Particuology 36 (2018) 59–69

Contents lists available at ScienceDirect

Particuology
journal homepage: www.elsevier.com/locate/partic

Modeling and optimization of the parameters affecting the in-situ

microencapsulation process for producing epoxy-based self-healing
anti-corrosion coatings
Ali Ebrahiminiya a , Mohammad Khorram a,∗ , Shadi Hassanajili a , Mehdi Javidi b
a
School of Chemical and Petroleum Engineering, Shiraz University, Shiraz, Iran
b
Department of Materials Science and Engineering, School of Engineering, Shiraz University, Shiraz, Iran

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

Article history: Micro/nanocapsules of urea–formaldehyde resin loaded with linseed oil, which are a self-healing agent in
Received 22 May 2016 glass flake epoxy anti-corrosion paint, were prepared using a combination of ultrasonic homogenization
Received in revised form and in-situ polymerization. The main objective of this study was to model and optimize the microen-
21 December 2016
capsulation process. Five-level central composite design was used to design, model, and optimize the
Accepted 9 January 2017
microencapsulation process. A quadratic model was constructed to show the dependency of the per-
Available online 20 June 2017
centage of encapsulated linseed oil and capsule size, as model responses, on the studied independent
variables (the rotational speed of the agitator and the power and duration of sonication). Analysis of vari-
Keywords:
Urea–formaldehyde
ance showed that all of the variables have significant effects on the encapsulated linseed oil percentage,
Response surface methodology while the rotational speed of the agitator and sonication time is effective variables for controlling the
Central composite design capsule size. Under the determined optimum conditions, a maximum encapsulated linseed oil percentage
Ultrasonic homogenization (ELO%) of 93.9% and a minimum micro/nanocapsule size of 0.574 ␮m were achieved at 594 rpm agitation,
Linseed oil 350 W sonication power, and 3 min sonication time. Validation of the model was performed. The percent-
Self-healing coating age relative errors between the predicted and experimental values of the ELO% and micro/nanocapsule
size are 1.28% and 3.66%, respectively. The efficacy of the optimum micro/nanocapsules in healing cracks
in a glass flake epoxy paint and corrosion protection was investigated by the salt spray test and Tafel
polarization technique.
© 2017 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of
Sciences. Published by Elsevier B.V. All rights reserved.

Introduction
Microencapsulation of active materials in hollow spheres is
an attractive subject in various fields, such as controlled release
systems (Möbus, Siepmann, & Bodmeier, 2012), preservation of fla-
vor (Gharsallaoui et al., 2012), biotechnology (Nazzaro, Orlando,
Fratianni, & Coppola, 2012), and food additives (Otálora, Carriazo,
Iturriaga, Nazareno, & Osorio, 2015). Recently, the use of microcap-
sules containing self-healing agents for healing cracks generated
in polymeric coatings and paints has attracted considerable inter-
est (Liu, Zhang, Wang, Wang, & Wang, 2012; Sauvant-Moynot,
Gonzalez, & Kittel, 2008; Suryanarayana, Rao, & Kumar, 2008).
These capsules loaded with a self-healing agent are incorporated in
∗ Corresponding author. Fax: +98 71 36473748.
E-mail addresses: ali ebrahiminiya@yahoo.com (A. Ebrahiminiya),
mkhorram@shirazu.ac.i, khorram m@hamoon.usb.ac.ir (M. Khorram),
ajili@shirazu.ac.ir (S. Hassanajili), mjavidi@shirazu.ac.ir (M. Javidi).
the coating. The microcapsules rupture and release the self-healing
agent during crack formation. The released self-healing agent reacts
with a catalyst dispersed in the polymeric coating or undergoes
oxidative polymerization and drying, leading to crack repair.
The most important techniques for preparing microcapsule-
loaded active agents are interfacial polymerization, in-situ
polymerization, coacervation, and meltable dispersion (Karger-
Kocsis, 2010). Among the above-mentioned methods, in-situ
polymerization is a popular method to prepare microcapsules for
self-healing applications. However, in standard emulsion microen-
capsulation methods, the lower limit of the capsule diameter is
approximately 10 ␮m (Blaiszik, Sottos, & White, 2008). To prepare
sub-micrometer and nanoscale capsules, some researchers have
combined ultrasonic homogenization and in-situ encapsulation
techniques (Blaiszik et al., 2008; Kouhi, Mohebbi, Mirzaei, & Peikari,
2013). In this combination technique, some parameters, such as the
rotational speed of the agitator and the power and time of sonica-
tion, have a strong effect on the size of the prepared capsules and

