ISSN 1451 - 9372(Print)
ISSN 2217 - 7434(Online)
APRIL-JUNE 2016
Vol.22, Number 2, 127-234
www.ache.org.rs/ciceq
Journal of the
Association of Chemical Engineers of
Serbia, Belgrade, Serbia
Vol. 22
Belgrade, April-June 2016
Chemical Industry & Chemical Engineering
Quarterly (ISSN 1451-9372) is published
quarterly by the Association of Chemical
Engineers of Serbia, Kneza Miloša 9/I,
11000 Belgrade, Serbia
Editor:
Vlada B. Veljković
veljkovic@yahoo.com
Editorial Office:
Kneza Miloša 9/I, 11000 Belgrade, Serbia
Phone/Fax: +381 (0)11 3240 018
E-mail: shi@yubc.net
www.ache.org.rs
For publisher:
Tatijana Duduković
Secretary of the Editorial Office:
Slavica Desnica
Marketing and advertising:
AChE Marketing Office
Kneza Miloša 9/I, 11000 Belgrade, Serbia
Phone/Fax: +381 (0)11 3240 018
Publication of this Journal is supported by the
Ministry of Education and Science of the
Republic of Serbia
Subscription and advertisements make payable
to the account of the Association of Chemical
Engineers of Serbia, Belgrade, No. 205-217271, Komercijalna banka a.d., Beograd
Computer typeface and paging:
Vladimir Panić
Printed by:
Faculty of Technology and Metallurgy,
Research and Development Centre of Printing
Technology, Karnegijeva 4, P. O. Box 3503,
11120 Belgrade, Serbia
Abstracting/Indexing:
Articles published in this Journal are indexed in
Thompson Reuters products: Science Citation
TM
Index - Expanded - access via Web of
®
SM
Science , part of ISI Web of Knowledge
No. 2
CONTENTS
Congli Mei, Ming Yang, Dongxin Shu, Hui Jiang, Guohai Liu,
Zhiling Liao, Soft sensor based on Gaussian process
regression and its application in erythromycin
fermentation process .......................................................... 127
Marija R. Miladinović, Marija B. Tasić, Olivera S. Stamenković, Vlada B. Veljković, Dejan U. Skala, Further
study on kinetic modeling of sunflower oil methanolysis
catalyzed by calcium-based catalysts ................................ 137
Dejan Markovic, Ivana Karadzic, Vukoman Jokanovic, Ana
Vukovic, Vesna Vucic, Biological aspects of application of nanomaterials in tissue engineering .................... 145
Safiye Bağcı, Ayhan Abdullah Ceyhan, Adsorption of
methylene blue onto activated carbon prepared from
Lupinus albus ..................................................................... 155
Didem Özçimen, Tufan Salan, Removal of reactive dye
Remazol Brilliant Blue R from aqueous solutions by
using anaerobically digested sewage sludge based
adsorbents .......................................................................... 167
Sonja Pecić, Ninoslav Nikićević, Mile Veljović, Milka Jadranin, Vele Tešević, Miona Belović, Miomir Nikšić, The
Influence of extraction parameters on physicochemical
properties of special grain brandies with Ganoderma
lucidum ............................................................................... 181
Monika Lutovska, Vangelce Mitrevski, Ivan Pavkov, Vladimir
Mijakovski, Milivoj Radojčin, Mathematical modelling
of thin layer drying of pear .................................................. 191
Biljana Damjanović-Vratnica, Svetlana Perović, Tiejun Lu,
Regina Santos, Effect of matrix pretreatment on the
supercritical CO2 extraction of Satureja montana
essential oil ......................................................................... 201
M. Barahoei, A. Zeinolabedini Hezave, S. Sabbaghi, Sh.
Ayatollahi, Copper oxide nano-fluid stabilized by ionic
liquid for enhancing thermal conductivity of reservoir
formation: Applicable for thermal enhanced oil recovery
processes............................................................................ 211
Naïma Moudir, Nadji Moulaï-Mostefa, Yacine Boukennous,
Silver micro- and nano-particles obtained using
different glycols as reducing agents and measurement
of their conductivity ............................................................. 227
Activities of the Association of Chemical Engineers of Serbia are supported by:
- Ministry of Education, Science and Technological Development, Republic of Serbia
- Hemofarm Koncern AD, Vršac, Serbia
- Faculty of Technology and Metallurgy, University of Belgrade, Belgrade, Serbia
- Faculty of Technology, University of Novi Sad, Novi Sad, Serbia
- Faculty of Technology, University of Niš, Leskovac, Serbia
- Institute of Chemistry, Technology and Metallurgy, University of Belgrade, Belgrade, Serbia
Journal of the
Association of Chemical Engineers of
Serbia, Belgrade, Serbia
EDITOR-In-Chief
Vlada B. Veljković
Faculty of Technology, University of Niš, Leskovac, Serbia
E-mail: veljkovicvb@yahoo.com
ASSOCIATE EDITORS
Jonjaua Ranogajec
Srđan Pejanović
Milan Jakšić
Faculty of Technology, University of
Novi Sad, Novi Sad, Serbia
Department of Chemical Engineering,
Faculty of Technology and Metallurgy,
University of Belgrade, Belgrade, Serbia
ICEHT/FORTH, University of Patras,
Patras, Greece
EDITORIAL BOARD (Serbia)
Đorđe Janaćković, Sanja Podunavac-Kuzmanović, Viktor Nedović, Sandra Konstantinović, Ivanka Popović
Siniša Dodić, Zoran Todorović, Olivera Stamenković, Marija Tasić, Jelena Avramović
ADVISORY BOARD (International)
Dragomir Bukur
Ljubisa Radovic
Texas A&M University,
College Station, TX, USA
Pen State University,
PA, USA
Milorad Dudukovic
Peter Raspor
Washington University,
St. Luis, MO, USA
University of Ljubljana,
Ljubljana, Slovenia
Jiri Hanika
Constantinos Vayenas
Institute of Chemical Process Fundamentals, Academy of Sciences
of the Czech Republic, Prague, Czech Republic
University of Patras,
Patras, Greece
Maria Jose Cocero
Xenophon Verykios
University of Valladolid,
Valladolid, Spain
University of Patras,
Patras, Greece
Tajalli Keshavarz
Ronnie Willaert
University of Westminster,
London, UK
Vrije Universiteit,
Brussel, Belgium
Zeljko Knez
Gordana Vunjak Novakovic
University of Maribor,
Maribor, Slovenia
Columbia University,
New York, USA
Igor Lacik
Dimitrios P. Tassios
Polymer Intitute of the Slovak Academy of Sciences,
Bratislava, Slovakia
Denis Poncelet
ENITIAA, Nantes, France
National Technical University of Athens,
Athens, Greece
Hui Liu
China University of Geosciences, Wuhan, China
FORMER EDITOR (2005-2007)
Professor Dejan Skala
University of Belgrade, Faculty of Technology and Metallurgy, Belgrade, Serbia
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 127−135 (2016)
CONGLI MEI
MING YANG
DONGXIN SHU
HUI JIANG
GUOHAI LIU
ZHILING LIAO
School of Electrical and
Information Engineering, Jiangsu
University, Zhenjiang, China
SCIENTIFIC PAPER
UDC 004:681.586:615.33:663.14
DOI 10.2298/CICEQ150125026M
CI&CEQ
SOFT SENSOR BASED ON GAUSSIAN
PROCESS REGRESSION AND ITS
APPLICATION IN ERYTHROMYCIN
FERMENTATION PROCESS
Article Highlights
• A systematic soft sensor modeling method based on GPR and PCA is proposed
• The variance of the predicted output was designed on the output uncertainty of the
GPR model
• Practical applications show the superiority of the proposed soft sensor method
Abstract
Erythromycin fermentation is a typical microbial fermentation process. Soft
sensors can be used to estimate the biomass of Erythromycin fermentation
process due to their relative low cost, simple development, and ability to predict difficult-to-measure variables. However, traditional soft sensors, e.g. artificial neural network (ANN) soft sensors, support vector machine (SVM) soft
sensors, etc., cannot represent the uncertainty (measurement precision) of
outputs, which results in difficulties in practice. Gaussian process regression
(GPR) provides a novel framework to solve regression problems. The output
uncertainty of a GPR model follows Gaussian distribution, expressed in terms
of mean and variance. The mean represents the predicted output. The variance can be viewed as the measure of confidence in the predicted output that
distinguishes the GPR from NN and SVM soft sensor models. We propose a
systematic approach based on GPR and principal component analysis (PCA)
to establish a soft sensor to estimate biomass of erythromycin fermentation
process. Simulations on industrial data from an erythromycin fermentation
process show the proposed GPR soft sensor has high performance of modeling the uncertainty of estimates.
Keywords: fermentation process; soft sensor; uncertainly; Gaussian
process regression; principle component analysis.
Online control and optimization methods of
fermentation processes depend on reliable and accurate measurements of key variables. However, not all
important variables are available in real-time to allow
timely action. Lack of key measurements can be
contributed to two factors: 1) insufficient automation
of some sensitive analysis; 2) even if real-time measurement is possible, the cost of installing an additional sensor may not be economically attractive.
Advanced control algorithms cannot always be used
in microbial fermentation processes due to the lack of
Correspondence: C. Mei, School of Electrical and Information
Engineering, Jiangsu University, Zhenjiang 212013, P.R. China.
E-mail: clmei@ujs.edu.cn
Paper received: 25 January, 2015
Paper revised and accepted: 10 July, 2015
measurements of important biological parameters.
Thus, people usually rely on past experience to operate microbial fermentation process. This situation
increases costs of production and operation in the
fermentation industry. In order to achieve effective
monitoring and control of the fermentation process,
on-line measurement problems of the key parameters, such as biomass concentration, etc., should be
resolved [1].
Recently, the soft sensor technique has seen
wide applications in bioprocesses. Compared to traditional measurement methods, it is an innovative detection method developed to estimate unmeasurable
variables or difficult-to-measure variables through online and measurable secondary variables. Some typical methods have been recognized to have strong
127
C. MEI et al.: SOFT SENSOR BASED ON GAUSSIAN PROCESS…
potential in bioprocesses, such as mechanism models
[2], adaptive observers [3], PLS (partial least squares)
[4], filtering techniques (e.g., Kalman filter, extended
Kalman filter, etc.) [5,6]. In addition, data-driven
methods have been also widely investigated in the
field of microbial fermentation, e.g., SVM (support
vector machine) [7–10] and ANN (artificial neural
network) [11–13]. The root mean square error (RMSE)
is usually used as a measure to evaluate soft sensors
[10]. According to the measurement principle, measurements are meaningless without precision information (uncertainty). For measuring instruments, the
uncertainty of measurements is easy to be estimated
by the precision of instruments. However, for the
above-mentioned soft sensors, the uncertainty of outputs is difficult to determine. This is an obvious drawback of existing soft sensor methods.
Gaussian process regression (GPR) is nonparametric and flexible, meaning that the complexity of the
model grows as more data points are received and
predictions can be obtained without giving the
unknown function y(x) an explicit parameterization. As
a Bayesian probabilistic model, GPR is able to model
the uncertainty inherent in noisy data, and give full
probabilistic predictions or error bars. They naturally
grow in regions away from training data where there
is high uncertainty about the prediction model. The
Gaussian process was first used by O’Hagan [14] as
an alternative approach to the artificial neural network
approach. Williams and Rasmussen [15] first described GPR in a machine learning context. With new
knowledge in machine learning, Gaussian process
has recently seen explosive growth. However, only a
limited number of soft sensing applications based on
GPR have been reported [16–19].
Modern industry processes are always loaded
with many sensors. It is well known that it is not
generally possible to use all available sensor variables as soft sensor inputs, because measurement
redundancy generally makes the calibration of the
regression model troublesome. The satisfactory performance of soft sensors is likely to be achieved if
only those sensor variables, named secondary variables, that are most sensitive to the primary variables
are employed. Several methods were proposed to
select best variables. Simple statistical analyses have
been conducted for process variables to identify a
subset of measurements for use in the soft sensors
[20]. Ma et al. used a conventional stepwise variable
selection method to develop a soft sensor [21]. Alternative Methods on principal component analysis
(PCA) [22] and singular value decomposition (SVD)
[23] were also investigated. For the effectiveness and
128
Chem. Ind. Chem. Eng. Q. 22 (2) 127−135 (2016)
simplicity, the PCA based method was widely used in
the field of soft sensor modeling to select secondary
variables [22,24].
To the best of authors’ knowledge, little information in literature has been devoted to soft sensor
development based on GPR combined PCA. In this
paper, a systematic soft sensor modeling approach
using GPR and PCA for the Erythromycin fermentation process was presented. The focus of this
method was modeling the uncertainty using GPR in
bioprocesses. To simplify the complexity of the GPR
based soft sensor, the PCA-based variable selection
method was used to select input variables of the soft
sensor. Finally, the proposed soft sensor was evaluated on industrial data from an actual Erythromycin
fermentation process.
MATERIAL AND METHODS
Erythromycin fermentation process
Erythromycin is a macrolide antibiotic. In terms
of structure, this macrocyclic compound contains a
14-membered lactone ring with ten asymmetric centers and two sugars (L-cladinose and D-desosamine),
making it a compound very difficult to produce via
synthetic methods. Erythromycin is produced from a
strain of the actinomycete Saccharopolyspora erythraea [25]. The flow chart of fermentation reaction is
shown in Figure 1. Also, the simplified structure and
parameters of reactor are shown in Figure 2.
Generally, the process parameters of microbial
fermentation processes include physical parameters,
chemical parameters and biological parameters.
Physical parameters typically include temperature (T)
and pressure (P) of the fermenter, air flow (FA), flow of
cooling water (FW), water temperatures of the inlet
and the outlet (T1, T2), rotate speed of the stirring
motor (RMP), etc. Chemical parameters include pH,
dissolved oxygen content (DO), etc. The physical and
chemical parameters can be automatically measured
by instruments. However, biological parameters, such
as cell concentration, metabolite concentration, substrate concentration and growing rate of the cell, etc.,
cannot be measured by sophisticated real-time measuring instruments. Therefore, soft sensors were proposed to estimate biological parameters in bioprocesses.
For an erythromycin fermentation process, biomass concentration plays a decisive role in the final
product (erythromycin) concentration. Thus, the primary way of ensuring product quality of Erythromycin
is to control biomass concentration, which can be
affected by many process factors. In this process,
C. MEI et al.: SOFT SENSOR BASED ON GAUSSIAN PROCESS…
Chem. Ind. Chem. Eng. Q. 22 (2) 127−135 (2016)
Figure 1. The flow chart of fermentation reaction.
secondary variables can be determined by statistical
methods on process variables including time, dissolved oxygen, pH, dextrin flow, soybean oil flow,
isopropanol flow, water flow, volume of dextrin, volume of soybean, volume of isopropanol, volume of
water, temperature, air pressure, stirring speed, air
flow, etc.
Figure 2. The diagram of erythromycin fermentation process.
The microbial growth curve is nonlinear and can
be divided into four different phases: A) lag phase, B)
log phase or exponential phase, C) stationary phase,
and D) death phase. This basic batch culture growth
model draws out and emphasizes aspects of microbial growth, which may differ from the growth of any
other creatures. In fact, even in batch culture, the four
phases are not always well defined. Cells do not
reproduce in synchrony and their growth rate during
exponential phase is often not a constant, instead of a
slowly decaying rate [26], so it is difficult to accurately
predict the growth of microorganisms, especially
during the exponential phase.
Data acquisition
During this research, training and test data sets
were collected from Zhenjiang Medicine Co., Ltd,
China. The whole Erythromycin fermentation process
lasted about seven days. In this work, 180 group
points have been collected by physical sensors during
each batch fermentation process, and the biomass
concentration was obtained by off-line laboratory
analysis every hour synchronously. Nine batches ran
under similar environment and initial conditions. The
sample data set was divided into two parts: training
set and test set. The batches 1-6 were used as training data sets, and another batches were used as test
data sets. Fourteen process variables were measured
in the process besides the time variable. The curves
of all process variables are shown in Figure 3. The
legend numbers 1-9 denote the 1st to the 9th batch
fermentation process. The subfigures a-n represent
different process variables.
Gaussian Process Regression (GPR)
In this section, the basic GPR method [27,28] is
introduced.
Suppose a training set:
D = {( xi , yi )} , (i = 1,..., l )
(1)
129
CI&CEQ 22 (2) 127−135 (2016)
10
50
8
pH
100
6
200
100
soybean oil flow
0
100
0
water flow
5
0
volume of water volume of soybean
x 10
1
0
500
0
100
1000
500
0
4000
2000
0
0.1
air pressure
40
30
20
40
0.05
0
1000
30
20
0
0
4
2
50
200
10
air flow
stirring speed
temperature volume of isopropanol volume of dextrin isopropanol flow
dextrin flow
dissolved oxygen
C. MEI et al.: SOFT SENSOR BASED ON GAUSSIAN PROCESS…
0
50
100
Time(h)
150
200
500
0
0
50
100
Time(h)
150
200
Figure 3. Variation curves of process variables.
where xi ∈ Rn denotes the n-dimensional input
(observation) vector, y i ∈ R denotes a scalar output
or target, and l is the number of training samples. The
task of a common regression is to estimate the conditional expectation of the dependent variable given
the independent variables, i.e., learning the mapping
relationship between input x and output y in terms of
training set, and predict the most likely value of the
test data x * . A Gaussian process is commonly specified by its mean and covariance function. To make
130
calculation easy, we generally set a zero mean GP
model on the function variables as:
f ( x )  N (0, k ( x, x * ))
(2)
For the regression problems, considering the
following model:
y = f ( x ) + ε , ε  N (0,σ n2 )
(3)
C. MEI et al.: SOFT SENSOR BASED ON GAUSSIAN PROCESS…
where ε is the noise, and σ n2 is the variance of the
noise. In this paper, considering fermentation is a
highly complex process and is susceptible to interference, we should add the noise to the prior (i.e.,
observed value), such that:
y  N (0, K ( X , X ) + σ n2I )
(4)
(
)
where K ( X , X ) = k ( x i , x j ) is a symmetric and nonnegative definite covariance matrix, I is the n × n
dimensional identity matrix. Then we can write the
joint distribution of the observed target values and the
function values at the test point under the prior as:
y 
  N
f * 
  K ( X , X ) + σ n2I
 0, 
  K (x *, X )
 
K (X , x * )  

k ( x * , x * )  
(5)
where f * is the GP posterior (predicted value, here
we consider only 1-dimensional output), K ( X , x * ) is
the n ×1 covariance matrix between training and test
cases, K ( x * , X ) is the transposition of K ( X , x * ) ,
k ( x * , x * ) is the covariance function evaluated at x *
and itself. Then, we can get the predictive distribution
of GPR as:
(
f * | X , y, x *  N f * ,cov(f * )
)
(6)
where:
f * = K ( x * , X )[K ( X , X ) + σ n2I ]−1 y
cov(f * ) = k ( x * , x * ) − K ( x * , X )[K ( X , X ) +
+σ n2I ]−1K ( X , x * )
(7)
(8)
Obviously, the predicted mean and variance of
the test point x * can be respectively written as Eq. (7)
and Eq. (8).
The covariance function is the crucial ingredient
in a Gaussian process predictor, as it encodes our
assumptions about the function which we want to get
[27]. To be a valid covariance function, it must be
positive semidefinite. There are some covariance
functions that can be used. A suitable covariance
function is:
 1 n

w t ( x it − x tj )2  + v 1δ ij

 2 t =1

k ( x i , x j ) = v 0 exp  −
(9)
where Θ = [w 1,...,w n ,v 0 ,v 1]T are the hyperparameters
of the covariance functions. v 0 represents a overall
measure of the prior knowledge, or put simply, it
controls the typical amplitude of covariation. v 1 represents the variation of noise and it follows Gaussian
distribution. δ ij is Kronecker delta. w t can be seen
as the weight of the t-th dimension, i.e., how many
information contained in the corresponding input
Chem. Ind. Chem. Eng. Q. 22 (2) 127−135 (2016)
dimension t. The first term of Eq. (9) expresses the
belief that inputs that are close to each other give rise
to outputs that are close to each other or that are
highly correlated. The second term of Eq. (9) involving the hyperparameter v 1 = σ 2 , i.e., the variance of
the noise model for the outputs and therefore only
occurs when i = j .
The test value can be predicted if the covariance
function had been given. However, the property of the
GP model cannot be guaranteed if the parameters of
covariance function have not been optimized. The
GPR is in the framework of Bayesian rule [27,28].
Thus, the marginal likelihood function of Bayesian
formalism, typically negative log marginal likelihood
function, was chosen as an optimization objective
function of hyperparameters, which can be written as:
1
2
L(Θ) = − lg(| C |) −
1 T −1
l
y C y − log(2π )
2
2
(10)
Then the optimized hyperparameters can be
obtained by solving the partial derivatives of the
marginal likelihood function as:
∂L (Θ) 1 
∂C 
= tr  (ααT − C −1)

2 
∂Θi
∂Θi 
(11)
where α = C −1y . Then, by making the partial derivatives be equal to zero, the optimal solution of hyperparameters can be obtained.
RESULTS AND DISCUSSION
Results of data processing
The correspondence between labels and process variables is shown in Table 1.
Table 1. Variable labels
Label
Meaning
1
Time
2
Dissolved oxygen tension
3
pH value
4
Dextrin flow
5
Soybean oil flow,
6
Isopropanol flow
7
Water flow
8
Volume of dextrin
9
Volume of soybean
10
Volume of isopropanol
11
Volume of water
12
Temperature
13
Air pressure
14
Stirring speed
15
Air flow
131
C. MEI et al.: SOFT SENSOR BASED ON GAUSSIAN PROCESS…
Chem. Ind. Chem. Eng. Q. 22 (2) 127−135 (2016)
ensure the cumulative contribution rate higher than
95%.
The PCs, also called latent variables, can be
used to model processes. However, the PCs have no
obvious physical meaning. The PCs represent the
most sensitive directions to the primary variable. The
cumulative contribution rate of the first six PCs was
over 95%, so the portfolio coefficients of only the first
6 PCs are given in Table 3. The coefficients can be
interpreted as measures of the contribution of variables to the PCs. To reduce the dimensions of input
variables, a specific approach was used to select the
variables with the greatest contribution to the high
sensitive PCs as inputs of soft sensors: the coefficient
with the largest absolute value corresponding process
variable in every PC is selected as a secondary variable.
Then, in Table 3, process variables 8, 4, 4, 2, 11
and 3, corresponding to the V1 to V6 respectively,
were selected as secondary variables. Finally, process variable 2, 3, 4, 8 and 11 were determined as
input variables of a soft sensor those denote dissolved oxygen, pH, dextrin flow, volume of dextrin
and volume of water. The output variable is biomass
concentration.
To avoid attributes in greater numeric dominating those in smaller numeric ranges, the collected
data was normalized by the following expression:
x i' = −1 +
x i − x i ,min
x i ,max − x i ,min
(12)
×2
where x i is the ith original data, x i' is the normalized
value of x i , x i ,max and x i ,min are respectively the
maximum and minimum value of x i , i represents the
dimension of data. Another advantage of the normalization method is to effectively reduce numerical difficulties during the calculation. Furthermore, we employed the PCA method to reduce the dimensions (see
in [29], concrete methods see in [30]). The results of
PCA are shown in Table 2.
Table 2. The contribution of the tenth principle component in
PCA algorithm
k-th principal
component
Contribution rate
%
Cumulative contribution
rate, %
1
55.3062
55.3062
2
21.5165
76.8227
3
7.9241
84.7468
4
5.7096
90.4564
5
3.2777
93.7341
6
2.4002
96.1343
Results of the GPR soft sensor
7
1.4697
97.6040
8
1.2109
98.8149
The proposed GPR soft sensor was built on
training data set after variable selection. Then, process data of three batches was used as test data set
to evaluate the proposed GPR soft sensor. Figure 4
shows the test results.
Figure 4a-c gives the output dynamics of the
GPR soft sensor and true data synchronously. The
predictions are expressed as mean (solid line) with
2*std (std, standard deviation) error bars (dotted
9
0.4661
99.2810
10
0.3592
99.6402
Table 2 shows that the first five principal components (PCs) accounted for 93.73% cumulative contribution, after adding the sixth one, which reaches
96.13%. Hence, the first six PCs were selected to
Table 3. Portfolio coefficients of PCA
1
2
3
4
5
6
7
V1
-0.3394
-0.0660
-0.1784
0.0259
-0.0047
0.2299
-0.1412
V2
0.1776
-0.0980
-0.2148
0.5741
-0.0112
0.5619
-0.0330
V3
0.0653
0.1314
0.0602
-0.7042
-0.0118
0.5692
-0.2591
V4
-0.0239
-0.5572
0.4876
0.0145
-0.0368
0.2819
0.3608
V5
-0.1043
0.2963
-0.1224
0.1188
0.0532
-0.0411
0.3835
V6
-0.0367
0.5091
0.6137
0.0817
-0.1681
-0.0503
0.0541
8
9
10
11
12
V1
-0.4106
-0.3434
-0.3907
-0.2595
V2
0.2090
0.1527
0.2262
-0.2325
V3
0.1067
0.0547
0.0873
V4
-0.1382
-0.0464
V5
-0.0205
V6
0.0135
132
13
14
15
0.3550
0.1204
-0.3368
0.1206
0.2093
-0.1132
0.1325
0.1406
-0.2103
-0.1181
-0.0721
0.0468
0.0074
-0.0500
0.1250
-0.2879
-0.0797
-0.1259
0.3091
-0.1057
-0.0245
-0.5674
-0.5248
-0.2695
-0.1603
-0.1135
-0.0250
-0.0384
-0.0856
0.3870
-0.3177
0.0639
0.2409
30
20
true data
mean for GP
mean +/- 2*std
10
0
0
50
100
Time(h)
(a) batch 7
150
200
Chem. Ind. Chem. Eng. Q. 22 (2) 127−135 (2016)
40
30
20
true data
mean for GP
mean +/- 2*std
10
0
0
50
100
Time(h)
(b) batch 8
150
Biomass concentration(g/L)
40
Biomass concentration(g/L)
Biomass concentration(g/L)
C. MEI et al.: SOFT SENSOR BASED ON GAUSSIAN PROCESS…
200
40
30
20
true data
mean for GP
mean +/- 2*std
10
0
0
50
100
150
Time(h)
(c) batch 9
200
Figure 4. Variation curves of biomass prediction by the GPR soft sensor.
lines). The regions between the two dotted lines
depict the confidence intervals.
To investigate the output uncertainty of the GPR
soft sensor with different conditions, we set an abnormal condition in batch 9. We assumed that the DO
sensor in batch 9 was abnormal (Figure 5). Figure 6
shows the output of the GPR soft sensor with the
abnormal DO sensor.
Discussions
Figure 4 shows that the predicted microbial
growth (output, i.e., mean curves of the GPR soft
sensor) approximately fits the microbial growth curve.
The predicted biomass concentration curves, shown
in Figure 4a-c, can also be divided into four stages:
lag phase (in the beginning, though it is not obvious),
log phase (during 10th-50th h), stationary phase
(during 50th-150th h), and death phase (after the 150th
h). We can find that the predicted area of uncertainty,
between the mean ±2std curves, in the log phase is
bigger than that in other phases. That can be interpreted by fermentation mechanisms. Actually, biomass varies greatly and is with high uncertainty in the
exponential phase. We can also found that only a few
Dissolved oxygen
80
70
60
50
40
30
0
50
100
Time(h)
150
200
Biomass concentration(g/L)
Figure 5. Variation curve of the abnormal DO sensor.
50
40
30
20
10
mean for GP
mean +/- 2*std
0
-10
0
50
100
Time(h)
150
200
Figure 6. Output uncertainty of the GPR soft sensor with the abnormal DO sensor.
133
C. MEI et al.: SOFT SENSOR BASED ON GAUSSIAN PROCESS…
true measured points were out of the range of the
area of uncertainty (see Figure 4a). This shows the
predicted uncertainty reflects well the distribution of
outputs. Therefore, we can conclude that the GPR
soft sensor is suitable to model the uncertainty in fermentation processes for its probabilistic characteristic.
The performance of the GPR soft sensor with an
abnormal condition was also investigated. As Figure 6
shows, when the DO sensor is abnormal, uncertainty
output of the soft sensor is significantly greater than
that in Figure 4c. That shows the uncertainty is sensitive to the distribution of input variables. The main
reason is that the output uncertainty (i.e., prediction
distribution) of a GPR model depends on the distribution of input variables [27]. From the view of soft
sensor maintenance, the uncertainty output can be
used to design indexes of abnormal instruments or
process states.
Chem. Ind. Chem. Eng. Q. 22 (2) 127−135 (2016)
REFERENCES
[1]
J.R. Leigh, Modelling and Control of Fermentation Processes, Peter Peregrinus Ltd., London, 1987
[2]
Y. Zhao, Studies on modeling and control of continuous
biotechnical processes, Ph.D. Thesis, Norwegian University of Science and Technology, 1996
[3]
G. Bastin, D. Dochain, O.L. Estimation, Adaptive control
of bioreactors, Elsevier Science, Amsterdam, 1990
[4]
H. Zhang, B Lennox, J. Process. Contr. 14 (2004) 41-50
[5]
A. Jahanmiri, H. Rasooli, Biochem. Eng. J. 24 (2005)
115-123
[6]
D.A. Gee, W.F. Ramirez, Biotechnol. Prog. 12 (1996)
132-140
[7]
G. Liu, D. Zhou, H. Xu, C. Mei, Expert Syst. Appl. 37
(2010) 2708-2713
[8]
L. Tian, Q.Q. Wang, A.Z. Cao, Adv. Mater. Res. 734
(2013) 2998-3002
[9]
X. Dai, W. Wang, Y. Ding, Z. Sun, Comput. Chem. Eng.
30 (2006) 1203-1225
[10]
K. Desai, Y. Badhe, S.S. Tambe, B.D. Kulkarni, Biochem.
Eng. J. 27 (2006) 225-239
[11]
A. Borin, M.F. Ferrão, C. Mello, L. Cordi, L.C. Pataca, N.
Durán, R.J. Poppi, Anal. Bioanal. Chem. 387 (2007)
1105-1112
[12]
J. Peng, F. Meng, Y. Ai, Bioresour. Technol. 138 (2013)
345-352
[13]
S.G. Kulkarni, A.K. Chaudhary, S. Nandi, S.S. Tambe,
B.D. Kulkarni, Biochem. Eng. J. 18 (2004) 193-210
[14]
A. O'Hagan, J. Kingman, J. R. Stat. Soc., B 40 (1978) 1–42
[15]
C.K. Williams, C.E. Rasmussen, in: Advances in Neural
Information Processing Systems 8, D. S. Touretzky, Ed.,,
The MIT Press, Cambridge, 1996, p. 514
[16]
R. Grbić, D Slišković, P Kadlec, Comput. Chem. Eng. 58
(2013), 84-97
[17]
F. di Sciascio, A.N. Amicarelli, Comput. Chem. Eng. 32
(2008) 3264-3273
[18]
X. Li, H. Su, J. Chu, Chinese. J. Chem. Eng. 17 (2009)
95-99
[19]
X. Yuan, Z. Ge, Z. Song, Chemometr. Intell. Lab. 138
(2014) 97-109
[20]
S. Bhartiya, J. R. Whiteley, ISA. T. 40 (2001) 307-323
[21]
M.D. Ma, J.W. Ko, S.J. Wang, M.F. Wu, S.S. Jang, S.S.
Shieh, D.S.H. Wong, Control. Eng. Pract. 17 (2009)
1026-1034
[22]
E. Zamprogna, M. Barolo, D.E. Seborg, J. Process.
Control 15 (2005) 39-52
[23]
T. Mejdell, S. Skogestad, Ind. Eng. Chem. Res. 30 (1991)
2543-2555
[24]
M. Shakil, M. Elshafei, M. A. Habib, F. Maleki, Comput.
Electr. Eng. 35 (2009) 578-586
[25]
C. Mims, H. Dockrell, R. Goering, I. Roitt, D. Wakelin,
Medical Microbiology (3rd ed.), Mosby Ltd., London,
2004, p. 489
[26]
A. Novick, Annu. Rev. Microbiol. 9 (1955) 97-110
CONCLUSION
Microbial fermentation process is complex and
nonlinear. The growth rates of microorganisms in different phases vary greatly, resulting in significant
biomass concentration changes. For fermentation
processes, dramatic changes usually exist in the
exponential phase. Considering the uncertainty of
microorganism growth, GPR combined PCA were
used to build a soft sensor to measure biomass concentration on line in Erythromycin fermentation process. The proposed soft sensor not only gives the
predicted values of biomass, but also gives the
uncertainty of outputs synchronously. The adequate
performance of the proposed soft sensor of modeling
the uncertainty of estimates was evaluated with data
obtained off-line of an industrial Erythromycin fermentation process. The output uncertainty can be
regarded as measurement precision. Moreover, it is a
useful reliability index of soft sensor or process condition in real applications.
Acknowledgements
The authors gratefully acknowledge the financial
support provided by Natural Science Foundation of
Jiangsu Province of China (grant no. BK20130531),
the Priority Academic Program Development of
Jiangsu Higher Education Institutions (grant no.
PAPD 6) and Science & Technology Innovation
Foundation of Ministry of Science & Technology for
Small-Medium Enterprises, China (grant no.
12C26213202207).
134
C. MEI et al.: SOFT SENSOR BASED ON GAUSSIAN PROCESS…
[27]
C.E. Rasmussen, C.K. Williams, Gaussian processes for
machine learning, The MIT Press, Cambridge, 2006, p.
16
[28]
M. Gibbs, D.J. MacKay, Efficient implementation of Gaussian processes, Unpublished manuscript, Cavendish
Laboratory, Cambridge, 1997, p. 2
CONGLI MEI
MING YANG
DONGXIN SHU
HUI JIANG
GUOHAI LIU
ZHILING LIAO
School of Electrical and Information
Engineering, Jiangsu University,
Zhenjiang, China
NAUČNI RAD
Chem. Ind. Chem. Eng. Q. 22 (2) 127−135 (2016)
[29]
I. Jolliffe, Principal component analysis, John Wiley &
Sons, New York, 2005, p. 28
[30]
S. Valle, W. Li, S.J. Qin, Ind. Eng. Chem. Res. 38 (1999)
4389-4401.
SOFT SENZOR ZASNOVAN NA GAUSOVOJ
REGRESIJI I NJEGOVA PRIMENA U PROCESU
FERMENTACIJE ERITROMICINA
Fermentacija eritromicina je tipičan mikrobni process. Za određivanje biomase u procesu
fermentacije eritromicina mogu se koristiti soft senzori zbog njihove niske cene, jednostavne izrade i sposobnosti da predvide teško merljive promenljive. Međutim, tradicionalni
soft senzori, kao npr. soft senzori zasnovani na veštačkim neuronskim mrežama (ANN),
mašini sa vektorima podrške (SVM) i tako dalje, ne mogu da predstavljaju mernu nesigurnost (preciznost merenja), što dovodi do teškoća u praksi. Gausova regresija (GPR) omogućuje novi okvir za rešavanje regresijskih problema. Merna nesigurnosti GPR modela
sledi Gausovu raspodelu, izraženu preko srednje vrednosti i varijanse. Srednja vrednost
prestavlja predviđeni izlaz. Varijansa se može posmatrati kao mera pouzdanosti predviđenog izlaza koji razdvaja GPR od ANN i SVM modela soft senzora. Predlaže se sistematski pristup zasnovan na GPR i analizi glavnih komponenti (PCA), da bi se uspostavio
soft senzor za procenu biomase u procesu fermentacije eritromicina. Simulacije sa industrijskim podacima iz procesa fermentacije eritromicina pokazuju da predloženi GPR model
soft senzora ima visok stepen modelovanja procena merne nesiguurnosti.
Ključne reči: fermentacije, soft senzori, merna nesigurnost, Gausova regresija,
analiza glavnih komponenata.
135
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 137−144 (2016)
MARIJA R. MILADINOVIĆ1
MARIJA B. TASIĆ1
OLIVERA S. STAMENKOVIĆ1
VLADA B. VELJKOVIĆ1
DEJAN U. SKALA2
1
University of Niš, Faculty of
Technology, Leskovac, Serbia
2
University of Belgrade, Faculty of
Technology and Metallurgy,
Belgrade, Serbia
SCIENTIFIC PAPER
UDC 662.756.3:544.4:663:66:546.41
DOI 10.2298/CICEQ150618027M
CI&CEQ
FURTHER STUDY ON KINETIC MODELING
OF SUNFLOWER OIL METHANOLYSIS
CATALYZED BY CALCIUM-BASED
CATALYSTS
Article Highlights
• Kinetics of methanolysis reaction catalyzed by calcium-based catalysts
• Kinetic model involves TAG mass transfer and chemically controlled regions
• Effects of reaction conditions on parameters of the kinetic models are analyzed
Abstract
The kinetic model, which was originally developed for sunflower oil methanolysis catalyzed by CaO⋅ZnO, was examined for several other calcium-based
catalysts like neat CaO, quicklime and Ca(OH)2. This model including triacylglycerols mass transfer- and chemically-controlled regimes demonstrated a
good agreement with the experimental data in terms of a high coefficient of
determination (0.971±0.022) and acceptable mean relative percentage deviation (±15.9%). Hence, this model is recommended for modeling the kinetics of
sunflower oil methanolysis over calcium-based catalysts under widely ranging
reaction conditions.
Keywords: biodiesel; calcium-based catalysts; kinetics; methanolysis;
modeling.
Biodiesel, which is commonly produced from
plant oils and animal fats by alcoholysis (usually
methanolysis), is an alternative fuel for diesel engines
as being based on renewable raw materials, easily
available, technically appropriate, economically competitive and environmentally beneficial. Although the
current production of biodiesel is mainly carried out by
homogeneous base-catalyzed methanolysis of edible
vegetable oils, it is expected that it will soon be replaced by heterogeneous catalysis based on the use of
basic solid metal oxides, hydroxides and complexes,
zeolites, hydrotalcites or supported catalysts. Because
of high basicity, mild reaction condition, high biodiesel
yield, low cost and easy preparation, calcium compounds, such as oxide [1–3], hydroxide [4] or methoxide [5], are increasingly used in biodiesel production from various oily or fatty feedstocks. Calcium
oxide (CaO) is used as neat [1–3,6], doped [7,8],
supported [9] or mixed [10,11]. Its great advantage is
Correspondence: V.B. Veljković, University of Niš, Faculty of
Technology, 16000 Leskovac, Bulevar oslobođenja 124, Serbia.
E-mail: veljkovicvb@yahoo.com
Paper received: 18 June, 2015
Paper revised: 12 July, 2015
Paper accepted: 12 July, 2015
the possibility of preparation from natural [12,13] or
waste [14,15] sources.
The kinetic models that have been used so far
for methanolysis reaction over calcium-based catalysts suppose zero [16,17] or first [2,3,5,10,11] order
with respect to triacylglycerols (TAGs). Some
researchers suppose [2,13] that the order of this
reaction varies from the zeroth order in the beginning
to the first one in the later period of methanolysis.
Also, separate correlations for mass transfer- and
reaction rate-controlled stages of the methanolysis
reaction of the first order with respect to TAGs are
reported [3,4]. Lukić et al. [10,11] have recently combined these two correlations into a single one that
predicts the reaction rate during the whole course of
methanolysis of refined and used sunflower oil over
CaO.ZnO. Moreover, this model has successfully
been employed by Sánchez et al. [15] for describing
the kinetics of jojoba oil methanolysis over mussel
shell-CaO. Hence, this model might be powerful for
modeling the kinetics of methanolysis of different
feedstocks over various calcium-based catalysts.
Therefore, the present work focuses on the adequacy, reliability and accuracy of the kinetic model of
137
M.R. MILADINOVIĆ et al.: FURTHER STUDY ON KINETIC MODELING…
Lukić et el. [10] for the sunflower oil methanolysis
catalyzed by several solid calcium-based catalysts
like neat CaO, quicklime and Ca(OH)2. This work was
aimed at testing the generalization capability of this
kinetic model under widely ranging reaction conditions in the presence of the above-mentioned calcium-based catalysts on the basis of statistical evaluation. In addition, this model was compared with the
model of Miladinović et al. [13] that has recently been
verified for several calcium-based catalyst [18].
Theoretic background
The accepted kinetic model is based on several
common assumptions [10]:
1. The overall methanolysis reaction is represented by the following stoichiometric equation:
Catalyst
⎯⎯⎯⎯
→ 3R + S
A + 3B ←⎯⎯⎯
⎯
(1)
where A is TAG, B is methanol, R is fatty acid methyl
esters (FAMEs), i.e., biodiesel and S is glycerol. In
fact, this reaction is a complex one that occurs via
three consecutive reversible reactions where monoand diacylglycerols (MAGs and DAGs, respectively)
are formed as intermediates. Since concentrations of
MAGs and DAGs are too small because of their much
faster consumption rates, compared to that of TAGs,
Eq. (1) can represent the complex methanolysis
reaction:
1, The reaction mixture is perfectly mixed, so its
composition and the catalyst space distribution are
uniform.
2. The methanolysis occurs at the catalyst particles surface between methoxide ions adsorbed on
the active centers and TAG molecules in the liquid
phase close to the active centers.
3. The rate of methanol mass transfer towards
catalytically active sites, the reverse reaction rate as
well as the adsorption/desorption rates of methanol,
FAME and glycerol do not limit the overall process
rate.
4. The methanolysis process is controlled by
TAG mass transfer limitation in the initial reaction
period and by the chemical reaction in the latter
period. The mass transfer limitation, caused by the
small available active specific catalyst surface [3,4],
depends on the formation of fine emulsion of methanol into the oil. In addition, FAMEs act as a cosolvent, enhancing the miscibility of the reactants and
increasing the TAG mass transfer rate and the methanolysis reaction rate. The first order reaction rate law
with respect to TAG is adopted:
138
Chem. Ind. Chem. Eng. Q. 22 (2) 137−144 (2016)
k ( k mt,A )0 1+ α x Aβ 
dx A
=
(1− x A )
dt
k + ( k mt,A ) 1+ α x Aβ 
(2)
0
where xA is the TAG conversion degree, t is time, k is
the pseudo first-order reaction rate constant, (kmt,A)0 is
the overall TAG volumetric mass transfer coefficient
at the beginning of the process, and α and β are fitting
parameters. Thus, the model includes four parameters: α, β, k and (kmt,A)0. The detailed derivation of
Eq. (2) can be found elsewhere [3,10].
The homogeneously catalyzed methanolysis
reaction is ignorable due to negligible leaching of the
catalyst. This is experimentally verified for neat CaO
[19], a CaO-based catalyst [20] and Ca(OH)2 [4] in
batch stirred reactors. Since CaO.ZnO is practically
insoluble in methanol, homogeneous catalysis is also
considered as negligible [10]. According to Granados
et al. [1], the contribution of the homogeneous methanolysis arising from the leached calcium species is
negligible if the catalyst loading is larger than 1% to
the oil, this condition being fulfilled in the above-mentioned studies.
The internal diffusion rate inside catalyst particles does not influence the methanolysis reaction
rate. This assumption is verified for neat CaO [3] and
Ca(OH)2 [4]. Also, powdered quicklime is a mesoporous material with pores significantly greater than
the diameter of a typical TAG molecule which minimizes internal diffusion limitations [13]. In addition,
CaO⋅ZnO is characterized as a catalyst with small
surface area and low porosity [10].
The neutralization of free fatty acids is negligible
because of their very small content in the oil used
(acid value of 0.24-0.29 mg KOH/g) [3,4,10,13]. Also,
the saponification reaction is ignorable and consequently, the catalyst amount remains constant during
the methanolysis.
EXPERIMENTAL
Experimental data
The experimental data for the sunflower oil
methanolysis reactions catalyzed by neat CaO [3],
quicklime [13], Ca(OH)2 [4] and CaO.ZnO [10] in
batch stirred reactors were taken from the earlier
investigations. Median particle size, median pore diameter, specific surface area, basic strength and
basicity of the used catalysts are presented in Table
1. The reaction conditions (type and volume of reactor, methanol-to-oil mole ratio, catalyst loading, temperature and stirring speed) applied in the abovementioned studies are given in Table 2.
M.R. MILADINOVIĆ et al.: FURTHER STUDY ON KINETIC MODELING…
Chem. Ind. Chem. Eng. Q. 22 (2) 137−144 (2016)
Table 1. Textural properties, basic strength and basicity of used catalyst; Dmed, particle - median particle size, Dmed, pore – median pore
diameter and SBET – specific surface area
Dmed, particle / μm Dmed, pore / nm
Sample
CaO
SBET / m2 g–1
Basic strength H_
Basicity, mmol/g
Reference
2.8
15.2
13.7
15<H_<18.4
2.46
[21]
<500
18.8
9.5
15<H_<18.4
2.16
[13]
Ca(OH)2
2.0
-
-
-
-
[4]
CaO⋅ZnO
4.3
-
3.1
9.3<H_<10
0.58
[11]
Qucklime
Modeling techniques and computer software
RESULTS AND DISCUSSION
Kinetic parameters of Eq. (2) were obtained by
fitting the sets of experimental data using the Mathematica™ v.9 (trial version). A demonstration [22] which
generated realistic-looking curves was modified and
rewritten using Eq. (2) in order to provide both good
initial value guessing and computing of each kinetic
parameter.
TAG and FAME concentrations, cA and cR, were
calculated from values of TAG conversion degree as
follows:
c A = c A0 (1 − x A )
(3)
c R = 3c A0 x A
(4)
Statistical estimation of kinetic models
The significance of the models was statistically
evaluated from the coefficient of determination (R2)
and the mean relative percent deviation (MRPD),
respectively:
n
R2 =
(y
i =1
n

i =1
MRPD =
p,i
− y a,i
)
( y p,i − y m )
100
n
n

i =1
2
(5)
2
y p,i − y a,i
y a,i
(6)
where yp and ya are predicted and experimental
values of the TAG conversion degree (%), ym is the
mean value of the TAG conversion degree (%), and n
is the number of experimental runs.
Kinetic modeling of sunflower oil methanolysis
Predicted values of parameters of the kinetic
model, Eq. (2), as well as R2- and MRPD-values as
the statistical measures of their adequacy, reliability
and accuracy, are presented in Table 3. The kinetic
model shows very high R2 (0.971±0.022) for all four
catalysts, showing that it predicts the time variation of
TAG conversion degree within the ranges of the reaction conditions applied reliablly. In addition, the
MRPD of ±15.9% (based on a set of 240 data) shows
a good agreement between the predicted and experimental values of TAG conversion degree. Thus, the
kinetic model is reliable and accurate not only for
CaO⋅ZnO but also for neat CaO, quicklime and
Ca(OH)2.
