Food Hydrocolloids 77 (2018) 474e481

Contents lists available at ScienceDirect

Food Hydrocolloids
journal homepage: www.elsevier.com/locate/foodhyd

Physical properties of tofu gel probed by water translational/rotational

dynamics
Tsubasa Kawaguchi a, Rio Kita a, Naoki Shinyashiki a, Shin Yagihara a, *, Minoru Fukuzaki b
a
Department of Physics, School of Science, Tokai University, 4-1-1 Kitakaname, Hiratuka, Kanagawa, Japan
b
Kumamoto Liberal Arts Education Center, Tokay University, 9-1-1 Toroku, Higashi, Kumamoto, Kumamoto, Japan

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

Article history: The dynamic behavior of water molecules in tofu gels is observed using dielectric spectroscopy (DS) and
Received 13 July 2017 pulsed field gradient nuclear magnetic resonance (PFG-NMR) methods. The breaking stress of tofu gels
Received in revised form are dependent on the heating time, which affects the water structures characterized by the dielectric
6 October 2017
relaxation time, t. On the other hand, the inhomogeneity of the gel structures decreases with increasing
Accepted 20 October 2017
Available online 29 October 2017
coagulant concentration was reflected as decreasing of the dielectric broadening parameter, b, that
related for a fluctuation of the dynamic water structure. In addition, a decrease in the translational
diffusion coefficient of water molecules is obtained with increasing coagulant concentration suggesting a
Keywords:
Tofu
reduction of the free space for the diffusional motion of water molecules. Negative correlations between
Soybean curd the broadening parameter, b, and the inverse of diffusion coefficient, 1/D, are found in tofu gels and
Food gel evaluation soymilk with distinct correlation coefficients of
0.95 and
0.87, respectively. These results suggest that
Water dynamics water dynamics analysis and complementary analysis using DS and PFG-NMR measuring techniques can
Dielectric spectroscopy be an effective tool to evaluate food gel physical properties. Further these technique could also explain
PFG-NMR the molecular mechanisms of gel and liquid structures and properties via dynamic water structures.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction its mesoscopic/microscopic properties investigated via mechanical

Tofu, also known as soybean curd, is an ordinary gelatinous food
with high water content that is traditionally eaten in Japan and
several Asian countries. In recent years, tofu has been recognized as
being healthy even in the Western countries because of its natural
high protein and low fat content and because it is inferred that tofu
reduces the development of some types of cancer (Messina, Persky,
Setchell, & Barnes, 1994; Wu et al., 1996). The typical recipe for tofu
involves heating soymilk with small amounts of coagulants, such as
magnesium chloride or calcium chloride. Interestingly, tofu shows
various textures depending on the coagulant concentration or type
(Prabhakaran, Perera, & Valiyaveettil, 2006); heating time
(Kohyama, Sano, & Doi, 1995); and amount of soybean protein
(Cheng, Shimizu, & Kimura, 2005; Kohyama, Murata, Tani, Sano,
Doi, 1995) which means that tofu is suitable for the food gel
model due to its controllable property. In the field of food science,
multiple papers have already reported on the gel structures of tofu
(Nishinari, Fang, Guo, & Phillips, 2014), especially, those based on
* Corresponding author.
E-mail address: yagihara@keyaki.cc.u-tokai.ac.jp (S. Yagihara).
measurements (Cheng et al., 2005; Fuchigami & Teramoto, 1997;
Huang & Kuo, 2015), ultrasound measurements (Kuo, Lien,
Huang, & Ting, 2011), and scanning electronic microscope (SEM)
or confocal laser scanning microscopy (CLSM) observations
(DeMan, DeMan, & Gupta, 1986; Onodera, Tomotada, & Nakasato,
2009; Pre stamo, Lesmes, Otero, & Arroyo, 2000; Urbonaite, De
Jongh, Van Der Linden, & Pouvreau, 2015). According to these
studies, the gelation model of tofu assumes, to a certain extent, that
the soy protein aggregates, thus holding water in pores of tens to
hundreds of micrometers (Kohyama et al., 1995a; Peng, Ren, & Guo,
2016). Experimental techniques, for example, nuclear magnetic
resonance (NMR), dielectric spectroscopy (DS), FT-IR, and circular
dichroism (Nagano, Akasaka, & Nishinari, 1995) provide informa-
tion of molecular dynamics and structures through analyses of
molecular behaviors, such as the molecular rotation, translation,
and vibration. Specifically, reorientations and translations of water
molecules respectively observed by DS and NMR methods could be
complementarily analyzed with larger scale behaviors obtained
from mechanical measurements (Li et al., 2014, 2015). Therefore,
there is not a sufficient understanding of the dynamic structure of
tofu gels on the molecular scale.

https://doi.org/10.1016/j.foodhyd.2017.10.025
0268-005X/© 2017 Elsevier Ltd. All rights reserved.
 T. Kawaguchi et al. / Food Hydrocolloids 77 (2018) 474e481 475

