Available online at www.sciencedirect.

com
Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756
Effect of Arabic gum, xanthan gum and orange oil contents on -potential,
conductivity, stability, size index and pH of orange beverage emulsion
Hamed Mirhosseini
a
, Chin Ping Tan
a,
, Nazimah S.A. Hamid
b
, Salmah Yusof
c
a
Department of Food Technology, Faculty of Food Science and Technology, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia
b
Department of Food Science, Faculty of Food Science and Technology, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia
c
Faculty of Science and Technology, Islamic Science University of Malaysia, 71800 Nilai, Negeri Sembilan, Malaysia
Received 2 April 2007; received in revised form 1 July 2007; accepted 10 July 2007
Available online 12 July 2007
Abstract
The main and interaction effects of main emulsion components namely Arabic gum content (1320%, w/w, x
1
), xanthan gum content (0.30.5%,
w/w, x
2
) and orange oil content (1014%, w/w, x
3
) on beverage emulsion characteristics were studied using the response surface methodology
(RSM). The physicochemical properties considered as response variables were: -potential (Y
1
), conductivity (Y
2
), emulsion stability (Y
3
), size
index (Y
4
) and pH (Y
5
). The results indicated that the response surface models were signicantly (p <0.05) tted for all response variables studied.
In contrast with -potential and pH, the independent variables had the most signicant (p <0.05) effect on size index. Regression models describing
the variations of the responses variables showed high coefcient of determination (R
2
) values ranging from 0.866 to 0.960. The main effect of
Arabic gum followed by its interaction with orange oil was observed to be signicant (p <0.05) in most of response surface models. Therefore, the
concentration of Arabic gum should be considered as a critical variable for the formulation of orange beverage emulsion in terms of the emulsion
characteristics studied. The overall optimum region resulted in a desirable orange beverage emulsion was predicted to be obtained by combined
level of 10.78% (w/w) Arabic gum, 0.24% (w/w) xanthan gum and 12.43% (w/w) orange oil. No signicant (p >0.05) difference was found
between the experimental and predicted values, thus ensuring the adequacy of the response surface models employed for describing the changes
in physicochemical properties as a function of main emulsion component contents.
2007 Elsevier B.V. All rights reserved.
Keywords: Arabic gum; Xanthan gum; Beverage emulsion; -Potential; Conductivity; Emulsion stability
1. Introduction
The perceived quality of emulsion-based food products, such
as milk, cream, sauces and beverages can be determined by the
emulsion microstructure or main constituents in a food emulsion
which interact with each other, either physically or chemically
[1,2]. Due to the non-thermodynamic nature of emulsions, the
formulation of a specic emulsion microstructure and speci-
cation of emulsion characteristics remains difcult to predict
Abbreviations: RSM, response surface methodology; CCD, central com-
posite design; R
2
, coefcient of determination; O/W, oil-in-water; DLVO,
DerjaguinLandauVerweyOverbeek; Min, minute; -potential, zeta potential;
HE, initial height of emulsion height; HC, height of cream layer; HS, height of
the sedimentation phase; ES, emulsion stability index; ANOVA, analysis of
variance; Eq., equation; 3D, three-dimensional

Corresponding author. Tel.: +603 89468418; fax: +603 89423552.

E-mail address: tancp@putra.upm.edu.my (C.P. Tan).
[3]. The effect of emulsion components on the physicochemical
properties of emulsions has been studied by many researchers
[46].
Beverage emulsions are oil-in-water (O/W) emulsions com-
prising two categories: avor/cloud emulsions and cloud
emulsions. Both types of beverage emulsions must have a sig-
nicant stability in both concentrated and diluted forms in the
nished emulsion-based products (soft drinks) [7]. The most
common manifestations of beverage emulsions deterioration are
formation of a whitish ring and/or shiny oil slick around the neck
of the container. Both emulsion defects are the result of a variety
of physiochemical mechanisms that occur within the beverage
emulsion, including creaming, occulation and coalescence [8].
An emulsion is traditionally dened as a dispersion of
droplets of one liquid in another, the two being immiscible
[2]. Emulsions and colloidal dispersions are thermodynamically
unstable systems from a physicochemical point of view, rapidly
or slowly separating into two immiscible phases over a period
0927-7757/$ see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.colsurfa.2007.07.007
48 H. Mirhosseini et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756
Nomenclature
w/w weight/weight
equal or more
% percent

C centigrade
min minute
MPa mega Pascal
mV milivolt
mS/cm microsiemens per centimeter
: per
cm centimeter
nm nanometer
< less
> more
of time [9] and are stabilized by improvement of their kinetic
stability [10], where stability may be dened as the resistance
to physical changes [11]. Stabilization is usually achieved by
adding small surfactant molecules (e.g. polysorbates, phospho-
lipids) and/or proteins (e.g. milk proteins) and/or thickening
agents (gums, gelatin) to the emulsion [12]. Physical destabiliza-
tion mechanisms of emulsions include the variation processes of
emulsion droplet size [3]. The breakdown of an emulsion may
manifest itself through the mechanisms, including: creaming,
occulation, ostwald ripening (partial) coalescence, and phase
inversion [8,13].
The stability of an emulsion is also governed by the
relative magnitude of attractive e.g. van der Waals force
and repulsive (e.g. electrostatic, steric, and hydration forces)
interactions between the emulsion droplets [8]. Oil-in-water
emulsions are usually stabilized by means of electrostatic
repulsion between similarly charged oil droplets as predicted
by DerjaguinLandauVerweyOverbeek (DLVO) theory [14].
However, the DLVOtheory takes into account only two kinds of
molecular interactions: electrostatic and van der Waals [15]. The
repulsive electrostatic interactions between similarly charged
emulsion droplets do not allow them to get as close together as
uncharged droplet and the cloud of counterions surrounding a
droplet moves less slowly than the droplet itself [16]. The -
potential can be used as a measure of the electrostatic repulsive
forces, which can give an indication of the potential stability of
the colloidal system.