http://dx.doi.org/10.1016/j.partic.2017.01.010
1674-2001/© 2017 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.
60 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69
the percentage of encapsulated self-healing agent. The capsule size
and loading percentage of the active agent have important effects
on the self-healing properties of the coating. For a microcapsule-
based self-healing coating, the size of the microcapsules strongly
influences the self-healing performance. The microcapsule diam-
eter affects the amount of released healing agent in the cracked
area (Sun, Zhang, Tang, & Yang, 2016). A smaller microcapsule
diameter results in better compatibility with the coating matrix
(Samadzadeh, Boura, Lochab, Kumar, & Roy, 2010). Rule, Sottos,
and White (2007) investigated the influence of the microcapsule
diameter on the performance of self-healing materials. They inves-
tigated the relationship between the microcapsule size, healing
agent delivery, and healing performance by the fracture test. They
demonstrated that the amount of healing agent delivered is deter-
mined by the product of the microcapsule weight fraction and
the microcapsule diameter. Matsuda et al. (2016) investigated the
effect of the particle size of pH-sensitive microcapsules produced
by water-in-oil emulsion polymerization. They found that smaller
size microcapsules showed better corrosion inhibition properties.
The core content is one of the most important characteristics of
microcapsules and their self-healing performance. Sharma, Shukla,
Lochab, Kumar, and Roy (2014) investigated the effect of the con-
centration of the encapsulating polymer on the core content. They
demonstrated that the core content can be increased by decreasing
the encapsulating polymer concentration in the feed solution.
Optimization of the process parameters affecting microen-
capsulation using the in-situ polymerization method has been
investigated by several researchers. Gil, Cuenca, Romero, Valverde,
and Sánchez-Silva (2014) optimized the synthesis procedure of
microparticles containing gold. They studied the effects of the
amount of organic solvent and the gold precursor to polymer mass
ratio (Au/PUF) on the properties of the prepared microcapsules.
The obtained results showed that increasing the amount of organic
solvent leads to a decrease of the gold encapsulation efficiency
and an increase of the average diameter of the encapsulated gold
microparticles. Increasing the Au/PUF ratio also increased the par-
ticle size. Kouhi et al. (2013) prepared microcapsules filled with
healing agents using in-situ polymerization of urea–formaldehyde.
They investigated the effect of the agitation rate on the average
diameter of the microcapsules and their self-healing performance.
Their results showed that increasing the agitation rate leads to a
decrease of the mean diameter of the microcapsules and improves
the performance in self-healing applications. Ollier, Penoff, and
Alvarez (2016) optimized the synthesis conditions of epoxy resin
microencapsulation. They analyzed the emulsification conditions
(time, and agitation method and rate), viscosity of the core phase,
stirring speed, core/shell mass ratio, and drying process. They found
that microcapsules prepared at the optimized reaction conditions
have a satisfactory size (below 150 ␮m). In another study (Hatami
Boura, Peikari, Ashrafi, & Samadzadeh, 2012), the effect of the
rotational speed of the agitator on the microcapsule size was the
only investigated process parameter, and the process parameter
was optimized. All of the above-mentioned studies have been per-
formed by investigating one variable at a time. This ignores the
interactions between affecting factors and does not reflect the
actual responses.
To optimize the process conditions and minimize the number
of experiments, response surface methodology (RSM) can be used.
RSM is an efficient mathematical and statistical tool for construct-
ing empirical models that describe the effects of the studied param-
eters and their interactions on the response (Myers, Montgomery,
& Anderson-Cook, 2009). Khoee, Payandeh, Jafarzadeh, and Asadi
(2016) modeled and optimized in-situ polymerization microencap-
sulation of epoxy resin in poly(urea–formaldehyde). To model and
optimize the microcapsule size and its core content, the central
composite design (CCD) method was used. The agitation rate, sur-
Table 1
Experimental ranges and levels of the independent variables.
Independent input parameter Levels and the actual values of parameter
−2 −1 0 1 2
Agitation speed, X1 (rpm) 300 400 500 600 700
Power of sonication, X2 (W) 200 250 300 350 400
Duration of sonication, X3 (min) 0 3 6 9 12
factant concentration, and sonication time were the parameters
analyzed.
The main objective of this study is to model and optimize the
effects of the rotational speed of the agitator, sonication power, and
sonication time on the micro/nanocapsule size and percentage of
encapsulated healing agent in capsules prepared by the combined
method of ultrasonic homogenization and in-situ polymerization.
The shell of the prepared micro/nanocapsules was composed of
urea–formaldehyde and its core material was linseed oil as a
healing agent owing to its film-forming ability by atmospheric oxi-
dation. RSM in conjunction with CCD was used to determine the
optimal experimental conditions for minimizing the capsule size
and maximizing the percentage of encapsulated healing agent.
Experimental
Materials
Urea, ammonium chloride, poly(vinyl alcohol) as a surfactant,
and resorcinol as a cross-linking agent were purchased from Dae-
jung Co. (Gyeonggi-Do, South Korea). Pure linseed oil as a healing
agent was obtained from Sigma-Aldrich Co. (St. Louis, USA) and
formaldehyde was obtained from Merck Chemicals Co. (Darmstadt,
Germany). Glass flake epoxy and hardener were procured from
Firoozeh Paint Co. (Shiraz, Iran).
Experimental design and statistical analysis
To investigate the effect of the rotational speed of agita-
tion, sonication power, and sonication time on the diameter
of the micro/nanocapsules and the linseed oil content in the
micro/nanocapsules, a three factor CCD was used. The agitation
speed (X1 ), sonication power (X2 ), sonication time (X3 ) were cho-
sen as the three independent factors for investigation (Table 1). The
other parameters that normally affect the in-situ polymerization
process were set at fixed values. As shown in Table 1, five levels
of the numerical factors were considered. The number of experi-
ments determined by CCD is N = k2 + 2k + cP , where k is the number
of factors and cP is the center point replicate number. In the present
study, the number of independent variables is three and hence
N = 20 for five repetitions in a center point experiment. The param-
eter that indicates the upper and lower levels of the independent
variables in the experiment design, ˛, is two for this design, because
there are three independent variables. The test factors were coded
according to the following equation:
Xi − X0
xi = , (1)
Xi
where Xi is the independent variable, X0 is the real value of the
independent variable in the center point of the independent vari-
able, and xi is the code value of the independent variable, and Xi
is the step change.
 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69 61