As other aqueous gel systems, tofu gels contain large amounts of molecular mechanisms in the water structures.
water, approximately 80%e90% (Wang, Swain, Kwolek, & Fehr,
1983). Therefore, the water content and the molecular in-
2. Experimental
teractions greatly affect the physical properties e.g. viscoelasticity
of tofu gels. Nevertheless, it is extremely difficult to directly observe
2.1. Sample preparation
dynamic behaviors of water molecules because of remarkably fast
dynamics of water molecules and the hydrogen-bonding networks
Tofu gels were made by mixing soymilk with a coagulant in a
with the characteristic time of 8 ps, in which the elementary pro-
water bath at 80±0:1+ C. The concentrations of magnesium chloride
cess of molecular motions is rotations of water molecules. Although
hexahydrate (Wako 1st Grade, Wako Pure Chemical Industries,
drying (Cai & Chang, 1997) or the centrifugation (Tang, Chen, &
Osaka, Japan) used as the coagulant were set from 0.4 wt% to 0.7 wt
Foegeding, 2011) methods are often employed to determine the
%. The heating time was varied with values of 10 min, 30 min, and
water content, dynamic behaviors of water molecules in tofu gel
60 min. After the heating procedure for the gelation process, each
cannot be obtained from these methods.
sample was stored during several minutes at room temperature to
For these reasons, a non-invasive and dynamic water observa-
make it cool down before the measurements. Two types of soymilk
tion technique is required to examine tofu in the food industry and
samples were prepared: the first was commercial soymilk
to gain scientific understanding. Therefore, we used the pulsed field
comprising a 9% soy protein component, and the second one was
gradient NMR (PFG-NMR) method and dielectric spectroscopy (DS)
made from a soy protein isolate (SPI) dissolved in ultrapure water
to observe the translational and rotational water dynamics,
with various concentrations of 5 wt%, 7 wt%, and 9 wt%. As SPI
respectively.
sample, Fujipro F kindly supplied by Fuji Oil Co., Ltd. (Osaka, Japan)
The PFG technique has been applied in various scientific
was used. The Fujipro F is composed of 21% 7S glycinin, 41% 11S
research fields. In the earlier period of its development, the po-
glycinin, 38% other lipoprotein, and 1.5% cations. The solutions were
tential of the field gradient method drew a lot of attention due to its
heated at 100±0:1+ C in a water bath for 3 min before adding the
usefulness of nuclear spin system controllability. The first study to
coagulant for complete dissolution of the SPI. These preparation
obtain a diffusion coefficient and expression was conducted by
procedures were carried out based on a report by Cheng et al.
Stejskal and Tanner (Stejskal & Tanner, 1965) by using PFG spin
(Cheng, Shimizu, & Kimura, 2004). The prepared sample conditions
echo sequence. In the field of medical science, the magnetic reso-
are listed in Table 1 for each heating times coagulant concentration,
nance imaging (MRI) with using the PFG method is indispensable in
and SPI concentration.
obtaining anatomical images for diagnoses. Currently, the PFG
The tofu samples were prepared in 5-ml vial tubes for the DS
method is used as a general method of NMR to examine physical
measurements. Bubbles were carefully removed via a pipette
properties, and to develop strategies for medical diagnoses. In this
before heating to ensure a smooth contact of the electrode surface
study, we applied the PFG technique to obtain the water trans-
with the sample. Isolated water that appeared after the gelling
lational diffusion coefficient in the tens of milliseconds region.
processes was also removed before measurements. A glass capillary
The DS analysis is based on the dynamic behaviors of the per-
was used for the PFG-NMR measurements as described below with
manent dipole moments of atomic groups or molecules under an
the same heating procedures. The sample amount required for DS
alternating electric field. Multiple dielectric studies have been
performed in the MHz-GHz frequency region for aqueous solution and NMR measurements were around 3 ml and 100 ml, respectively.
systems, e.g., polymer solutions, protein solutions (Shinyashiki
et al., 2009), and solutions in food systems (Maruyama et al., Table 1
2014; Miura, Yagihara, & Mashimo, 2003). Based on the study of The three dielectric relaxation parameters, the dielectric relaxation time t, the
Li et al. (Li et al., 2014), there are three types of waters in tofu gels: dielectric relaxation time broadening parameter b, the dielectric strength Dε, and
water that exists freely, near the surface of proteins, and tightly the diffusion coefficient D and the sample condition, the SPI concentration, coagu-
lant concentration c, and heating time t, obtained by the series of experiments.
bound with proteins. It is assumed that the latter two components
of water appear in the less than MHz region (Abe et al., 2017; SPI (wt%) c (wt%) t (min.) t (ps) b Dε Dðm2 =secÞ
Mashimo, Kuwabara, Yagihara, & Higasi, 1987 ). Unfortunately, in Soymilka 5 e e 7.93 1.000 72.1 2.0810
9
the less than MHz region, there are vast contribution of the direct 7 e e 9.33 0.910 58.9 2.0010
9
current (DC) component with the coagulant ions, which obstructs 9 e e 1.02 0.882 51.1 1.9210
9
Tofu Gela 7 0.5 30 9.21 0.913 63.0 1.8710
9
the relaxation process in certain regions. Therefore, we can only
7 0.6 30 10.12 0.895 61.4 1.8210
9
focus on the free water dynamics that usually appear in the GHz 7 0.7 30 8.40 0.926 64.8 1.9110
9
region with negligible DC components. The dielectric relaxation 9 0.5 30 11.26 0.911 63.5 1.9310
9
time of pure water, t, is 8.27 ps at 25+ C (Kaatze, 1989), which in- 9 0.6 10 9.67 0.907 66.3 1.8810
9
dicates the characteristic time of the dynamics of the hydrogen- 9 0.6 30 10.36 0.949 71.9 1.8110
9
9 0.6 60 8.44 0.983 73.3 1.7810
9
bonding network. This relaxation process provides valuable infor- 9 0.6 120 9.24 0.884 60.1 1.8810
9
mation concerning the dynamics of the water structures in aqueous 9 0.7 30 9.27 0.911 62.9 1.7910
9
systems and applies several analytical methods, such as a fractal Tofu Gelb e 0.4 10 8.26 0.974 63.3 2.0610
9
analysis (Ryabov, Feldman, Shinyashiki, & Yagihara, 2002) to eval- e 0.4 30 8.39 0.974 63.1 1.9910
9
uate the distribution of water molecules in aqueous materials from e 0.4 60 8.52 0.982 64.1 2.0110
9
the average and fluctuation of dynamic behaviors of water mole- e 0.5 10 8.13 0.966 65.9 1.9910
9
cules (Yagihara et al., 2007). e 0.5 30 8.45 0.974 63.3 1.9910
9
In this study, we performed high-frequency DS and PFG-NMR e 0.5 60 8.52 0.961 62.0 2.0010
9
measurements on tofu gels and soymilk to observe the rotational e 0.6 10 8.20 0.962 64.9 1.9610
9
and translational water dynamics directly in the gels. We propose e 0.6 30 8.45 0.968 62.9 1.9610
9
this method as a novel analysis method for food gels to characterize e 0.6 60 8.52 0.968 62.0 1.9910
9
the viscoelastic property. Furthermore, the correlation between the a
Made from a SPI dissolved in ultrapure water and measured by TDR.
dielectric relaxation and translational diffusion processes with b
Made from commercial soymilk consisting of a 9% soy protein component and
respective time and length scales was considered to explain these measured by VNA.
476 T. Kawaguchi et al. / Food Hydrocolloids 77 (2018) 474e481
Table 2
Linear regression analysis of the correlation between the dielectric relaxation
broadening parameter b and the inverse of the diffusion coefficient 1=D.
Slope Intercept R2 Pearson's r
Soymilk
2.8810
9 2.37 0.81
0.95
Tofu Gel
1.4510
9 1.70 0.74
0.87
2.2. PFG-NMR method
PFG-NMR measurements were performed using a 400-MHz
NMR (Ascend 400WB, Bruker, Germany) equipped with a Diff 50
probe (Bruker, Germany) at 25±0:1+ C. In this study, the diffusion
coefficient of the water molecules was obtained from the simulated
echo (Tanner, 1970) sequence. This method employs a field gradient
with various intensities up to 600 G/cm in the stimulated echo
sequence. Fig. 1 shows the stimulated-echo sequence, where D is
the diffusion time, d is the duration of the field gradient, g is the
gradient strength, and p=2 is the 90+ pulse. The diffusion time, D,
and the duration of field gradient, d, were fixed to 20 ms and 1 ms,
respectively. The resulting attenuation curves were fitted using the
Stejskal-Tanner equation (Stejskal & Tanner, 1965), that is written
as


