ARTICLE IN PRESS

FOOD
HYDROCOLLOIDS
Food Hydrocolloids 23 (2009) 271–280
www.elsevier.com/locate/foodhyd

Characterization of the influence of main emulsion components on the

physicochemical properties of orange beverage emulsion using response
surface methodology
Hamed Mirhosseinia, Chin Ping Tana,, Nazimah S.A. Hamidb, Salmah Yusofc,
Boo Huey Chernd
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
d
Department of Food Service and Management, Faculty of Food Science and Technology,
Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia
Received 13 December 2007; accepted 19 February 2008

Abstract

The present work was conducted to investigate the influence of main emulsion components, namely Arabic gum (7–13% w/w),
xanthan gum (0.1–0.3% w/w) and orange oil (6–10% w/w) contents on physical stability, viscosity, cloudiness and conductivity of orange
beverage emulsion. In this study, 20 orange beverage emulsions were established based on a three-factor central composite design (CCD)
involving 8 factorial points, 6 axial points and 6 center points. The main objective of the present study was to determine an optimal
concentration level of main emulsion components leading to an optimum orange beverage emulsion with desirable physicochemical
properties. In general, all response surface models were significantly (po0.05) fitted for describing the variability of physical stability,
viscosity, conductivity and cloudiness as a nonlinear function of the content of main emulsion components. More than 84% of the
variation of physicochemical properties of orange beverage emulsion could be explained as a function of the content of the main beverage
emulsion components. In general, the orange oil content appeared to be the most significant (po0.05) factor influencing all emulsion
characteristics studied except for conductivity. From the optimization procedure, the overall optimal region leading to the desirable
orange beverage emulsion was predicted to be achieved by the combined level of 13% (w/w) Arabic gum, 0.22% (w/w) xanthan gum and
10% (w/w) orange oil.
r 2008 Elsevier Ltd. All rights reserved.

Keywords: Arabic gum; Xanthan gum; Physical stability; Viscosity; Cloudiness; Conductivity; Beverage emulsion; Central composite design

1. Introduction stability implies no tendency towards structural changes

(Pittia, Gambib, & Lerici, 1997). Consequently, the
An emulsion is traditionally defined as a dispersion of droplets of the inner phase tend to cluster together
droplets of one liquid in another, the two being immiscible spontaneously, forming small or large flocs (flocculation),
(Dickinson & Stainsby, 1982). The emulsion and colloidal to coalesce giving larger spherical droplets and to cream,
systems are thermodynamically unstable systems and tend leading to a layer of the lower density phase on top of the
to destabilize due to an excess surface free energy, where emulsion (McClements, 1999). Long-term stability of a
dispersed system implies that the rate and extent of changes
Corresponding author. Tel.: +603 89468418; fax: +603 89423552. in emulsion structure are is adequately low in real time.
E-mail addresses: [email protected], [email protected] The long-term stability of beverage emulsions is normally
(C.P. Tan). extended by adding a variety of stabilizers (e.g. thickening

0268-005X/$ - see front matter r 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodhyd.2008.02.007
 ARTICLE IN PRESS
272 H. Mirhosseini et al. / Food Hydrocolloids 23 (2009) 271–280

agents, surfactants, etc) to retard the droplet aggregation conductivity. The main emulsion components studied as
or flocculation (Tesch & Schubert, 2002). independent variables were: Arabic gum (7–13% w/w),
The emulsion stability is influenced by the balance xanthan gum (0.1–0.3% w/w) and orange oil (6–10% w/w).
between attractive (Van der Waals and osmotic) and It should be noted that the other critical parameters such as
repulsive (electrostatic, steric and hydration) forces be- emulsion stability in the diluted form, average droplet size,
tween the emulsion droplets (McClements, 1999). These polydispersity index, z-potential, electrophoretic mobility,
forces mainly arise from excluded volume effects and ionic flow behavior, pH, density and flavor release of orange
forces between like charges, and they tend to be weak beverage emulsion under different concentration levels
except at close range (hard-sphere interactions) or very low have also been investigated in our previous studies
ionic strength (unscreened electrostatic interactions) (Dick- (Mirhosseini et al., 2007a–c, 2008). These optimization
inson, 1998). In the absence of the osmotic attraction, the studies allow the manufacturers for the pre-formulation of
net repulsion between emulsion droplets may be sufficiently an optimum orange beverage emulsion with desirable
large to retard the flocculation. On the contrary, as the physicochemical properties.
polysaccharide concentration is increased, the osmotic
attraction (added to the Van der Waals attraction) becomes
large enough to overcome the repulsive forces, and the 2. Materials and methods
droplets tend to flocculate into a secondary medium, i.e.
the overall attraction is sufficiently greater than the thermal 2.1. Materials
energy of the colloidal system (Dickinson & McClements,
1996). The electrostatic repulsion is one of the constituents Arabic gum (food grade) was provided by Colloides
that accounts for the suspending stabilization of particles Naturels International Co. (Rouen, France). Xanthan gum
in dispersion system (Genovese & Lozano, 2001). Thus, the was donated by CP Kelco (San Diego, CA, USA). Citric
determination of conductivity is crucial to understand the acid, sodium benzoate and potassium sorbate (X95%)
behavior and stability of colloids and suspensions over were purchased from Fisher Scientific (Pittsburgh, PA,
time. USA). Valencia cold pressed orange oil was provided by
Knowledge on the rheological properties of emulsion Danisco (Cultor, Aarhus, Denmark).
system is crucial for their proper formulation, process
evaluation (e.g. mixing, pump sizing, extraction, filtration,
extrusion, purification, pasteurization, evaporation, drying 2.2. Preparation of orange beverage emulsion
and aseptic processing), handling, storage and transporta-
tion, as well as for designing the relevant equipment As demonstrated in our previous study (Mirhosseini
and apparatus (Buffo & Reineccius, 2002). The shelf-life et al., 2007a–c), a representative orange beverage emulsion
of food emulsions depends on the rheological attributes of is composed of two phases: water phase and oil phase. To
the emulsion phases. For instance, the creaming of oil prepare the water phase, sodium benzoate, potassium
droplets depends on the viscosity of the aqueous phase. sorbate and citric acid were sequentially dispersed in
The rheological properties of these products are generally deionized water (60 1C) using a high-speed blender (Waring
controlled by varying the droplet volume fraction, the blender 32BL80, Torrington, CT, USA). It should be noted
droplet size distribution, the inter-droplet forces, the that these emulsion components were added in that order,
type and concentration of emulsion component, as well fully dissolving each component before adding the next
as the processing and storage conditions (McClements, compound. While mixing the mixture, Arabic gum was
1999). gradually added to deionized water (60 1C) and mixed for
The ability of food manufacturers to formulate emul- 3 min to facilitate hydration. The Arabic gum solution was
sion-based products with desirable and reproducible kept overnight at room temperature to fully hydrate
characteristics depends on knowledge on the relationship (Buffo et al., 2001). To prepare the water phase, xanthan
between the emulsion properties and matrix composition as gum solution was prepared separately by dissolving
well as microstructure. Several studies have been carried xanthan gum in deionized water and then mixed with
out on the beverage emulsion (Buffo, Reineccius, & Arabic gum solution by using a high-speed blender. While
Oehlert, 2001; Chanamai & McClements, 2001; Mirhossei- mixing the water phase, the cold pressed orange oil was
ni, Tan, Hamid, & Yusof, 2007a, b, 2008; Mirhosseini, gradually added into water phase to provide an initial
Yusof, Hamid, & Tan, 2007; Taherian, Fustier, & coarse emulsion. Fine emulsification (i.e. a small droplet
Ramaswamy, 2006; Tan, 1997). The present study was size o1 mm and narrow particle size distribution) was
conducted to characterize the main and interaction effects achieved by subjecting the pre-emulsions to pre-homo-
of main emulsion components on the physicochemical genization using the high shear homogenizer (Silverson
properties of orange beverage emulsion. The main objec- L4R, Buckinghamshire, UK) for 1 min and then passed
tive of present study was to determine an optimum level of through a high-pressure homogenizer (APV, Crawley,
independent variables leading to (i) the highest emulsion UK), for three passes (30, 28 and 25 MPa) (Mirhosseini
stability, viscosity and cloudiness, and (ii) the least et al., 2007a–c).
 ARTICLE IN PRESS
H. Mirhosseini et al. / Food Hydrocolloids 23 (2009) 271–280 273