Table 2 conditions listed in Table 2. Finally, the emulsion was cooled and
CCD experimental design and observed responses.
the microcapsules were recovered by vacuum filtration. The precip-
Run Independent input parameter Response itate was rinsed with water and washed with xylene to remove the
suspended oil. The obtained micro/nanocapsules were dried under
X1 (rpm) X2 (W) X3 (min) Y1 (ELO%) Y2 (␮m)
vacuum. The product was then placed in an oven at 35 ◦ C for 48 h.
1 400 250 3 92.09 0.841
2 600 250 3 92.19 0.572
3 400 350 3 92.35 0.582
4 600 350 3 93.92 0.554 Linseed oil content in the micro/nanocapsules
5 400 250 9 93.73 0.812
6 600 250 9 93.16 0.612 The linseed oil content was measured as described by
7 400 350 9 92.54 0.803
Suryanarayana et al. (2008). A certain amount of the obtained
8 600 350 9 92.9 0.932
9 300 300 6 92.51 0.822 microcapsules (Wi ) was crushed using a pestle and mortar, and
10 700 300 6 93.36 0.685 then placed in a thimble. For greater precision, the pestle and mor-
11 500 200 6 92.55 0.732 tar were washed with xylene and the xylene was added into the
12 500 400 6 93.38 0.702
thimble. Xylene was used as a solvent for linseed oil. Extraction
13 500 300 0 92.59 13.688
14 500 300 12 93.27 0.848
was carried out in a Soxhlet apparatus at an approximate tempera-
15 500 300 6 92.89 0.709 ture of 130 ◦ C. After 3 h, the thimble was carefully removed and the
16 500 300 6 92.82 0.682 product was placed in an oven at 60 ◦ C for 24 h. Finally, the dried
17 500 300 6 92.61 0.733 material was weighed (Wf ). The percentage of encapsulated linseed
18 500 300 6 92.73 0.703
oil (ELO%) was determined by
19 500 300 6 92.55 0.698
20 500 300 6 93.14 0.751
Wi − Wf
ELO% = × 100. (3)
Wi
In CCD, the response Y is related to the independent variables
by linear and quadratic terms:.
Scanning electron microscopy and image analysis

