Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
ISSN 1451 - 9372(Print)ISSN 2217 - 7434(Online)JANUARY-MARCH 2017Vol.23, Number 1, 1-150
www.ache.org.rs/ciceq
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: [email protected]
ASSOCIATE EDITORS Jonjaua Ranogajec
Faculty of Technology, University of Novi Sad, Novi Sad, Serbia
Srđan PejanovićDepartment of Chemical Engineering, Faculty of Technology and Metallurgy,
University of Belgrade, Belgrade, Serbia
Milan Jakšić 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 Texas A&M University,
College Station, TX, USA Milorad Dudukovic
Washington University, St. Luis, MO, USA
Jiri Hanika Institute of Chemical Process Fundamentals, Academy of Sciences
of the Czech Republic, Prague, Czech Republic Maria Jose Cocero
University of Valladolid, Valladolid, Spain Tajalli Keshavarz
University of Westminster, London, UK Zeljko Knez
University of Maribor, Maribor, Slovenia
Igor Lacik Polymer Intitute of the Slovak Academy of Sciences,
Bratislava, Slovakia Denis Poncelet
ENITIAA, Nantes, France
Ljubisa Radovic Pen State University,
PA, USA Peter Raspor
University of Ljubljana, Ljubljana, Slovenia
Constantinos Vayenas University of Patras,
Patras, Greece Xenophon Verykios University of Patras,
Patras, Greece Ronnie Willaert
Vrije Universiteit, Brussel, Belgium
Gordana Vunjak Novakovic Columbia University,
New York, USA Dimitrios P. Tassios
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
Journal of the Association of Chemical Engineers of Serbia, Belgrade, Serbia
Vol. 23 Belgrade, January-March 2017 No. 1
Chemical Industry & Chemical EngineeringQuarterly (ISSN 1451-9372) is published
quarterly by the Association of ChemicalEngineers of Serbia, Kneza Miloša 9/I,
11000 Belgrade, Serbia
Editor:Vlada B. Veljković
Editorial Office:Kneza Miloša 9/I, 11000 Belgrade, Serbia
Phone/Fax: +381 (0)11 3240 018E-mail: [email protected]
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, SerbiaPhone/Fax: +381 (0)11 3240 018
Publication of this Journal is supported by theMinistry of Education, Science and
Technological Development of the Republic of Serbia
Subscription and advertisements make payableto the account of the Association of Chemical
Engineers of Serbia, Belgrade, No. 205-2172-71, 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 inThompson Reuters products: Science Citation
Index - ExpandedTM - access via Web of Science®, part of ISI Web of KnowledgeSM
CONTENTS
Marija Mihajlović, Marija Stanojević, Mirjana Stojanović, Je-lena Petrović, Jelena Milojković, Marija Petrović, Zo-rica Lopičić, To what extent do soft mechanical activation and process parameters increase the efficiency of different zeolite/phosphate rock fertilizer mixtures? ................................................................................ 1
Fatih Ilhan, Kaan Yetilmezsoy, Harun Akif Kabuk, Kubra Ulu-can, Tamer Coskun, Busra Akoglu, Evaluation of operational parameters and its relation on the stoichiometry of Fenton’s oxidation to textile was-tewater .................................................................................. 11
Javad Ahmadishoar, S. Hajir Bahrami, Barahman Movas-sagh, Seyed Hosein Amirshahi, Mokhtar Arami, Rem-oval of Disperse Blue 56 and Disperse Red 135 dyes fromaqueous dispersions by modified montmorillonite nanoclay ............................................................................... 21
Alireza Ebrahiminezhad, Yahya Barzegar, Younes Ghasemi, Aydin Berenjian, Green synthesis and characterization of silver nanoparticles using Alcea rosea flower extract as a new generation of antimicrobials .................................. 31
Aleksandra Mišan, Bojana Šarić, Ivan Milovanović, Pavle Jovanov, Ivana Sedej, Vanja Tadić, Anamarija Mandić, Marijana Sakač, Phenolic profile and antioxidant properties of dried buckwheat leaf and flower extracts ........ 37
Yajing Zhang, Yu Zhang, Fu Ding, Kangjun Wang, Xiaolei Wang, Baojin Ren, Jing Wu, Synthesis of DME by CO2 hydrogenation over La2O3-modified CuO–ZnO– –ZrO2/HZSM-5 catalysts ....................................................... 49
Tatjana Kaluđerović Radoičić, Nevenka Bošković-Vragolović, Radmila Garić-Grulović, Mihal Đuriš, Željko Grbavčić, Friction factor for water flow through packed beds of spherical and non-spherical particles ................................... 57
Hai-Peng Gou, Guo-Hua Zhang, Kuo-Chih Chou, Prepar-ation of titanium carbide powder from ilmenite con-centrate ................................................................................. 67
Milana M. Zarić, Mirko Stijepovic, Patrick Linke, Jasna Stajić-Trošić, Branko Bugarski, Mirjana Kijevčanin, Targeting heat recovery and reuse in industrial zone ............................ 73
Sandra Raquel Kunst, Lilian Vanessa Rossa Beltrami, Mari-elen Longhy, Henrique Ribeiro Piaggio Cardoso, Tiago Lemos Menezes, Célia de Fraga Malfatti, Effect of diisodecyl adipate concentration in hybrid films applied to tinplate .............................................................................. 83
CONTENTS Continued Milovan Janković, Snežana Sinadinović-Fišer, Olga Goveda-rica, Jelena Pavličević, Jaroslava Budinski-Simendić, Kinetics of soybean oil epoxidation with peracetic acid formed in situ in the presence of an ion exchange resin: Pseudo-homogeneous model .................................... 97
Salah H. Aljbour, Sultan A. Tarawneh, Adnan M. Al-Harah-sheh, Evaluation of the use of steelmaking slag as an aggregate in concrete mix: A factorial design approach .... 113
Dušan Lj. Petković, Miloš J. Madić, Goran M. Radenković, The effects of passivation parameters on pitting potential of biomedical stainless steel ................................ 121
Marija Kodric, Sandra Stojanovic, Branka Markovic, Dragan Djordjevic, Modelling of polyester fabric dyeing in the presence of ultrasonic waves ............................................. 131
Aishi Zhu, Shanshan Liu, Kanfeng Wu, Chuan Ren, Maoqian Xu, Comparing of hot water and acid extraction of polysaccharides from proso millet ...................................... 141
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
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 19 (2017) CI&CEQ
1
MARIJA MIHAJLOVIĆ
MARIJA STANOJEVIĆ
MIRJANA STOJANOVIĆ
JELENA PETROVIĆ
JELENA MILOJKOVIĆ
MARIJA PETROVIĆ
ZORICA LOPIČIĆ
Institute for Technology of Nuclear
and Other Mineral Raw Materials,
Belgrade, Serbia
SCIENTIFIC PAPER
UDC 631.82/.85:549.67:553:66
https://doi.org/10.2298/CICEQ150622047M
TO WHAT EXTENT DO SOFT MECHANICAL ACTIVATION AND PROCESS PARAMETERS INCREASE THE EFFICIENCY OF DIFFERENT ZEOLITE/PHOSPHATE ROCK FERTILIZER MIXTURES?
Article Highlights
• Effect of mechanical activation on different exchange-fertilizer mixtures was investigated
• Substantially increased P content than that of the similar non-activated mixtures was
observed
• Influence of different process parameters was explored using multivariate statistics
• The results confirmed a high fertilization potential of the selected mechanically acti-
vated substrates
Abstract
In order to obtain effective mineral fertilizer, different mixtures of phosphate rock
(PR) with natural clinoptilolite (Cp) and NH4+ saturated clinoptilolite (NH4-Cp)
were subjected to soft mechanical activation. Mean concentrations of P released
from mechanically activated (MA) substrates, MACp/PR and MANH4-Cp/PR,
ranged between 2.81-3.19 mg L-1 and 2.02-7.74 mg L-1, respectively. These are
10 to 15 times higher P concentrations than those released from the corres-
ponding non-activated mixtures. Solution Ca2+, K+, Na+ and Mg2+ concentrations
varied according to the composition of the mixtures and the contact time
between the two minerals within their optimal values required for plant growth.
The obtained results suggest that the efficiency of the NH4-Cp/PR mixtures can
be significantly increased by the proposed mechanical activation. Influence of
process parameters on the observed concentrations of nutrients was shown
using multivariate statistics. The highest fertilization potential demonstrated
MANH4-Cp/PR mixture with the largest NH4-Cp share and the longest proposed
mixing time.
Keywords: exchange-fertilizer mixtures, phosphate rock, clinoptilolite, mechanical activation, multivariate analysis.
Correct use of fertilizers and natural resources
contributes to environmental sustainability. Inorganic
fertilizers account for about 80% of all phosphate
applications, of which more than 99% originate from
the phosphate rock (PR) [1]. Traditional technologies
for production of high soluble mineral fertilizers from
natural PR include acid treatments causing environ-
mental contamination and eutrophication [2]. To avoid
Correspondence: M. Mihajlovic, Central Laboratory for Testing,
Institute for Technology of Nuclear and Other Mineral Raw
Materials, Franše Deperea 86, 11000 Belgrade, Serbia. E-mail: [email protected] Paper received: 22 June, 2015 Paper revised: 12 November, 2015 Paper accepted: 7 December, 2015
these issues, it is necessary to find new methods to
obtain soluble fertilizers with less negative impact on
the environment. The direct application of PR seems
to be one of the effective strategies with lower energy
usage and production costs. However, the low sol-
ubility, therefore the low attainability of nutrients for
the plant growth, is the main disadvantage of the PR
direct utilization [3,4]. One of the proposed approaches,
in aim to increase dissolution of PR, is the application
of combined zeolite/PR mixtures. The addition of nat-
ural zeolites to the PR, usually clinoptilolite (Cp), imp-
roves the PRs agrochemical effect [5,6]. The high
cation-exchange capacity, soil texture enrichment and
water-retention ability are main features of zeolites,
M. MIHAJLOVIĆ et al.: TO WHAT EXTENT DO SOFT MECHANICAL ACTIVATION… Chem. Ind. Chem. Eng. Q. 23 (1) 19 (2017)
2
which determine their successful agricultural appli-
cations [7-11]. Characteristics of fertilizer mixtures
could be further enhanced through modification of
natural zeolites with nutrient elements such as NH4+
[12-14]. Thus, the zeolites, beside the increase of the
PR solubility can become a source of N, Ca2+ and K+
required for plant growth [6,14].
Another approach for increasing PR solubility is
its mechanical activation that produces physical and
chemical changes in close surface regions where
mechanical energy leads to contact between the
solids [15,16]. The extensive knowledge of solid-state
mechanochemistry has been expanded with the obs-
ervations of the influence of high pressure and shear
on the rate of chemical solid-state reactions [16].
However, some authors note that the particle size
reduction due to mechanical activation affects the
PRs reactivity to a greater extent than the resulting
deformations of its crystal structure [17]. Subse-
quently, it has been found that the efficiency of
exchange-fertilizer mixtures may be further enhanced
by their mechanical activation [18-22]. To what extent
this procedure increases the efficiency of different
zeolite/PR mixtures, so that its inclusion into the pro-
duction process could be considered rational, was the
subject of our survey. Therefore, the influence of soft
mechanical activation on improvement of the PR dis-
solution in the presence of two types of clinoptilolites,
natural (Cp) and partially NH4+ saturated (NH4-Cp)
was studied. The changes in mineralogical structure,
and concentrations of plant available nutrients rel-
eased from both MACp/PR and MANH4-Cp/PR mix-
tures, were investigated and compared with that of
the corresponding non-activated mixtures. Differ-
ences in concentrations of the released nutrients and
their dependencies on the process parameters
between various MA-substrates were analyzed using
multivariate statistical methods.
EXPERIMENTAL
Characterization of materials
Natural phosphate ore, PR, from deposit Lisina,
near Bosilegrad in Serbia with average content of 9%
P2O5 and zeolite tuff (with >75% of clinoptilolite (Cp))
from deposit Baia Mare, Romania, were used for the
preparation of the mixtures. Selected characteristics
of the Cp and PR samples are presented elsewhere
[6]. To obtain the NH4-exchanged Cp, a partial sat-
uration of the Cp with NH4+ at 1:7.5 ratios was per-
formed, according to the procedure described by
Mihajlović et al. [23].
Powder X-ray diffraction (XRD) was used to
determine the phase composition of the Cp/PR and
NH4−Cp/PR mixtures before and after mechanical
activation. The XRD patterns were obtained on a
Philips PW-1710 automated diffractometer using a Cu
tube operated at 40 kV and 30 mA. The instrument
was equipped with a diffracted beam curved graphite
monochromator and a Xe-filled proportional counter.
The diffraction data were collected in the 2 Bragg
angle range from 5 to 60°, counting for 2 s (qualitative
identification) at every 0.02° step. The divergence and
receiving slits were fixed at 1 and 0.1 units, res-
pectively. The XRD measurements were performed at
room temperature in a stationary sample holder.
Experimental procedure
Both groups of mixtures (Cp/PR and NH4-Cp/
/PR), in three replicates, were assembled in three
ratios of Cp and the PR; 5:1 10:1 and 15:1. Each
mixture originally contained 4 g of PR and the corres-
ponding share of zeolite (20, 40 and 60 g, respect-
ively). The mechanical activation of the mixtures was
carried out in a vibrating ring-mill (KHD, Humboldt
Wedag, AG). To avoid sticking, the period of mech-
anical activation was of 30 s per sample at room tem-
perature. After mechanical activation, mixtures were
placed in 300 ml volumetric flasks. Then, in each vol-
umetric flask was added 200 ml of distilled water and
mixtures were shaken on a rotary shaker for 24, 48
and 72 h at 220 rpm. In the resulting solutions, after
draining, the concentrations of Ca2+, K+, Na+ and Mg2+
were determined using a Perkin Elmer AAS 703
atomic absorption spectrometer. The concentration of
P from the solution was determined by colorimetry
[24] using a Spekol 1300 UV–Vis spectrophotometer.
Exploratory data analysis
Descriptive statistical analyses of the results
were expressed as the mean ± standard deviation
(SD). Analysis of variance (ANOVA) and the following
posthoc Tukey’s HSD were performed to determine
whether there are any significant differences between
the average values of the released nutrients between
the groups.
Multivariate statistical analysis
Principal component analysis (PCA) as a multi-
variate analytical method was used to display the
data in a multidimensional space, where the variables
determine the axes [25]. These axes are projected
into a few principal components (PCs), which are
linear combinations of the original variables and
define the maximum variation within the data. The first
principal component (PC1) accounts for the largest
M. MIHAJLOVIĆ et al.: TO WHAT EXTENT DO SOFT MECHANICAL ACTIVATION… Chem. Ind. Chem. Eng. Q. 23 (1) 19 (2017)
3
variance in the data set. The following principal com-
ponents account for the maximum of the remaining
variance in the data set. Each PC is characterized by
2-D scores plot maps, which shows a data overview
and similarities or dissimilarities within the data. In
this study PCA was employed to establish: i) the
differences between nutrients release from the pure
PR and mechanically activated substrates (MACp/PR
and MANH4-Cp/PR mixtures) by calculating score
plots and ii) the influence of process parameters (zeo-
lite/PR ratio and the mixing time) on the content of
nutrients (P, Ca2+, K+, Na+ and Mg2+) in the solutions
of both MA-mixtures by calculating correlation loading
plots. All PCA calculations were carried out by PLS
ToolBox, (Eigenvectors Inc., v. 7.9), for Matlab 7.12.0
(R2011a) and The Unscrambler (version 9.7, CAMO
Process AS, Oslo, Norway). PCA was applied by
using a 0.95 confidence level for Q and T2 Hotelling
limits for outliers and a singular value decomposition
algorithm (SVD). All data were auto scaled prior to
any PCA analysis.
RESULTS AND DISCUSSION
Mineral characterization
The mineralogical compositions of Cp/PR and
NH4-Cp/PR substrates (with highest zeolite share,
15:1) before and after mechanical treatment are
shown in Figure 1. All mixtures irrespective of the
composition were of similar mineralogical structure
(for which XRD patterns for mixtures at 5:1 and 10:1
Cp/PR ratio are omitted) and contained Heu-type
zeolite, quartz, plagioclase, muscovite, and apatite.
The crystallinity degree of minerals between non-act-
ivated and MA samples remained unchanged. Also,
systematic shifting of diffraction maximums of the
dominant mineral in the mixture, Cp, was rather neg-
ligible. Crystallite size for Heu-type zeolite measured
at (020) and (200) diffraction maximums were: Cp/PR
for <D> = 212 Å, MACp/PR for <D> = 283 Å, NH4-
-Cp/PR for <D> = 207 Å and MANH4-Cp/PR for <D> =
= 227 Å. The observed differences were due to the
presence of larger quantities of quartz and plagio-
clases in the Cp/PR and NH4Cp/PR, in relation to the
corresponding MA-mixtures, respectively (Figure 1a
and b).
The influence of mechanical activation on nutrient
release from the Cp/PR fertilizer mixtures
Solution P and Ca2+ concentrations released
from the MA-mixtures were compared with the con-
centration of the same elements leached from corres-
ponding non-activated mixtures [6], subject to their
composition and mixing time (Figure 2).
Figure 1. Diffractograms of Cp/PR and MACp/PR mixtures (a) and NH4-Cp/PR and MANH4-Cp/PR mixtures (b).
M. MIHAJLOVIĆ et al.: TO WHAT EXTENT DO SOFT MECHANICAL ACTIVATION… Chem. Ind. Chem. Eng. Q. 23 (1) 19 (2017)
4
The concentrations of P released from the
MACp/PR mixtures ranged between 2.81 and 3.19
mg L-1 (Figure 2a). This is on average fifteen times
higher P concentration than those released from the
corresponding non-activated mixtures (0.15 to 0.26
mg L-1) [6]. The observed increase in P concentra-
tions due to soft mechanical activation of the mixtures
was in accordance with the expected upward trend for
P found in the literature [21]. Solution Ca2+ concentra-
tions gradually increased with the increase of the
MACp/PR ratio and the mixing time and were
between 11 and 22.5 mg L−1. However, Cp/PR and
MACp/PR substrates had similar solution Ca2+ con-
centration implying that mechanical activation did not
affect the changes in solution Ca (Figure 2b).
Solution K+, Na+ and Mg2+ concentrations rel-
eased from the MACp/PR and the comparable Cp/PR
mixtures [6] are shown in Figure 3. The increase in
MACp/PR ratio and contact time caused a slight inc-
rease of K+, Na+ and Mg2+ concentrations in the sol-
ution. Solution K+ and Na+ concentrations of the
MACp/PR mixtures were between 5.66-7.99 mg L-1
and 11.7-17.15 mg L-1, respectively. This is on aver-
age 30% lower than that of the corresponding non-
activated mixtures. Petkova et al. [20] previously
reported that tribochemical activation of the Cp/PR
leads to decrease of the Cp ion-exchange ability
whereas at the same time dissolution of the PR has
been increased [20]. Solution Mg2+ concentration of
the activated mixtures ranged between 0.83-1.65 mg
Figure 2. Mean solution P (a) and Ca2+ (b) concentrations ± SD of the Cp/PR [6] and MACp/PR mixtures.
Figure 3. Mean solution K+, Na+ and Mg2+ concentrations ± SD of the Cp/PR [6] and MACp/PR mixtures at a) 5:1, b) 10:1 and
c) 15:1 zeolite/PR ratio.
M. MIHAJLOVIĆ et al.: TO WHAT EXTENT DO SOFT MECHANICAL ACTIVATION… Chem. Ind. Chem. Eng. Q. 23 (1) 19 (2017)
5
L-1 or around twofold higher in relation to the corres-
ponding Cp/PR mixtures (Figure 3).
The influence of mechanical activation on nutrient
release from the NH4-Cp/PR fertilizer mixtures
Solution P and Ca2+ concentrations of the
MANH4-Cp/PR mixtures were compared with the con-
tent of same nutrients leached from corresponding
non-activated mixtures [6], subject to their compo-
sition and the mixing time (Figure 4).
The mean concentrations of P released from the
MANH4-Cp/PR mixtures were between 2.02 and 7.74
mg L-1, which is ten times higher than those released
from the corresponding non-activated mixtures (0.36-
0.82 mg L-1) [6]. Similarly, 10-15 times higher extract-
ion of P to the acid media from the MA-phosphorite
has been reported previously [21]. Figure 4a shows
that in addition to mechanical activation, an increase
of the content of partially saturated NH4–zeolite in the
mixture and duration of contact between the two min-
erals significantly contributes the solubility of the PR.
Solution Ca2+ concentrations released from the
MANH4-Cp/PR mixtures varied between 445 and
510.5 mg L-1. The differences between dissolved Ca
from the NH4-Cp/PR and MANH4-Cp/PR were most
evident at 5:1 zeolite/PR ratio (Figure 4b). At higher
zeolite/PR ratios, the content of Ca in the solutions
was in favor of the non-activated mixtures. Further-
more, the Ca2+ concentration slightly decreased with
the increase of mixing time and Cp/PR ratio, similar to
the non-activated mixtures previously reported by
Mihajlovic et al. [6]. These results support the con-
clusions from the literature that the released Ca2+
from the PR, compelled by cation exchange with
NH4+, was partially withdrawn by the zeolite before
equilibrium [20,26].
Solution K+ and Na+ concentrations of the
MANH4-Cp/PR mixtures ranged from 40 to 59.4 mg L-1
and from 7.14 to15.45 mg L-1, respectively (Figure 5).
The content of both elements in the solution slightly
increased with the Cp/PR ratio increase, but ranged
around similar values in relation to the mixing time.
Nevertheless, in comparison to the non-activated mix-
tures, the K+ and Na+ concentrations in MA-fertilizers
with saturated zeolite were increased by 30 and 50%,
respectively. This supports the utilization of soft
mechanical activation, aiming to increase the leach-
ing of K from the tested fertilizer mixtures. Solution
Mg2+ concentrations in the MANH4-Cp/PR substrates
proportionally increased with increasing of zeolite
share in the mixtures and the mixing time, and ranged
between 20 and 54.5 mg L-1, very much alike to that
of the corresponding non-activated mixtures [6].
Figure 4. Mean solution P (a) and Ca2+ (b) concentrations ± SD of the NH4-Cp/PR [6] and MANH4-Cp/PR mixtures.
M. MIHAJLOVIĆ et al.: TO WHAT EXTENT DO SOFT MECHANICAL ACTIVATION… Chem. Ind. Chem. Eng. Q. 23 (1) 19 (2017)
6
ANOVA
One-way ANOVA was applied to compare the
means between the concentrations of nutrients rel-
eased from the MACp/PR and the MANH4-Cp/PR
mixtures of different composition at various mixing
times (Tables 1 and 2). Statistical differences were
determined at the p < 0.05 level, 95% confidence
limit, according to Tukey’s HSD test.
Calculated one-way Fisher’s statistic critical
value was Fcrit = 5.14 and Tukey’s HSD post hoc
critical value was 4.34 for each element in both group
of mixtures. These values numerically define the
levels of significant differences in nutrient content,
released from the various mixtures according to their
composition and mixing time. If the obtained statistical
value, F and/or Tukey HSD value (between labeled
groups) for each element is larger than its critical
value, the differences in concentrations of the rel-
eased nutrient from the compared groups of mixtures
are significant at 0.05 levels.
Conversely to solution P concentrations, the
differences between solution Ca2+, K+, Mg2+ and Na+
concentrations released from the MACp/PR sub-
strates, were found to be significant in relation to the
composition of the mixtures (Table 1). Regarding the
calculated Tukey HSD values from Table 1, the con-
centrations of K+ released from the MACp/PR mixture
at 5:1 ratio and Mg2+ released from the MACp/PR
mixture at 15:1 ratio, were found to be significantly
different in comparison to the other two groups. The
effect of mixing time did not affect the occurrence of
significant differences between solution concen-
trations of the elements within the groups (Table 1).
Significant differences in relation to the mixture
composition were also found for K+ and Mg2+ released
from MA-substrates with saturated Cp, while the con-
centrations of Na+ released from the mixtures with
lowest NH4+-Cp share were significantly different in
comparison to the other two groups. Differences in
composition of the mixtures did not significantly affect
solution Ca2+ and P concentration, while the effect of
mixing time in the MANH4-Cp/PR mixtures was sig-
nificant at the p < 0.05 level only for P (Table 2).
Figure 5. Mean solution K+, Na+ and Mg2+ concentrations ± SD of the NH4-Cp/PR [6] and MANH4-Cp/PR mixtures at a) 5:1, b) 10:1 and
c) 15:1 zeolite/PR ratio.
Table 1. ANOVA of the MACp/PR mixtures; F – one way Fisher`s statistic test; p-value - function of the observed sample results;
HSD – Tukey`s post hoc test
Parameter P ratio Ca ratio K ratio Na ratio Mg ratio Parameter P time Ca time K time Na time Mg time
F 2.07 85.75 76.14 84.97 8.57 F 1.94 0.09 0.06 0.09 0.22
p-Value 0.21 3.8610-5 5.4510-5 3.9710-5 0.02 p-Value 0.22 0.91 0.94 0.92 0.81
HSD|15:1-10:1| - 6.87 3.25 8.55 4.45 HSD|72h-48h| - - - - -
HSD|15:1-5:1| - 18.33 16.47 18.42 5.53 HSD|72h-24h| - - - - -
HSD|10:1-5:1| - 11.46 13.23 9.87 1.09 HSD|48h-24h| - - - - -
Table 2. ANOVA of the MANH4-Cp/PR mixtures; F – one way Fisher`s statistic test; p-value - function of the observed sample results;
HSD – Tukey`s post hoc test
Parameter P ratio Ca ratio K ratio Na ratio Mg ratio Parameter P time Ca time K time Na time Mg time
F 0.88 2.03 222.12 99.67 83.27 F 7.53 1.82 0.02 0.02 0.06
p-Value 0.46 0.21 2.3710-6 2.4910-5 4.210-5 p-Value 0.02 0.24 0.98 0.98 0.94
HSD|15:1-10:1| - - 11.904 0.32 8.72 HSD|72h-48h| 1.64 - - - -
HSD|15:1-5:1| - - 29.62 17.45 18.25 HSD|72h-24h| 5.36 - - - -
HSD|10:1-5:1| - - 17.71 17.13 9.53 HSD|48h-24h| 3.72 - - - -
M. MIHAJLOVIĆ et al.: TO WHAT EXTENT DO SOFT MECHANICAL ACTIVATION… Chem. Ind. Chem. Eng. Q. 23 (1) 19 (2017)
7
Principal component analysis
PCA was performed in order to explore the main
variation patterns between the data for pure PR,
MACp/PR and MANH4-Cp/PR mixtures in relation to
the different process parameters (Figure 6). The PCA
score plot with samples labeled in accordance to the
zeolite/PR ratio is presented in Figure 6a and the
PCA score plot with samples labeled in accordance to
mixing time is presented in Figure 6b. PCA resulted in
a two-component model, which explains 93.27% of
total variance. The first principal component, PC 1,
accounted for 70.67% and the second one, PC 2, for
22.60% of the overall data variance. The addition of
more PCs did not change the classification of the
samples. Taking into account PC1 and PC2 score
values, pure PR samples are grouped in the lower-left
part of the PCA score plot, MACp/PR samples in the
upper-left and MANH4-Cp/PR are separated in the
right part of the PCA score plots.
The zeolite share in the fertilizer mixtures caused
greater clustering of different MACp/PR samples, vis-
ible along positive PC2 axis. However, the overall imp-
act of the substrate composition is more evident in the
case of the MANH4-Cp/PR mixtures with obvious greater
separation of the samples along the first component.
Also, the highest zeolite share has the strongest
influence on solution cation concentrations (Figure 6a).
The different mixing times better affected the
separation of the MANH4-Cp/PR mixtures than of the
MACp/PR and the PR alone which is visible along
PC1 axis (Figure 6b). Furthermore, the influence of
mixing time increases as the zeolite/PR ratio inc-
reases, implying that the strongest effect of mixing
time is observed for the substrates with 15:1 zeo-
lite/PR ratio. Contrary to this, the influence of the
mixing time has a very weak effect on the MACp/PR
substrates. This indicates that changes in process
parameters had a greater impact on the release of
nutrients from the MANH4-Cp/PR mixtures.
To obtain a better overview of the effects of
zeolite/PR ratio in the mixtures and mixing time on
solution cation concentrations, two additional principal
component analyses were performed on the subsets
of data: for MACp/PR and for MANH4−Cp/PR samples
using correlation loading plots (Figure 6c and 6d).
From the correlation loading plot calculated for
MACp/PR substrates, it can be observed that the
highest zeolite/PR ratio exhibits the strongest influ-
ence on solution P, Ca2+, K+, Na+, and Mg2+ concen-
trations. The P content in the solution correlates the
best with MACp/PR ratio of 15:1.
Figure 4. Mean solution P (a) and Ca2+ (b) concentrations ± SD of the NH4-Cp/PR [6] and MANH4-Cp/PR mixtures.
M. MIHAJLOVIĆ et al.: TO WHAT EXTENT DO SOFT MECHANICAL ACTIVATION… Chem. Ind. Chem. Eng. Q. 23 (1) 19 (2017)
8
The correlation loading plot calculated for
MANH4–Cp/PR substrates shows that the highest
zeolite/PR ratio exhibits the strongest influence on
solution P, K+, Na+ and Mg2+ concentrations, while the
Ca content was in best correlation with the lowest
zeolite share in the mixture and the shortest mixing
time. Also, Ca is highly negatively correlated with P.
The best correlation between solutions P concentra-
tion, was found for 72 h mixing time and MANH4-
–Cp/PR ratio of 15:1.
CONCLUSION
The obtained results showed that the use of soft
mechanical activation of tested fertilizer mixtures fol-
lowed by inducted particle size reduction, although
did not produce significant structural changes in min-
erals, notably intensified the passage of nutrients to
the liquid media, especially P. Solution P concentra-
tions of both MA substrates were up to fifteen times
higher than that of the corresponding non-activated
mixtures, while the release of K+ from the MANH4-
–Cp/PR mixtures was increased by a third. Further-
more, all concentrations of the nutrients released from
the MANH4-Cp/PR substrates were within their opti-
mal values necessary for plant growth [27,28]. Sol-
ution Ca2+ concentrations were between 445 and
510.5 mg L−1, which suggests that the use of the
MANH4-Cp/PR mixtures may resolve a potential defi-
ciency of the Ca2+ in the solution, very common for
substrates of similar composition [6].
PCA revealed that the highest NH4-zeolite share
in the mixture had the most positive impact on the PR
dissolution. A growth of fertilization potential with
time, particularly of the MANH4-Cp/PR mixtures, was
observed. This supports the use of the selected
MANH4–Cp/PR mixture as a slow-release fertilizer,
very favorable for plants since fertilization can be per-
formed less frequently, which, besides the efficiency,
increases the cost-effectiveness of its utilization.
Acknowledgment
The authors are grateful to the Serbian Ministry
of Education, Science and Technological Develop-
ment for the financial support of this investigation
included in the project TR 31003, projects cycle
2011−2015.
REFERENCES
[1] L. Maene, Phosphate fertilizer production, consumption
and trade, the present situation and outlook to 2010, The
Sulphur Institute's 17th Sulphur Phosphate Symposium,
January 17-19, Boca Raton, FL, 1999
[2] M. Hart, B., Quin; M. Nguyen, J. Environ. Qual. 33 (2004)
1954−1972
[3] L. Leon, W. Fenster, L. Hammond, Soil Sci. Soc. Am. J.
50 (1986) 798−802
[4] S. Chien, In: Direct application of phosphate rock and
related technology:latest development and practical exp-
eriences; S. Rajan, S. Chien (Eds.), Special Publications
IFDCSP- 37, IFDC, Muscle Shoals, AL, 2003, pp. 50−62
[5] W. Pickering, N. Menzies, M. Hunter, Sci. Hortic. 94
(2002) 333−343
[6] M. Mihajlović, N. Perišić, L. Pezo, M. Stojanović, J.
Milojković, Z. Lopičić, M. Petrović, J. Agric. Food Chem.
62 (2014) 9965−9973
[7] D. Ming, E. Mumpton, in Minerals in Soil Environments,
2nd ed., J. Dixon, S. Weed (Eds.), Soil Science Society of
America, Madison, WI, 1989, pp. 873−911
[8] C. Cobzaru , S. Oprea, Chem. Ind. Chem. Eng. Q. 11
(2005) 206-212
[9] C. Cobzaru, C. Cibotaru, A. Rotariu, A. Marinoiu, S.
Oprea, Chem. Ind. Chem. Eng. Q. 15 (2009) 63−67
[10] M. Onyango, J. Kittinya, N. Hadebe, V. Ojijo, A. Ochieng,
Chem. Ind. Chem. Eng. Q. 17 (2011) 385−395
[11] M. Mihajlović, S. Lazarević, I. Janković-Častvan, B. Jokić,
Dj. Janaćković, R. Petrović, Chem. Ind. Chem. Eng. Q.
20 (2014) 283-293
[12] K. Barbarick, T. Lai, D. Eberl, Soil Sci. Soc. Am. J. 54
(1990) 911−916
[13] E. Allen, L. Hossner, D. Ming, D. Henninger,. Soil Sci.