2.3. Experimental tests measurement of cloudiness. The diluted emulsions were

transferred to a quartz cuvette with a 1 cm path length for
2.3.1. Emulsion stability (creaming test) the measurement of cloudiness. Cloudiness was then
In the present study, emulsion stability was monitored calculated from the absorbance reading at 660 nm
by measuring the extent of gravitational phase separation. (Kaufman & Garti, 1984). High absorbance reading values
For the measurement of physical stability, 15 ml of a correspond to the high emulsion cloudiness. Distilled water
prepared beverage emulsion was transferred into a 20 ml was used as a reference.
test tube and stored for 2 weeks at room temperature
(2571 1C). Emulsion stability index (ESI) was calculated 2.3.4. Viscosity
as percentage of the initial emulsion height (HE), height of The viscosity of orange beverage emulsions was mea-
cream layer (HC) and height of the sedimentation phase sured immediately after the sample preparation by means
(HS): ES (%)=(remaining HE/initial HE)  100 (Huang, of a steady stress Brookfield viscometer (Brookfield DV-
Kakuda, & Cui, 2001). In the experimental data, ESI was II+ Programmable Viscometer, Middleboro, MA, USA)
only shown to clarify the stability of 20 orange beverage equipped with the LV spindles. The viscosity measurement
emulsions. The higher emulsion stability was demonstrated range of a Brookfield programmable viscometer DV-II+
by the larger ESI value. The measurement was performed appears in the unit of cP or mPa S. For each emulsion, the
in duplicate and the average of two individual trials was viscosity was reported as an average of two individual
taken for data analysis. Creaming test has also been used to measurements. A fixed volume of emulsion in a 600 ml
determine emulsion stability by previous researchers beaker was measured each time and the spindle depth was
(Flinger, Flinger, & Mangino, 1990; Taherian et al., 2006). kept constant throughout the measurements. The proce-
dure for the measurement of viscosity has been described
2.3.2. Conductivity by previous study (İbanoğlu, 2002). In the present study,
The orange beverage emulsions were diluted (1:100) for the measurements were reported at a spindle speed of
the measurement of conductivity using a Malvern zeta sizer 100 rpm.
(Malvern series ZEN 3500, Malvern Instruments Ltd.,
Worcester, UK). To avoid multiple scattering effects, the 2.4. Experimental design and statistical analysis
emulsions should be diluted with deionized water prior to
analysis; then directly placed into the module. The As shown in Tables 1 and 2, 20 orange beverage
measurement range of a Malvern zeta sizer appears in the emulsions composed of Arabic gum (7–13% w/w), xanthan
unit of mS/cm for conductivity. The measurements of gum (0.1–0.3% w/w), orange oil (6–10% w/w), sodium
conductivity were carried out immediately after sample benzoate (0.1% w/w), potassium sorbate (0.1% w/w), citric
preparation. The measurements were reported as the acid (0.4% w/w) and deionized water were prepared for the
average of three separate injections, with four readings optimization procedure based on a three-factor central
made per injection. The averages of the triplicate values composite design (CCD). As shown in Table 1, the center
were taken as the response values for conductivity. point was repeated six times to calculate the repeatability
of the method (Montgomery, 2001). Analysis of variance
2.3.3. Cloudiness (ANOVA) and regression surface analysis were conducted
The measurement of cloudiness was carried out using a to determine the statistical significance of model terms and
UV–visible spectrophotometer (Spectronic GenesysTM 10, fit a regression relationship relating the experimental data
GENEQ Inc., Montreal, Canada). The measurements were to the independent variables. The generalized response
made using a standard single-beam arrangement, grating- surface model for describing the variation in response
based, double detectors. Wavelength accuracy of the variables is given below:
instrument was found to be approximately 71.0 nm with
Y ¼ b0 þ Sbi xi þ Sbii x2i þ Sbij xi xj (1)
wavelength repeatability equal to 70.5 nm. Spectra were
obtained over the wavelength range of 190–1100 nm. The where Y is the response value predicted by the model; b0 is
orange beverage emulsions were diluted (1:1000) for the a offset value; bi, bii and bij are main (linear), quadratic and