k 
k 
k
2
Y = ˇ0 + ˇi xi + ˇii xi + ˇij xi xj + ε, (2)
The size and shape of the capsules were investigated by scan-
i=1 i=1 i<j
ning electron microscopy (SEM, Vega, Tescan, Czech Republic). The
where ˇ0 is the constant or offset term, ˇi is the linear coefficient, capsule samples were dispersed in acetone and a drop of the solu-
ˇii is the quadratic effect coefficient, ˇij is the coefficient for inter- tion was poured on a metal stub. After evaporation of acetone, a
action, and ε is the statistical error (Hafizi, Ahmadpour, Heravi, gold film was coated under vacuum using a sputter coater (Bal-Tec
Bamoharram, & Khosroshahi, 2012; Noshadi, Amin, & Parnas, 2012). Company, Switzerland). To determine the diameter of the prepared
Experimental planning and the statistical analysis were performed micro/nanocapsules, the SEM images were analyzed with an image
by the Design–Expert statistical package (trial version 7.0). Table 2 analyzer (Image J2 software). In total, 25 capsules were randomly
shows the obtained standard matrix for a total of 20 experiments. selected and their average diameters were measured by the image
The experiments were performed in a random manner to avoid analyzer.
any systematic bias in the outcomes. The models obtained for the
diameter of the prepared capsules (Y2 ) and linseed oil content in
the capsules (Y1 ) were statistically analyzed by analysis of vari- Infrared spectroscopy
ance (ANOVA). The adequacy of the final model was also verified
by graphical and numerical analysis. The core and shell materials extracted from the Soxhlet appa-
ratus were analyzed by Fourier transform infrared spectroscopy
Micro/nanocapsulation process (FTIR, RX1, Perkin-Elmer, USA). The solid shell material collected
after Soxhlet extraction was crushed with a pestle in an agate mor-
Micro/nanocapsules containing healing agent were prepared by tar. The crushed material was mixed with potassium bromide and
the in-situ polymerization method described by Kouhi et al. (2013) dried at 40 ◦ C. The mixture was compressed into a disk. The FTIR
and Suryanarayana et al. (2008). In brief, 5 mL of 5% (w/v) aqueous spectrum of the extract after Soxhlet extraction was recorded. The
poly(vinyl alcohol) solution and 130 mL of deionized water were obtained spectrum was compared with the FTIR spectra of neat
mixed at room temperature. Under mechanical agitation, 2.5 g urea, linseed oil and urea–formaldehyde resin.
0.25 g ammonium chloride, and 0.25 g resorcinol were then added
to the obtained solution, and one or two drops of octanol was added
to the mixture as a co-stabilizer and anti-foaming agent. Prepara- Self-healing investigation
tion of urea–formaldehyde capsules should be performed under
acidic conditions. Therefore, the pH of the resulting solution was Optimized micro/nanocapsules in terms of the size and healing
adjusted to about 3 using 1 wt% chloric acid. Linseed oil (30 mL) was agent content were prepared and added to glass flake epoxy paint.
then slowly added, and to achieve a uniform emulsion, the solu- For this purpose, glass flake epoxy was mixed with the required
tion was agitated for 10 min. According to the procedure of Kouhi amount of hardener based on the manufacturer procedure and dif-
et al. (2013) and Blaiszik et al. (2008), tip sonication (TopSonics- ferent percentages of micro/nanocapsules (0, 10, and 20 wt%) were
UP400) was used to create a stable emulsion and form smaller added. To prevent fracture of the capsules, mixing was performed
capsules. After homogenization by sonication, 6.45 g formaldehyde at ambient temperature with a low agitation speed (200 rpm) for
was added to the emulsion. To complete the reaction, the solution 15 min. To achieve a dry film thickness of about 150 ␮m, the epoxy-
was slowly heated to 55 ◦ C and stirred by a mechanical agitator for containing micro/nanocapsules were coated on one side of a clean
4 h. The agitation speed in this step is one of the studied parame- carbon steel plate test piece (10 cm × 10 cm, 2 mm thick) by a film
ters (X1 ). For each test, the rotational speed of agitation, sonication applicator, and the coating was allowed to cure for 7 days at room
power, and sonication time were selected based on the operating temperature.
62 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69
Fig. 1. Comparisons of the predicted and actual values for the (a) percentage of encapsulated healing agent and (b) particle size.
Salt spray test
The paint film durability, qualitative corrosion response, and
ability of defects to repair and self heal were investigated by the
salt spray test. A 7 cm X-cut scratch was created on coated plates
with different amounts of micro/nanocapsules. The samples were
placed in the salt spray cabinet at an angle of 30◦ from vertical,
and the salt spray test was performed according to the ASTM B117
standard. A 5 wt% sodium chloride solution (pH ≈6.7) was used in
the test.
Tafel polarization test
The Tafel polarization test was performed in a 300 mL glass cell
containing 3.5 wt% sodium chloride solution as a corrosive medium
using an Ivium Stat (Ivium Technologies, Eindhoven, the Nether-
land) as the potentiostat/galvanostat device. A 1 cm2 coated carbon
steel plate was used as the working electrode, Ag/AgCl (saturated,
3 M KCl) was used as the reference electrode, and a platinum rod
was used as the counter electrode. The Tafel polarization test was
performed in the range −300 to 300 mV with respect to the open
circuit potential with a scan rate of 2 mV/s. All of the electrochem-
ical measurements were performed in accordance with the ASTM
G3 standard. Tafel extrapolation was used to determine the cor-
rosion current density, and the corrosion rate was calculated by
Faraday’s law.
Results and discussion
Statistical analysis
A five-level central composite experimental design and RSM
were used to investigate the effects of the three independent
variables on the response functions and to obtain the optimal
conditions for the maximum percentage of encapsulated linseed
oil (Y1 ) and minimum particle size (Y2 ). Design Expert statistical
software (trial version 7.0) was used to generate the experimen-
tal matrix and analyze the obtained results. Based on preliminary
studies, three factors were selected as independent variables: the
rotational speed of agitation (X1 ), sonication power (X2 ), and son-
ication time (X3 ). The obtained results from 20 experiments are
given in Table 2. The Fisher F-test, coefficient of determination R2 ,
adjusted R2 (adj-R2 ), and lack-of-fit test were used to confirm the
adequacy of the model. ANOVA showed a significant response sur-
face model (p-value ≤0.05) with a high determination coefficient. In
general, the results of the response surface may be poor or mislead-
ing. Therefore, the accuracy of the obtained results and models are
expressed by R2 and adj-R2 . Model summary statistics showed that
the Two-factor interaction (2FI) model has the maximum adj-R2
and predicted R2 values, and thus provides the best fitting model.
Table 3 shows the ANOVA results for each response. The valid-
ity of the models was confirmed by lack-of-fit testing, as shown
in Table 3. R2 values of 0.9071 and 0.9548 were obtained for the
percentage of encapsulated linseed oil and particle size responses,
respectively. The obtained results indicate that there are good cor-
relations between the responses and independent variables.
Fitting of polynomial equations
The statistical experimental design allowed the responses to
be modeled by fitting second-order polynomials. Two polynomial
equations to predict the responses were provided by the Design
Expert software. The equations obtained in terms of coded factors
are as follows:
Y 1 = 92.14900 − 0.011325X 1 − 0.010050X 2 + 1.31708X 3
+ (6.000 × 10−5 )X 1 X 2 − (7.83333 × 10−4 )X 1 X 3
− (2.86667 × 10−3 )X 2 X 3 , (4)
Y 2 = 4.08437 − (5.24125 × 10−3 )X 1 − 0.010097X 2 − 0.16987X 3
+ (1.42500 × 10−5 )X 1 X 2 + (9.41667 × 10−5 )X 1 X 3
+ (4.90000 × 10−4 )X 2 X 3 . (5)
To check the model adequacy, ANOVA was used and the results
are given in Table 3. The significance of the model terms was
estimated based on the F probability (Prob > F) at 95% confidence
intervals. Model F-values of 21.15 and 54.87 were obtained for the
percentage of encapsulated linseed oil and particle size responses,
respectively (see Table 3). The large F values show that most of
the variation in the responses can be explained by the regression
models. The associated p-value was used to estimate whether F
is sufficiently large to indicate statistical significance. It should be
noted that p-values less than 0.05 indicate that the model terms are
significant. Therefore, all of the investigated factors are significant.
Lack-of-fit p-values of 0.7811 and 0.5009 were obtained for the
percentage of encapsulated linseed oil and particle size responses,
respectively, indicating that the lack-of-fit is not significant rela-
tive to pure error (Table 3). The R2 value gives the proportion of
the total variation in the response predicted by the model, indicat-
ing the ratio of the sum of squares owing to regression to the total
sum of squares. A high R2 value (i.e., close to 1) is desirable and
reasonable agreement with the adj-R2 value is necessary. A high R2
 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69 63

Table 3
Results of ANOVA for each response.