AðgÞ 2
ln ¼
ðgg dÞ ðD
d=3ÞD: (1)
Aðg0 Þ
Here AðgÞ is the signal intensity at the field gradient strength g,
Aðg0 Þ is the signal intensity of the first plot, and g is the gyro-
magnetic ratio. In general, the right side of Eq. (1) except for D, is
called the B-value.
The gradient strength was calibrated using the diffusion coef-
ficient for ultrapure water, which is known to be
2:299  10
9 m2 =sec at 25+ C (Holz, Heil, & Sacco, 2000). To
compensate for convection (Chung, Yu, & Ahn, 2011; Iwashita,
Konuma, Harada, Mori, Sugase, 2016), a glass capillary with a
diameter of 1.8 mm was used as the sample tube and filled to a
depth of approximately 4 cm in a series of experiments. With this
condition, the required sample volume was approximately 80 ml.
Because these samples were prepared in water-rich conditions, the
Q-value for the resonance circuit was intentionally reduced on the
concept of Price and Walchli's (Price & Walchli, 2002) method to
diminish the radiation damping effect, which can often result in
larger errors in NMR measurements for aqueous systems.
2.3. DS measurement
DS measurements in the GHz frequency region were performed
to investigate the free water dynamics in tofu gels at 25±0:5+ C.
Frequency domain measurement with a vector network analyzer
Fig. 1. Schematic chart of the stimulated echo sequence.
(VNA) and time domain measurement with a time domain reflec-
tometry (TDR) technique were used to observe the dielectric
relaxation processes for the water dynamics analyses. A VNA
(N5230C, Agilent Technologies, CA., US) was used for the mea-
surement in the frequency range of 100 MHz-50 GHz with a high
performance probe (8570E, Agilent Technologies, CA., US) as an
electrode. The Ecal unit (N4693, Agilent Technologies, CA., US) was
connected between the VNA and the electrode via a flexible cable.
The OPEN, SHORT, and standard reference calibrations were per-
formed using air, mercury, and pure water, respectively, with an
automatically compensating transmission line and electrode via
the Ecal unit.
For the TDR measurements, a digitizing oscilloscope (DSA8300
and 80E10 TDR module, Tektronix, OR., US) was used, where the
rising time of the incident pulse was 12 ps. An open-ended semi-
rigid cable with a diameter of 2.2 mm and a length of 70 cm was
used as an electrode. The difference method (Cole, Mashimo, &
Winsor, 1980) was adopted for the analyses via following equations.
  