Table 1
Levels of independent variables established according to the central composite design (CCD)

Independent variables Independent variable level

Low Center High Axial (a) Axial (+a)

Arabic gum content (% w/w) 7.00 10.00 13.0 5.10 14.90

Xanthan gum content (% w/w) 0.10 0.20 0.30 0.04 0.36
Orange oil content (% w/w) 6.00 8.00 10.00 4.73 11.27
 ARTICLE IN PRESS
274 H. Mirhosseini et al. / Food Hydrocolloids 23 (2009) 271–280

Table 2
The matrix of central composite design (CCD) and experimental data obtained for the response variables studied (Yj) (mean7SD)

Run Block Independent variable Response variable

Arabic gum Xanthan gum Orange oil Cloudiness Conductivity Emulsion stability Viscosity (100 rpm)
(% w/w) (% w/w) (% w/w) (Y1, Å) (Y2, mS/cm) (Y3, %) (Y2, mPa s)

1 1 7.00 0.10 6.00 0.05870.0000 0.08970.0017 0.070.0 1370

2 1 7.00 0.30 10.00 0.14570.0007 0.07070.0021 55.771.7 7372
3 1 13.00 0.30 6.00 0.13570.0000 0.15270.0038 31.172.6 7075
4 (C) 1 10.00 0.20 8.00 0.12070.0007 0.09970.0047 61.673.9 4471
5 1 13.00 0.10 10.00 0.13170.0007 0.06970.0018 82.571.1 4573
6 (C) 1 10.00 0.20 8.00 0.10970.0014 0.10770.0057 55.170.7 3973
7 2 7.00 0.10 10.00 0.10970.0007 0.07070.0029 63.072.2 1870
8 2 7.00 0.30 6.00 0.05370.0000 0.07170.0016 38.473.2 7076
9 2 13.00 0.10 6.00 0.12970.0043 0.08270.0038 72.974.7 3072
10 (C) 2 10.00 0.20 8.00 0.16970.0007 0.07970.0027 63.571.3 5071
11 (C) 2 10.00 0.20 8.00 0.14270.0000 0.07670.0032 68.670.9 6174
12 2 13.00 0.30 10.00 0.23370.0007 0.08270.0052 79.372.3 10378
13 (C) 3 10.00 0.20 8.00 0.15070.0000 0.06970.0105 67.170.8 5673
14 3 14.90 0.20 8.00 0.15670.0007 0.08070.0054 79.475.3 7376
15 3 10.00 0.20 4.73 0.05970.0043 0.07270.0031 0.070.0 4272
16 3 10.00 0.20 11.27 0.16970.0014 0.05270.0062 79.670.0 5873
17 3 10.00 0.04 8.00 0.16570.0007 0.08370.0043 90.076.1 1671
18 3 5.10 0.20 8.00 0.02070.0021 0.06370.0057 65.170.9 2972
19 (C) 3 10.00 0.20 8.00 0.10770.0000 0.08770.0031 64.370.0 5674
20 3 10.00 0.36 8.00 0.16470.0000 0.08770.0028 64.370.6 11477

C, center point.

interaction regression coefficients, respectively. The ade- of design variables for one response (i.e. individual
quacy of the models was determined using model analysis, optimum region) might be far from the optimal region
lack-of fit test and coefficient of determination (R2) obtained for another response. Thus, those emulsion
analysis. Joglekar and May (1987) suggested that R2 properties should be simultaneously considered for multi-
should be at least 0.80 for a good fitness of a response ple optimization process. Minitab software provides two
model. The corresponding variables will be more signifi- commands (graphical and numerical multiple optimization
cant (po0.05) if the absolute t value becomes larger and procedures) to identify the combination of input variable
the p-value becomes smaller (Atkinson & Donev, 1992). settings that jointly optimize a set of responses.
The terms statistically found non-significant (p40.05) were In the present study, both multiple graphical and
dropped from the initial models and the experimental data numerical optimization procedures were applied to deter-
were refitted only to significant (po0.05) independent mine the optimum region of Arabic gum, xanthan gum and
variable effects in order to obtain the final reduced model. orange oil contents resulting in the overall response goals.
It should be noted that some variables were kept in the For graphical optimization procedure, the final reduced
reduced model despite non-significance. For example, models were expressed as three-dimensional (3D) response
linear terms were also kept in the model if a quadratic or surface plots to better visualize the interaction effect of main
interaction term containing this variable was significant emulsion components on the physicochemical emulsion
(po0.05). The experimental design matrix, data analysis properties. The 3D plots were drawn by keeping one
and optimization procedure were performed using the variable constant at the center point and varying the other
Minitab v. 13.2 statistical package (Minitab Inc., PA, two variables within the experimental range in order to show
USA). how each response variable related to two continuous design
variables. For numerical multiple optimizations, the re-
2.5. Optimization and validation procedures sponse optimizer was applied by using the Minitab software
for determining the exact optimum level of independent
In this study, a combined level of main emulsion variables leading to individual and overall response goals.
components (as input variable settings) was required to Response optimizer allows us to compromise among the
provide an orange beverage emulsion with desirable various responses. It helps us to identify the combination of
physicochemical properties (i.e. the highest emulsion input variable settings that jointly optimize a single response
stability, viscosity and cloudiness, and least conductivity). or a set of responses. This numerical response optimization
Each emulsion property was important in determining the allows us to interactively change the input variable settings
quality of the finished product, while the optimal settings to perform sensitive analyses and possibly improve the
 ARTICLE IN PRESS
H. Mirhosseini et al. / Food Hydrocolloids 23 (2009) 271–280 275

initial solution. For method validation, experimental data fitted for all response variables studied (Table 3). Thus,
were compared with predicted values in order to verify the more than 84% of variability of the physicochemical
adequacy of final reduced models. Close agreement and no emulsion properties could be explained by the RSM models
significant difference must exist between the experimental as the nonlinear function of the content of main emulsion
and predicted values. components. No indication of significant (p40.05) lack of
fit was observed for the final reduced models.
3. Results and discussion As shown in Table 4, orange oil concentration had the
most significant (po0.05) effect on the responses as
3.1. Data analysis compared to the other independent variables studied. As
shown in Table 4, the main effects of independent variables
The estimated regression coefficients for the response were found to be kept in all reduced models, while the
variables, along with the corresponding R2, R2 (adj), quadratic effect of Arabic gum was shown to be significant
F-value and p-value of lack of fit, are given in Table 3. (po0.05) only in the reduced model fitted for the
Each response (yi) was assessed as a function of main, cloudiness. As clearly shown in Tables 3 and 4, the main
quadratic and interaction effects of Arabic gum (x1), linear effects of independent variables had no significant
xanthan gum (x2) and orange oil. The individual signifi- (p40.05) effect on viscosity, but they should also be kept in
cance F-value and p-value of independent variables are the final reduced models since the interaction effect of
shown in Table 4. The results indicated that the response Arabic gum and orange oil as well as the quadratic effect of
surface models with high coefficient of determination (R2) xanthan gum exhibited significant (po0.05) effects on the
ranging from 0.842 to 0.975 were significantly (pp0.001) reduced model fitted for the viscosity. As stated by