Despite a large number of studies performed on avor/cloud
emulsion, there is a lack of sufcient knowledge on the for-
mulation of a desirable orange beverage emulsion in terms of
the critical physicochemical properties. The present study was
conducted to investigate the effect of proportion of main emul-
sion components (namely Arabic gum, xanthan gumand orange
oil) on physicochemical properties of orange beverage emul-
sion. In the present study, response surface methodology (RSM)
was applied for modeling the possible relationships between the
response and independent variables. Response surface analysis
can provide the response regression models for rapidly estimat-
ing the overall variation of physicochemical emulsion properties
as a function of main emulsion components in a relatively short
time. In addition, the present study was conducted to optimize
a desirable formulation for the orange beverage emulsion with-
out using the unhealthy synthetic compounds such as weighting
agents. The main objective of present study was to determine
an optimum level of independent variables resulting in (1) max-
imum emulsion stability, (2) minimum conductivity and size
index, (3) the largest magnitude of -potential and (4) a suitable
pH value depending on the other response goals. Therefore, a
three-factor central composite design (CCD) was used to study
the effect of three independent variables namely Arabic gum
content (1320%, w/w), xanthan gum content (0.30.5%, w/w)
and orange oil content (1014%, w/w) on the response vari-
ables studied. It should be noted that other critical parameters
such as turbidity, average droplet size, polydispersity index,
absorbance, viscosity, ow behavior, density and avor release
of orange beverage emulsion have also been investigated in our
other studies.
2. Materials and methods
2.1. Materials
Arabic gumwas provided by Colloides Naturels International
Co. (Rouen, France). Xanthan gum was donated by CP Kelco
(Chicago, USA). Citric acid, sodium benzoate and potassium
sorbate (95%) were purchased from Fisher Scientic (Pitts-
burgh, PA). Valencia cold pressed orange oil was provided by
Danisco (Danisco Cultor, Aarhus, Denmark).
2.2. Preparation of orange beverage emulsion
As shown in Table 2, 20 orange beverage emulsions that com-
posed of gum Arabic (1320%, w/w), xanthan gum (0.30.5%,
w/w), orange oil (1014%, w/w), sodiumbenzoate (0.1%, w/w),
potassium sorbate (0.1%, w/w), citric acid (0.4%, w/w) and
deionized water were prepared for the optimization procedure
based on a three factor CCD. As mentioned in our previous study
[17,18]. To prepare the water phase, sodium benzoate, potas-
sium sorbate and citric acid were dispersed in deionized water
(60

C) using a high speed blender (Waring blender 32BL80,

New Hartford, CO, USA). While mixing the mixture, Arabic
gum was gradually added to the deionized water (60

C) and
mixed for 3 min to facilitate hydration. The Arabic gum solu-
tion was kept overnight at room temperature to fully hydrate.
Xanthan gum solution was prepared separately by dissolving
xanthan gum in deionized water and then mixed with Arabic
gum solution by using a high speed blender. While mixing
the water phase, the cold pressed orange oil was gradually
added into water phase to provide an initial coarse emulsion
[1719]. Fine emulsication (i.e. small average droplet size
of <1 m with narrow particle size distribution) was achieved
by subjecting the pre-emulsions to pre-homogenization using
a high shear blender (Silverson L4R, Buckinghamshire, UK)
for 1 min and then passed through a high pressure homoge-
nizer (APV, Crawley, U.K.) for three passes (30, 28 and 25 MPa)
[1719].
H. Mirhosseini et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756 49
2.3. Analytical methods
2.3.1. Emulsion stability (creaming test)
In the present study, emulsion stability was determined
by measuring the extent of gravitational phase separation.
Creaming and sedimentation values were determined from
the ratio of cream and sediment volumes over total vol-
ume of emulsion samples up on standing [6]. For this
test, 15 mL of a prepared beverage emulsion was trans-
ferred into 20 mL test tube and stored for 2 weeks at room
temperature (25 1

C). Emulsion stability index (ESI) was

calculated as percentage of the initial emulsion height (HE),
height of cream layer (HC) and height of the sedimenta-
tion phase (HS):ESI(%) =100 (HE(HS+HC))/HE [2022].
In the experimental data, ESI was only shown to clarify
the stability of 20 orange beverage emulsions. The measure-
ment was performed in duplicate samples. Creaming test has
been used to determine emulsion stability in previous studies
[6,23].
2.3.2. -Potential and conductivity
The orange beverage emulsions were diluted (1:100) for the
measurement of -potential and conductivity using a Malvern
zeta sizer (Malvern series ZEN3500, Malvern Instruments Ltd.,
Worcester, UK). The measurement of zeta potential is calculated
based on the Henry equation as follow:
= 1.5 U
E

f(K
a
)
(1)
where is -potential calculating fromthe relative magnitude of
electrophoretic mobility (U
E
), viscosity (), dielectric constant
and Henry function (f(K
a
)). The electrophoresis experiment is
measuring the velocity of the particles in an electric eld using
laser Doppler velocimetry (LDV) technique. -potential is also
measured using a combination of the laser Doppler velocimetry
and electrophoresis (called laser Doppler electrophoresis). This
method measures how fast a particle moves in a liquid when
an electrical eld is applied. In fact, the essence of a classical
micro-electrophoresis system is a cell with electrodes at either
end to which a potential is applied. When the charged droplets
are suspended in a liquid, they move towards the electrode of
opposite charge, their velocity is measured and expressed in
unit eld strength as their mobility. The measurements of -
potential and conductivity were carried out immediately after
emulsion preparation using a maintenance-free capillary cell
and deionized water for diluting the sample in order to avoid
multiple scattering effects. In order to avoid the presence of air
bubbles, the lling was done using a 1 mL syringe. The cap-
illary cell was lled with diluted beverage emulsion (1:100)
between 7 and 10 mm; then fully inserted into the module. In
most cases, the instrument was equilibrated after 25 s especially
in term of the stabilized temperature and became ready for the
measurements. The measurement ranges of -potential and con-
ductivity appear in the units of mV and mS/cm, respectively.
For data analysis, the measurements were reported as the aver-
age of three separate injections, with four readings made per
injection.