1 þ ðcfs Þ juðgd Þεs ðuÞ r fx
εx ðuÞ ¼    ; (2)
1 þ juðgd Þεs ðuÞ ðcfs Þ r fs
where,
8
>
> Vs ðuÞ
Vx ðuÞ
>
> r ¼ ;
>
< Vs ðuÞ þ Vx ðuÞ
(3)
>
> fi ðzi Þ ¼ zi cotðzi Þ;
>
> pffiffiffiffiffiffiffiffiffiffiffiffi
>
: z ð uÞ
i ¼ ðud=cÞ εi ðuÞ:
Here εx ðuÞ and εs ðuÞ are the complex permittivities of the un-
known and reference samples, respectively, c is the light speed, j is
the imaginary number ðj2 ¼
1Þ, u is the angular frequency, gd is
the electrical length of the electrode, and d is the cell length. Vx ðuÞ
and Vs ðuÞ are the Laplace transforms of the reflected signals, Rx ðtÞ
and Rs ðtÞ, respectively, which are observable via the TDR method.
Ultrapure water (Millipore, Simplicity UV, resistivity 18.2 MU,cm)
and dimethyl sulfoxide (DMSO) were used as reference samples.
Comparing our data with that in the literature indicates that the
frequency covered by the TDR measurements was approximately
100 MHz-65 GHz.
The electrode for dielectric measurements was directly con-
tacted to the tofu gel inside the vial tube. Three times VNA and eight
times TDR measurements were respectively performed for one
sample to alternate gel-electrode contact surfaces. Ambiguities in
dielectric relaxation curves from the rugged surface of tofu gel
sample and non-removable isolated water were effectively sup-
pressed by virtue of the procedures of multiple measurements.
Finally, the obtained data of the complex permittivity ε ðuÞ, i.e.,
the dielectric constant ε0 and loss ε00 , were fitted using Cole-Cole
equation (Cole & Cole, 1941),
00 Dε sDC
ε ðuÞ ¼ ε0
jε ¼ ε∞ þ
j ; (4)
1 þ ðjutÞb ε0 u
where ε∞ is the extrapolated high-frequency permittivity, sDC is the
DC component, ε0 is the permittivity of vacuum, Dε is the dielectric
strength, t is the relaxation time, and b is the relaxation time
broadening parameter ð0 < b  1Þ. The three relaxation parameters
are especially important when explaining the results because t
reflects the characteristic time for the rotating mobility of the
molecule, Dε indicates a dipole moment and its number density
that directly contributes to the relaxation process, and b evaluates
the symmetrical broadening of the dielectric loss peak and reflects
 T. Kawaguchi et al. / Food Hydrocolloids 77 (2018) 474e481 477

a variety of molecular dynamics. Here the b value is unity for pure

water, and gels or dispersion systems have low values. For each
experiment, these three dielectric relaxation parameters were
determined via a curve fitting procedure using an in-house devel-
oped software.

3. Results and discussion

3.1. Diffusion coefficient

The 1 H-NMR spectra obtained for the soymilk is shown in Fig. 2.

The chemical shift was calibrated by via the TMS before of the
experiments. The large OH group peak of the water molecule is
observed at approximately 4 ppm. According to the gelation of the
soymilk, soy protein peaks in NMR spectra became smaller and
broader, e.g. the peak around 0.5 ppm in, Fig. 2b. It is well known
that the decrease in the molecular mobility causes a decrease in T2
NMR relaxation time, and hence NMR peaks became smaller and
broader (Slichter, 2013). In this study, only the OH group peak was
Fig. 3. Attenuation curve of the signal intensity obtained from the PFG-NMR mea-
analyzed to investigate the water dynamics. surement for soymilk and several coagulant concentrations, (open triangle) 0.4 wt%,
Fig. 3 shows the straight line of the logarithm of the signal in- (open square) 0.5 wt%, and (open circle) 0.6 wt%, of tofu gels heated for 10 min.
tensity ratio AðgÞ=Aðg0 Þ against the B-value for the PFG-NMR
measurements. The slope gives the value of the diffusion coeffi-
cient, D. Although a former study (Hong & Lee, 2006) reported that
the attenuation curve obtained for tofu had a multi-component
diffusion coefficient, only one component was observed in this
study. It is thought that the former study was performed with low
field NMR measurements that had a proton resonance frequency of
63 MHz thus the resolution is inferior than present experiments.
Therefore, the water OH group peak could not be separated from
the soy protein peaks and the attenuation curves contributed by
both the protein and water peaks may have displayed multi-
component behaviors. Furthermore, this consideration is consis-
tent with the study performed by Li et al. (Li et al., 2014) that the
bound water has small T2 relaxation time around 2 ms that suffi-
ciently smaller than the diffusion time for PFG-NMR measure-
ments. Hence, the signals of bound water were completely
attenuated during the diffusion time 20 ms. Consequently, only the
free water diffusion was observed.
Values of the diffusion coefficient, D, obtained from the Stejskal-

Fig. 4. The relationship between the coagulant concentration and the diffusion coef-
ficient of water molecules in (open diamond) pure water, () soymilk, and tofu gels
with heating times of (open circle) 10 min, (open triangle) 30 min, and (open square)
60 min.

Fig. 2. 1 H NMR spectra for (a) soymilk and (b) tofu gel at 25  C. The large peak near
4 ppm is attributed to water protons. Other smaller peaks are attributed to 11S, 7S
glycinin and lipids.
Tanner equation are listed in Table 1 and plotted in Fig. 4 that shows
the diffusion coefficient of water decreasing with coagulant con-
centration and heating time increasing. Onodera et al. (Onodera
et al., 2009) reported that an increase in the coagulant concentra-
tion makes the tofu microstructure finer in the lower coagulant
concentration region. On the other hand, there exists a change
point of the coagulant concentration dependence, and the tofu
microstructure shows coarse networks in the higher concentration
region. Furthermore, these trends correspond to mechanical study,
e.g. at higher concentrations than the change point, the breaking
stress of tofu gel decreases with increasing concentration. However,
the coagulant concentration region treated in the present work is
considered to be lower than the change point, since only a mono-
tonic decreasing trend was found in Fig. 4.
The diffusion coefficient of the water molecules existing in the
tofu gel pores had values closed to the diffusion coefficient for pure
water; and it means that water observed here behaves as free
478 T. Kawaguchi et al. / Food Hydrocolloids 77 (2018) 474e481

pffiffiffiffiffiffiffiffiffiffi
water. The spatial scale, r ¼ 6DD, of the diffusion was estimated
around 10 mm that comparable with the pore sizes in tofu gel (Peng
et al., 2016). Furthermore, the correlation coefficient between the
diffusion coefficient and coagulant concentration was obtained
as
0.73 (p < 0.01) that clearly means a negative correlation. These
results suggest that the pore size in the tofu gel structure decreases
with increasing coagulant concentration, and then increasing
collision frequency of water molecules in the pore structures de-
creases the apparent diffusion coefficient. On the other hand,
Kohyama et al. (Kohyama et al., 1995a) reported that the breaking
stress of tofu gel was exponentially increased with the heating time
and finally saturated. However, our results indicate that there is no
correlation between the diffusion coefficient and heating time (p <
0.05). Therefore, it is suggested that the diffusion coefficient is
influenced by only the homogeneity of tofu gel microstructure.