Recently, Tasić et al. [16] have recommended
the kinetic model of Miladinović et al. [13] as general
for describing the kinetics of sunflower oil methanolysis over calcium-based catalyst. This is a three-parameter model:
(1 − x A )(c C0 + 3c A0 x A )
dx A
= km
dt
K + c A0 (1 − x A )
(7)
where xA is the conversion degree of TAG, t is time,
cA0 is the initial TAG concentration, K is the model
parameter defining the TAG affinity for the catalyst
active sites, cC0 is the hypothetic initial FAME concentration corresponding to the initial available active catalyst surface and km is the apparent reaction rate
constant:
Table 2. Reaction conditions (type and volume of reactor, methanol-to-oil molar ratio, catalyst loading, temperature and stirring speed)
applied in the studies of sunflower oil methanolysis catalyzed by the calcium compounds; BSR – batch stirred reactor
Reaction conditions
Catalyst
Catalyst amount
Temperature
%
°C
CaO
1-10
60
6:1
900
250
[3]
Quicklime
1-10
60
6:1 - 18:1
900
250
[13]
Ca(OH)2
1-10
60
6:1
900
250
[4])
2
60-96
10:1
300
300
[10]
0.5, 1, 2
60
6:1, 10:1
300, 1000
1000
[11]
CaO∙ZnO
MeOH/oil mole ratio
Stirring intensity
rpm
Reactor type
BSR
Reactor volume
Reference
mL
139
M.R. MILADINOVIĆ et al.: FURTHER STUDY ON KINETIC MODELING…
Chem. Ind. Chem. Eng. Q. 22 (2) 137−144 (2016)
Table 3. Parameters and statistical estimation of kinetic models for methanolysis of sunflower oil
Catalyst
Type
Amount
Temperature
°C
Methanol-to-oil
mole ratio
Parameters
β
k / min-1
mass%
CaO
1.0
Quicklime
min
60
6:1
0.990
±11.5
0.991
±11.7
5.0
20.750
80
0.991
±10.0
10.0
20.800
78
0.992
±19.2
6.219
200
0.975
±15.2
2.5
10.331
120
0.927
±16.0
5.0
18.993
80
0.958
±16.2
10.0
19.474
78
0.989
±10.5
12:1
b
0.042
6.219
200
0.977
±17.4
9.183
120
0.933
±34.0
5.0
13.718
80
0.938
±25.8
10.0
14.065
78
0.903
±28.6
0.047
±15.3
4.523
200
0.969
2.5
10.331
120
0.988
±9.6
5.0
13.717
80
0.954
±33.3
10.0
14.892
78
0.929
±18.9
18:1
±17.8
0.021
1.440
213
0.979
0.042
5.100
177
0.998
±5.5
5.0
0.064
10.200
95
0.996
±11.1
10.0
0.077
10.200
95
0.995
±15.8
6-10
60
0.048
2.5
2.0
Mussel shell CaO
a
1.00
2.5
1.0
CaO ZnO
%
120
1.0
a
MRPD
200
1.0
.
R2
-1
3.480
6:1
0.067
α
10.600
60
1.00
4
2.5
1.0
Ca(OH)2
Statistical evaluation
(kmt,A)0×10
6:1
60
1.00
1.37
0.043
2.100
378
0.960
±21.7
70
10:1
1.75
0.051
24.400
167
0.989
±11.4
84
1.27
0.083
28.500
192
0.992
±2.8
96
2.90
0.120
326.000
215
0.994
±1.8
c
59.7
na
45-65
-6
6:1-12:1
0.23 6.59×10 -1.68×10
-5
-
d
na
Values of parameters were taken from Lukić et al.[10]; values of parameters were taken Sánchez et al. [15] for the jojoba oil methanolysis; cthe parameter
is a function of methanol concentration; dnot available
k m = kc B0c cat
b
(8)
where k is the actual reaction rate constant, while cB0
and ccat are initial concentrations of methanol and
catalyst, respectively. Equation (7) is shown as a
reliable predictor of the time variation of TAG conversion degree in the sunflower oil methanolysis over
the same calcium-based catalysts used in the present
study within the ranges of the reaction conditions
applied as indicated by high coefficient of determination (R2 > 0.93) and relatively small MRPD (±9.1%)
[18]. Based on the MRPD, the model of Miladinović et
al., Eq. (7), is better for CaO, quicklime and CaO.ZnO,
while the model of Lukić et al., Eq. (2), is more
adequate for Ca(OH)2.
The kinetic model of Lukić et al., Eq. (2), was
used for calculation of TAG and FAME concentrations
during the sunflower oil methanolysis over the calcium-based catalysts. The accuracy of the kinetic
models was examined by comparing the predicted
and experimental values of FAME and TAG concen-
140
trations. Figures 1 and 2 show the variation of TAG
and FAME concentrations with the progress of the
sunflower oil methanolysis catalyzed by neat CaO,
quicklime and Ca(OH)2. As it can be seen, the calculation of FAME and TAG concentrations obtained by
Eq. (2) agree quite well with the experimental concentrations independently of methanol-to-oil molar ratio
and catalyst amount.
So far, different mechanisms of methanolysis
reaction catalyzed by calcium-based catalysts have
been proposed in the literature [2,3,23], but the exact
nature of any of them has not been confirmed yet.
The results of the present kinetic analysis might indicate the possible reaction mechanism of methanolysis reaction catalyzed by calcium-based catalysts.
Probably, methanolysis reaction proceeds through the
simplified Eley-Rideal mechanism when the chemical
reaction at the catalyst surface controls the reaction
rate [10]. According to this mechanism, the adsorbed
molecules of methanol react with TAG molecules
M.R. MILADINOVIĆ et al.: FURTHER STUDY ON KINETIC MODELING…
located near the catalyst surface. In the initial reaction
period the overall methanolysis process rate is limited
by the TAG mass transfer to active sites on the catalyst surface. In the later period of reaction, the reaction rate at the catalyst surface becomes slower than
the TAG mass transfer rate toward the catalyst surface as a consequence of FAME formation and better
oil dispersion, so TAG consumption at catalyst surface begins to control the overall rate of reaction.
Figure 1. The comparison of TAG and FAME concentrations
calculated by Eq (2) with the experimental data (TAG: solid
symbols and FAME: open symbols) at the methanol-to-oil mole
ratio of 6:1 (catalyst amount, %: 1 – ○, 2.5 – ∆, 5 – □ and 10 - ◊)
(a) neat CaO, (b) quicklime and (c) Ca(OH)2 (TAG: straight line
and FAME: dashed line).
Chem. Ind. Chem. Eng. Q. 22 (2) 137−144 (2016)
Influence of reaction conditions on kinetic parameters
Values of the parameters of Eq. (2), β, α, k and
(kmt,A)0, for the methanolysis reaction over CaO⋅ZnO
were taken from the works of Lukić et al. [10], while
values for the catalytic reaction systems using neat
CaO, quicklime and Ca(OH)2 were calculated by fitting the experimental data (20 data sets).
In general, the reaction rate constant, k, seems
to depend on type and amount of catalyst, methanolto-oil molar ratio and reaction temperature, as it can
be concluded from Table 3. The catalyst amount does
not affect the reaction rate constant in the case of
neat CaO and quicklime, as well as CaO⋅ZnO [11].
However, in the case of Ca(OH)2 it increases exponentially with increasing the catalyst amount, which is
attributed to the enhanced possibility of adsorbed
molecules for the reaction at higher catalyst amount
[4]. For three CaO-based catalysts the reaction rate
constant k reduces with decreasing their basicity in
the following order: neat CaO > quicklime > CaO.ZnO.
It seems that the methanol-to-oil molar ratio does not
or very slightly affect the reaction rate constant for
quicklime (Table 3) and CaO⋅ZnO [11]. As expected,
the reaction rate constant increases with increasing
the reaction temperature in the CaO⋅ZnO reaction
system [10]. The k-value depends on temperature in
accordance to the Arrhenius equation, as shown by
Sánchez et al. [15] for the jojoba oil methanolysis
over mussel shell CaO with the activation energy of
55.09 J/mol. However, the k-value for this process is
much lower than those for the sunflower oil methanolysis, as it can be seen in Table 3, because of the
difference in the chemical composition of the two oils.
While sunflower oil consists of TAGs predominantly,
jojoba oil has no TAGs in their structure but long
straight ester chains [15], which slows down the
methanolysis reaction.
β-Values for the CaO⋅ZnO reaction system
range between 1.37 and 2.90 [10], and no obvious
dependence on temperature is observed. However,
the constant β-value of 1.00±0.01 is determined for
other three reaction systems using neat CaO, quicklime and Ca(OH)2 at a constant temperature (60 °C).
Sánchez et al. [15] reported much lower β-value
(0.23) for the jojoba oil methanolysis over mussel
shell CaO that was not dependent on reaction temperature and methanol-to-oil molar ration in the
ranges of 45 to 65 °C and 6:1 to 12:1, respectively.
The fitting parameter α appears to depend on
the type and amount of catalyst. For the same other
conditions, its value increases in the following order:
CaO⋅ZnO, then Ca(OH)2 and finally neat CaO and
quicklime that have the same α-value. The α-value of
141
M.R. MILADINOVIĆ et al.: FURTHER STUDY ON KINETIC MODELING…
Chem. Ind. Chem. Eng. Q. 22 (2) 137−144 (2016)
Figure 2. The influence of methanol-to-oil mole ratio on the FAME concentration calculated by Eq. (2) (straight line) at various quicklime
amounts: a) 1, b) 2.5, c) 5 and d) 10% (methanol-to-oil mole ratio: 6:1 - ○; 12:1 - ∆ and 18:1 - □; reaction temperature: 60 °C).
59.7 for mussel shell CaO in the range of catalyst
amount of 6 to 10% [15] is close to those for neat
CaO and quicklime in the range of catalyst amount of
5 to 10%. For CaO⋅ZnO [11], neat CaO, quicklime
and Ca(OH)2 the parameter α decreases with increasing the catalyst amount according to a power law
function. The value of this parameter depends on the
time when the critical TAG conversion is achieved
and the change of the limiting reaction regime occurs
[11]. At lower catalyst amounts a longer reaction time
is required for achieving the critical TAG conversion
at which the reaction rate is enhanced. For the catalyst amount of 1%, the α-value is 1300 for CaO⋅ZnO
[11], while only 200-213 for neat CaO, quicklime and
Ca(OH)2. This difference could be attributed to different values of β-parameter for CaO⋅ZnO (β = 2.01), on
one hand and neat CaO, quicklime and Ca(OH)2
(β = 1.00), on the other hand. Methanol-to-oil mole
ratio does not influence the α-value in the quicklime
reaction system. In the case of CaO⋅ZnO, the dependence of α on reaction temperature is not obvious.
142
The mass transfer coefficient, (kmt,A)0, is dependent on the type and amount of catalyst, methanol-to-oil mole ratio and reaction temperature. Figure 3
shows the dependence of (kmt,A)0 on the amount of
various catalysts. The same value of (kmt,A)0 is observed for the neat CaO and quicklime reaction systems, which is higher than that for the Ca(OH)2 reaction system. For all reaction systems, (kmt,A)0 increases with increasing the catalyst amount, reaching
a plateau approximately at the catalyst amount of 5%.
This means that the increase of the catalyst amount
above 5% had no significant effect on the overall process rate in the initial reaction period. For the
CaO⋅ZnO reaction system, (kmt,A)0 increases with increasing the catalyst amount in the range from 0.5 to
2.0% [11].
In the quicklime reaction system at a constant
catalyst amount, (kmt,A)0 decreases with the increase
of methanol-to-oil molar ratio from 6:1 to 12:1, but it
remains almost constant with further increase of
methanol-to-oil mole ratio (Table 3). This observation
M.R. MILADINOVIĆ et al.: FURTHER STUDY ON KINETIC MODELING…
means that the mass transfer rate is enhanced at
higher methanol amount due to higher oil-methanol
interfacial area because of the lower density of the
reaction mixture enabling more efficient mixing. As
expected, (kmt,A)0 increases with increasing the reaction temperature in the CaO⋅ZnO reaction system
probably because of a faster formation of emulsifying
agents stabilizing the emulsion [10]. However, Sánchez et al. [15] noticed no effect of catalyst amount
and reaction temperature on the (kmt,A)0 for mussel
shell CaO and its increase with increasing the methanol concentration in the range of methanol-to-jojoba
oil molar ratio from 6:1 to 12:1.
Chem. Ind. Chem. Eng. Q. 22 (2) 137−144 (2016)
Nomenclature
Concentration of TAG, mol/dm3
Initial concentration of TAG, mol/dm3
Initial concentration of TAG, mol/dm3
Concentration of FAME, mol/dm3
Reaction rate constant, Eq. (2), min-1
Overall TAG volumetric mass transfer coefficient at the beginning of the process, min-1
MRPD Mean relative percent deviation, %
n
Number of experimental runs
2
R
Coefficient of determination
t
Time, min
xA
Degree of TAG conversion
ya
Experimental values of the TAG conversion
degree, %
ym
Mean value of the TAG conversion degree, %
yp
Predicted values of the TAG conversion degree, %
cA
cA0
cB0
cR
k
(kmt,A)0
Greek symbols
α, β
Fitting parameters
REFERENCES
Figure 3. Dependence of (kmt,A)0 on catalyst amount:
neat CaO - ○, quicklime - ∆ and Ca(OH)2 - □ (methanol-to-oil
mole ratio: 6:1; reaction temperature: 60 °C).
CONCLUSIONS
The kinetics of sunflower oil methanolysis catalyzed by neat CaO, quicklime, Ca(OH)2 and CaO⋅ZnO
was tested using the recently reported kinetic model
involving the initial region controlled by the TAG mass
transfer followed by the chemically controlled region.
This model adopts the pseudo-first order reaction rate
law or both the mass transfer and the chemical reaction. The model demonstrates a good agreement with
the experimental data and can be recommended for
describing the time variations of TAG and FAME concentrations during methanolysis of sunflower, used
vegetable and jojoba oil. However, its generalization
capability could be verified entirely after the analysis
of experimental data obtained with other solid base
catalysts and vegetable oils as it has been done for
calcium-based catalysts in the present study.
Acknowledgments
This work has been funded by the Ministry of
Education, Science and Technological Development
of the Republic of Serbia (Project III 45001).
[1]
M.L. Granados, M.D.Z. Poves, D.M. Alonso, R. Mariscal,
F.C. Galisteo, R. Moreno-Tost, J. Santamaría, J.L.G.
Fierro, Appl. Catal. B Environ. 73 (2007) 317–326
[2]
M. Kouzu, T. Kasuno, M. Tajika, Y. Sugimoto, S. Yamanaka, J. Hidaka, Fuel 87 (2008) 2798–2806
[3]
V.B. Veljković, O.S. Stamenković, Z.B. Todorović, M.L.
Lazić, D.U. Skala, Fuel 88 (2009) 1554–1562
[4]
O.S. Stamenković, V.B. Veljković, Z.B. Todorović, M.L.
Lazić, I.B. Banković-Ilić, D.U. Skala, Bioresour. Technol.
101 (2010) 4423–4430
[5]
V.G. Deshmane, Y.G. Adewuyi, Fuel 107 (2013) 474–482
[6]
M. Verziu, S.M. Coman, R. Richards, V.I. Parvulescu,
Catal. Today 167 (2011) 64–70
[7]
D.M. Alonso, R. Mariscal, M.L. Granados, P. MairelesTorres, Catal. Today 143 (2009) 167–171
[8]
C.S. MacLeod, A.P. Harvey, A.F. Lee, K. Wilson, Chem.
Eng. J. 135 (2008) 63–70
[9]
M.C.G. Albuquerque, I. Jiménez-Urbistondo, J. Santamaría-González, J.M. Mérida-Robles, R. Moreno-Tost, E.
Rodríguez-Castellón, et al., Appl. Catal., A: Gen. 334
(2008) 35–43
[10]
I. Lukić, Ž. Kesić, S. Maksimović, M. Zdujić, H. Liu, J.
Krstić, D. Skala, Fuel 113 (2013) 367–378
[11]
I. Lukic, Z. Kesic, S. Maksimovic, M. Zdujic, J. Krstic, D.
Skala, Chem. Ind. Chem. Eng. Q. 20 (2014) 425–439
[12]
M. Kouzu, J. Hidaka, Y. Komichi, H. Nakano, M. Yamamoto, Fuel 88 (2009) 1983–1990
[13]
M.R. Miladinović, J.B. Krstić, M.B. Tasić, O.S. Stamenković, V.B. Veljković, Chem. Eng. Res. Des. 92 (2014)
1740–1752
143
M.R. MILADINOVIĆ et al.: FURTHER STUDY ON KINETIC MODELING…
Chem. Ind. Chem. Eng. Q. 22 (2) 137−144 (2016)
[14]
P.L. Boey, G.P. Maniam, S.A. Hamid, Bioresour. Technol.
100 (2009) 6362–6368
[20]
A. Bazargan, M.D. Kostić, O.S. Stamenković, V.B. Veljković, G. McKay, Fuel 150 (2015) 519–525
[15]
M. Sánchez, J.M. Marchetti, N. El Boulifi, J. Aracil, M.
Martínez, Chem. Eng. J. 262 (2015) 640–647
[21]
[16]
P. Patil, V.G. Gude, S. Pinappu, S. Deng, Chem. Eng. J.
168 (2011) 1296–1300
Z.B. Todorovic, O.S. Stamenkovic, I.S. Stamenkovic, J.M.
Avramovic, A. V. Velickovic, I.B. Bankovic-Ilic, et al., Fuel
107 (2013) 493–502
[22]
[17]
A.K. Singh, S.D. Fernando, Chem. Eng. Technol. 30
(2007) 1716–1720
[18]
M.B. Tasić, M.R. Miladinović, O.S. Stamenković, V.B.
Veljković, D.U. Skala, Chem. Eng. Technol. 38(9) (2015)
1550-1556
M.D. Normand, M. Peleg, Wolfram Demonstr. Proj.
(2012). http://demonstrations.wolfram.com/SurvivalCurvesOfBacilliSporesWithAnActivationShoulder/ (accessed
September 5, 2014)
[23]
L.S. Hsieh, U. Kumar, J.C.S. Wu, Chem. Eng. J. 158
(2010) 250–256.
[19]
O.S. Stamenković, pHD Thesis, Faculty of Technology,
University of Niš, 2008 (in Serbian)
MARIJA R. MILADINOVIĆ1
MARIJA B. TASIĆ1
OLIVERA S. STAMENKOVIĆ1
VLADA B. VELJKOVIĆ1
DEJAN U. SKALA2
1
Univerzitet u Nišu, Tehnološki fakultet,
Leskovac, Srbija
2
Univerzitet u Beogradu, Tehnološko-metalurški fakultet, Beograd, Srbija
NAUČNI RAD
DALJE PROUČAVANJE KINETIČKOG
MODELOVANJA METANOLIZE SUNCOKRETOVOG
ULJA KATALIZOVANE NEKIM KALCIJUMOVIM
JEDINJENJIMA
Kinetički model, koji je prvobitno razvijen za metanolizu suncokretovog ulja katalizovanog
CaO⋅ZnO, primenjen je na nekoliko drugih katalizatora na bazi kalcijuma, kao što su čist
CaO, negašeni kreč i Ca(OH)2. Ovaj model, koji uključuje režime kontrolisane prenosom
mase triacilglicerola i hemijskom reakcijom, pokazao je dobro slaganje sa eksperimentalnim podacima, što je potvrđeno visokim koeficijentom determinacije (0,971±0,022) i
prihvatljivim srednjim relativnim procentim odstupanjem (±15,9%). Stoga se ovaj model
preporučuje za modelovanje kinetike metanolize suncokretovog ulja u prisustvu katalizatora na bazi kalcijuma u širokom opsegu reakcionih uslova.
Ključne reči: biodizel, katalizatori na bazi kalcijuma, kinetika, metanoliza, modelovanje
144
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 145−153 (2016)
DEJAN MARKOVIC1
IVANA KARADZIC2
VUKOMAN JOKANOVIC3
ANA VUKOVIC1
VESNA VUCIC2
1
Department of Paediatric and
Preventive Dentistry, Faculty of
Dentistry, University of Belgrade,
Belgrade, Serbia
2
Centre of Research Excellence in
Nutrition and Metabolism, Institute
for Medical Research, University of
Belgrade, Belgrade, Serbia
3
Laboratory of Radiation Chemistry
and Physics, Institute of Nuclear
Sciences “Vinca”, University of
Belgrade, Belgrade, Serbia
SCIENTIFIC PAPER
UDC 602.9:616:66.017/.018
DOI 10.2298/CICEQ141231028M
CI&CEQ
BIOLOGICAL ASPECTS OF APPLICATION OF
NANOMATERIALS IN TISSUE ENGINEERING*
Article Highlights
• Stem cells and scaffolds - an essential role in the production of new tissue by tissue
engineering
• Nanotechnology - a field of high importance and rapid development
• Functional necessities of scaffolds – biocompatibility, biodegradability and mechanical
properties
• The main challenge: transforming tissue engineering into regenerative engineering
Abstract
Millions of patients worldwide need surgery to repair or replace tissue that has
been damaged through trauma or disease. To solve the problem of lost tissue,
a major emphasis of tissue engineering (TE) is on tissue regeneration. Stem
cells and highly porous biomaterials used as cell carriers (scaffolds) have an
essential role in the production of new tissue by TE. The cellular component is
important for the generation and establishment of the extracellular matrix,
while a scaffold is necessary to determine the shape of the newly formed tissue and facilitate migration of cells into the desired location, as well as their
growth and differentiation. This review describes the types, characteristics and
classification of stem cells. Furthermore, it includes functional features of cell
carriers – biocompatibility, biodegradability and mechanical properties of biomaterials used in developing state-of-the-art scaffolds for TE applications, as
well as suitability for different tissues. Moreover, it explains the importance of
nanotechnology and defines the challenges and the purpose of future research
in this rapidly advancing field.
Keywords: tissue engineering, nanomaterials, scaffolds, stem cells,
tissue regeneration.
Tissue engineering
Millions of patients worldwide need surgical
procedures to repair or replace tissue that has been
damaged through trauma or disease [1]. Today, conventional therapy addresses the problem of lost tissue
by appropriate tissue replacement – tissue graft. The
majority of defects can be healed using standard conservative or surgical methods. However, large defects
occurring after tumor surgery, cysts or multiple frac-
Correspondence: I. Karadzic, Centre of Research Excellence in
Nutrition and Metabolism, Institute for Medical Research, University of Belgrade, Tadeusa Koscuska 1, P.O. Box 102, 11129
Belgrade, Serbia.
E-mail: ivana.colak@gmail.com
Paper received: 31 December, 2014
Paper revised: 11 May, 2015
Paper accepted: 21 July, 2015
* This paper was part of a plenary lecture at the Rosov Pin
Conference 2014.
tures require a more complex procedure of tissue reparation [2,3].
With respect to lost tissue treatment, the main
emphasis of tissue engineering is on tissue regeneration (TE) rather than tissue replacement [4–7]. Thus,
stem cells and highly porous biomaterials used as
scaffolds have an essential role in the production of
new tissue by TE (Figure 1). The cellular component
is important for the generation and establishment of
extracellular matrix (ECM) in the new tissue, while a
scaffold is necessary for providing mechanical stability and foundation for a new three-dimensional tissue
structure [6,8,9].
Since it has been demonstrated that biological
systems correspond better to nano- than micro-dimensional biomaterial structures, nanotechnology has
become a field of high importance and rapid development [10–12]. Also, composition, size, morphology
and geometry of nanostructured materials can be
145
D. MARKOVIC et al.: BIOLOGICAL ASPECTS OF APPLICATION…
Chem. Ind. Chem. Eng. Q. 22 (2) 145−153 (2016)
Figure 1. Basic concept of tissue engineering.
controlled. Further, the surface of these materials can
be modified in order to enhance biocompatibility,
immune compatibility and/or cell adhesion [13].
This review describes the functional necessities
and types of stem cells and biomaterials used in developing state-of-the-art scaffolds for tissue engineering applications. Furthermore, it defines the challenges and the purpose of future research in this fast
advancing field.
Stem cells and tissue engineering
Stem cells are unspecialized cells in the early
stage of the development, which under normal conditions have the ability to differentiate into specialized
mature cells and to divide in order to produce more
stem cells [14–17]. Two functions define stem cells:
unlimited self-renewal capacity, which makes them
potentially immortal, and pluripotency [18]. Stem cells
can be divided in several different ways, as shown in
Table 1 [15,19–22].
Although they have less ability to differentiate,
adult stem cells are far more applicable in regenerative medicine than embryonic stem cells, primarily
because of being relatively easy to isolate, lack of
oncogenic potential and no ethical constraints over
their application [23–26].
Architecture and nanotechnology of scaffolds
The ECM represents a biological 3D carrier for
the cells and provides appropriate environment and
architecture specific for each tissue [27]. Therefore,
the key to a successful TE is proper design of cell
carriers – scaffolds, which mimic the native ECM,
combined with adequate stem cells. The role of these
carriers is to determine the shape of the newly formed
tissue and facilitate the migration of cells into the desired location, their growth and differentiation [28,29].
The key characteristic of every scaffold is that it
must be biocompatible – to provide physical and
mechanical functions and provoke a preferred response without causing any undesirable reactions in
the host. Hence, the choice of material is a crucial
point in tissue engineering [28,30]. It is desirable for
the scaffold to disintegrate during the formation of
new tissue, to allow the body’s own cells, over time, to
eventually replace the implanted material [31]. Therefore, the biodegradability of the scaffold is also considered very important in these processes.
Developing scaffolds with adequate mechanical
properties is one of the greatest challenges in
attempting to engineer bone or cartilage [32,33]. A
balance must be achieved between the mechanical
Table 1. Stem cells classification according to different criteria
Criterion
Cell potency
Function
Sources
146
Stem cells type
Properties
Totipotent stem cells
Potential to differentiate in any human cell even whole organism
Pluripotent stem cells
Potential to differentiate in various tissue types but not whole organism
Multipotent stem cells
Potential to differentiate in various cell types within tissue – progenitor cell
Unipotent stem cells
Potential to differentiate in one cell type – precursor cell
Normal stem cells
Not involved in pathologic process
Cancer stem cells
Associated with most cancer disease
Embryonic stem cells
Derived from inner cell mass of blastocyst
Adult stem cells
Derived from endoderm, mesoderm or ectoderm
D. MARKOVIC et al.: BIOLOGICAL ASPECTS OF APPLICATION…
properties and a sufficiently porous architecture in
order to obtain the desired scaffold [34]. Adequate
porosity allows cell migration and provides a suitable
microenvironment for cell proliferation and differentiation, with adequate vascularization, flow of nutrients and oxygen and elimination of degradation
products [35–37]. The porosity should be in the optimal range: small enough to ensure mechanical integrity and sufficiently large to provide optimal bioactivity.
For this reason, the size of the pores should be less
than 300 nm [29].
Two types of materials are currently used in TE:
natural and synthetic [38]. The advantage of natural
materials is biological recognition regarding cell adhesion and function. However, the downsides are
uncontrolled mechanical properties and biodegradability, possible host immune reaction and the cost
[39,40]. Various natural materials have been evaluated to date. Derivatives of ECM have been investigated for supporting cell growth. Proteins collagen
and fibrin, as well as polysaccharides glycosaminoglycans, have all proved appropriate regarding cell
compatibility, but some potential immunogenic issues
still remain [35,41,42]. Hyaluronic acid, one of the
most exploited glycosaminoglycans, in combination
with glutaraldehyde or water soluble carbodiimide, is
considered suitable for scaffold materials [43]. Polysaccharide chitin and chitosan based nanofibers have
remarkable potential to be used as tissue engineering
scaffolds, as well as drug delivery systems, wound
dressing materials, antimicrobial agents and biosensors, due to their biocompatibility, biodegradability,
antibacterial activity, low immunogenicity, wound
healing capacity and cell binding capability [44].
Further, some scaffolds are tested for use in the delivery of small molecules (drugs) to specific tissues
[45]. Finally, decellularized tissue extracts in which
Chem. Ind. Chem. Eng. Q. 22 (2) 145−153 (2016)
the remaining cellular residues or ECM act as a scaffold, are another form of cell carriers undergoing investigation [46].
Synthetic materials, on the other hand, have the
advantage of a commercial production, together with
a control over mechanical properties, microstructure
and degradation rate.
A commonly used synthetic material is polylactic
acid (PLA). This is a polyester which degrades within
the human body to form lactic acid, a chemical compound that plays a role in various biochemical processes and is easily removed from the body. The
nanofibrous PLA mats incorporating carbon nanotubes and rectorite, fabricated using an electrospinning technique, have proven suitable for biomedical
applications due to their increased thermal stability
and low cytotoxicity [47]. Similar to PLA are polyglycolic acid (PGA), polylactic-co-glycolic acid (PLGA)
and polycaprolactone (PCL), with degradation mechanisms similar to that of PLA but a different rate of
degradation compared to PLA [48,49]. Other natural
and synthetic materials have also been used in scaffold synthesis (Figure 2).
Due to their unique chemical, physical and biological functions, nano-sized particles/fillers, of both
inorganic and organic origin, have been studied in
detail. They differ in their structure, composition,
design and application, and can be in the form of
nanofibers, nanogels, etc, as presented in Figure 2.
Nanocomposite hydrogels combine the advantages of
nano-fillers and hydrogel matrices and thus may
result in improved mechanical and biological properties and find their potential application in biomedicine
as drug delivery matrices and scaffolds [50]. Recently, injectable scaffolds have received attention due
to their potential for avoiding the invasive surgery
normally required for tissue implantation. Natural
Figure 2. Synthesis, structure and design of different forms of nanomaterials used as scaffolds in tissue engineering.
147
D. MARKOVIC et al.: BIOLOGICAL ASPECTS OF APPLICATION…
polymers chitosan and alginate are used as coating
materials to make positively and negatively charged
PLGA nanoparticles, respectively. All the results
demonstrate the potential use of the biodegradable
colloidal gels as injectable scaffolds in tissue engineering and drug release [51].
Application of scaffolds in tissue engineering
Bone disease or bone defects such as osteosarcoma, osteoporosis, and bone fractures affect millions
of individuals worldwide [1]. In order to solve the problem of lost bone tissue, bone tissue engineering
emphasizes tissue regeneration rather than tissue
replacement and is becoming a subject of growing
interest. For over two decades, bioceramic material –
hydroxyapatite (HAP) has been used as a substitute
for bone as it has physical properties similar to the
inorganic component of natural bone [52]. It is suitable material for hard tissue replacement due to its
osteoconductivity, biocompatibility and slow resorption. On the other hand, the porosity of hidroxyapatite
translates into poor mechanical properties [53]. In the
attempt to compensate for these disadvantages, various polymers have been examined, however, none
of these meet all of the requirements for the ideal cell
carrier in bone TE [54]. This has led to the development of composite carriers consisting of both an
inorganic and organic component where the inorganic
particles are embedded into the surface of a polymer
matrix [36,55]. Inorganic-organic composites are
designed to mimic natural bone by combining the
viscoelastic properties of polymers with the strength
of the inorganic part of the composite, to create bioactive materials with improved mechanical properties
and ability to degrade over time [56]. Also, the basic
products of decomposition of hydroxyapatite and tricalcium-phosphate neutralize the acidity of the polymeric compounds. Regarding higher bioactivity,
inorganic nanostructured components have received
more interest than equivalent microstructured ones
[57]. Nanocomposites based on hydroxyapatite-collagen are being particularly rapidly developed and
showing promising results [58,59]. Recently, negatively charged inorganic hydroxyapatite nanoparticles
(NPs) and positively charged organic PLGA NPs were
assembled to form a cohesive colloidal gel which
proved to be suitable as an injectable filling for the
purpose of bone tissue regeneration [60]. Further, in
vivo tests revealed that a similar colloidal gel, created
by mixing PLGA nanoparticles of opposite charge,
capable of controlled release, has shown good results
as a filler for repair of cranial bone defects [61]. Use
of a composite scaffold with high porosity (low mech-
148
Chem. Ind. Chem. Eng. Q. 22 (2) 145−153 (2016)
anical properties) and fast degradation kinetics has
led to the production of grafts that can be used in low
load sites. In the middle of the last decade, a unique
composite carrier consising of a combination of biodegradable PLGA and bioresorbable calcium phosphate cement was created. This carrier is characterized by high porosity (81–91%), with macropores of
0.8–1.8 mm, and improved mechanical properties due
to the polymer [62]. Another 3D scaffold suitable for
bone regeneration, with porous design and mechanical properties similar to the trabecular bone is
obtained by combining calcium phosphate ceramics
(low crystalline CaP) fused in biodegradable PLGA
microspheres [63].
Scaffolds are commonly applied in bone and
cartilage tissue engineering, although notable results
have been achieved in many other tissues, including
skin, nerves, heart etc. However, different tissues
have their own peculiarities. For instance, skin tissue
engineering is complex for various reasons: greatly
limited donor sites in patients with skin losses over
50–60%, transmission of infection (mostly human
immunodeficiency virus or hepatitis related to allogeneic skin grafting), pain, and scarring at donor sites
[64]. Several scaffolds have been investigated for
applications in skin TE. In a rat model, regeneration of
tissue with similar properties to the native dermis and
significantly enhanced formation of blood capillaries
has been achieved after a novel composite film of
salmon DNA and collagen was implanted in a fullthickness skin wound in the dorsal region [65]. Further, alginate is also considered a promising agent in
skin tissue engineering due to its ability to maintain a
physiological, moist, microenvironment, reducing possibility for bacterial infection, and facilitating wound
healing especially in deep-thickness wounds [66].
Polysaccharide chitin has been combined with other
marine-derived composites and developed into a hydrogel wound dressing, providing good moist healing
environment and encouraging capillary formation in a
full-thickness skin wound in rat [67]. Composite nanofibrous cellulose acetate 3D mat coated with positively charged lysozyme and negatively charged
layered silicate – rectorite, obtained by electrospraying, has shown promising results in pharmaceutical
uses and antimicrobial wound dressing [68]. In addition, a nanofibrous scaffold made of PLGA-chitosan/polyvinyl alcohol, fabricated by electrospinning, is
proving to be useful in skin tissue engineering [42].
Recently, novel nanofibrous mats have been developed, coated layer-by-layer with silk fibroin and lysozyme on a cellulose electrospun template via electrostatic interaction. These mats are promising tools in
D. MARKOVIC et al.: BIOLOGICAL ASPECTS OF APPLICATION…
dermal reconstruction due to their nontoxic, biodegradable, biocompatible, antibacterial and wound healing properties [69].
In contrast, in cardiac tissue engineering the aim
is to produce tissue constructs that are thick and compact, contain physiological densities of metabolically
active cardiac cells, and contract synchronously in
response to electrical stimulation with sufficient force
[70]. The most difficult requirements are probably
related to the establishment of blood flow and to the
integration and electromechanical coupling with the
host tissue. Shinoka [71] presented the first results of
tissue engineered heart valves implanted into the
juvenile sheep model. The scaffolds used for these
heart valves were created from biodegradable polymers which were seeded in vitro with autologous
valve cells. Recently, additional studies were performed to examine tissue activity at the mitral site of
the experimental model, allowing to assess the highest tissue stress that can be achieved by tissue engineered heart valves [72]. The next challenge is to
insert these heart valves with minimally invasive technique without alteration of the tissue during the implantation. Finally, the field of the whole organ engineering has been expanding, in which the bio-artificial heart [73] could overcome the problem of organ
deficiency for patients suffering from end-stage heart
failure.
In the case of cartilage produced for the auricle
and nose or for complex facial trauma, the emphasis
is on maintaining the shape and accurate reproduction of the intended geometry [74]. Also, it is very
important to imitate the mechanical characteristics
found in the native cartilage and to control the mechanical stimulus for chondrogenesis and ECM production. Chen et al. [74] have shown that collagen in
its natural form is a better surface modification material than gelatin for promoting cell adhesion, proliferation and secretion of ECM components. Transplantation of peptide hydrogels made of nanofiber scaffolds containing chondrocytes and growth factors, into
cartilage defects in a bovine model, resulted in extensive synthesis of glycosaminoglycans and type-II collagen similar to the native cartilage [75]. In vitro experiments revealed that peptide scaffolds comprising
growth factors are capable of inducing chondrogenic
differentiation of human mesenchymal stem cells
(MSCs). These scaffolds can promote substantial regeneration of the articular cartilage in full thickness
chondral microfractured defects in the trochlea of
adult rabbits in the presence of bone marrow MSCs
[76].
Chem. Ind. Chem. Eng. Q. 22 (2) 145−153 (2016)
In nerve tissue engineering, due to the complex
system, involving neural cells, the microenvironment
with a variety of cell receptors, the ECM and specific
chemo–physical properties, electrospun guidance
channels and hydrogels are considered to be the
most promising types of scaffolds [77]. The soft nature of the nervous tissue potentially makes hydrogels
the ideal material, considering also their biodegradability, flexibility and low inflammatory potential. Adequate matching between the mechanical properties of
different materials and specific neural environments is
crucial in achieving the correct morphology, neural
growth and differentiation. It has been demonstrated
that neurite extension in dorsal root ganglia cells
conversely correlates with the mechanical stiffness of
agarose gels [78]. Many natural and synthetic polymers have been investigated for use as neural scaffold materials. For example, a biodegradable glass
material was used to repair the facial or median nerve
in a sheep model [79] and carbon nanostructures,
including nanotubes, nanofibers and graphene have
been incorporated in some experimental neural prostheses and guides [80].
Various types of tissues and their stem cells that
are commonly used and described for application in
tissue engineering are shown in Figure 3.
Figure 3. Various types of adult stem cells used in tissue
engineering: (a) skin, (b) ligaments and cartilage, (c) brain,
(d) muscle, (e) bone and (f) cardiovascular.
Examples of nanostructured scaffolds applied in bone
TE
One of the most frequently used polymers for
composite scaffolds today is PLGA, primarily due to
149
D. MARKOVIC et al.: BIOLOGICAL ASPECTS OF APPLICATION…
its proven biocompatibility and a variable degradation
rate that can be regulated by modifying the proportions of its constituent polymers, PLA and PGA
[2,5,81,82]. Moreover, the reversibility of a colloidal
gel composed of oppositely charged PLGA nanoparticles makes it excellent material for molding, extrusion or injection of tissue engineering scaffolds[83].
There are, also, newly developed materials based on
PLGA/HAP composites that are interesting for TE
because of their high biocompatibility and ability to
mimic natural bone. These materials have become a
promising tool in load-bearing bone TE, and might
provide optimal cell differentiation and mineralization
of the bone tissue. Cells seeded on such materials
easily adhere, especially on hydroxyapatite surface,
which indicates good cell proliferation and integration
of the bone implants [84].
In order to obtain better characteristics of scaffolds, new materials and new scaffold producing techniques are currently in development. Appropriate morphology of scaffold walls can be attained by using
nanodesign hydroxyapatite particles inside a biomimetic medium, where they self-assemble on a polymer/ceramic scaffold structure [38]. It is an improvement on design of scaffold obtained using polymeric
foam template based in polyurethane. This biomimetic method has shown that it stimulates the growth of
“bone-like” structures on scaffold surfaces [85]. In this
way, nanodesigned biomimetic apatite is very similar
to biological apatite and very suitable for cell growth
and proliferation [8].
A composite scaffold PLGA/HAP registered
under the name ALBO-OSS has shown good cell
adhesion of cells grown in control medium. Cells were
spherical with clearly visible pseudopodia or cytoplasmic extensions (Figure 4a). After 7 days of culture in
osteogenic medium, polygonal cells with very elongated cytoplasmic extensions were detected. The org-
Chem. Ind. Chem. Eng. Q. 22 (2) 145−153 (2016)
anic fibrous-like structures have also been observed
(Figure 4b). These structures and cell morphology are
typical for stem cells undergoing osteogenic differentiation and indicate the beginning of the ECM formation. After 21 days, ECM dominated in the SEM
micrographs covering the scaffold pores (Figure 4c).
This highly developed ECM network demonstrated
extensive differentiation and good biocompatibility
between cells and materials, which is essential for
use in tissue engineering [36].
Comparing the nano- and microstructured scaffolds, it was demonstrated by a difference in the optical density obtained by MTT testing and ALP activity
that the larger surface area of the nanostructured
scaffold allows better adhesion and provides more
space for the differentiation of mesenchymal cells
than in microstructured scaffolds [86].
Comparing the adhesion and quantity of human
osteoblasts cultured on Bio Oss and synthetic bone
substitute – Nano Bone for 7 days by SEM analysis, a
significantly higher number of cells with cytoplasmic
extensions was observed in the presence of Nano
Bone [87].
SEM analysis showed similar results after studying cell morphology and adhesion of mesenchymal
cells grown for 7 days on PLLA and on nano- and
micro-HAP/PLLA composite scaffolds. The cells were
spherical and their number was much higher in the
presence of nano HAP/PLLA composite scaffold, indicating a superior biocompatibility of that material [88].
Hence, there are many scaffolds that exhibit good
biocompatibility and could replace materials, such as
golden standard Bio OSS, currently used in bone
tissue engineering.
Challenges and future research
The main requirement for a successful application of the scaffold is a high control level of their
Figure 4. Scanning electron microscopy of dental pulp stem cells from deciduous teeth on Albo-Oss (porous hydroxyapatite + PLGA
composite) scaffold: a) Spherical cells with cytoplasmic extensions indicate good cell adhesion after 7 days in control medium; (b)
polygonal cells with elongated cytoplasmic extensions pseudopodia indicate very good cell adhesion; fibrous-like organic structures
point toward beginning of ECM production after 7 days in osteogenic medium; (c) ECM was dominant in the SEM micrographs,
covering the scaffold and the pores after 21 days in osteogenic medium.
150
D. MARKOVIC et al.: BIOLOGICAL ASPECTS OF APPLICATION…
micro- and macrostructural properties during production process. Mechanical properties of today’s
composite scaffolds still do not fully satisfy the properties of natural bone nor succeed in reaching their
anisotropy [89].
Though a wide-range of strategies have been
employed to produce the ideal scaffold that possesses the optimum dimensions, porosity, topography
and mechanical properties, the clinical success of
such constructs remains elusive [90].
The rapid development of techniques and fabrication tools that have arisen in the recent years
have signified new beginnings in the field of TE. However, there are still obstacles in achieving the success
of the in vitro experiments in an in vivo system. Moreover, the complexity of multiple tissues that form a
functional organ poses a real challenge to tissue engineers. Further developments are awaited in reaching
the goal of creating a completely functional organ,
thus truly transforming tissue engineering into regenerative engineering.
Acknowledgements
This paper was supported by the Projects 41030
and 172026 financed by the Ministry of Education,
Science and Technological Development of the Republic of Serbia.
Chem. Ind. Chem. Eng. Q. 22 (2) 145−153 (2016)
[11]
H. Hosseinkhani, M. Hosseinkhani, S. Hattori, R. Matsuoka, N. Kawaguchi, J. Biomed. Mater. Res., A 94 (2010)
1-8
[12]
H. Hosseinkhani, P.D. Hong, D.S. Yu, Y.R. Chen, D.
Ickowicz, I.Y. Farber, A.J. Domb, Int. J. Nanomed. 7
(2012) 3035-3043
[13]
H. Peng, X. Liu, R. Wang, F. Jia, L. Dong, Q. Wang, J.
Mater. Chem., B 2 (2014) 6435-6461
[14]
I. Singec, R. Jandial, A. Crain, G. Nikkhah, E. Y. Snyder,
Annu. Rev. Med. 58 (2007) 313-328
[15]
A. Vats, R.C. Bielby, N.S. Tolley, R. Nerem, J.M. Polak,
Lancet. 366 (2005) 592-602
[16]
E.A. de Wynter, A.J. Emmerson, N.G. Testa, Best Pract.
Res., Clin. Haematol. 12 (1999) 1-17
[17]
R.J. Deans, A.B. Moseley, Exp. Hematol. 28 (2000) 875–884
[18]
A. Saunders, F. Faiola, J. Wang, Stem. Cells. 31 (2013)
1227-1236
[19]
I. Wilmut, A. E. Schnieke, J. McWhir, A.J. Kind, K.H.
Campbell, Nature 385 (1997) 810-813
[20]
H.S. Dua, A. Azuara-Blanco, Surv. Ophthalmol. 44 (2000)
415-425
[21]
I. Malanchi, H. Peinado, D. Kassen, T. Hussenet, D.
Metzger, P. Chambon, M. Huber, D. Hohl, A. Cano, W.
Birchmeier, J. Huelsken, Nature 452 (2008) 650-653
[22]
D. Brasanac, I. Boricic, V. Todorovic, N. Tomanovic, S.
Radojevic, Br. J. Dermatol. 153 (2005) 1166-1175
[23]
E. Fuchs, J. A. Segre, Cell 100 (2000) 143-155
[24]
D. Markovic, A. Milenkovic, G. Koliakos, E. Kostidou, I.
Karadzic, J. Debeljak Martacic, V. Jokanovic, T. Peric, B.
Petrovic, N. Arsenijevic, T. Kanjevac, Balk. J. Stom. 14
(2010) 4-7
[25]
B.C. Perry, D. Zhou, X. Wu, F.C. Yang, M.A. Byers, T.M.
Chu, J.J. Hockema, E.J. Woods, W.S. Goebel, Tissue
Eng., C: Methods 14 (2008) 149-156
[26]
A. Graziano, R. d'Aquino, G. Laino, G. Papaccio, Stem.
Cell. Rev. 4 (2008) 21-26
[27]
S. Huang, D.E. Ingber, Nat. Cell. Biol. 1 (1999) E131-138
[28]
C. Wang, G. Meng, L. Zhang, Z. Xiong, J. Liu, J. Biomed.
Biotechnol. 2012 (2012) 579141
REFERENCES
[1]
M. Miura, S. Gronthos, M. Zhao, B. Lu, L.W. Fisher, P.G.
Robey, S. Shi, Proc. Natl. Acad. Sci. U.S.A. 100 (2003)
5807-5812
[2]
S.S. Kim, M. Sun Park, O. Jeon, C. Yong Choi, B.S. Kim,
Biomaterials 27 (2006) 1399-1409
[3]
S.C. Rizzi, D.J. Heath, A.G. Coombes, N. Bock, M.
Textor, S. Downes, J. Biomed. Mater. Res. 55 (2001)
475-486
[4]
E.M. Aarden, E.H. Burger, P.J. Nijweide, J. Cell.
Biochem. 55 (1994) 287-299
[29]
[5]
X. Miao, D.M. Tan, J. Li, Y. Xiao, R. Crawford, Acta
Biomater. 4 (2008) 638-645
D.W. Hutmacher, J.T. Schantz, C.X. Lam, K.C. Tan, T.C.
Lim, J. Tissue. Eng. Regen. Med. 1 (2007) 245-260
[30]
[6]
H. Sittisak, P. Vinai, Asian Biomed. 1 (2007) 229-238
K.S. Chan, W. Liang, W.L. Francis, D.P. Nicolella, J.
Mech. Behav. Biomed. Mater. 3 (2010) 584-593
[7]
L. Meinel, V. Karageorgiou, R. Fajardo, B. Snyder, V.
Shinde-Patil, L. Zichner, D. Kaplan, R. Langer, G. Vunjak-Novakovic, Ann. Biomed. Eng. 32 (2004) 112-122
[31]
B.N. Brown, C.A. Barnes, R.T. Kasick, R. Michel, T.W.
Gilbert, D. Beer-Stolz, D.G. Castner, B.D. Ratner, S.F.
Badylak, Biomaterials 31 (2010) 428-437
[8]
I. Karadzic, V. Vucic, V. Jokanovic, J. Debeljak-Martacic,
D. Markovic, S. Petrovic, M. Glibetic, J. Biomed. Mater.
Res., A 103 (2015) 350-357
[32]
M. Kamitakahara, C. Ohtsuki, T. Miyazaki, J. Biomater.
Appl. 23 (2008) 197-212
[33]
F.J. O'Brien, Mater. Today. 14 (2011) 88-95.
[9]
G.F. Muschler, C. Nakamoto, L.G. Griffith, J. Bone. Joint.
Surg. Am. 86A (2004) 1541-1558
[34]
D.W. Hutmacher, Biomaterials 21 (2000) 2529-2543
[35]
K.E. McCracken, P.L. Tran, D.J. You, M.J. Slepian, J.Y.