Source Y1 : ELO% Y2 : Size of microcapsules

Sum of square DF F value p-Value (Prob > F) Sum of square DF F value p-Value (Prob > F)

Model 4.18 6 21.15 <0.0001 Significant 0.18 5 54.87 <0.0001 Significant

X1 (rpm) 0.62 1 18.93 0.0008 0.026 1 39.67 <0.0001
X2 (W) 0.30 1 9.17 0.0097 – – – –
X3 (min) 0.62 1 18.68 0.0008 0.062 1 95.75 <0.0001
X1 X2 0.72 1 21.84 0.0004 0.041 1 62.54 <0.0001
X1 X3 0.44 1 13.40 0.0029 6.384 × 10−3 1 9.83 0.0079
X2 X3 1.48 1 44.86 <0.0001 0.043 1 66.55 <0.0001
Residual 0.43 13 8.442 × 10−3 13
Lack of fit 0.20 8 0.56 0.7811 Not significant 5.297 × 10−3 8 1.05 0.5009 Not significant
Pure error 0.23 5 3.145 × 10−3 5
Cor total 4.61 19 0.19 18

R2 0.9071 0.9548
Adj-R2 0.8642 0.9374
CV% 0.20 3.52
Adeq-precision 18.617 24.755

Fig. 2. Normal probability plots of the studentized residuals for the (a) percentage of encapsulated healing agent and (b) particle size.

coefficient ensures satisfactory adjustment of the quadratic model
to the experimental data. The adj-R2 value shows the variation
in the response predicted by the model adjusted for the number
of terms in the model. In this case, the values of the determi-
nation coefficients (R2 = 0.9071 for the ELO% and R2 = 0.9548 for
the microcapsule size) indicate that only 9.29% and 4.52% of the
total variation is not explained by the ELO% and microcapsule size
models, respectively. The adj-R2 values are also sufficiently high
to verify the high accuracy of the models. Adequate precision is
determined by the signal to noise ratio. Both models show adequate
precision (signal to noise ratio of values greater than 4), indicating
model discrimination.
Two approaches were used to validate the developed mathe-
matical models: plotting the actual responses versus the predicted
values and normal probability plots of the residuals. Fig. 1(a) and
(b) shows the plots of the actual against predicted responses, which
indicate that the predicted response values are in good agreement
with the actual values. Therefore, the developed models are valid.
The actual responses were obtained based on the experiments
and the predicted values were calculated based on the proposed
model. The studentized residuals are plotted against the predicted
responses in Fig. 2(a) and (b). It is clear that the errors are ran-
domly spread in the experimental matrix because the residuals fall
on a straight line. This result verifies that the variance for all of the
response values is constant, which indicates model term validation.
Effects of the independent parameters
First, the effects of the studied independent variables on the
responses were investigated without considering the interaction
with other variables. For this purpose, all of the other variables
were fixed at the center point.
Fig. 3(a)–(c) shows the effects of the rotational speed of agi-
tation (X1 ), sonication power (X2 ), and sonication time (X3 ) on
the percentage of the encapsulated linseed oil in the prepared
micro/nanocapsules. Considering the p-value (0.0008, Table 3) of
the rotational speed of agitation, it can be concluded that the rota-
tional speed has a significant effect on the ELO%. This means that
in this range of the rotational speed (300–700 rpm), the increase in
the ELO% can be mainly attributed to the rotational speed increase.
Fig. 3(a) shows the effect of the rotational speed on the ELO%. From
this figure, by increasing the rotational speed, the ELO% increases. In
emulsion systems, coalescence and coagulation of oil droplets leads
to phase separation and improper encapsulation of the oil (Chern,
2008). Therefore, as expected, increasing the rotational speed leads
to an increased intensity of mixing and prevents oil droplet coagu-
lation, consequently increasing the encapsulation efficiency.
The p-value of the sonication power is 0.0097 (Table 3), so it
can be concluded that the sonication power has a significant effect
on the ELO%. This means that for the examined range of sonication
power, the increase in the ELO% can be definitely attributed to the
sonication power increase. The effect of the sonication power on
the ELO% is shown in Fig. 3(b), which shows that increasing the
sonication power increases the ELO%.
Considering the p-value for the sonication time (0.0008, Table 3),
it can be concluded that the sonication time has a significant
effect on the ELO%. As shown in Fig. 3(c), increasing the sonica-
tion time increases the ELO%. According to the results obtained by
other researchers, oil droplets can be broken and dispersed with
increasing sonication time. Therefore, owing to the lack of suffi-
64 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69
Fig. 3. Effects of the independent parameters on the percentage of encapsulated linseed oil: (a) rotational speed of agitation (X1 ), (b) sonication power (X2 ), and (c) sonication
time (X3 ).
cient surfactant in the emulsion, the oil droplets coagulate and form
larger droplets, which leads to improper encapsulation (Kourniatis,
Spinelli, Piombini, & Mansur, 2010).
The effects of rotational speed of agitation (X1 ) and sonica-
tion time (X3 ) on the size of the micro/nanocapsules are shown
in Fig. 4(a) and (b). In emulsion systems, the rotational speed of
the agitator is the most important parameter affecting the particle
size. The p-values of this variable are <0.0001 (Table 3). This means
that the rotational speed of the agitator has a significant effect on
the micro/nanocapsule size. As shown in Fig. 4(a), increasing the
rotation speed increases the intensity of mixing, and consequently
the particle size is reduced.
Considering the p-value for the sonication time (<0.0001,
Table 3), it can be concluded that the sonication time has a signifi-
cant effect on the particle size. As shown in Fig. 4(b), increasing the
sonication time increases the particle size. The size of oil droplets in
an emulsion depends on various parameters, including the amount
of oil in the water phase, the type and amount of surfactant, and
the sonication intensity (Leong, Wooster, Kentish, & Ashokkumar,
2009). Because of the lack of available surfactants in the emulsion
at a constant surfactant concentration, increasing the homogeniza-
tion time causes an increase in the size of the emulsion droplets, and
consequently the size of the produced capsules (Kourniatis et al.,
2010). The same results have been obtained by other researchers
(Buffo, Reineccius, & Oehlert, 2001; Chanamai & McClements, 2001;
Gunji, Ueda, Ogata, & Nakagaki, 1992).
Interaction of parameters
A three-dimensional response surface is used to show the
effect of the interaction between two variables on the response.
The effects of the studied independent parameters on the ELO%
and micro/nanocapsule size are shown in Fig. 5(a)–(f). The ELO%
increases with increasing each of the studied parameters (Fig. 3).
However, to achieve the maximum ELO% at long homogenization
time, the other studied parameters (i.e., the rotational speed of the
agitator and sonication power) must be set at the minimum studied
levels. The main reason for this behavior is because of dispersion
of the surfactant and oil droplets in the long-term homogenization
process. Because of the higher density of the oil than the surfactant,
oil droplets coagulate before their surfaces are cover with surfac-
tant. Therefore, the number of produced micro/nanocapsules and
the ELO% decrease. In addition, increasing the homogenization time
increases the micro/nanocapsule size. As shown in Fig. 5(d), at con-
stant homogenization time, the micro/nanocapsule size increases
with increasing sonication power. Thus, a low rotational speed
of the agitator and high sonication power gives the smallest
micro/nanocapsules.
Optimization of the parameters
After statistical analysis of the data, the optimal variables were
determined to achieve minimum capsule size and maximum ELO%.
The optimum conditions to obtain minimum particle size and max-
imum ELO%, and the predicted values are given in Table 4. One
sample was prepared according to the optimum conditions spec-
ified in Table 4. The ELO% and particle size of this sample were
measured. This data was then compared with the predicted data.
The experimental and predicted values of each response are given
in Table 4. The percentage relative error between the predicted and
experimental values of each response was calculated by
 PV − EV 
%Bias = × 100, (6)
PV
where PV is the predicted value and EV is the experimental value.
The percentage bias shows that there is reasonable agreement
between the predicted and experimental values in all cases. The
results obtained in other studies that used the same materials for
 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69 65

