity (a.u.)

Intensity (a.u.)

0,010
0,020
0,008
0,015
0,006
0,010
0,004

0,002 0,005

0,000 0,000
280 300 320 340 360 380 400 420 440 460 480 500 300 350 400 450 500 550 600 650
Wavelength (nm) Wavelength (nm)
0,030
25 oC C
0,025 45 oC
o
70 C
0,020
Intensity (a.u.)

0,015

0,010

0,005

0,000
240 260 280 300 320 340 360
Wavelength (nm)
Fig. 2. (2A) Tryptophan emission spectra of the medium fat (MF) semi-hard cheese at 25, 45 and 70  C, (2B) Vitamin A emission spectra of the medium fat (MF) semi-hard cheese at
25, 45 and 70  C, (2C) Vitamin A excitation spectra of the medium fat (MF) semi-hard cheese at 25, 45 and 70  C.
 Z. Ozbekova, A. Kulmyrzaev / Journal of Food Engineering 210 (2017) 42e49
fat with the temperature increase (Dufour and Riaublanc, 1997).
Similar spectral patterns were also observed in the emission and
excitation spectra of vitamin A of the LF and HF cheeses obtained at
different temperatures (data not shown).
3.4. Multivariate statistical analysis
The analysis of the obtained spectra conducted in the previous
section has demonstrated that the characteristic spectral patterns
(maximum photon emission and excitation, shape and width of
spectra) have changed depending on the cheese constituents
(protein, moisture and fat contents) and temperature. It should be
noted that change of temperature caused modification of rheo-
logical properties of the cheese samples. Principal component
analysis (PCA) is a multivariate tool which is normally applied first
in order to confirm the results of tentative analysis of spectral
patterns observed under effect of different factors. PCA was applied
to the normalized spectra to extract information on the influence of
chemical composition and temperature on the cheese properties.
270 spectra (3 types of spectra, 3 types of cheese, 10 temperatures,
and 3 repetitions) were analyzed.
The results of PCA applied to the spectra (tryptophan emission,
vitamin A emission and excitation) obtained at 25  C showed
discrimination of the cheese samples depending on the cheese
chemical composition (data not shown).
Since the principal objective of this study was to predict the
rheological characteristics of cheeses such as storage modulus G0 ,
loss modulus G00 , yield stress tL, and flow stress tF measured at
three oscillation (yield stress, flow stress and LVE) regions and
these characteristics significantly affected by temperature, partic-
ular attention was drawn to the results of the PCA conducted on the
spectra obtained at different temperatures. As an example, the PCA
score plot obtained with the vitamin A excitation spectra of the MF
cheese samples is presented in Fig. 3. Principal component A1 and
principal component A2 explained 69.7% and 26.4% of the total
-3
x 10
3
1
A2 (26.4%)
MF30 MF30MF30
0 MF25
MF25
MF25
-1
-2
-5 -4 -3
Fig. 3. PCA similarity map obtained using vitamin A excitation spectra of the medium fat (MF) semi-hard cheese.
47
variance of the spectra, respectively. Relative to A1, the MF cheese
samples at different temperature conditions were discriminated.
The cheese samples, the spectra of which were recorded at the
temperatures higher than 40e45  C, were scored positively relative
to the principal component A1, while those, the spectra of which
recorded at the temperatures below 40e45  C, were scored nega-
tively Fig. 3. As the temperature-sweep dynamic shear experiments
demonstrated (Fig. 1), at 40e45  C the difference between G0 and G00
considerably decreased until the two moduli became equal each
other at the cross-point. Thus, the cheese samples scored positively
along A1 on the PCA scatter-plot (Fig. 3) oppositely differed from
those scored negatively because of the difference in the physical
state of triglycerides in the two groups of the cheese samples. The
cheese samples scored negatively exhibited solid-like behavior
(G’ > G00 ), while those scored positively because of melting of fat
became liquid-like (G’ < G”) (Fig. 3). Similar discriminations
depending on temperature were observed when PCA was applied
to the tryptophan and vitamin A emission spectra of the MF cheese
samples, as well as tryptophan emission, vitamin A emission and
excitation spectra of the LF and HF cheese samples (data not
shown). The results of PCA proved the close relationship between
the spectral data and temperature conditions and, consequently,
between the spectral data and rheological characteristics of the
cheeses.
The PLSR algorithm with cross-validation was used to develop
regression models of the chemical properties, melting point, and
rheological properties of the experimental cheeses. The numbers of
PLSR factors used for the final models were the numbers giving first
local minimum for the root mean squared error of validation
(RMSEV). The results of PLSR applied to the spectral, chemical and
rheological data sets are presented in Table 3. Considering LVE re-
gion of the oscillation tests, the best regression model including 7
components (R2 ¼ 0.82) to predict G0 and G00 was obtained applying
PLSR on the tryptophan emission spectra of the cheeses. Vitamin A
emission and excitation spectra are actually correlated with the
MF70
MF70
MF70
MF65
MF65
MF35
MF35
MF35 MF40
MF40
MF40 MF65
MF45
MF45
MF45
MF50
MF60
MF50
MF50 MF60
MF60
MF55
MF55
MF55
-2 -1 0 1 2
A1 (69.7%) x 10
-3
48 Z. Ozbekova, A. Kulmyrzaev / Journal of Food Engineering 210 (2017) 42e49