Table 3
Regression coefficients, R2, adjusted R2, probability values and lack of fit for the final reduced models

Regression coefficient Cloudiness (Y1, Å) Conductivity (Y2, mS/cm) Emulsion stability (Y3, %) Viscosity (100 rpm) (Y5, mPa S)

b0 0.125 0.0281 285.3 17.308

b1 0.047 0.0052 9.6 2.918
b2 1.105 0.3627 273.7 89.070
b3 0.001 0.0314 55.8 5.087
b21 0.002 – – –
b22 1.308 – – 471.502
b23 – 0.0014 2.9 –
b12 – 0.0417 31.6 –
b13 – 0.0013 – 0.818
b23 0.086 – – –
R2 0.901 0.877 0.919 0.975
R2 (adj) 0.83 0.788 0.872 0.96
a
Regression (p-value) 0.000 0.001 a 0.000 a
0.000 a

Lack of fit (F-value) 0.91 1.47 7.75 1.10

b
Lack of fit (p-value) 0.595 0.412 b 0.060 b
0.524 b

bi: the estimated regression coefficient for the main linear effects. bii: the estimated regression coefficient for the quadratic effects. bij: the estimated
regression coefficient for the interaction effects. 1: Arabic gum; 2: Xanthan gum; 3: Orange oil.aSignificant (po0.05).bNot significant (p40.05).

Table 4
The significance probability (p-value, F-ratio) of regression coefficients in the final reduced models

Variables Main effects Quadratic effects Interaction effects

x1 x2 x3 x21 x22 x23 x1x2 x1x3 x2x3

Cloudiness p-value 0.004 0.013 0.873* 0.018 0.040 – – – 0.038

(Y1, Å) F-ratio 13.67 8.84 0.03 7.67 5.40 – – – 5.57
Conductivity p-value 0.327* 0.009 0.023 – – 0.048 0.003 0.036 –
(Y2, mS/cm) F-ratio 1.05 9.95 6.99 – – 4.95 13.88 5.73 –
Emulsion stability (Y3, %) p-value 0.001 0.024 0.000 – – 0.000 0.010 – –
F-value 18.84 6.65 33.26 – – 22.70 9.33 – –
Viscosity (100 rpm) p-value 0.278* 0.166* 0.142* – 0.007 – – 0.023 –
(Y5, mPa.S) F-ratio 1.29 2.18 2.47 – 10.36 – – 6.74 –

*Not significant at po0.05.

 ARTICLE IN PRESS
276 H. Mirhosseini et al. / Food Hydrocolloids 23 (2009) 271–280

Montgomery (2001), the polynomial regression models and of cloudiness value (Y1) (Table 4). Thus, they were kept as
recommended optimum region may be significant (po0.05) critical parameters in the final reduced model fitted for
only in the studied independent variable ranges. Thus, it cloudiness (Table 3). Unji, Ueda, Ogata and Nakagaki
may not be true beyond the ranges of the factors. (1994) also found that the concentration of Arabic gum
Therefore, the recommeded models cannot be extrapolated significantly (po0.05) affected the cloudiness of oil in
beyond these ranges. water (o/w) emulsions stabilized with Arabic gum. The
results pointed out that the quadratic effect of orange oil,
3.1.1. Cloudiness interaction between Arabic gum and xanthan gum, as well
As shown in Table 3, the cloudiness value (Y1) was as the interaction effect of Arabic gum and orange oil had
negatively associated with main effects of xanthan gum and no significant (p40.05) effect on the cloudiness. Thus, they
orange oil as well as the quadratic effects of Arabic gum; should be dropped from the final reduced model fitted for
while the main effect of Arabic gum, quadratic effect of cloudiness data.
xanthan gum and its interaction with orange oil exhibited a As shown in Table 4, the main effect of orange oil
significant (po0.05) positive effect on the variation of appeared to be a not-significant (p40.05) parameter in the
cloudiness (Table 3). In general, the main effect of Arabic regression model fitted for the cloudiness, but it was kept in
gum followed by the main effect of xanthan gum was found the final reduced model as its interaction with xanthan gum
to have the most significant (po0.05) effect on the changes was found to be significant (po0.05). Fig. 1(a) exhibited

Cloudiness Conductivity
Hold values: Arabic gum: 10.0 Hold values: Orange oil: 8.0
Conductivity value (mS/cm)
Cloudiness value (A)

0.3 0.15

0.2 0.10

0.1 11.5 0.4 )

0.05
) 0.2 w/w
0.0 0.1 0.2
8.0 w/w 5 79 11 13 m(
%
0.3 0.4 4.5
i l (% 15 0.0 gu
o Arabic gum an
Xanthan gum nge (%w/w) nth
(%w/w) Ora Xa

Conductivity Emulsion stability

Hold values: Xanthan gum: 0.2 Hold values: Orange oil: 8.0
Emulsion stability value (%)
Conductivity value (mS/cm)

0.125
110
0.105 90
0.085 70
0.065 50
11.5 0.4 w)
0.045 w /w) 30 0.2 w/
5 7 9 8.0 (% 5 7 9 11 m (%
11 13 15 4.5 g um 13 15 0.0 gu
Arabic gum an an
(%w/w) nth Arabic gum nth
Xa (%w/w) Xa
Viscosity (100 RPM)
Hold values: Xanthan gum: 0.2
Viscosity value (mPa.s)