Soc. Am. J. 57(1993)1368−1374.
[14] [14] E. Allen, L. Hossner, D. Ming, D. Henninger, Soil Sci.
Soc. Am. J. 60 (1996) 1467−1472
[15] S. Ibrahim, A. El-Midany, T. Boulos, Physicochem. Probl.
Miner. Process. 44 (2010) 63-78
[16] B. Fotoohi, A study of mechanochemical activation in
solid-state synthesis of advanced ceramic composites, A
thesis submitted to The University of Birmingham, 2010
[17] N. Shulga, Russ. J. Appl. Chem. 85 (2012)1297−1306
[18] L. Shumskaya, E. Kirillova, T. Yusupov, J. Mining Sci. 35
(1999) 96-100
[19] N. Petrova, V. Petkova, Bulgarian Chem. Commun. 43
(2011) 301–307
[20] V. Petkova, E. Serafimova, N. Petrova, Y. Pelovski, J
Therm. Anal. Calorim. 105 (2011) 535–544
[21] T. Yusupov, L. Shumskaya, J. Mining Sci. 38 (2002) 177–181
[22] T. Yusupov, L. Shumskaya, E. Kirillova, V. Boldyrev, J.
Mining Sci. 42 (2006) 189-194
[23] M. Mihajlović, N. Perisic, L. Pezo, M. Stojanovic, J.
Milojković, M. Petrović, J. Petrović, Clay Minerals 49
(2014) 735-745
[24] R. Koenig, C. Johnson, Ind. Eng. 14 (1942) 155−156
[25] K. Pearson, Philos. Mag. 2 (1901) 559−572
[26] T. Lai, D. Eberl, Zeolites 6 (1986) 129−132
[27] S. Tisdale, W. Nelson, J. Beaton, In: Soil fertility and
fertilizers; Macmillan Publ. Co., New York, 1985
[28] G. Bryson, A. Barker, Commun. Soil Sci. Plant Anal. 33
(2002) 759−777.
M. MIHAJLOVIĆ et al.: TO WHAT EXTENT DO SOFT MECHANICAL ACTIVATION… Chem. Ind. Chem. Eng. Q. 23 (1) 19 (2017)
9
MARIJA MIHAJLOVIĆ
MARIJA STANOJEVIĆ
MIRJANA STOJANOVIĆ
JELENA PETROVIĆ
JELENA MILOJKOVIĆ
MARIJA PETROVIĆ
ZORICA LOPIČIĆ
Institut za tehnologiju nuklearnih i
drugih mineralnih sirovina, Franše
D'Eperea 86, 11000 Beograd, Srbija
NAUČNI RAD
U KOJOJ MERI BLAGA MEHANIČKA AKTIVACIJA I PROCESNI PARAMETRI POVEĆAVAJU EFIKASNOST RAZLIČITIH ZEOLIT/RUDA FOSFORA SMEŠA PRIRODNIH ĐUBRIVA?
U cilju dobijanja efikasnog mineralnog đubriva, različite smeše rude fosfora (PR) sa pri-
rodnim klinoptilolitom (Cp) i NH4+-klinoptilolitom (NH4-Cp) podvrgnute su blagoj meha-
ničkoj aktivaciji. Srednje koncentracije P otpuštene iz mehanički aktiviranih (MA) smeša,
MACp /PR i MANH4-Cp/PR, varirale su u rasponu od 2,81-3,19 mg L-1 i 2,02-7,74 mg L-1,
redom. Ovo su 10 do 15 puta vece koncentracije P od onih otpuštenih iz odgovarajucih
neaktiviranih smeša. Koncentracije Ca2+, K+, Na+ i Mg2+ u rastvorima smeša varirale su u
zavisnosti od sastava smeše i vremena kontakta dva minerala, a u okviru njihovih opti-
malnih vrednosti potrebnih za rast i razvoj biljaka. Dobijeni rezultati pokazuju da se efikas-
nost NH4-Cp/PR smeše može značajno povecati predloženim postupkom mehaničke
aktivacije. Uticaj procesnih parametara na sadržaj posmatranih nutrijenata prikazan je upo-
trebom multivarijantne statističke analize. Najveci potencijal đubrenja pokazala je MANH4-
-Cp/PR smeša sa najvecim udelom zeolita i najdužim predloženim vremenom mešanja.
Ključne reči: smeše đubriva, ruda fosfora, klinoptilolit, mehanička aktivacija, multi-
varijantna analiza.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 1119 (2017) CI&CEQ
11
FATIH ILHAN
KAAN YETILMEZSOY
HARUN AKIF KABUK*
KUBRA ULUCAN
TAMER COSKUN*
BUSRA AKOGLU
Department of Environmental
Engineering, Faculty of Civil
Engineering, Yildiz Technical
University, Davutpasa, Esenler,
Istanbul, Turkey
SCIENTIFIC PAPER
UDC 66.094.3:628.3:677
https://doi.org/10.2298/CICEQ150907048I
EVALUATION OF OPERATIONAL PARAMETERS AND ITS RELATION ON THE STOICHIOMETRY OF FENTON’S OXIDATION TO TEXTILE WASTEWATER
Article Highlights
• Operation conditions for implementation of the Fenton process were re-evaluated
• COD/H2O2, COD/Fe2+ and H2O2/Fe2+ ratios were optimized by physico-chemical studies
• The effects of important parameters on TOC and color removal were investigated
• A new operational table was produced on the basis of stoichiometric ratios
• Time-dependant dosing of H2O2 helps to obtain higher efficiency levels
Abstract
The operation conditions for the implementation of the Fenton process are of
utmost importance because there are problems related to the proportion of H2O2
dosage to COD and Fe2+ dosage to the specified H2O2 amount. The relevant lite-
rature shows that COD/H2O2 ratios range between 0.0084 and 113.9. Similarly,
the COD/Fe2+ ratio varies between 0.079 and 292.6, while the H2O2/Fe2+ ratio
varies between 0.09 and 287. Moreover, the ratio of the maximum value to the
minimum value used in the operations on the basis of COD is 13560 for COD/
/H2O2 (0.0084-113.9), 2210 for COD/Fe2+ (0.079-174.7), and finally 3190 for
H2O2/Fe2+ (0,09-287). The aim of this study was to re-evaluate these values that
significantly differ from each other with specific emphasis on textile wastewater
and considering stoichiometric ratios. Results showed that values ranging
between 0.43 and 4.0 for COD/H2O2 and those ranging between 0.75 and 3.0 for
H2O2/Fe2+ are more suitable. The results showed that in a Fenton process con-
ducted by dosing H2O2 at different times, the TOC reduction efficiencies inc-
reased from 80.8 up to 88.9%. Similarly, the color reduction efficiency also rose
from 96.5 to 98.7%.
Keywords: Fenton’s oxidation; operational parameters; oxidation; stoi-chiometric ratio, TOC removal.
Wastewaters are treated physically, chemically,
and biologically. The physical methods use separ-
ation processes rather than a treatment processes. At
the end of these processes, contaminants are gener-
ated in a concentrated manner [1]. At the end of the
biological processes, on the other hand, one of the
contaminants that one emerges is sludge [2]. Even
Correspondence: K. Yetilmezsoy, Department of Environmental
Engineering, Faculty of Civil Engineering, Yildiz Technical Uni-
versity, 34220, Davutpasa, Esenler, Istanbul, Turkey. E-mail: [email protected] * These authors do not have an active academic position at the
Faculty of Civil Engineering, Yildiz Technical University. Paper received: 7 September, 2015 Paper revised: 3 December, 2015 Paper accepted: 8 December, 2015
though the contaminants generated by chemical
methods, such as clarification-type processes, are dif-
ferent, they are still present inside the chemical sludge
[3], whereas the contaminants can be fully avoided by
using advanced treatment processes. The oxidation
processes in particular aid in the complete elimination
of these contaminants. The Fenton process has a
special characteristic among the other oxidation pro-
cesses. The difference of the Fenton process is that it
generates ●OH, which helps introduce the synergistic
effect of peroxide, an oxidant, the iron catalysis [4].
This process is especially efficient for COD parame-
ters that are difficult to reduce, when the BOD/COD
ratio is low. When a Fenton process is applied to a
COD parameter, which cannot be reduced by means
F. ILHAN et al.: EVALUATION OF OPERATIONAL PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 1119 (2017)
12
of other processes, high efficiencies can be obtained.
The reason for this is that the oxidation capacity of
hydroxyl radicals is higher than the other oxidants [4].
Based on this, one can conclude that the Fenton pro-
cess can be used on a variety of wastewaters such as
those produced by activities related to textiles [5],
leachates [6], paints [7], olive mills [8], pharmaceut-
icals [9], tanneries [10] and surfactants [11]. In addi-
tion, whenever the desired level of efficiency cannot
be obtained by biological refinement, the Fenton pro-
cess can also be used as a final treatment method to
shift COD values to the desired levels [12].
In spite of these advantageous characteristics,
the Fenton process brings about a number of import-
ant economic problems [13]. The major reason behind
this is that normally the peroxide dose and the
amount of iron to be used cannot be finely adjusted.
The addition of peroxide and iron in measurements
that are less than needed causes low efficiency
levels. On the other hand, excessive use of peroxide
creates problems, not only because it requires reduct-
ion (reduction of excessive peroxide), but also it
results in additional costs. Similarly, the addition of
excessive iron also brings about extra costs and also
results in extremely significant operational problems
such as the sludge caused by too much iron. For
these reasons, the selection of operational conditions
for a Fenton process is of utmost importance in terms
of a correct and efficient operation [4].
The aim of this study was to avoid eventual red-
undant and excessive H2O2 and Fe2+ additions, and
hence to prevent the formation of redundant sludge
and excessive peroxide, based on the stoichiometric
ratios that should be ensured by selecting the appro-
priate operational conditions. In addition, a physico-
chemical study was carried out on a textile waste-
water sample, by using less chemicals on the basis of
stoichiometric ratios, and it was aimed to demonstrate
that high reduction efficiencies can actually be
obtained by using low amounts of chemicals.
MATERIALS AND METHODS
Selection of operational parameters
In this study, the effects of the operational con-
ditions such as initial pH value, and COD/H2O2 and
H2O2/Fe2+ ratios on the Fenton process were inves-
tigated. Since the priority of this study was to define
COD/H2O2 and H2O2/Fe2+, these parameters were
investigated in more detail. After defining the optimum
values, an optimization study was carried out on the
basis of pH values. Finally, the effect of adding the
initial peroxide by dosing on the results was analysed.
Also, a comparative analysis between the first 9
studies and the study-set between 29 and 38 will pro-
vide further insight regarding the effect of dosing the
peroxide to add to the results.
Wastewater characteristics
Textile wastewater was selected as the study
case. The textile wastewater samples were taken
from the Akinal textile factory (Cevizlibag, Istanbul,
Turkey) and kept at 4 C. The wastewater parameters
were regularly measured during this study. The main
characteristics of the textile wastewater samples used
in the experiments are shown in Table 1.
Table 1. Characterization of textile wastewater from Akinal tex-
tile factory; SD – standard deviation
Parameter Unit Mean + SD
Chemical oxygen demand (COD) mg/L 1625±40
Five-day biochemical oxygen demand
(BOD5)
mg/L 570±15
BOD5/COD - 0.35
TOC/COD - 0.32
pH - 4.3±0.1
Total organic carbon (TOC) mg/L 524±10
Color Pt-Co 545±10
Conductivity µS/cm 2370±50
Fenton’s oxidation
A stock solution of 10 g/L of Fe2+ was prepared
by dissolving FeSO47H2O (Merck Chemical Corp.) in
distilled water. In addition to iron sulfate reagent, 30%
H2O2 solution (Merck Chemical Corp.) with a density
of 1.11 kg/L was used in the oxidation process. In
each oxidation test, 500 mL of textile wastewater
sample was collected from the textile industry efflu-
ent. In the first step of Fenton’s oxidation process, the
pH of the textile wastewater was adjusted to the
desired value by the addition of 1 M H2SO4 and 1 M
NaOH. During the whole oxidation process, the pH of
samples were also set at the desired value by adding
these reagents (1 M H2SO4 and 1 M NaOH) gradually
in addition to the pre-adjustment of the pH. The
FeSO47H2O and H2O2 solutions were then added to
the effluent sample and conducted for 5 min of rapid
mixing at 120 rpm using a Jar Test Equipment (VELP
Scientifica, FC6S). The effluent sample was then
gently stirred at 10 rpm for 25 min. After the floccul-
ation process, the sample transferred to a graduated
settling column for 30 min of settling. 100 mL of
supernatant sample was then collected for the further
analyses (TOC and color) after the settling process.
In order to prevent interferences in analytical mea-
surements, the pH of collected supernatant sample
F. ILHAN et al.: EVALUATION OF OPERATIONAL PARAMETERS… Chem. Ind. Chem. Eng. Q. .23 (1) 1119 (2017)
13
was increased to about 6.0 by adding 6 M NaOH gra-
dually to precipitate Fe2+ in the form of Fe(OH)3. Fin-
ally, the MnO2 reagent was then added to remove the
residual H2O2 from the collected supernatant [26].
Analytical procedure
The pH of wastewater samples was measured
by a pH meter (WTW series pH 720) and a pH probe
(WTW, pH-Electrode Sentix 41). The color of waste-
water samples was measured with a Hach Lange
spectrofotometer (model: DR 5000) and determined
as platinum-cobalt (Pt-Co) color unit according to
method 120. Electrical conductivity was measured by
using a multimeter instrument (Hach Lange HQ 40D).
The total organic carbon (TOC) was measured by
using a Hach Lange IL 550 TOC/TN analyzer. All
other experimental analyses were performed by the
procedures described in the Standard Methods of
APHA [34]. These parameters were determined by
the procedures described in method numbers of 5220
C (closed reflux, titrimetric method for COD) and 5210
B (5-day BOD test). The deionized water used in the
experiments was supplied from a purification system
(Meck Millipore Direct-Q 3, 5, 8 Ultrapure Water Sys-
tems). The analyses were carried out at least three
times for each sample to assess method precision.
Stability of the oxidation process and components of
wastewater samples were monitored in the Environ-
mental Engineering Laboratory at Yildiz Technical
University in Istanbul, Turkey.
RESULTS AND DISCUSSION
Literature review on the Fenton’s oxidation process
When the data reported in the relevant literature
are investigated, it is seen that the doses selected
both for peroxide and Fe2+ parameters differ greatly
from each other, even when they are proportional to
the COD values. A variety of the most recent studies,
selected among those available in the literature, are
shown in Table 2.
The cells in normal characters in Table 2 are
those directly reported by the mentioned studies. The
cells given in bold characters, on the other hand,
were calculated based on the values reported in
those studies, because each study reported a dif-
ferent ratio (H2O2/Fe2+, COD/H2O2, COD/Fe2+, Fe2+/
/H2O2, H2O2/COD and Fe2+/H2O2). The obtained res-
ults are quite interesting for a number of reasons. As
seen in Table 2, the H2O2 values determined on the
basis of COD are given in terms of concentrations.
Similarly, Fe2+, which are determined based on the
added H2O2 amount in terms of a mole ratio, are also
reported in terms of concentrations. However,
although it is more practical to report these values in
terms of concentrations, this brings about an impor-
Table 2. Comparison of different process typologies on Fenton’s oxidation
COD / mg L–1 pH Time
min
H2O2 dosage
mg/L
Fe2+ dosage
mg/L t / C COD/H2O2 COD/Fe2+ H2O2/Fe2+ Reference
5320 2-9 60 340-15300 280-5600 - 0.35-15.65 0.95-19 0.2-10 [14]
- 2.5-7 5-240 60-450 0-15 - - - 4-45 [15]
3242 2-6 30-120 16950-42375 200-1000 - 0.076-0.19 3.24-16.21 33.9-169.5 [16]
1140 2-8 0-60 1000-6500 150-1000 - 0.175-1.14 1.14-7.6 3.2-21.67 [17]
11987 2-4 90 13185- 52742.8 217-8686 - 0.23-0.91 1.38-55.2 10-100 [18]
314-404 2-5 0-100 - - - 0.126-0.315 0.609-3.125 1.99-9.90 [19]
564 2.5-7.0 0-300 102-510 22.4-100.8 30-60 1.11-5.53 5.6-25.2 0.61-2.43 [5]
176 ± 13.2 3-5 0-180 187-2240 560-2240 - 0.079-0.94 0.079-0.314 1-3 [20]
725 4 0-30 0-100 0-50 - 7.25-72.5 14.5-145 5-10 [21]
25624 0.75-3.75 35-255 225-900 500-4500 - 28.47-113.9 5.69-51.25 0.09-1.35 [22]
1200 5 - 50-300 20-160 - 4-24 7.5-60 0.5-5 [23]
2150-2770 2-5.5 - 19300-57750 222-2196 25-70 0.043- 0.127 1.12-11.08 96-287 [24]
7500-8400 5.4-9.1 0-150 5780-18,020 300-3000 - 0.44-1.37 2.65–26.5 1.93-34 [25]
1750 2-7 60 200-1200 100-1000 - 1.46-8.75 1.75-17.5 0.5-12 [26]
2533 1.5-3.5 30 44000-266000 552-2210 20-40 0.0095-0.058 1.15-4.59 19.91-201.1 [27]
69600-174000 3-5 0-240 1700-5100 204-3050 - 13.65-102 22.82-57 1.67-8.33 [28]
1670 3-7 0-180 15000-200000 13.8-1380 5-20 0.0084-0.111 1.21-121 0-5 [29]
4528 1-7 30-240 100-800 100-800 - 5.66-45.28 5.66-45.28 0.4-5 [30]
11620-85300 3 - 19800-55200 1250-5000 - 0.23-3.40 6.4-25.6 8-22 [31]
575-2271 2.25-2.47 40 39.0-252.9 13.0-84.3 - 2.27-58 6.82-174.69 3.0 [32]
F. ILHAN et al.: EVALUATION OF OPERATIONAL PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 1119 (2017)
14
tant problem. When these values are proportioned to
COD or as H2O2/Fe2+, they result in values that greatly
differ from each other.
As can be seen from Table 2, the COD/H2O2
ratio ranged between 0.0084 and 113.9. Similarly,
COD/Fe2+ ratios varied between 0.079 and 174.7, and
H2O2/Fe2+ ratios varied between 0.09 and 287. These
intervals are quite high. Since adding less H2O2 than
necessary brings about problems in terms of effi-
ciency, and excessive H2O2 causes problems not only
in efficiency but also in the reduction of excessive
peroxide, these large intervals pose a great challenge
in this oxidation process. For this reason, the ratio of
COD to the amount of H2O2 used should be deter-
mined stoichiometrically [33]. Reactions occur
smoothly by using 10% more than the stoichiometric
ratios. More than this amount should not be used as
this will cause excessive ●OH in water. The stoichio-
metric COD/H2O2 ratios to be used in the Fenton pro-
cesses can be found in the literature [8]. Based on the
stoichiometric ratio [8], it can be stated that for 2.125
g of H2O2 (or 0.0625 mol H2O2), 1 g of COD (or 1 g O2
= 0.03125 mol O2) is needed with a ratio of about
0.471. Considering a 10% margin of safety, it was
determined that the COD/H2O2 ratio should be equal
to 0.428 or higher. This is due to the fact that it is imp-
ortant for economic reasons to evaluate this ratio in
proportions such as 1:1, even 1:0.5 and 1:0.25, rather
than 1:2.125, considering the synergistic effect.
Therefore, it is quite plausible to use values ranging
between 0.428 and 4.0 for the COD/H2O2 ratio. The
COD/H2O2 ratios lower than 0.428 indicate use of
excessive peroxide, which might result in significant
problems from an engineering perspective.
Another important parameter is the H2O2/Fe2+
ratios. When the effect of H2O2 ions is determined (in
those studies where different doses of H2O2 are
added), a single dose is specified and used for the
Fe2+. However, while the specified Fe2+ amounts are
excessive for low doses of H2O2, it is generally insuf-
ficient for high doses of peroxide additions. Although
the higher doses normally induce better results, this is
not completely due to the oxidation effect of ●OH. The
added Fe2+ transform a part of the H2O2 into ●OH,
while the oxidation effect of the remaining H2O2 plays
an important role in the increase of efficiency. More
Fe2+ additions might help transform all H2O2 into ●OH
and therefore increase the oxidation capacity; how-
ever, it is not economically feasible to use the oxid-
ation capacity of peroxide rather than forming more ●OH. For this reason, in order to obtain more correct
results, the ratio of H2O2 to Fe2+ should be kept cons-
tant, rather than fixing the H2O2 amount in Fe2+ opti-
mization or the Fe2+ amount in H2O2 optimization. To
this end, for the first peroxide optimization, a H2O2/
/Fe2+ ratio, which is well-balanced in terms of its stoi-
chiometric value and reported to be successful by the
relevant literature, should be specified and iron
dosing should be carried out based on this ratio.
Then, the most suitable H2O2/Fe2+ ratios should be
defined by keeping the specified optimum H2O2 dose
constant and modifying the Fe2+ amount. When the
literature regarding the H2O2/Fe2+ ratios are inves-
tigated, it is seen that the values reported change
range within a very wide interval. In certain studies
[22] the H2O2/Fe2+ ratio is equal to 0.09, while in some
others [24] this value may rise to 287. The ratio of
these two values is as high as 3190, which clearly
shows that a suitable value for the H2O2/Fe2+ ratio
should be defined by means of optimization studies.
By taking the added Fe2+ ions and the amount of ●OH
formed by H2O2 into account, the stoichiometric ratios
to be used should be defined. The stoichiometric
equation and ratio to be used for H2O2/Fe2+ is given
below:
2 3 •2 2Fe H O Fe OH OH (1)
When this equation is stoichiometrically inves-
tigated, it is seen that for 56 g of Fe2+, 34 g of H2O2 is
needed with a ratio of about 0.607. By also con-
sidering a 10% margin of safety for the addition of
Fe2+, the new ratio can be defined as 0.552. An addi-
ton of Fe2+ greater than the amount resulting in this
ratio will bring about extra costs and also undesired
chemical sludge at the end of the reaction. Therefore,
lower ratios should not be used. It should be kept in
mind that the proportional lower limit for Fe2+ concen-
tration corresponding to the upper limit of the ratio to
use is 0.55. In this study, as can be seen also in
Table 3, the performance of the reaction was eval-
uated by using the lower amounts of Fe2+, such as the
ratios equal to 1 and 2.
Another proportional expression that is repre-
sentative for a Fenton process is COD/Fe2+. As a mat-
ter of fact, these two parameters are not directly inter-
related. This value is specified based on the H2O2
concentration. When the stoichiometric ratios are con-
sidered, the proportional values suggested to obtain a
synergistic effect and less chemical use are as fol-
lows:
Stoichiometric ratio: COD/H2O2/Fe2+: 1/2.125/3.5
Economical ratio: COD/H2O2/Fe2+: 1/2.125–0.25/3.5–0.35
The calculated ratios were obtained by con-
sidering 10% more than the maximum stoichiometric
ratio. The minimum values, on the other hand, were
F. ILHAN et al.: EVALUATION OF OPERATIONAL PARAMETERS… Chem. Ind. Chem. Eng. Q. .23 (1) 1119 (2017)
15
Table 3. Experimental runs and operational parameters
Runs COD/H2O2 H2O2/Fe2+ Initial pH Dosing
1 0.428 0.55 3.00 -
2 0.5 0.55 3.00 -
3 0.6 0.55 3.00 -
4 0.8 0.55 3.00 -
5 1.0 0.55 3.00 -
6 1.2 0.55 3.00 -
7 1.4 0.55 3.00 -
8 1.6 0.55 3.00 -
9 1.8 0.55 3.00 -
10 2.0 0.55 3.00 -
11 0.5 0.55 3.00 -
12 0.5 0.75 3.00 -
13 0.5 1.0 3.00 -
14 0.5 1.2 3.00 -
15 0.5 1.4 3.00 -
16 0.5 1.6 3.00 -
17 0.5 1.7 3.00 -
18 0.5 1.8 3.00 -
19 0.5 1.9 3.00 -
20 0.5 2.0 3.00 -
21 0.5 1.0 2.00 -
22 0.5 1.0 2.25 -
23 0.5 1.0 2.50 -
24 0.5 1.0 2.75 -
25 0.5 1.0 3.25 -
26 0.5 1.0 3.50 -
27 0.5 1.0 3.75 -
28 0.5 1.0 4.00 -
29 0.5 1.0 3.00 1
30 0.5 1.0 3.00 2
31 0.5 1.0 3.00 3
32 0.5 1.0 3.00 4
33 0.5 1.0 3.00 5
34 0.5 1.0 3.00 10
35 0.5 1.0 3.00 15
36 0.5 1.0 3.00 30
37 0.5 1.0 3.00 45
38 0.5 1.0 3.00 60
39 0.5 1.0 3.00 60
40 0.5 1.0 3.00 60
41 0.5 1.0 3.00 60
42 0.5 1.0 3.00 60
43 0.5 1.0 3.00 60
defined as one tenth of the maximum value, specified
considering the synergistic effect. Optimization work
to be carried out between these values is thought to
help define more appropriate operational condition,
addressing higher efficiencies and lower chemical
consumptions. The operational conditions specified in
this study were optimized at each stage before pas-
sing to the next one, with the aim to increase the effi-
ciency by using optimum values in the subsequent
study-sets. The obtained results were discussed
under different sub-titles in this section.
Effect of COD/H2O2 ratio
The most important parameter in a Fenton pro-
cess is the hydrogen peroxide ratio. As the H2O2
amount increases, the oxidation capacity and effi-
ciency also increase. However, the dose to apply
should be suitable to the amount of organic contam-
inants in the water. Otherwise, less H2O2 than needed
will reduce the efficiency level, while excessive H2O2
amounts will increase operational costs. Therefore,
the optimization of H2O2 amount is of utmost impor-
tance in the Fenton oxidation process. The color and
TOC reduction efficiencies obtained for different
doses specified on the basis of stoichiometric ratios
are given in Figure 1. The studies were carried out
with a pH value equal to 3.0, as an average of the
interval 2-4. The H2O2/Fe2+ ratio was taken as 0.55,
which is the most suitable value from a stoichiometric
perspective.
When Figure 1 is closely examined, it is seen
that increasing COD/H2O2 ratios (or decreasing per-
oxide amounts) adversely affect efficiency, since
lower peroxide amounts decrease the formation rate
of the ●OH radicals, and consequently lower the oxid-
ation capacity of the process. Therefore, the highest
efficiency level was obtained when COD/H2O2 ratio
was equal to 0.5. When the COD/H2O2 ratio was
taken as 0.5, the resulting TOC and color reduction
values were 80.2 and 96.1%, respectively. Based on
these results, 0.5 was accepted to be the optimum
value for the COD/H2O2 ratio and used for the opti-
mization of H2O2/Fe2+ ratio in the next step.
Effect of H2O2/Fe2+ ratio
In redox reactions, the Fe2+ ratio that helps form ●OH is also as important as the H2O2 in the Fenton
oxidation process. Similarly to the hydrogen peroxide
dosing, the use of higher than necessary Fe2+ doses
will result in additional costs. Using an insufficient
amount of Fe2+, on the other hand, will result in ineffi-
ciently completed redox reactions, which are needed
for the formation of ●OH. Therefore, the Fe2+ doses to
add must be optimized for better results. To this end,
a series of studies were carried out using H2O2/Fe2+
doses, determined in compliance with stoichiometric
ratios. The obtained results are given in Figure 2. In
these studies, the pH was 3.0, which is the average
value for the 2-4 interval. The COD/H2O2 ratio, on the
F. ILHAN et al.: EVALUATION OF OPERATIONAL PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 1119 (2017)
16
other hand, was 0.5, which was found to be the opti-
mum value.
When the obtained results are investigated, it is
clearly seen that increasing H2O2/Fe2+ ratios (or
decreasing Fe2+ amounts) slightly decrease the color
and TOC reduction levels. Increasing the H2O2/Fe2+
ratios indicates insufficient Fe2+ additions. Conse-
quently, a lower efficiency level can be observed due
to the inhibition of the formation of ●OH. The optimum
value for the H2O2/Fe2+ ratio was found to be 1.0. The
color and TOC reduction values determined by using
these values were 80.8 and 96.2%, respectively.
Effect of pH on TOC and color removal
Like in all chemical reactions, for a Fenton oxid-
ation process to be able to be carried out efficiently,
there is a suitable pH interval. For a Fenton reaction,
it is known that this interval corresponds to the acidic
range (generally between 2.0 and 4.0) [33,35]. An
optimization study on pH level was carried out to
further elaborate on this value between 2.0 and 4.0.
The results obtained from this study are shown in
Figure 3. This optimization was carried out by taking
H2O2/Fe2+ and COD/H2O2 ratios as 1.0 and 0.5,
respectively, which were found to be the optimum
values in previous studies.
When the obtained results are investigated, it is
seen that the highest efficiency levels are obtained
with a pH value equal to 3.0, because maintaining the
pH value constant at 3.0 (also at the previous opti-
mization studies) increased the reliability of the res-
ults obtained from these optimization studies. Under
these circumstances, the color and TOC reduction
efficiencies were found to be 80.8 and 96.2%, res-
pectively. The results indicated that both TOC and
color removal levels were lower, compared to results
obtained when the pH value was kept equal to 3.0.
This could be due to the decrease in the synergistic
effect of H2O2 and Fe2+ [26].
Effect of H2O2 dosing on TOC and color removal
●OH plays an active role in the reaction mech-
anism of the Fenton oxidation process as oxidant. For
the formation of these radicals, the addition of H2O2 –
Figure 1. Effects of COD/H2O2 ratio on color and TOC removal efficiencies (H2O2/Fe2+ = 0.55, pH 3.0, reaction time = 30 min).
Figure 2. Effects of H2O2/Fe2+ ratio on color and TOC removal efficiencies (COD/H2O2 =0.5, pH 3.0, reaction time = 30 min).
F. ILHAN et al.: EVALUATION OF OPERATIONAL PARAMETERS… Chem. Ind. Chem. Eng. Q. .23 (1) 1119 (2017)
17
in one go, or in different time periods during the react-
ion – is of great importance. For this aim, a series of
studies have been conducted regarding different dos-
ing types in photo-Fenton and solar photo-Fenton
procedures [36-37]. In classical Fenton studies, on
the other hand, the addition of single-step H2O2 is
generally preferred [38-40]. At this stage of the study,
the effects of different H2O2 dosing on color and TOC
reduction were attempted to be investigated. To this
end, the single-step method was not preferred and
the dosing of H2O2 addition was carried out in multiple
steps (2 times (per 15 min), 3 times (per 10 min), 4
times (per 7.5 min), 5 times (per 6 min), 6 times (per 5
min), 10 times (per 3 min), 15 times (per 2 min), 30
times (per 1 min), and 45 times (per 40 s)] in 30 s and
in continuous mode.
The obtained results clearly showed that time-
-dependent dosing of H2O2 might help to obtain higher
efficiency levels. It was seen that by carrying out
dosing at an extended period of time, TOC reduction
efficiency rose from 80.8 up to 88.9% for the same
amount of dosing. Similarly, the color reduction effi-
ciency also increased from 96.5 to 98.7% by time-
-dependent dosing of H2O2. The higher occurrence
rate of the reactions by dosing is also reflected in the
efficiency levels. An instantaneous dosing will play
the role of the oxidation capacity of peroxide instead
of turning all the peroxide into ●OH at one time.
Therefore, it is apparent that the dosing of hydrogen
peroxide for an extended period of time will consi-
derably improve the reduction efficiency levels.
CONCLUSIONS
This study aimed to provide the synchronization
between studies carried out on the operation con-
ditions intrinsic to the Fenton process that is effect-
ively used for strong organic compounds. When the
stoichiometric ratios are not considered, the exces-
sive use of chemicals results in not only additional
costs, but also in the formation of undesired amounts
of sludge. Similarly, the use of insufficient amounts of
reactant results in low reduction efficiencies. There-
fore, an evaluation of the most recent studies on the
Fenton process in the relevant literature was made,
and the need for such a study was demonstrated.
From a dosing point of view, it was shown that an
apparent increase in efficiency levels can be obtained
in the Fenton processes. At the same time, synchro-
nization in operational conditions can be provided and
the efficiency of different studies carried out in this
field can be compared to each other more easily. An
equivalent reduction in efficiency levels can be
obtained with lower chemical consumptions when the
stoichiometric ratios are taken into consideration.
REFERENCES
[1] D. Squire, J. Murrer, P. Holden, C. Fitzpatrick, Desalin-
ation 108 (1997) 143-147
[2] P. Kampas, S.A. Parsons, P. Pearce, S. Ledoux, P. Vale,
J. Churchley, E. Cartmell, Water Res. 41 (2007) 1734-
–1742
[3] F. Ilhan, U. Kurt, O. Apaydin, M.T. Gonullu, J. Hazard.
Mater. 154 (2008) 381-389
[4] E. Neyens, J. Baeyens, J. Hazard. Mater. 98 (2003) 33-
–50
[5] S. Karthikeyan, A. Titus, A. Gnanamani, A.B. Mandal, G.
Sekaran, Desalination 281 (2011) 438-445
[6] D. Hermosilla, M. Cortijo, C.P. Huang, Sci. Total Environ.