2.3.3. Size index
The measurement of size index was carried out using a
UVvis spectrophotometer (Spectronic Genesys
TM
10, GENEQ
inc., Montreal, Canada). The orange beverage emulsion were
diluted (1:100) and then contained in quartz cuvettes with a 1 cm
path length. Spectra were obtained over the wavelength range
of 1901100 nm. The absorbance readings were performed in
triplicate at two wavelengths 800 and 400 nm. The ratio of
absorbance at 800 nm over 400 nm is dened as the size index.
Distilled water was used as a reference. The measurements were
made using a standard single-beam arrangement, grating-based,
double detectors. Wavelength accuracy of the instrument was
found to be approximately 1.0 nm with wavelength repeata-
bility equal to 0.05 nm. This parameter has been studied for
emulsions by previous researchers [24,25].
2.3.4. pH
The pH values of orange beverage emulsions were measured
by means of a glass pH electrode (Mettler Toledo, Schwerzen-
bach, Switzerland).
2.4. Experimental design and statistical analysis
The effect of three independent variables namely Arabic
gum content, xanthan gum content and orange oil content on -
potential, conductivity, emulsion stability, size index and pH of
orange beverage emulsion was studied using a three factor CCD.
Twenty orange beverage emulsions were established based on
the CCD with three independent variables at ve levels of each
variable (Tables 1 and 2). The center point was repeated six
times to calculate the repeatability of the method [26]. Multiple
regressions were applied for predicting the linear, quadratic and
interaction terms of independent variables in the response sur-
face models. The generalized polynomial model for relating the
response to independent variables is given below:
Y =
0
+

i
x
i
+

ii
x
2
i
+

ij
x
i
x
j
(2)
Table 1
Levels of independent variables established according to the central composite design (CCD)
Variable (independent variables) Independent variables level
Low Center High Axial () Axial (+)
Arabic gum content (w/w%) 13.00 16.50 20.00 10.78 22.22
Xanthan gum content (w/w%) 0.30 0.40 0.50 0.24 0.56
Orange oil content (w/w%) 10.00 12.00 14.00 8.73 15.27
50 H. Mirhosseini et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756
Table 2
Matrix of the central composite design (CCD)
Treatment runs Blocks Arabic gum Xanthan gum Orange oil
1 1 20.00 0.50 10.00
2 1 13.00 0.50 14.00
3 1 13.00 0.30 10.00
4(C) 1 16.50 0.40 12.00
5 1 20.00 0.30 14.00
6(C) 1 16.50 0.40 12.00
7 2 13.00 0.50 10.00
8 2 20.00 0.50 14.00
9 2 20.00 0.30 10.00
10 2 13.00 0.30 14.00
11(C) 2 16.50 0.40 12.00
12(C) 2 16.50 0.40 12.00
13(C) 3 16.50 0.40 12.00
14(C) 3 16.50 0.40 12.00
15 3 16.50 0.24 12.00
16 3 10.78 0.40 12.00
17 3 16.50 0.40 15.27
18 3 16.50 0.40 8.73
19 3 22.22 0.40 12.00
20 3 16.50 0.56 12.00
C, center point.
where Y is response calculated by the model;
0
is a constant;

i
,
ii
and
ij
are linear, squared and interaction coefcients,
respectively. The adequacy of the models was determined using
model analysis, lack-of t test and coefcient of determination
(R
2
). Joglekar and May [27] suggested that R
2
should be at least
0.80 for a good t of a model. The corresponding variables will
be more signicant (p <0.05) if the absolute t value becomes
larger and the p-value becomes smaller. Only terms found sta-
tistically signicant (p <0.05) were included in the reduced
models. It should be noted that some independent variable
terms were kept in the reduced model despite non-signicance
(p >0.05). For example, linear terms were kept in the model
if a quadratic or interaction term containing this variable was
signicant (p <0.05). The experimental design matrix, data
analysis and optimization procedure were performed using the
Minitab v. 13.2 statistical package (Minitab Inc., Pennsylvania,
USA).
2.5. Optimization and verication procedures
The optimal level of three independent variables (x
1
, x
2
and
x
3
) which led to the desirable response goals was determined by
graphical and numerical optimization procedures. The graphical
technique was employed in order to deduce workable optimum
conditions [28]. For graphical optimization, a three-dimensional
(3D) response surface was plotted by keeping one variable con-
stant at the center point and varying the other two variables
within the experimental range. It should be noted that the 3D
plotting was performed only for the signicant (p <0.05) inter-
action effects. A numerical optimization procedure was also
carried out by the response optimizer using the Minitab soft-
ware for determining the exact optimum value of individual and
simultaneous multiple response optimizations resulting in the
desirable response goals. Experimental data was compared with
the tted values predicted by the models in order to verify the
adequacy of the regression models.
3. Results and discussion
3.1. Response surface analysis
Experimental data obtained for the response variables studied
are given in Table 3. Application of response surface method-
ology (RSM) allowed simultaneously determining main and
Table 3
Experimental data obtained for the response variables studied (Y
i
) (mean S.D.)
Run Block Response variable (mean S.D.)