90

70

50
30 11.5
)
5 7 9 8.0 w/w
11 13 15 4.5 m (%
u
Arabic gum ang
(%w/w) a nth
X

Fig. 1. Response surface plots showing the significant (po0.050) interaction effects on the variation of cloudiness (a), conductivity (b,c), emulsion stability
(d) and viscosity (e) (100 rpm).
 ARTICLE IN PRESS
H. Mirhosseini et al. / Food Hydrocolloids 23 (2009) 271–280 277

that the cloudiness (Y1) increased when the concentration conductivity (Y2). However, the interaction effect of Arabic
of orange oil in the emulsion formulation was increased. gum and orange oil was found to be significant (po0.05) in
This finding could be interpreted by the direct effect of oil the reduced model fitted for the conductivity (Tables 3 and
phase concentration on the refractive index followed by 4). Thus, the main effect of Arabic gum should also be kept
increasing the light scattering efficiency induced by the oil in the final reduced model (Table 4).
globules. The positive effect of oil phase concentration on The 3D surface plot was drawn to visualize the
the cloudiness has also been reported by previous studies significant (po0.05) interaction effect of independent
(D"użewska, Stobiecka, & Maszewska, 2006; Hernandez & variables on the conductivity (Fig. 1b,c). The curvatures
Baker, 1991; Taherian et al., 2006). Hernandez and Baker shown in Fig. 1(b,c) confirmed that the relationship found
(1991) also described that the cloudiness of emulsion between the independent interaction effects and conduc-
depended on the average droplet size and concentration of tivity value (Y2) appeared to be nonlinear. This may be
oil phase in emulsions. explained by the fact that the conductivity of orange
The individual optimum region leading to the highest beverage emulsion was affected not only by the main
_
cloudiness value (Y 1 ¼ 0:221 Å) was predicted to be at the emulsion components but also by the other factors such as
set level of 13% (w/w) Arabic gum, 0.3% (w/w) xanthan the presence of electrolytes, viscosity of the medium,
gum and 10% (w/w) orange oil. From the optimization particle size and distribution. The individual optimal
results, high concentration of main emulsion components region indicated that an orange beverage emulsion
provided the high cloudiness value. This phenomenon could composed of 7% (w/w) Arabic gum, 0.1% (w/w) xanthan
be interpreted by the positive effects of both hydrocolloids gum and 6% (w/w) orange oil was predicted to provide the
on the droplet concentration thereby increasing the cloudi- optimum conductivity (the least conductivity value,
ness, since, theoretically, the emulsion cloudiness is due Y2 ¼ 0.064 mS/cm). Conversely, set level of 13% (w/w)
largely to the suspended solid particles in the emulsion and Arabic gum, 0.3% (w/w) xanthan gum and 6% (w/w)
dispersion system. As also shown in our previous study orange oil was estimated to result in the highest con-
(Mirhosseini, Tan, Hamid, & Yusof, 2007b), the addition of ductivity value (Y2 ¼ 0.118 mS/cm).
high amount of beverage emulsion components resulted in
the increase in average droplet size. The direct correlation 3.1.3. Emulsion stability
between average droplet size and cloudiness has also been The results demonstrated that the emulsion stability (Y3)
reported by previous researchers Chanamai & McClements, was positively proportional to the main effects of
2001; Harnsilawat, Pongsawatmanit, & McClements, 2006), independent variables, while it was negatively affected by
whereas the orange beverage emulsion formulated by the quadratic term of orange oil and interaction effect
combination of 7% (w/w) Arabic gum, 0.22% (w/w) between Arabic gum and xanthan gum (Table 3). As shown
xanthan gum and 6% (w/w) orange oil was estimated to in Table 4, the main effect of orange oil followed by its
_
provide the least cloudiness value (Y 1 ¼ 0:044 Å). quadratic effect showed the most significant (po0.05)
effect on the emulsion stability (Y3), thus ensuring the
3.1.2. Conductivity content of orange oil should be considered as a primary
As shown in Tables 3 and 4, the conductivity value (Y2) factor for determining the physical stability of orange
was directly proportional to the significant (po0.05) main beverage emulsion in the independent variable ranges
effect of orange oil and the interaction effect of Arabic gum studied. In addition to the significant (po0.05) effect of
and xanthan gum, while the main effect of xanthan gum, its orange oil, the main effects of Arabic gum and xanthan
interaction with orange oil and quadratic effect of orange gum as well as their interaction effect were also found to be
oil showed a significant (po0.05) negative effect on the a significant (po0.05) parameter on the emulsion stability
variation of the conductivity value (Y2) (Tables 3 and 4). (Y3) (Table 4).
This may be interpreted by the fact that xanthan gum is a As also stated by the previous researchers (Papalmprou,
negatively charged polysaccharide (Zhan, 2003). Thus, the Makri, Kiosseoglou, & Doxastakis, 2005; Tsaliki, Pegia-
increase in xanthan gum content in the emulsion formula- dou, & Doxastakis, 2004), the stability of oil-in-water
tion can result in increase in negatively charged zeta emulsion with regard to flocculation, coalescence and
potential followed by decrease in the conductivity value. creaming has proven to be strongly affected by the presence
As shown in the results (Table 4), the interaction effect of polysaccharides. The influence of Arabic gum and
between Arabic gum and xanthan gum followed by the xanthan gum on emulsion viscosity may be due to the
main effect of xanthan gum exhibited the most significant specific effect of polysaccharides on the emulsion viscosity
(po0.05) effect on the changes of conductivity value (Y2). (Chanamai & McClements, 2001; Dickinson & McCle-
In addition, the main effect of xanthan gum and orange oil, ments, 1996; Philips & Williams, 1995), z-potential (Zhan,
quadratic effect of orange oil and its interaction with Arabic 2003), particle size and distribution (Taherian et al., 2006).
gum were found to be significant (po0.05) parameters in the Jacome-Gouth, Aserin, and Garti (1991) found that the
model fitted for the conductivity (Y2) (Tables 3 and 4). The proportion of Arabic gum to oil phase had a significant
other insignificant (p40.05) independent variable effects (po0.05) influence on the stability of beverage emulsion. In
must be dropped from the reduced model fitted to contrast, the quadratic effects of Arabic gum and xanthan
 ARTICLE IN PRESS
278 H. Mirhosseini et al. / Food Hydrocolloids 23 (2009) 271–280