407 (2009) 3473-3481
[7] U. Kurt, Y. Avsar, M.T. Gonullu, Chemosphere 64 (2006)
1536-1540
Figure 3. Effects of pH on color and TOC removal efficiencies (COD/H2O2 = 0.5, H2O2/Fe2+ = 1.0, reaction time = 30 min).
F. ILHAN et al.: EVALUATION OF OPERATIONAL PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 1119 (2017)
18
[8] M.S. Lucas, J.A. Peres, J. Hazard. Mater. 168 (2009)
1253-1259
[9] N.S.S. Martinez, J.F. Fernandes, X.F. Segura, A.S.
Ferrer, J. Hazard. Mater., B 101 (2003) 315-322
[10] U. Kurt, O. Apaydın, M.T. Gonullu, J. Hazard. Mater. 143
(2007) 33-40
[11] S.H. Lin, C.M. Lin, H.G. Leu, Water Res. 33 (1999) 1735-
–1741
[12] K. Yetilmezsoy, Environ. Sci. Pollut. Res. 19 (2012) 2227-
–2237
[13] P. Canizares, R. Paz, C. Saez, M.A. Rodrigo, J. Environ.
Manag. 90 (2009) 410-422
[14] J.H. Park, I.H. Cho, S.W. Chang, J. Environ. Sci. Heal., B
41 (2006) 109-120
[15] M.J. Abdul, M. Kumar, S. Vigneswaran, J. Kandasamy, J.
Ind. Eng. Chem. 18 (2012) 137-144
[16] R. Li, C. Yang, H. Chen, G. Zeng, G. Yu, J. Guo, J.
Hazard. Mater. 167 (2009) 1028-1032
[17] D. Guclu, N. Sirin, S. Sahinkaya, M.F. Sevimli, Environ.
Prog. Sustain. 32 (2013) 176-180
[18] M.I. Badawy, R.A. Wahaab, A.S. El-Kalliny, J. Hazard.
Mater. 167 (2009) 567-574
[19] C.C. Su, M.P. Asa, C. Ratanatamskul, M.C. Lu, Sep.
Purif. Technol. 83 (2011) 100-105
[20] X. Zhu, J. Tian, R. Liu, L. Chen, Sep. Purif. Technol. 81
(2011) 444-450
[21] S.F. Kang, C.H. Liao, M.C. Chen, Chemosphere 46
(2002) 923-928
[22] K.V. Padoley, S.N. Mudliar, S.K. Banerjee, S.C. Desh-
mukh, R.A. Pandey, Chem. Eng. J. 166 (2011) 1-9
[23] X. Wang, G. Zeng, J. Zu, J. Hazard. Mater. 153 (2008)
810-816
[24] I. Gulkaya, G.A. Surucu, F.B. Dilek, J. Hazard. Mater. 136
(2006) 763-769
[25] L. Chu, J. Wang, J. Dong, H. Liu, X. Sun, Chemosphere
86 (2012) 409-414
[26] K. Yetilmezsoy, S. Sakar, J. Hazard. Mater. 151 (2008)
547-558
[27] T. Mandal, D. Dasgupta, S. Mandal, S. Datta, J. Hazard.
Mater. 180 (2010) 204-211
[28] M. Ahmadi, F. Vahabzadeh, B. Bonakdarpour, E. Mofar-
rah, M. Mehranian, J. Hazard. Mater. 123 (2005) 187-195
[29] L.M. Nieto, G. Hodaifa, S. Rodriquez, J.A. Gimenez, J.
Ochando, Chem. Eng. J. 173 (2011) 503-510
[30] J. Li, Z. Luan, L. Yu, Z. Ji, Desalination 284 (2012) 62-65
[31] B. Bianco, I.D. Michelis, F. Veglio, J. Hazard. Mater. 186
(2011) 1733-1738
[32] R.F. Yu, H.W. Chen, K.Y. Liu, W.P. Cheng, P.H. Hsieh, J.
Chem. Technol. Biotech. 85 (2010) 267-278
[33] J. Blanco, F. Torrades, M.D. Varga, J.G. Montano, Desal-
ination 286 (2012) 394-399
[34] APHA (American Public Health Association), Standard
Methods for the Examination of Water & Wastewater, 21th
ed., Washington DC, 2005
[35] E.L. Kochany, J. Kochany, Chemosphere 73 (2008) 745-
–750
[36] D.P. Garcia, G. Buitron, J. Hazard. Mater. 217-218 (2012)
293-300
[37] E.O. Gomez, J.C.M. Ubeda, J.D.A. Hervas, J.L.C. Lopez,
L.S.J. Jorda, J.A.S. Perez, J. Hazard. Mater. 237-238
(2012) 223-230
[38] H. Zhang, H.J. Choi, C.P. Huang, J. Hazard. Mater., B
125 (2005) 166-174
[39] D.L. Wu, W. Wang, Q.W. Guo, Y.H. Shen, Chem. Eng. J.
214 (2013) 278-284
[40] S. Dogruel, T.O. Hanci, Z. Kartal, I.A. Alaton, D. Orhon,
Water Res. 43 (2009) 3974-3983.
F. ILHAN et al.: EVALUATION OF OPERATIONAL PARAMETERS… Chem. Ind. Chem. Eng. Q. .23 (1) 1119 (2017)
19
FATIH ILHAN
KAAN YETILMEZSOY
HARUN AKIF KABUK
KUBRA ULUCAN
TAMER COSKUN
BUSRA AKOGLU
Department of Environmental
Engineering, Faculty of Civil
Engineering, Yildiz Technical
University, Davutpasa, Esenler,
Istanbul, Turkey
NAUČNI RAD
PROCENA RADNIH PARAMETARA I NJIHOV ODNOS SA STEHIOMETRIJOM FENTON OKSIDACIJE OTPADNE VODE TEKSTILNE INDUSTRIJE
Radni uslovi Fentonovog procesa su od najveće važnosti jer postoje problemi vezani za
odnos doziranja H2O2 sa HPK i doziranja Fe2+ sa specifičnom količinom H2O2. Relevantna
literatura pokazuje da su vrednosti odnosa HPK/H2O2 u opsegu između 0,0084 i 113,9.
Slično, odnos HPK/Fe2+ je u opsegu između 0,079 i 292,6 dok odnos H2O2/Fe2+ varira iz-
među 0,09 i 287. Pored toga, odnos maksimalne i minimalne vrednosti korišćenih na bazi
HPK ima vrednost 13560 za odnos HPK/H2O2 (u opsegu od 0,0084 do 113,9), 2210 za
odnos HPK/Fe2+ (u opsegu od 0,079 do 174,7) i konačno 3190 za odnos H2O2/Fe2+ (u
opsegu od 0,09 do 287). Cilj ovog rada je da ponovo proceni ove vrednosti koje se zna-
čajno razlikuju jedna od druge sa posebnim naglaskom na otpadne vode tekstilne indus-
trije uz razmatranje stehiometrijskih odnosa. Ovi rezultati pokazuju da su vrednosti u op-
segu od 0,43 do 4,0 za HPK/H2O2 dok su vrednosti u opsegu od 0,75 do 3,0 mnogo
pogodnije za H2O2/Fe2+. Takođe, ovi rezultati pokazuju da doziranje H2O2 u različitim vre-
menima Fentonovog procesa može da poveća efikasnost smanjenja ukupnog organskog
ugljenika od 80,8 do 88,9%. Slično, efikasnost smanjenja obojenosti se povećava sa 96,5
na 98,7%.
Ključne reči: Fentonova oksidacija, radni parametri, oksidacija, stehiometrijski
odnos, uklanjanje ukupnog organskog ugljenika.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 2129 (2017) CI&CEQ
21
JAVAD AHMADISHOAR1
S. HAJIR BAHRAMI1
BARAHMAN MOVASSAGH2
SEYED HOSEIN AMIRSHAHI1
MOKHTAR ARAMI1
1Textile Engineering Department,
Amirkabir University of
Technology, Tehran, Iran 2Chemistry Department, K.N. Toosi
University of Technology, Tehran,
Iran
SCIENTIFIC PAPER
UDC 631.82/.85:549.67:553:66
https://doi.org/10.2298/CICEQ150116049A
REMOVAL OF DISPERSE BLUE 56 AND DISPERSE RED 135 DYES FROM AQUEOUS DISPERSIONS BY MODIFIED MONTMORILLONITE NANOCLAY
Article Highlights
• A modified montmorillonite was used for adsorption of two disperse dyes
• Modified nanoclay could effectively adsorb the disperse dyes from their aqueous dis-
persion
• The molecular weight and structure of dyes affected the dye adsorption from their
aqueous dispersion
• The major interactions between the nonionic dyes and organoclay are of hydrophobic
nature
• Modified montmorillonite can be used for dye removal from textile wastewater indus-
tries
Abstract
In this study modified montmorillonite was used as an adsorbent for the removal
of two selected disperse dyes i.e., Disperse Blue 56 (DB) and Disperse Red 135
(DR) from dye dispersions. The adsorption equilibrium data of dyes adsorption
were investigated by using Nernst, Freundlich and Langmuir isotherm models.
The adsorption kinetics was analyzed by using different models including
pseudo-first-order, pseudo-second-order, Elovich and Intraparticle diffusion
model. The Freundlich isotherm was found to be the most appropriate model for
describing the sorption of the dyes on modified nanoclay. The best fit to the
experimental results was obtained by using the pseudo-second-order kinetic
equation, which satisfactorily described the process of dye adsorption. Although
different kinetic models may control the rate of the adsorption process, the
results indicated that the main rate limiting step was the intraparticle diffusion.
The results showed that the proposed modified montmorillonite could be used as
an effective adsorbent for the removal of disperse dyes even from highly
concentrated dispersions.
Keywords: isotherm, kinetic, modified nanoclay, disperse dye, waste water.
The use and application of synthetic dyes in
number of industries such as textile, plastic, paper,
leather and cosmetic has gradually increased in
recent years. The discharge of large amount of col-
ored effluents, produced from these industries, into
water sources leads to water pollution and causes
major environmental problems. The removal of dyes
Correspondence: H. Bahrami, Textile Engineering Department,
Amirkabir University of Technology, Tehran, Iran. E-mail: [email protected] Paper received: 16 January, 2015 Paper revised: 19 November, 2015 Paper accepted: 28 December, 2015
from wastewater is one of the critical issues for the
dye manufactures and consumer industries.
The aromatic rings in the structure of most syn-
thetic dyes make them biologically non-degradable
[1]. A variety of methods could be suggested to eli-
minate dyes from wastewater. Some methods, such
as ozonation and oxidation, decompose [2,3] the dye
molecules to simpler compounds, which are less envi-
ronmentally hazardous. Techniques such as coagul-
ation and adsorption extract the dye molecules from
the solutions, i.e., water.
In general, the most important characteristics of
the adsorption based techniques are their lower cost,
J. AHMADISHOAR et al.: REMOVAL OF DISPERSE BLUE 56… Chem. Ind. Chem. Eng. Q. 23 (1) 2129 (2017)
22
better availability, high efficiency, and the possibility
to treat dyes in higher concentration forms in com-
parison to the other methods [4]. Various materials
have been used as adsorbents for the dye removal,
i.e., bagasse [5] and TiO2 [6]. As described in the lit-
erature, the activated carbon is the most widely used
among all conventional adsorbents. However eco-
nomical issues make it an expensive material for
water treatment applications [7].
Nanoclays are low-cost adsorbents, which have
been widely employed in dye user industries to ads-
orb dyes from wastewater [8]. Sodium montmorillonite
belongs to the smectite group and is composed of 2:1
type layered silicate with two tetrahedral silicates and
one octahedral layer with a considerable cation
exchange capacity in the octahedral layer. The
sodium montmorillonite is the more common form of
the five natural varieties of smectite [9-12]. The main
characteristic of sodium montmorillonite is its high
cation exchange capacity. Due to substitution of cat-
ions in the layers, there is a negative charge in the
octahedral layer which can be compensated by some
cations. Sodium cations in sodium montmorillonite
can be replaced by some chemical components. This
could lead to surface modifications of the clays [12].
These modifications not only change the surface
characteristics of the clays from hydrophobic nature
to hydrophilic one, but also increase the adsorption
capacity of the clays [13].
The literature survey indicated that the use of
such modifications of the clay surface, led to improve-
ment in the adsorption capacity of the different dye
classes [14]. Different chemical components have
been used for surface modification of nanoclays.
Grafting of functional polymers to the surface of the
clays [15], using ionic liquids such as imidozolium,
pyridinium [16,17] and phosphonium derivatives [18],
1,6-diamino hexane [19] and cationic surfactant [20]
are some of the reported modification types on nano-
clays. The quaternary ammonium compounds are
also typical materials used for nanoclay modification.
Modified clays have been used for removal of differ-
ent classes of dyes from wastewater. Zohra et al. [7]
used modified bentonite for removal of direct dyes.
Modified bentonite was used to remove reactive dyes
from waste water [21]. In another report acid dyes
were removed from wastewater using modified bento-
nite [22]. There is another report on removal of cation
dyes from wastewater using modified montmorillonite
and sepiolite [23,24]. To the best of our knowledge,
there have been relatively few works investigating the
removal of nonionic dyestuffs, i.e., disperse dyes,
from wastewater by using nanoclays.
In this study, a modified montmorillonite with an
aromatic quaternary ammonium modification was
used for the removal of aqueous dispersions of two
disperse dyes with different structures and molecular
weights. Some physicochemical aspects of the rem-
oval reaction such as adsorption isotherm, equilibrium
conditions and adsorption kinetics were investigated.
MATERIALS AND METHODS
A modified montmorillonite nanoclay (Nanofil
3010, QA-MMN) was purchased from Süd-Chemie
Company. Sodium montmorillonite (Cloisite-Na+,
MMN) with a cation exchange capacity of 92.6
cmol/kg, was obtained from Southern Clay Products,
Inc. Two commercial grade disperse dyes, i.e., Ser-
ilen Blue Rl (Disperse Blue 56 from Yourkshire Chem-
ical) and Sumikaron Red S-GG (Disperse Red 135
from Sumitomo Chemical) were selected as disperse
dyes and denoted as DB and DR, respectively. The
chemical structures of the employed dyes are illus-
trated in Figure 1. The other chemicals, including
sodium hydroxide, hydrochloride acid and ethanol,
were of analytical grade, supplied by Merck and used
without any purification.
NH2
NH2
O
O
OH
OH
Cl
(a)
N
NN
HN
N
O
O
O
O
O
O
O
(b)
Figure 1. Chemical structure of dye: a) Disperse Blue 56,
b) Disperse Red 135.
Adsorption experiment
All the adsorption experiments were carried out
at 24 C with a constant agitation speed of 500 rpm.
Dye dispersions with specified concentrations were
prepared. Then the nanoclays (MMN and QA-MMN)
were added to the separate dispersions and stirred by
a mechanical stirrer. After the adsorption processes
J. AHMADISHOAR et al.: REMOVAL OF DISPERSE BLUE 56… Chem. Ind. Chem. Eng. Q. 23 (1) 2129 (2017)
23
the nanoclay was centrifuged for 5 min at 5000 rpm.
The concentration of the residual dye in the bath was
measured using a Carry 100 UV-Vis spectrophoto-
meter. The decrease in the dye concentration in the
dye bath was accepted to be the amount of dye ads-
orbed by the nanoclays. In order to ensure the com-
plete dissolution of the dye dispersion, a 10 volumet-
ric parts of the dispersion was mixed with 90 volu-
metric parts of pure ethanol. The adsorption capacity
(qe) was calculated as follows:
e 0 e( ) /q c c V W (1)
where c0 (g/L) is the initial concentration of the dye, ce
(g/L) is the concentration of the dye at equilibrium, V
(L) is the volume of the dye dispersion, and W (g) is
the mass of the adsorbent. The quantity of the ads-
orbed dye at time t, qt (g/g), was calculated using the
following equation:
0( ) /t tq c c V W (2)
where ct (g/L) is the concentration of the dye at any
time t.
Adsorption Isotherm
The dye adsorption isotherm was investigated
by treating 200 ml dye dispersions having concen-
trations in the range of 0.1-1 g/L at pH 5 with 1 g of
the modified nanoclay. In order to analyze the iso-
therm data, three models, i.e., Nernst, Langmuir and
Freundlich, were employed.
Adsorption kinetic models
Although several adsorption kinetic models were
suggested for such an adsorption process, they
should be confirmed by the mechanism of the dye
adsorption on a sorbent [4]. The adsorption experi-
ments were carried out by treating 100 ml dye dis-
persions with concentration of 1 g/L at pH 5 with 1 g
of the nanoclay for different times. In order to inves-
tigate the dye adsorption kinetics, three adsorption
models were used - intra-particle diffusion, pseudo-
-first-order and pseudo-second-order.
RESULTS AND DISCUSSION
The effect of pH on the adsorption process
The initial pH value of the dye dispersion is an
important parameter, which controls the adsorption
process and affects the surface charges of nanoclay.
The adsorption study for disperse dyes on modified
nanoclay was carried out using 100 ml dye dispersion
with concentration of 0.2 g/L and 0.2 g adsorbent at
24 C and pH in the range of 3–11 for 2 h.
The results showed that the adsorption of the
two disperse dyes on clay depended on the initial pH
of the dye dispersions. It could be seen from Fig. 2
that the amount of dye adsorbed on the nanoclay
increased as the pH of DR and DB decreased from 7
to 3 and from 9 to 3, respectively. At higher values of
pH, a small increase in the adsorbed amount of dyes
was observed. Non-ionic dyes are negatively charged
in solution [25,26] and the modified nanoclay has
positively charged surface [7] especially at lower pH.
The electrostatic interactions between the negatively
charged dye molecules and the positively charged
clay surface resulted in a higher dye removal at lower
pH. At higher pH, the abundance of OH- competing
with the negatively charged dye molecules for the
adsorption sites led to lower adsorption of both dis-
perse dyes on the modified nanoclay. Similar obser-
vations were reported in other studies [19,25,27].
Figure 2. Effect of initial pH on the adsorption of disperse dyes
on modified nanoclays (QA-MMN).
Rate of adsorption
In order to study the adsorption rate and to
ensure complete adsorption of the dyes, the experi-
ments were carried out for more than 4 h. The amount
of the adsorbed dye on two types of clays (g/L) as a
function of the contact time (in minutes) is shown in
Fig. 3. The results showed that the rate of dye ads-
orption on nanoclay in the initial stage of the contact
period was significantly high and the maximum ads-
orption rate was registered in the initial 20 min of the
contact time for both dyes. After the initial stage of the
adsorption process, the rate of dye adsorption gradu-
ally decreased and reached a steady-state value
identified as the equilibrium loading capacity, qe.
Similar trend was reported in other studies [1,27-29].
The sharp slope in the first stage (higher initial
sorption rate) could be attributed to a large number of
vacant adsorption sites on the surface of the clay
available in the initial stage of the adsorption process
[7]. This led to an increase in the dye concentration
gradient between the dispersion and the surface of
J. AHMADISHOAR et al.: REMOVAL OF DISPERSE BLUE 56… Chem. Ind. Chem. Eng. Q. 23 (1) 2129 (2017)
24
the adsorbent [7]. As time passed, the surface of the
clay was gradually occupied by dye molecules, which
resulted in less number of sites on the clay surface
accessible to the dye molecules. This led to decrease
in the tendency of the dye towards the clay surface
[1,28]. The reduced adsorption rate showed a pos-
sible presence of a mono-layer of dye molecules on
the nanoclay surface [1,4,29,30], which was evident
for both disperse dyes.
Figure 3. Effect of contact time for adsorption of DB and DR on
MMN and QA-MMN
Figure 3 also showed differences in the dye
adsorption in the first and last stage of the adsorption
process for both disperse dyes. In the first stage of
the adsorption process (the first 20 min), the amount
of DB dye adsorbed on the nanoclay surface was
rather higher than the amount of DR dye. However
after this period, the adsorption behavior of the dyes
changed and the DB dye showed slower adsorption
rate in comparison to DR dye. Consequently at
steady state the amount of the adsorbed DR dye on
the clay surface is higher than the one of DB dye. It
seemed that, the differences could be related to the
molecular structures of the employed dyestuffs. In
fact, the smaller structure and lower molecular weight
of DB led to a higher mobility of the dye during the
first stage of the adsorption process in comparison to
DR and consequently to a higher adsorption rate. In
the next step it seemed that the major contributive
forces between the nonionic dye and organoclay were
the van der Waals forces and hydrophobic interact-
ions [7,14].
The negatively charged surface of the nanoclay
may adsorb the cationic surfactants in two steps.
Firstly via an ion exchange mechanism, a monolayer
of cationic surfactants on the surface of the clay was
formed. The positively charged ends of the cationic
surfactants were exchanged with the interlayer
exchangeable cations of the clay (Na+) and the hyd-
rophobic head of the cationic surfactants was arranged
outward (Fig. 4a). Secondly through hydrophobic-hyd-
rophobic interactions, cationic surfactant alkyl chains
were attached to the outer alkyl chain of the mono-
layer (Fig. 4b) [31]. For the Nanofil 3010 sample, a
diffraction peak at 2θ = 7.65 corresponding to a
basal spacing of 11.56 Å was observed. According to
the calculated basal spacing, it could be ensured that
the monolayer arrangement was formed [32].
In our study, according to the adsorption/par-
tition model, the organic fraction of the surfactant
modified clay, containing a long alkyl chain, behaved
as a partition medium and the partition occured
through interaction of the dyes with the cationic surf-
actant of the modified nanoclay. The cationic surf-
Figure 4. Schematic diagram of the formation of surfactant on the surface of the clay a) monolayer, b) bilayer formation and
c) the interactions of DR molecules with the surfactant modified adsorbent.
J. AHMADISHOAR et al.: REMOVAL OF DISPERSE BLUE 56… Chem. Ind. Chem. Eng. Q. 23 (1) 2129 (2017)
25
actants created hydrophobic regions [33]. The hydro-
phobic portion of the adsorbent surface had greater
affinity for dissociated species of dyes in aqueous
solution and the major contributive forces between
the nonionic dye and the organoclay were the Van der
Waals forces and hydrophobic interactions (Figure
4c). The partition of the dyes into hydrophobic regions
plays an important role in the dye uptake.
The dye with the larger molecular structure (DR)
could be adsorbed better than the DB dye on the
nanoclay surface due to the stronger van der Waals
and hydrophobic interactions (interaction between the
phenyl ring of the dyes and the CH2 group of the
modified adsorbent). A good example of a high sorp-
tion of the disperse dyes on the clay surface modified
with quaternary ammonium with aromatic rings was
the sorption of disperse dyes on Cloisite 10A, where
the Van der Waals forces were the main interaction
forces between the aromatic systems [14]. The affinity
of montmorillonite (MMN) toward these dyes from dis-
persion was lower in comparison to the modified
nanoclay (QA-MMN). From the results shown in Fig. 3
it could also be concluded that the times for achieving
the maximum adsorption for DB and DR were 60 and
120 min respectively, which were chosen for the next
experiments in this study.
Adsorption isotherm
The adsorption isotherm indicated the distribut-
ion of dye adsorbed on the nanoclay and dye in bath
in equilibrium state. The isotherm plots for both dyes
are shown in Fig. 5.
Figure 5 showed the experimental and predicted
data for three isotherms. The parameters of these
isotherms for the employed dyes are given in Table 1.
Based on the individual plots and the high regression
correlation coefficient R2, it could be concluded that
the Freundlich isotherm had better fits than the other
isotherms.
Adsorption kinetic models
The determination of the adsorption kinetics can
provide information about the rate of dye adsorption
on the adsorbent surface form dye dispersion as well
as the adsorption mechanism [4,30]. In order to study
the adsorption kinetics in this study, the experimental
data were analyzed using pseudo-first-order, pseudo-
-second-order, Elovich equation and intra-particle dif-
fusion models. The linear regression correlation coef-
ficient, R2, was used as a criterion to select the model,
which gives the best fit to the experimental data.
Figure. 5. Adsorption isotherm of two disperse dyes on
QA-MMN.
Pseudo-first-order model
The pseudo-first-order model is expressed by
Eq. (3):
Table 1. Calculated isotherms parameters
Dye Freundlich Nerst Langmuir
kf / L g-1 n R2 K / mL g–1 R2 qm / mg g–1 kb / L mg–1 R2
DR 135.64 0.872 0.99 126.4 0.979 0.2154 0.942 0.744
DB 128.38 0.9 0.994 127.5 0.991 0.3279 0.592 0.749
J. AHMADISHOAR et al.: REMOVAL OF DISPERSE BLUE 56… Chem. Ind. Chem. Eng. Q. 23 (1) 2129 (2017)
26
1 e
d( )
dt
t
qk q q
t (3)
where qt is the amount of the adsorbed dye at time t;
k1 is the rate constant of the pseudo-first-order model
(min-1); t is the time (min) and qe is the adsorption
capacity in equilibrium (mg/g). After definite integ-
ration by applying the conditions: qt = 0 at t = 0 and
qt = qt at t = t, Eq. (3) becomes [34]:
e e 1ln( ) lntq q q k t (4)
The linear relation between the eln( )tq q and t
confirmed the validity of the model while k1 and qe are
the slop and intercept of the proposed line, respect-
ively. The results listed in Table 2, showed that value
of the the correlation coefficient for DR is higher than
the one for DB and the values of qe obtained by Eq.
(4) for both dyes are not in agreement with the real
values (0.08056 and 0.0977 g/g for DB and DR, res-
pectively). The results showed that the experimental
data are not in agreement with the pseudo-first-order
kinetic model and this model cannot fully describe the
adsorption kinetics. In Fig. 6a and b are shown the
pseudo-first-order plots for DB and DR, respectively.
Pseudo-second-order model
The experimental data were also analyzed by
using pseudo-second-order model. The obtained
results showed that the pseudo-second-order model
was suitable for low initial concentration of the dye
solution [34]. Dogan et al. [1, 29] suggested that the
rate of this model depends on the amount of the
adsorbate on the surface of the adsorbent and the
capacity of the adsorbent at equilibrium. The model is
represented by Eq. (5):
22 e
d( )
dt
t
qk q q
t (5)
where k2 is the pseudo-second-order rate constant
(g/(mol min)). By integration at boundary conditions:
qt = 0 at t = 0 and qt = qt at t = t, Eq. (5) becomes [34]:
2
e2 e
1
t
t t
q qk q (6)
Again, the model was confirmed by linear rel-
ation between t/qt and t. As seen from Table 2, the
correlation coefficient values for DR and DB are
0.9995 and 0.9992, respectively and qe for both dyes
are in agreement with the real values (80.56 and
97.72 mg/g for DB and DR, respectively). The plots of
(a) (b)
(c) (d)
Figure 6. Plot of ln(qe - qt) vs. time (t),a: DB, b: DR and Plot of t/qt vs. t, c:DB, d: DR.
J. AHMADISHOAR et al.: REMOVAL OF DISPERSE BLUE 56… Chem. Ind. Chem. Eng. Q. 23 (1) 2129 (2017)
27
the pseudo-second-order model for DB and DR are
presented in Fig. 6c and d, respectively.
Clearly, the pseudo-second-order kinetic equa-
tion could satisfactory describe the dye adsorption
and the experimental results were best fitted by using
this model.
Elovich equation
The validity of this equation suggests the pre-
sence of reactions involving chemical adsorption of
adsorbate on the adsorbent [4,35]. Equation (8) rep-
resents the mathematical form of the Elovich equa-
tion:
d
exp( )d
tt
t (7)
where, is the initial adsorption rate (mol/(g min))
and is the desorption constant (g/mol). To simplify
the Elvoich equation, it was assumed that 1t . By
taking this into account and applying the boundary
conditions: qt = 0 at t = 0, the simple form of the
Elovich equation could be expressed as [1 ]:
( )tq Ln Ln t (8)
The straight line plot of qt vs. lnt confirmed the
validity of the Elovich equation as a suitable model
describing the kinetics of the adsorption process. The
correlation coefficients of the plots of qt vs. lnt for DB
and DR are 0.739 and 0.8476, respectively which
indicated that this model is not valid for this system.
Intraparticle diffusion model
Any adsorption process may consist of the fol-
lowing transportation steps: a) diffusion of the adsor-
bate in the surface of the adsorbent, b) intraparticle or
pore diffusion and c) sorption of adsorbate on the
adsorbent [1,4,36]. The intra particle diffusion model
would be valid, if the plot of qt vs. t0.5 showed a linear
relation and the intraparticle diffusion is the rate-limit-
ing step. According to Weber et al. [37], the equation
of intraparticle diffusion could be expressed by Eq.
(9):
1
2t pq K t C (9)
where qt is the amount of the adsorbed dye (mg/g) at
time t, Kp (mg/(g min1/2)) is the rate constant for int-
raparticle diffusion and C is the intercept. The intrapar-
ticle diffusion plots for both dyes are shown in Fig. 7.
The values of the intercept are proportional to
the thickness of the boundary layer and the larger
intercept corresponds to a greater boundary layer
effect [1,38,39]. As shown in Fig. 7 the adsorption
process occurred in two steps and the plots of qt vs.
t0.5 consist of two linear steps with different slopes for
each of the employed dyes. Similar trends were rep-
orted in other studies [1,4,13,35,40]. They suggested
that the sorption process occurred in two steps, i.e.,
surface sorption and intraparticle diffusion [4]. In the
first step, the dye molecules adsorbed on the external
surface of the clays and the rate of this step was
relatively fast. The second step occurred when the
external surface of the clays was saturated by the
dye. In this case, the dye molecules diffused in the
clays and were adsorbed on their internal surface.
Alkan et al. [1,4] suggested that the first step of the
diffusion process could be attributed to the macro-
-pore diffusion indicating boundary layer effect and
Table 2. Kinetics data calculated for adsorption of DB and DR on QA-MMN
Dye qe, exp
mg/g
Pseudo-first order Pseudo-second order Elovich equation
R2 K1102 / min–1 qe,cal / mg g–1 R2 K2 / g mg–1 min–1 qe,cal / mg g–1 α β R2
DB 80.56 0.5371 0.39 11.056 0.9995 0.00889 76.92 1.711011 2.352 0.739
DR 97.72 0.9956 5.9 81.451 0.9992 0.00135 111.11 0.819 13.74 0.847
Figure 7. Intra-particle diffusion plots for adsorption of DR and DB on QA-MMN.
J. AHMADISHOAR et al.: REMOVAL OF DISPERSE BLUE 56… Chem. Ind. Chem. Eng. Q. 23 (1) 2129 (2017)
28
the second step of the diffusion process was due to
the intraparticle or pore diffusion and was attributed to
micro-pore diffusion. The rate constants (Kp1 and Kp2)
of the intraparticle diffusion model for the employed
dyes are shown in Table 3.
The results listed in Table 3 showed that the
value of the first diffusion rate parameter (kp1) for DR
was different from the one for DB [3]. As discussed in
the previous paragraph, the first diffusion rate was
correlated to the surface sorption. The stronger
hydrophobic properties of the DR dye in comparison
to the one of DB and its affinity to the hydrophobic
nanoclay were the reasons for the higher adsorption
rate of the DR dye.
The slope of the second linear stage determined
the rate parameter corresponding to the intraparticle
diffusion. The value of this parameter for DR is higher
than the one for DB. The linear regression and the
comparison of the regression coefficients R12 and R2
2
for both dyes showed that the regression for both
dyes is linear but the plots do not pass through the
origin. This result is more obvious for DB. It sug-
gested that the adsorption involved intraparticle dif-
fusion and the diffusion could be accepted as a rate-
-limiting step. However, the diffusion is not the only
rate-controlling step [1,7,25,26,39] and other kinetic
models may also control the rate of adsorption.
CONCLUSIONS
In this study, the adsorption behavior of two
disperse dyes on modified montmorillonite was inves-
tigated. Modified montmorillonite could effectively
adsorb the disperse dyes from their aqueous dis-
persion and could be used for dye removal from tex-
tile wastewater industries. Changing the pH of the dye
dispersion could affect the dye removal from dye dis-
persion. An increase of the pH to values of above 7
led to increase in the adsorption of both disperse dye
on modified nanoclays. The adsorption rate study
showed that the adsorption of dyes on modified clays
involved two stages and the first stage of the ads-
orption process was faster than the second stage for
both dyes. The selected dyes had different molecular
weight and structure. This affected the dye adsorption
process on the absorbent surface. The isotherm
model, which showed the best fit to the experimental
data for both dyes was the Freundlich linear model
(R2 > 0.99). The kinetics studies showed that the
adsorption of the two disperse dyes could be well
defined by pseudo-second-order kinetic equation with
a high correlation coefficient (R2 > 0.99). The results
showed that the adsorption process involved intra-
particle diffusion and the diffusion could be a rate-
limiting step. However, the diffusion was not the only
rate-controlling step and other kinetic models could
also control the rate of adsorption.
REFERENCES
[1] M. Dogan, Y. Ozdemir, M. Alkan, Dyes Pigm. 75 (2007)
701-713
[2] H. Selcuk, Dyes Pigm. 64 (2005) 217-222
[3] A.E. Papadopoulos, D. Fatta, M. Loizidou, J. Hazard.