3.2. Dielectric relaxation

Results of dielectric spectroscopy measurements were well

characterized by depicting relation of the permittivity between
dielectric loss that so-called Cole-Cole diagram (Damez, Clerjon,
Abouelkaram, & Lepetit, 2007; Nelson & Trabelsi, 2006;
Watanabe, Taka, & Fujiwara, 2009), or to perform non-linear least
square curve fitting for frequency dependences of permittivity and
dielectric loss (Grosse, 2014). In this study we took the latter
method as described in the experimental section. A typical
dielectric relaxation curve obtained from VNA measurements for a
tofu gel sample prepared with a coagulant concentration of 0.4 wt%
and a heating time of 10 min was well fitted by two Cole-Cole
equations, as shown in Fig. 5. It is known that the relaxation pro-
cess with a loss peak near 20 GHz is caused by free water (Kaatze,
2007).
The lower frequency process near 1 GHz is much smaller than
the high-frequency free water process that is thought to involve
bound water or local protein dynamics. Miura et al. (Miura et al.,
2003) performed dielectric spectroscopy measurements for Fig. 5. Fitting curve example for the Cole-Cole equation for a tofu gel with a 0.4 wt%
various food systems and reported that the other food systems also coagulant concentration and 10 min of heating time. The dashed, and dotted lines
indicate the first process, and the second process, respectively. The direct current (DC)
have two dielectric relaxation at around 10 MHz to 1 GHz, and
component that only appeared in lower figure, was depicted by chain line. The solid
around 20 GHz. The lower relaxation is considered to bound water line indicates the sum of these components.
or micro-Brownian chain motion of the denatured protein. The
same characterization could be applied for this study. However, we
do not examine the lower process in more detail in this study In contrast, the relaxation time distribution parameter b does
because our aim is to examine the free water dynamics. The DC not depend on the heating time (r ¼ 0.17, p < 0.1) but does depends
component reflects the dynamic behavior of ions from the soymilk on the coagulant concentration (r ¼ 0.69, p < 0.05). Levy et al.
and the residual MgCl2 used as a coagulant to form the tofu gel (Levy, Puzenko, Kaatze, Ishai, & Feldman, 2012) reported that the b
structure. The dielectric relaxation curve obtained via TDR shows a value could be decreased by increasing the salt concentration in
similar relaxation processes, and the same analysis procedure was ionic aqueous solutions. However, the extent of a decrease of b is
performed. The relaxation parameters obtained for the around from 1.00 to 0.97 with addition of 3 wt% potassium chloride to the
20 GHz frequency process are listed in Table 1. solution. On the contrary, Fig. 6 shows a decrease in b from 1.00 to
Fig. 6 shows the coagulant concentration and the heating time 0.96e0.97 even at 0.7 wt% salt concentration. Consequently, the
dependencies of the three dielectric relaxation parameters (the decreasing trend of the b value shown in this study should have a
relaxation time t, the relaxation time broadening parameter b, and different origin from that reported by Levy et al. In the case of the
the dielectric strength Dε) obtained from the VNA measurements. present work, the decrease in b was induced by intricate con-
Despite the appropriate device calibration, the present results for straints to water molecules in the gel structure. Generally restricted
tofu show error bars of about 3% as maximum. It is considered that water molecules exhibit an increase in the relaxation time and a
the incomplete contact between surfaces of the electrode and decrease in b values, since the restricted motion of water molecules
sample and heterogeneity of the tofu gel sample. The relaxation also induces a structure formation of water and surrounding mol-
time t does not depend on the coagulant concentration (r ¼ 0.00, p ecules. These water structures are universally observed in aqueous
< 0.01) but does depend on the heating time (r ¼ 0.90, p < 0.01). complex systems (Donth, 2013), and the structure formation is
That is, the mobility of the water decreases with increasing heating called slow dynamics. In other words, the b values reflect a fluc-
time. It is known that the breaking stress increase exponentially tuation of the dynamic behavior of water molecules restricted in
with heating time (Kohyama et al., 1995a) due to the creation of the pores with inhomogeneous sizes, which depend on the coag-
additional cross-linking points, and the dielectric relaxation time of ulant concentration. This heterogeneity is supported by a SEM
water reflects this phenomenon.
 T. Kawaguchi et al. / Food Hydrocolloids 77 (2018) 474e481
Fig. 6. Dielectric relaxation parameters obtained via the curve fitting procedure for
tofu gels with heating times of (open circle) 10 min, (open triangle) 30 min, and (open
square) 60 min. The error bars shows the standard error for multiple measurements.
study (Onodera et al., 2009) indicating that increases in coagulation
resulted in various pore sizes.
The dielectric relaxation strength Dε is directly related with free
water content as discussed as below section. However, the rela-
tionship between the dielectric relaxation strength Dε and the
coagulant concentration with the heating time was not confirmed.
In other words, it is difficult to characterize the tofu gel structure
and the water dynamics from mere free water contents.
3.3. Complementary analysis
The observation timescales of PFG-NMR and the DS are
completely different; translational diffusion was observed at 20 ms