-Potential (Y
1
, mV) Conductivity (Y
2
, mS/cm) Emulsion stability (Y
3
, %) Size index (Y
4
) pH (Y
5
)
1 1 27.0 1.1 0.114 0.007 88.6 1.3 0.763 0.010 3.97 0.06
2 1 29.4 0.7 0.111 0.001 100.0 0.0 0.800 0.006 3.88 0.02
3 1 28.8 2.4 0.087 0.004 90.7 0.8 0.710 0.017 3.85 0.01
4 1 28.1 1.0 0.102 0.009 92.5 2.1 0.772 0.000 3.91 0.03
5 1 27.7 0.2 0.096 0.008 89.9 1.1 0.851 0.009 3.97 0.00
6 1 28.3 0.2 0.100 0.008 90.6 0.4 0.759 0.012 3.93 0.03
7 2 29.2 1.2 0.094 0.005 92.1 0.7 0.814 0.002 3.85 0.00
8 2 29.2 0.7 0.103 0.014 90.2 0.5 0.816 0.007 3.97 0.04
9 2 26.8 1.3 0.179 0.009 86.1 0.8 0.762 0.011 3.97 0.02
10 2 29.2 1.0 0.093 0.008 100.0 0.0 0.728 0.001 3.87 0.04
11 2 27.8 0.4 0.116 0.011 89.7 2.6 0.782 0.020 3.95 0.00
12 2 28.3 0.7 0.095 0.007 88.6 0.3 0.774 0.017 3.92 0.01
13 3 27.9 0.4 0.102 0.011 88.9 1.4 0.768 0.009 3.93 0.00
14 3 28.0 0.8 0.096 0.009 89.5 0.9 0.781 0.003 3.92 0.03
15 3 27.4 0.7 0.101 0.004 88.6 1.2 0.766 0.013 3.93 0.02
16 3 29.5 0.7 0.088 0.013 98.0 0.0 0.752 0.004 3.83 0.01
17 3 28.0 1.3 0.101 0.006 93.7 2.2 0.838 0.007 3.90 0.01
18 3 27.8 0.3 0.108 0.010 82.5 0.6 0.817 0.008 3.94 0.00
19 3 27.9 1.1 0.109 0.006 90.2 1.1 0.800 0.005 4.04 0.04
20 3 30.0 0.7 0.096 0.006 87.3 0.4 0.858 0.012 3.93 0.00
H. Mirhosseini et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756 51
Table 4
Regression coefcients, R
2
, adjusted R
2
, probability values and lack of t for the response resurface models
Regression coefcient -Potential (Y
1
, mV) Conductivity (Y
2
, mS/cm) Emulsion stability (Y
3
, %) Size index (Y
4
) pH (Y
5
)
b
0
47.52 0.1134 70.50 0.8440 3.65311
b
1
1.48 0.0374 4.10 0.0191 0.01656
b
2
28.33 0.2346 41.33 1.3065 0.0075
b
3
0.33 0.0078 7.11 0.1022 0.00057
b
2
1
0.02 0.0001 0.17 0.0004
b
2
2
23.53 0.0992 25.68 0.8156
b
2
3
0.02 0.0008 0.05 0.0035
b
12
0.39 0.0297 0.47 0.0750
b
13
0.04 0.0021 0.21 0.0025
b
23
0.65 0.0513 2.25 0.0419
R
2
0.866 0.906 0.96 0.94 0.932
R
2
(adj.) 0.681 0.778 0.905 0.859 0.908
Regression (p-value) 0.011
a
0.004
a
0.000
a
0.001
a
0.000
a
Lack of t (F-value) 5.93 1.09 2.75 4.15 0.101
Lack of t (p-value) 0.087
b
0.505
b
0.217
b
0.136
b
0.568
b
b
i
: The estimated regression coefcient for the main linear effects. b
ii
: The estimated regression coefcient for the quadratic effects. b
ij
: The estimated regression
coefcient for the interaction effects. 1: Arabic gum; 2: Xanthan gum; 3: Orange oil.
a
Signicant (p <0.05).
b
Not signicant (p >0.05).
interaction effects of main beverage emulsion components on
each response variable studied and offered ve response surface
models relating experimental data to the independent variables.
The variation of each response variable (y
i
) was assessed as a
function of linear, quadratic and interaction terms of Arabic gum
content (x
1
), xanthan gum content (x
2
) and orange oil content
(Table 4). The estimated regression coefcients for the response
variables, along with the corresponding R, R
2
(adj), F-value and
p-value of lack of t are given in Table 4. Regression analysis
and ANOVA were conducted for tting the model relating the
response to independent variables and examining the statisti-
cal signicance of the model terms. The results indicated that
a second order polynomial regression model was tted to all
response variables (Y
i
) except for pH. ANOVA results showed
that the response surface models were signicant (p <0.05) for
all response regressionequations (Table 4). The R
2
values ranged
from 0.866 to 0.960 were obtained for the response variables,
indicating that a high percentage (>86%) of the response vari-
ations was explained by the response surface models (Table 4).
The lack of t (Table 4), which measures the tness of the
model, did not result in a signicant (p <0.05) F-value in case of
response variables studied, indicating that the proposed regres-
sion models are sufciently accurate for describing the variation
of the responses. In general, only the linear effect of Arabic gum
showed a signicant (p <0.05) effect on -potential and pH;
whereas independent effects had the most signicant (p <0.05)
effect on size index (Table 5). The results showed that the lin-
ear effect of Arabic gum followed by its interaction with orange
oil were signicant (p <0.05) on most of the response variables
(Table 5). Conversely, the quadratic term of xanthan gum was
found to be non-signicant (p >0.05) in all response regres-
sion models. It should be noted that the polynomial regression
equation was only a statistical empirical model in the selected
ranges. It may not be true beyond the ranges of the factors.
Therefore, the model cannot be extrapolatedbeyondthese ranges
[26].
3.1.1. -Potential and conductivity
As shown in Table 3, -potential values (Y
1
) of orange
beverage emulsions varied from 26.82 to 29.98 (mV). An
absolute value, less than or greater than 25 mV is indicative
of occulated and deocculated emulsions, respectively [29].
Therefore, no occulation was shown in all orange beverage
emulsions. In general, the change in the -potential was not
large enough to signicantly (p <0.05) affect the interpreta-
tion of the results in the studied ranges of the independent
variables. The main effect of Arabic gum was only signif-
icant (p <0.05) on -potential (Y
1
). Previous study [30] has
also reported that the electrical charge (-potential) of the
emulsion droplets was strongly dependent on nal pH, polysac-
charide type, and polysaccharide concentration. The individual
optimization results showed that the optimum -potential (the
largest negative value, Y
1
=30.73 mV) was estimated to be
obtained by an orange beverage emulsion containing 10.78%
(w/w) Arabic gum, 0.56% (w/w) xanthan gum and 12.09%
(w/w) orange oil. The effect of hydrocolloid concentration was
more pronounced with xanthan gum(negatively charged or ionic
polysaccharide composed of a 14 linked -d-glucose backbone
substituted with a side chain including two mannoses and one
glucuronic acid). As shown in the results, high concentration
of xanthan gum provided large negatively charged -potential.