Mater. 146 (2007) 558-563
[4] M. Alkan, M. Dogan, Y. Turhan, O. Demirbas, P. Turan,
Chem. Eng. J. 139 (2008) 213-223
[5] J. X. Yu, J. Zhu, L.Y. Feng, R.A. Chi, J. Colloid Interface
Sci. 451 (2015) 153-160
[6] J. Feng, J. Zhu, W. Lv, J. Li, W. Yan, Chem. Eng. J. 269
(2015) 316-322
[7] B. Zohra, Kh.Aicha, S. Fatima, B. Nourredine, D. Zoubir,
Chem. Eng. J. 136 (2008) 295-305
[8] T.S. Anirudhan, M. Ramachandran, Process Saf. Environ.
Prot. 95 (2015) 215-225
[9] H.H. Murray, Applied Clay Mineralogy, Elsevier, Amster-
dam, 2007, pp. 5-32
[10] M.J. Wilson, Clay Minerology, Chapman & Hall, London,
1995, pp. 300-326
[11] M.J. Wilson, A Hand Book Of Determinative Methods in
Clay Mineralogy, Chapman & Hall, London, 1987, pp. 1-23
[12] A.C.D. Newman, Chemistry of Clays and Clay Minerals,
Wiley, New York, 1987, pp. 12-116
[13] A. Khenifi, B. Zohra, B. Kahina, H. Houari, D. Zoudir,
Chem. Eng. J. 146 (2009) 345-354
[14] Y. Yang, Sh. Han, Text. Res. J. 75 (2005) 622-627
[15] P. Liu, Appl. Clay Sci. 38 (2007) 64-76
[16] N.H. Kim, S.V. Malhotra, M. Xanthos, Microporous Meso-
porous Mater. 96 (2006) 29-35
[17] S. Livi, J. Duchet-Rumeau, J. Gerard, J. Colloid Interface
Sci. 353 (2011) 225-230
[18] G. Huang, J. Gao, X. Wang, Appl. Surf. Sci. 258 (2012)
4054-4062
Table 3. Intraparticle diffusion model parameter
C1 / mg g–1 R22 Kp2 / mg g–1 min–½ R1
2 Kp1 / mg g–1 min–½ Dye 59.7 0.0165 0 0.877 1.62 DB
28.55 0.711 0.55 0.989 9.49 DR
J. AHMADISHOAR et al.: REMOVAL OF DISPERSE BLUE 56… Chem. Ind. Chem. Eng. Q. 23 (1) 2129 (2017)
29
[19] Ö. Gök, A. S. Özcan, A. Özcan, Appl. Surf. Sci. 256 (2010)
5439-5443
[20] M. Kozak, L. Domka, J. Phys. Chem. Solids 65 (2004) 441-
–445
[21] A. Tabak, N. Baltas, B. Afshin, M. Emirik, B. Caglar, E.
Eren, J. Chem. Technol. Biotechnol. 85 (2010) 1199-1207
[22] P. Baskaralingam, M. Pulikesi, D. Elango, V. Ramamurthi,
S. Sivanesan, J. Hazard. Mater., B 128 (2006) 138-144
[23] G. Rytwo, S. Nir, M. Grespin, L. Margulies, J. Colloid
Interface Sci. 222 (2000) 12-19
[24] C.A.P. Almeida, N.A. Debacher, A.J. Downs, L. Cottet,
C.A.D. Mello, J. Colloid Interface Sci. 332 (2009) 46-53
[25] Q. Li, Q. Y. Yue, H. J. Sun, Y. Su, B. Y. Gao, J. Environ.
Manage. 91 (2010)1601-1611
[26] Q.Y. Yue, Q. Li, B.Y. Gao, Y. Wang, Sep. Purif. Technol.
54 (2007) 279-290
[27] M. Dogan, M. Alkan, Chemosphere 50 (2003)517-528
[28] I. Mall, V. Sirvastava, N. Agarwal, Dyes Pigm. 69 (2006)
210-223
[29] M. Dogan, M. H. Karaoglu, M. Alkan, J. Hazard. Mater. 165
(2009) 1142-1151
[30] D. Shen, J. Fan, W. Zhou, B. Gao, Q. Yue, Q. Kang, J.
Hazard. Mater. 172 (2009) 99-107
[31] M. Foroughi, H. Abolghasemi, M. Esmaielia, Gh. Nazari, B.
Rasem, Process Saf. Environ. Prot. 95 (2015) 226-236
[32] Z. Jianxi, H. Hongping, G. Jiugao, Y. Dan, X. Xiande, Chin.
Sci. Bull. 48 (2003) 368-372
[33] T.S. Anirudhan, M. Ramachandran, Process Saf. Environ.
Prot. 95 (2015) 215-225
[34] S. Azizian, J. Colloid Interface Sci. 276 (2004) 47-52
[35] M. Dogan, M. Alkan, O. Demirbas, Y. Ozdemir, C. Ozme-
tin, Chem. Eng. J. 124 (2006) 89-101
[36] J. Ahmadishoar, S.H. Amirshahi, R.M. Malek, J. Appl.
Polym. Sci. 112 (2009) 1030-1036
[37] W.J. Weber, J.C. Morris, J. Sanit. Eng. Div. 89 (1963) 31-
–60
[38] N. Kannan, M. M. Sundaram, Dyes Pigm. 51 (2001) 25-40
[39] B.H. Hameed, Colloids Surfaces, A 307 (2007) 45-52
[40] E. Eren, O. Cubuk, H. Ciftci, B. Eren, B. Caglar, Desalin-
ation 252 (2010) 88-96.
JAVAD AHMADISHOAR1
S. HAJIR BAHRAMI1
BARAHMAN MOVASSAGH2
SEYED HOSEIN AMIRSHAHI1
MOKHTAR ARAMI1
1Textile Engineering Department,
Amirkabir University of Technology,
Tehran, Iran 2Chemistry Department, K.N. Toosi
University of Technology, Tehran, Iran
NAUČNI RAD
UKLANJANJE DISPERZNIH BOJA PLAVA 56 I CRVENA 135 IZ VODENIH DISPERZIJA POMOĆU MODIFIKOVANE MONTMORILONITNE NANOGLINE
U ovom radu korišcen je modifikovani montmorilonit kao adsorbent za uklanjanje dve
izabrane disperzne boje disperzno plavo 56 (DB) i disperzno crveno 135 (DR) iz disperzija
boja. Podaci za adsorpcionu ravnotežu su analizirani korišćenjem Nernstovog, Frojndliho-
vog i Lengmirovog izotermskog modela. Kinetika adsorpcije je analizirana korišcenjem
modela pseudo-prvog i pseudo-drugog reda, Elovičevog i modela unutrašnje difuzije.
Frojndlihov izotermski model se pokazao kao nabolji za sorpciju boje na modifikovanoj
nano glini. Jednačina kinetike pseudo-drugog reda na zadovoljavajuci način opisuje brzinu
adsorpciju boje i najbolje fituje dobijene eksperimentalne podatke. Na osnovu dobijenih
rezultata može se zaključiti da se brzina adsorpcija može opisati modelom unutrašnje difu-
zije. Ova difuzija se može smatrati kao stupanj koji limitira brzine, ali ne i kao jedini, s
obzirom na to da i ostali modeli mogu kontrolisati brzinu adsorpcije. Dobijeni rezultati uka-
zuju da predloženi modifikovani montmorilonit može biti efikasan adsorbent za uklanjanje
disperznih boja čak i u slučaju jako koncentrovanim disperzijama.
Ključne reči: izoterma, kinetika, modifikovana nano glina, disperzna boja, otpadna
voda.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 3137 (2017) CI&CEQ
31
ALIREZA
EBRAHIMINEZHAD1–3
YAHYA BARZEGAR3,4
YOUNES GHASEMI3
AYDIN BERENJIAN5
1Noncommunicable Diseases
Research Centre, Fasa University
of Medical Sciences, Fasa, Iran 2Department of Medical
Biotechnology, School of Medicine,
Fasa University of Medical
Sciences, Fasa, Iran 3Department of Pharmaceutical
Biotechnology, School of
Pharmacy and Pharmaceutical
Sciences Research Centre, Shiraz
University of Medical Sciences,
Shiraz, Iran 4Department of Biochemistry,
Islamic Azad University, Shiraz
Branch, Shiraz, Iran 5School of Engineering, Faculty of
Science and Engineering, The
University of Waikato, Hamilton,
New Zealand
SCIENTIFIC PAPER
UDC 66.098:582.685.2:546.57:615
https://doi.org/10.2298/CICEQ150824002E
GREEN SYNTHESIS AND CHARACTERIZATION OF SILVER NANOPARTICLES USING Alcea rosea FLOWER EXTRACT AS A NEW GENERATION OF ANTIMICROBIALS
Article Highlights
• Synthesis of silver nanoparticles was developed using Alcea rosea flower extract
• AgNO3 concentration, flower extracts quantity, and reaction temperatures were deter-
mined to be significant factors in the AgNPs biosynthesis
• Prepared AgNPs were spherical in shape with 7.2 nm mean particle size
• Oxygen-bearing functional groups in biochemical compounds from A. rosea were res-
ponsible for reduction of Ag+
• The MIC for AgNPs against E. coli and S. aureus was determined to be 37.5 µg/mL
Abstract
Green synthesis of silver nanoparticles (AgNPs) was developed by treating Ag+
with Alcea rosea flower extract. AgNO3 concentration, flower extract quantity,
and reaction temperature were found to be significant factors in the bioreduction
reaction. Synthesized AgNPs were almost spherical in shape with an average
diameter of 7.2 nm. Fourier transform infrared spectroscopy (FTIR) analysis
revealed that oxygen-bearing functional groups in the A. rosea flower extract are
responsible for reduction of Ag+. The minimum inhibitory concentration (MIC) of
AgNPs against a Gram-positive (Staphylococcus aureus) and Gram-negative
(Escherichia coli) bacteria was determined to be 37.5 µg/ml.
Keywords: Alcea rosea; biochemical reduction; biosynthesis; green syn-thesis; Ag nanoparticles.
AgNPs have found widespread technological
applications due to their unique physicochemical,
optical and catalytic properties [1,2]. There is an inc-
reasing interest for application of AgNPs in house-
holds, medicine and industry [3]. Nowadays, there are
a lot of commercially available products containing
AgNPs, ranging from burn treating materials to anti-
microbial fabrics and paints [3]. The increasing applic-
ation of AgNPs will lead to increased demand for
AgNPs production. To date, various chemical and
Correspondence: A. Berenjian, School of Engineering, Faculty
of Science and Engineering, The University of Waikato, Hamil-
ton, New Zealand. E-mail: [email protected] Paper received: 24 August, 2015 Paper revised: 30 November, 2015 Paper accepted: 20 January, 2016
physicochemical techniques have been used for the
production of AgNPs [4,5]. However, all these
methods suffer from high energy consumption and
the use of toxic chemicals which are potentially dan-
gerous to the environment and human health. There-
fore, there is a need for development of reliable and
green process of AgNPs synthesis.
Green chemistry has emerged as a new concept
for development and implementation of chemical pro-
cesses in order to reduce or eliminate the use of
hazardous substances. Chemically synthesized nano-
particles are not colloidal or physicochemically stable
in aqueous media and therefore capping agents must
be used to increase the particle stability. In biosyn-
thesis reactions, biochemical species attach to the
surface of nanoparticles and act as capping and stab-
ilizing agent in a one-pot reaction [6-12]. Bioactive
A. EBRAHIMINEZHAD et al.: GREEN SYNTHESIS… Chem. Ind. Chem. Eng. Q. 23 (1) 3137 (2017)
32
compounds from microorganisms and plants have a
valuable capability for reduction and capping of
AgNPs without the use of any toxic chemicals and
harsh reaction conditions [6-8,10]. Carbohydrates and
proteins from microbial cells can be effective in red-
uction of Ag+ [12]. In comparison to usage of plant
extracts, biosynthesis of AgNPs using microorg-
anisms needs an elaborated process of culturing and
maintaining microbial cells that, in some cases, could
be pathogenic to humans. The use of plant extract
has advantages such as ease of handling, availability
and a broad viability of metabolites. AgNPs have
been synthesized using leaf extract of various plants
such as black tea, Lippia citriodora (Lemon Verbena),
maple (Acer sp.) and eucalyptus [7,8,13,14]. Other
parts of plants have also been used such as Piper
longum and Crataegus douglasii fruit extract, coffee
powder extract, Nephelium lappaceum, orange peel
extract, oil of Plukenetia volubilis L., Chrysanthemum
morifolium Ramat extract, Medicago sativa and Ster-
culia foetida seed exudate, sorghum bran extract and
Cinnamon zeylanicum bark extract [6,7,10,11,15-21].
Alcea rosea (Althaea rosea) is an important
medicinal herb in many countries. It was used trad-
itionally as expectorant, cooling, diuretic and emme-
nagogue substance. Alcea rosea flowers extract is
used as an anti-inflammatory, febrifuge, demulcent
and astringent agent. Flowers as well as their roots
are used in the treatment of inflammation of the kid-
neys and the uterus. Alcea rosea contains high mole-
cular weight acidic polysaccharides (1.3 to 1.6 million
Dalton) known as mucilages which are abundant in
flowers and leaves. These mucilages are composed
of glucoronic acid, galacturonic acid, rhamnose and
galactose. It also contains proteins, alkaloids and fla-
vonoids [22]. All of these biochemical compounds are
reported to be effective in bioreduction of Ag+ [6-8,10].
According to our best knowledge there is no report on
biosynthesis of AgNPs using Alcea rosea. Therefore,
the current research work aims to investigate: i) the
potential of Alcea rosea flower aqueous extract for
biosynthesis of AgNPs and ii) the antimicrobial effect
of the prepared AgNPs.
EXPERIMENTAL
Materials
Silver nitrate was purchased from Merck. All glass-
ware have been acid washed and then rinsed with
deionised water. All the solutions were prepared using
deionized-Millipore water (resistance >18 M cm).
Preparation of flower extract
Dried flowers of Alcea rosea were initially
washed in deionized water to remove the soil and
dust particles. The aqueous extract was consequently
prepared by mixing 2.5 g of dried flowers with 100 mL
of deionized water in a 250 mL Erlenmeyer flask. The
prepared mixture was boiled for 15 min, then filtered
through Whatman filter paper (Reeve angel®, grade
201) and stored at -20 C.
Preparation of AgNO3 solutions
Solutions of AgNO3 were prepared at 100, 50
and 10 mM concentration. For the initial/stock con-
centration, 1.7 g AgNO3 was dissolved in 100 mL
deionized water to obtain 100 mM solution. Solutions
with lower concentrations (50 and 10 mM) were pre-
pared by two and 10-fold dilutions, respectively.
Desired AgNO3 concentrations in the bioreduction
reactions were achieved by adding 1 mL of corres-
ponding solution to the reaction mixture.
Biosynthesis and characterization of AgNPs
Alcea rosea flowers extract was used as a
source for reducing and capping agent for synthesis
and stabilization of AgNPs in a simplified one-pot
reaction. The impact of various parameters such as
the amount of Alcea rosea flower extract, AgNO3 con-
centration and the reaction temperature was evalu-
ated by conducting several sets of experiments. The
AgNO3 concentration was tested at 1, 5 and 10 mM in
10 mL total reaction volume containing 4 mL (40
vol.%) flower extract at room temperature (27 C).
Impact of various flower extract amounts was also
investigated in the range from 10 to 70% of total
reaction volume at room temperature and 5 mM
AgNO3. The effect of reaction temperature on silver
ions reduction was evaluated at 15, 28, 50 and 75 C
using 5 mM AgNO3 and 4 mL flower extract. All the
reactions were monitored for 24 h.
The optical properties of the produced particles
were analyzed by ultraviolet and visible absorption
spectroscopy (T80+ UV/Vis spectrometer, PG Instru-
ments Ltd.) operated at a resolution of 1 nm within the
range of 300-700 nm. In each analysis, 0.1 mL of the
sample was diluted to 1 mL with deionized water [11].
Further characterizations were done by Transmission
Electron Microscopy (TEM, Philips, CM 10; HT 100
kV), Fourier-transform infrared spectroscopy (FTIR,
Bruker, Vertex 70, FT-IR spectrometer) and X-ray
powder diffraction (XRD, Siemens D5000).
Antimicrobial assay
Escherichia coli PTCC 1399 (ATCC 25922) and
Staphylococcus aureus PTCC 1112 (ATCC 6538)
A. EBRAHIMINEZHAD et al.: GREEN SYNTHESIS… Chem. Ind. Chem. Eng. Q. 23 (1) 3137 (2017)
33
were purchased from Persian Type Culture Collection
(PTCC). Minimum inhibitory concentration (MIC) was
determined using standard microdilution method
(CLSI M07-A8) [23]. In the experiment, AgNPs sus-
pension was prepared in Mueller-Hinton broth (MHB).
To prepare bacterial suspensions for inoculation, bac-
terial cells were cultured in MHB up to turbidity of the
BaSO4 0.5 McFarland standard (OD600 0.11). Then
the 0.5 McFarland suspensions were diluted to 1:20.
Finally, 10 µL of the prepared inoculums’ suspension
was transferred to each well in the 96-well plate con-
taining 90 µL MHB media with AgNPs. For blank
wells, 10 µL of fresh MHB was added to 90 µL MHB
with AgNPs. After 24 h incubation at 37 C, the
OD600 was measured by microplate reader (BioTek,
Power Wave XS2).
RESULTS AND DISCUSSION
Biosynthesis of AgNPs
Reduction of Ag+ to AgNPs resulted in color
change of the reaction solution due to excitation of
surface plasmon resonance (SPR) in the AgNPs.
AgNPs have a typical surface plasmon band absorp-
tion at about 400-450 nm. The UV-Vis spectroscopy,
therefore, can be used as an indirect method to
examine the formation and to some extend charac-
terisation of AgNPs [8,11]. The UV-Vis spectra of the
prepared particles in various concentrations of silver
nitrate are shown in Fig. 1. Increasing the silver nit-
rate concentration from 1 to 5 mM resulted in a major
increase of AgNPs content as indicated by the hyper-
chromic shift in SPR band. No significant increase in
AgNPs concentration was observed when using
higher AgNO3 concentration (>5 mM).
Figure 1. UV-Vis spectra of AgNPs prepared at various
concentrations of silver nitrate: a) 1, b) 5 and c) 10 mM.
SPR bands of the prepared AgNPs using differ-
ent amounts of flower extracts are shown in Fig. 2. By
increasing the flower extract up to 40% of the total
reaction volume, an obvious hyperchromic shift was
observed. However, further increase in the flower ext-
ract content, above 40 vol.% resulted in a significant
hypochromic shift in SPR band. Synthesis of the
metal nanoparticles was conducted in two main steps
namely, a) nucleation and b) growth of nanoparticles.
Organic compounds presence in the reaction mixture
has an inhibitory effect on the particle growth [24-26].
Thus, an optimal value of the Alcea rosea flower ext-
ract is required for reduction of Ag+ to AgNPs.
Figure 2. UV-Vis spectra of AgNPs prepared in various amounts
of Alcea rosea flower extract: a) 10, b) 20, c) 40 and d) 70% of
the reaction volume.
By increasing the volume of flower extract to
more than 40 vol.% of the total reaction volume, a
second absorption peak appeared at about 660 nm.
As shown in TEM micrographs (Fig. 3b), appearance
of this second peak is due to the formation of the
second population of large particles.
Figure 3. TEM micrographs of the prepared AgNPs at various
amounts of flower extract: a) 40 and b) 70 vol.% of the total
reaction volume.
The UV-Vis spectra of the prepared particles at
different temperatures are shown in Fig. 4. Increasing
A. EBRAHIMINEZHAD et al.: GREEN SYNTHESIS… Chem. Ind. Chem. Eng. Q. 23 (1) 3137 (2017)
34
the reaction temperature resulted in significant inc-
rease of the AgNPs concentration, as shown by the
hyperchromic shift in the SPR band. However, the
reaction temperature increase resulted in the appear-
ance of a shoulder in UV-Vis absorption spectra,
indicating the formation of poly-disperse AgNPs [12].
According to the results, room temperature is the
optimal temperature for bioreduction of AgNPs by
Alcea rosea flower extract. Conducting reaction in the
ambient condition could considerably reduce the
energy cost, which is one of the most important
issues in the scale-up process. A flowchart diagram of
the biosynthesis process is illustrated in Fig. 5.
Figure 4: UV-Vis spectra of the prepared AgNPs at various
temperatures: a) 27, b) 40, c) 50 and d) 70 C.
Figure 5. The flowchart diagram of the biosynthesis process.
Characterizations of AgNPs
Particle size distribution was determined by
measuring diameters of one hundred nanoparticles
randomly selected on the TEM images [27]. As shown
in Fig 6, the prepared particles were spherical in
shape with the average diameter of 7.2 nm. The pre-
pared particles were spherical in shape with the aver-
age diameter of 7.2 nm. The crystallinity of the par-
ticles was evaluated by X-ray powder diffraction pat-
terns (XRD, Siemens D5000) using drop coated films
on a glass slide [6,9]. As shown in Fig. 7, four main
characteristic diffraction peaks for silver were obs-
erved at 2θ values 38.2, 44.4, 64.7 and 77.4° due to
reflection from the crystal facets of (111), (200), (220)
and (311), respectively (JCPDS, silver file No. 04-
-0783) [11,13]. Three peaks around 2θ = 32° are indi-
cated by asterisks. Some researchers have attributed
these peaks to the interaction of silver nitrate with
biologic matrixes [28,29].
Figure 6. Particles size distribution of the prepared AgNPs.
Figure 7. XRD pattern of AgNPs indicating four main charact-
eristic peaks for silver, the peaks indicated by asterisk are from
mineral complexes.
FTIR spectra of AgNPs and Alcea rosea flower
extract are depicted in Fig. 8. The bands at 1059 and
1261 cm-1 are from C−O and C−C stretching vib-
rations, respectively. The peak with medium intensity
at 1421 cm-1 could be due to C−H bending vibrations.
Stretching vibrations of aliphatic C−H absorbed IR
radiation at about 2925 cm-1. The absorption peak
from carbonyl groups appeared at 1630 cm-1.
The broad absorption peak of hydrogen bonds
from O−H groups can be seen at 3394 cm-1 which
could overlap with the absorption from N−H bonds
[23,30-32]. Similarity to the flower extract, AgNPs
FTIR spectra indicates that AgNPs are capped with
biochemical compounds from Alcea rosea flower
extract. As it could be observed in the FTIR spectrum,
the main peaks come from oxygen-bearing functional
groups. It is widely believed that oxygen-containing
functionalities are necessary for anchoring of the
metal nanoparticles, and silver ions could easily oxid-
ize these groups [25].
A. EBRAHIMINEZHAD et al.: GREEN SYNTHESIS… Chem. Ind. Chem. Eng. Q. 23 (1) 3137 (2017)
35
Figure 8. FTIR spectra of AgNPs (a) and Alcea rosea flower
extract (b).
Antibacterial assay
Prepared nanoparticles have shown intense
effect on the bacterial growth (Fig. 9). The MIC con-
centrations for S. aureus and E. coli were determined
to be 37.5 µg/ml, which is acceptable compared to the
previously reported concentrations, 10-60 µg/ml [33-
–35]. Silver ions and silver based compounds have
strong antimicrobial effects and have been used for
decades as antimicrobial agents in various fields
[36,37]. AgNPs provide high fraction of exposed
atoms due to their extremely small size and thus
expand the contact surface of silver with microorg-
anisms. It has been confirmed that antimicrobial pro-
perties of AgNPs are due to oxidation of the exposed
silver atoms and release of Ag+ from the surface of
AgNPs [37]. Exposure to the air promoted Ag+ release
and resulted in 2.3-fold increase in the AgNPs anti-
microbial effects [36]. Silver ions are potent oxidants
and can destroy variety of cellular structures. The Ag+
enter into the bacterial cells by penetrating through
the cell wall and consequently turn the DNA molecule
into condensed form which results in the cell death. In
addition, it was also shown that Ag+ binds to funct-
ional groups of proteins, resulting in protein denatur-
ation [35,38]. Metal nanoparticles and particularly
AgNPs are able to destroy the permeability of the
bacterial membranes [34,39]. Exposure of bacterial
membranes to the AgNPs resulted in the leakage of
reducing sugars and proteins and induced the res-
piratory chain dehydrogenases into inactive state [34].
Commonly used antibiotics act very specifically
and target exact physiological points in the microorg-
anisms. This precise strategy provides a chance for
some mutant strains to escape and distribute. To
alleviate this condition and reduce the probability of
new resistant strain appearance, multidrug therapies
have been developed and used; however, this stra-
tegy is ineffective against multidrug resistant strains.
Interestingly, released Ag+ from AgNPs with multi-tar-
geting antimicrobial mechanism of action, significantly
reduced the chance for mutation and development of
a bacterial resistance mechanism. In addition, AgNPs
could increase the potency of the common antibiotics
[33,40-42].
Figure 9. Antimicrobial effect of AgNPs against E. coli
and S. aureus.
CONCLUSION
Alcea rosea flower extract contains bioactive
compounds that are effective in bioreduction of Ag+.
These biochemical compounds contain oxygen bear-
ing functional groups, which act as an anchor for Ag+.
Subsequently, produced AgNPs are capped with hyd-
rophilic biochemical compounds, which make them
colloidally stable. Reaction conditions such as silver
precursor concentration, amount of flower extract and
reaction temperature are the key factors in prepar-
ation of quality-based AgNPs. Therefore, these fac-
tors should be controlled in order to produce uniform
particles with narrow particle size distribution.
Acknowledgments
This investigation was financially supported by
The Fasa University of Medical Sciences, Fasa, Iran,
and The University of Waikato, Hamilton, New Zeal-
and.
REFERENCES
[1] Z.-J. Jiang, C.-Y. Liu, L.-W. Sun, J. Phys. Chem., B 109
(2005) 1730-1735
[2] A.D. McFarland, R.P. Van Duyne, Nano Lett. 3 (2003)
691057-1062
[3] A. Haider, I.-K. Kang, Adv. Mater. Sci. Eng. 2015 (2015)
A. EBRAHIMINEZHAD et al.: GREEN SYNTHESIS… Chem. Ind. Chem. Eng. Q. 23 (1) 3137 (2017)
36
[4] K.C. Song, S.M. Lee, T.S. Park, B. S. Lee, Korean J.
Chem. Eng. 26 (2009) 153-155
[5] V.G. Pol, D. Srivastava, O. Palchik, V. Palchik, M. Slifkin,
A. Weiss, A. Gedanken, Langmuir 18 (2002) 3352-3357
[6] G.A. Kahrilas, L.M. Wally, S.J. Fredrick, M. Hiskey, A.L.
Prieto, J.E. Owens, ACS Sustain. Chem. Eng. 2 (2013)
367-376
[7] M.N. Nadagouda, R.S. Varma, Green Chem. 10 (2008)
859-862
[8] D. Cruz, P.L. Falé, A. Mourato, P.D. Vaz, M. Luisa
Serralheiro, A.R.L. Lino, Colloids Surfaces, B 81 (2010)
67-73
[9] A.I. Lukman, B. Gong, C.E. Marjo, U. Roessner, A.T.
Harris, J. Colloid Interface Sci. 353 (2011) 433-444
[10] E.C. Njagi, H. Huang, L. Stafford, H. Genuino, H.M.
Galindo, J.B. Collins, G.E. Hoag, S.L. Suib, Langmuir 27
(2010) 264-271
[11] M. Sathishkumar, K. Sneha, S. Won, C.-W. Cho, S. Kim,
Y.-S. Yun, Colloids Surfaces, B 73 (2009) 332-338
[12] J. Xie, J.Y. Lee, D.I. Wang, Y.P. Ting, ACS Nano 1
(2007) 429-439
[13] S. Vivekanandhan, M. Schreiber, C. Mason, A.K. Moh-
anty, M. Misra, Colloids Surfaces, B 113 (2014) 169-175
[14] Y.-Y. Mo, Y.-K. Tang, S.-Y. Wang, J.-M. Ling, H.-B.
Zhang, D.-Y. Luo, Mater. Lett. 144 (2015) 165–167
[15] M. Ghaffari-Moghaddam, R. Hadi-Dabanlou, J. Ind. Eng.
Chem. 20 (2014) 739-744
[16] P. Rajasekharreddy, P.U. Rani, Mater. Sci. Eng., C 39
(2014) 203-212
[17] N.J. Reddy, D. Nagoor Vali, M. Rani, S.S. Rani, Mater.
Sci. Eng., C 34 (2014) 115-122
[18] B. Kumar, K. Smita, L. Cumbal, A. Debut, Ind. Crops
Prod. 58 (2014) 238-243
[19] B. Kumar, K. Smita, L. Cumbal, A. Debut, J. Saudi Chem.
Soc. 19 (2015) 574-580
[20] B. Kumar, K. Smita, L. Cumbal, Y. Angulo, J. Mol. Liq.
211 (2015) 476-480
[21] V. Dhand, L. Soumya, S. Bharadwaj, S. Chakra, D. Bhatt,
B. Sreedhar, Mat. Sci. Eng., C 58 (2016) 36-43
[22] S. Khoei, M. Azarian, S. Khoee, Curr. Nanosci. 8 (2012)
690-696
[23] A. Ebrahiminezhad, S. Davaran, S. Rasoul-Amini, J.
Barar, M. Moghadam, Y. Ghasemi, Curr. Nanosci. 8
(2012) 868-874
[24] A. Ebrahiminezhad, Y. Ghasemi, S. Rasoul-Amini, J.
Barar, S. Davaran, Colloids Surfaces, B 102 (2013) 534-
–539
[25] A. Ebrahiminezhad, Y. Ghasemi, S. Rasoul-Amini, J.
Barar, S. Davaran, Bull. Korean Chem. Soc. 33 (2012)
3957-3962
[26] A. Gholami, S. Rasoul-amini, A. Ebrahiminezhad, S.H.
Seradj, Y. Ghasemi, J. Nanomater. 2015 (2015)
[27] A. Ebrahiminezhad, S. Rasoul-Amini, S. Davaran, J.
Barar, Y. Ghasemi, Curr. Nanosci. 2014 (2014) 382-388
[28] N. Vigneshwaran, N. Ashtaputre, P. Varadarajan, R.
Nachane, K. Paralikar, R. Balasubramanya, Mater. Lett.
61 (2007) 1413-1418
[29] K. Chitra, G. Annadurai, Biomed Res. Int. 2014 (2014)
[30] A. Ebrahiminezhad, S. Najafipour, A. Kouhpayeh, A. Ber-
enjian, S. Rasoul-Amini, Y. Ghasemi, Colloids Surfaces,
B 118 (2014) 249-253
[31] A. Ebrahiminezhad, S. Rasoul-Amini, A. Kouhpayeh, S.
Davaran, J. Barar, Y. Ghasemi, Curr. Nanosci. 11 (2015)
113-119
[32] A. Ebrahiminezhad, V. Varma, S. Yang, A. Berenjian,
Appl. Microbiol. Biotechnol. (2015) 1-8
[33] S. Ruden, K. Hilpert, M. Berditsch, P. Wadhwani, A.S.
Ulrich, Antimicrob. Agents Chemother. 53 (2009) 3538-3540
[34] W.-R. Li, X.-B. Xie, Q.-S. Shi, H.-Y. Zeng, O.-Y. You-
-Sheng, Y.-B. Chen, Appl. Microbiol. Biotechnol. 85
(2010) 1115-1122
[35] I. Sondi, B. Salopek-Sondi, J. Colloid Interface Sci. 275
(2004) 177-182
[36] Z.-M. Xiu, J. Ma, P.J. Alvarez, Environ. Sci. Technol. 45
(2011) 9003-9008
[37] Z.-M. Xiu, Q.-B. Zhang, H.L. Puppala, V.L. Colvin, P.J.
Alvarez, Nano Lett. 12 (2012) 4271-4275
[38] Q. Feng, J. Wu, G. Chen, F. Cui, T. Kim and J. Kim, J.
Biomed. Mater. Res. 52 (2000) 662-668
[39] F. Ansari, P. Grigoriev, S. Libor, I.E. Tothill, J.J. Rams-
den, Biotechnol. Bioeng. 102 (2009) 1505-1512
[40] L.S. Devi, S. Joshi, Mycobiology 40 (2012) 27-34
[41] A.M. Fayaz, K. Balaji, M. Girilal, R. Yadav, P.T. Kalaichel-
van, R. Venketesan, Nanomed. Nanotechnol. Biol. Med.
6 (2010) 103-109
[42] A.R. Shahverdi, A. Fakhimi, H.R. Shahverdi, S. Minaian,
Nanomed. Nanotechnol. Biol. Med. 3 (2007) 168-171.