Yoon, J. Biol. Eng. 7 (2013) 11
C.M. Murphy, M.G. Haugh, F.J. O'Brien, Biomaterials 31
(2010) 461-466
[36]
I. Karadzic, Effects of various 3D biomaterials on osteoblastic differentiation of dental pulp mesenchymal cells,
PhD Thesis, Faculty of Stomatology, Belgrade, 2013
[10]
151
D. MARKOVIC et al.: BIOLOGICAL ASPECTS OF APPLICATION…
Chem. Ind. Chem. Eng. Q. 22 (2) 145−153 (2016)
[37]
E.A. Phelps, A.J. Garcia, Regen. Med. 4 (2009) 65-80
[64]
[38]
M. Petrovic, B. Colovic, V. Jokanovic, D. Markovic, J.
Ceram. Process. Res. 13 (2012) 398-404
M. Loss, V. Wedler, W. Kunzi, C. Meuli-Simmen, V.E.
Meyer, Burns 26 (2000) 644-652
[65]
[39]
I. Armentano, M. Dottori, E. Fortunati, S. Mattioli, J.M.
Kenny, Polym. Degrad. Stab. 95 (2010) 2126-2146
X. Shen, N. Nagai, M. Murata, D. Nishimura, M. Sugi, M.
Munekata, J. Mater. Sci. Mater. Med. 19 (2008) 3473–3479
[40]
O.O. Ige, L.E. Umoru, S. Aribo, ISRN Mater. Sci. 2012
(2012) 20
[66]
W.R. Gombotz, S.F. Wee, Adv. Drug. Deliv. Rev. 64
Suppl. (2012) 194-205
[41]
E. Chung, J.A. Rytlewski, A.G. Merchant, K.S. Dhada,
E.W. Lewis, L.J. Suggs, Acta Biomater. 17 (2015) 78-88
[67]
[42]
B. Duan, L. Wu, X. Li, X. Yuan, Y. Zhang, K. Yao, J.
Biomater. Sci. Polym. Ed. 18 (2007) 95-115
K. Murakami, H. Aoki, S. Nakamura, M. Takikawa, M.
Hanzawa, S. Kishimoto, H. Hattori, Y. Tanaka, T. Kiyosawa, Y. Sato, M. Ishihara, Biomaterials 31 (2010) 83-90
[68]
[43]
J.Y. Lai, D.H. Ma, H.Y. Cheng, C.C. Sun, S.J. Huang,
Y.T. Li, G.H. Hsiue, J. Biomater. Sci. Polym. Ed. 21
(2010) 359-376
W. Li, X. Li, Q. Wang, Y. Pan, T. Wang, H. Wang, R.
Song, H. Deng, Carbohydr. Polym. 99 (2014) 218-225
[69]
S. Xin, X. Li, Q. Wang, R. Huang, X. Xu, Z. Lei, H. Deng,
J. Biomed. Nanotechnol. 10 (2014) 803-810
[44]
F. Ding, H. Deng, Y. Du, X. Shi, Q. Wang, Nanoscale 6
(2014) 9477-9493
[70]
[45]
K.T. Shalumon, J.P. Chen, Curr. Pharm. Des. 21 (2015)
1979-1990.
G. Vunjak-Novakovic, N. Tandon, A. Godier, R. Maidhof,
A. Marsano, T. P. Martens, M. Radisic, Tissue Eng., B 16
(2010) 169-187
[71]
[46]
A.B. Daly, J.M. Wallis, Z.D. Borg, R.W. Bonvillain, B.
Deng, B.A. Ballif, D.M. Jaworski, G.B. Allen, D.J. Weiss,
Tissue Eng., A 18 (2012) 1-16
T. Shinoka, C.K. Breuer, R.E. Tanel, G. Zund, T. Miura,
P.X. Ma, R. Langer, J.P. Vacanti, J.E. Mayer, Jr., Ann.
Thorac. Surg. 60 (1995) S513-516
[72]
J.L. Honge, J.A. Funder, H. Jensen, P.M. Dohmen, W.F.
Konertz, J.M. Hasenkam, J. Heart Valve Dis. 19 (2010)
584-592
[73]
F.E. Smit, P.M. Dohmen, Med. Sci. Monit. Basic Res. 20
(2014) 161-163
[74]
C.H. Chen, M.Y. Lee, V.B. Shyu, Y.C. Chen, C.T. Chen,
J.P. Chen, Mater. Sci. Eng., C: Mater. Biol. Appl. 40
(2014) 389-397
[75]
S.A. Maher, R.L. Mauck, L. Rackwitz, R.S. Tuan, J.
Tissue. Eng. Regen. Med. 4 (2010) 25-29
[76]
R.N. Shah, N.A. Shah, M.M. Del Rosario Lim, C. Hsieh,
G. Nuber, S. I. Stupp, Proc. Natl. Acad. Sci. U.S.A. 107
(2010) 3293-3298
[77]
G.A. Saracino, D. Cigognini, D. Silva, A. Caprini, F.
Gelain, Chem. Soc. Rev. 42 (2013) 225-262
[78]
A.P. Balgude, X. Yu, A. Szymanski, R.V. Bellamkonda,
Biomaterials 22 (2001) 1077-1084
[79]
N.E. Starritt, S.A. Kettle, M.A. Glasby, Laryngoscope 121
(2011) 1614-1619
[47]
Y. Lu, X. Li, X. Zhou, Q. Wang, X. Shi, Y. Du, H. Deng, L.
Jiang, RSC Adv. 4 (2014) 33355-33361
[48]
Y. Li, Z.G. Wu, X.K. Li, Z. Guo, S.H. Wu, Y.Q. Zhang, L.
Shi, S.H. Teoh, Y.C. Liu, Z.Y. Zhang, Biomaterials 35
(2014) 5647-5659
[49]
F. Qi, J. Wu, T. Yang, G. Ma, Z. Su, Acta Biomater. 10
(2014) 4247-4256
[50]
F. Song, X. Li, Q. Wang, L. Liao, C. Zhang, J. Biomed.
Nanotechnol. 11 (2015) 40-52
[51]
Q. Wang, S. Jamal, M.S. Detamore, C. Berkland, J.
Biomed. Mater. Res., A 96 (2011) 520-527
[52]
A.M. Ambrosio, J.S. Sahota, Y. Khan, C.T. Laurencin, J.
Biomed. Mater. Res. 58 (2001) 295-301
[53]
M. Wang, Biomaterials 24 (2003) 2133-2151
[54]
I. Hofmann, L. Müller, P. Greil, F.A. Müller, Surf. Coat.
Technol. 201 (2006) 2392-2398
[55]
J.B. Lee, S.I. Jeong, M.S. Bae, D.H. Yang, D.N. Heo,
C.H. Kim, E. Alsberg, I.K. Kwon, Tissue Eng., A 17
(2011) 2695-2702
[56]
T. Douglas, E. Pamula, D. Hauk, J. Wiltfang, S. Sivananthan, E. Sherry, P.H. Warnke, J. Mater. Sci. Mater. Med.
20 (2009) 1909-1915
[80]
F. Tavangarian, Y. Li, Ceram. Int. 38 (2012) 6075-6090
[81]
S.W. Kang, H.S. Yang, S.W. Seo, D.K. Han, B.S. Kim, J.
Biomed. Mater. Res., A 85 (2008) 747-756
[82]
M. Ngiam, S. Liao, A.J. Patil, Z. Cheng, C.K. Chan, S.
Ramakrishna, Bone 45 (2009) 4-16
[57]
S.S. Liao, F.Z. Cui, Y. Zhu, J. Bioact. Compat. Polym. 19
(2004) 117-130
[83]
[58]
M.M. Pereira, J.R. Jones, R.L. Orefice, L.L. Hench, J.
Mater. Sci. Mater. Med. 16 (2005) 1045-1050
Q. Wang, L. Wang, M. S. Detamore, C. Berkland, Adv.
Mater. 20 (2008) 236-239
[84]
V. Jokanovic, Nanomedicine – the Biggest Challenge of
the 21st Century, Data Status, Belgrade, 2012 (in Serbian)
[59]
T. Lu, Y. Li, T. Chen, Int. J. Nanomed. 8 (2013) 337-350
[60]
Q. Wang, Z. Gu, S. Jamal, M.S. Detamore, C. Berkland,
Tissue Eng., A 19 (2013) 2586-2593
[85]
[61]
Q. Wang, J. Wang, Q. Lu, M.S. Detamore, C. Berkland,
Biomaterials 31 (2010) 4980-4986
B. Colovic, V. Jokanovic, B. Markovic-Todorovic, Z. Markovic, J. Optoelectron Adv. Mater. 11 (2009) 70-75
[86]
[62]
D. Lickorish, L. Guan, J.E. Davies, Biomaterials 28 (2007)
1495-1502
M.B. Eslaminejad, F. Bagheri, M. Zandi, E. Nejati, E.
Zomorodian, Cell J. 12 (2011) 469-476
[87]
[63]
Y.M. Khan, D.S. Katti, C.T. Laurencin, J. Biomed. Mater.
Res., A 69 (2004) 728-737
Q. Liu, T. Douglas, C. Zamponi, S. T. Becker, E. Sherry,
S. Sivananthan, F. Warnke, J. Wiltfang, P.H. Warnke,
Clin. Oral. Implan. Res. 22 (2011) 1259-1264
152
D. MARKOVIC et al.: BIOLOGICAL ASPECTS OF APPLICATION…
[88]
R.R. Recker, In: Disorders of Bone and Mineral Metabolism, F.L. Coe, M.J. Favus (Eds.), Raven Press, New
York, 1992, pp`. 219-240
DEJAN MARKOVIĆ1
IVANA KARADŽIĆ2
VUKOMAN JOKANOVIĆ3
ANA VUKOVIĆ1
VESNA VUČIĆ2
1
Klinika za dečju i preventivnu
stomatologiju, Stomatološki
fakultet, Univerzitet u Beogradu, Dr
Subotića 11, 11000 Beograd,
Srbija
2
Centar izuzetnih vrednosti za
ishranu i metabolizam, Institut za
medicinska istraživanja, Univerzitet
u Beogradu, Tadeuša Košćuška 1,
11000 Beograd, Srbija
3
Laboratorija za radijacionu hemiju
i fiziku, Institut za nuklearne nauke
“Vinča”, University of Belgrade,
Mike Petrovića Alasa 12-14, 11000
Beograd, Srbija
Chem. Ind. Chem. Eng. Q. 22 (2) 145−153 (2016)
[89]
M. M. Stevens, Mater. Today. 11 (2008) 18-25
[90]
Y. Ikada, J. R. Soc. Interface 3 (2006) 589-601.
BIOLOŠKI ASPEKTI PRIMENE NANOMATERIJALA
U TKIVNOM INŽENJERSTVU
Milioni pacijenata širom sveta imaju potrebu za hirurškim procedurama radi reparacije ili
nadoknade oštećenog tkiva nakon traume ili oboljenja. U cilju rešavanja problema izgubljenog tkiva, tkivno inžinjerstvo glavni akcenat stavlja na tkivnu regeneraciju a ne na zamenu tkiva, zbog čega postaje predmet sve većeg interesovanja. Matične ćelije i ćelijski
nosači - visoko porozni biomaterijali, tzv. skafoldi, imaju esencijalnu ulogu u stvaranju
novog tkiva putem tkivnog inžinjerstva. Ćelijska komponenta je neophodna zbog stvaranja
i uspostavljanja ekstracelularnog matriksa, dok je skafold zadužen za da odredi oblik novostvorenog tkiva i olakša migraciju ćelija na željeno mesto, njihov rast i diferencijaciju. Ovaj
pregledni rad opisuje vrste, karakteristike i klasifikaciju matičnih ćelija. Osim toga, uključuje
i neophodne funkcionalne osobine ćelijskih nosača – biokompatibilnost, biorazgradljivost i
mehanička svojstva biomaterijala u primeni tkivnog inženjerstva. Takođe, objašnjava razlog interesovanja za praktičnu primenu nanomaterijala i nanotehnologije i definiše izazove i
značaj daljih istraživanja u ovoj oblasti.
Ključne reči: tkivno inženjerstvo, nanomaterijali, ćelijski nosači (skafoldi),
matične ćelije, regeneracija tkiva.
NAUČNI RAD
153
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 155−165 (2016)
SAFIYE BAĞCI
AYHAN ABDULLAH CEYHAN
Selcuk University, Department of
Chemical Engineering, Konya,
Turkey
SCIENTIFIC PAPER
UDC 66.081.3:661.183:544.4
DOI 10.2298/CICEQ150522030B
CI&CEQ
ADSORPTION OF METHYLENE BLUE ONTO
ACTIVATED CARBON PREPARED FROM
Lupinus albus
Article Highlights
• Activated carbon was prepared from Lupinus albus by zinc chloride activation
• Activated carbon was used to remove methylene blue from aqueous solutions
2
• Surface area of 1254 m /g was characterized for activated carbon
• Maximum MB capacity of 109.89 mg/g was reported
• The kinetic data were well described by pseudo-second-order model
Abstract
The adsorption of methylene blue (MB) from synthetic aqueous solutions in
batch experiments using Lupinus albus-activated carbon (LAAC) by chemical
activation with zinc chloride was investigated. Prior to adsorption experiments,
surface/physical properties of LAAC were determined using scanning electron
microscopy, Fourier transform infrared spectroscopy and nitrogen adsorption
isotherm. In the adsorption experiments, effects of adsorption time, solution
pH, MB concentration and amount of LAAC were investigated. The isotherm
and kinetic parameters were used to describe the experimental data. The BET
surface area was 1254 m2/g while its total pore volume was found to be 0.484
cm3/g. Maximum adsorption capacity occurred at solution pH value 10 and was
recorded as 109.89 mg/g. Adsorption data were modeled using Langmuir,
Freundlich and Temkin adsorption isotherms. Langmuir isotherm and pseudosecond-order models fit to the process and reaction kinetics correspondingly.
Keywords: activated carbon, adsorption, Lupinus albus, methylene blue,
zinc chloride.
Activated carbon is a carbon based amorphous
material having high surface area and high porosity.
Due to its high adsorption capacity, activated carbon
has a wide range of applications, mostly used as a
removing agent for both organic and inorganic materials [1].
There are two basic methods used in activation
of activated carbon: chemical activation and physical
activation. The natures of the raw materials, impregnation time and ratio, nature of the chemical activator,
activation time and temperature have direct impact on
the micro- and macro-porosity of the prepared activated carbon [2].
Correspondence: A.A. Ceyhan, Selcuk University, Department
of Chemical Engineering, Konya, Turkey
E-mail: bagcisafiye@gmail.com; ceyhan@selcuk.edu.tr
Paper received: 22 May, 2015
Paper revised: 3 July, 2015
Paper accepted: 31 July, 2015
There is an increasing trend in environmental
and water pollution due to the advancements in industrial fields. One of the sources of environmental and
water pollution is anionic and cationic dyes. The presence of these waste dyes affects the life of a number
of organisms in nature. Dyes are used as coloring
material in many industries such as textile, leather,
cosmetics, pulp and paper, plastics, pharmaceuticals
and food industry (3). Dye residues from paint and
other industries are among the basic causes of
coloring observed in waste water. These residues in
waste water are carcinogenic [3,4]. Relatively mild
concentrations of the dyes are capable of coloring a
huge amount of water and their removal is costly.
This tends to limit subsequent use of the waste water
in other fields [3,4].
When the toxic effects of dyes are taken into
account, the importance of removing the materials
from waste water becomes more vivid. Various physical, chemical and biological waste water purification
155
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
methods are used. One of these purification methods
is adsorption. Due to its inert nature towards toxin
impurities, ease of design, being user friendly and
incurring lower establishing costs, adsorption has
become one of the most preferred methods in removing dyes [5].
The activated carbon used in adsorption process can be obtained from various sources. Due to
the high production costs, production methods that
involve cheap raw materials are preferred. Recently,
agricultural and industrial by-products have been
investigated and found to be favorable for the preparation of activated carbon.
Lupinus albus (LA) plant is mostly grown in the
Mediterranean and North America. In Turkey, the plant
is grown in Ege, Marmara, inner-Anatolia and around
the Mediterranean zones. These seeds of LA are used
in production of pharmaceuticals and as animal feeds
[6]. Used seeds of this plant are directly disposed. To
our knowledge, there is no study that deals with LA
plant seeds being used for preparation of activated
carbon.
In this study, for the first time, LA plant was used
in preparation of activated carbon. By using the
obtained LAAC, removal of the MB from the aqueous
solution was studied.
EXPERIMENTAL
Materials
The LA plant seeds were obtained from Konya
region-Turkey. The seeds were ground (Retsch SR
300) and classified (Retsch AS200) based on their
respective particle sizes. The particles with sizes
under 250 µm were used for preparation of activated
carbon. Then, they were washed with de-ionized
water, dried at 105 °C for 48 h and stored in an
enclosed container ready to be used in the preparation of activated carbon. During the experiments,
deionized water and chemicals at analytical purity
were used. MB was selected for adsorption experiments because of its known strong adsorption onto
solids. MB dye was purchased from Merck Chemicals
Company, Turkey. MB (C16H18N3SCl, molecular
weight, 319.85 g/mol, wave length, 668 nm) was used
without further purification.
Preparation of LAAC
First, thermogravimetric analysis (TGA) of the
LA plant seeds considered for the preparation of
LAAC was carried out. Thermogravimetric analysis
was performed using a Perkin Elmer TG/DTA 6300. A
10±0.5 mg LA plant sample was carbonized to be
156
Chem. Ind. Chem. Eng. Q. 22 (2) 155−165 (2016)
heated from ambient temperature to 900 °C in N2
atmosphere with a flow rate of 40 ml/min, at a linear
heating rate of 10 °C/min. The test was carried out at
least twice to decrease the test error, and good
reproducibility was established. The DTG curve was
derived from the TG curve.
In obtaining activated carbon, the chemical activation method with zinc chloride was used. The activation process was carried out in a tube furnace
having inner diameter of 6 cm and a heat zone of
length of 25 cm (Magma Therm MTTF12/75/600-E-4).
After mixing the raw material (5 g) with zinc chloride
at the ratio of 1:1, 1:3, 1:5, 1:7 and 1:9 (g raw material/g zinc chloride activator), 20 ml of de-ionized
water was added after which impregnation took place
for periods of 6, 12, 24, 48, 72, 96 and 120 h. The
impregnated raw materials were subjected to activation process, which was carried out at the heating
rate of 10 °C/min. under operating temperatures of
350, 400, 500 and 600 °C. The activation periods
were selected to be 15, 30, 45 and 60 min. The activation and cooling processes were completely
accomplished under N2 gas medium. In order to rinse
the zinc compounds off the obtained activated
carbons and open blocked pores, the activated carbons were washed with 0.5 M HCl acid. Then, the
rinsing of the LAAC with hot de-ionized water was
prolonged until the solution pH value became the
same as that of de-ionized water.
In this study, the activated carbon obtained from
the conditions of impregnation ratio of (1:7), impregnation time of 48 h, activation temperature of 400 °C
and activation time of 30 min was used.
Sample characterization
Prediction of the specific surface area, porosity
volume and porous diameter of the LAAC was done
with BET device (Quantachrome Nova 1200),
whereas the micro porosity volume was calculated by
using the t-plot micro porosity volume method. Distribution of the porosity sizes was specified by making
use of the Barret-Joyner-Halenda (BJH). The porosity
structure and surface morphologies of the LAAC
obtained were found by using an SEM device (Zeiss
Evo/LS 10). The surface chemistry was characterized
by determination of surface acidity/basicity with FT-IR
spectroscopy (Perkin Elmer 1100 series FT-IR device
at the range of 4000-400 cm-1 by using KBr pallets at
an accuracy of 1 cm-1). Zeta potential measurements
were also carried out using a Malvern Zeta-sizer Zen
3600.
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
Adsorption experiments
During the adsorption process, 0.2 L solutions of
MB having initial concentrations that range from 50 to
300 mg/L were used. The experiments were conducted into a thermostatic shaker (Memmert WNB
7-45) and at constant temperature of 30 °C. At the
beginning, the solution temperature was taken to be
30 °C, whereas the initial solution concentration was
set at 100 mg/L. The adsorption of MB onto LAAC
was investigated as a function of time in order to find
out the equilibrium time for maximum adsorption capacity. Optimum adsorption time intervals for different
pH values ranging between 2 to 10 were determined.
Samples were taken from the solution (range: 0–
–180 min) and the concentrations were determined.
With the given optimum adsorption time and pH
values, effects of the amount of LAAC having weight
range of 0.05 to 0.5 g on the adsorption process were
studied. At the end of the adsorption process, a UV
spectrophotometer (Biochrom Libra S22 UV) was
used to determine the concentrations of MB in the
solution. The amount of MB adsorbed by the adsor-
Chem. Ind. Chem. Eng. Q. 22 (2) 155−165 (2016)
bent at equilibrium, qe (mg/g), is calculated by using
Eq. (1) below:
qe =
(c 0 − c e ) V
w
(1)
where c0 (mg/L) is the initial concentration of MB in
the solution, ce (mg/L) is the concentration of MB in
the solution at equilibrium, qe (mg/g) is the amount of
MB adsorption at equilibrium, and V (L) is the solution
volume.
RESULTS AND DISCUSSION
First, 10 mg sample was carbonized to be
heated from ambient temperature to 900 °C. Experiment was conducted under the N2 atmosphere. DTG
curve was derived from TG curve. The temperature of
activated carbon preparation was determined as 400
°C by using DTG curve. Thermogravimetric analysis
(TGA, DTG) of the LA plant under N2 atmosphere are
shown in Figure 1.
Figure 1. The TG and DTG graphs of LA plant seeds.
157
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
TG and DTG curves for LA plant seed are
illustrated in Figure 1. The TG curve shows the percentage loss of mass for LA plant seeds at different
temperatures. It can also be observed the thermal
decomposition of seeds take place in two steps. The
TG curves for the 10 °C/min. heating rate shows a
gradual weight loss starting at 50 °C temperature and
ending at about 177 °C with a DTG peak at 100 °C.
The total loss of weight up to 177 °C is related to the
loss of absorbed water by LA plant seeds. One broad
peak is observed in the temperature range from 177
up to 511 °C with a DTG peak at 303 °C. The total
loss of weight up to 511 °C is related to pyrolysis of
PA plant seeds. The DTG curve is identical to the TG
curve as well. According to the TG and DTG results,
the thermal decomposition of LA plant seed can be
written as follows;
Chem. Ind. Chem. Eng. Q. 22 (2) 155−165 (2016)

50 −177 C
LA seed ⋅ xH2O ⎯⎯⎯⎯⎯
→ LA seed + xH2O(gas)

177 − 511 C
LA seed ⎯⎯⎯⎯⎯⎯
→ LAACsolid + XY(gas)
80% weight loss
(I)
(II)
As shown in n reaction steps above, the thermal
decomposition of LA plant seeds (LAseed) can be subdivided into two main stages – dehydration and pyrolysis, respectively. In the first step, the absorbed
water exists in the LA plant seeds at the low temperature range. In the next step, pyrolysis of LA plant
seeds occurs at high temperature range from 177–511 °C, resulting in the formation of LAAC. Different
gases compounds are represented as XY(gas).
The FT-IR spectrum given in Figure 2a presents
functional groups found in the structure of pure and
LAAC obtained with a 2-day chemical activation with
350% ZnCl2 at 400 °C. According to the given spec-
100
95
2853
3282
2925
1746
1541
1643
% Transmittance
1027
90
85
65
2925
60
2516 2193
2853
55
50
45
4000
1569
1139 874
pure
activated carbon
3500
3000
2500
559
2000
1500
1000
500
-1
Wavenumber, cm
(a)
(b)
Figure 2. a) FT-IR spectra for the activated carbon obtained at 400 °C temperature, 1:7 impregnation ratio, 48 h of impregnation time
and 30 min of activation period; b) N2 gas adsorption and desorption isotherm at –196 °C for the activated carbon obtained at 400 °C
temperature, 1:7 impregnation ratio, 48 h of impregnation time and 30 min of activation period.
158
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
trum, the spread peak seen in the band ranging
between 3200 and 3600 cm-1 is caused by the —OH
stretching vibration mode of the hydroxyl functional
groups. This is generally caused by the moisture
found in the shell structure. The peaks observed at
2900 and 2800 cm-1 band indicate the aliphatic
groups found in the structure of the LA shells. The
peaks observed at 1750 cm-1 show carbonyl (C=O)
stretching, which indicates that there might be derivatives of aldehyde, ketone and ester. The peaks
observed at about 1650 cm-1 were caused by the
stretching of the C=C structure. The peak found at
about 1100 cm-1 in the spectrum is caused by the C—O
stretching which indicates possible existence of the
acids, alcohols, phenols, ethers and esters in the raw
material.
When the spectrum of the produced LAAC was
compared with the FT-IR spectrum of the original
walnut shell specimen, substantial changes in the
functional groups were found [7]. At around 3400 cm-1
band, the observed maxima resulted from the —OH
stretching disappeared completely. Even the peaks
observed at 2900 and 2800 cm-1 that show aliphatic
groups existing in the LA structure were not in the
structure of the LAAC. However, the peaks observed
at 1600 and 1100 cm-1 were found to have increased
their intensities.
In Figure 2b, isothermal absorption and desorption of N2 gas at –196 °C, for the LAAC were obtained
by setting operating conditions at 400 °C, impregnation ratio of 1:7, impregnation time of 48 h and
activation time of 30 min. According to the IUPAC
classification, these types of isotherms cope with the
type IV isotherm class. This situation can be described with the narrow outlet discharging. Generally,
adsorption isothermals on micro- and mezzo-porous
layers resemble this type [8]. The surface area and
micropore volume of the LAAC prepared are 1254
and 0.3584 cm3/g, respectively. The average width of
the LAAC’s BJH adsorption is 1.54 nm. This result
indicates that the average porous width for the BJH
adsorptionis of micro type.
SEM photographs of the LAAC are shown in
Figure 3. As seen from the SEM images, with well-developed pores, the obtained LAAC has a random
and heterogeneous surface. The surface of the LAAC
has cracks, grooves and large pores. In addition,
pores of different sizes are also present.
Adsorption studies
Effect of pH
The adsorption procedures started with the
investigation of effects of solution pH. The studies
Chem. Ind. Chem. Eng. Q. 22 (2) 155−165 (2016)
were carried out at a constant temperature of 30 °C in
MB concentration of 100 mg/L and with 0.1 g of the
LAAC. The solution pH was varied between 2 and 10.
The results obtained are presented in Figure 4a.
Figure 3. SEM views of the produced activated carbon.
The increase in solution pH from 2 to 10 has led
to significant increase in the adsorption capacity. The
most probable reason for this increase is the pH dependent electrostatic interaction between the activated
carbon and the MB molecules. When the solution pH
increased from 2 to 10, the adsorption capacity of the
activated carbon increased from 45.19 to 89.80 mg/g.
As for the adsorption of the MB, the presence of two
possible mechanisms may be pronounced. These
are:
159
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
-
-
electrostatic interaction between MB molecules and positively charged groups of the
activated carbon and
chemical reaction between the adsorbent
and the MB molecules.
(a)
(b)
Chem. Ind. Chem. Eng. Q. 22 (2) 155−165 (2016)
of the activated carbon surface and increase in the
adsorption of the MB [10-12]. Other researchers also
found that the optimum solution pH value in removing
of MB with activated carbon is 10 [9,13-17].
Effect of contact time
Contact time varies with respect to features
such as the pore diameter, volume and surface areas
of activated carbon [12]. In this section of the study,
efforts have been made to determine adsorption
capacity at equilibrium by maintaining parameter
values such that pH is at 10, the amount of LAAC is
0.1 g, solution temperature is kept at 30 °C and concentration is kept at 100 mg/L (Figure 5a). It was
found that the maximum adsorption capacity for the
MB at the end of 120 min interval is 89.80 mg/g. The
amount of MB adsorbed increases with the contact
time and at the onset of the 120th min, the amount
reaches its equilibrium value.
(a)
Figure 4. a) Variation of the adsorption capacity, qe, with
solution pH (T: 30 °C, MB concentration: 100 mg/L, LAAC
mass: 0.1 g); b) the point of zero charge values.
Activated carbon is amphoteric in nature. Its
surface electrical charge can vary between positive
and negative depending on the solution pH [9]. The
point at which the net charge of adsorbent is equal to
zero (pHPZC) can be used to characterize the effect of
pH. It means smaller pH solution of adsorbent than
pHPZC results positive surface and bigger one negative surface. Variation of surface charge of the LAAC
with pH was investigated and the results are as
shown in Figure 4b. The adsorption of the MB onto
the LAAC having cationic character is more remarkable if it acquires the condition that is pH > pHPZC
(point of zero charge) which is equal to 3.14.
When the solution pH is low, the H+ relocate
themselves into active surface sites onto the activated
carbons and hence prevent adsorptions of the MB
molecules having cationic characters. With increasing
pH, the surface charge reaches a zero point of charge
(pHPZC), which is the zero net charge of adsorbents
(at pH values of 3.14). After this point, an increase in
pH value leads to the increase of negative character
160
(b)
(c)
Figure 5. a) Time for the adsorption to reach equilibrium; b)
effects of initial MB concentration on adsorption; c) effects of the
amount of activated carbon used on adsorption of MB (T: 30 °C,
MB concentration: 100 mg/L, pH 10, contact time: 120 min).
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
During adsorption process from the solution to
the surface of the activated carbon, the followings
happen:
- transfer of MB molecules from the solution to
the solution-adsorbent boundary layer,
- diffusion from boundary layer to the adsorbent surface,
- diffusion phases from adsorbent surface into
its pores [15].
The results show that the adsorption process
occurs faster at the beginning of the contact time
(≈40%). However, the process proceeds gradually
step by step towards equilibrium value. The vacant
active surface sites on the surface fill quickly as the
adsorption process starts. As the number of vacant
active surface sites on the surface decreases with
time and the diffusion rate attenuates, the time to
reach equilibrium tends to increase [18].
Effect of initial MB concentration
In this section of the study, effects of MB solutions with concentrations of 50, 75, 100, 150, 200,
250 and 300 mg/L on the adsorption capacity of
LAAC are investigated. Figure 5b shows variation of
the initial concentration of MB with the amount of MB
adsorbed by the LAAC. It is seen that when equilibrium is reached, adsorption capacity becomes greater
for higher initial concentrations. At higher initial concentrations, mass transfer resistance between solid
surface and solution is easily overcome. With increasing MB concentration, vacant active surface sites
on the adsorbent surface tend to fill quickly. However,
the rate of removal percentage of the MB (R%) is
inversely proportional to the increasing initial concentration. For low initial concentration, competition of
MB molecules for adsorption to the surface is at much
lower rates [19]. Moreover, the adsorption equilibrium
is achieved more quickly. Even the ratio of MB concentration in the solution to the number of active sites
on the adsorbent is lower. That’s why the concentration of removed MB becomes higher [12]. Similar to
the results of this study, in studies conducted on
removal of MB by using activated carbons, it was
determined that with the increasing solution concentration, the removal percentage of the MB tends to fall
[20-22].
Effect of adsorbent dosage
The size, amount and pore volume of the activated carbon used in the adsorption are among the
most important parameters. As the amount of activated carbon increases, the surface area tends to increase and hence adsorption capacity also increases.
In this phase of the study, the operating conditions
Chem. Ind. Chem. Eng. Q. 22 (2) 155−165 (2016)
specified previously (100 mg/L, 30 °C, pH 10, contact
time 120 min) were kept constant and the experiments were conducted (Figure 5c). The amount of
LAAC used was varied between 0.05 and 0.5 g. By
increasing the amount of LAAC to 0.3 g, the removal
percentage of the MB jumps to a value of 99%.
However, beyond this point, additional LAAC was
found to have no significant effects in removing the
MB. Another result indicates that the adsorption capacity (qe) deteriorates substantially with the increasing
amount of activated carbon. Possible reasons for this
are the presence of more vacant active surface sites
in the medium than the MB concentration needs,
lowering of surface area as a result of accumulation
of adsorbent particles, and elongation of the path to
be taken during the diffusion step [12,23].
Adsorption isotherms
Adsorption isotherms play an important role in
estimating real time adsorption capacities of adsorbents of different nature. The equilibrium values
obtained after the adsorption of MB were investigated
by using Langmiur, Freundlich, Temkin and Dubinin–Radushkevich isotherms. The adsorption data
obtained experimentally are given in Figure 6 and
Table 1.
The Langmuir isotherm assumes that there is a
certain number of active surface sites with similar
features on the surface. These centers are evenly
distributed over the surface and the adsorption is of a
monolayer type [24]. The theory can be represented
by the following linear form:
ce
c
1
=
+ e
q e q mK L q m
RL =
1
1 + K Lc 0
(2)
(3)
where ce is equilibrium concentration (mg/L), qe is the
amount adsorbed at equilibrium (mg/g), KL is a constant related to adsorption energy and capacity (L/mg),
qm is the monolayer adsorption capacity (mg/g) and c0
refers to initial concentration of the solution (mg/L).
The separation factor (RL) which the basic characteristic of the Langmuir adsorption isotherm is calculated
by using Eq. (3). The RL value between 0 and 1 indicates that the activated carbon is favorable for using
MB adsorption [25]. The RL values found for different
initial concentrations were in the range 0 < RL < 1.
This result shows that the activated carbon prepared
from LA plant for adsorption of MB is favorable.
The Freundlich isotherm assumes that the surface possesses heterogeneous energy distribution.
This isotherm, which also covers intra-molecular inter-
161
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
CI&CEQ 22 (2) 155−165 (2016)
(a)
(b)
(c)
(d)
Figure 6. Adsorption isotherms: a) Langmuir, b) Freundlich, c) Temkin and d) D-R.
Table 1. Isotherm constants for the adsorption isotherm models
Isotherm
Langmuir
Freundlich
Temkin
Dubinin-Radushkevich
Isotherm constants
KL = 0.03
R2 = 0.995
n = 3.51
R2 = 0.974
Kt = 0.486 L/g
B = 122.053 J/mol
R2 = 0.984
qm = 86.606 mg/g
E = 93 J/mol
R2 = 0.829
qmax = 109.89 mg/g
Kf = 21.09 (mg/g)(mg/L)
action, is favorable for heterogeneous systems; and
instead of monolayer adsorption, it takes the
possibility of multilayer adsorption into account.
q e = K fc e1/ n
(4)
In Eq. (4), Kf is a constant for the system, related
to the bonding energy and n is the adsorption
intensity, which is used to account for the adsorption
process is favorable. 1/n is a measure of adsorption
intensity or surface heterogeneity. A higher value of
adsorption capacity Kf indicates that the capacity of
the prepared activated carbon is high, whereas higher
adsorption intensity, n, indicates that the adsorption
process takes place thoroughly throughout the concentration interval of solution employed. Lower n
value at higher initial MB concentration implies that
the adsorption occurs perfectly. As for lower initial MB
concentrations, it can be said that there is a little correlation between experimental data and the Freundlich isotherm.
162
1/n
In Table 1, n and Kf constants for the Freundlich
isotherm are presented. The adsorption intensity (n)
was calculated to be 3.51. This result shows that the
activated carbon prepared is favorable for the high
MB concentration. The value of 1/n = 0.289, being in
the range of 0.1 < 1/n < 1, indicates favorably of the
prepared activated carbon for adsorption process [26].
The Temkin isotherm is similar to the Freundlich
isotherm as it also takes intra-molecular interaction
into account. It is with the image that if the heat of
adsorption for all molecules covers the surface perfectly, then the heats must fall linearly. The postulate,
in addition, considers that binding energy is distributed uniformly to the maximum value.
q e = B ln K t + B ln c e
(5)
B = RT / b
(6)
In Eqs. (5) and (6), b is the Temkin isotherm
constant (J/mol), Kt is the equilibrium binding constant
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
corresponding to the maximum binding energy (L/g),
B is a dimensionless constant related to heat of adsorption, R is the ideal gas constant (8.314 J/(mol K)),
T is the temperature of the medium (K).
The Dubinin-Radushkevich isotherm aims at
establishing the view that adsorption occurs either
physically or chemically. Based on this isotherm, it is
described that if the average adsorption energy is
lower than 8 kJ/mol, then physical adsorption takes
place, and if it is between 8 and 16 kJ/mol then ion
variation adsorption occurs, whereas higher adsorption energy values is characterised by chemical adsorption [13].
B ln q e = ln q m − k ε
2
(7)
ε = RT ln(1 + 1/ c e )
(8)
E = ( 2k )
(9)
−1/ 2
In Eqs. (7)-(9), qs is denoted as the theoretical
isotherm saturation capacity(mg/g), ε is the Polanyi
potential (J/mol), k is the activity coefficient (J2/mol2),
and E is the mean free energy (J/mol). The calculated
mean free energy (E) value was smaller than 8
kJ/mol. This result shows that physical interaction
between the adsorbent and the adsorbate is dominant
and that the process occurred by physical adsorption.
As shown in Table 1, the largest correlation
coefficient (R2) is from the Langmuir isotherm. The
phenomenon implies that the removal of the MB is
well described by the Langmuir isotherm. In addition,
it also indicates that the adsorption has essentially
taken place in a monolayered form. However, due to
energy distribution discrepancies on the surface, some
sites were marked with multi-site adsorption [27].
Adsorption kinetics
In this section of the study, kinetic behaviors of
the adsorption process are taken into consideration
by making use of the adsorption experimental data.
With this aim, three different kinetic models were
used, namely, the pseudo-first-order, pseudo-secondorder and intraparticle diffusion model.
The variation of the adsorption capacity with
time of the pseudo-first-order model developed by
Lagergren and Svenska [28] is described by Eq. (10):
1
qt
=
1
qe
+
k1 1
qe t
(10)
where k1 is the apparent kinetic rate constants of first-order reaction kinetic (min-1) and t is the reaction time
(min).
Chem. Ind. Chem. Eng. Q. 22 (2) 155−165 (2016)
The variation of adsorption capacity with time in
the pseudo-second-order model developed by Ho and
McKay [29] is expressed by Eq. (11):
t
1
1
=
+ t
2
q t k 2q e q e
(11)
where k2 is the apparent kinetic rate constants of
second-order reaction kinetic (g/(mg min)).
The intraparticle diffusion model developed by
Weber and Morris [30] is expressed by Eq. (12):
qt = k idt 1/2 + C
(12)
where kid is the intraparticle diffusion rate constant
(mg/(g min1/2)), and C is a constant that gives an idea
about the thickness of the boundary layer (mg/g). If
intraparticle diffusion is involved in the overall adsorption process, the plot of uptake (qt)versus the square
root of time (t0.5) should be linear. Moreover, if this line
passes through the origin, the intraparticle diffusion is
the rate controlling step of the process. If the plots do
not pass through the origin, it points out of some
degree of boundary layer control. Besides, it shows
that the intraparticle diffusion is not the only rate-limiting step, but also other kinetic models may control the
rate of adsorption process.
These plots usually are depicted as three stages
such as curve, linear and plateau in turn. The initial
stage is caused by external mass transfer. The intermediate linear stage is caused by intra-particle diffusion. The last plateau stage where intraparticle diffusion starts to slow down is caused by extremely low
solute concentrations in the solution [31]. The adsorption graphs of the models considered in the study are
given in Figure 7.
When deciding on the validity of a model,
besides the correlation coefficient (R2), closeness of
the adsorption capacities (qe) is evaluated as well
[32]. The normalized standard deviation Δq is calculated using the following equation:
(
)
 q exp − q cal / q exp 

Δq = 100 
N −1
2
(14)
Values of Δq for the pseudo-first-order model,
pseudo-second-order model and the intraparticle
diffusion model as well as kinetic parameters were
calculated and presented in Table 2.
From Table 2, the largest correlation value
(0.999) and the lowest normalized standard deviation
(1.002) occur on the pseudo-second-order model.
More concentrations (50 and 150 ppm) were used to
establish the kinetics models. Similar results were
found for 100 pm (figures not shown).
163
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
CI&CEQ 22 (2) 155−165 (2016)
(a)
(b)
(c)
Figure 7. Adsorption kinetic models: a) pseudo-first order, b) pseudo-second order, c) intra-particle model (T: 30 °C, MB concentration:
100 mg/L, LAAC mass: 0.1 g, pH: 10, contact time: 120 min).
The pseudo-second-order model assumes that
the dominant process is chemical adsorption, and that
there are chemical electrostatic interactions between
the adsorbent surface and the adsorbate. The surface
is monolayered but it is most likely that some sites
can be multilayered as well [32].
The intra-particle diffusion model (Weber-Morris
model) is widely used for the purpose of determining
the rate-specifying stage. On the graph of intraparticle
diffusion model, though the variation is almost linear,
and the lines do not cross the origin. This phenomenon indicates that the adsorption is not only ratecontrolled but also a boundary layer diffusion controlled. In addition, it infers to macroporous diffusion
and suggests that the active surface sites on the
surface fill up quickly [33]. The calculated α value was
found to be 0.12. The value of α being smaller than
0.5 indicates that the intraparticle diffusion is not the
only step that depicts diffusion rate [34].
CONCLUSION
In this study, activated carbons were prepared
from LA plant by using chemical activation with ZnCl2
164
under N2 gas atmosphere. Effects of parameters like
impregnation ratio, impregnation time, activation temperature and activation time were studied. The impregnation ratio of 1:7 and impregnation time of 48 h
were selected. Activation temperature and activation
time were 400 °C and 30 min, respectively. The BET
surface area of the LAAC was determined to be 1254
m2/g. Micropore volume of the activated carbon was
found to be 0.3584 cm3/g based on the t-plot method.
According to the BJH adsorption average, the porous
diameter is 1.54 nm. During the adsorption process it
was also found that the maximum adsorption capacity
is 109.89 mg/g and occurs at pH value of 10. It was
found favorable to describe the adsorption based on
the Langmuir isotherm. As for the reaction kinetics,
the pseudo-second-order model was found to be a
convenient match.
Acknowledgements
This study has been financially supported by the
Office of Scientific Projects Coordinator of Selcuk
University (BAP) (Project No: 12201090).
S. BAĞCI, A.A. CEYHAN: ADSORPTION OF METHYLENE BLUE…
Chem. Ind. Chem. Eng. Q. 22 (2) 155−165 (2016)
REFERENCES
[18]
M. Auta, B.H. Hameed Chem. Eng. J. 175 (2011) 233–
–243
[1]
J. Kong, Q. Yue, L. Huang, Y. Gao, Y. Sun, B. Gao, Q. Li,
Y. Wang, Chem. Eng. J. 221 (2013) 62–71
[19]
[2]
M.O. Abdullah, I.A.W. Tan, L.S. Lim, Renewable Sustainable Energy Rev. 15 (2011) 2061–2072
Z. Zhang, Z. Zhang, Y. Ferna´ndez, J.A. Mene´ndez; H.
Niu, J, Peng, L. Zhang, S. Guo, Appl. Surf. Sci 256 (2010)
2569–2576
[20]
[3]
M. Rafatullah, O. Sulaiman, R. Hashim, A. Ahmad, J.
Hazard. Mater. 177 (2010) 70–80
F. Raposo, M.A. De La Rubia, R. Borja, J. Hazard. Mater.
165 (2009) 291–299
[21]
[4]
M. Arulkumar, P. Sathishkumar, T. Palvannan, J. Hazard.
Mater. 186 (2011) 827–834
Y. Guo, J. Qi, S. Yang, K. Yu, Z Wang, H. Xu, Mater.
Chem. Phys. 78 (2003) 132–137
[22]
[5]
M. Berrios, M.A. Martín, A. Martín, J. Ind. Eng. Chem. 18
(2012) 780–784
M. Ghaedi, Sh. Heidarpour, S.N. Kokhdan, R. Sahraie, A.
Daneshfar, B. Brazesh, Powder Technol. 228 (2012) 18–
–25
[6]
O. Güldemir, N. Işık, F. Tosun, Sokaklardan evlere:
Termiye, 3. Geleneksel Gıdalar Sempozyumu, Konya,
Turkey, 2012, p. 549
[23]
J. Yang, K. Qiu, Chem. Eng. J. 165 (2010) 209–217
[24]
A.W. Adamson, Physical Chemistry of Surface, Interscience Publication, New York, 1960, p. 603
M. Kılıç, E.A. Varol, A.E. Pütün, Appl. Surf. Sci. 261
(2012) 247–254
[25]
N. Kannan, M.M. Sundaram, Dyes Pigm. 51 (2001) 25–40
[26]
T. Karthikeyan, S. Rajgopal, L.R. Miranda, J. Hazard.
Mater. 124 (2005) 192–199
[27]
Y. Gao, Q. Yue, B. Gao, Y. Sun, W. Wang, Q. Li, Y.
Wang, Chem. Eng. J. 217 (2013) 345–353
[7]
[8]
Y. Sarıkaya, Fizikokimya, Gazi Kitabevi, Ankara, 2000, p.
633
[9]
Y. Bao, G. Zhang, Energy Procedia 16 (2012) 1141–1146
[10]
G. Mezohegyi, F.P. van der Zee, J. Font, A. Fortuny, A.
Fabregat, J. Environ. Manage. 102 (2012) 148-164
[28]
S. Lagergren, K. Sven, Vetenskapsakad. Handl. 24
(1898) 1–39
[11]
K. Gobi, M.D. Mashitah, V.M. Vadivelu, Chem. Eng. J.
171 (2011) 1246– 1252
[29]
Y.S. Ho, G. Mckay, Process Biochem. 34 (1999) 451–465
[30]
W.J. Weber, J.C. Morris, J. Sanit. Eng. Div., Am. Soc.
Civil Eng. 89 (1963) 31–59
[31]
A.A.M. Daifullah, S.M. Yakout, S.A. Elreefy, J. Hazard.
Mater. 147 (2007) 633-644
[12]
M. Ghaedi, A.G. Nasab, S. Khodadoust, M. Rajabi, S.
Azizian, J. Ind. Eng. Chem. 20 (2014) 2317–2324
[13]
C. Saka, Ö. Şahin, Color. Technol. 127 (2011) 246–255
[14]
H. Cherifi, B. Fatiha, H. Salah, Appl. Surf. Sci. 282 (2013)
52-59
[32]
S.K. Theydan, M.J. Ahmet, J. Anal. Appl. Pyrolysis 97
(2012) 116–122
[15]
S. Senthilkumaar, P.R. Varadarajan, K. Porkodi, C.V.
Subbhuraam, J. Colloid Interface Sci. 284 (2005) 78–82
[33]
D. Kavitha, C. Namasivayam, Bioresour. Technol. 98
(2007) 14–21
[16]
H. Deng, L. Yang, G. Tao, J. Dai, J. Hazard. Mater. 166
(2009) 1514-1521
[34]
K.K. Singh, R. Rastogi, S.H. Hasan, J. Colloid Interface
Sci. 290 (2005) 61–68.
[17]
M. Ertaş, B. Acemioğlu, M.H. Alma, M. Usta, J. Hazard.
Mater. 183 (2010) 421–427
SAFIYE BAĞCI
AYHAN ABDULLAH CEYHAN
Selcuk University, Department of
Chemical Engineering, Konya, Turkey
NAUČNI RAD
ADSORPCIJA METILENSKOG PLAVOG NA
AKTIVNOM UGLJU PRIPREMLJENOM OD
Lupinus albus
U radu je analizirana adsorpcija metilenskog plavog (MB) iz vodenih rastvora u šaržnim
eksperimentima na aktivnom uglju dobijenom od Lupinus albus (LAAC), koji je hemijski
aktiviran cink-hloridom. Pre ispitivanja adsorpcije, površinske i fizičke osobine LAAC su
određene elektronskom mikrosopijom, infracrvenom spektroskopijom sa Furijeovom transformacijom i adsorpcijom azota. U adsorpcionim eksperimentima istraživani su uticaji
vremena adsorpcije, pH rastvora, koncentracije MB i količine LAAC. Na osnovu dobijenih
eksperimentalnih podataka određeni su kinetički parmetri i parametri adsorpcionih izotermi. Specifična površina po BET metodi je bila 1254 m2/g, dok je ukupna zapremina pora
bila 0,484 cm3/g. Maksimalni adsorpcioni kapacitet je 109,89 mg/g sa rastvorom čiji je pH
10. Eksperimentalni podaci su modelovani Lengmirovom, Frojndlihovom i Temkinovom
adsorpcionom izotermom. Kinetika i ravnoteža adsorpcije slede model pseudo drugog
reda i Langmuirovu izotermu, redom.
Ključne reči: aktivni ugalj, adsorpcija, Lupinus albus, metilensko plavo, cink-hlorid.