the dielectric relaxation was observed at approximately 10e100 ps.
479
Therefore, it is interesting to confirm the correlation between these
data (Maruyama et al., 2014). Typical complementary analyses us-
ing both experimental techniques are shown in Figs. 7 and 8 with
the physical quantities listed in Table 1.
Fig. 7 shows the free water content dependence of the diffusion
coefficient of the water molecules in the tofu gels. The free water
content of the tofu gels was calculated from the ratio of the
dielectric strength of the tofu gels and that of pure water, 73.2 at
25+ C (Kaatze, 1989). When the free water content of the soymilk
decreases with increasing soy protein concentration, the diffusion
coefficient of the water molecules decreases. This result is easily
understood as indicating the relationship between the decreasing
free water content and the increasing viscosity. However, this cor-
relation of the molecular diffusion of water with the free water
content was not shown for the tofu gels. Some plots shown in the
upper left region indicates the water content of almost 100%. These
large values are considered to be from syneresis due to the direct
contact between the gel and electrode surfaces used in the
dielectric spectroscopy measurements for gel. In this case, the
electrode surface is surrounded by water. This difficulty is an
essential issue of the evaluation system with the contact type
electrode for gel materials. Therefore, we analyzed the present data
without the two plots, since other plots show reasonable values
which means that the measurements were properly conducted.
Here the correlation coefficients for the soymilk and the tofu gel are
0.99 and 0.18, which are shown as the solid and dashed lines in
Fig. 7, respectively. These results suggest that it is not sufficient to
characterize the water translational dynamics of the tofu gel via
free water contents.
On the other hand, the dielectric relaxation time broadening
parameter b shows a linear correlation with the inverse of the
diffusion coefficient 1=D as shown in Fig. 8. There exist the same
types of tendencies for soymilk and tofu gels. Such tendency im-
plies that the distribution of relaxation times broadens when the
diffusion coefficient for water molecules is decreased by dynamical
restrictions. Interestingly, the correlation of dielectric relaxation
time t with diffusion coefficient D was not found. It is reasonable to
Fig. 7. The relationship between the diffusion coefficient and the free water content
for () soymilk and tofu gels with coagulant concentrations of (triangle) 0.4 wt%,
(square) 0.5 wt%, (diamond) 0.6 wt%, and (circle) 0.7 wt%. The open and closed symbols
indicate the commercial soymilk samples and those prepared using dissolved SPI,
respectively. The solid and dashed lines indicate the regression analyses for the soy-
milk and tofu gels, respectively. The error bars show the standard error for multiple
measurements.
480 T. Kawaguchi et al. / Food Hydrocolloids 77 (2018) 474e481
Fig. 8. The 1=D - b plot for () soymilk and tofu gels with coagulant concentrations of
(triangle) 0.4 wt%, (square) 0.5 wt%, (diamond) 0.6 wt%, and (circle) 0.7 wt%. The open
and closed symbols indicate gels made from commercial soymilk and dissolved SPI,
respectively. The solid and dashed lines indicate the regression analyses for the soy-
milk and tofu gels, respectively. The error bars show the standard error for multiple
times measurements.
consider each different rate determining fashion for the trans-
lational or rotational dynamics, although both values of D and t
represent the average value of the characteristic properties. The
rotational dynamics observed by the DS method as 20 GHz fre-
quency process generally reflect more local interactions than those
for the translational diffusion process observed by the NMR
method. The length scale of inter molecular interactions reflected
in the translational diffusion process corresponds to the time scale
of observation, for example 20 ms. On the other hand, the length
scale for the rotational dynamics is expressed by a region of
cooperative motions. Especially, in the case of water, the length
scale is apparently larger than usual expectations, since the length
is extended by the hydrogen-bonding network (Shinyashiki, Imoto,
& Yagihara, 2007). Authors suggested a simple picture of molecular
behaviors for water molecules in aqueous materials indicating
dielectric behaviors in which water molecules apparently exhibit as
large molecules (Yagihara et al., 2005). These considerations sup-
port that the b values imply quite large scale interactions of dy-
namic properties, comparing with the average value, t, and this is
the reason why D values well correspond to b.
It is also meaningful to confirm the difference between the
tendencies for soymilk and tofu gels. In Fig. 8, the solid and dashed
lines indicate the results of the linear regression analyses for the
soymilk and the tofu gel; these correlation coefficients are
0.87
and
0.95, respectively, which indicates a significant difference, as
shown in Table 2. The former tendency is caused by the restricted
dynamics resulting from the increase in the viscosity. However, the
latter tendency is caused by the structural formation of the tofu gels
depending upon the heating time and the coagulant concentration.
Even if the different structures of the gels and solutions cannot be
distinguished using only 1=D, which is the average value of the
dynamic behavior of the water molecules, a complementary anal-
ysis with b will reflect the dispersion of the dynamic behavior and
can characterize these structures. In this case, the b values reflects