A. EBRAHIMINEZHAD et al.: GREEN SYNTHESIS… Chem. Ind. Chem. Eng. Q. 23 (1) 3137 (2017)
37
ALIREZA EBRAHIMINEZHAD1–3
YAHYA BARZEGAR3,4
YOUNES GHASEMI3
AYDIN BERENJIAN5
1Noncommunicable Diseases
Research Centre, Fasa University of
Medical Sciences, Fasa, Iran 2Department of Medical Biotechnology,
School of Medicine, Fasa University of
Medical Sciences, Fasa, Iran 3Department of Pharmaceutical
Biotechnology, School of Pharmacy
and Pharmaceutical Sciences
Research Centre, Shiraz University of
Medical Sciences, Shiraz, Iran 4Department of Biochemistry, Islamic
Azad University, Shiraz Branch, Shiraz,
Iran 5School of Engineering, Faculty of Sci-
ence and Engineering, The University
of Waikato, Hamilton, New Zealand
NAUČNI RAD
ZELENA SINTEZA I KARAKTERIZACIJA NANOČESTICA SREBRA POMOĆU EKSTRAKTA CVETA Alcea rosea KAO ANTIMIKROBNOG SREDSTVA NOVE GENERACIJE
U radu je razvijena zelena sinteza nanočestica srebra (AgNP) tretiranjem Ag+ ekstraktom
cveta Alcea rosea. Utvrđeno je da su koncentracija AgNO3, količina ekstrakta cveta i
temperatura reakcije značajni faktori ove bioredukcione reakcije. Dobijene AgNP čestice
su sfernog oblika sa prosečnim prečnikom od 7,2 nm. FTIR analiza je pokazala da su za
redukciju Ag+ odgovorne funkcionalne grupe sa kiseonikom prisutne u ekstraktu cveta A.
rosea. Određena je minimalna inhibitorna koncentracija (MIC) AgNP za Gram-pozitivne
(Staphylococcus aureus) i Gram-negativne (Escherichia coli) bakterije i ona iznosi 37,5
µg/ml.
Ključne reči: Alcea rosea, biohemijska redukcija, biosinteza, zelena sinteza, Ag
nanočestice.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 3947 (2016) CI&CEQ
39
ALEKSANDRA MIŠAN1
BOJANA ŠARIĆ1
IVAN MILOVANOVIĆ1
PAVLE JOVANOV1
IVANA SEDEJ1
VANJA TADIĆ2
ANAMARIJA MANDIĆ1
MARIJANA SAKAČ1
1Institute of Food Technology in
Novi Sad (FINS), University of Novi
Sad, Novi Sad, Serbia 2Institute for Medicinal Plant
Research “Dr Josif Pančić”,
Belgrade, Serbia
SCIENTIFIC PAPER
UDC 633.12:66.061:616:615
https://doi.org/10.2298/CICEQ150704004M
PHENOLIC PROFILE AND ANTIOXIDANT PROPERTIES OF DRIED BUCKWHEAT LEAF AND FLOWER EXTRACTS
Article Highlights
• Buckwheat leaf and flower extracts were obtained by different extraction procedures
• Rutin and chlorogenic acid were identified as most abundant phenolic compounds
• Some of the extracts were as efficient as BHT in β-carotene bleaching test
• The extracts demonstrated strong ability to inhibit the destruction of erythrocytes
• The extracts prolonged the beginning of the oxidation process in sunflower oil
Abstract
Due to a high content of rutin (2-10%), dried buckwheat leaf and flower (DBLF)
formulations were shown to be efficient in the treatment of vascular diseases. In
order to find a cost effective way for the extraction of antioxidants, the effects of
ethanol/water ratio and temperature on the extraction efficiency of phenolic
compounds and the mechanisms of antioxidant action of the extracts were
tested. Extraction with ethanol/water mixture (80:20, v/v) for 24 h at room tempe-
rature, after the mixture was just brought to boil was demonstrated to be an
efficient and cheap way for obtaining a high yield of rutin (49.94±0.623 mg/g
DBLF). The most abundant phenolic compounds in DBLF extracts were rutin and
chlorogenic acid. Flavonoids, especially rutin, were shown to be the most res-
ponsible for the antioxidant activity in all investigated lipid model systems, acting
as free radical scavengers, electron-donating substances and chelators of iron
ions. In β-carotene bleaching tests, the extracts with the highest activity were as
efficient as BHT (butylated hydroxytoluole). Regarding the results of antihemo-
lytic and Schaal oven tests, the extracts demonstrated remarkable ability to
inhibit the oxidative destruction of erythrocytes and to prolong the beginning of
the oxidation process in sunflower oil.
Keywords: dried buckwheat leaf and flower, rutin, antioxidant activity, lipid oxidation, extraction.
Common buckwheat (Fagopyrum esculentum
Moench) is a highly nutritious pseudocereal known as
a very rich source of antioxidants, especially rutin [1].
The nutritive profile and antioxidant potential of buck-
wheat seeds has been extensively investigated [2-4],
and buckwheat seeds can now be regarded a “func-
tional food” [5]. However, recent studies indicate that
the highest concentration of rutin, up to 10%, is accu-
Correspondence: A. Mišan, Institute of Food Technology in Novi
Sad, University of Novi Sad, Bulevar cara Lazara 1, 21000 Novi
Sad, Serbia. E-mail: [email protected] Paper received: 4 July, 2015 Paper revised: 3 February, 2016 Paper accepted: 12 February, 2016
mulated in leaves and blossoms of the buckwheat
plant [1,6].
Plant phenolic compounds are well known as
highly effective free radical scavengers and antioxi-
dants [7]. The role of an antioxidant in a food product
is related to its ability to inhibit or stop rancidity and/or
deterioration of the nutritional quality [8]. Also, many
natural antioxidants are supposed to have protective
effects against chronic diseases, mainly by scaveng-
ing oxygen radicals, which can deteriorate biological
membranes.
Fagopyri herba is a herbal drug derived from
dried areal tissues of common buckwheat (Fagopy-
rum esculentum Moench) and has been used in the
treatment of vascular diseases [9]. Rutin (quercetin-3-
A. MIŠAN et al.: PHENOLIC PROFILE AND ANTIOXIDANT PROPERTIES… Chem. Ind. Chem. Eng. Q. 23 (1) 3947 (2017)
40
-O-rutinoside), the dominant flavonol glycoside in
DBLF, has been reported to possess antioxidant act-
ivity, to antagonize the increase of capillary fragility
associated with haemorrhagic disease, to reduce high
blood pressure [10], to decrease the permeability of
the blood vessels, to have an anti-oedema effect, and
to reduce the risk of atherosclerosis [11].
Since DBLF can be considered as an ingredient
for designing functional food products, the objective of
this research was to find a cost effective way for the
extraction of phenolic compounds from DBLF in order
to obtain the highest antioxidant activity of the ext-
racts. Water and ethanol/water mixtures were chosen
as nontoxic and environmentally friendly solvents,
which have been shown to be effective in the extract-
ion of quercetin glycosides [12]. Besides the phenol
profiling, the aim of this work was to test the mech-
anisms of antioxidant action of the obtained extracts
in chosen model systems to be able to predict their
possible use.
EXPERIMENTAL
Materials
DBLF (dried areal parts of Fagopyrum escul-
entum Moench collected during the flowering season)
was obtained from the Institute for Medicinal Plants
Research “Dr Josif Pančić” (Belgrade, Serbia) where
a herbarium voucher specimen (No. 31210911) was
deposited. For replicates, three packages of herbal
drug were provided, each containing 500 g of mat-
erial.
Butylated hydroxytoluole (BHT), 1,1-diphenyl-2-
-picrylhydrazyl (DPPH), ethylene diamine tetraacetic
acid disodium salt dihydrate (EDTA), 3-(2-pyridyl)-5-
-6-bis(4-phenyl-sulfonic acid)-1,2,4-triazine (ferro-
zine), ferrous sulfate heptahydrate, linoleic acid
(99%), potassium ferricyanide, sodium carbonate,
Tween 40, trichloracetic acid (TCA), Folin-Ciocalteu's
reagent, standard substances including gallic acid,
protocatechuic acid, caffeic acid, vanillic acid, chloro-
genic acid, syringic acid, ferulic acid, rutin, myricetin,
rosmarinic acid, trans cinnamic acid, naringenin, lut-
eolin, kaempferol, and apigenin were obtained from
Sigma (Sigma-Aldrich GmbH, Sternheim, Germany).
Quercetin was a product of J. T. Baker (Deventer, the
Netherlands), while high-performance liquid chro-
matography (HPLC) grade methanol, formic acid
(HPLC grade) and ethanol 96% were purchased from
Merck (Darmstadt, Germany). Water was purified
using Millipore Elix 10 UV water purification system
(Molsheim, France), and ultrapure water used for
HPLC mobile phase preparation was obtained using
Simplicity UV, Millipore (Molsheim, France).
Preparation of extracts
DBLF (2 g) was mixed either with 50 mL of
water or ethanol/water mixtures (50:50 and 80:20,
V/V). Maceration was performed for 24 h at room
temperature followed by extraction in an ultrasonic
bath (10 min at room temperature). Corresponding
extracts brought to boil before the maceration were
also prepared. The extracts were filtered through the
filter paper (Whatman, Grade 4 Chr, UK) and stored
at -4 C (up to two days) until further use. Preparation
of the extracts for HPLC-MS/MS analyses included
additional evaporation to dryness and redissolving in
an appropriate mobile phase.
Total flavonoid content
Colorimetric aluminum chloride method was
used for determination of total flavonoid content [13],
which is based on the formation of a complex flavo-
noid-aluminium. The probes were prepared by mixing
5 mL of extract, 1 mL of distilled water, and 2.5 mL of
AlCl3 solution (26.6 mg AlCl36H2O and 80 mg
CH3COONa dissolved in 20 mL distilled water). A
blank probe was prepared by replacing AlCl3 solution
with distilled water. The absorbance of probes and
blank probe were measured immediately at 430 nm.
Rutin was used as a standard and results were exp-
ressed as rutin equivalents (RE, g RE per 100 g of
sample). Absorption readings at 415 nm were taken
against a blank sample (reaction mixture without
AlCl3). Rutin (2.5 to 50 μg/mL) was used for the calib-
ration curve construction.
Identification of phenolic compounds by LC-MS/MS
Rapid resolution liquid chromatography with
mass selective detection was performed using an
Agilent Technologies 1200 Series liquid chromato-
graph coupled with Agilent Technologies 6410 triple-
quadropole (QQQ) mass spectrometer. An Eclipse
XDB-C18, 1.8 μm, 4.6 mm50 mm column (Agilent)
was used for separation of 18 phenolic compounds.
The solvent gradient program was created by varying
the proportion of solvent A (0.1 vol.% formic acid in
water) and solvent B (methanol). The following grad-
ient mode was used for phenolic acids identification:
initial 10% B; 0-2 min, 10-43% B; 2-7 min, 43% B;
7-9.5 min, 43-100% B; 9.5-10.5 min, 100% B with flow
rate of 1 mL/min. Identification of flavonoids was per-
formed using the following gradient mode: initial 4%
B; 0-1.5 min, 4-4.5% B; 1.5-4 min, 4.5-10% B; 4-11
min, 10% B; 11-18 min, 10-100% B, with flow rate of
1.2 mL/min. Injection volume was 1 μL. The eluted
A. MIŠAN et al.: PHENOLIC PROFILE AND ANTIOXIDANT PROPERTIES… Chem. Ind. Chem. Eng. Q. 23 (1) 3947 (2017)
41
components were ionized in negative electrospray
ionization (ESI) mode, using nitrogen as nebulizer
(pressure of 50 psi) and drying gas (temperature of
350 C, flow 8 L/min). Extracts used for LC–MS/MS
quantification were dissolved in starting mobile phase
solvent to the concentration of 0.2 mg/mL. All used
standards were dissolved in methanol to prepare
stock solutions of 1 mg/mL and the mix of stock sol-
utions was prepared, with concentration of each com-
pound being 100 μg/mL. Analyses were performed in
MRM (multiple reaction monitoring) mode. Com-
pound-specific, optimized MS/MS parameters are
given in Table 1.
Quantification of phenolic compounds by HPLC-DAD
A liquid chromatograph (Agilent 1200 series),
equipped with a DAD detector and an Eclipse XDB-
-C18, 1.8 μm, 4.6 mm50 mm column (Agilent) was
used for quantification of identified phenolic com-
pounds in DBLF extracts. A single rapid resolution
HPLC method suitable for the determination of 17
phenolic compounds, developed and validated as
previously reported by Mišan et al. [14], was used. In
brief, solvent gradient was performed by varying the
proportion of solvent A (methanol) to solvent B (1
vol.% formic acid in water) as follows: initial 10% A;
0-10 min, 10-25 % A; 10-20 min, 25-60 % A; 20-30
min, 60-70 % A at a flow-rate of 1 mL/min. The total
running time and post-running time were 45 and 10
min, respectively. The column temperature was 30
C. The injected volume of samples and standards
was 5 μL and it was done automatically using an
autosampler. The spectra were acquired in the range
of 210-400 nm and chromatograms plotted at 280,
330 and 350 nm with a bandwidth of 4 nm, and with
reference wavelength/bandwidth of 500/100 nm.
DPPH radical scavenging activity
A modified method of Hatano et al. [15] was
used to determine effect of different extracts on sca-
venging of 1,1-diphenyl-2-picrylhydrazyl (DPPH•) rad-
icals. 0.1 mL of examined extract previously diluted to
obtain at least four different concentrations (0.1 to 5
mg plant material/mL), 1.0 mL of DPPH• (90 μmol/L)
and 2.9 mL of methanol were shaken vigorously and
left to stand in the dark for 60 min. The absorbance
was measured at 515 nm (Cintra 101, GBC scientific,
UV/Vis) against the control (above mentioned mixture
without extract). Results were expressed as the con-
centration (mg plant material/mL) of the extract lead-
ing to 50% reduction of the initial DPPH• concen-
tration (IC50 value). BHT in the concentration range
0.006-0.600 mg/mL was used as a control.
Chelating activity on Fe2+
Chelating activity of the examined extracts on
Fe2+ was measured according to the method of
Decker and Welch [16]. Plant extracts were dissolved
in ethanol, in an appropriate manner to obtain a
series of dilutions (10 to 40 mg plant material/mL).
Table 1. Retention time, MRM transition, collision energy (CE), and fragmentor voltage of 18 phenolic compounds
Phenolic acid Time, min MRM (m/z) CE / eV Fragmentor voltage, V
Quinic acid 0.612 191>127; 109 12;12 99
Gallic acid 1.399 169>125 8 103
Protocatechuic acid 1.787 153>109 8 67
p-Hydroxybenzoic acid 4.787 137>93 8 82
Chlorogenic acid 6.612 353>191 20 113
Vanillic acid 7.135 167>123; 152 4;8 93
Caffeic acid 7.615 179>107; 135 16;8 73
Syringic acid 9.873 197>106; 121 16;8 73
Ferulic acid 14.254 193>134; 149; 178 4;4;4 119
Sinapic acid 14.569 223>208; 164 4;0 118
Rosmarinic acid 14.569 359>161;179 40;56 170
Flavonoids
Catechin 1.070 289>245; 203; 137 8;12;16 134
Epicatechin 1.830 289>109 20 134
Rutin 3.192 609>179; 300 40;36 118
Myricetin 3.528 317>109; 137; 151 32;28;20 144
Quercetin 4.578 301>121; 151; 179 28;16;12 135
Naringenin 4.761 271>151; 119 8;20 93
Kaempferol 6.246 285>151; 93; 117 16;36;40 155
A. MIŠAN et al.: PHENOLIC PROFILE AND ANTIOXIDANT PROPERTIES… Chem. Ind. Chem. Eng. Q. 23 (1) 3947 (2017)
42
The IC50 value (mg/mL) was defined as the concen-
tration of extract that chelates 50% of Fe2+ under the
experimental conditions. It was obtained by interpol-
ation from linear regression analysis. EDTA in the
concentration range 0.006-0.600 mg/mL was used as
a control.
Reducing power
The reducing power of the examined extracts, in
series of different concentrations (0.100 to 15 mg
plant material/mL), was determined according to the
method of Oyaizu [17]. This method is based on the
reduction of Fe3+ to Fe2+ and measuring absorbance
(at 700 nm) of the Perl’s Prussian blue complex. The
IC50 value (mg/mL) was defined as an effective con-
centration of extract at which the absorbance of react-
ion mixture reaches 0.5 for reducing power. It was
obtained by interpolation from linear regression ana-
lysis. BHT in the concentration range 0.006-0.600
mg/mL was used as a control.
β-Carotene bleaching method
Antioxidant activity (AA) of the series of exam-
ined extracts (1 to 40 mg plant material/mL) was det-
ermined using β-carotene bleaching method, an in
vitro assay which measures oxidative loss of β-caro-
tene in β-carotene/linoleic acid emulsion. Thermal
autoxidation at 50 C was performed for 2 h. The deg-
radation rate of β-carotene was calculated according
to first-order kinetics and the AA was expressed as
percent of inhibition relative to the control [18]. The
IC50 value (mg/mL) was defined as the concentration
at which AA was 50% under the experimental con-
ditions, and it was obtained by interpolation from
linear regression analysis. BHT in the concentration
range 0.2-2 mg/mL was used as a control.
Antihemolytic activity
Antihemolytic activity of the extracts was det-
ermined using the method of Ko et al. [19], which was
further optimized for DBLF extracts as previously
described by Šarić et al. [20]. A solution of hydrogen
peroxyde (0.0625 vol.%) in phosphate-buffered saline
(PBS) solution was used instead of concentrated sol-
ution, and incubation time was set to 2 h instead of 4
h from the original method. The extent of the hemo-
lysis in every sample was calculated as: 100(Asample/
/Atotal hemolysis) = % of hemolysis. Ascorbic acid solution
in the concentration range 0.005-0.500 mg/mL was
used as a control. Since the percentage of hemolysis
was calculated for different sample concentrations (1
to 40 mg plant material/mL), these values were plot-
ted against sample concentrations and, using linear
regression, IC50 values (concentration of the inves-
tigated extract at which 50% of hemolysis inhibition is
achieved) of every investigated extract, as well as the
ascorbic acid solution, were calculated.
Schaal oven test
The Schaal oven test at 70 C was conducted to
evaluate the antioxidant effectiveness of ethanolic
extracts of DBLF in retarding the rancidity of com-
mercially available refined sunflower oil during 12
days of storage. For that purpose, 5 g of oil was
mixed with 20 mass% tested extracts and spread in 1
cm layer in glass containers. The test was carried out
in the dark and oxidative changes were monitored
gravimetrically. Experiments were also carried out
with synthetic antioxidant, BHT at 10 ppm level and
the control sample with no added antioxidants. The
extracts were evaporated to dryness and redissolved
in methanol. A control sample was prepared by using
the same amount of methanol used to dissolve the
antioxidant (BHT) and the extracts. In order to monitor
the kinetics of oxidation process [21], first derivatives
of samples weight gain were calculated using the
software SciDAVis 0.2.4. [22].
Statistical analysis
Apart from the extraction procedures and Schaal
oven test, which were done in duplicate, all analyses
were performed in triplicate, and the mean values
with the standard deviations are reported. Analysis of
variance and Duncan's multiple range tests were
used. Statistical data analysis software Statistica
(StatSoft, Inc. 2011) [23] was used for analysis. P
values < 0.05 were regarded as significant.
RESULTS AND DISCUSSION
Referring to Kim et al. [24] who tested different
solvents in order to find the most effective way for the
extraction of rutin from buckwheat, the use of aque-
ous ethanol and acetone (both 50 vol.%) as extraction
solvents produced the highest yields of rutin. Kim et
al. [24] also found that the extraction temperatures in
the range 60–80 C and the extraction time from 0.5 to
1 h were optimal for achieving a high recovery yield of
rutin. On the other hand, Hinneburg and Neubert [12]
suggested that an extract with good antioxidant act-
ivity, a high content of phenolics, and a low content of
the phototoxic fagopyrin can be yielded by agitated
maceration with 30% ethanol at 60 C for 2 h.
Due to their low toxicity and supposed effici-
ency, water and ethanol/water mixtures were chosen
for the extraction of phenolic compounds from DBLF
in our experiment. Instead of prolonged heating which
is energy consuming and may result in destruction of
A. MIŠAN et al.: PHENOLIC PROFILE AND ANTIOXIDANT PROPERTIES… Chem. Ind. Chem. Eng. Q. 23 (1) 3947 (2017)
43
thermally labile constituents of interest, the effect of
short term heating, just to bring the mixture to boil and
24 h maceration afterwards on the extraction effici-
ency was examined. The following extraction yields
were obtained: water extraction = 20.34%; boiling
water extraction = 19.72%; ethanol/water (50:50)
extraction = 21.17%; boiling ethanol/water (50:50)
extraction = 21.28%; ethanol/water (80:20) = 21.28%
; boiling ethanol/water (80:20) extraction = 19.70%.
Total flavonoid content of plant extracts and HPLC
identification and quantification of phenolic
compounds
Besides the total flavonoid content determin-
ation, chemical characterization of the extracts inc-
luded identification of phenolic compounds by LC-
-MS/MS and quantification of identified compounds by
HPLC/DAD. Eighteen secondary biomolecules were
included into the identification (LC-MS/MS) method,
based on the availability of reference standards in the
laboratory. Method development started with the sel-
ection of precursor ions, which was done in MS2Scan
mode. The ionisation predominantly resulted in the
formation of [M−H]–. To assure high yield of precursor
ions while simultaneously preventing in-source frag-
mentation, fragmentor voltage (V) was optimised for
each compound. For this purpose, a standard mixture
was analyzed in MS2SIM mode, using fragmentor
voltages from 60 to 200 V in 10 V increments. In order
to optimise collision energy, the standard mixture was
subsequently analysed in Product Ion Scan mode,
using [M−H]– as precursors, optimal fragmentor volt-
age, and collision cell voltages ranging from 0-50 V
(in 10 V increments). Identification of the phenolic
constituents of the extracts was based on the com-
parison of the retention times and ionization patterns
in MRM mode with those of the external standards
(Table 1). Out of 18 analyzed compounds, the pre-
sence of 7 was confirmed in the extracts. After the
identification, phenolic compounds were quantified by
HPLC/DAD method, which was previously optimized
and validated for the determination of phenolic com-
pounds in crude medicinal plant extracts [14]. Content
of phenolic compounds in each extract is presented in
Table 2. Referring to the obtained results, the most
abundant phenolic compound in DBLF extracts was
rutin with the exception of Extract A where chloro-
genic acid was present in the highest concentration
(Table 2). The highest yield of rutin was obtained by
using boiling ethanol/water (80:20, V/V) as extracting
solvent, while the highest yields of chlorogenic acid
were obtained by water extraction. Ethanol/water
(50:50) mixture was shown to be the most efficient for
the extraction of investigated phenolic acids. Accord-
ing to Hinneburg and Neubert [12], quercetin origin-
ates as a product of rutin degradation by flavonol-3-
-O-β-heterodisaccharidase in buckwheat herb or in an
extract. Referring to the obtained results, quercetin
was present in all extracts at a concentration, which
was not highly influenced by the difference in applied
extraction procedures.
Our results are in accordance with the findings
of Hinneburg and Neubert [12], who reported that the
main phenolics of DBLF are rutin, chlorgenic acid,
and hyperoside and that rutin content in buckwheat
herb can be up to 8%. Besides rutin and quercetin,
the other flavonoids like vitexin, isovitexin, orientin
and isoorientin, which could exhibit 4-40% of total
antioxidant activity, were also reported to be present
in green parts of buckwheat [25]. However, due to the
lack of external standards, those compounds could
not have been quantified by applied HPLC method.
Instead, total flavonoid content of each extract was
estimated (Table 3). Due to the difference in applied
methods, these results differ from the results obtained
by HPLC. However, strong positive correlation
between HPLC-determined rutin and total flavonoid
contents was observed (r = 0.931, P < 0.05), indi-
cating that rutin is by far the most abundant flavonoid
in the extracts.
Table 2. Content of phenolic compounds (mg/g DBLF) in extracts: A-water; B-boiling water; C-ethanol/water (50:50); D-boiling ethanol/
/water (50:50); E-ethanol/water (80:20); F-boiling ethanol/water (80:20); values are means of three determinations ± standard deviation.
Values a,b,c,d, in each row with the same superscript are not significantly different (P < 0.05)
Compound Extract A Extract B Extract C Extract D Extract E Extract F
Gallic acid 0.052±0.009 a 0.037±0.004 a 0.874±0.065 c 0.864±0.03 c 0.779±0.021 b 0.766±0.071b
Protocatechuic acid 0.172±0.011a 0.175±0.016 a 0.266±0.021 c 0.236±0.019 b 0.209±0.014 b 0.228±0.048 b
Caffeic acid 0.214±0.013 a 0.240±0.027 a 1.084±0.101 c 0.953±0.091 b 0.988±0.077 b 0.966±0.068 b
Catehin 0.093±0.007 b 0.062±0.007 a 0.088±0.005 b 0.087±0.005 b 0.090±0.003 b 0.092±0.008 b
Chlorogenic acid 1.652±0.075 d 1.621±0.111 d 0.855±0.042b 0.950±0.036c 0.772±0.052 a 0.743±0.026 a
Rutin 0.251±0.013a 2.274±0.125b 40.35±0.693d 39.87±0.265d 34.93±0.213 c 49.94±0.623e
Quercetin 0.252±0.031 a 0.254±0.026 a 0.310±0.031 bc 0.306±0.034 bc 0.344±0.029bc 0.260±0.022 ab
A. MIŠAN et al.: PHENOLIC PROFILE AND ANTIOXIDANT PROPERTIES… Chem. Ind. Chem. Eng. Q. 23 (1) 3947 (2017)
44
Antioxidant activity (AOA) of the extracts
Due to the complexity of oxidative processes in
food matrices and cells and the versatile nature of
plant extracts a single method for the screening of
antioxidant potential is not recommended. Therefore,
the antioxidant effects of plant products must be
evaluated by combining at least two or more different
in vitro assays to obtain relevant data. With respect to
this, the antioxidant properties of the examined ext-
racts were tested through their ability to donate elec-
tons or H atoms, chelate/deactivate transition metal
ions and to inhibit the process of lipid oxidation.
Free radical scavenging activity of the extracts
was tested by applying the method towards long-lived
DPPH• while the electron-donating capacity was eval-
uated by measuring their reducing power. Referring to
the results (Table 3) expressed as IC50 values, the
extracts showed lower DPPH• scavenging activity and
reducing power than the reference compound BHT
(Table 4). In comparison with other investigated
samples, water extracts had significantly lower red-
ucing power and DPPH• scavenging activity than the
ethanol/water extracts, which were not significantly
different in both of the tests. Thermal treatment had a
significant positive influence only on water extraction.
Also, the correlation between reducing activity and
scavenging activity on DPPH• (Table 4) was highly
positive. Furthermore, reducing activity significantly
correlated with rutin and total flavonoid contents (Table
4). Similarly, strong positive correlation was observed
between scavenging activity on DPPH• and rutin and
total flavonoid contents (Table 4).
As the process of lipid oxidation occurs both in
the high-fat food products causing rancidity and in
living organisms resulting in cell damage, three dif-
ferent lipid model systems were used to measure the
antioxidant activity of the extracts: β-carotene bleach-
ing method, antihemolytic test and Schaal test.
β-Carotene bleaching test measures the loss of
the yellow colour of β-carotene due to its reaction with
radicals that are formed by linoleic acid oxidation in
an emulsion. Antioxidant activity of an extract in this
case refers to its ability to protect β-carotene from
oxidative damage. The test results (Table 3) indicated
the following order in antioxidant activities: boiling
ethanol/water (50:50) = ethanol/water (50:50) = boil-
ing ethanol/water (80:20) > ethanol/water (80:20) >
water extracts. The extracts with the highest activity
Table 3. Antioxidant activity and total flavonoid content of dried buckwheat leaf and flower (DBLF) extracts: A-water; B-boiling water; C-
ethanol/water (50:50); D-boiling ethanol/water (50:50); E-ethanol/water (80:20); F-boiling ethanol/water (80:20) and the control sub-
stances; values are means of three determinations ± standard deviation. Values in each column with the same superscript a,b,c,d are not
significantly different (P < 0.05)
Extract/control
substance
IC50 in mg plant material/mL or mg control substance/mL Total flavonoid content
g/100g plant material DPPH• scavenging
activity
Chelating activity
on Fe2+
Reducing
power
β-Carotene
bleaching method
Antihemolytic
activity
A 3.79±0.177c 272.8±31.1d 15.10±2.43d 5.04±0.083c 42.1±7.27d 0.22±0.026a
B 2.78±0.044b 64.8±0.961c 12.46±1.89c 4.76±0.400c 29.4±9.04c 0.72±0.035b
C 0.800±0.074a 34.4±0.559b 3.19±0.135b 1.88±0.142a 3.43±0.207b 2.57±0.031d
D 0.679±0.018a 25.7±1.01b 2.81±0.093b 1.87±0.054a 3.12±0.370b 4.16±0.032e
E 0.816±0.025a 39.8±0.240b 3.40±0.342b 3.82±0.197b 5.53±0.339b 2.42±0.021c
F 0.672±0.009a 30.6±0.580b 2.66±0.040b 1.78±0.003a 2.30±0.377b 4.92±0.026f
BHT 0.560±0.001a – 0.360±0.010a 1.78±0.060a – –
Ascorbic acid – – – – 0.18±0.010a –
EDTA – 0.039±0.001a –– – – –
Table 4. Correlation coefficients between IC50 values of antioxidant activity and phenolic compounds (total flavonoid and rutin) content
(P < 0.05)
Parameter Chelating activity on
Fe2+
Antihemolytic
activity
β-Carotene
bleaching method
Reducing
power
DPPH• scavenging
activity
Rutin content Not significant - 0.963 - 0.924 -0.9587 -0.974
Total flavonoid content Not significant - 0.874 - 0.905 -0.871 -0.879
DPPH• scavenging activity 0.871 0.999 0.867 0.996 -
Reducing power 0.823 0.995 0.880 - -
β-carotene bleaching method Not significant 0.881 - - -
Antihemolytic activity 0.871 - - - -
A. MIŠAN et al.: PHENOLIC PROFILE AND ANTIOXIDANT PROPERTIES… Chem. Ind. Chem. Eng. Q. 23 (1) 3947 (2017)
45
were as efficient as BHT. Significant correlation was
found between these results and rutin content and
total flavonoid content (Table 4). Also, these results
significantly correlated with scavenging activity on
DPPH• and the results of reducing power (Table 4).
Antihemolytic activity assay shows the ability of
the tested extracts to inhibit hydrogen-peroxide ind-
uced oxidation of the lipids in the phospholipid bilayer
of erythrocyte membranes and has the advantage of
using biological system instead of simpler membrane
models like phosphatidylcholine liposomes [26].
Although much higher than the IC50 of ascorbic acid,
calculated IC50 values of the investigated extracts
(Table 3) significantly correlated with the total flavo-
noid content and even better with the rutin content
(Table 4) of the samples, which indicates that flavo-
noids play a crucial role in preventing oxidative dam-
age to erythrocyte cell membranes in vitro. Also, the
highly significant correlation of antihemolytic activity
with scavenging activity on DPPH• and reducing
power (Table 4) of the extracts implies the major
mechanisms of antioxidant protection under the given
experimental conditions.
In the third model system, which could be rel-
evant to lipid oxidation processes in high-fat food pro-
ducts, sunflower oil was used as a lipid substrate. The
courses of oxidation were followed gravimetrically
and demonstrated by using derivative plots (Figure 1).
The inflection points obtained in this manner corres-
ponded to the maximum change in weight gain over
time and are convenient for the induction period det-
ermination. As it can be seen from the inflection
points in Figure 1, the oxidation process of the sun-
flower oil (control) began on the fifth day. With the
addition of DBLF ethanolic extracts, the induction
period in vegetable oil was prolonged. After BHT,
which showed the highest efficiency in delaying oxid-
ative changes (induction period of 8 days), boiling
ethanol/water (80:20) extract showed the highest acti-
vity prolonging the induction period of sunflower oil for
7.5 days. The ethanol/water (80:20) extract was cap-
able of prolonging the beginning of the oxidation
process until 7th day. However, boiling ethanol/water
(50:50) and ethanol/water (50:50) extracts were able
to postpone the oxidation process only for 5 and 5.5
days, respectively. Water extracts were not tested as
they had been previously demonstrated to be less
efficient by other tests (Table 3).
The ability of an antioxidant to chelate/deact-
ivate transition metal ions, which can catalyze hydro-
peroxide decomposition and Fenton-type reactions, is
considered an important mechanism of AOA, and
therefore the chelating activity test was used to eval-
uate the chelating efficiency of investigated extracts.
Obtained IC50 values indicate that investigated ext-
racts possess significant chelating activities and may
be able to play a protective role against oxidative
damage by sequestering Fe2+, but they showed much
lower chelating activity than the reference compound
EDTA. The water extracts were shown to be less effi-
cient than the ethanol/water extracts, which were not
statistically different (Table 3). The correlation
between Fe2+ chelating activity rutin and total flavo-
noid content was positive, but not statistically signific-
ant, which indicates that the other compounds pre-
sent in the extracts, along with flavonoids contribute
to the chelating activity. Chlorogenic acid was rep-
orted to be a powerful chelator of iron ions, capable of
forming iron/chlorogenic acid complex, which was not
capable of generating the hydroxyl radicals in Fenton
model system [27]. However, the correlation between
chlorogenic acid content and chelating activity of the
Figure 1. Antioxidant effectiveness of dried buckwheat leaf and flower (DBLF) ethanolic extracts tested by Schaal oven test. Changes in
the mass per day (%/day) of sunflower oil samples supplemented with: A–boiling ethanol/water (80:20) DBLF extract; B–ethanol/water
(80:20) DBLF extract; C–boiling ethanol/water (50:50) DBLF extract; D–ethanol/water (50:50) DBLF extract; E–control;
F–0.01% BHT during accelerated oxidation.