165
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
DIDEM ÖZÇIMEN1
TUFAN SALAN2
1
Department of Bioengineering,
Faculty of Chemical and
Metallurgical Engineering, Yıldız
Technical University, Davutpasa
Campus, Istanbul, Turkey
2
Department of Material Science
and Engineering, Kahramanmaras
Sutcu Imam University, Avsar
Campus, Kahramanmaras, Turkey
SCIENTIFIC PAPER
UDC
DOI 10.2298/CICEQ141128029O
CI&CEQ
REMOVAL OF REACTIVE DYE REMAZOL
BRILLIANT BLUE R FROM AQUEOUS
SOLUTIONS BY USING ANAEROBICALLY
DIGESTED SEWAGE SLUDGE BASED
ADSORBENTS
Article Highlights
• Adsorbents were produced via thermal and chemical activation processes
• Adsorption studies were conducted in batch equilibrium conditions for different dye
concentrations
• Chemically activated sewage sludge indicated better porous structure
• Adsorption process demonstrated better compliance with pseudo-second order kinetic model
• Freundlich isotherm provided better fit to the equilibrium data for all adsorbents
Abstract
In this study, adsorbents were produced from sewage sludge via chemical and
thermal activation processes. Experiments were carried out in a tubular furnace at a heating rate of 20 °C min-1 and temperature of 550 °C with a nitrogen
flow rate of 400 mL min-1 for 1 h. Dye adsorption experiments were performed
with Remazol Brilliant Blue R at several concentrations under batch equilibrium
conditions by comparing sewage sludge based adsorbents with raw material
and a commercial activated carbon. Maximum adsorption capacities of carbonized sewage sludge (CSWS) and activated sewage sludge (ASWS) were
found as 7.413 and 9.376 mg g-1 for 100 mg L-1 dye solution, whereas commercial activated carbon had a capacity of 11.561 mg g-1. Freundlich and
Langmuir isotherms were used to explain the adsorption mechanism together
with pseudo-first-order and pseudo-second-order kinetic models. The Langmuir isotherm, which had adsorption capacities of 34.60 (CSWS) and 72.99
mg g-1 (ASWS), provided a better fit to the equilibrium data than that of the
Freundlich isotherm. The pseudo second-order model, which had adsorption
capacities of 7.451 (CSWS) and 9.319 mg g-1 (ASWS), was very favorable to
explain the adsorption kinetics of the dye with high regression coefficients.
Keywords: sewage sludge, carbonization, activated carbon, adsorption,
Remazol Brilliant Blue R.
Synthetic dyes frequently cause pollution problems when the colored wastewater is discharged into
surface waters after its applications. Because of their
resistance to conventional wastewater treatment processes, brightly colored water-soluble reactive and
acid dyes are among the most problematic dyestuff
Correspondence: T. Salan, Department of Material Science and
Engineering, Kahramanmaras Sutcu Imam University, Avsar
Campus, 46100, Kahramanmaras, Turkey.
E-mail: tufansalan@gmail.com
Paper received: 28 November, 2014
Paper revised: 3 July, 2015
Paper accepted: 31 July, 2015
[1]. Problems of synthetic dyes pollution have arisen
day by day since they are extensively used due to
their easy application techniques, inexpensive synthesis costs, low energy consumption and variety of
color compared with natural dyes [2]. Furthermore,
synthetic dyes are not easily biodegradable; therefore, after broad treatment applications colors may
still stay in the wastewater stream. These streams are
a growing threat to aquatic life because of their high
chemical oxygen (CODs) and biological oxygen demands (BODs), toxic, allergic, skin irritating, mutagenic
and carcinogenic features [3-6]. Because of these
167
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
problems, many investigations on the removal of dye
color from industrial effluents have appeared in recent
years effective physical and chemical treatment
methods including adsorption, chemical coagulation,
precipitation, ultra-filtration, electro-dialysis, ionizing
radiation, ozone oxidation and photo-catalytic degradation for the treatment of dye containing wastewater
streams [2,7-9]. Moreover, biological materials such
as algae, fungi, bacteria and yeast have been alternatively used for the removal of dyestuff from wastewater [10]. Among all these methods, adsorption process was considered excellent as compared to other
techniques for its easy design and operation, effectiveness and high efficiency [11,12].
Previous research studies have used several
different adsorbents like carbon nanotubes (CNTs) for
dye uptake from wastewater environments [13-15].
Among these adsorbents, activated carbon was found
to achieve great performance in the adsorption of
various pollutants, not only at the laboratory scale, but
also for industrial applications [16-18]. However, despite these excellent properties, utilization of commercial activated carbon is sometimes restricted due
to relatively expensive starting material and therefore
higher application costs. Furthermore, after a few
uses of the original activated carbon, a saturation
phenomenon occurs on the active surface sites and
subsequently the activated carbon cannot no longer
adsorb the adsorbates from the wastewater due to
diffusion limitations. Once the activated carbon is
saturated, a regeneration step is required for further
treatment applications [19-21].
Attempts have been made to produce low-cost
carbon-based adsorbents from alternative feedstock
such as agricultural [22-24], industrial byproducts or
waste [25-28], such as sewage sludge that may be a
substitute for commercial activated carbons in the
adsorption of various pollutants in wastewater treatment applications. Agricultural waste and industrial
byproducts are very important feedstock for the
manufacture of activated carbon since they are abundant, renewable, easy available and sustainable alternative sources.
Sewage sludge is a waste material produced as
a byproduct of wastewater purification processes in
urban and industrial wastewater treatment plants.
Nowadays, its production is rapidly increasing and will
continue to rise, because more municipal wastewater
will be treated due to environmental necessity and
legal requirements to reach better standards for
wastewater treatment along with the urbanization and
industrial development [29,30]. Sewage sludge is a
very complex substance consisting of biological, org-
168
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
anic and inorganic components and water. It contains
undigested organic compounds remained after the
treatment, which are proteins and peptides, lipids,
polysaccharides, plant macromolecules and aliphatic
structures [31].
Considering that sludge reduction is restricted
due to sustainability of wastewater treatment to reach
the higher standards, it will be desirable to develop
innovative, eco-friendly and effective new routes for
the valorization of this “waste of the waste”, so it will
turn into a useful feedstock [32]. Thus, various handling methods have been utilized for the sewage
sludge disposal in a sustainable way. These methods
can be categorized into two main strategies consisting of traditional disposal or reuse and energy applications. Until now, sewage sludge has been used as
a renewable feedstock to produce energy via anaerobic digestion, combustion, gasification, pyrolysis/carbonization and novel technologies such as wet oxidation and supercritical water oxidation [33].
Among these methods, pyrolysis (carbonization)
– the thermal destruction of biomass under inert
atmosphere – is a clean and efficient application, and
it generates valuable liquid, gas and solid products
depending on the process conditions [34]. Because of
carbonaceous structure, abundant organic compositions and volatile components, sewage sludge is a
reasonably appropriate candidate for manufacturing
porous adsorbents under controlled carbonization
conditions [35]. Activated carbon production by using
sewage sludge means both a considerable saving for
starting material costs, and a way of making economically valuable utilization of a waste via producing
a useful material. Utilization of sewage sludge for the
production of activated carbon used for the advanced
treatment process of wastewater reduces the operating expenses of the plant. Moreover, this strategy
solves pollution, odor and space occupation problems
derived from sewage sludge in the plant [33,36].
Many studies have demonstrated that the sewage sludge was a great in-expensive and easy available feedstock for the production of activated carbon
[37]. These studies mainly related with the adsorption
of phenol and phenolic compounds, heavy metals,
pollutant gases and dyes onto sewage sludge based
adsorbents [38-40]. Among these studies dye adsorption was the most commonly applied method to
determine uptake capacity of these adsorbents and
the methylene blue was frequently used as an adsorbate in the dye adsorption experiments [41,42].
Remazol Brilliant Blue R dye used as an adsorbate in our present study is one of the most important
reactive dyes in the textile industry. It is an anthra-
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
quinone-based dye and represents an important class
of toxic and recalcitrant organo-pollutants. Remazol
Brilliant Blue R (RBBR) is frequently used as a starting material in the production of polymeric dyes [43].
So far, several attempts have been made to remove
RBBR from aqueous solutions by using various adsorbents and alternative adsorption techniques.
Although several studies have been reported in the
literature about the RBBR adsorption on the various
adsorbents produced from different feedstock and
other dyes on the sewage sludge based adsorbents,
there has been no research concerning the adsorption of RBBR onto adsorbents produced from sewage
sludge. Thus, this study focuses on investigating the
RBBR removal efficiencies of low-cost and ecofriendly adsorbents produced from sewage sludge by
using chemical and thermal activation methods.
an average dry solid content of 90%. Proximate
analysis of sewage sludge sample was carried out
according to ASTM D3172-13 (Standard Practice for
Proximate Analysis of Coal and Coke). Ultimate
analysis of sewage sludge sample was performed by
using an ultimate analyzer (LECO, CHNS-932), following ASTMD3176 (Standard Practice for Ultimate
Analysis of Coal and Coke). Finally, macro and trace
element composition of sewage sludge sample was
determined by the inductively coupled plasma atomic
emission spectrometry, ICP-OES (Perkin-Elmer,
Optima 2100 DV). The results of proximate, ultimate
and elementary analyses for sewage sludge sample
are shown in Table 1. Additionally, a commercial activated carbon (Analiz Kimya Co., Turkey) was used for
the adsorption experiments for the comparison. Zinc
chloride (reagent grade, Merck) was used in the adsorption experiments as a chemical activation reagent.
A stock solution of Anthraquinone dye Remazol Brilliant Blue R (Reactive Blue 19; C22H16N2O11S3) was
used for the adsorption studies.
EXPERIMENTAL
Materials
Sewage sludge samples were collected from
Atakoy advanced biological wastewater treatment
plant, which is an urban treatment plant located in
Istanbul, Turkey. The sludge had been anaerobically
stabilized by means of biogas production process
subsequent to activated sludge treatment. Furthermore, it was dewatered via centrifugation process and
was dried by using a rotary disc drier at the plant to
Preparation of the adsorbents
Before the carbonization experiments, powdered
raw sewage sludge was dried at 105 °C in an oven
(Binder) for 24 h. It was not ground and sieved
because of its uniform particle size distribution. Thermogravimetric analysis (TGA) was performed by
using TA SDT Q600. Simultaneous TGA/DSC equip-
Table 1. Proximate and ultimate analyses of sewage sludge sample along with macro and trace element concentrations
Proximate analysis
Parameter
a
Ultimate analysis
Content,wt.%
Element
b
Content, wt.%
Moisture
5.14
Carbon (C)
38.97
Volatile matter
52.12
Hydrogen (H)
5.10
6.2
Nitrogen (N)
4.11
36.54
Sulphur (S)
0.81
Fixed carbon
Ash
c
Oxygen (O)
-1
HHV (MJ kg )
50.99
16.18
Elemental composition
d
Element
-1
Concentration , mg kg
Element
d
Calcium (Ca)
30187.55
Manganese (Mn)
202.66
Aluminum (Al)
9733.33
Boron (B)
71.46
Potassium (K)
9367.81
Lead (Pb)
64.96
Magnesium (Mg)
8927.82
Nickel (Ni)
45.41
Sulfur (S)
7366.66
Cobalt (Co)
35.55
Iron (Fe)
7358.18
Molybdenum (Mo)
5.84
Sodium (Na)
3259.73
Cadmium (Cd)
0.92
Zinc (Zn)
767.38
Arsenic (As)
nd
Chromium (Cr)
369.55
Mercury (Hg)
nd
Cupper (Cu)
a
b
-1
Concentration , mg kg
e
e
263.32
c
d
e
As received; ash free dry basis; calculated by difference; dry base weight; not detected
169
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
ment for the determination of thermal degradation
properties of sewage sludge and produced adsorbents. Thermal analyses were conducted at a heating
rate of 15 °C min-1 from 20 to 1000 °C with a nitrogen
flow rate of 100 mL min-1. Two different methods were
used for the production of adsorbents. In the first
method, dried sewage sludge samples were physically impregnated with ZnCl2 by employing a mass
ratio of 1:1 (ZnCl2 to raw material) and then, samples
were activated in a quartz pipe inserted into a horizontal split tubular furnace (Protherm ASP) with a
nitrogen flow rate of 400 mL min-1. The heating rate of
the furnace was 20 °C min-1 and the reaction was
carried out at a final temperature of 550 °C for 1 h.
Adsorbents produced via this method were referred to
as ASWS (activated sewage sludge) throughout the
manuscript. On the other hand, in the second method
dried sewage sludge was thermally activated without
activation agent at the same reaction conditions. Adsorbents produced via this method were referred to as
CSWS (carbonized sewage sludge) throughout the
manuscript. After completion of the processes, these
adsorbents were washed several times with diluted
HCl solution and hot distilled water. Finally, they were
dried at 105 °C for 24 h in the oven. The particle size
range of the obtained adsorbents was quite variable
due to nonhomogeneous structure of sewage sludge
powder. Therefore, it could roughly be determined
that particle size was generally less than 180 µm
(Mesh No. 80).
Characterization of the adsorbents
Some physical and chemical properties such as
porosity, surface area and functional groups of the
adsorbents were determined through various instruments. The scanning electron microscopy (SEM) micrographs of the adsorbents were obtained by using a
JEOL JSM 5410 LV model equipment. Specimens
were imaged at 20.0 kV with 10.0 µm in 1000× time
magnification. Spectroscopic analysis of the both of
the raw material and adsorbents was performed by
using Agilent Cary 630 ATR/FTIR (attenuated total
reflectance/Fourier-transform infrared) equipment.
The spectra of the samples were recorded in the
wavenumber region between 4000 and 650 cm-1. The
Brunauer–Emmett–Teller (BET) surface analysis of the
samples was determined by using Micromeritics
TriStar II 320 equipment after degassing process,
which was carried out at 110 °C for 1 h.
Adsorption experiments
In the adsorption experiments, sewage sludge,
produced adsorbents and commercial activated carbon (AC) as different adsorbents were used to rem-
170
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
ove RBBR dye from aqueous solutions in different
concentrations. Firstly, RBBR was dissolved in distilled water to prepare 1000 mg L-1 stock solution. Then,
this stock solution was diluted to obtain standard
concentrations of 20, 40, 60, 80 and 100 mg L-1 for
the adsorption experiments. The pH values of the
standard solutions were adjusted to 3 in all cases with
HCl/NaOH solutions. According to the following literature survey, all adsorption studies were carried out
at constant acidic conditions (pH 3).
The pH of aqueous dye solution plays a significant role for the adsorption process of dye molecules. Dye adsorption is highly pH-dependent due to
effect of pH on the surface binding sites of the adsorbent as well as the degree of ionization process of
the dye molecule [54]. Therefore, setting up the pH of
the solution is critical to obtain a suitable surface
charge for the functional hydroxyl groups of adsorbent
that interact with cations/anions of dye [55]. At lower
pH, the reactive azo dyes such as anthraquinonic
RBBR dissolves and releases colored negatively
charged dye anions into aqueous solution, which will
exhibit electrostatic attraction towards positively
charged surfaces. Moreover, at acidic pH values
some functional groups of adsorbent are also protonated. The dissociated anions of dye molecules are
transferred from solution to the surface of adsorbent
and adsorption occurs via the electrostatic interactions between the ions of negatively charged dye
molecules and positively charged absorbent surface.
In case of RBBR, the adsorption must be due to the
interaction of sulfonic (–SO3−) groups of dyes with –
OH groups on the surface of the adsorbent [56-58].
At the first stage, an initial kinetic experiment
was conducted in which the optimum adsorption time
(equilibrium time) was determined by using 100 mg L-1
dye solution for 120 min. Amount of the adsorbed
dyestuff was determined by using a UV-Vis equipment (Scinco S-3100) at 593 nm (λmax) which is the
maximum absorbance wavelength value of RBBR at
the adsorption process. In the batch adsorption experiments, 40 mg adsorbent was thoroughly mixed with
the 5 ml aqueous solution of dye in a sealed conical
centrifuge tube placed into a tube rack. This rack was
placed in a thermostated water bath shaker at a
rolling speed of 200 rpm at a constant temperature of
25 °C for 60 min (equilibrium time). At the end of the
equilibrium period, the supernatant parts of the tubes
were subsequently analyzed for the residual concentration of RBBR by using UV/Vis spectrophotometer.
The amount of the remained RBBR was determined
from the calibration curve obtained according to concentration-absorbance chart of starting standard sol-
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
utions. Finally, the amount (mg) of dyestuff absorbed
by per unit weight (g) of the adsorbents (SWS, AC,
ASWS and CSWS) was calculated via the following
equation:
Q=
(c 0 –
c e )V
W
(1)
where c0 and ce (mg L-1) are the amount of initial and
remaining RBBR in the solution at time of equilibrium
respectively, V is the volume (L) of the solution, and
W is the weight (g) of the adsorbent.
Besides, attempts were made to fit these kinetic
data by employing the pseudo-first-order [59] and
pseudo-second-order [60] models, as expressed by
Eqs. (2) and (3), respectively:
Pseudo-first-order equation:
ln(Q e − Qt ) = lnQ e − (k 1t )
(2)
Pseudo-second-order equation:
1
t
t
=
+
2
Qt k 2Q e Q e
(3)
where Qe and Qt are the amounts (mg g-1) of RBBR
adsorbed onto the adsorbents at the equilibrium and
at the time of t, respectively, while k1 and k2 are the
kinetic rate constants for the pseudo-first-order (min-1)
and the pseudo-second-order (g mg-1 min-1) adsorption processes, respectively.
After determination of Q values for all solutions,
in order to develop the quantitative data further, the
experimental data obtained for the equilibrium adsorption of RBBR onto the adsorbents were analyzed
employing the Freundlich [61] and Langmuir [62]
isotherms using Eqs. (4) and (5) given below:
Freundlich isotherm equation:
logQ e = log K F +
1
n
log c e
(4)
Langmuir isotherm equation:
 1
= 
Q e  Qmaxb
1
 1
 
  ce

1
 +
 Qmax
(5)
where Qe (mg g-1) and ce (mg L-1) are the equilibrium
concentrations of RBBR dye in the solid and liquid
phases, respectively, while KF ((mg g-1) (mg L-1)-1/n)
and n are the Freundlich constants related to the adsorption capacity and intensity, respectively. Similarly,
Qmax (mg g-1) and b (L g-1) are the Langmuir constants
related to the adsorption capacity. The values of the
all results were given as average of the three identical
trials for each experiment. Uncertainties were pre-
sented as standard deviations in triplicate for each
experiment in the related charts.
RESULTS AND DISCUSSION
Figure 1 and Table 2 present thermal analysis
and obtained data from curves of TGA and DTG for
the SWS, CSWS and ASWS samples. The graph
shows the weight loss of sewage sludge in percentage based on the starting weight with the increasing
pyrolysis temperature. As can be seen from Figure 1,
weight loss of raw sludge and activated sewage
sludge was separated into three discrete stages (start
of decomposition, main decomposition and final
decomposition) in the 20-1000 °C temperature range.
However, carbonized sewage sludge showed
different decomposition characteristics compared to
the other two samples; it had a decomposition range
that lacked exact boundaries, with the exception of
dehydration and drying process (20-150 °C).
For the sewage sludge, the first step roughly
occurred between 20-150 °C with a weight loss of
3.77%, which was caused by the evaporation of adsorbed and bound water molecules in the structure of
sludge. This range was also very similar to activated
and carbonized sludge; however, ASWS had a significant mass loss of 18.08% in this range, which
could be attributed to high moisture adsorption capacity due to its porous structure. In the second step, a
weight loss of 25.21% occurred between 170-600 °C
for SWS. This important weight loss was explained
due to decomposition of volatile organic components
such as carbohydrates, proteins and lipids that constitutes a significant part of sewage sludge. Finally, a
weight loss of 6.75% was observed at the third step
between 600-760 °C. This loss was related with thermal decomposition of inorganic compounds such as
ash, which is another important ingredient in the composition of sewage sludge.
When all three samples were compared in terms
of residual mass, CSWS had the maximum residue of
76.52%, followed by SWS (59.45%) and ASWS
(35.32%). This could indicate that after the carbonization process, the structure of the sludge became
stricter, and weakly bonded chemical components
were removed from structure, eventually leading to a
lower mass loss in CSWS compared to the other
samples. On the other hand, the chemical activation
process provided loosening of the structure and this
caused easy removal of the components from the
sludge structure. According to the TGA results, temperature of 550 °C was determined as optimal for the
production of adsorbents from sewage sludge.
171
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
CI&CEQ 22 (2) 167−179 (2016)
Figure 1. TGA (a) and DTG (b) curves of SWS, CSWS and ASWS.
Table 2. Characteristic parameters obtained from TGA and DTG curves for basic degradation stages
Sample
Parameter
SWS
CSWS
ASWS
DTG
a
Temperature range , °C
170-600
600-760
340-500
500-620
620-930
320-700
700-950
338.95
716.38
456.35
579.63
813.89
527.26
839.71
DTGmaxc / % min-1
1.85
1.39
0.24
0.43
0.63
4.78
1.07
WLd / %
25.21
6.75
2.14
2.87
10.06
33.81
10.50
Tmaxb / °C
TGA
DSPe / °C
170
300
320
-
-
592.74
59.45
76.52
HTf / °C
Rg / %
a
b
c
35.32
d
e
Not included dehydration temperature range; temperature of the maximum weight loss rate; maximum weight loss rate; weight loss; temperature of
f
g
basic degradation start point; half-life temperature; residue remained after analysis
Figure 2 shows the FT-IR spectrum of sewage
sludge (SWS), carbonized sewage sludge (CSWS)
and activated sewage sludge (ASWS) and commercial activated carbon (AC). As can be seen in Figure
2, SWS and CSWS are very alike with the exception
of a few peaks at around 2900 and 1500 cm-1, and
between 650 and 800 cm-1 wavenumber, due to ther-
172
mal destruction of some chemical functional groups.
The broad band between 3100 and 3600 cm-1 was
associated with O–H stretching, which were hydroxyl-containing compounds such as water and alcohol for
all samples. Moreover, it could be related with N-H
stretching vibrations of amine- and amide-containing
compounds.
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
Figure 2. FT-IR spectra of SWS, CSWS, ASWS and AC.
In the spectra of SWS, the most characteristic
bands at 2851 and 2922 cm-1 were due to stretching
vibrations of aliphatic C–H bonds. These bonds could
be ascribed to methylene groups and lipid compounds of SWS. These bands disappeared after thermal and chemical activation process of sewage
sludge. The medium broad band at 1633 cm-1 related
to C=O (Amide I band) stretching vibration of carbonyl
groups originated from carboxylic acids and its derivatives or C=C stretching resulting from alkene compounds of SWS. It was very obvious from Figure 2
that this broad band quite shifted toward a lower
wavenumber of 1620 and 1584 cm-1 for CSWS and
ASWS, respectively. This situation could be caused
by some changes in the functional groups due to thermal and chemical interactions during adsorbent production processes.
The week absorption peak at 1520 cm-1 showed
the presence of protein content (Amide II band) of
SWS like the Amide I band. It is clear from Figure 2
that the N–H bending derived band disappeared in the
spectra of CSWS and ASWS. At 1422 cm-1 wavenumber, medium broad band was probably caused by
C–H and carboxyl O–H vibrations because of amine
and hydroxyl compounds. This band completely disappeared after activation of SWS in the presence of
ZnCl2. In the fingerprint region, the strong narrow
bands between 1000 and 1100 cm-1 could indicate two
different chemical groups. One of them was C–O
stretching vibrations of alcohols and phenol compounds (aliphatic ether), and the other one was Si–O–C
or Si–O–Si stretching vibrations, which related to silicon components in sewage sludge and adsorbents.
Finally, for all samples the peaks in the 700-900 cm-1
wavenumber range in the spectra of the samples
were attributed to an aromatic C–H stretching vibration
indicating the presence of adjacent aromatic hydrogens. On the other hand, the spectra of AC showed
that AC had a homogenous chemical structure due to
its high content of carbon atom comparing the other
samples.
As a result of abundant organic and inorganic
contents of sewage sludge, FTIR spectra of raw material and adsorbents demonstrated the presence of
various functional groups such as aliphatic, carboxylic, aromatic, silicon and phosphorous which are
based on various components of bacterial residues,
carbohydrates, proteins, lipids and ash.
Figure 3 presents the micrographs of the surface properties of SWS, CSWS, ASWS and AC. The
structure of raw sludge (Figure 3a) was dense and
there were almost no pores on it. The structure of raw
material became significantly different due to the
decomposition of organic matters after the carbonization and activation process. Commercial activated
carbon has a more porous structure and higher surface area than other sewage sludge based adsorbents.
It can be seen easily from the micrographs that
the surface of carbonized product (Figure 3b) was
collapsed and some pores appeared. On the other
hand, after the chemical activation process, pores in
various shapes and sizes (Figure 3c) were formed
due to some ZnCl2 evaporation from carbon skeleton
and interaction between ZnCl2 and other compounds
(e.g., H2O) of sludge during the activation process.
Moreover, opening of the closed gaps previously
occupied by ZnCl2 also contributed to porous structure subsequent to the washing process. On the other
hand, on the surface of commercial activated carbon
173
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
pores were more distributed regularly, and pore
volumes and the number of pores were higher than
those of activated sewage sludge (Figure 3d).
Figure 3. SEM images of SWS (a), CSWS (b), ASWS (c) and
AC (d).
The BET analysis strongly supported the SEM
micrographs. According to the BET results, the surface area of the materials increased considerably
from SWS to ASWS. BET surface area values of
SWS, CSWS, ASWS and AC were 0.94, 19.2, 348.93
and 899.95 m2 g-1, respectively. It was very clear that
chemical activation method was the best option for
the production of adsorbents with high surface area.
The value of 348.93 m2 g-1 was very notable when it
compared with other studies in the literature. So far,
many investigations on sewage sludge based adsorbents features different BET values in a wide range of
7–1705 m2 g-1 [37].
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
Before the batch adsorption experiments, in
order to determine the optimum adsorption time and
adsorption kinetics, the effect of contact time on adsorption was evaluated at 25 °C by using 100 mg L-1
dye solution. As can be seen in Figure 4, RBBR
uptake increased rapidly at the beginning of adsorption process with the increasing contact time and
thereafter reached to the equilibrium. The data of the
SWS were not included in the graph due to its very
low value that restricted the comparison of other adsorbents. The adsorption trends were very similar for
AC and ASWS, along with SWS and CSWS, separately. AC showed the best adsorption performance
while SWS showed worst performance among the
four samples.
The minimum contact time required to reach to
the equilibrium was found as 60 min for both of sewage sludge based adsorbents similarly to commercial
activated carbon. Thus, the contact time was fixed at
this value to make sure that equilibrium was established for the other batch adsorption experiments.
Figure 4 indicated that the extent of adsorption increased up to a particular value with increasing time,
after which no further increase occurred for the adsorption of RBBR. Furthermore, when adsorbents were
saturated by RBBR after about 70 min, a desorption
process started.
Kinetic parameters calculated from Figure 4 for
the adsorption of RBBR onto adsorbents are listed in
Table 3. The results showed that the adsorption process better fitted to the pseudo-second order kinetic
model than the pseudo-first order kinetic model for all
adsorbents samples. The values of Qe agreed well
with the experimental data (Qexp) for the pseudo-second order kinetic model. The correlation coefficients
of pseudo-second-order kinetic model were also
higher than that of the pseudo-first order kinetic
Figure 4. Adsorption capacities of RBBR onto the adsorbents.
174
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
Table 3. Pseudo-first-order, Pseudo-second-order kinetic models, Freundlich and Langmuir isotherms constants and correlation coef–
ficients
Adsorbent
Qe(exp) / mg g-1
Pseudo-first-order
Qcal / mg g
-1
k1 / min
Pseudo-second-order
-1
2
R
Qcal / mg g
-1
k2 / g (mg min)-1
R2
SWS
0.557
0.124
0.006
0.127
0.590
0.287
0.984
AC
11.561
0.294
0.017
0.609
11.534
0.515
0.999
CSWS
7.413
0.601
0.035
0.947
7.451
0.199
0.999
ASWS
9.376
1.390
0.017
0.675
9.319
0.343
0.999
Freundlich constants
Kf / mg g-1
Langmuir constants
n
R2
Qmax / mg g
-1
b×103 / L g-1
R2
SWS
0.019
1.51
0.870
0.87
6.76
0.964
AC
1.469
0.998
0.866
109.89
13.82
0.923
CSWS
0.138
0.947
0.973
34.60
4.27
0.978
ASWS
0.296
0.962
0.974
72.99
4.20
0.984
model. As can be seen from Table 3, Qe values from
the pseudo-first order kinetic model were not in
agreement with experimental data although the correlation coefficient of CSWS was quite close the value
of 1, therefore the adsorption process did not comply
with this model. According to these results, adsorption
of RBBR onto adsorbents was based on chemisorption mechanisms with a rate-limiting exchange
reaction step, which controlled by adsorption process
due to inter-particular diffusion [63,64].
Figure 5 shows the amount of adsorbed RBBR
onto the adsorbents at equilibrium. Experimental data
was plotted as a function of the equilibrium concentration of the standard dye solutions. The data of the
SWS were not included in the graph due to its very
low value that restricted the comparison of other adsorbents. Adsorption capacity of ASWS increased
quite linearly with increasing initial concentration of
12
RBBR at the equilibrium time similarly to AC. However, unlike ASWS, the adsorbed amount of dye for
CSWS increased with lower slope after initial dye
concentration of 60 mg L-1. The reason of this situation was due to limited pore structure devoid of
mesopore and micropore formations and with a low
surface area of CSWS.
According to the batch experimental results,
both of the sewage sludge based adsorbents had a
maximum dye adsorption capacity of 4.95 (SWS),
54.4 (CSWS), 71.14 (ASWS) and 92.3% (AC) at 20
mg L-1 dye concentration. These values were slightly
higher than other initial dye concentrations because
of rapidly increasing pore saturation with the increasing concentrations at the same conditions. On the
contrary, adsorbed percentage of dye generally increased with the increasing concentrations of RBBR
and, the adsorption capacities of SWS, CSWS,
100 ppm
10
80 ppm
100 ppm
Qe (mg g-1 )
8
60 ppm
100 ppm
80 ppm
6
60 ppm
40 ppm
4
80 ppm
40 ppm
20 ppm
2
20 ppm
CSWS
60 ppm
ASWS
40 ppm
AC
20 ppm
0
0
5
10
15
20
25
30
35
40
45
Ce (mg L -1 )
Figure 5. Absorbed amount of the dye stuff by per unit weight of the adsorbents at different equilibrium
of standard solution concentrations.
175
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
ASWS and AC were found as 3.46, 59.42, 74.65 and
92.11% at initial dye concentration of 100 mg L-1,
respectively.
Adsorption isotherms are very important for the
design of adsorption systems since they represent
how the dye molecules are partitioned between the
adsorbent and liquid phase at equilibrium as a function of concentrations. In this study, obtained equilibrium data for the adsorption of RBBR onto adsorbents
were analyzed by considering the Freundlich [61] and
Langmuir [62] model equations and isotherms. It was
observed from the isotherm charts that both models
indicated a decent representation of the experimental
results by linear Langmuir or Freundlich isotherm
equations. However, according to the regression
coefficients (R2), Langmuir isotherm provided better
conformity to the equilibrium data than that of Freundlich isotherm. Freundlich and Langmuir isotherm constants were also listed in Table 3 for the batch adsorption experiments along with kinetic parameters.
As can be seen from Table 3, Kf and Qmax values
of ASWS indicated that the adsorption capacity were
found higher than that of CSWS. Moreover, these
values were significantly close to the experimental
values. Furthermore, n and b values denoted that
ASWS had higher sorption intensity and affinity
energy than CSWS. Based on the n values ranging
between 1 and 10 (very close to the value of 1, with
the exception of SWS), it can be concluded that the
adsorption process was very favorable. The results
indicated that AC showed great consistency with both
Freundlich and Langmuir isotherms with a correlation
coefficient of 0.866 and 0.923, respectively. Moreover, it had the highest Qmax value of 109.89 mg g-1
for Langmuir among the four samples. The isotherm
plots also showed that the Langmuir isotherm provided the best fit with a correlation coefficient of 0.978
for CSWS. On the other hand, the correlation value of
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
the Freundlich isotherm (0.973) was quite close to
this value. Therefore, the adsorption characteristics of
CSWS could be explained with both Langmuir and
Freundlich isotherms, as homogeneous surface and
monomolecular or heterogeneous surface adsorption
mechanisms, respectively.
On the other hand, it was very clear from Table
3 that the adsorption characteristics of ASWS were in
accordance with both Freundlich and Langmuir isotherms due to very close regression coefficients.
Therefore, it could be inferred that the adsorption
characteristics of ASWS was a heterogeneous surface system with the interaction between adsorbed
molecules and exponential distribution of active centers in the adsorbent. Negatively charged reactive
RBBR compounds were affected with an electrostatic
attraction towards positively charged adsorbent surface and thereby above-mentioned adsorption phenomena occurred.
However, if a system is explained with the
Langmuir isotherm, it is very important to determine
the dimensionless separation factor [65] called the
equilibrium parameter (RL), represented by Eq. (6):
RL =
1
1 + bc 0
where b (mg L-1) is the Langmuir constant and c0 (mg
L-1) is the initial dye concentration. The value of RL
shows the kind of the Langmuir isotherm to be unfavorable (RL > 1), linear (RL = 1), favorable (0 < RL < 1)
or irreversible (RL = 0). Figure 6 presents the RL
values of the adsorption process at different initial dye
concentrations for all adsorbents. As can be seen in
Figure 6, all RL values were between 0 and 1 and
decreased with increasing dye concentration from
one to zero. These values demonstrated that the
adsorption behavior of RBBR dye onto all adsorbents
Figure 6. Dimensionless separation factor of RBBR dye adsorption onto adsorbents.
176
(6)
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
was significantly favorable and increasingly irreversible with increasing concentration of adsorbate.
Table 4 presents a summary of related studies
about RBBR adsorption onto different adsorbents and
their adsorption performance comparing the results of
this study. Considering the previous studies in Table
4 about the RBBR uptake by activated carbons and
adsorbents obtained from alternative sources, CSWS
and ASWS showed considerable performance.
close performance to commercial activated carbon in
terms of adsorption efficiency. As a result, the present
study proved that sewage sludge, a problematic byproduct of wastewater treatment plant, could be
effectively used in treatment of the same wastewater
streams after it was activated thermally and chemically via pyrolysis in an eco-friendly recycle.
Table 4. Comparison of adsorption capacity of various adsorbents for the adsorption of RBBR
-1
Adsorbent
Adsorption capacity, mg g or %
-1
ZnO fine powder
345 mg g
Sawdust based activated carbon
368.5 mg g
-1
-1
95%; 50 mg L -25 mL dye:0.2 g AC
Jatropha curcas pods based activated carbon
-1
Immobilized Scenedesmus quadricauda
48.3 mg g
Rambutan peel based activated carbon
78.38%;100 mg L -200 mL:0.3 g AC
-1
Peanut hull based activated carbon
149.25 mg g
MgO nanoparticles
166.7 mg g
Polyaniline/bacterial extracellular polysaccharides
composite
Mangosteen peel based activated carbon
Pine cone based activated carbon
-1
-1
361.82 mg g
-1
-1
pH
Ref.
4
[4]
10.7
[16]
3
[37]
2
[44]
-
[45]
-
[46]
8
[47]
3
[48]
80.35%;100 mg L -200 mL dye:0.3 g AC
Natural value
[49]
>98%; Different initial dye concentration:0.1 g AC
2 and 11
[50]
3
[51]
-1
Polyurethane-type foam prepared from peanut shell
5 mg g
Red mud
27.8 mg g
Metal hydroxide sludge
91 mg g
-1
-1
2
[52]
7
[53]
Carbonized sewage sludge
34.60 mg g
-1
3
This study
Activated sewage sludge
72.99 mg g
-1
3
This study
CONCLUSIONS
This study demonstrated that chemically activated adsorbents (ASWS) could be employed as
effective adsorbents for the removal of reactive RBBR
dye from wastewater streams. SEM micrographs of
the adsorbents proved that chemical activation is an
effective way to manufacture useful high porosity adsorbents. Chemical analysis of adsorbents showed
that there were many functional groups that are very
important for the nature of adsorption process on the
surface of adsorbents, and these compounds vary
due to used raw material. The adsorption capacity
calculated from kinetic models was quite closer to the
experimental data, which indicated that the pseudo
first-order was a suitable model to explain the RBBR
adsorption kinetics. On the other hand, R2 values
from the plots of Langmuir and Freundlich models
indicated that the Freundlich isotherm was a more
suitable model to describe the RBBR dye adsorption.
In all cases, ASWS showed better performance than
that of CSWS due to its highly porous structure and
high surface area. Moreover, ASWS exhibited quite
Acknowledgments
This research has been supported by Yıldız
Technical University Scientific Research Projects
Coordination Department. Project Number: 2014-07-04-KAP01.
REFERENCES
[1]
N. Willmott, J. Guthrie, G. Nelson, J. Soc. Dye Colour.
114 (1998) 38-41
[2]
Z. Aksu, A.B. Akın, Chem. Eng. J. 165 (2010) 184-193
[3]
M. Ertaş, B. Acemioğlu, M.H. Alma, M. Usta, J. Hazard.
Mater. 183 (2010) 421-427
[4]
K. Ada, A. Ergene, S. Tan, E. Yalçın, J. Hazard. Mater.
165 (2009) 637-644
[5]
A. Mittal, J. Mittal, A. Malviya, D. Kaur, V.K. Gupta, J.
Colloid. Interf. Sci. 342 (2010) 518-527
[6]
V.K. Gupta, R. Jain, A. Nayak, S. Agarwal, M. Shrivastava, Mater. Sci. Eng. C Mater. Biol. Appl. 31 (2011)
1062-1067
[7]
V.K. Gupta, R. Jain, A. Mittal, T.A. Saleh, A. Nayak, S.
Agarwal, S. Sikarwar, Mater. Sci. Eng. C: Mater. Biol.
Appl. 32 (2012) 12-17
177
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
[8]
A. Mittal, J. Mittal, A. Malviya, V.K. Gupta, J. Colloid
Interface Sci. 340 (2009) 16-26
[38]
M. Seredych, T.J. Bandosz, J. Colloid. Interf. Sci. 302
(2006) 379-388
[9]
S. Karthikeyan, V.K. Gupta, R. Boopathy, A. Titus, G.
Sekaran, J. Mol. Liquids 173 (2012) 153-163
[39]
C.C. Leng, N.G. Pinto, Carbon 35 (1997) 1375-1385
[40]
Q. Wen, C. Li, Z. Cai, W. Zhang, H. Gao, L. Chen, G.
Zenga, X. Shua, Y. Zhaoa, Bioresour. Technol. 102
(2011) 942-947
[10]
M. Iqbal, A. Saeed, Process Biochem. 42 (2007) 1160–1164
[11]
G. Crini, Bioresour. Technol. 97 (2006) 1061-1085
[41]
[12]
A. Mittal, D. Kaur, A. Malviya, J. Mittal, V.K. Gupta, J.
Colloid Interface Sci. 337 (2009) 345-354
H.D. Utomoa, X.C. Onga, S.M.S. Lim, G.C.B. Onga, P. Li,
Int. Biodeter. Biodegr. 85 (2013) 460-465
[42]
[13]
T.A. Saleh, V.K. Gupta, J. Colloid Interface Sci. 371
(2012) 101-106
L.F. Calvo, M. Otero, A. Moran, A.I. Garcia, Bioresour.
Technol. 80 (2001) 143-148
[43]
[14]
T.A. Saleh, V.K. Gupta, Environ. Sci. Pollut. Res. 19
(2012) 1224-1228
P. Sathishkumara, M. Arulkumarb, T.J. Palvannan,
Clean. Prod. 22 (2012) 67-75
[44]
[15]
H. Khani, M.K. Rofouei, P. Arab, V.K. Gupta, Z. Vafaei, J.
Hazard. Mater. 183 (2010) 402-409
A. Ergene, K. Ada, S. Tan, H. Katırcıoğlu, Desalination
249 (2009) 1308-1314
[45]
[16]
K. Vijayaraghavan, S.W. Won, Y-S. Yun, J. Hazard.
Mater. 167 (2009) 790-796
M.A. Ahmad, R. Alrozi, Chem. Eng. J. 168 (2011) 280–285
[46]
[17]
V.K. Gupta, S.K. Srivastava, D. Mohan, S. Sharma,
Waste Manage. 17 (1998) 517-522
Z-Y. Zhong, Q. Yanga, X-M. Li, K. Luo, Y. Liu, G-M.
Zenga, Ind. Crop. Prod. 37 (2012) 178-185
[47]
[18]
V.K. Gupta, S. Agarwal, T.A. Saleh, J. Hazard. Mater.
185 (2011) 17-23
G. Moussavi, M. Mahmoudi, J. Hazard. Mater. 168 (2009)
806–812
[48]
[19]
V.K. Gupta, Suhas, J. Environ. Manage. 90 (2009) 23132342
V. Janaki, K. Vijayaraghavan, A.K. Ramasamy, K.J. Lee,
B.T. Oh, S. Kamala-Kannan, J. Hazard. Mater. 241-242
(2012) 110-117
[20]
I. Ali, V.K. Gupta, Nat. Protoc. 1 (2007) 2661-2667
[49]
[21]
V.K. Gupta, A. Nayak, Chem. Eng. J. 180 (2012) 81-90
M.A. Ahmad, R. Alrozi, Chem. Eng. J. 165 (2010) 883–890
[22]
D. Özçimen, A. Ersoy-Meriçboyu, Adsorpt. Sci. Technol.
28 (2010) 327-340
[50]
Ü. Geçgel, H. Kolancılar, Nat. Prod. Lett. 26 (2012) 659–664
[23]
D. Özçimen, A. Ersoy-Meriçboyu, J. Hazard. Mater. 168
(2009) 1118-1125
[51]
M.H. Bilir, N. Şakalar, B. Acemioğlu, E. Baran, M.H.
Alma, J. Appl. Polym. Sci. 127 (2013) 4340-4351
[24]
D. Özçimen, A. Ersoy-Meriçboyu, Fuel Process Technol.
89 (2008) 1041-1046
[52]
G. Ratnamala, K.V. Shetty, G. Srinikethan, Water Air Soil.
Pollut. 223 (2012) 6187-6199
[25]
V.K. Gupta, B. Gupta, A. Rastogi, S. Agarwal, A. Nayak,
J. Hazard. Mat. 186 (2011) 891-901
[53]
S.C.R. Santos, V.J.P. Vilar, R.A.R. Boaventura, J.
Hazard. Mater. 153 (2008) 999-1008
[26]
N. Graham, X.G. Chen, S. Jeyaseelan, Water Sci. Technol. 43 (2001) 245-252
[54]
L.S. Oliveira, A.S. Franca, T.M. Alves, S.D.F. Rocha, J.
Hazard. Mater. 155 (2008) 507-512
[27]
A. Mittal, J. Mittal, A. Malviya, V.K. Gupta, J. Colloid
Interface Sci. 344 (2010) 497-507
[55]
T.L. Seey, M.J.N.M. Kassim, Int. J. Pure Appl. Sci.
Technol. 9 (2012) 9-19
[28]
A.K. Jain, V.K. Gupta, A. Bhatnagar, Suhas, Separ. Sci.
Technol. 38 (2003) 463-481
[56]
Y. Al-degs, M.A.M. Khraisheh, S.J. Allen, M.N. Ahmad,
Water Res. 34 (2000) 927–935
[29]
W.H. Rulkens, Water Sci. Technol. 49 (2004) 11-19
[57]
[30]
V.K. Gupta, I. Ali, T.A. Saleh, A. Nayak, S. Agarwal, RSC
Adv. 2 (2012) 6380-6388
T.V.N. Padmesh, K. Vijayaraghavan, G. Sekaran, M.
Velan, Dyes Pigments 71 (2006) 77-82
[58]
[31]
I. Fonts, G. Gea, M. Azuara, J. Abrego, J. Arauzo,
Renew. Sust. Energ. Rev. 16 (2012) 2781-2805
N.S. Maurya, A.K. Mittal, P. Cornel, E. Rother, Bioresour,
Technol. 97 (2006) 512-521
[59]
S. Lagergren, Kung. Sven. Veten. Hand. 24 (1898) 1-39
F. Rozada, L.F. Calvo, A.I. Garcıa, J. Martın-Villacorta, M.
Otero, Bioresour. Technol. 87 (2003) 221-230
[60]
Y.S. Ho, G. McKay, Process Biochem. 34 (1999) 451-455
[61]
H. Freundlich, Zeitsch. Physik. Chem 57 (1906) 385-470
I. Langmuir, J. Am. Chem. Soc. 40 (1918)1361-1403
[32]
[33]
W. Rulkens, Energy Fuels 22 (2008) 9-15
[62]
[34]
A.V. Bridgwater, Biomass Bioenerg. 38 (2012) 68-94
[63]
[35]
X. Wang, N. Zhu, B.J. Yin, Hazard. Mater. 153 (2008) 22–27
C.W. Cheung, J.F. Porter, G., McKay, J. Chem. Technol.
Biotechnol. 75 (2000) 963-970
[64]
[36]
M. Otero, F. Rozada, L.F. Calvo, A.I. Garcia, A. Moran,
Dyes Pigments 57 (2003) 55-65
G. McKay, Y.S. Ho, J.C.P. Ng, Sep. Purif. Methods 28
(1999) 87-125
[65]
[37]
K.M. Smith, G.D. Fowler, S. Pullket, N.J.D. Graham,
Water Res. 43 (2009) 2569-2594
K.R. Hall, L.C. Eagleton, A. Acrivos, T. Vermeulen, Ind.
Eng. Chem. Fundam. 5 (1966) 212-223.
178
D. ÖZÇIMEN, T. SALAN: REMOVAL OF REACTIVE DYE…
DIDEM ÖZÇIMEN1
2
TUFAN SALAN
1
Department of Bioengineering,
Faculty of Chemical and
Metallurgical Engineering, Yıldız
Technical University, Davutpasa
Campus, Istanbul, Turkey
2
Department of Material Science
and Engineering, Kahramanmaras
Sutcu Imam University, Avsar
Campus, Kahramanmaras, Turkey
NAUČNI RAD
Chem. Ind. Chem. Eng. Q. 22 (2) 167−179 (2016)
UKLANJANJE REAKTIVNE BOJE REMAZOL
BRILLIANT BLUE R IZ VODENIH RASTVORA
ADSORBENTOM NA BAZI OTPADNOG MULJA IZ
ANAEROBNE DIGESTIJE
U ovom radu analiziran je adsorbent dobijen iz otpadnog mulja hemijskom i termalnom
aktivacijom. Eksperimenti su izvedeni u cevastoj peći pri brzini zagrevanja od na 20 °C
min-1 na temperaturi od 550 °C i sa protokom azota od 400 mL min-1 za 1 h. Šaržni
eksperimenti su izvedeni sa rastvorima boje Remazol Brilliant Blue R različitih koncentracija korišćenjem adsorbenta dobijenog iz otpadnog mulja i komercijalnog aktivnog uglja.
Maksimalni kapaciteti adsorpcije iz rastvora u kojima je koncentracija boje bila 100 mg L1
na karbonizovanom otpadnom mulju (CSWS), aktiviranom otpadnom mulju (ASWS) i
komercijalnom aktivnom uglju su bile: 7,413; 9,376 i 11,561 mg g-1, redom. Za objašnjavanje mehanizma adsorpcije korišćene su Frojndlihova i Lengmirova adsorpciona izoterma i kinetički modeli pseudo prvog i drugog reda. Langmuirova adsorpciona izoterma
koja ima adsorpcione kapacitete od 34,60 mg g-1 za CSWS I 72,99 mg g-1 za ASWS, bolje
fituje eksperimentalne podatke od Frojndlihove izoterme. Model pseudo-drugog reda koji
je imao adsorpcione kapacitete od 7,451 mg g-1 za CSWS i 9.319 mg g-1 za ASWS,
pokazao se jako dobrim u objašnjenju kinetike adsorpcije, sa jako visokim vrednostima
regresionih koeficijenata.
Ključne reči: otpadni mulj, karbonizacija, aktivni ugalj, adsorpcija, Remazol Brilliant Blue R.