It may be attributed to the reason that the increase of xanthan
gum content leads to increase the negative surface charge of
emulsion droplets followed by increasing the negatively charged
-potential.
As shown in Table 5, the conductivity value (Y
2
) was sig-
nicantly (p <0.05) inuenced by the main effect of Arabic
gum and all interaction effects of independent variables stud-
ied. The results indicated that the interaction effect of Arabic
gum and orange oil followed by the main effect of Arabic gum
had the most signicant (p <0.05) effects on conductivity value
(Y
2
) (Table 5). The 3D response surfaces was plotted to better
visualize the signicant (p <0.05) interaction effects of inde-
52 H. Mirhosseini et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756
Table 5
F-ratio and p-value for each independent variable effect in the polynomial response surface models
Variables Main effects Quadratic effects Interaction effects
x
1
x
2
x
3
x
2
1
x
2
2
x
2
3
x
1
x
2
x
1
x
3
x
2
x
3
-Potential (Y
1
, mV)
p-value 0.019
*
0.133 0.740 0.126 0.116 0.541 0.448 0.109 0.467
F-ratio 8.570 2.800 0.120 2.920 3.110 0.410 0.640 3.250 0.580
Conductivity (Y
2
, mS/cm)
p-value 0.004
*
0.478 0.675 0.727 0.700 0.251 0.011
*
0.002
*
0.012
*
F-ratio 15.78 0.550 0.190 0.130 0.160 1.530 10.64 20.67 10.34
Emulsion stability (Y
3
, %)
p-value 0.019
*
0.404 0.028
*
0.001
*
0.507 0.593 0.737 0.015
*
0.371
F-ratio 8.570 0.780 7.120 30.99 0.480 0.310 0.120 9.480 0.900
Size index (Y
4
)
p-value 0.244 0.033
*
0.008
*
0.219 0.076 0.008
*
0.001
*
0.009
*
0.142
F-ratio 1.580 6.600 12.50 1.780 4.150 12.33 26.02 11.56 2.650
pH (Y
5
)
p-value 0.000
*
0.861 0.788
F-ratio 190.93 0.032 0.075
x
1
, x
2
and x
3
: The main effect of Arabic gum, xanthan gum and orange oil, respectively. x
2
1
, x
2
2
and x
2
3
: The quadratic effect of Arabic gum, xanthan gum and orange
oil, respectively. x
1
x
2
: The interaction effect of Arabic gum and xanthan gum. x
1
x
3
: The interaction effect of Arabic gum and orange oil. x
2
x
3
: The interaction
effect of xanthan gum and orange oil.
*
Signicant at p <0.05
pendent variables on the conductivity (Fig. 1ac). As shown in
Fig. 1(ac), the presence of curvatures in the conductivity curves
conrmed that the variation of conductivity value (Y
2
) was
explained as a nonlinear function of main emulsion components.
The individual optimum region led to minimum conductivity
value (Y
2
=0.064 mS/cm) was predicted to be achieved by com-
bined level of 10.78% (w/w) Arabic gum, 0.56% (w/w) xanthan
gum and 8.73% (w/w) orange oil. It was found that the same
levels of independent variables (except for orange oil content)
resulted in the largest negatively charged -potential and mini-
mum conductivity.
3.1.2. Emulsion stability
As shown in Table 4, the emulsion stability (Y
3
) showed a
negative correlation with the main effect of Arabic gum. It may
be attributed to the depletion occulation induced by high con-
centration of Arabic gum. The proposed effect appeared to be
stimulated by the high content of xanthan gum (0.3%, w/w).
The orange beverage emulsions containing high concentrations
of Arabic gum and xanthan gum exhibited the high apparent
viscosity and relatively large average droplet size (data not
shown). In this case, there is a large excess of non-absorbed
polysaccharide in the aqueous phase of emulsions prepared
by high concentration of polysaccharide [7]. Under certain
circumstances, non-absorbed biopolymers are most probably
responsible for promoting droplet occulation through a deple-
tion mechanism [8]. Dickinson et al. [31] also reported that
combination of the two surface-active materials can lead to a
marked loss of stability, in some cases. When sufcient surface-
active agent is present to saturate the droplet surface, the increase
of surfactant concentrations can be attributed to depletion oc-
culation caused by excess bulk of two surface-active materials
[3133].
The results indicated that the quadratic term of Arabic gum
had the most signicant (p <0.05) effect on emulsion stability
(Y
3
) (Table 5). Besides the quadratic effect, the main effect of
Arabic gum and its interaction effect with orange oil were also
shown to be signicant (p <0.05) on emulsion stability (Y
3
), thus
ensuring the content of Arabic gum should be considered as a
primary factor for determining the stability of orange beverage
emulsion in the studied independent variable ranges. As also
demonstrated in previous studies [34,35], the stability of o/w
emulsion with regard to occulation, coalescence and creaming,
has proven to be strongly affected by the presence of polysac-
charides. In addition to Arabic gum, the main effect of orange
oil was found to be a signicant (p <0.05) parameter in the vari-
ation of emulsion stability (Table 5). As shown in Fig. 1(d), the
variation of emulsion stability could be explained as a nonlin-
ear function of signicant (p <0.05) interaction effect of Arabic
gum and orange oil. As shown in Fig. 1(d), the emulsion stabil-
ity increased when the content of xanthan gum was increased
within the studied variable ranges.