Fig. 4. Effects of the independent parameters on the size of the micro/nanocapsules: (a) rotational speed of agitation (X1 ) and (b) sonication time (X3 ).

Fig. 5. Effects of the interaction of each two parameters (X1 & X2 for (a) and (d); X1 & X3 for (b) and (e); X2 & X3 for (c) and (f)) on the (a)–(c) ELO% and (d)–(f) size of the
micro/nanocapsules.

Table 4
Optimized parameters and results.

Factors ELO% Particle size

X1 (rpm) X2 (W) X3 (min) Predicted Actual %Bias Predicted (␮m) Actual (␮m) %Bias

594 350 3 93.9 92.7 1.28 0.574 0.553 3.66

Comparison of this study with previously published studies

This study Kouhi et al. (2013) Suryanarayana et al. (2008) Behzadnasab et al. (2014)

Approach In-situ polymerization In-situ polymerization In-situ polymerization In-situ polymerization

Core material Linseed oil Linseed oil Linseed oil Linseed oil
Shell material Urea–formaldehyde Urea–formaldehyde Urea–formaldehyde Urea–formaldehyde
Core content (%) 92–94 91–94 80 Not reported
Capsule size (␮m) 0.553–0.932 0.45–6 5–100 10–150
66 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69

Fig. 6. SEM micrographs of the micro/nanocapsules: (a) sample 6, (b) sample 8, (c) sample 9, (d) sample 13, (e) sample 18, and (f) sample 19 (Table 2).

Fig. 7. FTIR spectra of the (a) urea–formaldehyde resin and extracted shell material and (b) pure linseed oil and extracted core oil.

the core and shell, and in-situ polymerization for microencapsu-
lation are also given in Table 4, and they are compared with the
results obtained in this study. It is worth noting that in the current
study, the optimal conditions were determined to simultaneously
achieve minimum capsule size and maximum ELO%.
Micro/nanocapsule characterization
SEM micrographs of some micro/nanocapsules are shown in
Fig. 6. All of the particles show an acceptable spherical shape with
a rough surface. The spherical shape of the particles leads to better
dispersion of the particles in the epoxy resin and the rough sur-
face is suitable for their adhesion to the epoxy matrix. The capsules
have a smooth inner surface, and the wall thickness of the prepared
capsules is very low relative to their total volume.
FTIR analysis was performed to confirm formation of the
urea–formaldehyde shell and encapsulation of linseed oil in the
prepared micro/nanocapsules. The FTIR spectra of the pure lin-
seed oil, urea–formaldehyde resin, extracted shell materials, and
extracted oil are shown in Fig. 7. The characteristic absorption
bands for the urea–formaldehyde resin (Fig. 7(a)) in the wave
number ranges 1560–1680, 1440–1470, 1670–1760, 1142–1286,
and 3200–3500 cm−1 are assigned to the N H, C H, C O, C N,
and O H groups in the urea–formaldehyde resin, respectively. The
presence of the same absorption bands in the FTIR spectrum of
the shell material confirms formation of the urea–formaldehyde
shell. The FTIR spectra of the pure linseed oil and extracted core oil
are shown in Fig. 7(b). The characteristic peaks in the wave num-
ber ranges 1600–1675, 1625–1750, 1440–1500, and 2926 cm−1
correspond to C C, C O, C H, and C H stretching, respectively.
Therefore, these spectra confirm that linseed oil is encapsulated as
a core by a urea–formaldehyde shell.
Qualitative corrosion study
Salt spray tests were performed on three samples coated
with glass flake epoxy paint containing different percentages
of micro/nanocapsules. The first was an epoxy coating without
micro/nanocapsules (M-0), the second included 10 wt% optimized
micro/nanocapsules (M-10), and the third sample included 20 wt%
optimized micro/nanocapsules (M-20). The onset of corrosion is
observed after 10 h exposure for the M-0 sample, as shown in
Fig. 8. In comparison, the severity of corrosion is very low for
the M-10 and M-20 samples. Fig. 9 shows optical microscopy
 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69 67