A. MIŠAN et al.: PHENOLIC PROFILE AND ANTIOXIDANT PROPERTIES… Chem. Ind. Chem. Eng. Q. 23 (1) 3947 (2017)
46
extracts was not established. On the contrary, the
results of chelating activity significantly correlated
with those of scavenging activity on DPPH• and red-
ucing power (Table 4). Unlike the poor correlation
with β-carotene bleaching test, the results of anti-
hemolytic activity assay significantly correlated with
those of chelating activity on Fe2+ (Table 4). Contrary
to the β-carotene bleaching test, the antihemolytic
activity assay represents an iron ion-dependent lipid
peroxidation system, where a potential of an antioxid-
ant to act as a chelator becomes important [28].
CONCLUSION
DBLF was shown to be a rich source of anti-
oxidants. Extraction with ethanol/water mixture (80:20
V/V) for 24 h at room temperature, after the mixture
was just brought to boil was demonstrated to be an
efficient and cheap way for the extraction of antioxid-
ants.
Flavonoids, primarily rutin, were shown to be the
most responsible for the antioxidant activity in all
investigated lipid model systems, acting as free rad-
ical scavengers, electron-donating substances and
iron ion chelators. In the β-carotene bleaching test,
the extracts with the highest activity were as efficient
as BHT. Furthermore, the extracts were able to pro-
tect erythrocyte cell membranes from oxidative dam-
age and to prolong the induction period of sunflower
oil for 3.5 days under the accelerated oxidation con-
ditions.
Finally, the DBLF extracts obtained by this rel-
atively cheap and environmentally friendly extraction
procedure show a very good potential for incorpor-
ation in certain fat-containing food products as a rich
source of natural antioxidants, which, beside their
already proven health benefits, would provide a sub-
stantial protection against oxidative-induced rancidity.
However, in order to fully optimize the extraction
method, further research, which would include testing
of ratio of liquid to raw material, extraction time, the
ultrasonic power and radiation time on the extraction
efficiency needs to be done.
Acknowledgment
This work is a part of the National Project (TR-
-31029) financially supported by the Ministry of
Education, Science and Technological Development,
Republic of Serbia.
REFERENCES
[1] I. Kreft, N. Fabjan, K. Yasumoto, Food Chem. 98 (2006)
508-512
[2] I. Sedej, A. Mandić, M. Sakač, A. Mišan, V. Tumbas,
Cereal Chem. 87 (2010) 387-392
[3] I. Sedej, M. Sakač, A. Mandić, A. Mišan, M. Pestorić, O.
Šimurina, J. Čanadanović-Brunet, LWT-Food Sci. Tech-
nol. 44 (2011) 694-699
[4] I. Sedej, M. Sakač, A. Mandić, A. Mišan, V. Tumbas, M.
Hadnađev, J. Cereal. Sci. 54 (2011) 347-353
[5] S. Li, Q.H. Zhang, Crit. Rev. Food Sci. Nutr. 41 (2001)
451-464
[6] E. Margitanová, A. Vollmannová, L. Krížová, I. Čičová, J.
Microbiol. Biotech. Food Sci. 1 (2012) 949-956
[7] H. Jaberian, K. Piri, J. Nazari, Food Chem. 136 (2013)
237-244
[8] A. Mišan, N. Mimica-Dukić, M. Sakač, A. Mandić, I.
Sedej, O. Šimurina, V. Tumbas, J. Food Sci. 76 (2011)
C1239-C1244
[9] N. Ihme, H. Kiesewetter, F. Jung, K.H. Hoffmann, A. Birk,
A. Müller, K.I. Grützner, Eur. J. Clin. Pharm. 50 (1996)
443-447
[10] M.Y. Abeywardena, R.J. Head, Prostaglandins, Leuko-
trienes Essential Fatty Acids 65 (2001) 91-97
[11] J.-S. Lee, S.-H. Bok, S.-M. Jeon, H.-J. Kim, K.-M. Do,
Y.-B. Park, M.-S. Choi, Food Chem. 119 (2010) 235-240
[12] I. Hinneburg, R.H. Neubert, J. Agric. Food Chem. 53
(2005) 3-7
[13] C.C. Chang, M.H. Yang, H.M. Wen, J.C. Chern, J. Food
Drug Anal. 10 (2002) 178-182
[14] A. Mišan, N. Mimica-Dukić, A. Mandić, M. Sakač, I. Milo-
vanović, I. Sedej, Cent. Eur. J. Chem. 9 (2011) 133-142
[15] T. Hatano, H. Kagawa, T. Yasuhara, T. Okuda, Chem.
Pharm. Bull. 36 (1988) 2090-2097
[16] E. Decker, A.B. Welch, J. Agric. Food. Chem. 38 (1990)
674-677
[17] M. Oyaizu, Jpn. J. Nutr. 44 (1986) 307-315
[18] M.S. Al-Saikhan, L.R. Howard, J.C.Jr. Miller, J. Food Sci.
60 (1995) 341-347
[19] F.N. Ko, G. Hsiao, Y.H. Kuo, Free Radical Biol. Med. 22
(1997) 215-222
[20] B. Šarić, I. Milovanović, A. Mišan, A. Mandić, J. Kos, P.
Jovanov, N. Nedeljković, in Proceedings of the 6th Central
European Congress on Food, Cefood 2012, Novi Sad,
Serbia, 2012, pp. 172-176
[21] O.R. Fennema, Principles of food science. Part I. Food
Chemistry, Marcel Dekker Inc, 1976
[22] SciDAVis, 0.2.4 (Scientific Data Analysis and Visualiz-
ation), available from: http://scidavis.sourceforge.net/ (20
September 2012)
[23] StatSoft, Inc. 2011, Statistica (data analysis software sys-
tem), version 10. http://www.statsoft.com (30 November
2011)
[24] K.H. Kim, K.W. Lee, D.Y. Kim, H.H. Park, I.B. Kwon, H.J.
Lee, Bioresour. Technol. 96 (2005) 1709-1712
[25] P. Sharma, A.K. Ghimeray, A. Gurung, C.W. Jin, H.S.
Rho, D.H. Cho, Afr. J. Biotechnol. 11 (2012) 184-190
[26] S. Chaudhuri, A. Banerjee, K. Basu, B. Sengupta, P.K.
Sengupta, Int. J. Biol. Macromol. 41 (2007) 42-48
A. MIŠAN et al.: PHENOLIC PROFILE AND ANTIOXIDANT PROPERTIES… Chem. Ind. Chem. Eng. Q. 23 (1) 3947 (2017)
47
[27] Y. Kono, S. Kashine, T. Yoneyama, Y. Sakamoto, Y. Mat-
sui, H. Shibata, Biosci. Biotechnol. Biochem. 62 (1998)
22-27
[28] M. Sakač, J. Čanadanović-Brunet, A. Mišan, V. Tumbas,
Đ. Medić, Food Technol. Biotechnol. 48 (2010) 524-529.
ALEKSANDRA MIŠAN1
BOJANA ŠARIĆ1
IVAN MILOVANOVIĆ1
PAVLE JOVANOV1
IVANA SEDEJ1
VANJA TADIĆ2
ANAMARIJA MANDIĆ1
MARIJANA SAKAČ1
1Institut za prehrambene tehnologije u
Novom Sadu (FINS), Univerzitet u
Novom Sadu, Bulevar cara Lazara 1,
21000 Novi Sad, Srbija 2Institut za proučavanje lekovitog bilja
„Dr Josif Pančić“, Tadeuša Košćuška 1,
11 000 Beograd, Srbija
NAUČNI RAD
FENOLNI PROFIL I ANTIOKSIDANTNA SVOJSTVA EKSTRAKATA OSUŠENIH LISTOVA I CVETOVA HELJDE
Formulacije na bazi osušenih listova i cvetova heljde (DBLF) su dokazano efikasne u
tretiranju bolesti krvnih sudova, a što se dovodi u vezu sa visokim sadržajem rutina
(2-10%). U cilju pronalaženja ekonomičnog načina ekstrakcije antioksidanata testirani su
uticaj odnosa etanola i vode, kao i temperature na efikasnost ekstrakcije fenolnih jedinjenja
i mehanizme antioksidantnog delovanja ekstrakata. Ekstrakcija smešom etanol/voda
(zapreminski odnos 80:20) tokom 24 h na sobnoj temperaturi, nakon što je smeša zagre-
jana do ključanja, pokazala se kao efikasan i ekonomičan način za dobijanje visokog sadr-
žaja rutina (49,94±0,623 mg/g DBLF). Najzastupljenija fenolna jedinjenja u DBLF ekstrak-
tima bila su rutin i hlorogenska kiselina. Pokazano je da su flavonoidi, a posebno rutin,
najodgovorniji za antioksidantnu aktivnost ekstrakata u lipidnim model sistemima, delujući
kao “hvatači” slobodnih radikala, elektron-donorske supstance i helatori jona gvožđa. U
testu ispitivanja antioksidantne aktivnosti β-karoten metodom najpotentniji ekstrakti bili su
jednako efikasni kao i BHT (butilovani hidroksitoluen). Rezultati antihemolitičkog i Schaal
testa ukazali su na izuzetnu sposobnost ekstrakata da inhibiraju oksidativnu razgradnju
eritrocita i produže period do početka oksidativnih promena u suncokretovom ulju.
Ključne reči: osušeni listovi i cvetovi heljde; rutin; antioksidantna aktivnost; oksi-
dacija lipida; ekstrakcija.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 4956 (2017) CI&CEQ
49
YAJING ZHANG1,3
YU ZHANG1
FU DING1,3
KANGJUN WANG1,3
XIAOLEI WANG2
BAOJIN REN1
JING WU1
1College of Chemical Engineering,
Shenyang University of Chemical
Technology, Shenyang, China 2School of Science, Shenyang
University of Technology,
Shenyang, China 3Liaoning Co-innovation Center of
Fine Chemical Industry, Shenyang,
China
SCIENTIFIC PAPER
UDC 66.094.25:546.264–31:544.47
https://doi.org/10.2298/CICEQ150711005Z
SYNTHESIS OF DME BY CO2 HYDROGEN-ATION OVER La2O3-MODIFIED CuO–ZnO–ZrO2/HZSM-5 CATALYSTS
Article Highlights
• La-modified CuO-ZnO-ZrO2/HZSM-5 catalysts were prepared by an oxalate co-precipit-
ation method
• La-modified CuO-ZnO-ZrO2/HZSM-5 catalysts show higher catalytic performances
• The catalytic performances of the catalysts are strongly dependent on the La content
Abstract
A series of La2O3-modified CuO-ZnO-ZrO2/HZSM-5 catalysts were prepared by
an oxalate co-precipitation method. The catalysts were fully characterized by
X-ray diffraction (XRD), N2 adsorption-desorption, hydrogen temperature pro-
grammed reduction (H2-TPR), ammonia temperature programmed desorption
(NH3-TPD), and X-ray photoelectron spectroscopy (XPS) techniques. The effect
of the La2O3 content on the structure and performance of the catalysts was
thoroughly investigated. The catalysts were evaluated for the direct synthesis of
dimethyl ether (DME) from CO2 hydrogenation. The results displayed that La2O3
addition enhanced catalytic performance, and the maximal CO2 conversion
(34.3%) and DME selectivity (57.3%) were obtained over the catalyst with 1%
La2O3, which due to the smaller size of Cu species and a larger ratio of Cu+/Cu.
Keywords: CO2 hydrogenation; dimethyl ether; La2O3 promoter.
Carbon dioxide emission has caused irrever-
sible climate changes due to its greenhouse effect [1].
CO2 capture and storage is one of efficient method for
reducing the CO2 emissions. However, this techno-
logy requires high cost [2]. Chemical recycling of CO2
is another main option, which is presented as an eco-
nomically attractive and sustainable method. CO2 can
be converted into methanol, dimethyl ether (DME),
methane and syngas (CO+H2) [3], carboxylic acids,
etc. Among them, synthesis of DME has been paid
special attention since DME can be applied as a clean
fuel, coolant, propellant, and is an important chemical
intermediate [4]. The chemical reactions occurring in
direct conversion of CO2 to DME can be described by
the following equations [5]:
Correspondence: F Ding, K. Wang, College of Chemical Eng-
ineering, Shenyang University of Chemical Technology, Shen-
yang 110142, PR China. E-mail: F. Ding, [email protected];
K. Wang, [email protected] Paper received: 11 July, 2015 Paper revised: 8 January, 2016 Paper accepted: 19 February, 2016
H2 2 3 2CO 3H CH OH H O, 49.4kJ/ mol (1)
H3 3 3 22CH OH CH OCH H O, 23.4 kJ / mol (2)
H2 2 2CO H CO H O, 41.2 kJ/ mol (3)
For the methanol synthesis (reaction (1)), CuO–
–ZnO–ZrO2 catalytic system is considered as more
favorable than the traditional CuO–ZnO–Al2O3, because
the interaction of Cu metal particles with ZnO and
ZrO2 lead to the stabilization of a “mix” of Cu and Cuδ+
(not Cu2+, Cu1+ and Cu) [6]. For the methanol dehyd-
ration process (reaction (3)), solid-acid catalysts
(HZSM-5, γ-Al2O3 and sulfated zirconia) are employed
[7]. HZSM-5 is widely applied for methanol dehyd-
ration because: 1) it has very high catalytic activity at
the optimum reaction temperature; 2) it is more res-
istant toward poisoning of acid sites by the water due
to more hydrophobic character; 3) predominance of
Brønsted-type acidity, which can promote the DME
yield [8]. However, H-ZSM-5 presents the disadvant-
ages of narrow pore size and strong acid sites, which
Y. ZHANG et al.: SYNTHESIS OF DME BY CO2 HYDROGENATION… Chem. Ind. Chem. Eng. Q. 23 (1) 4956 (2017)
50
will limit reactant molecules diffusion and lead to
formation of secondary products, respectively. Witoon
et al. developed sulfated zirconia catalyst with 20%
sulfur-loaded on ZrO2 as methanol dehydration cat-
alyst, the yield of methanol and DME over CuO–ZnO–
–ZrO2/20S-ZrO2 both higher than those over CuO–
-ZnO–ZrO2/H-ZSM-5, but the stability of the former
catalyst is a little poor [9].
Up to now, the conversion of CO2 and the sel-
ectivity of DME are still not high. To enhance the cat-
alytic activity, efforts have been addressed by adding
promoters for methanol synthesis catalyst, besides
developing new preparation methods [10-14]. Our
previous results showed that V and Mn oxides mod-
ified CuO–ZnO–ZrO2/HZSM-5 exhibited higher cat-
alytic performance [15-16]. La2O3, as a rare earth
metal oxide, is considered as owing some basic char-
acter [17], and it can promote many metal oxide cat-
alytic reactions. Guo et al. [18] investigated catalysis
performance of La doping Cu/ZrO2 for CO2 hydro-
genation to methanol, and they found the amount of
basic site increases with La loading and the presence
of La enhance the selectivity of methanol. Gao et al.
reported appropriate amount of La can decrease the
crystallite size of CuO and enhance the dispersion of
Cu [19]. Sun’s group has investigated the effect of La
on the performance of Cu/Zn/Al catalysts via hydro-
talcite-like precursors for CO2 hydrogenation to meth-
anol, and the results showed La addition not only led
to higher BET specific surface area and Cu disper-
sion, but also increased the total number of basic
sites and proportion of strongly basic sites [20].
Furthermore, Sun’s group also reported that addition
of Zr to La-Cu-Zn-O with perovskite structure catalyst,
for synthesis of methanol from CO2 hydrogenation,
will lead to smaller particles, lower reduction tempe-
rature, higher Cu dispersion, larger amount of hydro-
gen desorption at low temperature as well as higher
concentration of basic sites are obtained [21]. How-
ever, the introduction of La2O3 into the CuO–ZnO–
-ZrO2/HZSM-5 catalyst for CO2 to DME has not been
reported.
Herein, La2O3-modified CuO-ZnO-ZrO2/HZSM-5
catalysts with a different La2O3 content were pre-
pared, aiming at investigating the effect of La2O3 mod-
ification on the structure and performance the cat-
alysts.
EXPERIMENTAL
Catalyst preparation
The La2O3 modified CuO-ZnO-ZrO2/HZSM-5
(CuO:ZnO:ZrO2 mass ratio: 5:4:0.2) catalysts, abbre-
viated as CZZLxH where x stands for theoretical
La2O3/CZZ wt.%, were prepared by an oxalate co-pre-
cipitation method (CZZLx/HZSM-5 mass ratio: 2:1),
HZSM-5 (SiO2/Al2O3 mole ratio: 50) was purchased
from Catalyst Plant of Nankai University (China). The
preparation method is same as described in other
work [15]. In brief, first, metal nitrates were dissolved
into a certain amount of ethanol (denoted as solution
A); H2C2O42H2O (200 mol.% of metal nitrate) was
also dissolved into ethanol (solution B). Then, sol-
utions A and B were slowly dropped into a beaker
containing a HZSM-5 suspension in ethanol suspen-
sion kept under stirring at 333 K. The suspension was
sealed and aged for 2 h and then the ethanol was
evaporated at 353 K to get a precipitate. Finally, the
precipitate was dried at 393 K for 12 h and calcined in
air at 673 K for 4 h.
Catalyst testing
Catalytic performance was evaluated in a con-
tinuous-flow fixed-bed reactor made of stainless steel
with inside diameter of 0.01 m. First, the catalyst was
reduced with 10% H2/N2 at 573 K for 3 h under atmo-
spheric pressure. Then it was cooled to 653 K and
reactant gas flow was introduced, raising the pressure
to 3.0 MPa, the reaction temperature was 543 K. The
exit line was heated to prevent condensation. The
products were analyzed on line with a gas chroma-
tograph (SP2100A) equipped with both a TCD (for CO
and CO2, GDX-101 connected with Porapak T col-
umn) and a FID (for CH4, CH3OH and CH3OCH3,
Porapak Q column). Conversion and selectivity
values were calculated by internal standard method
[22]. XCO2, Spi and YDME represent the conversion of
CO2, the selectivity of the product (DME, MeOH and
CO) and the yield of DME, respectively. Each expe-
rimental data was corresponds to an average of three
independent measurements, with error of ±2%.
2
2,in 2,out
2,in 2,outCO
2,in
2,in
CO CON N
CON
X (4)
iPi
2,out1 CO
PS (5)
where Pi stands for the concentration of a specific i
product (DME, MeOH orCO):
2DME DME COY S X (6)
Catalyst characterization
XRD measurements were performed on a Rig-
aku D/max 2500pc X-ray diffractometer with Cu-Kα
Y. ZHANG et al.: SYNTHESIS OF DME BY CO2 HYDROGENATION… Chem. Ind. Chem. Eng. Q. 23 (1) 4956 (2017)
51
radiation ( = 1.54156 Å) at a scan rate of 4 min-1 at
50 kV and 250 mA. For the reduced catalysts were
first reduced with 10% H2/N2 at 573 K for 3 h, and
cooled to room temperature under the N2 flow, and
then keep it into a bottle full of N2 and send it to XRD
chamber immediately. The crystallite size was calcul-
ated using the Scherrer equation. BET surface areas
were measured by N2 adsorption at 77 K using a
Quantachrome Autosorb 1-C. Before measurements,
samples were degassed under vacuum at 573 K for 4
h. H2-TPR was carried out in 10% H2/Ar flowing at 50
mL min-1, using a ramp rate of 10 K min-1 to 773 K.
NH3-TPD was conducted on a Chemisorb from 373 to
873 K. XPS measurements were performed on an
ESCALAB-250. The catalysts were first reduced with
10% H2/N2 at 573 K for 3 h, and cooled to room tem-
perature under the N2 flow, then put into the chamber
of X-ray photoelectron spectrometer immediately for
measurement. The exact composition of the surface
of the catalyst was determined by XPS. Acidity on the
catalyst was measured on a NICOLET 500 FT-IR
through pyridine adsorption. The sample was pre-
pared to the load slice, which was subject to purific-
ation under vacuum pressure 0.0133 Pa at 673 K,
after cooling down to the room temperature, pyridine
was adsorbed for 0.5 h. Desorption was carried out by
programming temperature to 423 K. The infrared
spectrum was generated within the scope of 1400–
–1700 cm–1 at room temperature.
RESULTS AND DISCUSSION
Catalytic performance of catalysts
The catalytic performances of CZZLxH catalysts
with varying La2O3 content are summarized in Table
1. The major product was DME, and the side products
were methanol, CO and trace hydrocarbons. The
amount of hydrocarbons was less than 1% and there-
fore they were neglected. The CO2 conversions and
DME selectivities over the La2O3-modified catalysts
are higher than those over unmodified one (CZZL0H),
indicating that La2O3-modification can efficiently
enhance the catalytic performances. The CZZL1H
exhibited the maximum CO2 conversion and DME sel-
ectivity of 34.3 and 57.3%, respectively. CO2 conver-
sion over CZZL1H increases 18.6%, compared with
CZZL0H. The influence of the catalyst composition on
the performance does not seem distinct, which is due
to the thermodynamic regime of reaction test because
of combination of high temperature and long contact
time. If the temperature is decreased to 533 K, the
differences of conversion and DME selectivity between
CZZL1H and CZZL0H become much larger (Table 2).
It is also noted that the selectivity of CO is high (about
30%), which may due to the higher reaction tempe-
rature of 543 K. CO was produced by reverse water-
gas reaction, the reaction has endothermic character,
as shown in reaction (2). Meanwhile, compared to
methanol synthesis, the RWGS reaction has a higher
apparent activation energy [23], indicating that the
increase in CO production is faster than that of meth-
anol with higher reaction temperature. The selectivity
of CO over CZZL1H exhibited the minimal value, indi-
cating that a certain amount of La2O3 can inhibit the
RWGS reaction. However, as Gao et al. pointed out,
further work needs to be carried out to investigate
how La2O3 changes the RWGS reaction [19].
In order to further increase the DME yield, the
catalytic performance of the CZZL1H catalyst was
also investigated at lower gas hourly space velocity.
As shown in Table 2, when the GHSV was as low as
1800 h-1, the CO2 conversion and DME selectivity
increase to 36.4 and 58.2%, respectively. The inc-
rease of the conversion can be attributed to the longer
contact time of CO2 and H2 over the catalyst at lower
GHSV. Similarly, methanol dehydration can proceed
to a higher degree with increasing contact time, lead-
ing to higher DME selectivity. In addition, a contrast
experiment was carried out at a reaction pressure of 5
MPa while keeping other conditions constant. DME
selectivity increased by approximately 7.3% and CO
selectivity decreased by 24.6% compared with those
obtained at 3 MPa. This result suggests that an inc-
rease of reaction pressure can improve the catalytic
performance.
The structure of catalyst
Figure 1A shows the XRD patterns of CZZLxH
catalysts. The peaks appearing at 35.5, 38.7, 48.7,
Table 1. Catalytic performances of the catalysts; reaction conditions: T = 543 K; p = 3.0 MPa; CO2:H2 = 1:3; GHSV = 4200·h-1
Catalyst Conversion of CO2, % Selectivity, %
DME Yield, % DME Productivity, g g cat-1 h-1 DME CH3OH CO
CZZL0H 28.9 55.1 12.9 32.0 15.9 0.171
CZZL0.5H 30.4 55.2 12.4 31.7 16.8 0.181
CZZL1H 34.3 57.3 13.3 29.4 19.6 0.212
CZZL2H 31.1 53.6 12.7 33.7 16.7 0.180
Y. ZHANG et al.: SYNTHESIS OF DME BY CO2 HYDROGENATION… Chem. Ind. Chem. Eng. Q. 23 (1) 4956 (2017)
52
and 66.2 can be ascribed to CuO phase (tenorite,
JCPDS 48-1548), the peaks at 31.7, 34.3, 36.6, 56.5,
62.8 and 67.8 can be indexed to ZnO phase (JCPDS
65-3411), and the peaks appearing in the 2 range of
21-25 belong to HZSM-5 (JCPDS 44-0003). No
peaks belonging to ZrO2 or La2O3 are observed, indi-
cating that ZrO2 and La2O3 are amorphous or well
dispersed in catalyst body. The intensities of the
peaks assigned to CuO and ZnO weakened and the
widths of the peaks broadened gradually with inc-
reasing La2O3 content from 0 to 1%, then they
became sharper and narrower again when La2O3
content was 2%. This result indicates that the addition
of a proper amount of La2O3 can enhance the dis-
persion of CuO and ZnO [24]. This changing trend
can be well reflected in the changing trends of Cu and
CuO grain size, as shown in Table 3.
Table 3. Physicochemical properties of the catalysts; diffraction
peak at 2θ 38.7 for CuO and 43.3 for Cu
Catalyst SBET / m2 g–1 DXRD / nm
CuO Cu
CZZL0H 135.4 11.3 13.5
CZZL0.5H 136.7 9.6 11.9
CZZL1H 138.2 8.4 10.5
CZZL2H 135.8 9.3 11.4
The XRD patterns of the reduced catalysts are
shown in Figure 1B. The diffraction peaks at 2θ
values of 43.3, 50.4 and 74.1 can be indexed to the
crystal planes of (111), (200) and (220) of metallic
copper phase, respectively (JCPDS 04-0836). No dif-
fraction peaks belonging to the CuO phase could be
detected, suggesting all CuO species had been red-
uced to copper. The intensities of the peaks assigned
to Cu and ZnO changed in a similar trend as those of
CuO and ZnO (Figure 1A).
The specific surface areas and the calculated
crystallite sizes of the catalysts using Scherrer’s
equation are listed in Table 3. The SBET increased
from 135.4 of CZZL0H to 138.2 m2 g-1 of CZZL1H, and
then decreased to 135.8 m2 g-1 for CZZL2H. The
changing trend of the DCu is opposite to the trend of
the SBET, a minimum of 10.5 nm is obtained over
CZZL1H. Although the SBET change is not as signific-
ant as DCu, the trend is in accordance with XRD
results. Combining the results in both Tables 1 and 3,
it can be observed that CZZL1H shows the smallest
Cu crystallite size and the best catalytic performance,
which indicates the activity is closely related to the
crystallite size of Cu. Guo et al. also reported the cat-
alytic performance of Cu-TiO2-ZrO2 related to the
crystallite size of CuO [25].
Table 2. Catalytic performances of the catalysts at different reaction conditions; reaction conditions: CO2:H2 = 1:3
Catalyst T
K
Conversion of CO2
%
p
MPa
GHSV
h-1
Selectivity, % DME Yield
%
DME Productivity
g g cat-1 h-1 DME CH3OH CO
CZZL1H 543 34.3 3 4200 57.3 13.3 29.4 19.6 0.212
CZZL1H 543 36.4 3 1800 58.2 14.2 27.6 21.1 0.098
CZZL1H 543 38.5 5 1800 62.5 16.7 20.8 24.0 0.111
CZZL0H 533 24.2 3 4200 57.1 13.7 29.2 13.8 0.148
CZZL1H 533 30.4 3 4200 61.8 14.2 24.0 18.8 0.203
20 30 40 50 60 70 80
d
b
c
Inte
nsi
ties
(a. u
.)
2Theta (deg)
a
CuO
ZnO
HZSM-5
A
20 40 60 80
B
**
*
Cu*
ZnOHZSM-5
c
b
d
a
Inte
nsi
ties
(a.
u.)
2 Theta (deg)
Figure 1. A) XRD patterns of La2O3-modified CuO–ZnO–ZrO2/HZSM-5 and B) reduced catalysts: a) CZZL0H;
b) CZZL0.5H; c) CZZL1H; d) CZZL2H.
Y. ZHANG et al.: SYNTHESIS OF DME BY CO2 HYDROGENATION… Chem. Ind. Chem. Eng. Q. 23 (1) 4956 (2017)
53
The reducibility of catalyst
H2-TPR results are depicted in Figure 2. It is
obvious that La2O3 modification displays a significant
effect on the interaction between the metal and the
carriers. Each catalyst exhibits a broad peak with
peak maximum in 423-563 K, corresponding to the
reduction of CuO to Cu [26]. For the catalysts with
increasing La2O3 content from 0 to 0.5 or 1%, the
peaks maxima shift to higher temperature from 529 to
554 or 556 K while no shape changes are observed,
which indicates the interaction between ZnO and CuO
or La2O3 and CuO becomes stronger [19,23,27]. This
result seems conflicting with XRD result. According to
XRD result, the dispersion of CuO becomes better
and the crystalline size becomes gradually smaller
with La2O3 addition amount from 0 to 1%, suggesting
that the reducibility of CuO should become easier.
This fact implies La2O3 modification still plays another
role. For CZZL2H catalyst, it is evident that the H2-
-TPR change tendency becomes more different and
the peak maximum shifts towards the lower tempe-
rature, indicating the reducibility of CuO becomes
easier [28,29]. It is concluded that La2O3 modification
has at least two roles. On one hand, La modification
promotes the dispersion of CuO, leading to easier
reduction of CuO; on the other hand, La2O3 modific-
ation enhances the interaction between CuO and
other metal oxides resulting in a more difficult red-
uction process. The two effects compete with each
other and the reduction temperature is dependent on
which effect is predominant. Xiong et al. also found
that the La2O3 can increase the dispersion of Co/AC
catalysts, whereas the reduction temperature shifted
to a higher position due to stronger interaction [30].
350 400 450 500 550 600 650
a
b
c
Co
nsu
mp
tio
n o
f H
2 (
a.u
.)
Temperature (K)
d
Figure 2. H2-TPR profiles of catalysts: a) CZZL0H; b) CZZL0.5H;
c) CZZL1H; d) CZZL2H.
Surface acidity of catalyst
Figure 3 shows the NH3-TPD results obtained
for pure HZSM-5 and CZZLxH catalysts. The total
acidic amount, the strength and the fraction of various
acid sites are summarized (supporting info, available
from the authors upon request). On pure HZSM-5 pro-
file, two NH3 desorption peaks are observed, indicat-
ing the existence of at least two different acid
strengths. In general, the peak located in 393-523 K
and 573-773 K can be attributed to weak and strong
acid strengths, respectively [31]. But for all catalysts,
three NH3 desorption peaks, in the temperature
regions of 373-473 K, 473-573 K and 573-673 K are
observed, denoted as α, β and γ peak, which can be
assigned to weak, medium and strong acid strengths
of HZSM-5, respectively [32]. This result is similar
with that of the reported V-modified catalyst [15].
According to the study, the reason why the catalysts
showed another more NH3 desorption peak than pure
HZSM-5 is mainly ascribed to the fact that strong acid
strength on pure HZSM-5 are blocked and modified
by metal oxides and oxalic acid, respectively. Com-
pared with La2O3-free catalyst, peaks α and β on the
other curves shift a little to higher temperature with
the increasing La2O3 content, implying that the weak
and medium acid strengths become stronger; on the
contrary, the peak γ shifts a little towards lower tem-
perature, indicating strong acid strength becomes
weaker. It is also worth noting that the total acid
amount decreases with the increasing amount of
La2O3 from 0.5 to 1%, which can be explained by the
basic character of La2O3. Sugi et al. also observed
this effect and suggested that La2O3-modification
resulted in reducing the support acidity [33]. However,
continuous addition of La2O3 to 2% increases the total
400 500 600 700
427K
421K
597K
500 K
599K499 K
656K
461 K
e
d
c
b
a
Am
mo
nia
des
orp
tio
n/a
.u
Temperatue (K)
Temperature, K
Figure 3. NH3-TPD profiles of pure HZSM-5 and the catalysts:
a) HZSM-5; b) CZZL0H; c) CZZL0.5H; d) CZZL1H; e) CZZL2H.
Y. ZHANG et al.: SYNTHESIS OF DME BY CO2 HYDROGENATION… Chem. Ind. Chem. Eng. Q. 23 (1) 4956 (2017)
54
acid amount again. Previously published studies pro-
posed that there are two reasons for the acidity
amount increase [34]. One is that the La3+ has some
Lewis acidic property originated from an empty f orb-
ital; the other reason is that Si-OH and Al-OH in the
zeolite framework are polarized by La3+, which result
in a stronger acidity. Therefore, the acidity of the
CZZLxH catalysts is dependent on the La2O3 content.
Moreover, La2O3 modification can also affect the
concentration and distribution of the three acid
strengths of HZSM-5 in the catalysts. With the inc-
reasing amount of La2O3, the concentration and fract-
ion of medium acid strength increase and those of
strong acid strength become smaller as compared to
those on CZZL0H, but the difference is too small
(supporting info). It is generally considered that the
strong acidic strength on HZSM-5 zeolite promotes
the generation of secondary products like hydrocar-
bons, resulting in low selectivity to DME [6]. However,
here the change of the acidity could not be a factor
accounting for the higher DME selectivity because the
methanol selectivity of the various catalysts was
almost the same, as shown in Table 1. This result
suggests that the acidity of the catalysts is strong
enough to efficiently convert the produced methanol
to DME. In addition, the acidity of Brønsted acid sites
and Lewis acid sites for CZZL0H and CZZL1H cat-
alysts were determined (supporting info, available
from the authors upon request). The results show the
acidity of both type acid sites of CZZL1H decreased
compared CZZL0H, thus it can be inferred that there
is no direct relationship between the selectivity of
DME and the change of acid type of the two catalysts.