gum as well as their interactions with orange oil appeared protein-deficient (AG). The AGP fraction is mostly
to be not-significant (p40.05) on the emulsion stability responsible for the emulsifying properties: the hydrophilic
(Y3). Thus, they should not be kept in the final reduced carbohydrate blocks are linked to the protein chain that
model. strongly adsorbs at o/w interface promoting emulsion
Fig. 1(d) exhibits the variation in emulsion stability as a stability (Williams et al., 1990). In fact, the formation of a
nonlinear function of main emulsion components. As thick, sterically stabilizing layer around emulsion droplets
reported in a previous study (Tesch & Schubert, 2002), enables the flavor oil emulsion to be stabilized in both
stabilizers can improve the stability of emulsions by concentrated and diluted forms.
hydrodynamic effects. On the other hand, macromolecular
stabilizers can also cause instability of the emulsion. 3.1.4. Viscosity
Macromolecular emulsifying agents enter into interactions In most cases, the viscosity (Y5) of orange beverage
which may have different effects on the stability of the emulsions decreased when the spindle speed was increased.
emulsion, depending on the affinity of the molecules. The This observation could be explained by the reason that as
present study also showed that in some cases the individual the shear rate increased the droplet–droplet interaction was
main effects of the hydrocolloids studied were different deformed and eventually disrupted, which resulted in the
from their interaction effects. As exhibited in Fig. 1(d), the size reduction of the flocs thereby decreasing of viscosity
optimization procedure demonstrated that the increase of (Peamprasart & Chiewchan, 2006). This finding exhibited
orange oil up to 9.6% led to increase in the emulsion that the prepared orange beverage emulsions behaved as a
stability, while the addition of higher content of orange oil shear thinning fluids (or non-Newtonian pseudoplastic) in
(49.6%) tended to a slight decrease in the emulsion most cases. D"użewska et al.(2006) also found that Arabic
stability in the independent variable ranges studied. gum-based beverage emulsion behaved as a non-New-
From the optimization results, an orange beverage tonian pseudoplastic fluid. Sun, Gunasekaran, and Ri-
emulsion containing 13% (w/w) Arabic gum, 0.1% (w/w) chardsc (2007) similarly described that the emulsions with
xanthan gum and 9.6% (w/w) orange oil was predicted to little or no xanthan gum (0 and 0.02 wt%) behaved like a
be the individual optimum region that resulted in highest Newtonian fluid, while the emulsions containing relatively
emulsion stability (Y3 ¼ 87.5%). As reported in our higher xanthan gum concentration (X0.05% wt) exhibited
previous study (Mirhosseini et al., 2007b), the average shear-thinning behavior. As shown in Tables 3 and 4, the
droplet size significantly (po0.05) decreased with decreas- viscosity (100 rpm, Y5) was directly proportional to the
ing xanthan gum content. Thus, the beverage emulsion main and quadratic effects of xanthan gum, and interaction
containing 0.1% (w/w) xanthan gum provided the smaller effect of Arabic gum and orange oil; while the main effects
average droplet size and thereby higher emulsion stability of Arabic gum and orange oil had a negative effect on the
than the emulsion containing higher xanthan gum content viscosity (100 rpm, Y5). However, viscosity (100 rpm, Y5)
(40.1% w/w); whereas the least emulsion stability was significantly (po0.05) influenced only by the quadratic
(Y3 ¼ 18.8%) was estimated to be at the combined level term of xanthan gum and interaction effect of Arabic gum
of 7% (w/w) Arabic gum, 0.3% (w/w) xanthan gum and and orange oil.
6% (w/w) orange oil. As shown in the optimization results, As shown in Fig. 1(e), the viscosity (100 rpm, Y5) of
the high concentration of Arabic gum along with low orange beverage emulsions significantly (po0.05) increased
content of xanthan gum led to the highest emulsion when the xanthan gum content was increased. This
stability. Under individual optimum stability region, these observation has also been reported by previous researchers
combination levels of Arabic gum and xanthan gum were (Sun et al., 2007; Yemar, Tamehana, Munro, & Singh,
predicted to provide a sufficient excess of absorbed 2001). Xanthan gum has high molecular weight molecules
polysaccharide in the aqueous phase of emulsions. When (3,000,000 Da) composed of a 1–4 linked b-D-glucose
sufficient content of emulsifier fraction (i.e. Arabic gum backbone with three side-chains: a D-glucuronosyl unit
and xanthan gum) is present to saturate the droplet between two D-mannosyl units. The chemical structure of
surface, the extra increase of emulsifier fraction content xanthan gum also includes a large number of free carboxyl
can be attributed to the larger droplet size, and thereby groups. When xanthan gum is dispersed in water system, its
depletion of flocculation caused by excess bulk of non- complex molecules form complicated aggregates through
absorbed biopolymers (Dickinson, Ritzoulis, & Povey, hydrogen bonds and polymer entanglement (Hemar,
1999). Tamehana, Munro, & Singh, 2002). In this condition,
The positive effect of Arabic gum on emulsion stability xanthan undergoes a conformational transition from a
may be explained by arabinogalactan polysaccharide double helix to a complex aggregate through hydrogen
fraction (AGP) in the chemical structure of Arabic gum. bonds and polymer entanglement. Because of these highly
As demonstrated by Williams, Phillips, and Randall (1990), ordered networks and entanglements, a little increase of
Arabic gum consists of three distinct fractions, including: a xanthan gum content provides a high viscosity at low shear
high molecular mass, arabinogalactan, plus protein com- rate and vice versa. The presence of these active sites in the
plex (AGP), a glycoprotein (G1) and a lower molecular xanthan gum structure provides a great water absorption
mass fraction, an arabinogalactan polysaccharide, which is capacity thereby increase in the viscosity.
 ARTICLE IN PRESS
H. Mirhosseini et al. / Food Hydrocolloids 23 (2009) 271–280 279