Table 3
Results of the PLSR statistics to predict rheological and physico-chemical properties of the experimental cheeses from fluorescence spectra.

Region Characteristics Triptophan emission spectra Vitamin A emission spectra Vtamin A excitation spectra

R2 PLS components R2 PLS components R2 PLS components

0
LVE G 0.82 7 0.80 10 0.78 6
G00 0.82 7 0.81 10 0.78 6
Yield stress G0 0.53 5 0.53 8 0.62 6
G00 0.59 5 0.62 8 0.67 6
tL 0.75 9 0.90 10 0.74 10
Flow stress G0 0.90 7 0.85 8 0.85 6
G00 0.90 7 0.85 8 0.85 6
tF 0.90 7 0.85 8 0.85 6
Physico-chemical characteristics
Tmelting( C) 0.99 6 0.98 9 0.89 6
Moisture (%) 0.99 6 0.98 10 0.89 6
Protein (%) 0.98 6 0.98 10 0.88 6
Fat (%) 0.98 6 0.98 10 0.88 6

physical state of triglycerides in the fat globules. In connection with 4. Conclusion

that, processing the fluorescence spectra of vitamin A yielded the
PLSR models, which were able to predict G’ and G” with relatively
high accuracy (Table 3).
Yield stress region is characterized by G0 and G00 as well, and
additionally yield stress tL can be derived from this part of the
oscillation tests. As mentioned above, tL indicates the magnitude of
the stress at which the deformation of a sample becomes irre-
versible. Therefore, tL is a critical characteristic to be taken into
account when, for example, the ability of any viscoelastic matter to
resist applied external forces in order to maintain given shape and
sizes is considered. Yield stress tL was well predicted with the R2
value of 0.90 using the PLSR model with 10 components, which was
derived using the vitamin A emission spectra (Table 3). The model
taking into account 6 components derived from the vitamin A
excitation spectra predicted fairly well the values of G0 and G00
related to the yield stress region. Corresponding values of R2 for G’
and G” were 0.62 and 0.67, respectively (Table 3).
Flow stress region yields flow stress tF, which indicates the
magnitude of the applied external mechanical stress inducing
viscous flow of a viscoelastic matter. At this point the viscoelastic
matter exhibits liquid-like behavior. The PLSR models allowing to
accurately predict G’, G” and tF of the cheese samples related to the
flow stress region were derived using the spectra of tryptophan and
vitamin A. The R2 values of the models developed on the trypto-
phan emission spectra were 0.95, while those based upon the
vitamin A emission and excitation spectra were 0.85 (Table 3).
PLSR statistics for the prediction of the moisture, protein and fat
contents and melting points of the cheese samples was also con-
ducted and presented in Table 3. Considering the accuracy of pre-
diction, the regression models derived from the tryptophan
emission spectra and vitamin A emission spectra demonstrated the
highest values of R2. The melting temperatures of the cheeses
(Tmelting) and moisture content were predicted with R2 ¼ 0.99 using
the tryptophan emission spectra (Table 3). As the PLSR statistics
showed, the vitamin A emission and excitation spectra could also
be considered as a reliable source of information on the chemical
composition and melting temperatures of cheeses. Melting tem-
peratures and moisture, protein and fat contents were predicted
with R2 ¼ 0.98 applying PLSR on the vitamin A emission spectra
(Table 3). The regression with the vitamin A excitation spectra
resulted in the prediction of melting temperatures and moisture
content with R2 ¼ 0.89 and protein and fat content with R2 ¼ 0.88.
PLSR analysis proved the fluorescence spectra of tryptophan and
vitamin A as useful information source of rheological and chemical
properties of cheeses.
Fluorescence spectroscopy coupled with multivariate statistical