So, the improvement of DME selectivity is due to the
decrease in CO selectivity resulted from the Cu-based
catalyst but not from the acid component of the
bifunctional catalysts. In addition, according to the lit-
erature, it could be speculated that the introduction of
La2O3 into catalyst will increase the surface basicity of
catalyst, which in turn promotes the adsorption of CO2
and sequence enhances the yield for methanol, finally
DME selectivity is increased after methanol dehyd-
ration [18].
Results of XPS investigations
The reduced CZZLxH catalysts characterized by
XPS, the binding energy of Cu2p3/2, Zn2p3/2, as well
as the surface compositions of the catalysts are sum-
marized (supporting info, available from the author
upon request). For all the reduced catalysts, binding
energies (BE) of Cu2p3/2 are located at about 932.3
eV, which are the characteristic peaks of reduced
Cu+/Cu species [35]. The binding energy shifted to
higher positions with increasing amounts of La2O3,
which indicated stronger interaction between CuO
and other metal oxides carriers [36]. For the purpose
of distinguishing Cu+ from Cu species, their kinetic
energies in the XAES Cu LMM line positions were
measured (Figure 4). The Cu LMM spectra show a
broad and asymmetrical peak, implying the coexist-
ence of Cu+ and Cu in the surface of the catalysts.
Two symmetrical peaks centered at near 916.6 and
918.7 eV can be obtained by deconvolution, which
are corresponding to Cu+ and Cu species, respect-
ively [37]. Additionally, Cu+/Cu can be calculated
based on the results. Volcanic shape change trends
of the Cu+/Cu versus La addition content are obs-
erved, the CZZL1H exhibited the maximum of 0.144,
which may consequently lead to higher activity due to
the stabilization of Cu+ favoring the hydrogenation of
CO2 [38]. The binding energies (BE) of Zn2p3/2 for all
catalysts are located at about 1021.8 eV, which are
close to the characteristic peaks of ZnO species [39].
Compared to the nominal surface compositions of the
catalysts, it can be seen that the actual surface com-
position Cu/Zn decreased significantly, implying
enrichment of Zn. Similarly, La2O3 content is much
higher on the surface. It is worth noting that is La2O3
is not detectable in CZZL0.5H, possibly because the
content is too small to give enough signal.
915 918 920
d
c
b
a918.7916.6
Inte
nsi
ties
(a. u
.)
Kinetic Energy (eV)
Figure 4. Cu LMM XAES spectra of the reduced catalysts:
a) CZZL0H; b) CZZL0.5H; c) CZZL1H; d) CZZL2H.
CONCLUSIONS
La2O3 modification has great impact on the cat-
alytic performance of CuO-ZnO-ZrO2/HZSM-5 cat-
alysts for promoting direct CO2-to-DME. La2O3-mod-
ification can efficiently enhance the catalytic perform-
ance of CuO-ZnO-ZrO2/HZSM-5 catalysts. The
sample containing a nominal amount of 1% La2O3
Y. ZHANG et al.: SYNTHESIS OF DME BY CO2 HYDROGENATION… Chem. Ind. Chem. Eng. Q. 23 (1) 4956 (2017)
55
gave the maximum CO2 conversion and DME sel-
ectivity of 34.3 and 57.3%, respectively, benefiting
from smaller Cu particles and a larger Cu+/Cu ratio.
When the La2O3 content is low (from 0.5 to 1%), it can
strengthen the interaction between CuO and other
metal oxides and inhibit the reduction of CuO species;
meanwhile, the total acid amount decreases slightly
but the medium strong acid concentration and strength
increase a little. However, excess of La2O3 content,
e.g., 2%, will lead to an opposite effect. In summary,
suitable La2O3 addition can improve the catalytic per-
formance of CuO-ZnO-ZrO2/HZSM-5 for one step
CO2 to DME transformation.
Acknowledgement
The authors thank National Nature Science
Foundation of China (51301114, 21201123,
21203125 and 61403263), SRF for ROCS, SEM (No.
[2010]1174), Natural Science Foundation of Liaoning
Province (2015020649), Liaoning Educational Depart-
ment Foundation (L2013161), LNET (LJQ2013044)
for financial support.
REFERENCES
[1] P. Cox, C. Jones, Science 321 (2008)1642-1644
[2] S.C.G. Santos, S.W.M. Machado, A.M. Garrido Pedrosa,
M.J.B. Souza, J. Porous Mater. 22 (2015) 1145-1151
[3] A. García-Trenco, A. Vidal-Moya, A. Martínez, Catal.
Today 179 (2012) 43-51
[4] T.A. Semelsberger, K.C. Ott, R.L. Borup, H.L. Greene,
Appl. Catal., A: Gen. 309 (2006) 210-223
[5] G. Bonura, M. Cordaro, C. Cannilla, A. Mezzapica, L.
Spadaro, F. Arena, F. Fusteri, Catal. Today 228 (2014)
51-57
[6] T. Takeguchi, K. Yanagisawa, T. Inui, M. Inoue, Appl.
Catal., A: Gen. 192 (2000) 201–209
[7] T. Witoo, S.Bumrungsalee, M. Chareonpanich, J. Limtra-
kul, Energy Conv. Manage. 103 (2015) 886-894
[8] A. García-Trenco, A. Martínez, Appl. Catal., A: Gen. 411-
–412 (2012) 170-179
[9] T. Witoo, T. Permsirivanich, N. Kanjanasoontorn, C.
Akkaraphataworn, A. Seubsai, K. Faungnawakij, C.
Warakulwit, M. Chareonpanich, J. Limtrakul, Catal. Sci.
Technol. 5 (2015) 2347-2357
[10] L. Shi, K. Tao, R.Q. Yang, F.Z. Meng, C. Xing, N.
Tsubaki, Appl. Catal., A: Gen. 401 (2011) 46-55
[11] P. Gao, F. Li, F. K. Xiao, N. Zhao, W. Wei, L.S. Zhong,
Y.H. Sun, Catal. Today 194 (2012) 9-15
[12] F. Zha, J. Ding, Y. Chang, J.F. Ding, J.Y. Wang, J. Ma,
Ind. Eng. Chem. Res. 51 (2012) 345-352
[13] X.M. Guo, D.S. Mao, G.Z. Lu, S. Wang, G.S. Wu, J. Mol.
Catal., A: Chem. 345 (2011) 60-68
[14] J. Xiao, D.S. Mao, X.M. Guo, J. Yu, Energy Technol. 3
(2015) 32–39
[15] Y.J. Zhang, D.B. Li, Y. Zhang, Y. Cao, S.J. Zhang, K.J.
Wang, F. Ding, J. Wu, Catal. Commun. 55 (2014) 49–52
[16] Y.J. Zhang, D.B. Li, D. Jiang, Y. Zhu, Y. Wu, S. Zhang, K.
Wang, J. Mol. Catal. (China) 28 (2014) 344-350
[17] K. Wakui, K. Satoh, G. Sawada, K. Shiozawa, K. Matano,
K. Suzuki, Y. Yoshimura, T. Hayakawa, K. Murata, F.
Mizukami, Catal. Lett. 81 (2002) 83-88
[18] X.M. Guo, D.S. Mao, G.Z. Lu, S. Wang, G.S. Wu, J. Mol.
Catal., A: Chem. 345 (2011) 60-68
[19] W.G. Gao, H. Wang, Y.H. Wang, W. Guo, M.Y. Jia, J.
Rare Earths 31 (2013) 370-375
[20] P. Gao, F. Li, N. Zhao, F.K. Xiao, W. Wei, L.S. Zhong,
Y.H. Sun, Appl. Catal., A: Gen. 468 (2013) 442-452
[21] H.J. Zhan, F. Li, P. Gao, N. Zhao, F.K. Xiao, W. Wei, L.S.
Zhong, Y.H. Sun, J. Power Sources 251 (2014) 113-124
[22] G. Bonura, M. Cordaro, L. Spadaro, C. Cannilla, F.
Arena, F. Frusteri, Appl. Catal., B: Environ. 140–141
(2013) 16–24
[23] L. Li, D.S. Mao, J. Yu, X.M. Guo, J. Power Sources 279
(2015) 394-404
[24] S.G. Li, H.J. Guo, C.R. Luo, H.R. Zhang, L. Xiong, X.D.
Chen, L.L. Ma, Catal. Lett. 143 (2013) 345-355
[25] X.M. Guo, D.S. Mao, S. Wang, G.S. Wu, G.Z. Lu, Catal.
Commun. 10 (2009) 1661–1664
[26] R. Nie, H. Lei, S. Pan, L. Wang, J. Fei, Z. Hou, Fuel 96
(2012) 419-425
[27] P. Gao, F. Li, H.J. Zhan, N. Zhao, F.K. Xiao, W. Wei, L.S.
Zhong. J. Catal. 298 (2013) 51-60
[28] R.W. Liu, Z.Z. Qin, H.B. Ji, T.M. Su, Ind. Eng. Chem.
Res. 5 (2013) 16648−16655
[29] Y.W. Zhang, Y.M. Zhou, H. Liu, Y. Wang, Y. Xu, P.C. Wu,
Appl. Catal., A: Gen. 323 (2007) 202-210
[30] G.P. Jiao, Y.J. Ding, H.J. Zhu, X.M. Li, W.D. Dong, J.W.
Li, Y. Lv, J. Catal. (China) 30 (2009) 92-94
[31] C.V. Hidalgo, H. Itoh, T. Hattori, M. Niwa, Y. Murakami, J.
Catal. 45 (1984) 362-370
[32] G. Bonura, M. Cordaro, L. Spadaro, C. Cannilla, F.
Arena, F. Frusteri, Appl. Catal., B: Environ. 140-141
(2013) 16-24
[33] Y.Y. Shu, D. Ma, X.H. Bao, Y. Xu, Catal. Lett. 66 (2000)
161-167
[34] X.N. Wang, Z. Zhao, C.M. Xu, A.J. Duan, L. Zhang, G.Y.
Jiang, J. Rare Earths 25 (2007) 321-328
[35] P.H. Matter, D.J. Braden, U.S. Ozkan, J. Catal. 223
(2004) 340–351
[36] J.M. Camposmartin, A. Guerreroruiz, J.L.G. Fierro, J.
Catal. 156 (1995) 208–218
[37] K.P. Sun, W.W. Lu, F.Y. Qiu, X.L. Xu, Appl. Catal., A:
Gen. 252 (2003) 243-249
[38] D.S. Mao, W.M. Yang, J.C. Xia, B. Zhang, Q.Y. Song,
Q.L. Chen, J. Catal. 230 (2005) 140-149
[39] D.G. Cantrell, L.J. Gillie, A.F. Lee, K. Wilson, Appl.
Catal., A: Gen. 287 (2005) 183-190.
Y. ZHANG et al.: SYNTHESIS OF DME BY CO2 HYDROGENATION… Chem. Ind. Chem. Eng. Q. 23 (1) 4956 (2017)
56
YAJING ZHANG1,3
YU ZHANG1
FU DING1,3
KANGJUN WANG1,3
XIAOLEI WANG2
BAOJIN REN1
JING WU1
1College of Chemical Engineering,
Shenyang University of Chemical
Technology, Shenyang, China 2School of Science, Shenyang Univer-
sity of Technology, Shenyang, China 3Liaoning Co-innovation Center of Fine
Chemical Industry, Shenyang, China
NAUČNI RAD
SINTEZA DIMETIL ETRA HIDROGENIZACIJOM CO2 POMOĆU LA2O3-MODOFOKIVANIH CuO–ZnO–ZrO2/HZSM-5 KATALIZATORA
Oksalatnom ko-precipitacionom metodom je pripremljena serija katalizatora CuO-ZnO-
–ZrO2/HZSM-5 modifikovanih pomoću La2O3. Katalizatori su okarakterisani X-difrakcionom
metodom (XRD), N2 adsorpcijom-desorpcijom, termoprogramiranom redukcijom (H2-TPR),
termoprogramiranom desorpcijom (NH3-TPD) i fotoelektronskom spektroskopijom X-zraka
(XPS). Istražen je i uticaj sadržaja La2O3 na strukturu i performanse katalizatora. Kataliza-
tori su testirani u procesu CO2 hidrogenizacije i direktne sinteze dimetil etra (DME). Rezul-
tati pokazuju da dodatak La2O3 poboljšava performanse katalizatora, kao i da su maksi-
malna konverzija CO2 (34,3%) i DME selektivnost (57,3%) dobijeni upotrebom katalizatora
sa 1% La2O3, zbog manjih Cu čestica i veceg odnosa Cu+/Cu0.
Ključne reči: hidrogenizacija CO2, dimetil etar La2O3 promoter.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017) CI&CEQ
57
TATJANA
KALUĐEROVIĆ RADOIČIĆ1
NEVENKA
BOŠKOVIĆ-VRAGOLOVIĆ1
RADMILA GARIĆ-GRULOVIĆ2
MIHAL ĐURIŠ2
ŽELJKO GRBAVČIĆ1
1Faculty of Technology and
Metallurgy, University of Belgrade,
Belgrade, Serbia 2Intitute of Chemistry, Technology
and Metallurgy, Department for
Catalysis and Chemical
Engineering, University of
Belgrade, Belgrade, Serbia
SCIENTIFIC PAPER
UDC 66.021.1:621.798
https://doi.org/10.2298/CICEQ150506006K
FRICTION FACTOR FOR WATER FLOW THROUGH PACKED BEDS OF SPHERICAL AND NON-SPHERICAL PARTICLES
Article Highlights
• Experimental evaluation of pressure drop correlations in packed beds was conducted
• Pressure drop across beds of spherical and non-spherical particles was measured
• Spherical glass particles and quartz filtration sand were used as packing material
• Correlations in the form of Ergun equation gave the best results
• The coefficients in Ergun equation are system-specific
Abstract
The aim of this work was the experimental evaluation of different friction factor
correlations for water flow through packed beds of spherical and non-spherical
particles at ambient temperature. The experiments were performed by mea-
suring the pressure drop across the bed. Packed beds made of monosized
glass spherical particles of seven different diameters were used, as well as beds
made of 16 fractions of quartz filtration sand obtained by sieving (polydisperse
non-spherical particles). The range of bed voidages was 0.359–0.486, while the
range of bed particle Reynolds numbers was from 0.3 to 286 for spherical
particles and from 0.1 to 50 for non-spherical particles. The obtained results
were compared using a number of available literature correlations. In order to
improve the correlation results for spherical particles, a new simple equation was
proposed in the form of Ergun’s equation, with modified coefficients. The new
correlation had a mean absolute deviation between experimental and calculated
values of pressure drop of 9.04%. For non-spherical quartz filtration sand par-
ticles the best fit was obtained using Ergun’s equation, with a mean absolute
deviation of 10.36%. Surface-volume diameter (dSV) necessary for correlating the
data for filtration sand particles was calculated based on correlations for
dV = f(dm) and = f(dm).
Keywords: pressure drop, packed bed, spherical particles, quartz filtra-tion sand, non-spherical particles.
Packed beds of particles permit a widespread
means of contact between fluid and solid phases and
are used in many different industrial processes. Some
examples of their application include filtration pro-
cesses, ion-exchange, catalytic reactions, heat trans-
fer, gas scrubbing, grain drying and others. The
shape and size of particles that make up the bed are
chosen for the characteristics of the specific process.
Correspondence: T. Kaluđerović Radoičić, Faculty of Techno-
logy and Metallurgy, University of Belgrade, Karnegijeva 4, Bel-
grade, Serbia. E-mail: [email protected] Paper received: 6 May, 2015 Paper revised: 12 February, 2016 Paper accepted: 23 February, 2016
The particle size and shape always aim at high pro-
cess effectiveness, so a wide range of particles are
used. In some applications, like in down-flow granular
filters, polydisperse natural materials are used as the
particulate phase. When natural materials are used,
the shape of the particles is irregular and their size
falls into some granulometric interval. Differently
shaped particles pack with different degrees of bed
voidage, which results in different pressure drop
across the bed. The pressure drop through the
packed bed is one of the most important parameters
to be known for the adequate design of the process
as well as for the estimation of the capital and oper-
T. KALUĐEROVIĆ RADOIČIĆ et al.: FRICTION FACTOR FOR WATER FLOW… Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017)
58
ating costs and sizing the pumps or fans required to
force the fluid through the bed.
The pressure gradient through packed beds has
been studied extensively and a large number of cor-
relations were proposed [1-17]. The most widely used
equation for pressure drop calculation was proposed
by Ergun [1]:
22
3 2 3
(1 ) (1 )150 1.75 f
p p
P- = U U
H d d (1)
The friction factor introduced by Ergun is:
3
2 1
pp
f
dPf
H U (2)
According to Eqs. (1) and (2), Ergun’s equation
for friction factor is:
p'
1501.75
Repf (3)
where:
p'Re
(1 )
p fd U (4)
For spherical particles, dp is the diameter of the
particles that constitute the packed bed, while for non-
spherical particles dp is usually taken to be the sur-
face-volume diameter dsv [18,19]. By analogy, all of
the correlations proposed for spherical particles can
be used for non-spherical particles using surface-vol-
ume diameter. Note that particle sphericity is defined
as:
SV
V
d
d (5)
where dV represents the volume diameter of the par-
ticle.
The other literature correlations for friction factor
in packed beds of spherical and non-spherical par-
ticles are shown in Table 1. Note that some authors
used different forms for friction factor and Reynolds
number with respect to Ergun’s definitions of fp and
p'Re , as shown in Table 1.
Table 1. Some important literature correlations for friction factor in packed beds of spherical particles
Reference Friction factor Eq. Re number range
Ergun [1]
p
1501.75
'Repf
(3) 1<Rep<2.4103
Macdonald et al. [2]
p
1801.80
'Repf
(6) -
Gibilaro et al. [3]
'
4.8p
18 (1 )0.33
Repf ; Note: ' 3 / (1 )p pf f
(7) -
Montillet et al. [4]
0.2
'p 3 0.5
p p
1 1000 6012
Re Re
c
p
Df = a
d
a = 0.061 ( < 0.39), a = 0.050 (( > 0.39); For (Dc/dp) > 50, term (Dc/dp)0.2 = 2.2
(8) 10 < Rep < 2.5103
3.8 ≤ Dc/dp ≤ 40-50
Kuerten, ref. in [5]
2'
3 0.5p p
25 1 ) 21 60.28
Re4 Repf =
(9) 0.1 < Rep < 4000
Hicks 6
1.2' 0.2
p3
(1 )6.8 Repf =
(10) 500 < Rep < 6104
Tallmadge 7
2 1.166' 1/6
p3 3p
150 (1 ) 4.2(1 )Re
Repf =
(11) 0.1 < Rep < 105
Lee and Ogawa 10
2' 2
3p p
1 12.5(1 ) 29.32 1.560.1 , where 0.352 0.1 0.275
2 Re Rep n
f = n
(12) 1 < Rep < 105
Cheng [9]
'pRe
pAM
f BM ,
2 11
3 1
p
c
dM
D,
2
2
1185 17
1
c
c p
DA
D d M
21/31 1
1.3 0.03 c
c p
DB
D d M
(13a)
(13)
-
T. KALUĐEROVIĆ RADOIČIĆ et al.: FRICTION FACTOR FOR WATER FLOW… Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017)
59
Table 1. Continued
Reference Friction factor Eq. Re number range
Eisfeld and Schnitzlein [10]
2
'p
154
Rep
M Mf
B, M by Eq. (13a),
221.15( / ) 0.87p cB d D
(14) 0.01 < Rep <
< 1.76105
Reichelt [11]
2
'
150
Rep
p
M Mf
B, M by Eq. (13a),
22
1.5 / 0.88p cB d D (15) -
Zhavoronkov et al. [12]
2
'p
165.31.2
Rep
Af B ,
11
2 / (1 )c p
A BD d
(16) -
Raichura [13]
2
'pRe
pAM
f BM , M by Eq. (13a),
2
2
1036(1 ) 80( / )
1p cA d D
M
22.81 1.82( / )
1p cB d D
M
(17) -
Allen et al. [14]
3
2 cA A
18
1 Re Re
pA
pf
VP a bf =
H AU;
Coefficients a, b and c depend on particle type and packing structure.
Note:
ARe 4 ,
(1 ) 6
p p SVf
p p
V V dU
A A, p A
3 3, Re Re
4 2p Af f
(18) 75 < Rep < 3000
Nemec and Levec [15]
3/2 ' 4/3p
150 1.75
Repf
(19) 10 < Rep < 500
Singh et al. [16]
0.2 0.696 2.945 2
24.666Re exp[11.85(log ) ]V
S pvf
dPf
H U
Note:
pVRe ,
f VU d '
p pV3 / (1 ), Re Re / (1 )p Sf f
(20) 1257 < RepV < 2674
Ozahi et al. [17]
* 32 2
'pV
2761.76
2 1 ReO
VdPf =
L
Note:
* 21/ ,
2fP P U
'pVRe ,
(1 )
f VU d / ,p of f ' '
p pVRe Re
(21) 708 < RepV < 7773
The overview of the pressure drop correlations
for spherical particles shown in Table 1 is given in our
previous paper [20], together with the experimental
data for the friction factor for air flow through packed
beds of spherical glass particles at ambient and ele-
vated temperatures. The main conclusion of this
study was that the overall best fit of all our expe-
rimental data is given by Ergun’s [1] correlation, with
a mean absolute deviation of 10.90% [20].
Some authors proposed friction factor correl-
ations for non-spherical particles that included particle
sphericity directly in the equations (Table 1) [14-17].
Allen et al. [14] reviewed the use of different correl-
ations for pressure drop through packed beds of
spherical and non-spherical particles. They have
shown that the particle shape, arrangement, packing
method as well as surface roughness influence the
pressure drop significantly. The authors also con-
ducted pressure drop measurements in air-particles
system, using randomly packed beds of smooth and
rough glass spheres, wooden cubes, wooden cylin-
ders, acorns (ellipsoids), mixed smooth spheres of
different sizes and rounded and crushed rock with
equivalent diameters from 10.5 to 24.4 mm. The
range of p'Re numbers was 75-3000. Based on their
experimental work, they proposed the correlation for
non-spherical particles represented by Eq. (18), Table
1.
Nemec and Levec [15] investigated single-
-phase flow through packed bed reactors in the range
of p'Re numbers of 10 < p
'Re < 500 with dense and
loose packing of different uniformly sized spherical
and non-spherical particles (glass and Al2O3 spheres,
Al2O3 cylinders and rings as well as Ni-Mo trilobes
and quadralobes) in the size range of 1.26–3.49 mm.
The fluid used in their experiments was nitrogen at 10
bar. The authors concluded that Ergun’s equation
represents a good approximation of the fluid flow
through the packed bed of spherical particles in the
investigated p'Re range, while it under-predicts the
pressure drop over non-spherical particles under the
same conditions. For non-spherical particles they pro-
T. KALUĐEROVIĆ RADOIČIĆ et al.: FRICTION FACTOR FOR WATER FLOW… Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017)
60
posed a new correlation which includes the particles
sphericity, given by Eq. (19), Table 1.
Singh et al. [16] investigated the pressure drop
of air flow through packed bed solar energy storage
system having large sized elements of different
shapes (spheres and cubes) with dV 125-186 mm in
the range of RepV numbers from 1257 to 2674 and
sphericities from 0.55 to 1. The correlation the authors
proposed is given by Eq. (20), Table 1.
Ozahi et al. [17] investigated the pressure drop
in air-particle systems, with particle size 6–19 mm,
and the range of sphericity of 0.55 to 1. The range of
RepV numbers in their experiments was 708–7773.
The correlation for friction factor they proposed is
given by Eq. (21), Table 1.
Several authors 9-13,21 investigated the inf-
luence of Dc/dp on pressure drop in packed beds of
particles with a general conclusion that the effect of
Dc/dp is negligible for Dc/dp >10. In addition, some
recent studies investigated the possibility to extend
the use of Ergun’s equation to polydisperse particles
systems taking into account the particle size distri-
bution [22,23].
Flow through porous media was also studied in
the field of hydrology and for different applications in
civil engineering. The experimental results obtained
were used to validate the semi-empirical relations for
non-Darcy flow [24–26].
The present study was conducted in order to
investigate the optimal choice of friction factor correl-
ation for calculating the pressure drop for water flow
through packed beds of spherical and non-spherical
particles. The experimental evaluation of literature
correlations was conducted by measuring the pres-
sure drop across packed beds of different packing
and for different water flow rates. The spherical par-
ticles used were monosized glass beads, while the
non-spherical material used was polydisperse quartz
filtration sand. The values of the equivalent diameter
and the sphericity of the quartz sand particles needed
for the calculations were obtained by using the cor-
relations for volume diameter and sphericity as a
function of mean sieve diameter proposed in our
previous paper [27]. These correlations were derived
for polydisperse fractions of quartz filtration sand with
sieve diameters in the dm interval 0.359 to 2.415 mm
[27]:
1.0787 0.0355V md d (22)
0.7942 0.063 md (23)
where dV and dm are in mm.
EXPERIMENTAL APPARATUS
The experiments were performed in the water-
–particle system schematically shown in Figure 1. The
packed bed column (f) was used for pressure drop
measurements. It was equipped with a distributor and
the calming section (e) in order to ensure the uniform
flow of water through the bed. The upwards water
flow was induced using a pump (b) and the flow rate
was measured using an electromagnetic flow meter
(d). The packed bed bulk temperature was measured
using the temperature indicator (TI). The pressure
drop in packed beds of different particles was mea-
sured using piezometers (h). The experiments were
performed with two types of particles: glass spherical
particles and polydisperse quartz filtration sand non-
-spherical particles, as shown in Figure 2. The fluid
used was deaerated water at a nearly constant tem-
perature of 20 C. In each run, water temperature was
recorded and water density and viscosity were cal-
culated. The particle characteristics and range of the
experimental conditions are summarized in Table 2.
Figure 1. Schematic diagram of the experimental system
(a-reservoir; b-pump; c-valve; d-electromagnetic flow meter;
e-calming section; f-column; g-overflow; h-piezometers;
TI-temperature indicator).
Seven kinds of mono-sized spherical glass par-
ticles were used. The experiments with glass spher-
ical particles were conducted in two cylindrical col-
umns: first column of the diameter of 40 mm was
used for 0.840–3.020 mm particles and the second
column of the diameter of 62 mm for 4.140–6.180 mm
particles. The ratio of the column diameter to the par-
ticle diameter (geometric aspect ratio) in the experi-
T. KALUĐEROVIĆ RADOIČIĆ et al.: FRICTION FACTOR FOR WATER FLOW… Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017)
61
ments was between 10.0 and 47.6. As the wall effects
were not studied experimentally in this paper, it
should be noted that the literature on the range of
geometric aspect ratio in which the wall effects are
negligible is somewhat divided. Generally, the wall
effects are considered negligible at Dc/dp ratios less
than 10, but there are some researchers who found
wall effects to be significant at Dc/dp ratios as high as
15-20 [10,21]. As in the case of some of our expe-
riments the Dc/dp ratios were in that range, the exist-
ence of wall effects cannot be excluded.
The measurements of the pressure drop in the
beds of non-spherical particles were conducted using
16 fractions of quartz filtration sand obtained from the
company ”Kaolin”-Valjevo. The raw material was first
washed by fluidization to eliminate fine dust, then
dried and sieved through a number of standard sieves
with sieve openings ranging from 2.830 to 0.297 mm.
The obtained fractions had the sieve diameters in the
interval of dm = (ds,n + ds,n+1)/ 2 = 0.359 to 2.415
mm and the ratio between the two successive sieve
sizes dR =ds,n/ds,n + 1 was in the interval 1.132 to
1.715, where ds,n is the size of the opening of the
sieve through which the particle had passed
and ds,n+1 is the size of the opening of the sieve on
which the particle was retained, as shown in Table 2.
Table 2. Particle characteristics and the range of the experimental conditions
dp or dm in mm ds,n+1 / mm ds,n / mm Dc / mm p / kg m–3 U / cm s–1
Spherical particles
6.180 – – 62 2521 0.359-0.366 0.221-2.576
5.040 – – 62 2504 0.372-0.447 0.092-2.576
4.140 – – 62 2514 0.370-0.442 0.097-3.163
3.020 – – 40 2465 0.360-0.423 0.135-2.950
2.120 – – 40 2461 0.366-0.417 0.079-2.576
1.120 – – 40 2895 0.366-0.423 0.027-2.576
0.840 – – 40 2875 0.367-0428 0.018-0.448
Non-spherical particles
2.415 2.000 2.830 64 2638 0.464 0.070-1.860
1.800 1.600 2.000 64 2638 0.458 0.181-1.166
1.700 1.400 2.000 64 2638 0.453 0.098-1.036
1.583 1.166 2.000 64 2638 0.475 0.088-1.145
1.545 1.410 1.680 64 2638 0.457 0.090-1.290
1.500 1.400 1.600 64 2638 0.441 0.078-0.933
1.283 1.166 1.400 64 2638 0.449 0.067-0.943
1.201 0.991 1.410 64 2638 0.500 0.046-1.178
1.098 1.030 1.166 64 2638 0.461 0.052-0.984
1.000 0.750 1.250 64 2638 0.426 0.037-0.423
Figure 2. Some of the particles used in the experiments.
T. KALUĐEROVIĆ RADOIČIĆ et al.: FRICTION FACTOR FOR WATER FLOW… Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017)
62
Table 2. Continued
dp or dm in mm ds,n+1 / mm ds,n / mm Dc / mm p / kg m–3 U / cm s–1
Non-spherical particles
0.940 0.850 1.030 64 2638 0.468 0.067-0.788
0.781 0.711 0.850 64 2638 0.470 0.015-0.626
0.656 0.600 0.711 64 2638 0.464 0.017-0.254
0.560 0.519 0.600 64 2638 0.471 0.033-0.298
0.505 0.420 0.589 64 2638 0.483 0.011-0.321
0.359 0.297 0.420 64 2638 0.486 0.011-0.091
A total of 23 runs were conducted (7 with spher-
ical particles and 16 with non-spherical particles) and
a total of 725 data points were collected (511 for
spherical and 214 for non-spherical particles). The
bed particle Reynolds number, p'Re , varied between
0.3 and 286 for spherical particles and between 0.1
and 50 for non-spherical particles. All of the water
superficial velocities used in the experiments were
below the minimum fluidization velocity for the res-
pective particles. The velocities were in the range of
0.03-0.94 UmF, where UmF represents minimum fluid-
ization velocity.
RESULTS AND DISCUSSION
The results of the friction factor fp vs. p'Re
obtained by the experimental measurements of pres-
sure drop are shown in Figures 3 and 4 for spherical
and non-spherical particles, respectively. The com-
parison between the experimental results and the sel-
ected literature correlations is given in Table 3 and in
Figures 5 and 6. The literature correlations for spher-
ical particles shown in Table 1 were tested for all the
experimental data. The surface-volume diameter and
the sphericity needed for the calculations for non-
spherical particles were obtained using the correl-
ations from our previous paper [27] for quartz filtration
sand (Eqs. (22) and (23)). For quartz filtration sand
beds, the literature correlations specifically defined for
non-spherical particles were also tested: Allen et al.
[14], Nemec and Levec [15], Singh et al. [16] and
Ozahi et al. [17] correlations.
The mean absolute deviation between the mea-
sured values of the pressure gradient and the values
obtained from the literature correlations were cal-
culated according to the following equation:
calc measured
measured1
( / ) ( / )1
( / )
NP H P H
N P H
(24)
where N is the number of data points, (P/H)calc and
(P/H)measured are the calculated and the measured
pressure gradients.
Figure 3. fp vs. p'Re for spherical particles.
Figure 4. fp vs. p'Re for non-spherical particles.
As can be seen from Table 3, the best fit of our
experimental data for pressure drop in beds of spher-
ical particles was obtained using Cheng [9] correl-
ation, with mean absolute deviation of 10.89%. The
correlations of Macdonald et al. [2] and Montillet et al.
T. KALUĐEROVIĆ RADOIČIĆ et al.: FRICTION FACTOR FOR WATER FLOW… Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017)
63
[4] also gave very good results in fitting our expe-
rimental data with mean errors of 12.18 and 13.13%,
respectively. A number of other correlations tested
gave results with mean absolute deviation in the
range of 16-20%. It should be noted that Hicks [6]
correlation gave the mean error of 50.19% for spher-
ical particles and 84.34% for non-spherical particles.
The reason for such a large error is that the range of
Rep numbers in this paper was below 183, while the
specified range of applicability of Hicks correlation is
Rep > 500.
Table 3. Comparison of experimental data for friction factor, σ
(%), with different correlations from the literature
Reference Particles
Spherical Non-spherical
Ergun [1] 19.82 10.36a
Macdonald et al. [2] 12.18 21.28 a
Gibilaro et al. [3] 18.19 13.28 a
Montillet et al. [4] 13.13 45.53 a
Kuerten, ref. in [5] 29.36 24.85 a
Hicks [6] 50.19 84.34 a
Tallmadge [7] 16.04 11.19 a
Lee and Ogawa [8] 18.61 25.30 a
Cheng [9] 10.89 31.74 a
Eisfeld and Schnitzlein [10] 18.78 11.31 a
Raichura [13] 39.07 111.39 a
Reichelt [11] 20.02 10.64 a
Zhavoronkov et al. [12] 19.04 13.58 a
Allen et al. [14] – 59.40 b
Nemec and Levec [15] – 63.52 c
Singh et al. [16] – 85.22
Ozahi et al. [17] – 78.53
This paper (Eq.(25)) 9.04 –
aUsing dSV; bUsing coefficients for crushed rock, dSVn = 24.4 mm; cUsing
coefficients for cylindrical particles
The correlations with mean absolute deviation
between experimental and correlated pressure drop
less than 20% for beds of spherical particles are
shown in Figure 5. The correlations shown in Figure 5
were calculated for Dc/dp = 25 and bed porosity of
= 0.40 (the mean values in our experiments) in
order to be able to show the correlations with direct
dependence on as lines.