From the optimization results, an orange beverage values (data not shown). No significant (p40.05) difference
emulsion composed of 13% (w/w) Arabic gum, 0.3% was found between those values. The experimental
(w/w) xanthan gum and 10% (w/w) orange oil was esti- response values were shown to be close to the predicted
mated to be the individual optimum region leading to the ones. These observations verified the adequacy of
highest viscosity (Y5 ¼ 104 mPa s, 100 rpm). As clearly corresponding response surface models relating the
shown, a set level of high concentration of the main experimental data to three independent variables studied.
emulsion components led to the highest viscosity value The desirable recommended beverage emulsion was
(100 rpm, Y5). This observation may be interpreted by the practically prepared and then evaluated by monitoring
reason that the presence of large number of particles the physicochemical emulsion properties studied. Under
increased resistance to the flow, which hence resulted in an the optimum condition, the corresponding experimental
increase in the apparent viscosity. On the other hand, this values for physical stability, viscosity, conductivity
phenomenon could also be interpreted by the positive and cloudiness of the predicted orange beverage emulsion
effect of main emulsion components, especially xanthan were 84.7, 82.3, 0.071 and 0.197, respectively. The
gum, on the average droplet size (Mirhosseini et al., corresponding response values obtained from the experi-
2007b), resulting in increase in the emulsion viscosity. mental data and those values predicted by the models
As also shown in our previous study (Mirhosseini et al., were observed to be close together. No significant (p40.05)
2007b), the addition of high amount of main emulsion difference between the experimental and predicted values
components, especially xanthan gum, led to increase in the verified the adequacy of final reduced models fitted by
average droplet size. D"użewska et al. (2006) also reported RSM.
that the increase in oil phase concentration led to
significant (po0.05) increase of apparent emulsion viscos-
4. Conclusions
ity. They explained the observation by the fact that the
increase of oil phase content induced an increase in average
In the present work, the dependence of physical stability,
droplet size and considerable extension of droplet size
viscosity, conductivity and cloudiness of orange beverage
distribution (Mirhosseini et al., 2007b), whereas the
emulsion on the content of main emulsion components
combined level of 7% (w/w) Arabic gum, 0.1% (w/w)
was studied by the RSM. The results indicated that the
xanthan gum and 6% (w/w) orange oil was predicted to
interaction of Arabic gum and xanthan gum behaved
provide the least emulsion viscosity (Y5 ¼ 14.5 mPa s).
differently from their individual main effects in some
cases. For instance, emulsion stability was negatively
3.2. Optimization procedure for predicting a desirable
influenced by xanthan gum. However, the hydrocolloid
orange beverage emulsion
usually tends to improve the stability of emulsion system.
Therefore, this study offered that the interaction between
For the graphical interpretation, three-dimensional (3D)
the beverage emulsion components should be considered
response surface plotting was highly recommended (Ma-
as critical parameters for pre-formulation of orange
son, Gunst, & Hess, 2003). Thus, the fitted polynomial
beverage emulsion. In all cases, reduced analysis offered
equation was expressed as response surface plot in order to
the significant (po0.05) reduced models with high R2
visualize the relationship between the response experi-
(40.84), thus ensuring reliable adjustment of experi-
mental levels of independent variables. The overall opti-
mental data to the independent variables studied. In
mum condition was then determined by superimposing
general, the present study indicated that the studied
all response surface plots. A numerical optimization was
emulsion characteristics were significantly (po0.05) influ-
also carried out by the response optimizer using the
enced by the proportion of main emulsion components.
Minitab software for determining the exact optimum
The present study indicated that the instability of beverage
level of independent variables leading to the overall
emulsion could be controlled by an appropriate pro-
optimum condition. The numerical optimization results
portion of main emulsion components. Under the corre-
indicated that the overall optimum region was predicted to
sponding optimum condition, no significant (po0.05)
be at the combined level of 13% (w/w), 0.22% (w/w) and
difference between experimental and predicted values
10% (w/w) for Arabic gum, xanthan gum and orange oil,
verified the accuracy of the response surface models
respectively. The corresponding response values for physi-
employed for predicting the variation of emulsion proper-
cal stability, viscosity, conductivity and cloudiness of
ties as a function of the content of main emulsion
orange beverage emulsion predicted under the recom-
components.
mended optimum condition were 81.5, 78.5, 0.075 and
0.188, respectively.
Appendix A. Supplementary Materials
3.3. Verification of the response surface models
Supplementary data associated with this article can
The adequacy of the response surface equations was be found in the online version at doi:10.1016/j.foodhyd.
checked by the comparison of experimental and predicted 2008.02.007.
 ARTICLE IN PRESS
280 H. Mirhosseini et al. / Food Hydrocolloids 23 (2009) 271–280