179
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 181−189 (2016)
SONJA PECIĆ1,2
NINOSLAV NIKIĆEVIĆ1
MILE VELJOVIĆ1
MILKA JADRANIN3
VELE TEŠEVIĆ4
MIONA BELOVIĆ5
MIOMIR NIKŠIĆ1
1
Department for Food Technology
and Biochemistry, Faculty of
Agriculture, University of Belgrade,
Belgrade, Serbia
2
Economics Institute, University of
Belgrade, Belgrade
3
Institute of Chemistry, Technology
and Metallurgy – Center for
Chemistry, University of Belgrade,
Belgrade, Serbia
4
Department for Instrumental
Methods of Chemical Analysis,
Faculty of Chemistry, University of
Belgrade, Belgrade, Serbia
5
Institute of Food Technology,
University of Novi Sad, Novi Sad,
Serbia
SCIENTIFIC PAPER
UDC 66.061.3:663.8:544:582.284
DOI 10.2298/CICEQ150426033P
CI&CEQ
THE INFLUENCE OF EXTRACTION
PARAMETERS ON PHYSICOCHEMICAL
PROPERTIES OF SPECIAL GRAIN
BRANDIES WITH Ganoderma lucidum
Article Highlights
• Ganoderma lucidum is an interesting raw material in the production of special grain
brandy
• The extraction parameter had important influence on the content of identified triterpenoid acids
• The special grain brandy with G. lucidum shows a considerable antioxidant potential
• The addition of G. lucidum can be an alternative to long-time aging in wooden casks
Abstract
Ganoderma lucidum is one of the five major medicinal mushrooms. In Asian
countries, alcoholic beverages with Ganoderma are traditionally produced and
sold in local markets as a symbol of healthy products. The aim of this study
was to examine the possibility of producing brandy enhanced with this mushroom and to investigate the influence of extraction parameters (time, concentration) on color, total phenolic content, antioxidant capacity, sensory characteristics and the composition and content of triterpenoid acids within the
brandy. HPLC-DAD/ESI-ToF-MS analysis was used to identify triterpenoid
acids. In brandy samples, 15 triterpenoid acids were determined, with the total
content in the range of 2.63-4.06 mg/100 mg. In these samples, the most commonly detected triterpenoid acid was ganoderic acid A. In our study, the total
phenolic content of analyzed samples ranged from 34.07 to 118.1 mg/L GAE.
The color and sensory characteristics of analyzed brandies were significantly
improved in comparison with samples without G. lucidum. The obtained
samples represent an interesting new product for market worldwide with improved antioxidant capacity.
Keywords: Ganoderma lucidum, special grain brandy, triterpenoid
acids, antioxidant capacity, color.
Traditional medicine of the far east countries is
based on using wide diversity of medicinal mushrooms. Ganoderma lucidum (Lingzhi) is one of the
five major medicinal mushrooms, which was marked
as an upper class herb in Sheng Pen Tsao Ching [1].
In nature, the fruit body of this mushroom was very
rare and during the ancient times in the far east it was
available only for high-ranking officials and royal
family [2]. In recent decades, the successful artificial
Correspondence: S. Pecić, Department for Food Technology
and Biochemistry, Faculty of Agriculture, University of Belgrade,
Nemanjina 6, 11080 Belgrade, Serbia.
E-mail: pecic84@hotmail.com; sonja.pecic@ecinst.org
Paper received: 26 April, 2015
Paper revised: 28 July, 2015
Paper accepted: 15 August, 2015
cultivation provided sufficient amount of fungi for
commercial exploitations and production of the different drugs and food supplements. Therefore, the interest for this brilliant mushroom has expanded from
Eastern countries to all around the world, especially in
the Western countries.
The fruit bodies of G. lucidum have woody texture and in food industry are used in different forms,
such as alcohol or water extract, powder, syrup and
liquors [3]. The world observes the continued growth
in usage of all kinds of products made from this mushroom for the promotion of health, but it has also been
used to prevent and treat various diseases, including
some widespread and deadly diseases like cancer,
HIV, hypertension and hepatitis [4]. According to pre-
181
S. PECIĆ et al.: THE INFLUENCE OF EXTRACTION PARAMETERS…
vious research, most of its pharmaceutical activities
were assumed to correlate with its antioxidant activity
[5]. Based on many scientific reports, many compounds from Lingzhi have proven the antioxidant activity in vitro assays and the most important components of Lingzhi with antioxidant effect are phenolics,
triterpenoids and polysaccharides [6,7].
Triterpenoid components isolated from G. lucidum are one of the most important group of bioactive
compounds and they show important medicinal
effects, including anticancer, anti-HIV-1, anti-inflammatory and antioxidant properties [8-13]. The extracts
of Lingzhi’s fruit bodies, mycelia and spores contain
more than 150 highly oxygenated lanostane-type triterpenoids [14]. These oxidized species can be easily
extracted by any organic solvent [15]. Extraction of
Lingzhi’s wooden fruit bodies in alcohol-water solution
can be more successful than extraction in water,
because in some cases more bioactive compounds
can be dissolved [16].
The important characteristic of terpenoids is
their bitter taste of different intensity. Based on the
intensity of bitterness, terpenoids are divided into
three groups: intensely bitter (ganoderic acid A, C1, J;
lucidenic acid A, D1; lucidon A, C), slightly bitter
(ganoderic acid B, C2, K) and very slightly bitter (no
bitter) (ganoderic acid D; lucidenic acid B,C, E1, G, H;
ganolucidic acid C, D; lucidon B) [17].
Spirits are alcoholic beverages with content of
ethanol over 15-20 vol.%, produced by distillation
from fermented agricultural products containing carbohydrates [18]. According to the Regulation on categories, quality and labeling of brandy and other
alcohol spirits of the Republic of Serbia, grain brandy
is produced by the distillation of a fermented mash of
cereals; these spirits have to contain at least 37.5 vol.
% of ethanol [19]. In the process of spirits production,
distillation is the main step by which the volatile compounds are partially separated. The new product is
colorless and often characterized by a raw, unharmonious taste and odor [20]. Distilled beverages have a
negligible amount of biologically active compounds.
Their composition and biological activity could be
improved by maturation in wooden barrels or by the
addition of herbs.
In Asian countries, G. lucidum is traditionally
used as a raw material for the production of alcohol
Chem. Ind. Chem. Eng. Q. 22 (2) 181−189 (2016)
beverages which are sold in local markets as a symbol of healthy products. The main aim of the addition
of G. lucidum is to improve the functional properties of
beverages, but moreover to have additional important
effects on sensory characteristics. The Japanese
sake beverage is manufactured with the addition of
Ganoderma extract or its flavor [21]. G. lucidum have
been used as raw material in the production of bitter
liqueur “Bitter 55” (Ganoderma bitter) [22]. A study
has found that the addition of G. lucidum improved
both sensory and functional characteristics of traditional Korean rice wine yakja, and also affected its
color [23].
The aim of this study was to examine the possibility of producing the special brandy with G. lucidum and to investigate the influence of extraction
parameters (time, concentration) on color, total phenol content, antioxidant capacity, sensory characteristic and the composition and content of triterpenoid
acids of obtained brandy.
EXPERIMENTAL
G. lucidum was isolated from the collection of
the Department of Microbiology, Faculty of Agriculture, University of Belgrade, Serbia. Grain brandy
used in experiment for the production of the special
brandies with G. lucidum was obtained from local
homemade manufacture. Air-dried fruit bodies of fungi
G. lucidum were cut in to pieces (about 1 cm) and
mixed with 45 vol.% alcohol medium (grain alcohol).
Extraction was performed using shaker in dark place
at room temperature for 7, 21, and 60 days with three
different concentrations of mushroom: 10, 25 and 40
g/L (Table 1). After the extractions, the solutions were
filtered and the samples of special brandies were
stored in glass bottles in dark place at room temperature. All samples were made in triplicate.
HPLC-DAD/ESI-ToF-MS analysis of special grain
brandy samples
The analyzed samples (100 mL) were vacuum-evaporated (45 °C) to obtain the desired volume (∼10
mL) and then lyophilizates were dissolved in methanol to the concentration of 10000 mg/mL.
HPLC-DAD/ESI-ToF-MS analyses were carried
out on an Agilent 1200 series HPLC system (Agilent
Table 1. The labels for various combinations of special brandy samples with G. lucidum
Factor
Sample
1
2
3
4
5
6
7
8
9
Concentration of Ganoderma lucidum, g/L
10
10
10
25
25
25
40
40
40
Time of extraction, days
7
21
60
7
21
60
7
21
60
182
S. PECIĆ et al.: THE INFLUENCE OF EXTRACTION PARAMETERS…
Technologies, Waldbronn, Germany) equipped with a
degasser, a binary pump, an autosampler, a termostated column compartment and a diode array detector (DAD) coupled with a 6210 time-of-flight LC/MS
system (Agilent Technologies, Santa Clara, CA, USA)
via an electrospray ionization (ESI) interface. The
chromatographic separation was achieved on a Zorbax EclipsePlus C18 column (100 mm×2.1 mm i. d.;
1.8 μm). The mobile phase consisted of water containing 0.2% formic acid (A) and acetonitrile (B). A
combination of isocratic and linear gradient modes of
elution was applied as follows: 0-2 min 20% B, 2-30
min 20-95% B, 30-35 min 95% B, 35-36 min 95-20%
B, 36-40 min 20% B. The mobile phase flow-rate was
0.40 mL/min, the column temperature was set at 40
°C, and the injection volume was 2 μL. The spectral
data were accumulated in the range of 190-450 nm,
and representative chromatograms were recorded at
254 nm. The HPLC effluent was directed into the
atmospheric pressure ESI ion source of the mass
spectrometer. The eluted compounds were mixed
with nitrogen in the heated nebulizer interface and
negatively charged ions were obtained by applying
following ES parameters: capillary voltage, 4000 V;
gas temperature, 350 °C, drying gas (N2) flow, 12
L/min, nebulizer pressure, 45 psig (310.26 Pa), fragmentor voltage 140 V, and masses were measured in
the range of 100-1500 m/z. A personal computer
system running MassHunter Workstation software
was used for data acquisition and processing. The
molecular feature extractor of MassHunter Workstation was used to predict chemical formulas.
To confirm the identity of compounds for which
molecular formula was calculated from measured
high-accurate masses, an HPLC-DAD/ESI-MS-MS
experiment was performed on a Waters TQ (Tandem
Quadrupole) instrument coupled with a Waters
Acquity UPLC H-Class HPLC system. The HPLC system consisted of a quaternary pump (Waters Quaternary Solvent Manager), an injector (Waters Sample
Manager-FTN), and a photodiode array detector
(Waters 2998 PDA). The HPLC conditions were the
same as those for HPLC-DAD/ESI-ToF-MS. The
HPLC effluent was introduced into the ESI ion source
of the mass spectrometer without splitting. The detector was operated at low resolution in the full scan
mode under the following MS conditions: negative ion
mode; capillary voltage, 3000 V; cone voltage, 25 V;
source temperature 120 °C; desolvation temperature,
250 °C; drying gas (N2) flow, 500 L/h; scan range,
100-1500 m/z. The ultrahigh purity argon was used as
the collision gas for collision induced dissociation
(CID) experiments, and the collision energy was set
Chem. Ind. Chem. Eng. Q. 22 (2) 181−189 (2016)
at 20 eV. In the MRM (multiple reaction monitoring)
mode, the characteristic transitions of the deprotonated ([M-H]–) and/or [M-H-H2O]– molecular ions of
components (Table 2) were measured. The data
acquisition and analysis were performed by MassLynx
V4.1 software.
Above mentioned HPLC-DAD/ESI-MS-MS experiment in scan mode was used for estimation of
amounts of components. Namely, the amounts of
compounds were estimated by comparing the peak
areas obtained for the particular component (1-15)
with the peak area obtained for the internal standard
(cholic acid).
Determination of total phenolics
Determination of total phenolic content (TPC) in
the samples of grain brandy with G. lucidum was
conducted by the Folin-Ciocalteu method described
by Singleton and Rossi [24].
DPPH radical scavenging activity
DPPH-reducing activity was evaluated following
the modified procedure described by Kaneda et al.
[25]. The analyzed samples (0.2 mL, diluted in different ration with 96 vol.% ethanol) were added to the
DPPH working solution (2.8 mL) (mixture of 0.186×10-4
mmol/L DPPH in ethanol and 0.1 M acetate buffer (pH
4.3) in ratio 2:1). The absorbance at 525 nm was
measured after 90 min of incubation in the dark.
DPPH reagent and distilled water were used as a
blank reference. The Trolox calibration curve was
plotted as a function of the inhibition percentage of
DPPH radical. The results were expressed as mM of
Trolox equivalents per liter of brandy. The percentage
of DPPH inhibition was calculated by following
equation:
Activity (% of DPPH reduction) = 100
A − As
A
(1)
where A is the absorbance of DPPH solution with
ethanol and As-absorbance of a DPPH solution with
sample. All experiments were performed in triplicate.
FRAP assay
FRAP assay was performed according to the
procedure by Benzie and Strain [26].
Color determination
Color determination of distillate beverages was
performed according to the AOAC method [27]. The
standard curve was produced with solutions of
K2Cr2O7 (0.05-0.5 g/L) in 0.005 M H2SO4, whose
absorbance was determined at 430 nm and expressed in color units (CU), ranging from 1 to 10.
183
S. PECIĆ et al.: THE INFLUENCE OF EXTRACTION PARAMETERS…
CI&CEQ 22 (2) 181−189 (2016)
Table 2. The contents of triterpenoid acids in special grain brandy samples (mg/100mg)
No
tR / min
DAD/MS
1
8.00/8.08
Content of components in the liophilizate of the sample, mg/100 mg
G1
G2
G3
G4
G5
G6
G7
G8
G9
Molecular Mass and
MRM
Compound
formula m/z (meaTransition
nm
and mass
sured)
λmax
0.2379 0.1893 0.2115 0.2115 0.2621 0.2586 0.2483 0.0690 0.2345 256 C30H46O7 518.3244; 517→499 Ganoderic
(518.3087) 517.3168;
acid C2
563.3232;
1035.6402
2
8.36/8.45
0.1212 0.1275 0.0866 0.1113 0.1545 0.1841 0.1427 0.2004 0.1033 256 C27H40O6 460.2824; 459→441
(460.2825) 459.2750; 459→385
Lucidenic
acid LM1
505.2812;
919.5583
3
8.51/8.59
0.0890 0.1210 0.1062 0.0938 0.1157 0.1333 0.0993 0.1010 0.0897 254 C30H42O8 530.2879 529→511 Ganoderic
(530.2879) 529.2806 511→467 acid C 6
575.2862
4
8.83/8.92
0.3525 0.3836 0.4092 0.3495 0.4258 0.4551 0.4382 0.0877 0.4316 256 C30H44O8 532.2927; 531→513 Ganoderic
(532.3036) 531.2851; 513→469
acid G
1063.583
5
9.10/9.19
0.2155 0.2070 0.2951 0.2576 0.3214 0.2948 0.2991 0.2160 0.2829 254 C30H44O7 516.3087; 515→497 Ganoderic
(516.3087) 515.3016; 497→453
acid B
561.3054;
1031.6068
6
9.24/9.33
7
9.55/9.65
0.2441 0.2789 0.2553 0.2246 0.3034 0.3306 0.2586 0.2258 0.2431 258 C29H40O8 516.2810; 515→473
(516.2723) 515.2738
Lucidenic
acid E
0.2540 0.2525 0.2407 0.2966 0.3205 0.3212 0.3070 0.2330 0.0330 254 C30H40O8 528.2723 527→509
(528.2723) 527.2651 509→465
Elfvingic
acid A
573.2757
8
9.93/10.04
0.5480 0.5433 0.6177 0.5852 0.7813 0.6589 0.6664 0.5285 0.6340 256 C30H44O7 516.3088; 515→497 Ganoderic
(516.3087) 515.3015;
acid A
561.3073;
1031.6099
9
10.52/10.63
0.2195 0.1226 0.1384 0.2448 0.2490 0.2466 0.2497 0.1722 0.2375 254 C27H38O6 458.2669; 457→439
(458.2668) 457.2595; 457→287
Lucidenic
acid A
503.2652;
915.5264
0.0896 0.0677 0.0639 0.0913 0.0887 0.1128 0.1047 0.0238 0.0870 254 C30H42O8 530.2877; 529→511 12-Hydroxy(530.2879) 529.2804; 511→467 -ganoderic
10.63/10.74
acid D
575.2854;
10
1059.5595
0.2141 0.1874 0.2181 0.2959 0.3049 0.2682 0.3051 0.2186 0.3075 254 C30H42O7 514.2930; 513→495 Ganoderic
(514.2930) 513.2854; 495→451
acid D
11.11/11.22
11
559.2932;
1027.5754
12
11.29/11.40
13
11.40/11.52
0.2211 0.1538 0.1820 0.1882 0.2106 0.2264 0.2227 0.1733 0.1960 254 C29H38O8 514.2627; 513→471
(514.2566) 513.2554
Lucidenic
acid D2
0.0844 0.0888 0.0886 0.1029 0.0832 0.1011 0.1228 0.0848 0.1298 254 C30H40O7 512.2773; 511→493 Ganoderenic
(512.2774) 511.2697; 493→449
acid D
557.2763;
1023.5455
14
11.94/12.06
0.2613 0.2527 0.3089 0.3291 0.3510 0.3847 0.3277 0.2935 0.3845 254 C32H42O9 570.2829; 569→551 Ganoderic
(570.2829) 569.2757; 551→509
acid F
1139.5577
184
S. PECIĆ et al.: THE INFLUENCE OF EXTRACTION PARAMETERS…
Chem. Ind. Chem. Eng. Q. 22 (2) 181−189 (2016)
Table 2. Continued
No
tR / min
DAD/MS
Content of components in the liophilizate of the sample, mg/100 mg
G1
G2
G3
G4
G5
G6
G7
G8
G9
Molecular Mass and
MRM
Compound
formula m/z (meaTransition
nm
and mass
sured)
λmax
514.2930;
15
12.10/12.23
C30H42O7 513.2854; 513→451 Ganoderic
0.0505 0.0420 0.0477 0.0667 0.0846 0.0722 0.0810 0.0666 0.0715 254
(514.2930) 559.2932; 513→437
acid J
Σ
0.8180 0.7079 0.8038 0.8967 1.1149 0.9777 0.9971 0.7673 0.9430
-
-
-
-
Bitter triterpenoid
acids
Σ
3.2027 3.0181 3.2699 3.449 4.0567 4.40487 3.873 2.6942 3.4659
-
-
-
-
Total triterpenoid
acids
1027.5754
CIEL*a*b* chromatic parameters
Color measurements were performed on brandy
samples using a portable tristimulus Chroma Meter
model CR-400 (Konica Minolta, Osaka, Japan).
Results were expressed in Commission Internationaled’Eclairage L*, a* and b* color space coordinates.
The following parameters were measured: L* (lightness), a* (+a* = redness, -a* = greenness), b* (+b* =
yellowness, -b* = blueness), C* (chroma or saturation) and h (hue angle). CIEL*a*b* parameters were
read using CIE illuminant D65 and the observer angle
at 2°.
Sensory analyses
Sensory characteristics of the brandies enriched
with mushroom were determined using a modified
Buxbaum model of positive ranking. The common
quality parameters were evaluated: clearness, color,
distinction, odor and taste. In this evaluation a brandy
sample may have a maximal score of 20 points [28].
The analysis was conducted by evaluation panel,
made of 5 sensory experts. All evaluation experts had
long tradition in evaluation of alcohol beverages.
Samples were diluted with distilled water to reach an
alcohol proof of 45° (vol.%).
Statistical analyses
The determination of polyphenol, antioxidant
capacity, sensory characteristics and color were done
in triplicate, and data were expressed as mean value
± standard deviation (SD). The experimental data
were subjected to the analysis of variance (ANOVA).
Analysis was conducted in a factorial arrangement
where time extraction and concentration of added G.
lucidum were analyzed factors. Tukey’s test was used
to determine difference (p ≤ 0.01) between the mean
values. Statistical analyses were performed with the
statistical program Statistica 12 [29].
RESULTS AND DISCUSSION
The study was conducted to identify and compare the composition and concentration of triterpenoid
acids in 9 samples of special grain brandies. In the
research following, 15 triterpenoid acids were determined in all samples (Table 2): ganoderic acid (A, B,
C2, C6, D, F, G and J), ganoderenic acid (D), lucidenic acid (A, E, D2 and LM1), 12-hydroxyganoderic
acid D and elfingenic acid A. Based on these results,
the extraction parameters did not effect on the composition of identified triterpenoid acids in special grain
brandies. Previous research reported that the chemical composition of fungi G. lucidum, including triterpenoid acids, depends on the geographical distributions, growth conditions and substrates [30].
However, the extraction parameters had an
important effect on the content of total triterpenoid
acids in analyzed brandy samples (2.63-4.06 mg/100
mg, Table 2). According to the content of terpenoid
acids, the analyzed samples can be ranged as:
G5 > G6 > G7 > G4 > G9 > G3 > G1 > G2 > G8.
Wang et al. [31] estimated the quantity of six major
triterpenoids in 36 Ganoderma samples. The average
content of total triterpenoids in G. lucidum samples
was 0.18-1.15 mg/100 mg. Hence, the content of triterpenoids in brandy samples was higher than in the
analyzed G. lucidum samples.
Ganoderic acid A was the most abundant ganoderic acid in analyzed samples, which has been confirmed in some previous research [32]. The extraction
time influenced ganoderic acid A content; with increasing concentration of added mushroom, the extraction of this compound completed faster. Addition of
the higher amount of mushroom in brandy samples
had limited effect on increasing the ganoderic acid A
content and the highest value was found in sample
G5 with 25 g/L mushroom extracted after 21 days.
185
S. PECIĆ et al.: THE INFLUENCE OF EXTRACTION PARAMETERS…
Total phenolic content and antioxidant properties
Chem. Ind. Chem. Eng. Q. 22 (2) 181−189 (2016)
Two antioxidant assays were carried out to
evaluate the antioxidant characteristics of analyzed
brandy samples: DPPH and FRAP assay. Our results
showed a considerable antioxidant potential of analyzed brandy samples, which strongly depends on the
concentration of added fungi. The correlation between
TPC and antioxidant capacity was very high and presented in Table 4, with values r(TPC-FRAP) = 0.9702 and
r(TFC-DPPH) = 0.9618. Phenolic compounds of fungi significantly improve the antioxidant capacity of grain
brandy, which is in correlation with previous research
conducted by Kim et al. [3].
TPC and antioxidant properties were presented
in Table 3. The results of ANOVA showed that the
concentration of mushroom and its extraction time
had very significant influence on TPC of these special
grain brandies (p < 0.01). The interaction fungi concentration×extraction time did not affect the phenolic
content of analyzed samples, which indicates that
these factors affect independently.
Grain brandy used as an alcohol medium for the
production of special grain brandy (5.1 mg/L) had a
low TPC content, which is consistent with all distillated unaged beverages. The phenolic components in
the distilled beverages originated from the wooden
barrels in which they are stored after distillation.
According to the research of Ziyatdinova et al. [33],
regular brandy shows TPC in the range of 59-334
mg/L gallic acid equivalents (GAE) depending on the
type and origin of brandy. In our study, the TPC content of analyzed samples ranged from 34.07 to 118.1
mg/L GAE. It can be concluded that the TPC of these
analyzed samples were also significantly increased
by adding fungi to the alcoholic beverage. The regular
brandies aged for at least two years (3 stars) did not
have a significantly higher TPC then special brandies
G7, G8, G9 made with 40 g/L G. lucidum [33]. Based
on the results, the fungi concentration had a strong
influence on the TPC. Also, extraction time did not
have a considerable effect on the parameters investigated in the following samples: 25 (G4, G5, G6) and
40 g/L (G7, G8, G9). This outcome indicates that the
extraction process was completed within seven days.
The only exception of this behavior was the samples
G1, G2 and G3 with 10 g/L of mushroom. Based on
the results, extraction of phenolics was not finished
after 60 days for these samples.
Table 4. Correlation between TPC, antioxidant characteristics
and color intensity
FRAP
TPC
FRAP
DPPH
Color
r
p
r
p
r
p
0.9702
0.000
0.9618
0.000
0.9618
0.000
0.000
0.9422
0.000
0.9291
0.000
0.9231
0.000
Color
Color measurement
Color is the most important visual feature, which
creates the first and very important impression among
consumers. Hence, the color of a distillate beverage
is an important characteristic, especially for beverages matured in wooden casks in which the dark
golden color signifies the highest quality of beverage.
According to this research, the compounds extracted
from this mushroom also affect the color of the alcoholic beverage.
The color intensity of grain brandy used as basic
alcohol medium was 1.25 CU. After extraction of components from G. lucidum, the color of the brandy
changed. Based on the results, it can be concluded
that compounds of fungi significantly increased the
color intensity of special brandy samples. The differ-
Table 3. Total phenol content, antioxidant activity and color of special brandy with Ganoderma lucidum; TPC – Total phenol content,
expressed as milligram of gallic acid equivalents per liter of brandy; DPPH – DPPH radical scavenging activity expressed as mmol of
Trolox equivalent; different letters in same row denote a not significant difference according Tuckey’s test, at p ˂ 0.01
Sample
FRAP
DPPH / mmol TE
G1
35.13±0.42
G2
34.07±0.64
a
0.138±0.001
ab
0.330±0.005
ab
4.84±0.00
G3
51.40±6.20
0.151±0.006
ab
0.405±0.034
ab
8.99±0.00
a
G4
77.13±3.01
b
c
8.99±0.01
a
cd
9.32±0.00
0.148±0.008
a
0.296±0.000
c
0.338±0.012
0.867±0.009
G5
84.60±8.23
b
0.311±0.004
cd
G6
88.07±7.64
b
1.014±0.115
0.321±0.012
cd
0.924±0.026
G7
114.60±0.80
c
0.432±0.015
G8
110.90±1.90
c
0.417±0.013
fg
0.976±0.008
118.10±2.30
c
0.438±0.005
fg
1.043±0.031
f
a
CUI / Color units
a
G9
186
Parameter
TPC / mg GAE/L
cdf
1.318±0.037
4.93±0.02
8.48±0.00
14.18±0.00
cdfg
13.11±0.00
dfg
12.62±0.00
S. PECIĆ et al.: THE INFLUENCE OF EXTRACTION PARAMETERS…
ent extraction time did not significantly increase the
color intensity of samples with equal concentration of
fungi. Hence, addition of higher amount of fungi
makes the color more intensive for samples with the
same extraction time. The correlation of TPC and
color was very high (r = 0.9618). It can be concluded
that the color intensity of samples was significantly
correlated with phenolic content.
According to previous research, the color of
plum brandies after 11 years of maturation in sessile
oak and 18 year and in mulberry cask [28] was not
significantly higher than the color value for the sample
of special brandy G7. Therefore, the addition of this
mushroom can be regarded as an innovative process
and can replace the long period of aging in wooden
casks.
The results of CIEL*a*b* method are presented
in Table 5. The L* value decreased with increasing
concentration of added mushroom. The value of parameter a* for samples with 10 g/L (G1, G2, G3) was
significantly different from the samples with higher
concentration (G4-G9), because it was defined with
the light tone of green color, and the other samples
had light tone of red color. The parameter b* for all
samples describes the different intensity of the yellow
color. According to the results for the hue angle, and
Chem. Ind. Chem. Eng. Q. 22 (2) 181−189 (2016)
parameters a* and b*, it can be concluded that all
samples had yellow color, with very small proportion
of red or green color. Based on these results, it can
be concluded that with addition of the higher amount
of G. lucidum, the lightness of samples is reduced.
Also, the addition of higher amount of mushroom
affected the values of a* and b* parameters and
therefore increased the proportion of yellow color and
red color of samples.
Sensory evaluation
The results of sensory evaluation were between
17.78 and 18.12 (Table 6). It is obvious that the
brandy samples enriched with G. lucidum were well
accepted by sensory experts. The results of factorial
ANOVA indicate that the extraction parameters and
their interaction did not have statistically significant
effect on the sensory scores of special brandy
samples. Therefore, the sensory characteristics of
analyzed samples were improved with extracted components of fungi, but the quantity of extracted compounds did not have significant influence on the final
sensory impression.
The triterpenoid acids were the most important
compounds of fungi which define the taste of the
brandy samples. Although the lowest content of total
Table 5. CIElab chromatic parameters of special brandy samples with Ganoderma lucidum; values represent means of triplicate determinations ± standard deviation
L*(D65)
a*(D65)
b*(D65)
C*(D65)
h(D65)
G1
54.84±0.01
-0.55±0.00
20.59±0.02
20.60±0.02
91.54±0.02
G2
54.98±0.01
-0.54±0.02
19.90±0.02
19.91±0.01
91.56±0.06
G3
54.55±0.01
-0.78±0.02
23.41±0.01
23.43±0.01
91.91±0.08
G4
50.07±0.01
2.35±0.01
34.16±0.00
34.24±0.00
86.07±0.01
G5
51.93±0.01
0.86±0.01
28.90±0.02
28.91±0.02
88.29±0.02
G6
50.06±0.00
2.57±0.01
35.76±0.01
35.85±0.01
85.89±0.02
G7
48.22±0.01
4.67±0.03
36.52±0.02
36.82±0.02
82.72±0.05
G8
49.47±0.01
3.46±0.03
35.42±0.01
35.59±0.01
84.42±0.05
G9
47.72±0.01
5.39±0.04
39.18±0.04
39.55±0.03
82.16±0.07
Sample
Table 6. Sensory characteristic of special brandies; values represent means of triplicate determinations ± standard deviation. G-grain brandy
Sample
Assessments characteristics
Color
Clearness
Distinction
Odor
Taste
G1
1
1
2
5.66±0.11
8.48±0.08
G2
1
1
2
5.64±0.11
8.36±0.11
G3
1
1
2
5.50±0.16
8.40±0.07
G4
1
1
2
5.68±0.22
8.34±0.05
G5
1
1
2
5.70±0.16
8.26±0.05
G6
1
1
2
5.62±0.13
8.48±0.08
G7
1
1
2
5.48±0.08
8.54±0.05
G8
1
1
2
5.62±0.13
8.50±0.07
G9
1
1
2
5.38±0.08
8.40±0.07
187
S. PECIĆ et al.: THE INFLUENCE OF EXTRACTION PARAMETERS…
terpenoinds acids was found in the sample G8, the
share of bitter triterpenoids was higher than in other
samples. It can be concluded that the content of bitter
acids had the more important influence on the sensory marks than the content of total triterpenoid acids.
As reported in the previous research, the specific
triterpenoid acid had a different medical effect. The
higher content of bitter terpenoids, particularly ganoderic acid A, also improved the antioxidant capacity,
antinociceptive, anti-inflammatory and antitumor activity of samples [11].
CONCLUSION
Chemical composition and sensory characteristics of grain brandy were changed by the maceration of Ganoderma lucidum due to the transition of
soluble mushroom compounds in an alcohol-water
mixture. Based on the analysis of the content of total
phenolics, antioxidant properties and sensory characteristics, it can be concluded that extracted mushroom
compounds refined the chemical complex. In the production process, the optimal extraction parameters
have to be defined depending on the desired character of products. With the increase of fungi concentration the antioxidant capacity is enhanced, and improves the health characteristics of special brandies,
but the higher amount of fungi did not have an important effect on the sensory evaluation. The extracted
component colored the samples, and therefore the
addition of this mushroom can be regarded as an
innovative process and can be an alternative to the
process of aging in wooden casks over a long period.
The medicinal mushroom G. lucidum may be an interesting raw material in the production of special brandy
with a bitter taste.
Acknowledgement
This work was performed within the framework
of the research projects No. 46001 and 172053 supported by the Ministry of Education, Science and
Technological Development, Republic of Serbia.
Chem. Ind. Chem. Eng. Q. 22 (2) 181−189 (2016)
[5]
A. Sheikh, D. Vyas, M.A. Ganaie, K. Dehariya, V. Singh,
Int. J. Pharm. Pharm. Sci. 6 (2014) 679-684
[6]
S.W. Galor, J. Yuen, A.J. Buswell, F.F. Benzie, in: Herbal
medicine Bimolecular and clinical aspects, J. Barnes,
L.A. Anderson, J.D. Phillipson (Eds.), CRC Press, Boca
Raton, FL, 2011, pp. 175-200
[7]
P. Kalaĉ, in: Mushroom: Type, properties and Nutrition, S.
Andres, N. Braumann (Eds.), Nova Science publisher,
Inc., Hauppauge, NY, 2012, pp. 129-152
[8]
B. Ma, W. Rei, Y. Zhou, J. Ma, Y. Ruan, C.N. Wen. N.
Am. J. Med. Sci. 3 (2011) 495-498
[9]
N.H. Chen, J.W. Liu, J.J. Zhong, J. Pharmacol. Sci. 108
(2008) 212-216
[10]
Q.X. Yue, X.Y. Song, C. Ma, L.X. Feng, S.H. Guan, W.Y.
Wu, M. Yang, B.H. Jiang, X. Liu, Y.J. Cui, D.A. Guo,
Phytomedicine 17 (2010) 606-613
[11]
J.W. Xu, W. Zhao, J.J. Zhong, Appl. Microbiol. Biotechnol. 87 (2010) 457-466
[12]
S.R. El-Mekkaway, M. Meselhy, N. Nakamura, Y. Tezuka, M. Hattori, N. Kakiuchi, K. Shimotohno, T. Kawahata, T. Otake, Phytochemistry 49 (1998) 1651-1657
[13]
B.S. Min, N. Nakamura, K.W. Bac, H. Miyashiro, M.
Hattori, Chem. Pharm. Bull. 46 (1998) 1607-1612.
[14]
T. Akihisa, Y. Nakamura, M. Tagata, H. Takuda, K.
Yasukawa, E. Uchiyama, T. Suzuki, Y. Kimura, Chem.
Biodiversity 4 (2007) 224-231
[15]
Y.H. Gao, W.B. Tang, H. Gao, J. Lam, S.F. Zhou. Int. J.
Med. Mushrooms 6 (2004) 95-106
[16]
M. Nikšić, N. Nikićević, V. Tešević, A. Klaus, Int. J. Med.
Mushrooms 3 (2001) 223
[17]
T. Nishitoba, H. Sato, S. Sakamura, Agric. Biol. Chem. 52
(1988) 1791-1795
[18]
M.L. González-Sanjosé, S.P. Magariňo, in: Encyclopedia
of Food Science and Nutrition, B. Caballero, L.Z. Trugo,
P.M. Finglas (Eds.), Academic Press, Waltham, MA,
2003, pp. 3553-3561
[19]
The Regulation on categories, quality and labeling of
brandy and other alcohol spirits, 73 (2010) 6-8
[20]
N. Christoph, C. Bauer-Christoph, in: Flavour and fragst
rances, Chemistry, Bioprocessing and Sustainability,1 ed.,
R.G. Berger (Eds.), Springer, Berlin, 2004, pp. 219-239
[21]
S.C. Jong, J.M. Birmingham, inAdvanced in applied microbiology,S.I. Neidleman, A.I. Laskin (Eds.), Academic
Press Inc., San Diego, CA, 1992, pp. 101-134
[22]
S. Gorjanović, M. Novaković, P. Vukosavljević, F.Pastor,
V. Tešević, D. Sužnjević, J. Agric. Food Chem. 58 (2010)
8400-8406
[23]
J.H. Kim, D.H. Lee, S.H. Lee, S.Y. Choi, S.J. Lee, J.
Biosci. Biotechnol. 97 (2003) 24-28
[24]
V.L. Singleton, J.A. Rossi, Am. J. Enol. Vitic. 16 (1963)
144–158
[25]
H. Kaneda, N. Kobayashi, S. Furusho, H. Sahara, S.
Koshino, Tech. Q. Master Brew. Assoc. Am. 32 (1995)
90–94
[26]
I.F.F. Benzie, J.J. Strain, Anal. Biochem. 239 (1996) 70–
76
REFERENCES
[1]
H. Lee, H. Itokawa, M. Kozuka, in: Asian functional foods,
J. Shi, C.T. Ho, F. Shahidi (Eds.), Pub- CRC Press, Boca
Raton, FL, 2005, pp. 21-72
[2]
P. Stamets, Growing gourment and medicinal mushrooms, Ten Speed Press, Berkeley, CA, 1993, pp. 355-369
[3]
M. Berovič, J. Habijanič, I. Zore, B. Wraber, D. Hodžar, B.
Boh, F. Pohleven, J. Biotechnol. 103 (2003) 77-86
[4]
R. Peterson, Phytochemistry 67 (2006) 1985-2001
188
S. PECIĆ et al.: THE INFLUENCE OF EXTRACTION PARAMETERS…
[27]
AOAC, Method 992.8, Association of Official Analytical
Chemists, Arlington, VA, 1998
[28]
S. Pecić, M. Veljović, S. Despotović, I. Leskošek-Čukalović, M. Jadranin, V. Tešević, M. Nikšić, N. Nikićević,
Eur. Food Res. Technol. 235 (2012) 479-487
Chem. Ind. Chem. Eng. Q. 22 (2) 181−189 (2016)
[31]
X.M. Wang, M. Yang, S.H. Guan, R.X. Lui, J.M. Xia, K.S.
Bi, D.A. Guo, J. Pharmaceutical. Biomed. 41 (2006) 838–
–844
[32]
Y. Liu, Y. Liu, F. Qui, X. Di, J. Pharmaceutical. Biomed.
Anal. 54 (2010) 717-721
[29]
Statistica (Data Analysis Software System), v. 12,
StatSoft, Inc., Tulsa, OK, 2013
[33]
G. Ziyatdinova, I Salikhova, H. Budnikov, Food Chem.
150 (2014) 80-86
[30]
S. Keypour, H. Rafati, H. Raihi, F. Mirzajani, M.F.
Moradali, Food Chem. 11 (2010) 1704-1708
[34]
M.Y. Kim, P. Seguin, J.K. Ahn, J.J. Kim, S.C. Chun, E.H.
Kim, S.H. Seo, E.Y. Kang, S.L. Kim, Y.J. Park, H.M. Ro,
I.M. Chung, J. Agr. Food Chem. 56 (2008) 7265-7270.
SONJA PECIĆ1,2
NINOSLAV NIKIĆEVIĆ1
MILE VELJOVIĆ1
MILKA JADRANIN3
VELE TEŠEVIĆ4
MIONA BELOVIĆ5
MIOMIR NIKŠIĆ1
1
Institut za Prehrambenu tehnologiju i
biohemiju, Poljoprivredni fakultet,
Univerzitet u Beogradu, Nemanjina 6,
11080 Beograd
2
Ekonomski institut, Kralja Milana 16,
11000 Beograd
3
Institut za hemiju, tehnologiju i
metalurgiju - Centar za hemiju,
Univerzitet u Beogradu, Njegoševa 12,
11000 Beograd
4
Odeljenje za instrumentalne metode
hemijske analize, Hemijski fakultet,
Univerzitet u Beogradu, Studentski trg
16, 11000 Beograd
5
Institut za prehrambene tehnologije,
Univerzitet u Novom Sadu, Bulevar
cara Lazara 1, 21000 Novi Sad
UTICAJ PARAMETARA EKSTRAKCIJE NA
FIZIČKOHEMIJSKE KARAKTERISTIKE
SPECIJALNIH ŽITNIH RAKIJA SA DODATKOM
GLJIVE Ganoderma lucidum
Ganoderma lucidum spada u pet najznačajnijih medicinskih gljiva. U Azijskim zemljama,
alkoholna pića sa dodatkom G. lucidum se tradicionalno proizvode i prodaju u lokalnim
marketima kao simbol zdravih proizvoda. Cilj ove studije bio je ispitivanje mogućnosti
proizvodnje rakija obogaćenih ovom gljivom, kao i proučavanje uticaja parametara ekstrakcije (vremena ekstrakcije i koncentracije gljive) na boju, sadržaj ukupnih fenolnih jedinjenja, antioksidativni kapacitet, senzorne karakteristike i sadržaj triterpenskih kiselina
specijalnih rakija. HPLC-DAD/ESI-ToF-MS metoda je korišćena za identifikaciju triterpenskih kiselina. U uzorcima specijalnih rakija, detektovano je 15 triterpenskih kiselina sa
ukupnim sadržajem od 2,63 do 4,06 mg/100 mg. Ganoderinska kiselina A je bila najzastupljenija triterpenska kiselina u analiziranim uzorcima. Ukupan sadržaj fenola u analiziranim uzorcima se kretao od 34,07 do 118,1 mg/L GAE. Boja i senzorne karakteristike
analiziranih uzoraka specijalnih rakija su značajno poboljšani u poređenju sa uzorcima bez
dodatka G. lucidum. Dobijeni uzorci predstavljaju interesantan novi proizvod sa povećanim
antioksidativnim kapacitetom za tržišta širom sveta.
Ključne reči: Ganoderma lucidum, specijalne žitne rakije, triterpenske kiseline,
antioksidativni kapacitet, boja.
NAUČNI RAD
189
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 191−199 (2016)
MONIKA LUTOVSKA1
VANGELCE MITREVSKI2
IVAN PAVKOV3
VLADIMIR MIJAKOVSKI2
MILIVOJ RADOJČIN3
1
Faculty of Technical Sciences,
University St. Kliment Ohridski,
Bitola, Macedonia
2
Faculty of Technical Sciences,
University “St. Kliment Ohridski”,
Bitola, Macedonia
3
Faculty of Agriculture, University
of Novi Sad, Novi Sad, Serbia
SCIENTIFIC PAPER
UDC 634.13:66.047:543:519.87
DOI 10.2298/CICEQ150122032L
CI&CEQ
MATHEMATICAL MODELLING OF THIN
LAYER DRYING OF PEAR
Article Highlights
• We examined thin-layer drying of pear slices as a function of drying conditions
• The best model was determined numerically
–9
-8
2 -1
• Effective moisture diffusivity values ranged from 6.49×10 to 3.29×10 m s
-1
• The values of activation energy were in the range of 28.15–30.51 kJ mol
Abstract
In this study, thin-layer drying of pear slices as a function of drying conditions
was examined. The experimental data sets of thin-layer drying kinetics at five
drying air temperatures (30, 40, 50, 60 and 70 °C) and three drying air velocities (1, 1.5 and 2 m s-1) were obtained from an experimental setup designed
to emulate industrial convective dryer conditions. Five well-known thin-layer
drying models from scientific literature were used to approximate the experimental data in terms of moisture ratio. The best model was evaluated numerically. For each model and data set, the statistical performance index (φ) and
chi-squared (χ2) value were calculated and models were ranked afterwards.
The performed statistical analysis showed that the model of Midilli gave the
best statistical results. Since the effect of drying air temperature and drying air
velocity on the empirical parameters was not included in the basic Midilli
model, a generalized form of this model was developed. With this model, the
drying kinetic data of pear slices can be approximated with high accuracy. The
effective moisture diffusivity was determined by using Fick’s second law. The
obtained values of the effective moisture diffusivity (Deff) during drying ranged
between 6.49×10-9 and 3.29×10-8 m2 s-1, while the values of activation energy
(E0) varied between 28.15 and 30.51 kJ mol-1.
Keywords: mathematical modelling, thin-layer drying, pear, moisture diffusivity, activation energy.
Fruits play an important role in human diet and
nutrition as sources of vitamins and minerals. Pears
are a good source of dietary fiber, vitamins C and B6,
minerals like magnesium and potassium. The worldwide pear production in 2012 was estimated at
23580845 Mt (http://faostat3.fao.org/download/Q/QC/
/E). With 16266000 Mt, China was the largest producer, followed by the USA (778582 Mt), Argentina
(700000 Mt) and Italy (645540 Mt). Because fresh
pears have very short shelf life, preservation after
harvesting is necessary by using of different proCorrespondence: M. Lutovska, Faculty of Technical Sciences,
University St. Kliment Ohridski, Makedonska Falanga 33, 7000
Bitola, Republic of Macedonia.
E-mail: sahdooel@hotmail.com
Paper received: 22 January, 2015
Paper revised: 4 August, 2015
Paper accepted: 15 August, 2015
cesses such as drying, storing in cold at controlled
microclimatic conditions and canning. The common
processing techniques of pears are conserves in
syrup, purees for use in nectars, yogurts, and drying.
Dried pears can be used in bakery products, gravies,
compotes, and for consumption of the dry fruit [1].
Several drying methods are commercially used to
remove moisture from food products, but convective
hot air drying is the most widely used method. From a
mathematical point of view, the convective drying is a
complex process of simultaneous heat and mass
transfer within dried material and from its surface to
the surroundings caused by a number of transport
mechanisms. There are several different methods of
describing the complex simultaneous heat and moisture transport processes within drying material, but
there is no single theory for wet material drying pre-
191
M. LUTOVSKA et al.: MATHEMATICAL MODELLING OF THIN LAYER…
Chem. Ind. Chem. Eng. Q. 22 (2) 191−199 (2016)
diction that encompasses all transfer mechanisms. In
the approach initially proposed by Philip and De Vries
[2] and Luikov [3], the moisture and temperature fields
in the drying material are described by a system of
two coupled partial differential equations. The system
of equations incorporates coefficients that are functions of temperature and moisture content thus making it a non-linear system. Such a system has been
used for certain applications. However, for many
practical calculations, the influence of temperature
and moisture content on all transport coefficients has
often been neglected and the resulting system of two
linear partial differential equations has been used [4].
On the other hand, thin-layer drying models are
important tools in mathematical modeling of drying
curves. They are often used to estimate drying time,
generalize drying curves, and have wide application
due to their ease of use and requirement of less data
unlike in complex models. The thin-layer drying
models that describe the drying rate of food materials
are categorized into three groups: theoretical, semitheoretical and empirical [5].
In scientific literature there are many researches
on the experimental studies and mathematical modelling of the drying behavior of various fruits, such as:
apricot [6], apple [7,8], banana [9], cherry [10], grape
[11,12], kiwi [13], quince [14] and plum [15]. There
have been quite a few experimental investigations of
convective hot air drying characteristics and mathematical modeling of processes of drying of pear [1,16–18].
The objectives of this study were:
a) experimental investigation of the drying kinetics of pear for drying air conditions (drying air temperatures 30, 40, 50, 60 and 70 °C; drying air velocities, 1, 1.5 and 2 m s-1; absolute air humidity of
0.0154 kg water kg-1 dry air);
b) evaluation of suitability of thin-layer drying
model and comparison of their goodness of fit, and
development of the model as a function of drying
conditions;
c) determination of the effective moisture diffusivity and activation energy from drying data for the
above mentioned drying conditions.
spherical form, from the central medulla region, where
the cell structure is more uniform, were used in the
drying experiments. The initial moisture content of
fresh slices was determined gravimetrically by hot air
oven method at 105 °C and atmospheric pressure for
a period of 24 h. The average initial moisture content
of pear slices was obtained as 4.99±0.10 kg water kg-1
d.m.
MATERIAL AND METHODS
Uncertainty analysis is a powerful tool when it is
used in the planning and design of experiments. If the
wR is the uncertainty in the result and w1,w2,...,wn are
the uncertainties in the independent variables, then
the R is result in a given function of the independent
variables x1,x2,...,xn. If the uncertainties in the independent variables are all given with same odds, then
uncertainty in the result having these odds and can
be calculated by [20]:
Material
The material used in the experimental part of the
research was fresh pear, cultivar "William”. Until the
processing time, the fruit was stored in cold chamber
at temperature of 4 °C and relative air humidity of
75%. Samples with thickness of 4.0±0.1 mm and
192
Drying procedure
The obtained experimental data set for thin-layer
drying kinetics of pear slices was performed using an
experimental apparatus setup designed to reproduce
industrial convective dryer conditions [19]. The dryer
unit was started 1 h before each experiment in order
to achieve the desired steady state conditions of the
drying air flow. The drying experiments were
performed at drying air temperatures of 30, 40, 50, 60
and 70 °C, drying air velocities of 1, 1.5 and 2 m s-1,
while the absolute air humidity remained constant at
0.0154 kg water kg-1 dry air. The measuring of sample
mass change was conducted continually, without
interruptions to the drying process by a special action
of trays carrier, which was placed on the sensor
(model PW6CC3MR, HBM, Germany). The mass
measuring sensor was connected to a measuring
acquisition system (model NI 622225, National Instruments, USA), which recorded mass change during
the drying process. In the same time interval, the
acquisition system recorded the temperature of dry
and wet-bulb thermometer of the surrounding air
using set of micro-thermocouples K-type. For measurement of air drying temperature and temperature
of drying slices, micro-thermocouples type K also
were used. Air velocity was measured in the measuring pipe using a Pitot tube and a Testo 506 differential micro-manometer. Drying experiments were
stopped when the moisture content of samples decreased to 0.14 kg kg-1 d.m. from the initial value of
4.99 kg water kg-1 d.m. The experiments were replicated three times at each drying air temperature and
drying air velocity, and the average value of the moisture ratio was used for constructing drying curves.