Fig. 8. Qualitative corrosion study by the salt spray technique for samples coated with glass flake epoxy paint (150 ␮m thickness dry film) containing micro/nanocapsule
contents of 0 wt% (M-0), 10 wt% (M-10), and 20 wt% (M-20).

Fig. 9. Optical microscopy images of M-0 (a) before and (b) one year after the salt spray test, M-10 (c) before and (d) one year after the salt spray test, and M-20 (e) before
and (f) one year after the salt spray test.

images showing the extent of healing of the above-mentioned optimized micro/nanocapsules (Fig. 9(c) and (d)) does not com-
samples. The epoxy coating without micro/nanocapsules (Fig. 9(a) pletely heal, while the epoxy coating with 20 wt% optimized
and (b)) shows no healing action. The coating loaded with 10 wt% micro/nanocapsules completely heals, as shown in Fig. 9(e) and
68 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69
Table 5
Electrochemical parameters extracted from the Tafel polarization curves.
Samples Electrochemical parameters
Ecorr (V vs. Ag/AgCl) Icorr (A/cm2 ) Rp ( cm2 )
M-0 −0.500 5.975 × 10−7 5.341 × 104
M-10 −0.701 3.714 × 10−7 6.663 × 104
M-20 −0.247 1.535 × 10−8 4.519 × 106
a
ˇa = slope of the anodic Tafel reaction, when plotted on base 10 logarithmic scale in V/decade.
b
ˇc = slope of the cathodic Tafel reaction when plotted on base 10 logarithmic scale in V/decade.
Fig. 10. Tafel polarization curves for carbon steel test pieces coated with glass-flake
epoxy paint (150 ␮m thickness dry film) containing micro/nanocapsules contents
of 0 wt% (M-0), 10 wt% (M-10), and 20 wt% (M-20).
(f). This indicates the role of the healing agent in repairing the glass
flake paint inhomogeneity in the cut area and the successive healing
process.
Tafel polarization test
The Tafel polarization curves for carbon steel test pieces coated
with glass flake epoxy paint (150 ␮m thickness dry film) containing
different contents of micro/nanocapsules are shown in Fig. 10. The
electrochemical parameters extracted by the Tafel extrapolation
technique are given in Table 5. The results show that addition of
microcapsules to the epoxy significantly increases the corrosion
resistance of the coating. The corrosion rate of the M-0 sample is
on average 34.5 times greater than that of the M-20 sample.
Conclusions
Synthesized micro/nanocapsules were optimized in terms of
the particle size and percentage of encapsulated linseed oil as
the healing agent. The investigation showed that the rotational
speed of agitation, sonication time, and sonication power have
significant effects on the percentage of encapsulated oil. The per-
centage of encapsulated linseed oil ranged from 92.09% to 93.21%.
Increasing the rotational speed gives smaller particles, and long-
term sonication leads to larger capsules. The sizes of the prepared
micro/nanocapsules ranged from 554 nm to 13.68 ␮m. Therefore,
the use of ultrasonic homogenization for a short time with low
power gives the best results.
Linseed oil that is properly encapsulated within a shell of
urea–formaldehyde with a spherical capsule shape and rough sur-
face can be better dispersed in the paint medium. Release of linseed
oil as a healing agent in paint was successfully performed and the
modified epoxy showed excellent performance in terms of corro-
sion resistance.
ˇa a (V/decade) ˇc b (V/decade) Corrosion rate (mm/y)
0.093 0.345 0.0069
0.103 0.128 0.0043
0.276 0.379 0.0002
References
Behzadnasab, M., Esfandeh, M., Mirabedini, S. M., Zohuriaan-Mehr, M. J., &
Farnood, R. R. (2014). Preparation and characterization of linseed oil-filled
urea–formaldehyde microcapsules and their effect on mechanical properties of
an epoxy-based coating. Colloids and Surfaces A: Physicochemical and Engineering
Aspects, 457, 16–26.
Blaiszik, B. J., Sottos, N. R., & White, S. R. (2008). Nanocapsules for self-healing mate-
rials. Composites Science and Technology, 68(3), 978–986.
Buffo, R. A., Reineccius, G. A., & Oehlert, G. W. (2001). Factors affecting the emul-
sifying and rheological properties of gum acacia in beverage emulsions. Food
Hydrocolloids, 15(1), 53–66.
Chanamai, R., & McClements, D. J. (2001). Depletion flocculation of beverage emul-
sions by gum arabic and modified starch. Journal of Food Science, 66(3), 457–463.
Chern, C. S. (2008). Principles and applications of emulsion polymerization. New York:
John Wiley & Sons.
Gharsallaoui, A., Roudaut, G., Beney, L., Chambin, O., Voilley, A., & Saurel, R. (2012).
Properties of spray-dried food flavours microencapsulated with two-layered
membranes: Roles of interfacial interactions and water. Food Chemistry, 132(4),
1713–1720.
Gil, S., Cuenca, N., Romero, A., Valverde, J. L., & Sánchez-Silva, L. (2014). Optimization