the homogeneity of the tofu gel structure via the dynamic behavior.
Accordingly, the b
1=D diagram provides a characterization of the
average dynamic and structural behavior and the heterogeneity.
Furthermore, the present analysis and results support previous
reports of other measurements of mechanical properties and mi-
crostructures (Kohyama et al., 1995a; Onodera et al., 2009).
Therefore, this cooperative analysis is effective to understand and
characterize tofu and other gel structures.
4. Conclusions
In order to perform DS and PFG-NMR complementarily, rotation
and translational dynamics for water molecules were discussed in
different time scales. The dielectric relaxation time, t, and the
dielectric relaxation time broadening parameter, b, for water were
dependent on the heating time and the coagulant concentration,
respectively. The t and b are considered to reflect the tofu gel
breaking stress and microstructure homogeneity, respectively. The
diffusion coefficient, D, for water shows the same pattern with the b
and was affected by the tofu gel structure. Comparing the relation
of the b and the 1=D, distinctive correlation coefficient for the soy
milk and the tofu gel were found as
0.87 and
0.95, respectively.
It is concluded that, the water dynamics obtained by the DS and
PFG-NMR measurements could be applied as a new evaluation
method for gel and gel like foods.
Acknowledgement
We thank Fuji Oil Co., Ltd. for kindly providing SPI samples. The
series of NMR measurement were performed by Tokai University
Technology Joint Management Office facilities. This work was
supported by JSPS KAKENHI Grant Number 15K13554.
References
Abe, F., Nishi, A., Saito, H., Asano, M., Watanabe, S., Kita, R., et al. (2017). Dielectric
study on hierarchical water structures restricted in cement and wood materials.
Measurement Science and Technology, 28(4), 044008.
Cai, T., & Chang, K. (1997). Dry tofu characteristics affected by soymilk solid content
and coagulation time. Journal of Food Quality, 20(5), 391e402.
Cheng, Y., Shimizu, N., & Kimura, T. (2004). Rheological properties of mgcl2-
coagulated tofu and conditions for hard tofu production in China. Journal of
the Japanese Society of Agricultural Machinery, 66(3), 98e104.
Cheng, Y., Shimizu, N., & Kimura, T. (2005). The viscoelastic properties of soybean
curd (tofu) as affected by soymilk concentration and type of coagulant. Inter-
national Journal of Food Science & Technology, 40(4), 385e390.
Chung, K.-C., Yu, H.-Y., & Ahn, S. (2011). Convection effects on pgse-nmr self-
diffusion measurements at low temperature: Investigation into sources of
induced convective flows. Bulletin of the Korean Chemical Society, 32(6),
1970e1974.
Cole, K. S., & Cole, R. H. (1941). Dispersion and absorption in dielectrics i. alternating
current characteristics. The Journal of Chemical Physics, 9(4), 341e351.
Cole, R. H., Mashimo, S., & Winsor I. V, P. (1980). Evaluation of dielectric behavior by
time domain spectroscopy. 3. precision difference methods. The Journal of
Physical Chemistry, 84(7), 786e793.
Damez, J.-L., Clerjon, S., Abouelkaram, S., & Lepetit, J. (2007). Dielectric behavior of
beef meat in the 1e1500khz range: Simulation with the fricke/coleecole model.
Meat Science, 77(4), 512e519.
DeMan, J., DeMan, L., & Gupta, S. (1986). Texture and microstructure of soybean
curd (tofu) as affected by different coagulants. Food Structure, 5(1), 11.
Donth, E.-J. (2013). The glass transition: Relaxation dynamics in liquids and disordered
materials (Vol. 48). Springer Science & Business Media.
Fuchigami, M., & Teramoto, A. (1997). Structural and textural changes in kinu-tofu
due to high-pressure-freezing. Journal of Food Science, 62(4), 828e837.
Grosse, C. (2014). A program for the fitting of debye, coleecole, coleedavidson, and
havriliakenegami dispersions to dielectric data. Journal of Colloid and Interface
Science, 419, 102e106.
Holz, M., Heil, S. R., & Sacco, A. (2000). Temperature-dependent self-diffusion co-
efficients of water and six selected molecular liquids for calibration in accurate
1h nmr pfg measurements. Physical Chemistry Chemical Physics, 2(20),
4740e4742.
Hong, Y. S., & Lee, C. H. (2006). Self-diffusion coefficient of water in tofu determined
by pulsed field gradient nuclear magnetic resonance. Journal of Agricultural and
Food Chemistry, 54(1), 219e223.
Huang, Y.-C., & Kuo, M.-I. (2015). Rheological characteristics and gelation of tofu
made from ultra-high-pressure homogenized soymilk. Journal of Texture Studies,
46(5), 335e344.
Iwashita, T., Konuma, T., Harada, E., Mori, S., & Sugase, K. (2016). Use of glass cap-
illaries to suppress thermal convection in nmr tubes in diffusion
 T. Kawaguchi et al. / Food Hydrocolloids 77 (2018) 474e481
measurements. Magnetic Resonance in Chemistry, 54(9), 729e733.
Kaatze, U. (1989). Complex permittivity of water as a function of frequency and
temperature. Journal of Chemical and Engineering Data, 34(4), 371e374.
Kaatze, U. (2007). Reference liquids for the calibration of dielectric sensors and
measurement instruments. Measurement Science and Technology, 18(4), 967.
Kohyama, K., Murata, M., Tani, F., Sano, Y., & Doi, E. (1995a). Effects of protein
composition on gelation of mixtures containing soybean 7s and 11s globulins.
Bioscience, Biotechnology, and Biochemistry, 59(2), 240e245.
Kohyama, K., Sano, Y., & Doi, E. (1995b). Rheological characteristics and gelation
mechanism of tofu (soybean curd). Journal of Agricultural and Food Chemistry,
43(7), 1808e1812.
Kuo, F.-J., Lien, C.-C., Huang, Y.-Y., & Ting, C.-H. (2011). Use of ultrasound for
measuring tofu texture. Engineering in Agriculture, Environment and Food, 4(3),
83e89.