The highest emulsion stability (Y
3
=100%) was predicted
to be at combined level of 10.78% (w/w) Arabic gum, 0.56%
(w/w) xanthan gum and 12.02% (w/w) orange oil. The pre-
dicted optimum region also conrmed that Arabic gum had a
negative effect on the emulsion stability; while the high concen-
tration of xanthan gum was predicted to provide the optimum
emulsion stability. As shown in our previous study [18], the
stabilizing effect of xanthan gum on o/w emulsions may be
interpreted by its desirable effect on viscosity and pseudoplas-
tic behavior of orange beverage emulsion. Xanthan gum has
a complex structure with high molecular weight including a
large number of free carboxyl groups which provide great water
absorption capacity. When xanthan gum is dispersed in aque-
ous solutions, it undergoes a conformational transition from a
H. Mirhosseini et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756 53
Fig. 1. Response surface plots showing the signicant (p <0.05) interaction effects for conductivity (ac), emulsion stability (d) and size index (e and f).
double helix to a complex aggregates through hydrogen bonds
and polymer entanglement. Because of its complicated network
and entanglements, a little increase of xanthan gum resulted
in increase the apparent viscosity and pseudoplastic behavior
of orange beverage emulsion. As stated by Stokes law, the
velocity of creaming or sedimentation phenomenon is reversely
proportional to the viscosity of water phase. On the other hand,
the increase of a negatively charged polysaccharide like xan-
than gum in the emulsion formulation results in an increase
in negative repulsive forces (electrostatic and steric) between
emulsion droplets. Therefore, the presence of xanthan gum can
reduce the interfacial tension and form a cohesive interfacial
lm around the emulsion droplets thereby retarding emulsion
instability.
3.1.3. Size index and pH
As shown in Table 5, the independent variables had the most
signicant (p <0.05) effect on the variation of size index (Y
4
).
The results indicated that the interaction of Arabic gum and
xanthan gum had the most signicant (p <0.05) effect on size
index (Y
4
). The results clearly showed that the main effects
of xanthan gum and orange oil, quadratic term of orange oil,
and interaction effect of Arabic gum and orange oil also sig-
nicantly (p <0.05) inuenced the size index (Y
4
). As clearly
shown in Fig. 1(e and f), the signicant (p <0.05) relation-
ship between size index and interaction effect of independent
variables was strongly nonlinear. The interaction effect of inde-
pendent variables did not show the same effect pattern at
different concentration level (Fig. 1(e and f)). From the opti-
mization result, combination of 10.78% (w/w) Arabic gum,
0.24% (w/w) xanthan gum and 12.14% (w/w) orange oil was
predicted to provide the desirable size index (the least value,
Y
4
=0.648).
In contrast with other response variables, the variation of pH
was tted by a linear regression equation to the experimental
data. As shown in Table 5, only the main effect of Arabic gum
showed a signicant (p <0.05) positive effect on pH value (Y
5
).
The results indicated that the pH was negatively correlated to
54 H. Mirhosseini et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756
Table 6
Experimental and predicted values for the response variables studied
Run -Potential
a
(Y
1
, mV) Conductivity
a
(Y
2
, mS/cm) Emulsion stability
a
(Y
3
, %) Size index
a
(Y
4
) pH
a
(Y
5
)
Y
0
Y
i
Y
0
Y
i
Y
0
Y
i
Y
0
Y
i
Y
0
Y
i
Y
0
Y
i
Y
0
Y
i
Y
0
Y
i
Y
0
Y
i
Y
0
Y
i
1 27.0 27.6 0.6 0.114 0.111 0.003 88.6 88.7 0.1 0.763 0.771 0.008 3.97 3.98 0.01
2 29.4 29.4 0.1 0.111 0.119 0.008 100.0 99.7 0.3 0.800 0.801 0.001 3.88 3.86 0.02
3 28.8 28.8 0.0 0.087 0.087 0.000 90.7 90.7 0.1 0.710 0.712 0.002 3.85 3.86 0.01
4 28.1 28.0 0.1 0.102 0.098 0.003 92.5 91.0 1.5 0.772 0.764 0.008 3.91 3.92 0.01
5 27.7 27.6 0.1 0.096 0.096 0.000 89.9 91.2 1.3 0.851 0.841 0.010 3.97 3.97 0.00
6 28.3 28.0 0.3 0.100 0.098 0.002 90.6 91.0 0.4 0.759 0.764 0.006 3.93 3.92 0.01
7 29.2 29.4 0.2 0.094 0.093 0.001 92.1 90.6 1.5 0.814 0.829 0.015 3.85 3.87 0.02
8 29.2 29.3 0.1 0.103 0.102 0.001 90.2 90.1 0.1 0.816 0.819 0.003 3.97 3.98 0.01
9 26.8 26.8 0.0 0.179 0.170 0.009 86.1 86.2 0.1 0.762 0.766 0.004 3.97 3.98 0.01
10 29.2 28.7 0.5 0.093 0.095 0.002 100.0 99.7 0.3 0.728 0.724 0.003 3.87 3.86 0.01
11 27.8 28.1 0.4 0.116 0.110 0.006 89.7 90.1 0.4 0.782 0.768 0.014 3.95 3.92 0.03
12 28.3 28.1 0.1 0.095 0.110 0.015 88.6 90.1 1.5 0.774 0.768 0.006 3.92 3.92 0.00
13 27.9 28.0 0.2 0.102 0.097 0.005 88.9 88.8 0.2 0.768 0.786 0.019 3.93 3.93 0.00
14 28.0 28.0 0.0 0.096 0.097 0.001 89.5 88.8 0.7 0.781 0.786 0.005 3.92 3.93 0.01
15 27.4 27.9 0.5 0.101 0.104 0.003 88.6 87.8 0.7 0.766 0.772 0.007 3.93 3.93 0.00
16 29.5 29.7 0.2 0.088 0.082 0.006 98.0 99.2 1.2 0.752 0.745 0.007 3.83 3.83 0.00
17 28.0 28.3 0.3 0.101 0.095 0.006 93.7 93.2 0.4 0.838 0.846 0.008 3.90 3.93 0.03
18 27.8 27.3 0.4 0.108 0.115 0.007 82.5 83.2 0.7 0.817 0.802 0.015 3.94 3.93 0.01
19 27.9 27.6 0.3 0.109 0.116 0.008 90.2 89.3 0.9 0.800 0.799 0.001 4.04 4.02 0.02
20 30.0 29.4 0.5 0.096 0.095 0.002 87.3 88.3 1.0 0.858 0.844 0.014 3.93 3.93 0.00
(Y
0
): Experimental value. (Y
i
): Predicted value. (Y
0
Y
i
): Residue.
a
No signicant (p >0.05) difference between experimental (Y
0
) and predicted value (Y
i
).
the emulsion stability (r =0.578). Conversely, it was directly
proportional to -potential (r =0.574). In fact, the decrease in
pH value was correlated to the increase in emulsion stability
and the magnitude of -potential. As a rule, the increase of pH
leads to decrease the functional efciency of preservatives in the
beverage emulsion. Thus, the least pH value was considered as
anoptimumpHregionfor orange beverage emulsion. The results
demonstrated that the least pH value (Y
5
=3.83) was estimated
to be at set level of 10.78% (w/w) Arabic gum, 0.56% (w/w)
xanthan gum and 15.27% (w/w) orange oil.