Experimental uncertainty
M. LUTOVSKA et al.: MATHEMATICAL MODELLING OF THIN LAYER…
wR = (
∂R
∂R
∂R
w 1)2 + (
w 2 )2 + ... + (
w n )2
∂x 1
∂x 2
∂x n
(1)
The drying air temperature, the temperature of
drying slices, the relative humidity of the air drying,
the air drying velocity and change of mass of drying
samples are independent parameters measured in
the drying experiments of pears. To carry out these
experiments, the sensitivity of data acquisition system
is ±0.01 °C, with the measurement error is ±0.02 °C,
while the sensitivity of the micro-thermocouple is
±0.01 °C, with measurement errors of ±0.11 °C. The
sensitivity of mass sensor is of 0.01 g with accuracy
of ±2 g, while the used differential micro-manometer
has a measuring range of 0-100 hPa, resolution 1 Pa
and accuracy ±1 Pa. From these data based on
manufacturer’s specification, total uncertainties of the
moisture ratio (MR), and drying rate (DR), were
calculated:
2
2
2
w MR = w DR = w mt
+ w ml
+ w mq
= ±0.15
(2)
where wmt is the total uncertainty in the measurement
of time of mass loss values, wml is the total uncertainty
in the measurement of mass loss values, and wmq is
the total uncertainty in the measurement of the moisture quantity.
Mathematical modelling of drying curves
For approximation of experimental data of the drying kinetic of pear slices five thin-layer mathematical
models from scientific literature were used (Table 1).
For statistical evaluation of these models, the
values for performance index (φ) and chi-squared (χ2)
value were used. The value of performance index is
calculated based on the values of the coefficient of
determination (R2) the root mean squared error
(RMSE), and the mean relative deviation (MRD) [24]:
φ=
R2
RMSE × MRD
(3)
Higher values of the performance index indicate
that the thin-layer drying model better approximates
the experimental drying data.
Chem. Ind. Chem. Eng. Q. 22 (2) 191−199 (2016)
The D’Agostino-Pearson test of normality is the
most effective procedure for assessing the goodness
of fit for a normal distribution. This test is based on
the individual statistics for testing of the residual population of skewness (z1) and kurtosis (z2), the values
of which were calculated according to equations given
in [25]. Then the chi-squared value is computed as
[25]:
χ 2 = z 12 + z 22
(4)
The tabled critical 0.05 chi-square value for the
degree of freedom df = 2 is χ2 = 5.99. Therefore, if the
computed value of chi-square is equal to, or greater
than either of the aforementioned values, the null
hypothesis can be rejected at the appropriate level of
significance (p > 0.95), i.e., the thin-layer drying
model should be rejected [25]. The best model which
is describing the thin-layer drying characteristics of
pear slices has to be chosen on the basis of higher φ
value and lower χ2 value.
Determination of effective moisture diffusivity
In most studies carried out on drying, diffusion is
generally accepted to be the main mechanism during
the transport of moisture to the surface to be evaporated. In this study, the first term of the analytical
solution of Fick’s second law for infinite slab geometry
was used to determine the effective moisture diffusivity of pear slices [18]:
MR =
8
π
2
exp( −
π 2D efft
)
4L2
(5)
where Deff is the effective moisture diffusivity (m2 s-1), t
is the drying time (s), and L is the half-thickness of the
slices (m). From Eq. (5), plotting the data in terms of
ln MR versus drying time gives a straight line with a
slope of θ, from which the effective moisture diffusivity
can be calculated.
The dependence of the effective diffusivity of
food materials on temperature is generally described
by the Arrhenius equation [18]:
Table 1. Thin-layer drying models; MR = Mt/M0 – moisture ratio, Mt – moisture content at any time of drying (kg water kg-1 d.m.), M0 – initial
moisture content (kg water kg-1 d.m.). A, B, C, D, E – empirical coefficients, k1 – drying constant (min-1), t – drying time (min)
Model
Model equation
Name of model
M1
MR = Aexp(-k1t)+Bexp(-Ct)+Dexp(-Et)
Modified Henderson-Pabis
[5]
M2
MR = Aexp(-k1tB)+Ct
Midilli
[21]
M3
MR = Aexp(-k1t)+(1-A)exp(-k1Bt)
Diffusion approach
[5]
B
B
Reference
M4
MR = Aexp(-k1t )+Cexp(-Dt )
Hii
[22]
M5
MR = Aexp(-k1t+Bt0.5) + C
Jena and Das
[23]
193
M. LUTOVSKA et al.: MATHEMATICAL MODELLING OF THIN LAYER…
D eff = D 0 exp(−
E0
)
RTk
Chem. Ind. Chem. Eng. Q. 22 (2) 191−199 (2016)
temperature from 30 to 70 °C resulted in decreased
drying time. The time needed to reach the moisture
content of 0.136 kg kg-1 d.m. changed from 720 to
220 min, when temperature increased from 30 to 70
°C. The reduction in drying time resulted from the increase in vapour pressure within the dried samples
with increasing temperature, enabling faster migration
of moisture to the product surface. Several authors
reported considerable decrease in drying time when
higher temperatures were used for drying [7,10,14,18].
Figure 1b shows the influence of air drying temperature on the variation of the drying rate (DR = (Mt+dt–Mt)/dt) with moisture content at air drying velocity of 2
m s-1. It can be noticed that the drying rates were
higher in the beginning of drying processes and later
decreased with decrease in moisture for all samples
under all drying conditions. The higher drying air temperature produced a higher drying rate and consequently faster reduction in the moisture content, and
(6)
where D0 is the pre-exponential factor of the Arrhenius equation (m2 s-1), E0 is the activation energy for
moisture diffusion (kJ mol-1), R is the ideal gas
constant (8.314 J mol-1 K-1), and Tk is the absolute
temperature of drying air (K). The plot of ln Deff versus
1/Tk gives a straight line with a slope of E0/R and the
intercept equal to ln D0. After that, by using the Arrhenius relationship, the activation energy vcan be calculated.
RESULTS AND DISCUSSION
In Figure 1a, the variations of moisture content
of pear slices with drying time at drying air
temperature of 30, 40, 50, 60 and 70 °C at drying air
velocity of 2 m s-1 are shown. An increase in air drying
(a)
0.08
Drying Rate, DR (kg water/kg d.m.min)
o
0.07
o 30 C,
o
o
o
o
40 C, ◊ 50 C, x 60 C, ∆ 70 C
0.06
0.05
0.04
0.03
0.02
0.01
0.00
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
Moisture content, M (kg water/kg d.m.)
(b)
Figure 1. Drying curves of pear slices at different drying air temperatures and drying air velocity of 2 m s-1: a) moisture content vs. drying
time; b) drying rate vs. moisture content.
194
M. LUTOVSKA et al.: MATHEMATICAL MODELLING OF THIN LAYER…
hence a decrease in drying time. Similar results were
reported in literature for some various fruits
[11,12,14,18]. Figure 2a shows the variation of moisture content with drying time at air drying velocities 1,
1.5 and 2 m s-1 at air drying temperature of 70 °C. An
increase in air drying velocity from 1 to 2 m s-1 resulted in approximately 30% decreased drying time.
The decrease in drying time results from the increase
of heat and mass transfer coefficients between the
drying air and drying samples. Figure 2b shows the
changes in drying rate (DR) as a function of drying
time at the different air drying velocities. The influence
of the drying air velocity on drying rate is significant at
the beginning of the drying process, implying that the
evaporation initially takes place at the surface, being
therefore more directly affected by air velocity. The
predominance of drying air velocity is therefore succeeded by the moisture diffusion process [26]. These
results are in agreement with some other studies in
Chem. Ind. Chem. Eng. Q. 22 (2) 191−199 (2016)
the literature on drying of various vegetables and
fruits [1,26].
The experimental moisture content data
obtained at different drying air temperatures and
different drying air velocities were converted to the
moisture ratio (MR) and then fitted to the five well-known thin-layer drying models given in Table 1. The
method of indirect non-linear regression and quasiNewton estimation method from StatSoft Statistica
software (Statsoft Inc., USA) were used in numerical
experiments. Based on the thin-layer data of pear and
each model from Table 1, the average values of the
coefficient of determination (R2), root mean squared
error (RMSE), mean relative deviation (MRD), performance index (φ) and χ2 were calculated. After that,
the thin layer models were ranked based on average
values of the performance index, φa (Table 2).
From Table 2, it is evident that model M2, i.e.,
the Midilli model, has the highest value of average
Moisture content, M (kg water/kg d.m.)
5.5
5.0
o 1 m/s,
1.5 m/s, x 2 m/s
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0
50
100
150
200
250
300
Drying time, t (min)
(a)
Drying rate, DR (kgwater/kg d.m. min)
0.08
o 1 m/s,
0.07
1.5 m/s, x 2 m/s
0.06
0.05
0.04
0.03
0.02
0.01
0.00
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
Moisture content, M (kg water/kg d.m.)
(b)
Figure 2. Drying curves of pear slices at different drying air velocity at drying air temperature of 70 °C: a) moisture content vs. drying
time; b) drying rate vs. moisture content.
195
M. LUTOVSKA et al.: MATHEMATICAL MODELLING OF THIN LAYER…
performance index (φa = 7898) and lowest average
chi-squared value (χa2 = 1.801) compared to the other
models. Since the effect of drying air temperature and
drying air velocity on the empirical parameters was
not included in the basic Midilli model, a generalized
Midilli model was developed:
MR = ( A log(T ) + B )exp(−Ct Dv E ) + Ft
As shown in Figure 3, a good match was found
between experimental and calculated values of drying
data for pears with the generalized Midilli model.
The values of effective moisture diffusivity
obtained from different drying conditions are presented in Table 4. It can be seen that the values of
effective moisture diffusivity increased with the increase of drying air velocity and drying air temperature.
The Deff values obtained in this study varied in the
range from 6.49×10-9 to 3.29×10-8 m2 s-1 and are in
general range from 10-12 to 10-6 m2 s-1 for drying food
materials [27]. The estimated values of effective moisture diffusivity reported in this study were compared
with other results for pear published in scientific
literature (1.87-8.12×10−10 m2 s-1 for “d’Anjou” pear
after osmotic dehydration, and 1.59×10−10 to 7.64×
×10−10 m2 s-1 for natural pear [28]; 5.10–11.40×10−10
m2 s-1 for pear dried in convective dryer [16]; 8.56×
×10−11 to 2.25×10−10 m2 s-1 for pear dried in convective
dryer at 50, 57, 64 and 71 °C, within an air drying
velocity of 2 m s-1 [18]. The differences between the
results can be explained by the type, composition and
tissue characteristics of the pear slices used in experimental investigations, slice thickness, proposed
model for calculation and method for determination of
moisture diffusivity [18].
The values of pre-exponential factor for different
drying air velocities varied between 4.84 x 10-4 to 1.44
x 10-3, while the values for activation energy for
different drying air velocities ranged between 28.15 kJ
mol-1 and 30.51 kJ mol-1. Those values correspond to
the values given in the scientific literature for drying of
different food materials which are in the range from
12.7 to 110 kJ mol-1 [29]. In Figure 4, a plot of the
natural logarithm of Deff as a function of the reciprocal
of absolute temperature for drying air velocities of 1,
1.5 and 2 m s-1 is shown. The results show linear
relation, due to Arrhenius type dependence.
(7)
where A, B, C, D, E and F are empirical parameters,
T (°C) is drying air temperature, v (m s-1), is drying air
velocity, and t (min) is drying time. Furthermore, multiple indirect non-linear regression analysis was
adopted including all experimental data from drying
kinetics.
Table 2. Statistic summary of the regression analysis; index “a”
denotes average values were calculated for five air drying temperatures and three air drying velocities
R a2
RMSEa
MRDa
φa
χ a2
M1
0.998
0.010
0.098
4732
1.997
4
M2
0.999
0.005
0.037
7898
1.801
1
M3
0.997
0.009
0.118
4688
2.634
5
M4
0.999
0.005
0.042
7282
2.020
2
M5
0.999
0.008
0.070
5019
2.400
3
Model
Chem. Ind. Chem. Eng. Q. 22 (2) 191−199 (2016)
Rank
In order to investigate the influence of measurement error on the parameters of the generalized
model and other statistical parameters, a normally
distributed error with zero mean and standard deviation σ = 0.3 was added to the experimental values of
moisture content. The estimated values of parameters
on the generalized Midilli model, and values of statistical parameters with “exact” (without noise) moisture content data and with noise are given in Table 3.
The higher value of R2 = 0.989 and lower values
of RMSE = 0.065 and MRSE = 0.651 indicate that
generalized the Midilli model can be used to estimate
the moisture ratio of pear with a high accuracy in the
measurement ranges of drying air temperatures of 30
and 70 °C and drying air velocities of 1–2 m s-1.
Table 3 shows the influence of measurement
error on statistical parameters. The value of root
mean squared error increased from 6.5 to 9.5% when
the experimental values of moisture content were
adjusted with normally distributed error with zero
mean and standard deviation σ = 0.3.
CONCLUSIONS
In the present study, the drying characteristics of
pear slices under convective hot-air drying were investigated. The experimental drying data in terms of
moisture ratio were approximated with five well known
thin layer drying models and goodness of fit was
determined using performance index, φ, and chi-squared value, χ2. According to the statistical results,
Table 3. Non-linear regression and statistical parameters
Values obtained
Without noise
With added noise
196
σ
A
B
C
D
E
F×105
R2
RMSE
MRD
0
-0.661
3.575
0.014
0.904
0.308
3.500
0.989
0.065
0.651
0.3
-0.655
3.536
0.014
0.903
0.327
4.200
0.974
0.095
0.989
M. LUTOVSKA et al.: MATHEMATICAL MODELLING OF THIN LAYER…
1.0
CI&CEQ 22 (2) 191−199 (2016)
o
o 30 C, experimental, o calculated
o
40 C, experimental,
Moisture ratio, MR (-)
0.8
calculated
o
◊ 50 C, experimental, ◊ calculated
o
∆ 60 C, experimental, ∆ calculated
o
x 70 C, experimental, x calculated
0.6
0.4
0.2
0.0
0
100
200
300
400
500
600
700
800
Drying time, t (min)
(a)
Moisture ratio, MR (-) zzz
1.0
o 1 m/s, experimental, o calculated
1.5 m/s, experimental, calculated
◊ 2 m/s, experimental, ◊ calculated
0.8
0.6
0.4
0.2
0.0
0
50
100
150
200
250
300
Drying time, t (min)
(b)
Figure 3. Experimental and predicted moisture ratio: a) different air drying temperatures at drying air velocity of 2 m s-1;
b) different drying air velocities at temperature of 70 °C.
Table 4. The estimated values of effective moisture diffusivity
Air velocity, m s
1.0
1.5
2.0
-1
–8
2
-1
R2
Air temperature, °C
Deff / 10 m s
30
0.6491
0.999
40
0.9743
0.996
50
1.460
0.999
60
1.821
0.999
70
2.583
0.991
30
6.822
0.999
40
9.901
0.999
50
1.624
0.995
60
2.294
0.999
70
2.813
0.999
30
0.8113
0.990
40
1.142
0.992
50
1.692
0.999
60
2.303
0.999
70
3.291
0.999
197
M. LUTOVSKA et al.: MATHEMATICAL MODELLING OF THIN LAYER…
-17.0
Chem. Ind. Chem. Eng. Q. 22 (2) 191−199 (2016)
o 1 m/s,
1.5 m/s, x 2 m/s
ln(Deef)
-17.4
-17.8
-18.2
-18.6
-19.0
0.0028
0.0029
0.003
0.0031
0.0032
0.0033
0.0034
1/(T+273.15)
Figure 4. Effect of air drying temperature on the effective diffusivity for each air drying velocity of pear slices.
a new generalized Midilli model was developed. This
model can be used to predict the moisture content of
pear slices at any time of drying processes with high
ability between drying air temperatures of 30 and 70
°C and drying air velocities of 1 to 2 m s-1. The
effective moisture diffusivity values were estimated
from Fick’s diffusion and ranged between 6.49×10-9
and 3.29×10-8 m2 s-1, while the values of activation
energy were in the range of 28.15–30.51 kJ mol-1.
Acknowledgments
[11]
S. Azzouz, A. Guizani, W. Jomaa, A. Belghith, J. Food
Eng. 55 (2002) 323-330
[12]
I. Doymaz, M. Pala, J. Food Eng. 52 (2002) 413–417
[13]
I. Doymaz, J. Food Process. Preserv. 33 (2009) 145-160
[14]
D.A. Tzempelikos, A.P. Vouros, A.V. Bardakas, A.E. Filio,
D.P. Margaris, Case Studies in Therm. Eng. 3 (2014) 79–85
[15]
I. Doymaz, J. Food Eng. 64 (2004) 465–470
[16]
R.P.F. Guiné, Int. J. Food Sci. Technol. 41 (2006) 1177–1181
[17]
The authors would like to thank the anonymous
reviewers for their useful comments.
R.P.F. Guiné, D.M.S. Ferreira, M.J. Barroca, F.M.
Gonçalves, Biosyst. Eng. 98 (2007) 422–429
[18]
I. Doymaz, Int. J. Food Sci. Technol. 48 (2013) 1909–1915
REFERENCES
[19]
I. Pavkov, Ph.D. thesis, University of Novi Sad, Faculty
of Agriculture, Novi Sad, 2012
[1]
I. Doymaz, O. Ismail, Food Sci. Biotechnol. 21 (2012)
1377-1381
[20]
J.P. Holman, Experimental methods for engineers,
McGraw-Hill, New York, 2011
[2]
J.R. Philip, D.A. De Vries, Eos, Trans. Am. Geophys.
Union 38 (1957) 222-232
[21]
A. Midilli, H. Kucuk, Z. Yapar, Drying Technol. 20 (2002)
1503-1513
[3]
A.V. Luikov, Theory of Drying, Energyia, Moscow, 1968
(in Russian)
[22]
C.L. Hii, C.L. Law, M. Cloke, J. Eng. Sci. Technol. 3
(2008) 1-10
[4]
G. Kanevce, Lj. Kanevce, V. Mitrevski, G. Dulikravich,
H.R.B. Orlande, Int. J. Heat Mass Transfer 129 (2007)
379-386
[23]
S. Jena, H. Das, J. Food Eng. 79 (2007) 92-99
[24]
I.I. Ruiz-López, E. Herman-Lara, Drying Technol. 27
(2009) 726-738
[5]
Z. Erbay, F. Icier, Crit. Rev. Food Sci. Nutr. 50 (2009)
441-464
[25]
D.J. Sheskin, Handbook of parametric and nonparametric
statistical procedure, CRC Press, Boca Raton, FL, 2004
[6]
E. Mirzaee, S. Rafiee, A. Keyhani, Z. Emam-Djomeh,
Res. Agric. Eng. 55 (2009) 114-120
[26]
A. Zhu, F. Jiang, Chem. Ind. Chem. Eng. Q. 20 (2014)
171-181
[7]
S. Seiiedlou, R.H. Ghasemzadeh, N. Hamdami, F. Talati,
M. Mohammad, Int. J. Agric. Biol. 12 (2010) 171-178
[27]
D.G. Saravacos, B.Z. Maroulis, Transport properties of
foods, Marcel Dekker, New York, 2001
[8]
C.A. Cruz, P.F.R. Guine, C.J. Gonçalves, C.A. Correia, In
Proceedings of the International Conference of Agricultural Engineering, GIGR-AgEng, Valencia, 2012
[28]
K.J. Park, A. Bin, F.P.R. Brod, J. Food Eng. 56 (2003) 97–103
[9]
I. Doymaz, Int. J. Food Prop. 13 (2010) 486-497
[29]
N.P. Zogzas, Z.B. Maroulis, D. Marinos-Kouris, Drying
Technol. 14 (1996) 2225–2253.
[10]
I. Doymaz, O. Ismail, Food Bioprod. Process. 89 (2011)
31-38
198
M. LUTOVSKA et al.: MATHEMATICAL MODELLING OF THIN LAYER…
MONIKA LUTOVSKA1
VANGELCE MITREVSKI2
IVAN PAVKOV3
VLADIMIR MIJAKOVSKI2
MILIVOJ RADOJČIN3
1
Faculty of Technical Sciences,
University St. Kliment Ohridski, Bitola,
Macedonia
2
Faculty of Technical Sciences,
University “St. Kliment Ohridski”, Bitola,
Macedonia
3
Poljoprivredni fakultet, Univerzitet u
Novom Sadu, Novi Sad, Serbia
NAUČNI RAD
Chem. Ind. Chem. Eng. Q. 22 (2) 191−199 (2016)
MATEMATIČKO MODELOVANJE SUŠENJA
KRUŠKE U TANKOM SLOJU
U ovom radu je istraživano sušenje kriški kruške u tankom sloju kao u funkciji uslova
sušenja. Grupe eksperimentalnih podataka istraživanja kinetike sušenja u tankom sloju pri
pet različitih temperatura vazduha (30, 40, 50 , 60 i 70 °C) i tri različite brzine strujanja
vazduha (1, 1,5 i 2 m s- 1) u eksperimentalnoj sušari koja simulira industrijske uslove
konvektivnog sušenja. Pet dobro poznatih kinetičkih modela je korišćeno za fitovanje eksperimentalnih podataka za sadržaj vlage, a najbolji model je određen numerički. Za svaki
model i grupu podataka, izračunati su statistički indeks uspešnosti (φ) i χ2-vrednost, a zatim
su modeli rangirani. Izvršena statistička analiza pokazuje da model Midilli daje najbolje
statističke ocene. Kako uticaj temperature i brzine strujanja vazduha na empirijske parametre nije bio uključen u osnovni Midilli model, razvijen je generalizovani oblik ovog
modela. Sa ovim modelom, kinetički podaci sušenja kriški krušaka mogu biti aproksimirani
sa visokom tačnošću. Efektivna difuzivnost vlage je određena korišćenjem drugog Fikovog
zakona. Dobijene vrednosti efektivne difuzivnosti vlage (Deff) tokom sušenja su između
6,49×10-9 i 3,29×10-8 m2 s-1, dok vrednosti energije aktivacije (E0) variraju od 28,15 do
30,51 kJ mol-1.
Ključne reči: matematičko modelovanje, sušenje u tankom sloju, difuzivnost
vlage, aktivaciona energija.
199
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 201−209 (2016)
BILJANA
DAMJANOVIĆ-VRATNICA1
SVETLANA PEROVIĆ2
TIEJUN LU3
REGINA SANTOS3
1
Faculty of Metallurgy and Technology, University of Montenegro,
Podgorica, Montenegro
2
Faculty of Natural Sciences and
Mathematics, University of Montenegro, Podgorica, Montenegro
3
School of Chemical Engineering,
College of Engineering and
Physical Sciences, University of
Birmingham, Edgbaston,
United Kingdom
SCIENTIFIC PAPER
UDC 66.061.3:582:546.26-31:54
DOI 10.2298/CICEQ150504034D
CI&CEQ
EFFECT OF MATRIX PRETREATMENT ON
THE SUPERCRITICAL CO2 EXTRACTION
OF Satureja montana ESSENTIAL OIL
Article Highlights
• Supercritical CO2 extraction of Satureja montana was carried out with different pretreated matrices
• GC/FID analysis of extracts - dominant compound thymol was provided
• Gland physical disruption by fast decompression provides the highest content of thymol and carvacrol
• Extracts showed rather significant antimicrobial activity, similar to essential oil
Abstract
The effect of matrix pretreatment of winter savory (Satureja montana L.) on the
supercritical CO2(SC-CO2) extraction yield, composition and antimicrobial activity of extracts and essential oil (EO) was investigated. The herb matrix was
submitted to conventional mechanical grinding, physical disruption by fast
decompression of supercritical and subcritical CO2, and physical disruption by
mechanical compression. The analyses of the essential oil obtained by SC-CO2
extraction and hydrodistillation were done by GC/FID method. The major compounds in winter savory EO obtained by SC-CO2 extraction and hydrodistillation were: thymol (30.4-35.4 and 35.3%), carvacrol (11.5-14.1 and 14.1%),
γ-terpinene (10.2-11.4 and 9.1%) and p-cymene (8.3-10.1 and 8.6%), respectively. The attained results revealed that physical disruption of essential oils
glands by fast CO2 decompression in supercritical region (FDS) achieved the
highest essential oil yield as well as the highest content of thymol, carvacrol
and thymoquinone. Antimicrobial activity of obtained winter savory SC-CO2
extracts was the same (FDS) or weaker compared to essential oil obtained by
hydrodistillation.
Keywords: Satureja montana, supercritical CO2 extraction, herb matrix
pretreatment, essential oil yield, essential oil composition.
Satureja montana L., commonly known as
winter savory or mountain savory, belongs to the
Lamiaceae family, Nepetoideae subfamily and Mentheae tribe and is a perennial semi-shrub (20-30 cm)
that inhabits arid, sunny and rocky regions. S. montana L. is native to the Mediterranean and can be
found throughout Europe, Russia and Turkey [1]. The
whole herb is mildly antiseptic, aromatic, carminative,
digestive, mildly expectorant and stomachic and,
therefore, is widely used as a flavoring agent in food
Correspondence: B. Damjanović-Vratnica, Faculty of Metallurgy
and Technology, University of Montenegro, 81000 Podgorica,
Montenegro.
E-mail: biljanad@ac.me
Paper received: 4 May, 2015
Paper revised: 10 July, 2015
Paper accepted: 27 August, 2015
products and also as a traditional herbal medicine [2].
The positive effects of savory on human health as
well as its antibacterial activity are attributed to its
active constituents such as EO, triterpenes, flavonoids and rosmarinic acid [3-7].
Essential oils (EOs) are highly concentrated volatile, aromatic essences of aromatic plants that can
be obtained by expression, fermentation, enfleurage
or extraction with classical solvents; but the method of
hydrodistillation (HD) is most commonly used for
commercial production of Eos [8]. Supercritical CO2
(SC-CO2) extraction compares favorably with conventional processes because it allows high yields to
be attained while simultaneously shortening the process time, avoiding thermal and hydrolytic degrad-
201
B. DAMJANOVIĆ-VRATNICA et al.: EFFECT OF MATRIX PRETREATMENT…
ation of labile compounds, and avoiding toxic solvents
residue in the product [9].
The need for a matrix pretreatment is related to
the location of the EOs within the herbaceous matrixes. In many aromatic herbs, EOs are largely located within glandular trichomes that develop on the
surface of leaves and other organs of the plants. The
so-called “peltate hairs” appear to contain most of the
oil and will henceforth be called “the glands”. The
glands are well described in literature and an extensive “library” of electronmicrographs and photomicrographs has been built up showing these glands and
their location on the leaves of typical herbs [10]. Most
of EOs of aromatic herbs from Lamiaceae family is
found on the surface of the leaves in peltale gladular
trichomes (peltate glands) [11,12]. The transport of
material to or from the intact glands is highly restricted by the glandular walls which must be disrupted
by mechanical means or other techniques to reach
acceptable extraction rate of the oil using compressed
CO2 as solvent [13].
Commonly, mechanical processes like flaking,
grinding or pelletizing are applied [14]. However, during
mechanical pretreatments, losses by degradation
(oxidation and thermal degradation) and by evaporation of volatiles are observed, leading to a discrepancy between the EO composition of the herb and
that of the extract [15]. In previously published investigations [16-18] a positive effect on extraction kinetics was observed after CO2 decompression of Lamiaceae species, Aloe vera as well as valerian and ginger roots. Also, it was found that carvacrol content in
S. montana extracts can be significantly increased by
application of ultrasound and high pressure pretreatments [19].
The aim of this work was to determine the
effects of different S. montana matrix pretreatments
on the supercritical CO2 extraction rate, yield and
composition of extracts, as well as its antimicrobial
activity.
EXPERIMENTAL
Matrix
Many factors can influence the amount of EO in
aromatic herbs, such as climate and environmental
conditions and age of plants. To minimize this influence in the present work, leaves of wild growing S.
montana were collected manually from the same
collection site from the central part of Montenegro
(near Podgorica, 100 m a.s.l.) prior the flowering (in
May). Collected leaves were air-dried at room temperature for 7 days, packed in double walled paper
202
Chem. Ind. Chem. Eng. Q. 22 (2) 201−209 (2016)
bags and stored at 5 °C before use. Voucher specimen was confirmed and deposited in Herbarium,
Department of Biology, Faculty of Natural Sciences
and Mathematics, University of Montenegro, voucher
number: S809/08.
Matrix pretreatment procedure
The initial water inherent in the dried winter savory leaves was found to be 8.7 mass% using a Dean
and Stark apparatus with n-heptane as the reflux
solvent.
In order to determine effects of winter savory
matrix pretreatment on SC-CO2 extraction yield, following pretreatments were applied:
• Intact herb matrix.
• Conventional mechanical grinding of herb
matrix with different exposure periods: typically, 100 g
of material was milled in a domestic blender (Multi
Moulinex, 260 W) during 20, 40 and 60 s and, after
sieving (laboratory Erweka sieves, mesh from 0.1 to 2
mm), mean particle size was determined (d20 = 1.1
mm, d40 = 0.9 mm and d60 = 0.6 mm, respectively).
Herb batch was used immediately in extraction experiments in order to minimize losses of volatile compounds.
• Physical disruption of EOs glands by fast
decompression of supercritical and subcritical CO2
(FDS and FDL) respectively, as alternative physical
pre-treatment:
The physical disruption of EOs glands with compressed CO2 was studied at 40 (FDS) and 26 °C
(FDL). 80 g intact herb leaves was enclosed in an
extraction vessel and compressed CO2 was fed from
the bottom of extractor. When the desired pre-expansion pressure was achieved (90 bar), inlet valve
was closed and the bed was exposed to CO2 at this
pressure for a designed period (60 min) before fast
decompression of the bed was carried out (∼1.20 s).
FDS and FDL pretreatments were immediately followed by CO2 extraction in the same apparatus.
• Physical disruption of EOs glands by
mechanical compression (pellet): 30 g of intact herb
leaves was charged in a 25 mm diameter die, and
exposed 5 min to 15 t m–2 mechanical pressure in a
Graseby Specac 25.011 hydraulic press to make a
pellet. The obtained pellets were smashed using a
pestle and mortar, and used immediately in extraction
experiments.
Supercritical carbon dioxide extraction procedure
The supercritical carbon dioxide extraction was
carried out on extraction apparatus in the Laboratory
of Supercritical Fluid Technology Group, University of
Birmingham (Figure 1).
B. DAMJANOVIĆ-VRATNICA et al.: EFFECT OF MATRIX PRETREATMENT…
Chem. Ind. Chem. Eng. Q. 22 (2) 201−209 (2016)
Figure 1. Schematic diagram of the apparatus used in supercritical CO2 experiments.
Extraction procedure was previously described
in detail elsewhere [15]. Shortly, commercial CO2,
from storage cylinders at bottle pressure, was fed
through valves, V1 and NRV1, and a filter, F1. In
order to reduce the temperature of the CO2, the gas
passes through a coil immersed in a refrigeration
bath, HE1. The cooled CO2 was then pressurized in
the liquid compressor pump and passed through a 50
µm filter, F2 and NRV2. The CO2 then passes through
a heat exchanger (HE2), where now the solvent is a
supercritical fluid and enters the external insulated air
bath housing. The gas was transported through
another heat exchange before entering the pressure
vessel.
For each experiment, 80 g of pre-treated herb
was weighed and, before loading, some glass wool
was inserted, after which the herb was transferred in
the pressure vessel. Finally, two steel mesh filters of
different types were inserted to stop particles leaving
the vessel.
Obtained extract was collected in first collector,
after depressurized CO2. Additionally, EO was attached
to the “cold collector”, which was immersed in a mixture of acetone/dry-ice (–82 °C). Its purpose was to
collect the volatile components (solute) from the CO2
that may have escaped from first collector.
The extraction conditions for all experiments
were: temperature 40 °C and pressure 100 bar, extraction time 360 min and CO2 flow rate 0.3 kg CO2/h.
Commercial carbon dioxide (99.995% purity, BOC
grade N4.5-CP) as well as acetone (capillary GC
grade, ≥99.9%, Sigma-Aldrich, UK) and dry ice (-82
°C) were used for extractions.
Hydrodistillation procedure
80 g of herb material (60 s milled in a domestic
blender, d = 0.6 mm) was submitted to hydrodistillation in a Clevenger-type apparatus for 2 h according
to Yugoslav Pharmacopoeia IV. The obtained essential oil was dried over anhydrous sodium sulphate,
measured, poured in hermetically sealed dark-glass
containers and stored in a freezer at -4 °C until
analyzed by GC.
Gas chromatography
The yield of the gained essential oil was evaluated by GC-FID using a VARIAN 3400 gas chromatograph with a DB-5 capillary column (30 m×
×0.32 mm, film thickness 25 µm). The analysis conditions were: oven temperature was programmed at
50 °C for 5 min, and then increased to 140 °C at a
rate of 8 °C/min and then increased to 250 °C at a
rate of 13 °C/min; the injector and detector temperature were 120 and 270 °C, respectively. Helium as
carrier gas was adjusted to flow rate of 1.4 ml/min,
and the injection mode was splitless.
The constituents of extracts were identified by
comparing their retention times with authentic standards (Fluka, Great Britain) and their mass was calculated from a predetermined peak area response
factor.
203
B. DAMJANOVIĆ-VRATNICA et al.: EFFECT OF MATRIX PRETREATMENT…
Microbial strains
In order to evaluate the activity of the essential
oil of Satureja montana, the following microorganisms
were used: as reference strains Staphylococcus
aureus ATCC 25923, Escherichia coli ATCC 25922,
Pseudomonas aeruginosa ATCC27853 and Candida
albicans ATCC 10231 (Torlak, Belgrade, Serbia);
further, the clinically isolated S. aureus, E. coli, P.
aeruginosa and C. albicans.
The microorganisms were isolated from clinically treated patients of the Clinical Centre of Montenegro (Podgorica, Montenegro).
Determinations of minimum inhibitory concentration
(MIC) and minimum bactericidal concentration (MBC)
A broth microdilution method was used to determine the minimum inhibitory concentrations (MIC)
and minimum bactericidal concentration (MBC) [20].
All tests were performed in Mueller Hinton broth supplemented with Tween 80 at a final concentration of
0.5 vol.%. Briefly, serial doubling dilutions of the extracts were prepared in a 96-well microtiter plate
ranged from 0.09 to 25.00 mg/ml. To each well 10 μL
of resazurin indicator solution (prepared by dissolving
a 270-mg tablet in 40 mL of sterile distilled water) and
30 μL of Mueller–Hinton broth were added. Finally, 10
μL of bacterial suspension (106 CFU/mL) was added
to each well to achieve a concentration of 104 CFU/mL.
Two columns in each plate were used as controls:
one column with a broad-spectrum antibiotic as a
positive control (amykacine) and one column containing the methanol as negative controls. Plates
were wrapped loosely with cling film to ensure that
bacteria did not become dehydrated and prepared in
triplicate, and then they were placed in an incubator
at 37 °C for 18-24 h. Color change was then
assessed visually. The lowest concentration at which
color change occurred was taken as the MIC value.
The average of 3 values was calculated and that were
the MIC and the MBC for the tested extract. The MIC
is defined as the lowest concentration of the extract at
which the microorganism does not demonstrate visible growth. The microorganism growth was indicated
by the turbidity. The MBC was defined as the lowest
concentration of the extract at which incubated microorganism was completely killed.
Assay of in vitro antifungal activity
Broth microdilution assays were performed in
accordance with the guidelines [21]. Briefly, stock solutions were prepared in water for nystatin and in
methanol for oil. The final dilution was prepared in
RPMI 1640 medium, adjusted to pH 7.0 with 0.165 M
204
Chem. Ind. Chem. Eng. Q. 22 (2) 201−209 (2016)
morpholenepropanesulfonic acid buffer, in the range
from 0.09 to 25 mg/ml and inoculum size of 103
CFU/ml. The growth (drug free) and sterility controls
were also included. Microdilution trays were incubated in ambient air at 35 °C. MICs were determined
visually after 48 h of incubation, as the lowest concentration of the drug that caused no detectable
growth. The average of 3 values was calculated
giving the MIC and the MBC for the tested oil.
Statistical analysis
Extraction experiments were carried out in triplicate. Means and standard deviation (SD) were calculated using Origin Pro 8 (OriginLab, USA). Duncan’s test was conducted to analyze the difference
between various pre-treatments. A value of P < 0.05
was considered as statistically significant.
RESULTS AND DISCUSSION
The main volatile components and average yield
of Satureja montana essential oil, after 6 h supercritical CO2 extraction from differently treated matrix
and hydrodistillation are shown in Table 1.
Major compounds in winter savory EO obtained
by SC-CO2 extraction and HD were: thymol (30.4–35.4 and 35.3%), carvacrol (11.5-14.1 and 14.1%),
γ-terpinene (10.2-11.4 and 9.1%) and p-cymene (8.3–10.1 and 8.6%), respectively.
The extraction yield value of S. montana EO
was similar to that previously found [5,22–24] but
much higher than the yield reported by other
researchers [1,25]. The phytochemical profile of the
winter savory EO in this study was in agreement with
the results of several authors who have also evaluated this vegetal species [1,5,26,27]. In contrast, the
savory EO evaluated by Ćavar et al. [25] was characterized by a high content of alcohols, such as geraniol and terpinen-4-ol. It was found that the final
chemical depends on: each organ and its stage of
development; the climatic conditions of the plant
collection site; the degree of terrain hydration; macronutrient and micronutrient levels; and the plant material’s drying conditions [9,28].
The percentage of oxygenated monoterpenes
was rather high in all samples (45.1-54.5%), due to
high content of major compounds thymol and carvacrol (Figure 2).
The percentage of sesquiterpene hydrocarbons
was highest in essential oil obtained by hydrodistillation (16.3%) while in other extract varied from
14.5% (20 s grind matrix, d20 = 1.1 mm) to 7.1%
(FDS). The content of monoterpene hydrocarbons
was quite similar in all samples (27.0-28.4%) except
B. DAMJANOVIĆ-VRATNICA et al.: EFFECT OF MATRIX PRETREATMENT…
CI&CEQ 22 (2) 201−209 (2016)
Table 1. Major components (%) of S. montana essential oil obtained by SC-CO2 extraction (after 6 h) from different pretreated herb
matrices and hydrodistillation
Component
Pretreatment procedure
Pellet 15t
20 s grind
40 s grind
60 s grind
FDL 60min
FDS 60min
HD
α-Pinene
1.6
0.2
0.5
0.9
0.9
0.9
0.8
Sabinene
0.1
0.1
0.1
0.2
0.2
0.1
0.2
β-Pinene
0.3
1.1
1.2
1.1
0.4
0.3
0.3
β-Myrcene
1.9
2.9
2.8
2.4
2.7
1.3
0.7
α-Terpinene
1.3
1.6
1.6
1.3
1.2
0.9
3.1
p-Cymene
8.3
9.1
8.3
9.3
10.1
8.8
8.6
Limonene
3.3
2.5
2.4
2.6
2.4
3.3
1.3
γ-Terpinene
10.9
10.2
10.7
10.2
10.5
11.4
9.1
Linalool
1.9
2.1
2.3
2.8
3.8
2.4
3.8
Terpinene-4-ol
1.1
0.8
0.9
0.8
1.4
1.3
1.9
Thymoquinone
0.5
0.3
0.8
0.9
1.2
1.4
-
Thymol
30.7
30.4
30.5
31.3
32.5
35.4
35.3
Carvacrol
12.5
11.5
12.3
12.7
13.2
14.1
14.1
t-Caryophyllene
3.6
4.9
4.3
3.6
2.4
1.8
4.3
Germacrene D
0.2
0.2
0.1
0.3
0.2
0.1
2.1
β-Farnesene
3.4
4.5
4.1
4.2
2.3
2.1
3.7
δ-Cadinene
3.8
4.8
4.8
4.1
3.5
2.9
4.6
Caryophyllene oxide
0.5
0.1
0.2
0.2
0.4
0.2
1.6
Total identified
85.9
87.3
87.9
88.9
89.3
88.6
95.6
Yield, mass%
1.62
1.28
1.35
1.48
1.64
1.75
1.82
Figure 2. Yield (mass%) of S. montana essential oil isolated by SC-CO2 extraction and hydrodistillation (HD) with respect to grouped
components, expressed as mean ± standard deviation; a,b,c,dmeans for each pre-treatment, without a common letter, are significantly
different from each other (P < 0.05).
for the essential oil obtained by HD (24.1%). This
result is probably due to an uncontrolled loss of some
volatile components from the Clevenger apparatus
[24]. The major difference between these two techniques (SC-CO2 and HD) is the occurrence of thymoquinone, an oxygen-containing monoterpene in all
SC-CO2 extracts. This compound is of great importance to the pharmaceutical industry, due to its anti-
cancer, antioxidant and anti-inflammatory properties,
as well as the neuroprotective effect against forebrain
ischemia and Alzheimer disease [23].
From the general phytochemical knowledge and
also from the obtained results it is evident that the
preprocessing of plant material plays rather significant role in chemical composition of the essential oil.
Thus, highest content of thymol, carvacrol and γ-ter-
205
B. DAMJANOVIĆ-VRATNICA et al.: EFFECT OF MATRIX PRETREATMENT…
pinene (35.4, 14.1 and 11.4%, respectively) as well
as limonene (3.33%) was recorded in the EO
obtained from winter savory matrix treated by FDS.
The highest content of p-cymene and linalool (10.1
and 3.8%, respectively) was found in EO gained
savory matrix treated by FDL. The smallest content of
very important compounds as thymol and carvacrol
(30.4 and 11.5%, respectively) was recorded in EO
obtained from 20 s grinded herb matrix (d20 = 1.1
mm). For mechanically treated samples, it is noticeable that content of phenolic compound thymol, carvacrol and thymoquinone rise as particle diameter
size decreases (Table 1). Grosso et al. [23] observed
similar trend during SC-CO2 extraction (at 40 °C, 90
bar and particle size 0.4, 0.6 and 0.8 mm) of S. montana volatile oil for carvacrol and thymol content.
High antimicrobial activity of savory EO is probably due to the presence of phenolic components,
such as thymol and its isomer carvacrol as well as its
precursors, γ-terpinene and p-cymene which activities
have been confirmed. Moreover, a number of researchers had shown that components present in
lower amount in savory EO, such as terpinen-4-ol,
linalool and limonene, could also contribute to the
antimicrobial activity of the oil [7,29].
To test the effect of the winter savory leaves
pretreatment on the CO2 extraction of EOs, the extraction curves from differently treated matrixes were
compared to that of the untreated matrix (Figure 3).
The yields are expressed as the percentage ratio
between the mass of EO extracted and the initial
mass of dry herb sample.
Chem. Ind. Chem. Eng. Q. 22 (2) 201−209 (2016)
As can be seen, the extraction yields are considerably enhanced when pretreated matrices are
extracted, compared to intact herb. Relatively limited
extraction rate was observed from the intact herb matrix, probably because of intraparticle resistance due
to the location of the EO in the herb. It has been
shown in the literature that the ability of CO2 to penetrate intact glands is restricted by the low solubility of
both cuticle and cell wall components [30].
Also, as can be seen from Figure 3, two different
shapes of extraction curves can be observed among
pretreated herb matrices. Mechanically prepared matrices revealed high extraction rates in first 120 min of
extraction, which was followed by rather low extraction rates in the later stages. This trend is especially noticeable during extraction from mechanically
pretreated herb matrix with diameter particle size
d20 = 1.1 mm and d40 = 0.9 mm. On the other hand,
physically treated and mechanically pressured matrices as well as mechanically pretreated matrices
(with diameter particle size d60 = 0.6 mm) show a
smoother curve through extraction.
The disruption of glands during the mechanical
grinding of the leaves occurred either by direct contact with the blender blades or by particle collision
helped by the turbulence within the blender. This turbulence probably resulted in higher dislocation of the
EO from disrupted glands, so this oil is accessible to
the extraction solvent. This fraction was easily extracted in the earlier phase, whereas the remaining
EO was extracted very slowly in the later phase.
Figure 3. Essential oil extraction curves from different pretreated S. montana matrices isolated by SC-CO2 extraction, expressed as
mean ± standard deviation.
206
B. DAMJANOVIĆ-VRATNICA et al.: EFFECT OF MATRIX PRETREATMENT…
Chem. Ind. Chem. Eng. Q. 22 (2) 201−209 (2016)
rise to the extent of 42 to 93 °C during spice grinding,
causing loss of volatile oil and flavour constituents
[35]. The temperature rise can be minimized to some
extent by circulating cold air or water around the grinder, but this technique is not sufficient. It was previously found that black pepper volatile oil content
was 26% higher after the cryogenic grinding in comparison with grinding at room temperature [36].
After 6 h of extraction, rather similar extract
yields were obtained from FDL extraction and mechanically compressed matrices (1.64 and 1.62%, respectively), while the extraction yield from FDS pretreated herb matrix was the highest in this set of
experiments (1.75%) with further increase trend. It
should be noted that, in our study, the essential oil
yield obtained by hydrodistillation is 1.82%, which is
comparable to the estimate content within the FDS-treated matrix, i.e., 1.75%. In previously published
investigation, Stamenić et al. [18] also observed
higher yields for extraction of several materials –
among them hyssop, thyme and mint during extraction with CO2 with subsequent decompression to
atmospheric conditions.
The results of the bioassays (Table 2) show that
tested S. montana extracts obtained by SC–CO2 extraction from different pretreated herb matrices exhibited the same (FDS) or weaker antimicrobial activity
in comparison with essential oil obtained by hydrodistillation.
The highest antibacterial efficiency was shown
by the SC-CO2 extract after FDS pretreated herb
matrix where MIC ranged between 0.09 mg/ml and
25.0 mg/ml. The lowest activity of FDS extract was
observed for P. aeruginosa (25.0 mg/ml), while the
highest activity was observed against S. aureus and
E. coli (0.09 mg/ml). This efficiency could be attributed to the high content of compounds with known
antimicrobial activity, such as phenolic components
thymol and its isomer carvacrol as well as its pre-
In physically treated herb matrices by fast CO2
decompression and mechanical pressure, the lack of
turbulence during the preparation process results in
lower displacement of the EO from glands; consequently, extraction curves are more gradual. When
exposed to compressed CO2, the gas slowly penetrates the glands and dissolves in the intraglandular oil
until the solubility limit is reached. During the fast
decompression of the bed, the dissolved gas is desorbed from the oil phase and discharged to the bulk
solvent. The inability of the glands to discharge the
gas, at a rate dictated by the loss of solubility in the oil
with the decompression of the bed, generates a pressure gradient across the glands that may leads to its
rupture [30].
Particle size plays an important role in SC extraction processes, when internal mass transfer resistance is reduced and the extraction is controlled by
equilibrium conditions. Furthermore, the extraction
rate increases because of a shortening in the diffusion path. In our work, the yields of EO from herb
leaves treated by mechanical grinding in 20 and 40 s
period (particle size d20 = 1.1 mm and d40 = 0.9 mm)
were 1.28 and 1.35%, respectively. The smaller yield
could be explained with low efficiency of grinding
treatment in liberating intraglandular EO.