Levy, E., Puzenko, A., Kaatze, U., Ishai, P. B., & Feldman, Y. (2012). Dielectric spectra
broadening as the signature of dipole-matrix interaction. ii. water in ionic so-
lutions. The Journal of Chemical Physics, 136(11), 114503.
Li, T., Rui, X., Li, W., Chen, X., Jiang, M., & Dong, M. (2014). Water distribution in tofu
and application of T2 relaxation measurements in determination of tofu’s
water-holding capacity. Journal of agricultural and food chemistry, 62(34),
8594e8601.
Li, T., Rui, X., Wang, K., Jiang, M., Chen, X., Li, W., et al. (2015). Study of the dynamic
states of water and effects of high-pressure homogenization on water distri-
bution in tofu by using low-field nuclear magnetic resonance. Innovative Food
Science & Emerging Technologies, 30, 61e68.
Maruyama, Y., Numamoto, Y., Saito, H., Kita, R., Shinyashiki, N., Yagihara, S., et al.
(2014). Complementary analyses of fractal and dynamic water structures in
proteinewater mixtures and cheeses. Colloids and Surfaces a: Physicochemical
and Engineering Aspects, 440, 42e48.
Mashimo, S., Kuwabara, S., Yagihara, S., & Higasi, K. (1987). Dielectric relaxation
time and structure of bound water in biological materials. Journal of Physical
Chemistry, 91(25), 6337e6338.
Messina, M. J., Persky, V., Setchell, K. D. R., & Barnes, S. (1994). Soy intake and cancer
risk: A review of the in vitro and in vivo data. Nutrition and Cancer, 21(2),
113e131.
Miura, N., Yagihara, S., & Mashimo, S. (2003). Microwave dielectric properties of
solid and liquid foods investigated by time-domain reflectometry. Journal of
Food Science, 68(4), 1396e1403.
Nagano, T., Akasaka, T., & Nishinari, K. (1995). Study on the heat-induced confor-
mational changes of b-conglycinin by ftir and cd analysis. Food Hydrocolloids,
9(2), 83e89.
Nelson, S. O., & Trabelsi, S. (2006). Dielectric spectroscopy of wheat from 10 mhz to
1.8 ghz. Measurement Science and Technology, 17(8), 2294.
Nishinari, K., Fang, Y., Guo, S., & Phillips, G. (2014). Soy proteins: A review on
composition, aggregation and emulsification. Food Hydrocolloids, 39, 301e318.
Onodera, Y., Tomotada, O., & Nakasato, K. (2009). Homogeneity and microstructure
of tofu depends on 11s/7s globulin ratio in soymilk and coagulant concentra-
tion. Food Science and Technology Research, 15(3), 265e274.
Peng, X., Ren, C., & Guo, S. (2016). Particle formation and gelation of soymilk: Effect
481
of heat. Trends in Food Science & Technology, 54, 138e147.
Prabhakaran, M. P., Perera, C. O., & Valiyaveettil, S. (2006). Effect of different co-
agulants on the isoflavone levels and physical properties of prepared firm tofu.
Food Chemistry, 99(3), 492e499.
Prestamo, G., Lesmes, M., Otero, L., & Arroyo, G. (2000). Soybean vegetable protein
(tofu) preserved with high pressure. Journal of Agricultural and Food Chemistry,
48(7), 2943e2947.
Price, W. S., & Walchli, M. (2002). Nmr diffusion measurements of strong signals:
The pgse-q-switch experiment. Magnetic Resonance in Chemistry, 40(13),
S128eS132.
Ryabov, Y. E., Feldman, Y., Shinyashiki, N., & Yagihara, S. (2002). The symmetric
broadening of the water relaxation peak in polymerewater mixtures and its
relationship to the hydrophilic and hydrophobic properties of polymers. The
Journal of Chemical Physics, 116(19), 8610e8615.
Shinyashiki, N., Imoto, D., & Yagihara, S. (2007). Broadband dielectric study of dy-
namics of polymer and solvent in poly (vinyl pyrrolidone)/normal alcohol
mixtures. The Journal of Physical Chemistry B, 111(9), 2181e2187.
Shinyashiki, N., Yamamoto, W., Yokoyama, A., Yoshinari, T., Yagihara, S., Kita, R., et al.
(2009). Glass transitions in aqueous solutions of protein (bovine serum albu-
min). The Journal of Physical Chemistry B, 113(43), 14448e14456.
Slichter, C. P. (2013). Principles of magnetic resonance (Vol. 1). Springer Science &
Business Media.
Stejskal, E. O., & Tanner, J. E. (1965). Spin diffusion measurements: Spin echoes in
the presence of a time-dependent field gradient. The Journal of Chemical Physics,
42(1), 288e292.
Tang, C.-H., Chen, L., & Foegeding, E. A. (2011). Mechanical and water-holding
properties and microstructures of soy protein isolate emulsion gels induced
by cacl2, glucono-d-lactone (gdl), and transglutaminase: Influence of thermal
treatments before and/or after emulsification. Journal of Agricultural and Food
Chemistry, 59(8), 4071e4077.
Tanner, J. E. (1970). Use of the stimulated echo in nmr diffusion studies. The Journal
of Chemical Physics, 52(5), 2523e2526.
Urbonaite, V., De Jongh, H., Van Der Linden, E., & Pouvreau, L. (2015). Water holding
of soy protein gels is set by coarseness, modulated by calcium binding, rather
than gel stiffness. Food Hydrocolloids, 46, 103e111.
Wang, H., Swain, E., Kwolek, W., & Fehr, W. (1983). Effect of soybean varieties on the
yield and quality of tofu. Cereal Chemistry, 60(3), 245e248.
Watanabe, K., Taka, Y., & Fujiwara, O. (2009). Cole-cole measurement of dispersion
properties for quality evaluation of red wine. Measurement Science Review, 9(5),
113e116.
Wu, A. H., Ziegler, R. G., Horn-Ross, P. L., Nomura, A., West, D. W., Kolonel, L. N., et al.
(1996). Tofu and risk of breast cancer in asian-americans. Cancer Epidemiology
and Prevention Biomarkers, 5(11), 901e906.
Yagihara, S., Asano, M., Kosuge, M., Tsubotani, S., Imoto, D., & Shinyashiki, N. (2005).
Dynamical behavior of unfreezable molecules restricted in a frozen matrix.
Journal of Non-crystalline Solids, 351(33), 2629e2634.
Yagihara, S., Oyama, M., Inoue, A., Asano, M., Sudo, S., & Shinyashiki, N. (2007).
Dielectric relaxation measurement and analysis of restricted water structure in
rice kernels,. Measurement Science and Technology, 18(4), 983.