3.2. Optimization procedure
Multiple response optimizations were carried out to deter-
mine the optimum levels of independent variables leading
to the desirable response goals. For the graphical interpreta-
tion, 3D response surface plotting was highly recommended
[36]. Thus, the signicant (p <0.05) interaction effects of fac-
tors on the response variables was expressed as 3D response
surface plots to better visualize the variation of physicochem-
ical properties as function of main emulsion components. A
numerical optimization was also carried out for determining
the exact value of multiple response optimizations resulting
in the overall optimum condition. The results indicated that
an orange beverage emulsion containing 10.78% (w/w) Ara-
bic gum, 0.24% (w/w) xanthan gum and 12.43% (w/w) orange
oil was predicted to provide the overall optimumregion in terms
of all physicochemical properties studied. Under the optimum
condition, the corresponding predicted response values for -
potential, conductivity, emulsion stability, size index and pH
were predicted to be 29.58, 0.067, 100, 0.648, and 3.82,
respectively.
3.3. Verication of the models
The adequacy of the response surface equations was checked
by the comparison of experimental and tted values predicted
by the response regression models. The experimental and pre-
dicted values are given in Table 6. No signicant (p >0.05)
difference was found between those values. The experimen-
tal response values were shown to be in agreement with the
predicted ones. Closeness between the experimental and pre-
dicted values conrmed the adequacy of corresponding response
surface models employed for describing the variation of physic-
ochemical emulsion properties as function of main beverage
emulsion components.
4. Conclusions
In general, the present study indicated that the physicochemi-
cal emulsion properties were shown to be signicantly (p <0.05)
inuenced by the proportion of main emulsion components. The
results demonstrated that the behavior of orange beverage emul-
sions containing high contents of two surface active agents was
unpredictably complicated. The results exhibited that the mixed
surface active materials behaved totally different from one and
another. For instance, the negative relationship found between
Arabic gum and emulsion stability could be explained by its
complex interaction with high concentration of xanthan gum
(0.3% w/w). As shown in the optimum regions, low concen-
tration of Arabic gum provided the overall optimum region. As
also stated in previous study [37], emulsions containing Arabic
gum are more susceptible to depletion occulation, especially
when one considers that they are normally used at consider-
ably higher concentrations. Thus, the presence study offered
H. Mirhosseini et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756 55
that the accurate proportion of surfactants should be carefully
determined for the formulation of the orange beverage emul-
sion. Since, the simultaneous addition of high concentrations of
various types of surface active agents may lead to the negative
effect on the stability of nal beverage emulsion. In this study,
response surface analysis led to signicant (p <0.05) regres-
sion models which had no indication of lack of t, thus ensuring
reliable adjustments between the independent and response vari-
ables. The multiple response optimizations showed that set level
of 10.78% (w/w) Arabic gum, 0.24% (w/w) xanthan gum and
12.43% (w/w) orange oil was predicted to provide the overall
optimum region. As shown in the optimum regions, high con-
centration of xanthan gum as an anionic polysaccharide led to
increase the negatively charged -potential and emulsion stabil-
ity. This may be due to the enhancing effect of xanthan gum
on the repulsive interactions between emulsion droplets. Since,
the degree of occulation depends on the balance of attrac-
tive and repulsive interactions. Under certain circumstances, the
droplets will tend to repel each other and there is no tendency to
occulate.
Acknowledgements
The authors wish to thank Prof. Gary Reineccius and Dr. Ali
Reza Taherian for their invaluable guidance and suggestions at
the critical start of this project. The authors would like to express
their appreciation to Dr. Boo Huey Chern for her useful help in
this project. Sincere appreciation is also extended to Mr. Yap
fromSiber Hegner (M) Sdn. Bhd. for his assistance in -potential
and conductivity measurements.
References
[1] J.N. Coupland, D.J. McClements, Lipid oxidation in food emulsions,
Trends Food Sci. Technol. 7 (1996) 8391.
[2] E. Dickinson, G. Stainsby, Colloids in Foods, Applied Science Publishers,
London, 1982.
[3] D.I. Comas, J.R. Wagner, M.C. Tom as, Creaming stability of oil in water
(O/W) emulsions: Inuence of pH on soybean protein lecithin interaction,
Food Hyd. 20 (2006) 990996.
[4] G.D. Chanana, B.B. Sheth, Particle size reduction of emulsions by formu-
lation design-II: effect of oil and surfactant concentration, J. Pharm. Sci.
Technol. 49 (1995) 7176.
[5] K. Lindenstruth, B.W. M uller, Parameters with inuence on the droplet
size of w/o emulsions, Die Pharm. 59 (2004) 187190.
[6] A.R. Taherian, P. Fustier, H.S. Ramaswamy, Effect of added oil and mod-
ied starch on rheological properties, droplet size distribution, opacity
and stability of beverage cloud emulsions, J. Food Eng. 77 (2006) 687
696.
[7] C.T. Tan, Beverage emulsions, in: K. Larsson, S.E. Friberg (Eds.), Food
Emulsions, second ed., Marcel Dekker, New York, 1990, pp. 445478.
[8] D.J. McClements, Food emulsions; principles, practice and techniques,
CRC Press, Boca Raton, 1999.
[9] R.P. Borwankar, L.A. Lobo, D.T. Wasan, Emulsion stability-kinetics of
occulation and coalescence, J. Colloid Interf. Sci. 69 (1992) 135
146.
[10] B.A. Bergenstahl, Q.M. Claesson, Surface forces in emulsions, in: S.E.
Friberg, K. Larsson (Eds.), Food Emulsions, third ed., Marcel Dekker, New
York, 1997.