It was previously found [31-34] that reduced
substrate particle size positively effects essential oils
yield and SC-CO2 extraction rate. Smaller particles
caused increases in the specific surface area as well
as a disruption of the cell walls and other inner barriers, thus leaving the essential oil more accessible to
the SC-CO2.In our study, at the smallest diameter
particle size (d60 = 0.6 mm), the SC-CO2 extraction
yield was higher (1.48%), compared to mechanically
treated samples, but not as expected. We assume
there is some loss of essential oil during extended
grinding period (60 s) due to higher temperature
through this pretreatment process. Temperature could
Table 2. Antimicrobial activity ( MIC/MBC, mg/ml) of S. montana extracts obtained by SC-CO2 extraction (after 6 h) from different
pretreated herb matrices and essential oil obtained by hydrodistillation
Microorganism
Pretreatment procedure
Pellet 15 t
20 s grind
40 s grind
60 s grind
FDL 60 min
FDS 60 min
HD
Staphylococcus aureus
0.36/0.36
1.57/1.57
0.36/0.36
0.09/0.18
0.09/0.09
0.09/0.09
0.09/0.09
S. aureus ATCC 25923
0.18/0.36
0.36/0.36
0.18/0.36
0.09/0.09
0.09/0.09
0.09/0.09
0.09/0.09
Escherichia coli
3.13/3.13
6.25/12.5
3.13/3.13
0.36/0.36
0.36/0.36
0.18/0.36
0.18/0.18
E. coli ATCC 25922
1.57/1.57
6.25/6.25
1.57/1.57
0.36/0.36
0.18/0.36
0.18/0.18
0.09/0.18
25.0/25.0
Pseudomonas aeruginosa
>25.0
>25.0
>25.0
>25.0
25.0/25.0
25.0/25.0
P. aeruginosa ATCC 27853
12.5/12.5
12.5/25.0
12.5/12.5
6.25/12.5
6.25/6.25
3.13/3.13
Candida albicans
0.36/0.36
0.72/0.72
0.72/0.72
0.36/0.36
0.36/0.36
0.18/0.18
0.18/0.18
C. albicans ATCC 10231
0.36/0.36
0.72/0.72
0.72/0.72
0.36/0.36
0.36/0.36
0.18/0.18
0.18/0.18
207
B. DAMJANOVIĆ-VRATNICA et al.: EFFECT OF MATRIX PRETREATMENT…
cursors, γ-terpinene and p-cymene [29]. Thymol
disintegrates the outer membrane and increases the
permeability of the cytoplasmic membrane to ATP of
E. coli [37]. Carvacrol is an isomer of thymol and has
been shown to cause damage in B. subtilis cells [38].
The presence of the hydroxyl group seems to be
more important for the antimicrobial activities of these
compounds than the ability to expand and consequently to destabilize the bacterial membrane.
The weakest antimicrobial efficiency was shown
by the SC-CO2 extract after 20 s grinding (MIC ranged
between 0.72 and 25.0 mg/ml), probably due to the
smallest content of very important compounds such
as thymol and carvacrol.
The mentioned results suggest the significance
of individual oil components ratio in the antimicrobial
mixture. Antimicrobial action is often determined by
more than one component; each of them contributes
to the beneficial or adverse effects [9].
The obtained results showed that SC-CO2 extracts as well as essential oil obtained by hydrodistillation were more effective against ATCC strains
than against clinically isolated strains. The data indicated that Gram-positive S. aureus was the most
sensitive strain tested to the savory extracts while P.
aeruginosa was the most resistent. Gram-negative P.
aeruginosais known to have a high level of intrinsic
resistance to virtually all known antimicrobials and
antibiotics, due to a very restrictive outer membrane
barrier [39]. Pseudomonas species are known to have
the ability to metabolise a wide range of organic compounds and for this fact is used extensively in bioremediation; this may explain their high level of resistance. They may simply metabolise the compounds in
extracts that are inhibitory to many of the other bacteria [40]. Previous reports [28,30,44] showed that
Gram-positive bacteria are generally more sensitive
to the effects of the savory essential oil, which was
confirmed in this study. S. montana extracts and
essential oil generally exhibited relatively high antifungal activity, whether as clinically isolated or as
ATCC strain, regardless of the individual components’
percentage values, what confirms previous results on
this subject [5,7,40].
critical CO2) matrix pretreatment results in the highest
content of EO which is believed to be of better quality
than those obtained from mechanically treated matrix,
due to higher content of oxygenated compounds
thymol, carvacrol and thymoquinone.
The supercritical CO2 extracts, as well as the
essential oil obtained by hydrodistillation, were the
most effective against S. aureus and E. coli. Any microbiological activity depends on the chemical composition of oil and the investigated strain sensitivity.
Therefore, it can be concluded that the similar antimicrobial activities of essential oil and FDS extract
are probably caused by phenolic components, thymol
and carvacrol which are the main components in
extract as well as in essential oil. Microbial susceptibility tests confirm potential use of S. montana supercritical CO2 extracts in the food and pharmaceutical
industry.
Acknowledgements
The authors wish to thank Dr. Darren Westerman, School of Engineering, Chemical Engineering,
University of Birmingham for assistance during experimental work.
REFERENCES
[1]
T.L.C. de Oliveira, R. de Araújo Soares, E.M. Ramos, M.
Cardoso, E. Alves, R.H. Piccoli, Int. J. Food Microbiol.
144 (2011) 546–555
[2]
A.R. Gohari, S. Saeidnia, M.R. Gohari, F. Moradi-Afrapoli, M. Malmir, A. Hadjiakhoondi, Nat. Prod. Res. 23
(2009) 1609-1614
[3]
N. Bezić, M. Skočibušić, V. Dunkić, Acta Bot. Croat. 64
(2005) 313–322
[4]
G.S. Ćetković, A.I. Mandić, J.M. Čanadanović-Brunet,
S.M. Đilas, V.T. Tumbas, J. Liq. Chromatogr. Relat.
Technol. 30 (2007) 293–306
[5]
F.V.M. Silva, A. Martins, J. Salta, N.R. Neng, J.M.F.
Nogueira, D. Mira, N. Gaspar, J. Justino, C. Grosso, J.S.
Urieta, A.M.S. Palavra, A.P. Rauter, J. Agric. Food Chem.
57 (2009) 11557–11563
[6]
J. Vladić, Z. Zeković, A. Cvejin, D. Adamović, S. Vidović,
Separ. Sci. Technol. (Philadelphia, PA, U. S.) 49 (2014)
2066-2072
[7]
T. Mihajilov-Krstev, D. Radnović, D. Kitić, V. Stankov-Jovanović, V. Mitić, Z. Stojanović-Radić, B. Zlatković, Cent.
Eur. J. Bio. 7 (2014) 668-677
CONCLUSION
In this study, it was found that different matrix
pretreatment of S. montana leaves, namely mechanical grinding, FDS, FDL treatment and physical
disruption of EOs glands by mechanical compression
are effective for the supercritical CO2extraction of
EOs. However, FDS (fast decompression of super-
208
Chem. Ind. Chem. Eng. Q. 22 (2) 201−209 (2016)
[8]
S. Burt, Int. J. Food Microbiol. 94 (2004) 223–253
[9]
K. Araus, E. Uquiche, J. M. del Valle, J. Food Eng. 92
(2009) 438–447
[10]
S. Combrinck, G.W. Du Plooy, R.I. McCrindle, B.M.
Botha, Ann. Bot. (Oxford, U. K.) 99 (2007) 1111-1119
[11]
E. Werker, E. Putievsky, U. Ravid, N. Dudai, I. Katzir,
Ann. Bot. (Oxford, U. K.) 71 (1993) 43-50
B. DAMJANOVIĆ-VRATNICA et al.: EFFECT OF MATRIX PRETREATMENT…
Chem. Ind. Chem. Eng. Q. 22 (2) 201−209 (2016)
[12]
I. Žižović, M. Stamenić, A. Orlović, D. Skala, Chem. Ind.
Chem. Eng. Q. 12 (2006) 164-167
[25]
S. Ćavar, M. Maksimović, M.E. Šolic, A.J. Mujkić, R.
Bešta, Food Chem. 111 (2008) 648–653
[13]
I. Žižović, M. Stamenić, A. Orlović, D. Skala, Chem. Eng.
Sci. 60 (2005) 6747-6756
[26]
J. Mastelić, I. Jerković, Food Chem. 80 (2003) 135–140
[27]
D. Bezbradica, J. Tomović, M. Vukasinović, S. Šiler–Marinković, M. Ristić, J. Essent. Oil Res. 17 (2005) 462–465
[14]
F. Meyer, P. Jaeger, R. Eggers, M. Stamenić, S. Milovanović, I. Žižović, Chem. Eng. Process. 56 (2012) 37–45
[15]
F. Gaspar, R. Santos, M.B. King, Ind. Eng. Chem. Res.
39 (2000) 4603-4608
[28]
F. Bakkali, S. Averbeck, D. Averbeck, M. Idaomar, Food
Chem. Toxicol. 46 (2008) 446–475
[16]
F. Gaspar, G. Leeke, J. Essent. Oil Res. 16 (2004) 64-68
[29]
[17]
J. Ivanović, I. Žižović, S. Petrović, D. Skala, Chem. Ind.
Chem. Eng. Q. 15 (2009) 271-278
V. Slavkovska, R. Jančić, S. Bojović, S. Milosavljević, D.
Đoković, Phytochemistry 57 (2001) 71-76
[30]
[18]
M. Stamenić, I. Žižović, R. Eggers, P. Jaeger, H. Heinrich, E. Roj, J. Ivanović, D. Skala, J. Supercrit. Fluids 52
(2010) 125–133
F. Gaspar, R. Santos, M.B. King, J. Supercrit. Fluids 21
(2001) 11-22
[31]
E. Reverchon, AIChE J. 42 (1996) 1765–1771
[32]
[19]
S. Vidović, Z. Zeković, B. Marošanović, M. Pandurević–Todorović, J. Vladić, J. Supercrit. Fluids. 95 (2014) 468–
–473
E.M.C. Reis-Vasco, J.A.P. Coelho, A.M.F. Palavra, C.
Marrone, E. Reverchon, Chem. Eng. Sci. 55 (2000) 2917–
–2922
[33]
[20]
National Committee for Clinical Laboratory Standards.
Performance standards for antimicrobial susceptibility
th
testing, 9 ed., Approved standard M100-S9. Wayne, PA,
USA, NCCLS, 1999, pp. 23–26
S.A. Aleksovski, H. Sovová, J. Supercritic. Fluids 40
(2007) 239-245
[34]
T. Fornari, A. Ruiz-Rodriguez, G. Vicente, E. Vázquez, M.
R. García-Risco, G. Reglero, J. Supercritic. Fluids. 64
(2012) 1-8
[35]
T. Fornari, G. Vicente, E. Vázquez, M. R. García-Risco,
G. Reglero, J. Chromatogr. A. 1250 (2012) 34-38
[36]
C.T. Murthy, N. Krishnamurthy, T. Ramesh, P.N. Srinivasa Rao, J. Food Sci. Tech. Mys. 33 (1996) 299-301
[37]
S.C. Chao, D.G.Young, C.J. Oberg, J. Essent. Oil Res.
12 (2000) 639–649
[38]
A. Ultee, E.J. Smid, Int. J. Food Microbiol. 564 (2001)
373-378
[39]
C.M. Mann, S.D. Cox, J.L. Markham, Lett. Appl. Microbiol. 30 (2000) 294–297
[40]
C. Serrano, O. Matos, B. Teixeira, C. Ramos, N. Neng, J.
Nogueira, M. L. Nunes, A. Marques,J. Sci. Food. Agric.
91 (2011) 1554–1560.
[21]
National Committee for Clinical Laboratory Standards.
Method for Antifungal Disk Diffusion Susceptibility Testnd
ing of Yeasts, 2 ed., Approved Guideline M44-A.Wayne,
PA, USA, NCCLS, 2004
[22]
M. Skočibušić, N. Bežić, J. Essent. Oil. Res. 16 (2004)
387–391
[23]
C. Grosso, A.C. Figueiredo, J. Burillo, A.M. Mainar, J.S.
Urieta, J.G. Barroso, J.A. Coelho, A.M.F. Palabra, J. Sep.
Sci. 32 (2009) 328-334
[24]
J. Rzepa, M. Sajewicz, T. Baj, P. Gorczyca, M. Włodarek,
K. Głowniak, M. Waksmundzka-Hajnos, T. Kowalska,
Chromatogr. Res. Int. 1 (2012) 1-8
BILJANA
DAMJANOVIĆ-VRATNICA1
2
SVETLANA PEROVIĆ
3
TIEJUN LU
REGINA SANTOS3
1
Metalurško-tehnološki fakultet, Univerzitet Crne Gore, 81000 Podgorica,
Crna Gora
2
Prirodno-matematički fakultet, Univerzitet Crne Gore, 81000 Podgorica,
Crna Gora
3
School of Chemical Engineering,
College of Engineering and Physical
Sciences, University of Birmingham,
Edgbaston, United Kingdom
NAUČNI RAD
UTICAJ PREDTRETMANA BILJKE Satureja montana
NA ESKTRAKCIJU ETARSKOG ULJA
NATKRITIČNIM CO2
Ispitan je uticaj različitih predtretmana vrijeska (Satureja montana L.) na ekstrakciju etarskog ulja natkritičnim CO2 – prinos, hemijski sastav i antimikrobne aktivnosti dobijenih
ekstrakata i etarskog ulja. Matriks biljke je, prije ekstrakcije, podvrgnut konvencionalnom
mljevenju, fizičkom razaranju ćelija brzom dekompresijom natkritičnog i supkritičnog CO2
kao i fizičkom razaranju mehaničkom kompresijom. Analiza etarskog ulja dobijenog natkritičnom CO2 ekstrakcijom i hidrodestilacijom je urađena GC/FID metodom. Glavne komponente etarskog ulja vrijeska, koji je dobijeno natkritičnom CO2 ekstrakcijom i hidrodestilacijom, su: timol (30,4-35,4 i 35,3%), karvakrol (11,5-14,1 i 14,1%), i p-cimen (8,3-10,1 i
8,6%), redom. Rezultati ispitivanja ukazuju da je najveći prinos etarskog ulja, kao i najveći
sadržaj timola, karvakrola i timokinona u ulju, dobijen poslije fizičkog razaranja uljnih ćelija
brzom dekompresijom CO2 u natkritičnoj oblasti (FDS). Antimikrobna aktivnost ekstrakata
dobijenih natkritičnom CO2 ekstrakcijom bila je ista (FDS) ili slabija u poređenju sa etarskim uljem vrijeska koje je dobijeno hidrodestilacijom.
Ključne reči: Satureja montana, natkritična CO2 ekstrakcija, predtretman biljnog
matriksa, prinos etarskog ulja, hemijski sastav etarskog ulja.
209
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
M. BARAHOEI1
A. ZEINOLABEDINI HEZAVE2
S. SABBAGHI3
SH. AYATOLLAHI4
1
Nano Chemical Engineering
Department, Shiraz University,
Shiraz, Iran,
mlhbarahoei@gmail.com
2
Islamic Azad University,
Dashtestan Branch, Borazjan, Iran,
zeinolabedinihezave.ali@gmail.com
3
Nano Chemical Eng. Dep., Shiraz
University, Shiraz, Iran,
Sabbaghi@shirazu.ac.ir
4
School of Petroleum and
Chemical Engineering, Sharif
University of Technology, Tehran,
Iran, shahab@shirazu.ac.ir
SCIENTIFIC PAPER
UDC 553.98:546.562-31:66:544
DOI 10.2298/CICEQ150407035B
CI&CEQ
COPPER OXIDE NANO-FLUID STABILIZED BY
IONIC LIQUID FOR ENHANCING THERMAL
CONDUCTIVITY OF RESERVOIR
FORMATION: APPLICABLE FOR THERMAL
ENHANCED OIL RECOVERY PROCESSES
Article Highlights
• Three conventional surfactants and one ionic liquid type are used as stabilizing agents
• 1-dodecyl-3-methylimidazolium chloride ([C12mim] [Cl]) was used as a new kind of
surfactant
• Thermal conductivity of base fluid was increased up to 48% using nanoparticles of
copper oxide
Abstract
Since oil reservoirs are limited and energy demand is increasing, seeking for
high efficient EOR processes or enhancing the efficiency of current proposed
EOR methods for producing trapped oil from reservoirs are highly investigated.
As a way out, it is possible to couple the EOR and nanotechnology to utilize
the efficiency of both methods together. Regarding this possibility, in the present study, in the first stage of investigation stable and uniform water-based
solution of nano-sized particles of copper oxide with different concentrations
(0.01–0.05 M) were prepared and then injected into the core samples. In the
first stage, the effects of different surfactants with respect to their concentrations were investigated. Then, different scenarios of using nano-fluid as a thermal conductivity modifier were examined. The obtained results clearly demonstrate that changing concentration of nanoparticles of copper oxide from 0.01
to 0.05 M is able to enhance the thermal conductivity of rocks from 27 to 48%
compared with the thermal conductivity of dry core.
Keywords: thermal conductivity, nano-fluid, copper oxide, enhanced oil
recovery, ionic liquids.
A considerable amount of crude oil remains
unrecoverable underground after the primary and
secondary oil recovery processes, which is the target
of more oil recovery as the demand for energy increases rapidly [1]. These techniques are gaining more
momentum as the giant oil reserves are being rapidly
depleted and the search for new oil reserves is
becoming more expensive [2]. Hence, over the past
decades, many research studies have examined
methods called enhanced oil recovery (EOR) to find
Correspondence: Sh. Ayatollahi, School of Petroleum and
Chemical Engineering, Sharif University of Technology, Tehran,
Iran.
E-mail: shahab@shirazu.ac.ir
Paper received: 7 April, 2015
Paper revised: 25 July, 2015
Paper accepted: 27 August, 2015
the most suitable methods to extract larger amount of
trapped oil from the oil reservoirs [3-10]. Among the
different EOR methods, the thermal method is one of
the most favorable methods examined over the past
three decades due to the several advantages. Thermal EOR methods are generally applicable to heavy,
viscous crudes, and involve the introduction of thermal energy or heat into the reservoir to raise the temperature of the oil and reduce its viscosity. However,
the main limitations of thermal methods are the low
thermal conductivity of the reservoir rock and consequent heat loss during the thermal process affecting low reservoir area by the heat transfer process
[11,12]. For example, Kiasari et al. [13] have reported
that as the thermal conductivity of the rock reduced,
less heat was conducted into farther distances from
steam chamber in the core.
211
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
Generally, it can be concluded that in the case
of thermal EOR processes, it is well established that
the effective thermal conductivity (ETC) not only has
a direct effect on the efficiency of the thermal EOR
processes, but also measuring and estimation of ETC
of the dry or fluid-saturated porous materials are
essential for several other applications [14-31].
In other words, accurate knowledge of ways that
can enhance the ETC for certain purposes is crucial,
since heat transfer and temperature distributions
resulting from heat conduction in the solid matrix are
the basis of most of the above-mentioned applications.
In this regard, it was a great interest of many
researchers in the world to measure the experimental
value of ETC and investigate the effect of different
parameters, which can modify this parameter toward
a desired status [32-38]. As a way out, there are two
options to eliminate these limitations: a) injecting the
steam or other heat transferring fluid into a reservoir
with shallow depth and thick pay zone or b) using
method that it is able to deliver the heat into the larger
area of reservoir [38]. Among these, since the first
one seems applicable, several methods are proposed
to modify this parameter, especially the injection of
nanoparticles that introduce unique features.
During the past ten years, nanotechnology has
paved its way as a novel technique that can contribute to more efficient, less expensive, and more environmentally friendly technologies through different
industries including oil and gas industries [39,43]. For
example, Hascakir et al. [10] added three different
iron powders including iron (Fe), ferric oxide (Fe2O3),
and ferric chloride (FeCl3) into the heavy crude oil.
They reported that not only was iron oxide able to
increase the thermal conductivities, but it was also
able to assist in decreasing the percentage of polar
component in the oil, resulting in a reduction in the
viscosity of the oil caused by hydrogen bonding reduction, consequently enhancing the oil recovery.
In addition, Onyekonwu et al. [41] reported that
it is possible to alter the wettability of rock surfaces
using three kinds of polysilicon nano-fluids including
lipophobic and hydrophilic PSNP (LHPN), hydrophobic and lipophilic PSNP (HLPN) and neutrally wet
PSNP (NWPN). They claimed that NWPN and HLPN
introduced good capability to enhance oil recovery by
two different mechanisms, namely alteration of rock
wettability and reduction of interfacial tension [41].
Furthermore, Roustaei et al. [43] have reported
that using lipophilic polysilicon and naturally wet polysilicon (NWP) nanoparticles led to a change toward
less water-wet conditions and a drastic decrease in
212
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
oil-water interfacial tension from 26.3 to 1.75 mN/m
and 2.55 mN/m, respectively. Moreover, they reported
that oil recoveries increased by 32.2 and 28.57%,
respectively, when 4 g/L HLP and NWP nano-fluids
were injected into the core samples [43].
A point worth mentioning is that although the
nano-fluids are greatly used in different areas of oil
and gas industries, especially EOR methods as aforementioned [40-47], no application of nano-fluids for
thermal EOR processes has yet been reported.
Due to this shortcoming and the possibility of
utilizing this technology in this area, in the current
work metal-based nano-fluid of copper oxide has
been examined to if it is possible to enhance the
thermal conductivity of the rocks or not. The other
novelty of this work is the application of new kind of
surfactants, namely ionic liquid (IL)-based surfactant
utilized for the first time, to the best knowledge of the
authors, for stabilizing the nano-fluid solution. The
idea behind this novelty was raised from the recent
work performed by Hezave et al. [48-50], which introduced the ILs as a new kind of surfactants for oil
industries with special focus on EOR purposes.
In brief, the applications of ILs have been crucially increased due to their unique physicochemical
properties such as high thermal stability, large liquids
range, high ionic conductivity, high solvating capacity,
negligible vapor pressure, and non-flammability which
make them a very good candidate to replace by the
conventional solvents in the different fields of chemical engineering industries [48-50].
Based on these unique advantages, the effectiveness of using copper oxide nanoparticles to
enhance the thermal conductivity of rocks was investigated. Thus, besides the effect of 1-dodecyl-3-methylimidazolium chloride [C12mim] [Cl] on the
stability of the nano-fluid solution, the effects of different parameters including nanoparticles concentration in the range of 0.01-0.05 M were investigated.
EXPERIMENTAL
Materials
Copper oxide particles were supplied from
Merck, Germany with purity of >99.99%. The ionic
liquid under the name of 1-dodecyl-3-methylimidazolium chloride ([C12mim] [Cl]) was synthesized as
previously described [50]. In brief, 1-methylimidazole,
1-chlorododecyl and diethyl ether were supplied from
Merck/Fluka and used without any further purification.
According to the previously reported procedure the IL
1-dodecyl-3-methylimidazolium chloride ([C12mim] [Cl])
was synthesized by reacting 1-methylimidazolium
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
with excess amount of the 1-chlorododecane without
any additional solvent in a round bottomed flask fitted
with a reflux condenser (heating and stirring at 70 °C
for 48-72 h). The resulting viscous liquid was cooled
to room temperature and was washed by diethyl
ether. After drying overnight at 100 °C, the purity of
the products was assessed by HNMR. In addition, the
used cores in this investigation were prepared from
outcrop rocks of southern Iran cores for core-flooding
tests were prepared from samples in a similar formation. The majority content of the rocks was recognized to be dolomite (see Table 1).
Table 1. Properties of the used cores; length: 3.0 cm, diameter:
7.0 cm
No
Porosity (φ)
Permeability, mD
1
19.5
101.1
2
20.6
117.7
3
16.3
141.0
4
20.4
138.8
5
18.2
48.6
Preparation of nano-fluid
In the first stage of this study, copper oxide
nano-fluids with different concentrations were prepared to find the effect of copper oxide nanoparticles
on the thermal properties of the core. It is worth
mentioning that two different base fluids were used to
prepare the nano-fluid – water and ethylene glycol. In
this way, ionic liquid as a surface-active agent (surfactant) was used to stabilize the nanoparticles of copper oxide in the solution by modifying the scattering
factor. In fact, a layer that was coated around the
nanoparticles with the surfactant caused electrostatic
repulsion between double layer particles, so the
nanoparticles were dispersed into the base fluid.
Because of this phenomenon, adhesion and aggregation of particles was prevented.
IL was used as a surfactant since it has been
proven that this kind of surfactant is able to tolerate
harsh temperature conditions without any breakdown,
decomposition and degradation at high temperatures,
while common surfactants such as C16TAB corrupted
and lost their functionality at temperatures above 40
°C. In the light of above facts, IL seems a suitable
candidate for stabilizing nano-fluids for thermal applications. After finding the optimum surfactant concentration, optimum pH and exposure time solution under
ultrasonic irradiation, consistent nano-fluid copper
oxide was synthesized. Finally, the stability of prepared nano-fluids was examined by performing sedimentation tests for at least one month to investigate
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
the possible aggregation of the particles [51]. To
investigate the effects of other surfactants in thermal
conductivity of core, CuO nano-fluids were prepared
with different kinds of surfactants and then injected to
core samples. The C16TAB, SDS and PVP were used
for this purpose. After that, the CuO nano-fluids with
several kinds of surfactants (PVP, SDS, C16TAB and
IL) at different concentrations were synthesized and
then their stability and thermal influences were compared to each other. Moreover, the experiments were
designed in way that allowed investigating the effects
of base fluids, including water and ethylene glycol, in
preparing the nano-fluid as the nano-fluid IL was used
as the stabilizer.
Thermal conductivity measurement
After measuring the porosity (He-Porosimeter
32351, Vinci, France) and permeability (Coreval700
32372, Vinci, France) of cores, In order to measure
the thermal conductivity of rocks with different status
(dry, water-saturated and nano-fluid saturated) a
homemade thermal conductivity measurement apparatus worked based on the steady state method was
used (Figure 1). This apparatus designed in a way
that allowed measuring the thermal conductivity at
ambient pressure and elevated temperature up to 180
°C. A brief description of the used apparatus is as
follows. At first, the core was placed into the core
holder made of polytetrafluoroethylene (PTFE). The
core holder consisted of three different sections
demonstrated in Figure 1. As it is clear in this figure,
there is a section made of copper, which was used as
a reference for calculation of heat transmission.
Further, the apparatus was equipped with four different temperature sensors controlled by PID protocol
with accuracy of 0.1 °C to measure the temperature
through the apparatus (reference section and core
plug). In more detail, these four sensors as shown in
Figure 1 were used to measure the temperature of
beginning and end of reference copper and core plug.
The required heat in this apparatus was generated
using three electric heaters symmetrically placed
before the reference section made of copper. Since
there were two different kinds of materials (copper as
a reference section and core plug as the under investigation section) that are contacted to each other,
contact resistance had a significant role, which can
face the experiments with large order of error. To
overcome this problem, a silicon thermal compound
slot was used to minimize the resistance between two
different sections. The effectiveness of the used silicon thermal compound slot was evaluated by measuring the temperature of two sides. In addition, con-
213
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
Figure 1. Thermal conductivity measurement apparatus.
struction of the core holder using PTFE minimized the
heat loss through the system by reducing the convective heat transfer. It should be noted that that
besides the internal temperature sensors, four external temperature sensors have been used to ease
calculation of radial heat loss. On the other hand,
recording ambient temperature fluctuations is a crucial parameter necessary for calculating the thermal
conductivity.
Prior to any test, the accuracy of the data was
evaluated by measuring the thermal conductivity of
known samples such as aluminum and copper rods.
The measured thermal conductivities revealed that
there was less than 3% difference between the
results of the obtained thermal conductivities and
those reported in literature (Table 2).
Table 2. Thermal conductivity of the calibration set
Sample
Length Thermal conductivity, Calculated thermal
mm
W/mK [53]
conductivity, W/mK
Copper
48.5
400
405.8
Aluminum
72.4
250
246.67
Considering the measured temperatures and
utilizing Fourier’s law (Eq. 1) it is possible to calculate
the rock thermal conductivity as:
Q = KA ΔT
(1)
where K is the thermal conductivity (W/mK), A is the
area (m2) and ΔT is the temperature difference (°C). If
the above equation utilizes for constant heat transfer
between the copper section and core section, it is
possible to calculate the core thermal conductivity as
follows:
214
Q = K Cu ACuΔTCu
k Core =
Q
ACore ΔTCore
(2)
(3)
where KCu is the copper conductivity, ACu is the
surface area of copper cylinder faced with heat transfer, ΔTCu is the temperature difference of two faces of
the copper cylinder, kCore is core conductivity, ACore is
the surface area of core and ΔTCore is the temperature
difference of two faces of the core.
Core flooding procedures
In this study, the core-flooding apparatus (Figure
2) was used to flood the core plugs with the prepared
nano-fluid solutions to measure the conductivity of the
treated cores.
The core flooding system, previously described
in detail [48], was used to conduct the core flooding
experiments. In brief, after porosity and permeability
measurements, the core plugs were placed in the
core-holder. Then, the cores were saturated by injecting several pore volumes (PV) of nano-fluids at the
injection rate of 0.3 mL min-1 while the effluent solution was monitored. It is noteworthy that at the start
of injection stage, the effluent was completely clear.
In other words, at first it seemed that the core plug
was acting like a filter, but after injecting several pore
volumes of nano-fluid, the same effluent was produced, which indicated that the rock was fully saturated with the nano-fluids and further injection of nanofluid into the core could be halted. Then, the saturated
core was placed into the thermal conductivity measurement apparatus for further processing. Figure 3
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
shows the status of the effluent nano-fluid solutions at
different time periods.
Figure 2. Schematic of the used core flooding apparatus
(1 - HPLC pomp, 2 - chemicals cylinder,3 - brine cylinder, 4 - oil
cylinder, 5 - confining pressure, 6 - oven, 7 - core holder,
8 - outlet fluids tube, 9 - pressure transducer).
a)
b)
c)
d)
Figure 3. a) Fluid inlet; b) effluent after injection of 5 PV;
c) effluent after 10 PV; d) effluent after 13 PV.
RESULTS AND DISCUSSION
Effect of nano-fluid injection on thermal conductivity
of core
In the first stage, the effect of nano-fluid injection
on the enhancement of thermal conductivity was
examined using copper oxide as nanoparticle at a
concentration of 0.01 M. Then, the obtained results
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
were compared with the thermal conductivity of dry
core and saturated core by distilled water. The
obtained results demonstrated in Figure 4 revealed
that injection of water can reduce the temperature
difference along to the core to the value of 12.4 °C,
which means higher thermal conductivity of rock
compared to dry status (13.1 °C). This obtained result
can be explained based on the fact that the dry core
utilizes the porous media occupied by air, which acts
as an insulation, thereby reducing the thermal conductivity of core. However, if the dry core is saturated
with distilled water, the air will be replaced by distilled
water, which has a higher thermal conductivity compared to air; consequently, lower temperature difference between the two sides of the core will be observed.
Also, measuring the temperature difference of
two sides of the core saturated with nano-fluids of
copper oxide revealed that injection of nanoparticles
had a great effect on the thermal conductivity compared with two previously examined statuses. In other
words, injection of nano-fluid of copper oxide into the
core reduced the temperature different from to 10.1
°C, which is lower than 13.1 °C (for dry core) and 12.4
°C (for core saturated with distilled water). This reduction can be related to the presence of metallic
based particles in the fluid which can be oriented into
the pores in a way that increases the thermal conductivity by making a continuous conductive path.
Also, a closer examination in Figure 4 revealed
that the temperature variation profile is consisted of
two different areas named A and B. The B section is
the steady zone while the A section is a zone which
the temperature variation experienced a fluctuation
can rise from the two different parameters including
movement and settlement of the particles and the
steadiness of transferred heat into the core. Finally, to
observe the sole effect of nanoparticles on the thermal conductivity of the core, nano-fluid-saturated core
was dried at ambient temperature for a week. After
that, the temperature difference along to the core was
measured again and it was found that the presence of
nanoparticles in the core can reduced the temperature difference about 1 °C, which was 8% higher than
the temperature difference when only water was
injected into the core.
The interesting point is that measuring the temperature difference between the two sides of the core
(dried after saturating by distilled water) revealed temperature variation similar to when the core was dry
from the beginning. However, in the nano-fluid saturated case, after evaporation of base fluid during one
week, the particles remained on the surface of the
215
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
Figure 4. Temperature difference variation along the core for different cases.
rock, consequently decreasing the temperature difference compared to the dry core state.
Based on the obtained results, it can be concluded that the highest thermal conductivity enhancement can be achieved while the nanoparticles and
base fluid (distilled water) injected concurrently.
In the next stage of this investigation, the
obtained results of temperature variations converted
to thermal conductivity demonstrated an interesting
behavior. A close examination of the results depicted
in Figure 5 revealed a transient behavior of thermal
conductivity can be related to the existence of vapor
in the core holder chamber produced by introducing
heat into the core sample.
In more detail, because of time passing, vapors
started to be produced under the effect of heat passing through the system. In light of the produced
vapor, the thermal conductivity experienced a fluctuation, which faded away after a while, when the produced vapor escaped from the system. The results
depicted in Figure 5 show that the presence of nano-
particles of copper oxide significantly increased the
thermal conductivity of rock up to 33%. As aforementioned, this trend can be related to the fact that
the injected fluid enters into the core pores and produces a continuous conductive pass full of nano-fluid
copper oxide, increasing the thermal conductivity of
the core toward a more conductive state.
The calculated thermal conductivity of different
cases including dry core, water-saturated core and
nano-fluid-saturated core are given in Table 3. As can
be seen, injection of only distilled water enhances the
thermal conductivity of core about 6% while injection
of copper oxide nano-fluid solution with concentration
of 0.01 M enhances the thermal conductivity up to
26%. In other words, injection of copper oxide nanofluid even with low concentration of 0.01 M can significantly affect the thermal conductivity of the rock.
Effect of nano-copper oxide particles concentration
on thermal conductivity
In the last section of this study, the effect of
copper oxide nanoparticles on the thermal con-
Figure 5. Changes of sample thermal conductivity compares to base case (dry) as a function of time for 4 different cases.
216
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
ductivity of core was investigated. In this regard, the
concentration of nano-fluid of copper oxide was
ranged between 0.01 to 0.05 M (0.01, 0.02, 0.03, 0.04
and 0.05 M) while the other operational conditions
including pH, sonication time, temperature and injection rate were kept constant.
Table 3. Thermal conductivity of rock at injection of copper
oxide nano-fluid solution with concentration of 0.01 M
Thermal
Enhancement on the
conductivity, W/mK thermal conductivity, %
Test
Dry core
1.5
-
Water saturated
core
1.59
6
Nano-fluid
saturated core
1.88
25.91
The obtained results demonstrated in Figure 5
revealed that as the concentration of copper-oxide
nanoparticles increased from 0.01 to 0.05 M, the final
temperature difference between the two ends of the
core decreased from 10.1 to 8.8 °C, corresponding to
47% enhancement in thermal conductivity. It should
be mentioned that as the concentration of nano-fluid
increases, the required time for stabilization of temperature difference decrease, resulting in higher effectiveness of nano-fluid injection at higher concentration.
The reason behind this trend can be related to the
fact that as the concentration of the nanoparticles
increases, a denser continues convective path for
heat transfer through the porous media will be produced, therefore faster and higher heat transfer
through the core occur. In addition, comparing the
thermal conductivity of the dried cores after nano-fluid
injection (see Figure 6) revealed that although the
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
thermal conductivity of the samples for all of the concentrations were lower than the saturated cores by
nano-fluids, they leads to the higher thermal conductivity compared with thermal conductivity of cores
saturated only by distilled water. Finally, the obtained
thermal conductivities enhancement for different
cases were demonstrated in Figure 6 revealed that
thermal conductivity of nano-fluid saturated cores
were higher compared with those cores which were
dried after nano-fluid injection.
The point is completely obvious is that injection
of nano-fluid solution in both cases of before and after
drying leads to a significant increase in the thermal
conductivity enhancement compared with saturated
core with distilled water. In more details, when the
core was dried for one week at room temperature, the
thermal conductivity of the core reduced since water
was evaporated from the pores and the continuity of
the conductive path was reduced compared with the
initial status. However, the dried saturated core with
nano-fluid was still more conductive than the core
saturated by distilled water, due to the considerable
effect of copper oxide nanoparticles on the thermal
conductivity considering its metallic nature.
A closer examination of Table 4 reveals an interesting point that the percentage of reduction in the
thermal conductivity after drying of core saturated by
nano-fluid solution is equal to the increase of core
thermal conductivity if saturated by distilled water. In
other words, it can be concluded that thermal conductivity of core experienced an increase of about 6%
in the presence of water, which is similar to the percentage of thermal conductivity reduction if the
injected nano-fluid was dried. This trend was consistently observed for all of the examined cores with
Figure 6. Changes in thermal conductivity increases as a function of nanoparticle concentration of the injected fluid.
217
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
Table 4. Percentage of enhancement of cores’ thermal conductivity (W/mK)
Concentration of Nano-fluid, M
Water saturated, %
Nano-saturated, %
0.01
7.69
25
20
0.02
8
33.5
28.9
0.03
5
32.94
24
0.04
7.14
40.9
35
0.05
6.9
47.82
42.8
different concentrations of copper oxide nanoparticles. Hence, it can be concluded that the great portion of the thermal conductivity enhancement during
injection of nano-fluid solution into the core can rise
from the presence of the metallic-based nanoparticles
of copper oxide.
Finally, SEM analysis was used to find the effect
of the injected nanoparticles of copper oxide on the
surface alteration of the core samples. As it can be
seen from Figure 7, the injected nanoparticles of the
copper oxide are deposited on the surface of the rock,
and thus increase the thermal conductivity of rock.
Figure 7. SEM image of core saturated with CuO nano-fluid:
a) blank sample; b) sample treated with nano-fluid.
Effect of the base fluid on the thermal conductivity of
the core
In the previous step, the effect of water-based
CuO nano-fluid was investigated and the results was
determined. But the other crucial parameter influ-
218
Dried after 1 week, %
ences the effectiveness of the nano-fluids is the used
base fluid. Regarding this fact, in the current section
the effect of using another fluid, namely ethylene
glycol, was investigated. For this purpose, ethylene
glycol-based CuO nano-fluid was prepared and
injected in to the core and the temperature differences between its two end caps was measured. The
obtained results demonstrated that the used nanofluid decreases the temperature differences and increased thermal conductivity of core. However, the
obtained temperature differences using ethylene glycol were higher than the values obtained the waterbased fluid. This means the ethylene glycol-based
CuO nano-fluid caused less increase in thermal conductivity compared to the water-based CuO nano-fluid.
As it is clear from Figure 8, preparing the solution using ethylene glycol reduces the temperature
difference about 1.4 °C and enhances the thermal
conductivity about 15%. Comparing the obtained
results by the experiments performed with water and
those performed with ethylene glycol, one can conclude that ethylene glycol is not only unable to modify
the thermal conductivity, but also reduces the thermal
conductivity to some extent. The reason behind this
observed trend can be related to the lower thermal
conductivity of ethylene glycol compared with water.
In other words, since the thermal conductivity of
ethylene glycol is lower than water, injection of solution with ethylene glycol reduces the thermal conductivity. Based on the obtained results in this section, it can be concluded that in the way of choosing
the best base fluid for preparation of the nano-fluid,
one must consider the thermal conductivity of the
base fluid. In other words, higher thermal conductivity
of base fluid leads to better results for preparing the
nano-fluid for thermal conductivity enhancement
purposes.
According to Figure 9, the lower increase in
thermal conductivity of 17% in the case of ethylene
glycol can be due to lower thermal conductivity of
ethylene glycol, as compared to water.
One part of the increase in thermal conductivity
is caused by the fluid and depends on the thermal
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
Figure 8. Temperature difference variation along the core for different cases in ethylene glycol-based fluid.
Figure 9. Changes of sample thermal conductivity compare to base case (dry) as a function of time for 4 different cases in ethylene
glycol-based fluid.
conductivity of the base fluid. So this part is lower in
the case of ethylene glycol because of its lower thermal conductivity [53]. Finally, a comparison with the
use of water-base fluid, ethylene glycol is more
appropriate and enhances the heat transfer increases.
Effect of surfactant type on stability of nano-fluids and
the thermal conductivity of core
Four different surfactants, namely ionic liquid,
SDS (sodium dodecyl sulfate), C16TAB (cetyl trimethylammonium bromide) and PVP (polyvinylpyrrolidone) were used as stabilizer agents.
After preparation of the solutions, they were
injected into the cores and the effects of the solutions
were investigated through two stages. In the first
stage, the effect of injected nano-solution on the
thermal conductivity of rock samples was investigated
spontaneously after injection (see Figure 9). A glance
into Figure 9 reveals that injection of nano-CuO fluid
into cores leads to an increase in the thermal conductivity of core samples for all of the used surfactants. This observed trend can be related to the presence of the nanoparticles of CuO occupying the
pores, which enhances the effective path for heat
transfer.
A closer examination in Figure 10 also demonstrated that although all of the used surfactants
enhance the thermal conductivity of core samples, the
ionic liquid introduced a significant effect on the thermal conductivity compared with the other used surf-
219
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
Figure 10. Changes of sample thermal conductivity compare to base status (dry) as a function of time for 4 different kind of surfactants
after nano-fluid injection.
actants. In other words, utilization of IL had the highest effect on the thermal conductivity enhancement.
The reason for this observed trend arises from
the structure of the ionic liquid. The IL used in this
study comprises of a hydrophilic section with a benzene ring, which distribute the electrical charges
through the IL molecules. Due to this distribution, the
surfactant molecules have many opportunities to orient on the surface of the nanoparticles of copper
oxide in order to introduce proper repulsion force
between the nanoparticles. Because of these repulsion forces, double layers form, enhancing the stability of nanoparticles suspension in the solution. Due
to the produced repulsion forces, agglomeration and
coagulation of nanoparticles are reduced to a level
that considerably increases stability [54].
Furthermore, the required concentration of IL for
efficient stabilization is significantly lower than the
conventional surfactant utilized in the current study
including PVP, C16TAB and SDS. The reason of this
observed trend can be also be related to the structure
of the IL which enhances the active sites for occupation by nanoparticles with no steric hindrance,
consequently increasing the stability of the nano-fluid
and heat transfer capability (see Figure 11).
According to Figure 10, between SDS and
C16TAB, SDS (12 carbons) is more efficient for increasing the thermal conductivity since its carbon
chain is shorter than C16TAT (16 carbons). As the
carbon chain of surfactant increases the steric hindrance, it thus reduces the effectiveness of the nanofluid for thermal conductivity modification.
220
Figure 11. Schematic of nanoparticles coating by different kind
of surfactants.
Among all of the examined surfactants, PVP
exhibited the poorest results, which can rise from its
different structure. In more details, PVP is a polymerlike molecule that surrounds the nanoparticles and
introduces a shielding effect on them; consequently,
the heat transfer capability via nanoparticles reduces.
Moreover, it can be seen in Figure 10 that, compared to other surfactants used in this study, PVP is
the last in ranking of increasing heat transfer. This is
because of different its structure and its ability to
create a layer around the nanoparticles, strengthening the double-layer repulsion between the nanoparticles.
On the other hand, the polymer layer acted as
an insulator layer, in addition to reducing nano-sized
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
active participation, and prevented the formation of a
conductive surface for heat transfer. Therefore, the
heat transfer performance of PVP was the weakest
relative to other surfactants.
The other investigated case was the sole effect
of nanoparticles on the thermal conductivity. For this
purpose, after the injection of nano-fluid into the core
samples, the core was left at the room temperature to
let the water evaporate. After that, the thermal conductivity was measured, which can be considered as
the sole effect of nanoparticles. The obtained results
based on this procedure revealed that the thermal
conductivity of dried samples was lower than in the
case when the system was flooded by nano-fluid. In
other words, drying the cores leads to lower thermal
conductivity. This observed trend can be related to
the fact that as the water evaporated, the effect of
base fluid (water) on the thermal conductivity enhancement was eliminated, which reduced the overall
thermal conductivity enhancement. On the other
hand, the nanoparticles settled on the rock surfaces
and the role of surfactants will be motivated. Also, IL
due to its structure and low concentration, results in a
lower steric hindrance, which leads to higher available
active sites for orientation of nanoparticles, modifying
the thermal conductivity.
Figure 12 shows results after injection of nanofluids that have been prepared with different surfactants. The core is given time to dry about one week,
which allowed the tangible effects of nanoparticles to
be exhibited. The results showed that after a week of
drying, the heat transfer process was unchanged, but
the percentage increase in thermal conductivity compared to the previous state (immediately after injection) was lower. This is due to the evaporation of the
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
base fluid (water). On the other hand, nanoparticles
are adsorbed onto the surface of core and the effect
of surfactants on the absorption will be highlighted.
In addition, IL due to the mobility of charge on its
hydrophilic head and its lower required concentration,
does not prevent the nanoparticles from participating
in the heat transfer surface. Therefore, a greater area
of the nanoparticles is in contact with the rock and
increases the heat transfer.
SDS possesses concentrated electrical charges
on its hydrophilic head, which increases its required
concentration for a certain thermal conductivity enhancement compared with the other surfactants. In addition, C16TAB has a low efficiency due to its long chain,
which increases the steric hindrance for nanoparticles
and reduces the active surfaces necessary for interaction between nanoparticles and the rock surfaces.
Lastly, utilization of PVP produces a layer around the
nanoparticles, which reduces the thermal conductivity
(see Figure 13).
Effect of concentration of CuO nanoparticles
regarding different kind of surfactant on the
thermal conductivity of core
In this section, the functionality of thermal conductivity to concentration and type of surfactants has
been investigated (see Figure 14). As it is obvious,
among the used surfactants, IL showed the best
results for enhancing the thermal conductivity of rock.
On the other hand, it can be observed that as the
concentration of nanoparticles increases, the thermal
conductivity increases, while an increase in the concentration of surfactant leads to a reduction of thermal
conductivity. This observed trend can be related to
two different phenomena. As the concentration of
Figure 12. Changes of sample thermal conductivity compare to base status (dry) as a function of time for
4 different kind of surfactants for dried core.
221
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
Figure 13. Schematic of nanoparticles adsorbed on the surface of the core with different kind of surfactants.
Figure 14. Changes of sample thermal conductivity compare to base status (dry) as a function of time for 4 different kind of surfactants
after injection of nano-fluid solution.
nanoparticles increases, the active surface for heat
transfer related to the nanoparticles increases, while
as the concentration of the surfactant increases, the
steric hindrance increases, which reduces the functionality of the nano-fluid. In other words, it is the net
effect of these two competing factors that dictates
whether the thermal conductivity increases or decreases.
A close examination of Figure 14 revealed that
these two competing factors are more obvious for
PVP surfactant, while they can barely be realized for
IL. In conclusion, it is obvious that the PVP surfactant
introduces no significant change on the thermal
conductivity.
In the next step, the cores saturated by CuO
nano-fluid were dried for one week at ambient conditions. The results of these experiments (Figure 15)
revealed that after one week, the increase in thermal
conductivity was the highest for IL and the lowest for
PVP. However, in these conditions, tangible effects of
Figure 15. Changes of sample thermal conductivity compare to base status (dry) as a function of time for 4 different kind of surfactants
as the cores were dried after one week.
222
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
nanoparticles and surfactant are seen after vaporization of the base fluid. According to Figure 15, increasing the concentration of nano-fluid prepared by
PVP has a slight effect on the thermal conductivity.
CONCLUSION
In the present study, a homemade thermal conductivity measurement apparatus was used to investigate the effect of copper oxide nanoparticles on the
thermal conductivity of the core. Different solutions of
copper oxide nanoparticles, using 1-dodecyl-3-methylimidazolium chloride ([C12mim] [Cl]), sodium dodecyl
sulfate (SDS) C16TAB and PVP as stabilizing agents,
were used to prepare nano-fluids necessary for injecting into the core. The obtained results revealed that
injection of copper oxide nanoparticles prepared in
water with different concentrations of 0.01 to 0.05 M
was able to enhance the thermal conductivity of the
cores up to 48%. In addition, the obtained results
revealed that the thermal conductivity of core saturated by distilled water is significantly lower than the
thermal conductivity of a core saturated by nano-fluid
of copper oxide, since the presence of metallic base
particles of copper oxide prove a conductive path
through the core by orienting in the pores. In addition,
the results revealed that type of base fluid for preparation of the nano-fluid is a crucial parameter for
enhancing the thermal conductivity.