References Colloids and Surface A: Physicochemical and Engineering Aspects,

315(1–3), 47–56.
Atkinson, A. C., & Donev, A. N. (1992). Optimum experimental designs. Mirhosseini, H., Tan, C. P., Hamid, N. S. A., & Yusof, S. (2007a).
Oxford: Oxford University Press pp. 132–189. Modeling the relationship between the main emulsion components and
Buffo, R. A., & Reineccius, G. A. (2002). Modeling the rheology of stability, viscosity, fluid behavior, z-potential and electrophoretic
concentrated beverage emulsions. Journal of Food Engineering, 51, mobility of orange beverage emulsion using response surface
267–272. methodology. Journal of Agricultural and Food Chemistry, 55,
Buffo, R. A., Reineccius, G. A., & Oehlert, G. W. (2001). Factors affecting 7659–7666.
the emulsifying and rheological properties of gum acacia in beverage Mirhosseini, H., Tan, C. P., Hamid, N. S. A., & Yusof, S. (2007b).
emulsions. Food Hydrocolloids, 15, 53–66. Optimization of the contents of Arabic gum, xanthan gum and orange
Chanamai, R., & McClements, D. J. (2001). Prediction of emulsion color oil affecting turbidity, average particle size, polydispersity index and
from droplet characteristics: Dilute monodisperse oil in water density in orange beverage emulsion. Food Hydrocolloids, in press,
emulsions. Food Hydrocolloids, 15, 83–91. doi:10.1016/j.foodhyd.2007.06.011.
Dickinson, E. (1998). Stability and rheological implications of electrostatic Mirhosseini, H., Tan, C. P., Hamid, N. S. A., & Yusof, S. (2007). Solid-
milk protein–polysaccharide interactions. Trends in Food Science and phase microextraction for determining twelve orange flavour com-
Technology, 9, 347–354. pounds in a model beverage emulsion. Phytochemical Analysis, 105,
Dickinson, E., & McClements, D. J. (1996). Advances in food colloids. 1659–1670.
London: Chapman and Hall. Mirhosseini, H., Yusof, S., Hamid, N. S. A., & Tan C. P. (2007d). Solid-
Dickinson, E., & Stainsby, G. (1982). Colloids in foods. London: Applied phase microextraction for headspace analysis of key volatile com-
Science Publishers. pounds in orange beverage emulsion. Food Chemistry, in press.
Dickinson, E., Ritzoulis, C., & Povey, M. J. W. (1999). Stability of Montgomery, D. C. (2001). Design and analysis of experiments (5th ed).
emulsions containing both sodium caseinate and Tween 20. Journal of New York: Wiley.
Colloid and Interface Science, 212, 466–473. Papalmprou, E., Makri, E., Kiosseoglou, V., & Doxastakis, G. (2005).
D"użewska, E., Stobiecka, A., & Maszewska, M. (2006). Effect of oil Effect of medium molecular weight xanthan gum in rheology and
phase concentration on rheological properties and stability of stability of oil in water emulsion stabilized with legume proteins.
beverage emulsions. Acta Science Poland Technology Alimentarius, 5, Journal of Science of Food and Agriculture, 85, 1967–1973.
147–156. Peamprasart, T., & Chiewchan, N. (2006). Effect of fat content and
Flinger, K. L., Flinger, M. A., & Mangino, M. E. (1990). The effects of preheat treatment on the apparent viscosity of coconut milk after
compositional factors on the short-term physical stability of a homogenization. Journal of Food Engineering, 77, 653–658.
concentrated infant formula. Food Hydrocolloids, 4, 95–104. Philips, G. O., & Williams, P. A. (1995). Interaction of hydrocolloids in
Genovese, D. B., & Lozano, J. E. (2001). The effect of hydrocolloids on food systems. In A. G. Gaonkar (Ed.), Ingredient interactions: Effect
the stability and viscosity of cloudy apple juices. Food Hydrocolloids, on food quality (pp. 131–170). New York: Marcel Dekker.
15, 1–7. Pittia, P., Gambib, A., & Lerici, C. R. (1997). Hygrometric measurements
Harnsilawat, T., Pongsawatmanit, R., & McClements, D. J. (2006). for the evaluation of the stability of model food emulsions. Food
Stabilization of model beverage cloud emulsions using protein–poly- Research International, 30, 177–184.
saccharide electrostatic complexes formed at the oil–water interface. Sun, C., Gunasekaran, S., & Richardsc, M. P. (2007). Effect of xanthan
Journal of Agricultural and Food Chemistry, 54, 5540–5547. gum on physicochemical properties of whey protein isolate stabilized
Hemar, Y., Tamehana, M., Munro, P. A., & Singh, H. (2002). Influence of oil-in-water emulsions. Food Hydrocolloids, 21, 555–564.
xanthan gum on the formation and stability of sodium caseinate oil-in- Taherian, A. R., Fustier, P., & Ramaswamy, H. S. (2006). Effect of added
water emulsions. Food Hydrocolloids, 15, 513–519. oil and modified starch on rheological properties, droplet size
Hernandez, E., & Baker, R. A. (1991). Turbidity of beverages with citrus distribution, opacity and stability of beverage cloud emulsions. Journal
oil clouding agent. Journal of Science, 56, 1024–1030. of Food Engineering, 77, 687–696.
Huang, X., Kakuda, Y., & Cui, W. (2001). Hydrocolloids in emulsions: Tan, C. T. (1997). In S. Friberg, & K. Larsson (Eds.), Beverage emulsions
Particle size distribution and interfacial activity. Food Hydrocolloids, in food emulsions (3th ed, pp. 491–524). New York, NY: Marcel
15, 533–542. Dekker Inc.
İbanoğlu, E. (2002). Rheological behavior of whey protein stabilized Tesch, S., & Schubert, H. (2002). Influence of increasing viscosity of the
emulsions in the presence of gum Arabic. Journal of Food Engineering, aqueous phase on the short-term stability of protein stabilized
52, 273–277. emulsions. Journal of Food Engineering, 52, 305–312.
Jacome-Gouth, O., Aserin, A., & Garti, N. (1991). Stability and opacity of Tsaliki, E., Pegiadou, S., & Doxastakis, G. (2004). Evaluation of the
orange oil cloudy emulsions based on gum elemi and gum Arabic. emulsifying properties of cottonseed protein isolates. Food Hydro-
Journal of Food Science and Technology, 26, 249–257. colloids, 18, 631–637.
Joglekar, A. M., & May, A. T. (1987). Product excellence through design Unji, M., Ueda, H., Ogata, M., & Nakagaki, M. (1994). Studies on the oil-
of experiments. Cereal Foods World, 32, 857–868. in-water emulsions stabilized with gum Arabic by using the turbidity
Kaufman, V. R., & Garti, N. (1984). Effect of cloudy agents on the ratio method. Journal of Pharmaceutical Society of Japan, 112,
stability and opacity of cloudy emulsions for soft drinks. Food 906–913.
Technology, 19, 255–261. Williams, P. A., Phillips, G. O., & Randall, R. C. (1990). Structure–func-
Mason, R. L., Gunst, R. F., & Hess, J. J. (2003). Statistical design and tion relationships of gum Arabic. In G. O. Phillips, D. J. Wedlock, &
analysis of experiments with applications to engineering and science. P. A. Williams (Eds.), Gums and stabilizers for the food industry, Vol. 5
Europe, London, Hoboken, NJ: John Wiley and Sons Inc. (An (pp. 25–36). New York: Elsevier.
International Thomason Publishing. Yemar, Y., Tamehana, M., Munro, P. A., & Singh, H. (2001). Influence of
McClements, D. J. (1999). Food emulsions: Principles, practice and xanthan gum on the formation and stability of sodium caseinate oil-in
techniques. Boca Raton: CRC Press. water emulsions. Food Hydrocolloids, 15, 513–519.
Mirhosseini, H., Tan, C. P., Hamid, N. S. A., & Yusof, S. (2008). Effect of Zhan, X. B. (2003). Production, properties and application of food
Arabic gum, xanthan gum and orange oil contents on z-potential, hydrocolloids (pp. 8–9). Beijing: China Light Industry Press
conductivity, stability, size index and pH of orange beverage emulsion. (in Chinese).