[11] H. Mulder, P. Walstra, The Milk Fat Globule: Emulsion Science as Applied
toMilkProducts andComparable Foods, Pudoc, Wageningen, Netherlands,
1974.
[12] E. Dickinson, C.C. Woskett, Competitive adsorption between proteins and
small-molecule surfactants in food emulsions, in: R.D. Bee, P. Richmond,
J. Mingins (Eds.), Food Colloids, Royal Society of Chemistry, London,
1989.
[13] P. Walstra, Overviewof emulsion and foamstability, in: E. Dickinson (Ed.),
Food Emulsions and Foams, Royal Society of Chemistry, London, 1987.
[14] S.V. Verbich, S.D. Stanislav, A. Tarovski, O. Holt, O. Saether, J. Sjblom,
Evaluation of stability ratio in oil-in-water emulsions, Colloids Surf. A:
Physicochem. Eng. Aspects 123-124 (1997) 209223.
[15] T. Sato, Stability of dispersions, J. Coat. Technol. 65 (1993) 113121.
[16] C.T. Tan, J. Wu Holmes, Stability of beverage avor emulsions, Perf. Flav.
13 (1988) 2341.
[17] H. Mirhosseini, C.P. Tan, S.A.H. Nazimah, S. Yusof, Optimization the
contents of Arabic gum, xanthan gum and orange oil affecting on turbidity,
cloudiness, average particle size, polydispersity index and density in orange
beverage emulsion, Food Hyd., 2007, doi:10.1016/j.foodhyd.2007.06.011.
[18] H. Mirhosseini, C.P. Tan, S.A.H. Nazimah, S. Yusof, Modeling the rela-
tionship between the main emulsion components and stability, viscosity,
uid Behavior, -potential and electrophoretic mobility of orange beverage
emulsion using response surface methodology, J. Agric. Food Chem., in
press.
[19] H. Mirhosseini, S. Yusof, N.S.A. Hamid, C.P. Tan, Solid-phase microex-
traction for headspace analysis of key volatile compounds in orange
beverage emulsion, Food Chem. 105 (2007) 16591670.
[20] S.J. Radford, E. Dickinson, M. Golding, Stability and rheology of emul-
sions containing sodium caseinate: combined effects of ionic calcium and
alcohol, J. Colloid Interf. Sci. 274 (2004) 673686.
[21] E.A. Makri, G.I. Doxastakis, Study of emulsions stabilized with Phaseolus
Vulgaris or Phaseolus Coccineus with the addition of Arabic gum, locust
bean gum and xanthan gum, Food Hyd. 20 (2006) 11411152.
[22] X. Huang, Y. Kakuda, W. Cui, Hydrocolloids in emulsions: particle size
distribution and interfacial activity, Food Hyd. 15 (2001) 533542.
[23] E. Bylaite, T. Nylander, R. Venskutonis, B. J onsson, Emulsication of
carway essential oil in water by lecithin and -lactoglobulin: Emulsion
stability and properties of the formed oil aqueous interface, Colloids Surf.
B: Biointerf. 20 (2001) 327340.
[24] E. Du zewska, A. Stobiecka, M. Maszewska, Effect of oil phase concen-
tration on rheological properties and stability of beverage emulsions, Acta
Sci. Pol. Technol. Aliment. 5 (2006) 147156.
[25] G.S. Myung, H.J. Sung, Y.K. Jong, D.K. Jong, Rapid evaluation of water-
in-oil (w/o) emulsion stability by turbidity ratio measurements, J. Colloid
Interf. Sci. 230 (2000) 213215.
[26] D.C. Montgomery, Design and Analysis of Experiments, fth ed., Wiley,
New York, 2001.
[27] A.M. Joglekar, A.T. May, Product excellence through design of experi-
ments, Cereal Foods World 32 (1987) 857868.
[28] M. Giovanni, Response surface methodology and product optimization,
Food Technol. 37 (1983) 4145.
[29] H.A. Leiberman, M.M. Reiger, G.S. Banker, Pharmaceutical Dosage
Forms: Disperse Systems, vol. 2, Mercel Dekker, New York, NY,
1989.
[30] T. Harnsilawat, R. Pongsawatmanit, D.J. McClements, Stabilization of
model beverage cloud emulsions using protein-polysaccharide electrostatic
complexes formedat the oil-water interface, J. Agric. FoodChem. 54(2006)
55405547.
[31] E. Dickinson, C. Ritzoulis, M.J.W. Povey, Stability of emulsions containing
both sodium caseinate and tween 20, J. Colloid Interf. Sci. 212 (1999)
466473.
[32] Y. Cao, E. Dickinson, D.J. Wedlock, Creaming and occulation in emul-
sions containing polysaccharides, Food Hyd. 4 (1990) 185195.
[33] J. Bibbette, Depletion interactions and fractionated crystallization for poly-
disperse emulsion purication, J. Colloid Interf. Sci. 147 (1991) 474478.
[34] E. Papalmprou, E. Makri, V. Kiosseoglou, G. Doxastakis, Effect of medium
molecular weight xanthan gum in rheology and stability of oil-in-water
emulsion stabilized with legume proteins, J. Sci. Food Agric. 85 (2005)
19671973.
[35] E. Tsaliki, S. Pegiadou, G. Doxastakis, Evaluation of the emulsifying prop-
erties of cottonseed protein isolates, Food Hyd. 18 (2004) 631637.
56 H. Mirhosseini et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 315 (2008) 4756
[36] R.L. Mason, R.F. Gunst, J.J. Hess, Statistical Design and Analysis of Exper-
iments with Applications to Engineering and Science, John Wiley and Sons
Inc., An International Thomason Publishing, Europe, London, 1V7AA,
Hoboken, NJ, 2003.
[37] K.Y. Tse, G.A. Reineccius, Methods to predict the physical stability of
avorcloud emulsion, in: C.T. Ho, C.T. Tan, C.H. Tong (Eds.), Flavor
Technology: Physical Chemistry, Modication, and Process, American
Chemical Society, Washington, D.C., 1995, pp. 172182.