Finally, based on the obtained results it can be
concluded that injection of metallic-base nanoparticles of copper oxide is an applicable and feasible
method to enhance the thermal conductivity of cores
especially since stabilized by 1-dodecyl-3-methylimidazolium chloride ([C12mim] [Cl]), which is a new kind
of surfactant that can tolerate harsh temperature
conditions. Although the results in this study revealed
the potential of copper oxide nanoparticles for thermal
EOR methods, the application of nanoparticles needs
further systematic investigations.
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
Fractured Porous Media, Spec. Topics Rev. Porous
Media — Int. J. 1(2) (2010) 179–191
[4]
M. Lashkarbolooki, A. Zeinolabedini Hezave, Artificial
neural network As an applicable tool to predict the binary
heat capacity of mixtures containing ionic liquids, J. Fluid
Phase Equilib. 324 (2012) 102–107
[5]
A. Mohsenzadeh, M. Nabipour, S. Asadizadeh, M. Nekouie, H. Ameri, Sh. Ayatollahi, Experimental Investigation
of Different Steam Injection Scenarios during SAGD
Process, J. Spec. Topics Rev. orous Media (STRPM)
2(4) (2011)
[6]
A. Zeinolabedini Hezave, S. Dorostkar, S. Ayatollahi, M.
Nabipour, B. Hemmateenejad, Investigating the effect of
ionic liquid (1-dodecyl-3-methylimidazolium chloride
([C12mim] [Cl])) on the water/oil interfacial tension as a
novel surfactant, Colloids Surfaces, A: Physicochem.
Eng. Aspects 421 (2013) 63-71
[7]
A. Zeinolabedini Hezave, S. Dorostkar, S. Ayatollahi, M.
Nabipour, B. Hemmateenejad, Dynamic interfacial tension behavior between heavy crude oil and ionic liquid
solution
(1-dodecyl-3-methylimidazolium
chloride
([C12mim][Cl] + distilled or saline water/heavy crude oil))
as a new surfactant, J. Mol. Liquids 187 (2013) 83-89
[8]
P. Sarafzadeh, A. Zeinolabedini Hezave, M. Ravanbakhsh, A. Niazi, S. Ayatollahi, Enterobacter cloacae as
biosurfactant producing bacterium: Differentiating its
effects on interfacial tension and wettability alteration
Mechanisms for oil recovery during MEOR process,
Colloids Surfaces, B: Biointerfaces 105 (2013) 223-229
[9]
P. Sarafzadeh, A. Niazi, V. Oboodi, M. Ravanbakhsh, A.
Zeinolabedini Hezave, S.. Ayatollahi, S. Raeissi, Investigating the efficiency of MEOR processes using Enterobacter cloacae and Bacillus stearothermophilus
SUCPM#14
(biosurfactant-producing
strains)
in
carbonated reservoirs, J. Petroleum Sci. Eng. 113 (2014)
46-53
[10]
P. Sarafzadeh, A.i Zeinolabedini Hezave, S. Mohammadi,
A. Niazi, S. Ayatollahi, Modification of rock/fluid and
fluid/fluid interfaces during MEOR processes, using two
biosurfactant producing strains of Bacillus stearothermophilus SUCPM#14 and Enterobacter cloacae: A mechanistic study, Colloids Surfaces, B: Biointerfaces 117
(2014) 457-465
[11]
V. Vafaei, R. Eslamloueyan, L. Enfeali, Analysis and
Simulation of Steam Distillation Mechanism during Steam
Injection Process, J. Energy Fuels 23(1) (2009) 327-333
[12]
Sh. Ayatollahi, N. Saber, M. J. Amani, A. Bitaab, Mathematical Investigation of Effective Thermal Conductivity in
Fractured Porous Media, J. Porous Media 9 (2006) 625–635
[13]
H. Hashemi Kiasari, S. Sedaee; A. Naderifar, Investigation on the Effect of the Reservoir Variable and Operational Parameters on Sagd Performance, Braz. J. Petrol.
Gas 4(2) (2010) 059-069
[14]
MS. Yamano, S. Uyeda, Y. Aoki, TH. Shipley, Estimates
of heat flow derived from gas hydrates, Geology10 (1982)
339–343
REFERENCES
[1]
[2]
[3]
P. Maroufi, H. Rahmanifard, A. Jahanmiri, Experimental
Investigation of Secondary Oil Recovery in Fractured
th
Media, in Proceedings of 7 International Chemical Engineering Congress & Exihibition Kish, Iran, 21-24 November, 2011
P. Maroufi, H. Rahmanifard, M. Escrochi and H. Al-Hadhrami, Experimental Investigation of Wettability Effect and
Drainage Rate on Tertiary Oil Recovery from Fractured
Media, J. Porous Media (JPM), March 2012.
M. Rezaveisi, B. Rostami, R. Kharrat, C. Ghotbi, Experimental Investigation of Tertiary Oil Gravity Drainage in
223
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
[15]
V. Cermak, L. Rybach, Thermal conductivity and specific
heat of minerals and rocks, in: G. Angenheiser, Physical
properties of rocks, Vol. 1, Springer, Berlin, 1982, pp.
305–343
[33]
M.R.J. Wyllie, P.F. Southwick, An Experimental Investigation of the S.P. and Resistivity Phenomena in Dirty
Sands, J. Alloys Compd. 6 (1954) 44–57
[34]
[16]
K. Gallagher, Thermal conductivity of sedimentary and
basement rocks from the Eromanga and Cooper basins,
South Australia, Explor. Geophys, 18 (1987) 381–392
W. Woodside, J.H. Messmer, Thermal Conductivity of
Porous Media. I. Unconsolidated Sands, J. Appl. Phys.
32(9) (1961) 1688–1699
[35]
[17]
J.H. Schon, Physical properties of rocks: fundamentals
and principles of petrophysics, Pergamon, Oxford, 1996
W. Woodside, J.H. Messmer, Thermal Conductivity of
Porous Media. II. Consolidated Rocks, J. Appl. Phys.
32(9) (1961) 1699–1706
[18]
A.H. Lachenbruch, J.H. Sass, Heat flows in the United
State, in: J.G. Heacock, ed., The Earth’s crust, Amer.
Geophys. Union, Washington DC, 1977, pp. 626–675
[36]
[19]
J.H. Sass, A.H. Lachenbruch, R. Munroe, Thermal conductivity of rocks from measurements on fragments and
its application to heat-flow determinations, J. Geophys.
Res. 76 (1971) 3391–3401
A.D. Sullins, K. Daryabeigi, Effective Thermal Conductivity of High Porosity Open Cell Nickel Foam, in Proth
ceedings of 35 AIAA Thermophysics Conference,
Anaheim, CA, USA, June 11–14, 2001
[37]
C. Tseng, M. Yamaguchi, T. Ohmori, Thermal Conductivity of Polyurethane Foams From Room Temperature to
20 K, Cryogenics 37(6) (1997) 305–312
[38]
N. Oshima, N. Watari, The calculation of effective thermal
conductivity of media with regularly dispersed parallel
cylinders, Jpn. Soc. Mech. Eng. Int. J. 32 (1989) 225
[39]
A. Habibi, H. Al-Hadrami, A. Al-Ajmi, Y. Al-Wahaibi, Sh.
Ayatollahi, Effect of Nano-fluid Treatment on Water Sensitive Formation to Investigate water shock phenomenon,
an Experimental Study, Accepted, 9 Feb. 2013, for
Publication in Journal of Dispersion Science and
Technology.
[40]
M. Ahmadi, A. Habibi, P. Pourafshary, Sh. Ayatollahi, An
Experimental Study of Interaction between Nanoparticles’
Deposition on a Sintered Porous Medium and Migratory
Fines, J. Porous Media 16(5) (2013) 459-467
[41]
O.M. Onyekonwu, A.N. Ogolo, Department of Petroleum
and Gas Engineering, University of Port Harcourt, Rivers
State, Nigeria. Investigating the Use of Nanoparticles in
Enhancing Oil Recovery, SPE International Student
Paper Contest at the SPE Annual Technical Conference
and Exhibition being held in New Orleans, LA, USA, 2010
[42]
M. Parvazdavani, M. Masihi, M.H. Ghazanfari, M. Sherafati, L. Mashayekhi, Investigation of the Effect of Water
Based Nanoparticles Addition on Hysteresis of Oil andWater Relative Permeability Curves, Department of
Chemical and Petroleum Engineering, Sharif University of
Technology
[43]
A. Roustaei, J. Moghadasi, H. Bagherzadeh, A. Shahrabadi, An Experimental Investigation of Polysilicon Nanoparticles Recovery Efficiencies through Changes in Interfacial
Tension
and
Wettability
Alteration,
an
experimental", 12–14 June 2012.
[44]
N.A. Ogolo, O.A. Olafuyi, M.O. Onyekonwu, Enhanced
Oil Recovery Using Nanoparticles, SPE 160847,2012.
[45]
D. Shah, D. Rusheet, Application of Nanoparticle Saturated Injectant Gases for EOR of Heavy Oil", SPE129539, 2009.
[46]
M.M. Zerafat, Sh. Ayatollahi Nanotechnology-Assisted
EOR Techniques: New Solutions to Old Challenges,
SPE-157094, International Oil field Nanotechnology
Conference and Exhibitionheld in Noordwijk, The
Netherlands, 12–14 June, 2012
[47]
D. Shah, D. Rusheet, Application of Nanoparticle Saturated Injectant Gases for EOR of Heavy Oils", University
[20]
G.R. Hadley, Thermal conductivity of packed metal
powders, Int. J. Heat Mass Transfer 29 (1986) 909–920
[21]
C.T. Hsu, P. Cheng, K.W. Wong, A lumped-parameter
model for stagnant thermal conductivity of spatially periodic porous media, Trans. ASME 117 (1995) 264–269
[22]
A. Hasan, Optimising insulation thickness for buildings
using life cycle cost, in: Applied energy, Vol. 3, Elsevier,
Amsterdam, 1999, pp. 115–124
[23]
W.F. Waite, B.J. de Martin, S.H. Kirby, J. Pinkston, C.D.
Ruppel, Thermal conductivity measurements in porous
mixtures of methane hydrate and quartz sand, Geophys.
Res. Lett. 29 (2002) 1–4
[24]
A.D. Naidu, D.N. Singh, Field probe for measuring thermal resistivity of soils, J. Geotech. Geoenviron. Eng.
(2004) 213–216
[25]
O. Zkahraman, H.T. Selver, E.C. Isik, Determination of
the thermal conductivity of rock from P-wave velocity, Int.
J. Rock Mech. Min. Sci. 41 (2004) 703–708
[26]
A. Demirci, K. Görgülü, Y.S. Durutuk, Thermal conductivity of rocks and its variation with uniaxial and tri axial
stress, Int. J. Rock Mech. Min. Sci. 1 (2004) 1133–1138
[27]
J. Côte, M. Konrad, A generalized thermal conductivity
model for soils and construction materials, Can.
Geochem. J. 42 (2005) 443–458
[28]
A.J. Sievers, Solar energy conversion—solid-state,
physics aspects, in: B.O. Serahin, ed., Topics in Applied
Physics, Vol. 30, Springer, Berlin, 1979, p. 57
[29]
D.M. Staicu, D. Jeulin, M. Beauvy, M. Laurent, C. Berlanga, Effective thermal conductivity of heterogeneous
materials: calculation methods and application to different
microstructures, High Temp. High Press. 33 (2001) 293–
–301
[30]
C. Gruescu, A. Giraud, F. Homand, D. Kondo, DP. Do,
Effective thermal conductivity of partially saturated
porous rocks, Int. J. Solid Struct. 44 (2007) 811–833
[31]
K.A. Brucker, J. Majdalani, Effective thermal conductivity
of common geometric shapes, Int. J. Heat Mass Transfer
48 (2005) 4779–4796
[32]
Y.S. Touloukian, R.W. Powell, C.Y. Ho, P.G. Klemens,
Thermal Conductivity Nonmetallic Solids, IFI/Plenum,
New York, 1970
224
M. BARAHOEI et al.: COPPER OXIDE NANO-FLUID STABILIZED…
of Alaska Fairbanks, 2009, SPE International Student
Paper Contest at the SPE Annual Technical Conference
and Exhibition being held in New Orleans, LA, USA, 4-7
October, 2009
[48]
A. Zeinolabedini Hezave, S. Dorostkar, Sh. Ayatollahi, M.
Nabipour, B. Hemmateenejad, Investigating the effect of
ionic liquid (1-dodecyl-3-methylimidazolium chloride
([C12mim] [Cl])) on the water/oil interfacial tension as a
novel surfactant, Colloids Surfaces, A: Physicochem.
Eng. Aspects 421 (2013) 63– 71
[49]
A. Zeinolabedini Hezave, S. Dorostkar, Sh. Ayatollahi, M.
Nabipour, B. Hemmateenejad, Effect of Different Families
(Imidazolium and Pyridinium) of Ionic Liquids-Based
Surfactants on Interfacial Tension of Water/Crude Oil
System, Fluid Phase Equilibria (2013) 1-30
[50]
A. Zeinolabedini Hezave, S. Dorostkar, Sh. Ayatollahi, M.
Nabipour, B. Hemmateenejad, Dynamic interfacial tension behavior between heavy crude oil and ionic liquid
solution
(1-dodecyl-3-methylimidazolium
chloride
M. BARAHOEI1
A. ZEINOLABEDINI HEZAVE2
S. SABBAGHI3
SH. AYATOLLAHI4
1
Nano Chemical Engineering
Department, Shiraz University, Shiraz,
Iran, mlhbarahoei@gmail.com
2
Islamic Azad University, Dashtestan
Branch, Borazjan, Iran,
zeinolabedinihezave.ali@gmail.com
3
Nano Chemical Eng. Dep., Shiraz
University, Shiraz, Iran,
Sabbaghi@shirazu.ac.ir
4
School of Petroleum and Chemical
Engineering, Sharif University of
Technology, Tehran, Iran,
shahab@shirazu.ac.ir
NAUČNI RAD
Chem. Ind. Chem. Eng. Q. 22 (2) 211−225 (2016)
([C12mim] [Cl] + distilled or saline water/heavy crude oil))
as a new surfactant, J. Mol. Liquids 187 (2013) 83–89
[51]
M. Barahoei, S. Sabbaghi, A. Zeinolabedini Hezave, Sh.
Ayatollahi, Investigation on Stability of CuO Nano-fluid
Prepared at Presence of Ionic liquid and its Effectiveness
th
its thermal conductivity, in the Proceedings of the 8
International Chemical Engineering Congress & Exhibition (IChEC 2014), Kish, Iran, 24-27 February, 2014
[52]
R. David, Lide (Eds.), CRC Handbook of Chemistry and
th
Physics, 84 ed., CRC Press, Boca Raton, FL, 2003;
Section 12, Properties of Solids; Thermal and Physical
Properties of Pure Metals/Thermal Conductivity of Crystalline Dielectrics/Thermal Conductivity of Metals and
Semiconductors as a Function of Temperature
[53]
http://www.engineeringtoolbox.com/thermal-conductivity-liquids-d_1260.html
[54]
A. Safavi, S. Zeinali, Synthesis of highly stable gold
nanoparticles using conventional and germinal ionic
liquids, Colloids Surfaces, A: Physicochem. Eng. Aspects
362 (2010) 121–126.
NANO FLUID SA BAKAR OKSIDOM STABILIZOVAN
JONSKOM TEČNOŠĆU ZA POVEĆANJE
TOPLOTNE PROVODLJIVOSTI STENA LEŽIŠTA:
PRIMENA U TERMALNIM PROCESIMA SA
POVEĆANIM ISKORIŠĆENJEM NAFTNIH LEŽIŠTA
Pošto su rezerve nafte ograničene, a potražnja za energijom u porastu, nalaženje novih
efikasnih ili poboljšanje efikasnosti postojećih metoda za povećanja iskorišćenja naftnih
ležišta (EOR) je predmet brojnih istraživanja. Kao izlaz iz ovog problema moguća je kombinovana upotreba EOR metoda i nanotehnologija metoda da bi se iskoristile efikasnost
obe ove metode. U potrazi za takvom mogućnošću u prvom delu ovog rada pripremljeni su
stabilni vodeni rastvori sa nano česticama bakar oksida različitih koncentracija (0,01-0,05
M), koji su zatim ubrizganu u uzorke stena. Analiziran je efekat koncentracije različitih
površinski aktivnih komponenti. Zatim su proučavani različiti scenariji korišćenja nano
fluida kao modifikatora toplotne provodljivosti. Dobijeni rezultati jasno pokazuju da se
promenom koncentracije nano čestica bakar oksida u opsegu od 0,01 do 0,05 M može da
poboljša toplotna provodljivost stena od 27 do 48% u poređenju sa suvim stenama.
Ključne reči: toplotna provodljivost, nano-fluid, bakar-oksid, povećanje iskorišćenja naftnih ležišta, jonske tečnosti.
225
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chemical Industry & Chemical Engineering Quarterly
Chem. Ind. Chem. Eng. Q. 22 (2) 227−234 (2016)
NAÏMA MOUDIR1,2
NADJI MOULAÏ-MOSTEFA1,3
YACINE BOUKENNOUS2
1
LAFPC, University of Blida, Route
de Soumaa, Blida, Algeria
2
Research Center for
Semiconductor Technology for
Energy, Merveilles, Algeria
3
LME, University of Medea, Ain
D'Heb, Medea, Algeria
SCIENTIFIC PAPER
UDC 546.57:54:66.094.2
DOI 10.2298/CICEQ150106036M
CI&CEQ
SILVER MICRO- AND NANO-PARTICLES
OBTAINED USING DIFFERENT GLYCOLS AS
REDUCING AGENTS AND MEASUREMENT
OF THEIR CONDUCTIVITY
Article Highlights
• Silver micro- and nanoparticles were obtained by chemical reduction using glycols in the
presence and absence of PVP
• Silver particles were used for the preparation of conductive pastes for the metallization of solar cells
• Powder properties were determined by XRD, SEM and DSC/TGA
• The obtained samples showed good crystallinity, purity and spherical morphology
• The conductive behavior of pastes is dependent on the size, shape and type related
to the synthesis process
Abstract
Synthesis of silver micro- and nano-particles for the preparation of conductive
pastes for the metallization of solar cells was realized by chemical reduction in
the presence and absence of poly(vinyl-pyrrolidone) (PVP). Silver nitrate was
used as a precursor in the presence of three polyols (ethylene glycol, di-ethylene glycol and propylene glycol) tested at experimental temperatures near
their boiling points. Six samples were obtained by this protocol. Three silver
powders obtained without the use of PVP has a metallic luster appearance;
however, the samples produced using an excess of PVP were in the form of
stable colloidal dispersions of silver nano-particles. Structural characterizations
of samples using a scanning electron microscope and X-ray diffractometer
showed good crystallinity and spherical morphology. From DSC and TGA analyses, it was observed that all the nano-silvers present in the colloidal suspension have the same thermal behavior.
Keywords: silver nitrate, nano-particles, chemical reduction, glycol derivatives, PVP.
The operating characteristics of photovoltaic
solar cells depend upon their realization process and,
more particularly, on the nature and quality of the
front and rear electrical contact [1,2]. These contacts
in the form of a grid are generally realized using
screen printing technology. They consist of a homogeneous melted mixture of metallic oxides with additional chemicals [3]. Silver has the highest electrical
and thermal conductivity among metals, making it a
popular material for electrical contacts. Generally, silver powder is used as screen printed for making elecCorrespondence: N. Moudir, LAFPC, University of Blida, Route
de Soumaa, 09000 Blida, Algeria.
E-mail: moudir2001@yahoo.fr
Paper received: 6 January, 2015
Paper revised: 13 July, 2015
Paper accepted: 1 September, 2015
trical contacts on silicon solar cells due to its excellent
properties [4]. The thick film conductive paste contains spherical and/or flake-form silver powder of very
high conductivity, chemical stability, and low price.
In order to improve the layer adapted to the
desired properties, good knowledge of the physical
and morphological properties of silver powder is necessary. Given the range of applications in which silver
is important, better control of its properties would
have profound economic implications. It is well established that, by shrinking the size of a solid particle to
the nanometer regime, the chemical, electrical, mechanical, and optical properties can be altered [2]. Silver, which represents the most important component
in the paste, occurs in particles with various shapes
and sizes. It must have suitable properties (surface
227
N. MOUDIR et al.: SILVER MICRO- AND NANO-PARTICLES…
area, particle size and distribution), which depend on
the process parameters. Since silver is the main
component influencing the metallic contact, the
smaller are its particles sizes the greater is its specific
area in the powders with the best characteristics; the
specific area being the main factor favoring the film
continuity [4].
To manufacture these powders, several processes have been applied such as atomization and
milling of solid metals, precipitation from salt metal
solutions and electrodeposition [5,6], thermal method
[7], microwave irradiation [8], green synthesis [9],
chemical reduction [10], and gamma irradiation [11].
Each technique produces a powder with a characteristic morphology that influences its functional properties [12,13].
Nevertheless, in order to achieve better characteristics of silver particles, which affect directly the
mechanical and physical properties of the deposited
film, and consequently the solar cell performances,
the chemical process seems to be the best one giving
fine silver powder controlled in shape and size [14]. It
was chosen because it did not require special equipment. However, this method involves rigorous processing conditions.
Polyol synthesis was originally introduced by
Fievet et al. [13] as an excellent method for the production of metallic submicrometer-sized and nano-particles. This method was successfully developed
for the preparation of single-crystal silver nano-particles with uniform size and shape using polyvinylpyrrolidone (PVP) [15].
The polyol method involves heating a polyol with
a salt precursor and a polymeric capping agent to
generate metal colloids. Using this method, silver
nano-particles with different shapes have been manufactured [16,17]. All these improvements have promoted the scientific knowledge on nanomaterials [18].
Recently, Rao et al. [19] have included strategies
which have been employed for the synthesis of nano-materials of different dimensions.
In a typical polyol synthesis, silver atoms are
formed by reducing silver nitrate (AgNO3) used as
precursor with ethylene glycol [17]. PVP, a protective
agent, plays a crucial part in controlling superfine silver particle size and size distribution by reducing
silver nitrate with polyol. The particle size and particle
aggregation decrease with an increasing of PVP/
/AgNO3 weight ratio [20,21]. Several works focused
on the synthesis of micro- and/or nano-sized silver
particles using ethylene glycol [17,22-30].
In this document, we opted for the synthesis of
particles using the reduction of silver salt in three gly-
228
Chem. Ind. Chem. Eng. Q. 22 (2) 227−234 (2016)
colic media. It must be noted that ethylene glycol was
thoroughly employed, however propylene glycol and
di-ethylene glycol were rarely used [31,32].
The objective of this work was on one hand to
illustrate and assess the difference in the reduction
power of three glycols, used in similar concentrations
on silver cations, and on other hand to show their
influence on the crystallinity and morphological properties of the micro- and nano-particles obtained in
the absence and presence of PVP, respectively.
With thusly prepared powders, we have
attempted to develop our own purpose paste because
most commercial thick-film pastes are not especially
developed for solar cell metallization and had to be
modified for specific emitters.
EXPERIMENTAL
Materials
Silver samples were prepared through a chemical reduction of a polyol reduced silver nitrate solution using glycol derivatives as solvents. The following commercial reagents were employed: ethylene
glycol (EG) from Prolabo, di-ethylene glycol (DEG),
PVP and acetone, from Fluka, propylene glycol (PG)
from BDH Limited Poole and silver nitrate (AgNO3)
from Sigma-Aldrich, ethyl cellulose from BDH Prolabo, terpineol from Fluka, and polyvinyl- butyral from
Sigma Aldrich. All the chemicals were used as
received without further purification. De-ionized water
was used for washing the glassware.
Preparation of silver micro- and nano-particles
AgNO3 salt was dissolved under stirring in a
polyol reducing agent. Polyols have been tested near
their boiling points (197 °C for EG, 244 °C for DEG
and, 188 °C for PG). The reaction temperature was
fixed approximately between 10 and 20 °C below the
polyol boiling point. The moleratio of the precursor to
the polyol was set to 0.1, for each polyol. Each
sample batch was divided into two separate samples.
One of them was reacted with PVP as a surfactant
agent, and the other without PVP. The PVP was
weighted in a manner that the weight ratio of AgNO3/
/PVP was 2. For the whole of samples, the precipitation reaction occurs under continuous moderate stirring conditions. At the end of the reactions, the precipitates were separated by centrifugation. They were
washed in acetone several times until a clean solvent
was obtained. Further drying treatment at 80 °C overnight was performed on the micro-powders obtained
without PVP. Colloidal suspensions (nano-particles)
with PVP are noted SEG, SDEG and SPG, and the micro-
N. MOUDIR et al.: SILVER MICRO- AND NANO-PARTICLES…
-powders obtained without PVP are noted PEG, PDEG
and PPG. The subscripts EG, DEG and PG
correspond to the glycol solvents.
Preparation of conductive pastes
Samples of conductive pastes were formulated
using the same kind and amount of glass frit and
organic vehicle for comparative purposes with the
synthesized silver nano- and micro-powders. They
were prepared by dispersing the obtained silver
samples and an amount of lead-borosilicate glass frit
into an organic-polymeric vehicle (a mixture of ethyl
cellulose, terpineol and polyvinyl-butyral) with a mass
ratio of 75/5/20 of silver/glass frit/organic vehicle.
Conductive thick films were prepared by following the normal sequence of operations such as paste
preparation, screen-printing and firing the pastes. For
the back side a commercially aluminium paste from
Pemco was used. For the front side, the formulated
homemade pastes were screen printed through a
stainless steel mesh.
Solar cells fabrication
N+P solar cells were fabricated on as-cut
p-doped multicrystalline silicon wafers, 10 cm×10 cm,
and about 320 μm in thickness. The achieved process
is classic and involves the following steps: i) chemical
preparation, texturization, ii) emitter diffusion, iii) antireflection coating, iv) edge isolation, v) back metallization with commercially Ag/Al paste and vi) front
metallization using the pastes prepared by the synthesized powders.
Conductive thick films were prepared by following the normal sequence of operations such as paste
preparation, screen-printing and firing. For the back
side, a commercially aluminium paste from Pemco
was used. For the front side, the formulated homemade pastes were screenprinted through a stainless
steel mesh (mesh count 350). The resultant films for
the two sides were dried at ∼150 °C for 5 min to
volatilize the organic binder in the pastes, which
otherwise causes gas bubbles at higher temperatures
and results in cracking of the metallization [33]. Co-firing of the samples was done in a heated resistance
belt furnace with air atmosphere having peak firing
temperature of 720 °C for 3 min to sinter the inorganic
binder.
The obtained solar cells, hereafter corresponding to PEG, PDEG, PPG and SEG, SDEG, SPG were
produced following scrupulously the same process,
so that a direct comparison of the results could be
made. The pastes were fired on a conveyer belt
furnace at a peak set point temperature around 720
Chem. Ind. Chem. Eng. Q. 22 (2) 227−234 (2016)
°C with roughly 3 s in the hot zone to achieve a thickness of ∼12 μm.
Analysis methods
The crystalline structure of samples was performed using an X’pert Plus-PANalytical X-ray powder diffractometer using CuKα radiation with a tube
current of 50 mA and a voltage of 40 kV. The X’Pert
High Score Plus software was used for the phase
identification. The study of the morphology was carried out using a scanning electron microscope (Jeol,
JSM-6360 LV SEM/EDAX).
Thermal analysis of samples was performed
using coupled techniques such as differential scanning calorimeter and thermo-gravimetric analyzer
(DSC/TGA). The weight loss of samples was determined by TGA curve. This coupled analysis was
performed with a NetzschSTA 409 PC instrument, in
the temperature range of 25-1450 °C. This temperature range includes the boiling point of solvents, and
degradation temperature of polymers.
Because silver nano-particles absorb light significantly and their UV/Vis spectra show an intense
absorption peak in the visible region, their optical
properties were investigated by a Cary 500 DE Varian
model, in the spectral range of 175 to 3300 nm. The
samples were dispersed in absolute ethanol for the
analysis, and the powders were leaved in their final
granular form. For the whole, the absorbance measurement was carried out at room temperature, in the
spectral range of 300 to 700 nm.
The contact resistance of the deposited layers
was measured using standard transmission line measurement (TLM) method. Four contacts were used (2
for current and 2 for voltage).
RESULTS AND DISCUSSION
Chemical reaction description
During the reduction of silver, the transparent
mixture changed to yellow, while with continuous stirring the color of the reaction changed to gray and
dark gray. These observations show that with an
increase in reaction time, concentration of silver cluster gradually increased where silver cations are adsorbed. By analogy and for reason of similarity with
other glycols, a general mechanism of metal reduction in ethylene glycol can be represented by the following reactions:
CH2OH–CH2OH → CH3CHO + H2O
(1)
2CH3CHO + 2Ag (I) → 2Ag + 2H+ + CH3COCOCH3 (2)
229
N. MOUDIR et al.: SILVER MICRO- AND NANO-PARTICLES…
Chem. Ind. Chem. Eng. Q. 22 (2) 227−234 (2016)
The three silver powders obtained without the
use of surfactant have a metallic lustered appearance. However, the samples produced using an
excess of PVP are in the form of stable colloidal dispersions of silver nano-particles. All the micro and
nano-silver particles synthesized in the same conditions were obtained in a similar time indicating that
EG, DEG and PG have identical reduction power.
However, EG, DEG and PG have distinct reducibility.
In this work the reaction temperature is too high;
hence the reaction rates in EG, DEG and PG are
accelerated to the same level.
of PVP. The entire samples exhibit the same patterns.
Four strong peaks were noticeable in each pattern,
which are characteristics to fcc silver of JCPDS card
No. 04-0783 [34]. The results suggest that the powders show good crystallinity. No crystallographic
impurities phases were found. The high intensity of
the peaks indicated that the silver particles were well
crystallized. Nevertheless, the peaks pattern of
samples SEG, SDEG and SPG had not changed in crystallinity. These results confirm the negligible amorphous phase of surfactant in PVP samples.
XRD characterization
The morphologies of silver samples characterized by SEM images are shown in Figure 2 (without PVP for PEG powder) and 3 (with PVP for SEG
SEM micrographs
Figure 1 shows the XRD patterns of the synthesized silver particles in the presence and absence
3
8x10
P PG without PVP
Intensity (counts)
3
6x10
(111)
(200)
3
4x10
(220)
(311)
P DEG without PVP
3
2x10
P EG without PVP
10
20
30
40
50
60
70
80
2θ (degree)
3
8x10
SPG with PVP
Intensity (counts)
3
6x10
(111)
3
4x10
(200)
(220)
S DEG with PVP
(311)
3
2x10
SEG with PVP
0
10
20
30
40
50
60
70
80
2θ (degree)
Figure 1. XRD diffraction patterns of the synthesized Ag powders without and with PVP.
230
N. MOUDIR et al.: SILVER MICRO- AND NANO-PARTICLES…
Chem. Ind. Chem. Eng. Q. 22 (2) 227−234 (2016)
PPG
PDEG
PEG
Figure 2. SEM micrographs of the synthesized Ag micro-powder without PVP using respectively EG, DEG and PG.
SPG
SDEG
SEG
Figure 3. SEM micrographs of the synthesized Ag colloidal suspensions with PVP using respectively EG, DEG and PG.
colloidal suspension). The micrographs revealed that
the PEG powder is formed by agglomerated heterogeneous spheroidal particles of size ranging from 1-2
µm. However, the SEG sample prepared with PVP
featured homogeneous nano-spherical particles. In
addition, the presence of nanowires formed by
assembling clusters of some nano-particles was observed. The size was approximately 50-80 nm. The
micro-particles of PDEG were agglomerated in a narrow size distribution and their mean size was
estimated to be 1-2 µm (Figure 2). The structure of
SDEG (Figure 3) was slightly different, as there were no
assembled particles forming rods.
The micrograph of PPG presented in Figure 2
illustrates the same morphology for PEG and PDEG. For
SPG (Figure 3), spherical silver nano-particles are
observed. These results prove that both the size and
shape particles depend on the presence of surfactant.
These results are in agreement with those obtained
by Li et al. [35], Yu et al. [36] and Gasaymeh et al.
CH2
HC
CH2
n
HC
CH2
n
HC
[37]. The PVP restricts the mobility of silver ions
during the chemical reduction and, prevents aggregation among the particles and limits their sizes. It
should be noted that there is no difference in morphological properties of the silver products synthesized in
the same conditions.
These effects can be described according to the
above described reaction mechanism (Eqs. (3) and
(4)). This mechanism is explained by the formation of
coordination bonds between PVP and silver ions
involving the trapping of Ag+ by macromolecules of
PVP. Firstly, the complex of PVP silver ions is constructed; secondly, the complex promotes silver nucleation; thirdly, aggregation and limited dispersion
leads to the formation of silver nano-particles [38]. As
the reaction rate of silver particles increases with the
concentration of PVP, it appears that this could be the
reason for observing more regular and smaller particles and narrower particle size distributions at higher
concentration of PVP [39,40].
CH2
n
HC
n
+
:Ag :
+
N
N
O + 2 Ag
3
O
+ N
+
N
O :Ag :O
(3)
CH2
HC
n
:Ag+:
N
O
CH2
HC
+ N
CH2
n
HC
+
O :Ag :O
N
CH2
n
HC
+2e
HC
O
+ N
:Ag:
N
CH2
n
n
O :Ag: O
CH2
HC
n
N
(4)
231
N. MOUDIR et al.: SILVER MICRO- AND NANO-PARTICLES…
PEG without PVP
Mass change = 0.24%
100,2
100.2
-1,0
-1.0
100,0
100.0
TG (%)
DSC (mW/mg)
100,4
100.4
280 °C
-0,5
0.5
-1,5
-1.5
99,8
99.8
-2,0
-2.0
99.6
99,6
-2.5
-2,5
00
100
100
200
200
300
300
400
400
500
500
600
600
Temperature (Celsius)
Figure 4. DSC/TGA curves of Ag powders synthesized without
PVP using EG.
The organic phase could be however completely
removed by keeping the particles at a relatively low
temperature (400 °C) for several minutes. In comparison with the results shown in Figure 4, thermal
analysis of SEG shows that silver suspensions of
nano-particles had a higher degradation temperature
than that of PVP used alone. All the nano-silvers
present in the colloidal suspensions have almost the
232
SEG with PVP
0.0
0,0
100
100
9090
0.5
-0,5
8080
7070
-1.0
-1,0
6060
Mass change = 73.06%
-1.5
-1,5
5050
56.4 °C
4040
-2.0
-2,0
446.5 °C
-2.5
-2,5
00
100
100
200
200
300
300
400
400
3030
500
500
2020
600
600
Temperature (Celsius)
Figure 5. DSC/TGA curves of colloidal nano-suspensions
synthesized with PVP using EG.
UV-Vis absorption spectra
100,6
100.6
0,0
0.0
110
110
TG (%)
Figures 4 and 5 show the DSC and TGA curves
of PEG and SEG samples from which onset and final
decomposition temperatures were obtained. It should
be noted that only DSC/TGA data of PEG and SEG
were studied because the similarity of the results.
From the DSC analysis of silver micro-powders synthesized in EG without PVP (Figure 4), the presence
of an exothermic phase transition was observed
around 280 °C, while the TGA diagram shows a
weight loss of about 0.24%, which is a negligible
mass. It represents the amount of organic part that
persists as trace of the experimental process. Other
impurities located in the micrometer silver powders
result from the decomposition of the rest fraction of
the reducing agent. As silver does not melt until 961
°C and is not known to sublimate, it is reasonable to
assume that the observations made during TGA can
be attributed to the organic materials only. At the prominent temperatures identified by TGA and DSC, the
most important endothermic peak in DSC curve of
nano-silver synthesized with PVP, using EG (Figure
5) corresponds to the solvent volatilization of the colloidal suspension. This one coincides with the peaks
of weight loss of 18.47 and 20.62%.
same thermal behavior. Particles coated with stable
substances such as PVP are suitable, because the
protective layer can be removed at temperatures
below 500 °C. This amount of PVP will not affect the
conductivity of the metal powder [37,41].
DSC (mW/mg)
Thermal analysis
CI&CEQ 22 (2) 227−234 (2016)
Figure 6 shows the UV-Vis absorption spectra
for the samples prepared with 0.1 mole ratio of
AgNO3 to polyol, in excess of PVP with glycols. The
absorption band in visible light region (400-450 nm) is
typical for silver nano-particles. The absorbance
bands at 447, 441 and 436 nm are attributed to the
surface plasmon resonance phenomenon of free
electrons in the conduction bands of silver nanoparticle suspensions, respectively synthesized by EG,
DEG and PG. These results are in agreement with
that of silver nano-particles prepared with PVP chemical reaction by many authors, such as Dung Dang et
al. [30] and Slistan-Grijalva et al. [42]. The morphology and the particle size may play an important
role in the absorption spectra of Ag/PVP nano-particles.
Resistivity measurement
The sheet resistance of the fired films for nanoparticles was 3.6, 4.1 and 4.6×10-6 Ω cm-2 corresponding to SEG, SDEG and SPG, respectively, and 9.6,
13.6 and 11.1×10-6 Ω cm-2 to PEG, PDEG and PPG,
respectively. From these results, we deduce that the
conductive behavior of the obtained conductive
pastes is dependent on the size, shape and type related to the synthesis process (with or without PVP)
N. MOUDIR et al.: SILVER MICRO- AND NANO-PARTICLES…
and slightly on the firing range process (around 720
°C) of the screen-printed deposited metallic contacts.
2.5
2,5
SEG
SDEG
Absorbance (a.u)
2.0
2,0
SPG
SEG
1.5
1,5
SPG
SDEG
1.0
1,0
Chem. Ind. Chem. Eng. Q. 22 (2) 227−234 (2016)
mated from 1 to 2 µm. The structure of SDEG was
slightly different, while the morphologies of PEG and
PDEG were comparable. For SPG, spherical silver nanoparticles were observed. These results prove that
both the size and shape particles depend on the presence of surfactant. The DSC/TGA analysis showed
that the protective layer could be removed at temperatures below 500 °C. This amount of PVP will not
affect the conductivity of the metal powder. The surface plasmon bands appearing in the UV-visible
region are characteristic of silver nano-particles.
The obtained results also show that the highest
heating temperature gives similar powders. This
means that the heating at high temperature allows the
silver nitrate to react completely giving fine powders
with same properties.
0.5
0,5
REFERENCES
300
300
400
400
500
500
600
600
700
700
[1]
S.B. Rane, P.K. Khanna, T. Seth, G.J. Phatak, D.P.
Amalnerkar, B.K. Das, Mater. Chem. Phys. 82 (2003)
237-245
[2]
S.B. Rane, T. Seth, G.J. Phatak, D.P. Amalnerkar, B.K.
Das, Mater. Lett. 57 (2003) 3096-3100
[3]
J.C. Lin, C.Y. Wang, Mater. Chem. Phys. 45 (1996) 136–144
[4]
Z. Zhang, B. Zhao, L. Hu, J. Solid State Chem. 21 (1996)
105-110
[5]
J.C. Lin, C.Y. Wang, Mater. Chem. Phys. 45 (1996) 136–144
[6]
A. Sinha, B.P. Sharma, Bull. Mater. Sci. 28 (2005) 213–217
[7]
X. Sun, Y. Luo, Mater. Lett. 59 (2005) 3847-3850.
[8]
A. Pal, S. Shah, S. Devi, Mater. Chem. Phys. 114 (2009)
530-532
[9]
A. Jebali, F. Ramezani, B. Kazemi, J. Cluster Sci. 22
(2011) 225-232
[10]
H. Wang, X. Qiao, J. Chen, S. Ding, Colloids Surfaces A
56 (2005) 111-115
[11]
Y.N. Rao, D. Banerjee, A. Datta, S.K. Das, R. Guin, A.
Saha, Radiat. Phys. Chem. 76 (2010) 1240-1246
[12]
Y. Dai, T. Deng, S. Jia, L. Jin, F. Lu, J. Membr. Sci. 281
(2006) 685-691
[13]
F. Fievet, J.P. Lagier, B. Blin, B. Beaudoin, M. Figlarz,
Solid State Ionics 32/33 (1989) 198-205
[14]
B. Wiley, Y. Sun, Y. Xia, Acc. Chem. Res. 40 (2007)
1067-1076
[15]
Y. Sun, Y. Xia, Science 298 (2002) 2176-2179
[16]
B. Blin, F. Fievet, D. Beaupere, M. Figlarz, Nouv. J. Chim.
13 (1989) 67-72
[17]
D. Malina, A. Sobczak-Kupiec, Z. Wzorek, Z. Kowalski,
Dig. J. Nanomater. Bios. 7 (2012) 1527-1534
[18]
G. Carotenuto, G.P. Pepe, L. Nicolais, Eur. Phys. J., B 16
(2000) 11-17
Wavelength (nm)
Figure 6. UV/Vis spectra of Ag colloidal nano-suspensions
synthesized with PVP.
CONCLUSION
Series of experiments in glycolic medium (EG,
DEG and PG) were carried out at the boiling point of
each one for the reduction of silver nitrate to silver
micro-particles, and to nano-particles with the use of
PVP. The three silver powders obtained without the
use of surfactant had a metallic lustered appearance.
However, the samples produced using an excess of
PVP were in the form of stable colloidal dispersions of
silver nano-particles.
The obtained results from DRX, SEM and thermal analysis showed that the presence of PVP affects
the physicochemical properties of the synthesized
silver particles; however, the use of the different glycolic mediums at their boiling temperature has no influence.
The XRD results showed that the powders
exhibited good crystallinity, with negligible amorphous
phase of surfactant in PVP samples. The SEM of PEG
powder reveals that the powder is made by agglomerated heterogeneous spheroidal particles having a
size ranging from 1 to 2 µm. However, the SEG sample
prepared with PVP exhibits homogeneous nanospherical particles, with some nanowires formed by
assembling clusters of some nano-particles. The size
was approximately on the order of 50-80 nm. The
micro-particles of PDEG were agglomerated in a narrow size distribution and their mean size was esti-
233
N. MOUDIR et al.: SILVER MICRO- AND NANO-PARTICLES…
Chem. Ind. Chem. Eng. Q. 22 (2) 227−234 (2016)
[19]
C.N.R. Rao, H.S.S. Ramakrishna Matte, Rakesh Voggu,
A. Govindaraj, Dalton Trans. 41 (2012) 5089-5120
[31]
J.H. Sohn, L.Q. Pham, H.S. Kang, J.H. Park, B.C. Lee,
Y.S. Kang, Radiat. Phys. Chem. 79 (2010) 1149-1153
[20]
K. S. Chou, Y.S. Lai, Mater. Chem. Phys. 83 (2004) 82–88
[32]
T. Maiyalagan, Appl. Catal., A 340 (2008) 191-195
[33]
M.M. Hilali, Understanding and development of manufacturable screen-printed contacts on high sheet-resistance
emitters for low-cost silicon solar cells, PhD Thesis,
Georgia Institute of Technology, 2005
[34]
JCPDS data Philips X’Pert High Score Package, Diffraction data CD-ROM, International Center for Diffraction
Data, Newtown Square, PA, 2004
[35]
Z. Li, Y. Li, X-F. Qian, J. Yin, Z-K. Zhu, Appl. Surf. Sci.
250 (2005) 109-116
[21]
W. Songping, M. Shuyuan, Mater. Chem. Phys. 89 (2005)
423-427
[22]
Y. Sun, Y. Yin, B.T. Mayers, T. Herricks, Y. Xia, Chem.
Mater. 14 (2002) 4736-4745
[23]
Y. Gao, L. Song, P. Jiang, L.F. Liu, X.Q. Yan, Z.P. Zhou,
D.F. Liu, J.X. Wang, H.J. Yuan, Z.X. Zhang, X.W. Zhao,
X.Y. Dou, W.Y. Zhou, G. Wang, S.S. Xie, H.Y. Chen, J.Q.
Li, J. Crys. Growth 276 (2005) 606-612
[24]
C. Luo, Y. Zhang, X. Zeng, Y. Zeng, Y. Wang, J. Colloid
Interface Sci. 288 (2005) 444-448
[36]
D-G. Yu, W-C. Lin, C-H. Lin, L-M. Chang, M-C. Yang,
Mater. Chem. Phys. 101 (2007) 93-98
[25]
J.L. Elechiguerra, L. Larios-Lopez, C. Liu, D. Garcia-Gutierrez, A. Camacho- Bragado, M.J. Yacaman, Chem.
Mater. 17 (2005) 6042-6052
[37]
S-S. Gasaymeh, S. Radiman, L-Y.Heng, E. Saion, G.H.
Mohamed Saeed, Am. J. Appl. Sci. 7 (2010) 892-901
[38]
J.S. Kim, J. Ind. Eng. Chem. 13 (2007) 566-570
[26]
G. Dzido, A.B. Jarzebski, J. Nanopart. Res. 13 (2011)
2533-5541
[39]
Z. Zhang, B. Zhao, L. Hu, J. Solid State Chem. 121(1996)
105-110
[27]
J.H. Byeon, Y.W. Kim, Ultrason. Sonochem. 19 (2012)
209-215
[40]
J. Zhu, S. Liu, O. Palchik, Y. Koltypin, A. Gedanken,
Langmuir 16 (2000) 6396-6399
[28]
Y. Zhang, R. Zhang, N. Liu, Z. Yuan, J. Yang, Y. Zhu, Y.
Cao, A. Zhang, P. Yang, J. Inorg. Organomet. Polym.
Mater. 22 (2012) 514-518
[41]
S-A. Ketkar, G-G. Umarji, G.J. Phatak, J.D. Ambekar,
U.P. Mulik, D.P. Amalnerkar, Mater. Chem. Phys. 96
(2006) 145-153
[29]
T. Darmanin, P. Nativo, D. Gilliland, G. Ceccone, C.
Pascual, B.D. Berardis, F. Guittard, F. Rossi, Colloid
Surfaces, A 395 (2012) 145-151
[42]
[30]
T.M. Dung Dang, T.T.T. Le, E. Fribourg-Blanc, M. Chien
Dang, Adv. Nat. Sci. Nanosci. Nanotechnol. 3 (2012)
035004 (doi:10.1088/2043-6262/3/3/035004)
A. Slistan-Grijalva, R. Herrera-Urbina, J.F. Rivas-Silva,
M. Avalos-Boria, F.F. Castillon-Barraza, A. Posada-Amarillas, Physica E. Low Dimens. Syst. Nanostruct. 25
(2004) 438-448.
NAÏMA MOUDIR1,2
NADJI MOULAÏ-MOSTEFA1,3
YACINE BOUKENNOUS2
1
LAFPC, University of Blida, Route de
Soumaa, Blida, Algeria
2
Research Center for Semiconductor
Technology for Energy, Merveilles,
Algeria
3
LME, University of Medea, Ain D'Heb,
Medea, Algeria
NAUČNI RAD
MIKRO- I NANO-ČESTICE SREBRA DOBIJENE
POMOĆU RAZLIČITIH GLIKOLA KAO
REDUKUJUĆIH SUPSTANCI I MERENJE
NJIHOVE PROVODLJIVOSTI
Sinteza mikro i nano-čestica srebra za pripremu provodnih pasti za metalizaciju solarnih
ćelija, izvršena je hemijskom redukcijom u prisustvu i odsustvu polivinilpirolidona (PVP).
Kao prekursor u sintezi korišćen je srebro nitrat u prisustvu tri poliola (etilen glikol, dietilen
glikol i propilen glikol) testiranih na eksperimentalnim temperaturama u blizini njihovih tački
ključanja. Na ovaj nečin je dobijeno šest uzoraka. Bez upotrebe PVP, dobijena su tri
uzorka srebra u obliku praha sa metalnim sjajem; međutim, uzorci dobijeni u prisustvu
viška PVP se nalaze u formi stabilnih koloidnih disperzija nano-čestica srebra. Strukturna
karakterizacija uzoraka pomoću skenirajućeg elektronskog mikroskopa i X-analize pokazala je dobru kristalnost i sfernu morfologiju. DSC i TGA analiza pokazuju da sve nano-čestice srebra prisutne u koloidnoj suspenziji imaju isto termalno ponašanje.
Ključne reči: srebro-nitrat, nano-čestice, hemijska redukcija, derivati glikola, PVP.
234