of the synthesis procedure of microparticles containing gold for the selective
oxidation of glycerol. Applied Catalysis A: General, 472, 11–20.
Gunji, M., Ueda, H., Ogata, M., & Nakagaki, M. (1992). Studies on the oil-in-water
emulsions stabilized with gum-arabic by using the turbidity ratio method. Jour-
nal of the Pharmaceutical Society of Japan, 112(12), 906–913.
Hafizi, A., Ahmadpour, A., Heravi, M. M., Bamoharram, F. F., & Khosroshahi, M. (2012).
Alkylation of benzene with 1-decene using silica supported preyssler heteropoly
acid: Statistical design with response surface methodology. Chinese Journal of
Catalysis, 33(2), 494–501.
Hatami Boura, S., Peikari, M., Ashrafi, A., & Samadzadeh, M. (2012). Self-healing abil-
ity and adhesion strength of capsule embedded coatings—Micro and nano sized
capsules containing linseed oil. Progress in Organic Coatings, 75(4), 292–300.
Karger-Kocsis, J. (2010). Epoxy polymers new materials and innovations. New York:
Wiley Online Library.
Khoee, S., Payandeh, S. H., Jafarzadeh, P., & Asadi, H. (2016). Size and core content
optimization of epoxy nanocapsules by response surface methodology for use
in self-healing coatings. Smart Materials and Structures, 25(8), 084014.
Kouhi, M., Mohebbi, A., Mirzaei, M., & Peikari, M. (2013). Optimization of smart
self-healing coatings based on micro/nanocapsules in heavy metals emission
inhibition. Progress in Organic Coatings, 76(7), 1006–1015.
Kourniatis, L. R., Spinelli, L. S., Piombini, C. R., & Mansur, C. R. (2010). Formation of
orange oil-in-water nanoemullsions using nonionic surfactant mixtures by high
pressure homogenizer. Colloid Journal, 72(3), 396–402.
Leong, T. S. H., Wooster, T. J., Kentish, S. E., & Ashokkumar, M. (2009). Minimising
oil droplet size using ultrasonic emulsification. Ultrasonics Sonochemistry, 16(6),
721–727.
Liu, X., Zhang, H., Wang, J., Wang, Z., & Wang, S. (2012). Preparation of epoxy micro-
capsule based self-healing coatings and their behavior. Surface and Coatings
Technology, 206(23), 4976–4980.
Matsuda, T., Jadhav, N., Kashi, K. B., Jensen, M., Suryawanshi, A., & Gelling, V. J. (2016).
Self-healing ability and particle size effect of encapsulated cerium nitrate into
pH sensitive microcapsules. Progress in Organic Coatings, 90, 425–430.
Möbus, K., Siepmann, J., & Bodmeier, R. (2012). Zinc–alginate microparticles for
controlled pulmonary delivery of proteins prepared by spray-drying. European
Journal of Pharmaceutics and Biopharmaceutics, 81(1), 121–130.
Myers, R. H., Montgomery, D. C., & Anderson-Cook, C. M. (2009). Response surface
methodology: Process and product optimization using designed experiments (3rd
ed.). New York: John Wiley & Sons.
Nazzaro, F., Orlando, P., Fratianni, F., & Coppola, R. (2012). Microencapsulation in food
science and biotechnology. Current Opinion in Biotechnology, 23(2), 182–186.
Noshadi, I., Amin, N. A. S., & Parnas, R. S. (2012). Continuous production of biodiesel
from waste cooking oil in a reactive distillation column catalyzed by solid het-
eropolyacid: Optimization using response surface methodology (RSM). Fuel, 94,
156–164.
Ollier, R. P., Penoff, M. E., & Alvarez, V. A. (2016). Microencapsulation of epoxy resins:
Optimization of synthesis conditions. Colloids and Surfaces A: Physicochemical
and Engineering Aspects, 511, 27–38.
Otálora, M. C., Carriazo, J. G., Iturriaga, L., Nazareno, M. A., & Osorio, C. (2015).
Microencapsulation of betalains obtained from cactus fruit (Opuntia ficus-indica)
by spray drying using cactus cladode mucilage and maltodextrin as encapsulat-
ing agents. Food Chemistry, 187, 174–181.
 A. Ebrahiminiya et al. / Particuology 36 (2018) 59–69 69

Rule, J. D., Sottos, N. R., & White, S. R. (2007). Effect of microcapsule size on the Sun, D., Zhang, H., Tang, X. Z., & Yang, J. (2016). Water resistant reactive microcap-
performance of self-healing polymers. Polymer, 48(12), 3520–3529. sules for self-healing coatings in harsh environments. Polymer, 91, 33–40.
Samadzadeh, M., Boura, S. H., Peikari, M., Kasiriha, S. M., & Ashrafi, A. (2010). A review Suryanarayana, C., Rao, K. C., & Kumar, D. (2008). Preparation and characterization of
on self-healing coatings based on micro/nanocapsules. Progress in Organic Coat- microcapsules containing linseed oil and its use in self-healing coatings. Progress
ings, 68(3), 159–164. in Organic Coatings, 63(1), 72–78.
Sauvant-Moynot, V., Gonzalez, S., & Kittel, J. (2008). Self-healing coatings: An alter-
native route for anticorrosion protection. Progress in Organic Coatings, 63(3),
307–315.
Sharma, P., Shukla, S., Lochab, B., Kumar, D., & Roy, P. K. (2014). Microencapsulated
cardanol derived benzoxazines for self-healing applications. Materials Letters,
133, 266–268.