tools was successfully used to develop PLS regression models to
predict storage modulus (G0 ), loss modulus (G00 ), yield stress (tL)
and flow stress (tF) of the semi-hard cheeses at LVE, yield stress and
flow stress oscillation regions simultaneously. In addition, melting
temperatures, moisture, protein and fat contents were accurately
predicted using the regression models derived from the fluores-
cence spectral data collected from the experimental cheeses. It
should be particularly noted that the attempt to predict yield stress
(tL) and flow stress (tF) from the fluorescence spectra of the cheeses
was successful. Indeed, yield stress (tL) and flow stress (tF) were
predicted with R2 ¼ 0.90 applying PLSR to the vitamin A emission
spectra and vitamin A excitation spectra, respectively. In conclu-
sion, this study demonstrates that fluorescence spectroscopy
combined with multivariate statistical tools has potential applica-
tion for the accurate, non-destructive and rapid prediction of
storage modulus (G’), loss modulus (G”), yield stress (tL) and flow
stress (tF) of semi-hard cheeses at LVE, yield stress and flow stress
oscillation regions simultaneously. In order to test its robustness
the technique developed in this study will be examined using
greater number of cheese varieties.
References
Ak, M.M., Bogenrief, D., Gunasekaran, S., Olson, N.F., 1993. Rheological evaluation of
Mozzarella cheese by uniaxial horizontal extension. J. Texture Stud. 24,
437e453.
Andersen, C.M., Vishart, M., Holm, V.K., 2005. Application of fluorescence spec-
troscopy in the evaluation of light-induced oxidation in cheese. J. Agric. Food
Chem. 53, 9985e9992.
AOAC, 2000. Official Methods of Analysis, seventeenth ed. Association of Official
Analytical Chemists, Maryland (Chapter 33).
Bertrand, D., Scotter, C.N.G., 1992. Application of multivariate analyses to NIR
spectra of gelatinized starch. Appl. Spectrosc. 46, 1420e1425.
Buffa, M.N., Trujillo, A.J., Pavia, M., Guamis, B., 2001. Changes in textural, micro-
structural, and colour characteristics during ripening of cheeses made from raw,
pasteurized or high-pressure-treated goats milk. Int. Dairy J. 11, 927e934.
Campanella, O.H., Popplewell, L.M., Rosenau, J.R., Peleg, M., 1987. Elongational vis-
cosity measurements of melting American Process cheese. J. Food Sci. 52,
1249e1251.
Dijkstra, T.K., 2010. Latent variables and indices: herman Wold’s basic design and
partial least squares. In: Vinzi, V.E., Chin, W.W., Henseler, J., Wang, H. (Eds.),
Handbook of Partial Least Squares. Concepts, Methods and Applications.
Springer, Berlin, pp. 1e23.
Dufour, E., Riaublanc, A., 1997. Potentiality of spectroscopic methods for the char-
acterization of dairy products. I. Front-face fluorescence study of raw, heated
and homogenized milks. Le Lait 77 (6), 657e670.
Dufour, E., Mazerolles, G., Devaux, M.F., Duboz, G., Duployer, M.H., Mouhous-
Riou, N., 2000. Phase transition of triglycerides during semi-hard cheese
ripening. Int. Dairy J. 10, 81e93.
 Z. Ozbekova, A. Kulmyrzaev / Journal of Food Engineering 210 (2017) 42e49
Guinee, T.P., 2011. Cheese rheology. In: Fuquay, F.W., Fox, P.F., McSweeney, P.L.H.
(Eds.), Encyclopedia of Dairy Sciences. Elsevier Academic press, London,
pp. 685e697.
Herbert, S., Mouhous-Riou, N., Devaux, M.F., Riaublanc, A., Bouchet, B., Gallant, J.D.,
Dufour, E., 2000. Monitoring the identity and the structure of soft cheeses by
fluorescence spectroscopy. Le Lait 80, 621e634.
Hort, J., Grys, G., Woodman, J., 1997. The relationships between the chemical,
rheological and textural properties of Cheddar cheese. Lait 77, 587e600.
John, A.L., 2008. Some perspectives on the use of cheese as a food ingredient. Dairy
Sci. Technol. 88, 573e594.
Joshi, N.S., Jhala, R.P., Muthukumarappan, K., Acharya, M.R., Mistry, V.V., 2004.
Textural and rheological properties of processed cheese. Int. J. Food Prop. 7 (3),
519e530.
Karoui, R., Dufour, E., 2006. Prediction of the rheology parameters of ripened semi-
hard cheeses using fluorescence spectra in the UV and visible ranges recorded
at a young stage. Int. Dairy J. 16, 1490e1497.
Karoui, R., Laguet, A., Dufour, E., 2003. Fluorescence spectroscopy: a tool for the
investigation of cheese melting e correlation with rheological characteristics.
EDP Sci. 83, 251e264.
Karoui, R., Dufour, E., De Baerdemaeker, J., 2007. Monitoring the molecular changes
by front-face fluorescence spectroscopy throughout ripening of a semi-hard
cheese. Food Chem. 104, 409e420.
Kulmyrzaev, A., Dufour, E., Noe, Y., Hanafic, M., Karoui, R., Qannari, E.M.,
Mazerolles, G., 2005. Investigation at the molecular level of soft cheese quality
and ripening by infrared and fluorescence spectroscopies and chemometrics-
relationships with rheology properties. Int. Dairy J. 15, 669e678.
Ladokhin, A.S., 2000. Fluorescence spectroscopy in peptide and protein analysis. In:
Meyers, R.A. (Ed.), Encyclopedia of Analytical Chemistry. John Wiley & Sons Ltd,
Chichester, pp. 5762e5779.
Lakowicz, J.R., 2006. Fluorophores. In: Lacowicz, J.R. (Ed.), Principles of Fluorescence
Spectroscopy. Plenum Press, New York.
Leach, M.R., Farkas, B.E., Daubert, C.R., 2003. Rheological characterization of process
cheese using tube viscometry. Int. J. Food Prop. 6 (2), 259e267.
Mezger, T.G., 2011. The Rheology Handbook, third ed. Vincentz Network GmbH & Co.
49
KG, Hanover.
Monfreda, M., 2012. Principal component analysis: a powerful interpretative tool at
the service of analytical methodology. In: Sanguansat, P. (Ed.), Principal
Component Analysis. Rijeka: InTech, pp. 49e66.
Muliawan, Edward B., Hatzikiriakos, Savvas G., 2007. Rheology of mozzarella
cheese. Int. Dairy J. 17, 1063e1072.
O’Callaghan, D.J., Guinee, T.P., 2004. Rheology and texture of cheese. In: Fox, P.F.,
McSweeney, P.L.H., Cogan, T.M., Guinee, T.P. (Eds.), Cheese: Chemistry, Physics
and Microbiology. Elsevier Academic Press, London, pp. 511e540.
Purna, G.S.K., Prow, L.A., Metzger, L.E., 2005. Utilization of front-face fluorescence
spectroscopy for analysis of process cheese functionality. Dairy Sci. 88,
470e477.
Rampon, V., Genot, C., Riaublanc, A., Anton, M., Axelos, M.A.V., McClements, D.J.,
2003. Front-face fluorescence spectroscopy study of globular proteins in
emulsions: displacement of BSA by a nonionic surfactant. J. Agric. Food Chem.
51, 2482e2489.
€l, Y., 2003. Relation between a temperature-sweep dynamic shear
Reparet, J.-M., Noe
test and functional properties of cheeses. Le Lait 83 (4), 321e333.
Strasburg, G.M., Ludescher, R.D., 1995. Theory and application of fluorescence
spectroscopy in food research. Trends Food Sci. Technol. 6, 69e75.
Subramanian, R., Muthukumarappan, K., Gunasekaran, S., 2006. Linear viscoelastic
properties of regular- and reduced-fat pasteurized process cheese during
heating and cooling. Int. J. Food Prop. 9 (3), 377e393.
Tunick, M.H., 2010. Activation energy measurements in rheological analysis of
cheese. Int. Dairy J. 20, 680e685.
Tunick, M.H., Mackey, K.L., Shieh, J.J., Smith, P.W., Cooke, P., Malin, E.L., 1993.
Rheology and microstructure of low-fat Mozzarella cheese. Int. Dairy J. 3,
649e662.
Van Hekken, D.L., Tunick, M.H., Malin, E.L., Holsinger, V.H., 2007. Rheology and melt
characterization of low-fat and full fat Mozzarella cheese made from micro-
fluidized milk. LWT - Food Sci. Technol. 40, 89e98.
Venugopal, V., Muthukumarappan, K., 2003. Rheological properties of Cheddar
cheese during heating and cooling. Int. J. Food Prop. 6 (1), 99e114.