Compared to the data of our previous paper
[20], in which air-spherical particles system was
investigated, the results are in the similar range for
Macdonald et al. [2] (12%) and Cheng [9] (11 and
12%) correlations, which gave very good results at
ambient temperature both for air-particles and water-
particles systems. On the other hand, the correlations
of Ergun [1], Tallmadge [7], Reichelt [11], Gibilaro et
al. [3], Einsfeld and Schnitzlein [10] and Zhavoronkov
et al. [12] performed better in air-particles system,
while the correlation of Montillet et al. [4] performed
better in water-particles systems.
Figure 5. Comparison of experimental data of fp vs. p'Re with
chosen correlations for spherical particles.
In order to improve the correlation results for
spherical particles, a new simple equation is pro-
posed in the form of Ergun’s equation, with modified
coefficient:
'
2091.75
Rep
p
f (25)
The mean absolute deviation between the
values calculated from Eq. (25) and the experimental
data is 9.04%. The new correlation is shown in com-
parison to the experimental data in Figure 3.
For non-spherical particles (quartz filtration
sand), the best fit of the experimental data was
obtained using Ergun’s equation [1], with mean abs-
olute deviation of 10.36%. The correlations of Reich-
elt [11], Tallmadge [7] and Einsfeld and Schnitzlein
[10] also gave very good results in fitting the experi-
mental data with mean errors of 10.64, 11.19 and
11.31%, respectively. The correlations with mean
absolute deviation of pressure drop less than 20% for
quartz filtration sand packed beds are shown in Fig-
ure 6. The correlations shown in Figure 6 were cal-
culated for Dc/dp = 73 and bed porosity of = 0.40
(the mean values in our experiments) in order to be
able to show the correlations with direct dependence
on as lines.
T. KALUĐEROVIĆ RADOIČIĆ et al.: FRICTION FACTOR FOR WATER FLOW… Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017)
64
Figure 6. Comparison of experimental data of fp vs. p'Re with
chosen correlations for non-spherical particles.
The correlations specifically derived for non-
spherical particles gave very poor results in correl-
ating our experimental data. The mean absolute devi-
ations for these correlations were in the range of
59.40 to 85.22%. The reason for this is that the correl-
ations of Allen et al. [14], Sing et al. [16] and Ozahi et
al. [17] were derived from experiments in systems in
which the particles were of large diameters (6–20 mm
equivalent diameters) and the Rep numbers were
larger than 75, while in our system, the Rep numbers
of non-spherical particles were smaller than 25. Allen
et al. [14] and Singh et al. [16] correlations were
derived for packed bed solar energy storage systems
with particle materials being wooden cubes, crushed
rock, concrete and masonry bricks. On the other
hand, the correlation of Nemec and Levec [15] was
derived for the particles in the range of 1.26–3.49 mm,
but in the high pressure system (5–20 bar).
The very poor performance of the correlations
derived for the non-spherical particles in our quartz
sand packed beds emphasizes the fact that the
friction factor in packed beds strongly depends on
other variables besides the sphericity of the particles.
It can be concluded that the friction factor is very sys-
tem specific and that it depends on particle shape and
size, voidage, arrangement, packing method as well
as surface roughness.
It is interesting to note that almost all of the cor-
relations that gave good results in correlating our exp-
erimental data both for spherical and non-spherical
particles were in the form of Ergun’s equation, i.e.,
represented a modification of this equation with differ-
ent coefficients. From this fact it can be concluded
that the form of Ergun’s equation with two added
terms describing viscous and inertial effects is ade-
quate for representing the friction factor in packed
beds. However, the coefficients in the equation are
very system-specific and care should be taken when
choosing the adequate equation for pressure drop
calculation for the specific purpose. The correlation
chosen for the specific system should be obtained
from the data in the similar range of experimental con-
ditions as the system it is intended to be applied to.
CONCLUSIONS
The present study was conducted in order to
investigate the optimal choice of friction factor correl-
ation for water flow through packed bed of particles.
The best fit of experimental data for spherical par-
ticles was obtained using the Cheng [9] correlation
(mean absolute deviation of 10.89%). Ergun’s equa-
tion [1] gave better results in correlating the data for
non-spherical particles with mean absolute deviation
of 10.36% compared to 19.82% for spherical glass
particles. Ergun’s equation was modified in order to
improve the fit for spherical particles and a new cor-
relation was proposed. The mean absolute deviation
between the experimental data and the proposed
correlation is 9.04%.
The correlations specially derived for non-spher-
ical particles gave very poor results in correlating our
experimental data probably because they were
derived for systems with much larger particles packed
in a different arrangement. Most of the correlations
that gave good results in correlating experimental
data both for spherical and for non-spherical particles
were in the form of Ergun’s equation with modified
coefficients, thus showing that the form of Ergun’s
equation with two added terms describing viscous
and inertial effects is adequate for representing the
friction factor in packed beds. However, the coef-
ficients in the equation are very system-specific.
Acknowledgment
Financial support of the Serbian Ministry of
Education, Science and Technological Development.
(Project ON172022) is gratefully acknowledged.
Nomenclature
a empirical coefficient in Eq. (6), dimensionless
Ap area of a particle
b empirical coefficient in Eq. (6), dimensionless
c empirical coefficient in Eq. (6), dimensionless
dm particle sieve diameter
dp packed bed particle diameter
T. KALUĐEROVIĆ RADOIČIĆ et al.: FRICTION FACTOR FOR WATER FLOW… Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017)
65
ds sieve opening
dSV particle surface-volume diameter
ds,n the size of the opening of the sieve through
which the particle had passed
ds,n+1 the size of the opening of the sieve on which
the particle was retained
dV particle volume diameter
Dc column diameter
fA friction factor defined according to Eq. (18)
fo friction factor defined according to Eq. (21)
fS friction factor defined according to Eq. (20)
fp friction factor defined according to Eq. (2)
f’p modified friction factor defined according to
Eq. (7)
H bed height
P pressure
P bed pressure drop, Pa
P dimensionless bed pressure drop, Eq.(21)
Rep =(fUdp)/μ particle Reynolds number
p'Re =(fUdp)/(μ(1-ε)) bed particle Reynolds num-
ber
RepV particle Reynolds number defined according
to Eq. (20)
RepV’ bed particle Reynolds number defined
according to Eq. (21)
ReA bed particle Reynolds number defined
according to Eq. (18)
U superficial fluid velocity
UmF minimum fluidization velocity (superficial)
Vp volume of a particle
Greek letters
voidage
μ fluid viscosity
ρf fluid density
ρp particle density
σ mean absolute deviation
particle sphericity, dimensionless
Subscripts
f fluid
mF minimum fluidization
p particle
REFERENCES
[1] S.S. Ergun, Chem. Eng. Prog. 48 (1952) 89-94
[2] F. Macdonald, M.S. El-Sayed , K. Mow, F.A.L. Dullien,
Ind. Eng. Chem. Fundam. 18 (1979) 199-208
[3] L.G. Gibilaro, R. Di Felice, S.P. Waldram, Chem. Eng.
Sci. 40 (1985) 1817-1823
[4] A. Montillet, E. Akkari, J. Comiti, Chem. Eng. Process. 46
(2007) 329-333
[5] H. Watanabe, Inte. J. Eng. Fluid Mech. 2 (1989) 93-108
[6] R.E. Hicks, Ind. Eng. Chem. Fundam. 9 (1970) 500-502
[7] J.A. Tallmadge, AIChE J. 16 (1970) 1092-1093
[8] J. Lee, K. Ogawa, J. Chem. Eng. Jpn. 27 (1994) 691-693
[9] N.S. Cheng, Powder Technol. 210 (2011) 261-266
[10] B. Eisfeld, K. Schnitzlein, Chem. Eng. Sci. 56 (2001)
4321-4329
[11] W. Reichelt, Chem. Ing. Tech. 44 (1972) 1068-1071
[12] N.M. Zhavoronkov, M.E. Aerov, N.N. Umnik, J. Phys.
Chem. 23 (1949) 342-361 (in Russian)
[13] R.C. Raichura, Exp. Heat Transfer 12 (1999) 309-327
[14] K.G. Allen, T.W. von Backström, D.G. Kröger, Powder
Technol. 246 (2013) 590-600
[15] D. Nemec, J. Levec, Chem. Eng. Sci. 60 (2005) 6947-6957
[16] R.Singh, R.P. Saini, J.S. Saini, Sol. Energy 80 (2006)
760-771
[17] E.Ozahi, M.Y.Gundogdu, M.Ö. Carpinlioglu, Adv. Powder
Technol. 19 (2008) 369-381
[18] D. Geldart, Powder Technol. 60 (1990) 1-13
[19] W.-C. Yang, in: Handbook of Fluidization and Fluid-par-
ticle Systems, W.-C. Yang, Marcel Dekker, New York,
2003
[20] R. Pešić, T. Kaluđerović Radoičić, N. Bošković-Vra-
golović, Z. Arsenijević, Ž. Grbavčić, Chem. Ind. Chem.
Eng. Q. 21 (2015) 419–427
[21] R. Di Felice, L.G. Gibilaro, Chem. Eng. Sci. 59 (2004)
3037-3040
[22] M. Mayerhofer, J. Govaerts, N. Parmentier, H. Jeanmart,
L. Helsen, Powder Technol. 205 (2011) 30-35
[23] A. Luckos, J.R. Bunt, Fuel 90 (2011) 917-921
[24] M. Sedghi-Asl, R. Hassan, J. Hydrol. Eng. 49 (2011) 814-
–817
[25] K.N. Moutsopoulos, I.N. Papaspyros, V.A. Tsihrintzis, J.
Hydrol. 374 (2009) 242-254
[26] M.B. Salahi, M. Sedghi-Asl, M. Parvizi, J. Hydrol. Eng.
(2015) 04015003
[27] T. Kaluđerović Radoičić, M.Đuriš, R.Garić-Grulović, Z. Ar-
senijević, Ž.Grbavčić, Powder Technol. 254 (2014) 63-71.
T. KALUĐEROVIĆ RADOIČIĆ et al.: FRICTION FACTOR FOR WATER FLOW… Chem. Ind. Chem. Eng. Q. 23 (1) 5766 (2017)
66
TATJANA KALUĐEROVIĆ RADOIČIĆ1
NEVENKA
BOŠKOVIĆ-VRAGOLOVIĆ1
RADMILA GARIĆ-GRULOVIĆ2
MIHAL ĐURIŠ2
ŽELJKO GRBAVČIĆ1
1Tehnološko-metalurški fakultet,
Univerzitet u Beogradu, Karnegijeva 4,
Beograd, Srbija 1Institut za hemiju, tehnologiju i
metalurgiju, Univerzitet u Beogradu,
Njegoševa 12, Beograd, Srbija
NAUČNI RAD
KOEFICIJENT TRENJA FLUID-ČESTICE U PAKOVANIM SLOJEVIMA SFERIČNIH I NESFERIČNIH ČESTICA
Cilj ovog rada je bio eksperimentalno ispitivanje koeficijenta trenja fluid-čestice prilikom
strujanja vode kroz pakovani sloj sferičnih i nesferičnih čestica na sobnoj temperaturi. U
eksperimentima je meren pad pritiska prilikom strujanja fluida kroz pakovani sloj. Na osno-
vu dobijenih rezultata, izvršena je evaluacia različitih literaturnih korelacija. U eksperimen-
tima je korišćeno sedam vrsta monodisperznih sferičnih staklenih čestica, kao i 16 frakcija
polidisperznih nesferičnih čestica filtracionog peska različitih dimenzija, dobijenih proseja-
vanjem. Opseg poroznosti pakovanih slojeva je bio od 0,359 do 0,486, dok je opseg vred-
nosti Rejnlodsovog broja za čestice bio od 0,3 do 286 za sferične čestice i od 0,1 do 50 za
nesferične čestice. Dobijeni rezultati su korelisani korišćenjem većeg broja literaturnih
korelacija. U cilju poboljšanja rezultata korelisanja za sferične čestice, predložena je nova
jednačina u formi Ergunove jednačine sa modifikovanim koeficijentima. Srednje apsolutno
odstupanje eksperimentalnih od izračunatih vrednosti za predloženu korelaciju iznosilo je
9,04%. Za nesferične čestice kvarcnog filtracionog peska, najbolji rezultati su dobijenu
korišćenjem Ergunove jednačine, sa srednjim apsolutnim odstupanjem od 10,36%. Povr-
šinsko-zapreminski prečnik (dSV) koji je neophodan za korelisanje eksperimentalnih poda-
taka za nesferične čestice je računat na osnovu korelacija za dV = f(dm) i = f(dm) koje su
predložene u našem prethodnom radu [27].
Ključne reči: pad pritiska, pakovani sloj, sferične čestice, kvarcni filtracioni pesak,
nesferične čestice.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 6772 (2017) CI&CEQ
67
HAI-PENG GOU1,2
GUO-HUA ZHANG1
KUO-CHIH CHOU1,2
1State Key Laboratory of Advanced
Metallurgy, University of Science
and Technology Beijing, Beijing,
China 2Collaborative Innovation Center of
Universal Iron and Steel
Technology, University of Science
and Technology Beijing, Beijing,
China
SCIENTIFIC PAPER
UDC 621.762:669.295:546.72:66
https://doi.org/10.2298/CICEQ150521007G
PREPARATION OF TITANIUM CARBIDE POWDER FROM ILMENITE CONCENTRATE
Article Highlights
• Titanium carbide powder was prepared by the carbothermic reduction of ilmenite con-
centrate
• The iron and titanium carbide in the reduction products were economically separated by
using ferric chloride solution
• The iron in the waste liquid was recycled in the form of -FeOOH
Abstract
A new process of producing titanium carbide powder from ilmenite concentrate
was put forward. The ilmenite concentrate was reduced by graphite powder at
1500 C for 6 h, with the reduction products of TiC, Fe and MgO. The porous
reduction products were agitation leached in ferric chloride solution at 25 C for
60 min. After the filtration, TiC powder accompanied by little MgAl2O4 and
Mg2SiO4 were obtained. Finally, the element Fe was recycled from the filtrate in
the form of -FeOOH after blowing air for 3 hours at 80 C. The size of rod-like
-FeOOH particle was less than 1 µm.
Keywords: titanium carbide, ferric chloride solution, ilmenite concen-trate, -FeOOH.
Titanium carbide, TiCx, existing as a homogen-
eous phase within the limits 0.47 < x < 1.0 [1], has a
NaCl-type of structure. Because of its high melting
point (3067 C), high hardness (32.4 GPa), good
chemical inertness and good electrical conductivity
(310-7 S/cm), titanium carbide has been found tre-
mendous applications in various fields, such as wear-
resistant material, cutting tool and anodes in lithium-
-ion batteries [2-4].
Nowadays, various methods have been adopted
to prepare titanium carbide, e.g., self-propagating
high temperature synthesis (SHS) [1], mechanically
activated sintering [5], thermal plasma [6], carbo-
thermic reduction [7,8], etc. Considering the abundant
reserves of ilmenite concentrate in the world, the pre-
paration of titanium carbide by ilmenite concentrate is
receiving more and more attentions. Most of the
studies were focused on the carbothermic reduction
process [9-11], while how to separate the reduction
Correspondence: G.-H. Zhang, State Key Laboratory of Adv-
anced Metallurgy, University of Science and Technology Bei-
jing, Beijing, 100083, China. E-mail: [email protected] Paper received: 21 May, 2015 Paper revised: 19 February, 2016 Paper accepted: 26 February, 2016
products of iron and titanium carbide economically is
also quite significant. After the carbothermic reduction
of ilmenite concentrate, Welham and Williams [7]
used 3% HCl to leach the products for 24 h at room
temperature. As a result, almost all of the iron went
into the waste liquid and could not be recycled. On
the other hand, there are many reports on the pro-
duction of TiC reinforced iron-based composite from
ilmenite [12-16]. After being milled in a planetary ball
mill, TiC reinforced iron-based composite was syn-
thesized via microwave heating [12,13], electric dis-
charge assisted mechanical milling (EDAM) [14] or
carbothermic reduction [15,16]. It is economical to
convert raw materials directly to composite materials.
In this paper, a new process of separating iron
and titanium carbide via lixiviation was put forward.
The ferric chloride solution was used for leaching to
separate reduction productions of iron and titanium
carbide. After leaching, the main component of the fil-
trate was ferrous chloride. The iron in the filtrate could
be recycled in the form of -FeOOH [17,18]. -FeOOH
could be used as an important by-product to produce
ultrafine Fe2O3 [19]. After recycling the elemental Fe,
the main component of the rest filtrate was ferric
H.-P. GOU et al.: PREPARATION OF TITANIUM CARBIDE POWDER… Chem. Ind. Chem. Eng. Q. 23 (1) 6772 (2017)
68
chloride, which could be reused for the next leaching
process.
EXPERIMENTAL
The ilmenite concentrate, produced in Panz-
hihua, Sichuan, China, was examined by X-ray fluor-
escence (XRF) and X-ray diffraction (XRD). The
results of XRF and XRD are presented in Table 1 and
Figure 1a, respectively. The main mineral phase of the
ilmenite concentrate is FeTiO3, while the main impur-
ity element is Mg, which exists in solid-solution of
(Mg,Fe)(Ti,Fe)O3 [20]. As shown in Figure 1b, it should
be noted that the peaks of FeTiO3 and (Mg,Fe)(Ti,Fe)O3
are similar and overlapped in the XRD pattern
because of the same crystalline structure.
Figure 1. XRD patterns of the ilmenite concentrate.
According to the authors’ previous study [10],
the optimized carbothermic reduction parameters
were established as: molar ratio of C to FeTiO3 4:1,
reduction temperature 1500 C and reduction time 6
h. The main reactions occurred during the carbo-
thermic reduction are shown in Eqs. (1)-(4). All the
standard Gibbs energy changes, G , are calculated
by Factsage 6.4:
3FeTiO 4C Fe TiC 3CO
G = –497.7T + 711201.1 (1)
3MgTiO 3C MgO TiC 2CO
G = –342.2T + 558842.7 (2)
2 3 2 4MgO Al O MgAl O
G = –6.9T + 22840.7 (3)
2 2 42MgO SiO Mg SiO
G = –4.4T + 65772.6 (4)
The ilmenite concentrate and graphite powder
(Sinopharm Chemical Reagent Co., Ltd, Chemical
Pure, 98%) were mixed uniformly in an agate mortar
(Changzhou Putian Instrument Manufacture CO., Ltd,
diameter 105 mm). Then the mixtures were made into
cylindrical briquettes with the addition of PVA (2 wt.%).
The diameter and weight of the cylindrical briquettes
were 18 mm and 2 g, respectively. When the tempe-
rature of the vertical tube furnace reached 1500 C,
the alumina crucible with the briquettes was put into
the furnace under a protective argon gas atmosphere
(0.8 L/min). After reacting for 6 h, the crucible was
taken out of the furnace quickly and cooled by the
argon stream (1.5 L/min). The reduction products
were examined by XRD and scanning electron micro-
scope (SEM) to investigate their phase compositions
and microstructure.
The reduction products were agitation leached
by ferric chloride solution (0.5 mol/L) at 25 C in an
electro-thermostatic water bath. The stirring rate was
200 rpm and the pulp density for each experiment
was 20 g/L. The leaching time was 1, 2, 5, 10, 20, 30
and 60 min, respectively. After leaching, the solid
phase was separated from the liquid phase by suction
filtration as soon as possible. After being rinsed
thoroughly by deionized water, the obtained solid
phase was examined by XRD and SEM.
Table 1. Chemical compositions of ilmenite concentrate (wt.%)
Component FeO TiO2 SiO2 CaO Al2O3 MgO SO3
Content 39.30 43.68 3.15 1.28 2.91 7.99 0.62
Component Na2O MnO Cr2O3 ZnO P2O5 In2O3 Total
Content 0.28 0.69 0.03 0.02 0.03 0.02 100
H.-P. GOU et al.: PREPARATION OF TITANIUM CARBIDE POWDER… Chem. Ind. Chem. Eng. Q. 23 (1) 6772 (2017)
69
The filtrate after leaching was collected in a
beaker and then heated with blowing air (0.5 L/min).
The temperature was set to be 80 C and was not
changed during this process. The purpose of blowing
air into the filtrate was to accelerate the reaction. After
3 h, a yellow solid precipitate appeared. The preci-
pitate was collected by suction filtration and examined
by XRD and field emission scanning electron micro-
scope (FE-SEM).
RESULTS AND DISCUSSION
Carbothermic reduction of the ilmenite concentrate
After 6 h of the carbothermic reduction, the per-
centage of total mass loss ratio was 41.88, which was
consistent with the maximum theoretical mass loss
ratio [10]. It illustrates that the carbothermic reduction
reacted completely. The XRD patterns of the red-
uction products are presented in Figure 2, from which
it can be concluded that the reduction products were
mainly made up of TiC, Fe and MgO. The morphology
images of the ilmenite concentrate and the reduction
products morphology are shown in Figure 3a and b,
respectively. Compared with the ilmenite concentrate,
the reduction products were porous, which was due to
the generation and evolution of CO gas. Backscat-
tered electron (BSE) image of the main phases inside
the reduction products is displayed in Figure 3c,
which indicates that there are three different regions
in the samples. Based on the results, the energy dis-
persive spectrometer (EDS) analyses performed at
different regions are shown in Table 2. It is obvious
that the main impurity element Mg mostly existed in
the form of MgO after the carbothermic reduction, which
is consistent with the results of the XRD analyses.
Purification of titanium carbide powders
The percentages of mass loss with different
leaching time are shown in Figure 4a. After leaching
for 10 min, the mass loss vs. time curve reached a
plateau of 47.6%. As shown in Figure 3b, the porous
structure of the reduction products was advantageous
to the leaching process. XRD pattern of the leaching
products after leaching for 60 minutes is presented in
Figure 4b. Compared with Figure 2, the peaks of Fe
and MgO disappeared in Figure 4b. Based on Figures
2 and 4b, the following reactions occurred in the pro-
cess of lixiviation. The hydrogen ions in Eq. (6) were
resulted from the hydrolysis of ferric chloride solution:
3 22FeCl Fe 3FeCl (5)
222H MgO Mg H O (6)
Figure 2. XRD patterns of the reduction products.
Table 2. EDS results of different phases in Figure 3c
Phase Elements mass fraction
Fe 88.75%Fe; 5.99%C; 2.37%Si; 2.89%Ti
TiC 85.43%Ti; 13.85%C; 0.71%V
MgO 30.80%O; 53.68%Mg; 1.91%Ca; 9.80%Si; 0.78%Fe;
3.02%Ti
Figure 3. a) SEM morphology image of the ilmenite concentrate; b) SEM image of the reduction products; c) BSE image of the mainly
compositions inside the reduction products.
H.-P. GOU et al.: PREPARATION OF TITANIUM CARBIDE POWDER… Chem. Ind. Chem. Eng. Q. 23 (1) 6772 (2017)
70
After leaching, the main phases were TiC,
MgAl2O4 and Mg2SiO4. BSE images of the leaching
products are shown in Figure 5. The overall distri-
bution of the leaching products is displayed in Figure
5a. Most of the iron was removed by ferric chloride
solution. As a result, TiC particles which were integ-
rated with Fe tightly were scattered into individual
particles. The wetting angle between TiC and Fe at
1500 C is 30 [21]. As shown in Figure 5b, there was
still a little bit of Fe inside TiC particles. This part of
the iron was wrapped by TiC particles and could not
be removed by ferric chloride solution. From Figure
5c and d, it could be seen that a few impurities of
MgAl2O4 and Mg2SiO4 were combined with TiC par-
ticles. EDS analyses performed at different phases
are shown in Table 3, which were in agreement with
the XRD patterns in Figure 4b. The removal of MgAl2O4
and Mg2SiO4 are complicated and are now in prog-
ress. Or, the mixture of TiC, MgAl2O4 and Mg2SiO4
may be considered as a new TiC based composite
material since all of them have very high hardness.
Recycle of the waste liquid
When blowing air into the filtrate, the ferrous
chloride solution was oxidized to ferric chloride sol-
ution, and a yellow solid precipitate appeared. The
Figure 4. a) The percentages of mass loss with different leaching time; b) XRD patterns of the leaching products
with leaching for 60 min.
Figure 5. BSE images of the leaching products.
H.-P. GOU et al.: PREPARATION OF TITANIUM CARBIDE POWDER… Chem. Ind. Chem. Eng. Q. 23 (1) 6772 (2017)
71
XRD patterns of the precipitate are presented in
Figure 6a, which indicates that the precipitate is aka-
ganeite (-FeOOH). As seen from the FE-SEM mor-
phology image of -FeOOH shown in Figure 6b,
-FeOOH with the length less than 1 m was rod-like
shaped, which was consistent with the reported expe-
riment phenomena [22,23]. The rod-like particles of
-FeOOH could be dissolved and reprecipitate as
hematite (-Fe2O3). Under different conditions, -Fe2O3
could be in different morphologies of spheres, cubes
or double ellipsoids [22]. Therefore, -FeOOH is a
valuable by-product. The main reaction occurring dur-
ing blowing air into the waste liquid is shown in Eq.
(7):
2 2 2 312FeCl 3O 2H O 8FeCl 4 -FeOOH (7)
Table 3. EDS results of different phases in Figure 5
Phase Elements mass fraction
Fe 87.94%Fe; 9.94%C; 2.12%Ti
TiC 86.73%Ti; 12.64%C; 0.63%V
MgAl2O4 48.84%O; 15.84%Mg; 31.67%Al; 3.65%Ti
Mg2SiO4 45.40%O; 34.69%Mg; 17.34%Si; 1.02%Ca; 1.55%Ti
As can be concluded from Eq. (7), 1/3 Fe con-
tent is recovered in the form of FeOOH and the
remaining Fe still existed as ferric chloride in solution,
which could be reused in the next leaching process.
However, as the accumulation of the elemental Mg in
the new ferric chloride solution, the recycle of the
ferric chloride solution would be terminated when
MgCl2 reached saturation.
CONCLUSION
The purification of titanium carbide powder from
ilmenite concentrate was investigated in this article.
After the carbothermic reduction at 1500 C for 6 h
and lixiviation by ferric chloride solution at 25 C for
10 min, TiC powder with little MgAl2O4 and Mg2SiO4
was produced. The mass loss ratios during the carbo-
thermic reduction and lixiviation process were 41.88
and 47.6%, respectively. The carbothermic reduction
product Fe went into the waste liquid and was
recycled in the form of -FeOOH after blowing air for
3 h at 80 C. The waste liquid was oxidized to ferric
chloride solution for the next process of lixiviation.
Acknowledgements
Thanks are given to the financial supports form
the Fundamental Research Funds for the Central
Universities (FRF-TP-15-009A3) and the National
Natural Science Foundation of China (51474141).
REFERENCES
[1] J.B. Holt, Z.A. Munir, Combustion synthesis of titanium
carbide: theory and experiment. J. Mater. Sci. 21 (1986)
251-259
[2] G.E. Hollox, R.E. Smallman, J. Appl. Phys. 37 (1966)
818-823
[3] N. durlu, J. Eur. Ceram. Soc. 19 (1999) 2415-2419
[4] P. Patel, I.S. Kim, P.N. Kumta, Mater .Sci. Eng. B 116
(2005) 347-352
[5] M. Razavi, M. R. Rahimipour, R. Kaboli, J. Alloys Compd.
460 (2008) 694-698
[6] L. Tong, R.G. Reddy, Scripta Mater. 52 (2005) 1253-1258
[7] N.J. Welham, J.S. Williams, Metall. Mater. Trans. B 30
(1999) 1075-1081
[8] N.J. Welham, Miner. Eng. 9 (1996) 1189-1200
[9] K.S. Coley, B.S. Terry, P. Grieveson, Metall. Mater.
Trans. B 26 (1995) 485-494
[10] H.P. Gou, G.H. Zhang, K.C. Chou, Metall. Mater. Trans.
B 46 (2014) 48-56
[11] S.K. Gupta, V. Rajakumar, P. Grieveson, Metall. Trans. B
20 (1989) 735-745
Figure 6. a) XRD patterns of the precipitate; b) FE-SEM morphology image of -FeOOH.
H.-P. GOU et al.: PREPARATION OF TITANIUM CARBIDE POWDER… Chem. Ind. Chem. Eng. Q. 23 (1) 6772 (2017)
72
[12] M. Razavi, M.S. Yaghmaee, M.R. Rahimipour, S.S.
Razavi-Tousi, Int. J. Miner. Process. 94 (2010) 97-100
[13] M. Razavi, M.R. Rahimipour, T. Ebadzadeh, S.S. Razavi-
-Tousi, Bull. Mater. Sci. 32 (2009) 155-160
[14] A. Calka, D. Oleszak, N. Stanford, J. Alloys Compd. 459
(2008) 498-500
[15] Y. Chen, Scripta Mater. 36 (1997) 989-993
[16] M. Razavi, M.R. Rahimipour, Ceram. Int. 35 (2009) 3529-
-3532
[17] C. Rémazeilles, Ph. Refait, Corros. Sci. 49 (2007) 844-
-857
[18] P. Refait, J.M.R. Génin, Corros. Sci. 39 (1997) 539-553
[19] E.A. Deliyanni, D.N. Bakoyannakis, A.I. Zouboulis, K.A.
Matis, L.Nalbandian, Microporous Mesoporous Mater. 42
(2001) 49-57
[20] R.G. Berman, L.Y. Aranovich, Contrib. Mineral. Petrol.
126 (1996) 1-24
[21] R. Warren, J. Mater. Sci. 15 (1980) 2489-2496
[22] J.K. Bailey, C.J. Brinker, M.L. Mecartney, J.Colloid
Interface Sci. 157 (1993) 1-13
[23] S. Musić, S. Krehula, S. Popović, Mater. Lett. 58 (2004)
444-448.
HAI-PENG GOU1,2
GUO-HUA ZHANG1
KUO-CHIH CHOU1,2
1State Key Laboratory of Advanced
Metallurgy, University of Science and
Technology Beijing, Beijing, China 2Collaborative Innovation Center of
Universal Iron and Steel Technology,
University of Science and Technology
Beijing, Beijing, China
NAUČNI RAD
PRIPREMA PRAŠKASTOG TITAN-KARBIDA OD KONCENTRATA ILMENITA
Razvijen je novi proces za dobijanje praškastog titanijum karbida od koncentrata ilmenita.
Koncentrat ilmenita je usitnjen grafitnim prahom na 1500 C u toku 6 h, pri čemu su pro-
izvodi redukcije TIC, Fe i MgO. Porozni redukcioni proizvodi su luženi u rastvoru feri-hlo-
rida na 25 C uz mešanje 60 min. Nakon filtracije, dobijeni su prah TiC uz malo MgAl2O4 i
Mg2SiO4. Konačno, elementarno gvožđe je reciklisano iz filtrata u obliku -FeOOH nakon
uduvavanja vazduh tokom 3 h na 80 C. Veličina štapišastih čestica -FeOOH je manja od
1 m.
Ključne reči: Titan-karbid, gvožđe-hlorid, koncentrat ilmenite, -FeOOH.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017) CI&CEQ
73
MILANA M. ZARIĆ1
MIRKO STIJEPOVIC1
PATRICK LINKE2
JASNA STAJIĆ-TROŠIĆ1
BRANKO BUGARSKI3
MIRJANA KIJEVČANIN3
1Institute of Chemistry, Technology
and Metallurgy, University of
Belgrade, Belgrade, Serbia 2Department of Chemical
Engineering, Texas A&M
University at Qatar, Education City,
Doha, Qatar 3Faculty of Technology and
Metallurgy, University of Belgrade,
Belgrade, Serbia
SCIENTIFIC PAPER
UDC 620.92:66.021.4
https://doi.org/10.2298/CICEQ150622009Z
TARGETING HEAT RECOVERY AND REUSE IN INDUSTRIAL ZONE
Article Highlights
• Heat recovery and reuse of waste heat via indirect heat integration
• Increasing of energy efficiency and reducing consumption of fossil fuel
• Linear programming (LP) used for method formulation
• Industrial zone energy integration strategy
Abstract
In order to reduce the usage of fossil fuels in industrial sectors by meeting the
requirements of production processes, new heat integration and heat recovery
approaches are developed. The goal of this study is to develop an approach to
increase energy efficiency of an industrial zone by recovering and reusing waste
heat via indirect heat integration. Industrial zones usually consist of multiple inde-
pendent plants, where each plant is supplied by an independent utility system, as
a decentralized system. In this study, a new approach is developed to target
minimum energy requirements where an industrial zone would be supplied by a
centralized utility system instead of decentralized utility system. The approach
assumes that all process plants in an industrial zone are linked through the
central utility system. This method is formulated as a linear programming prob-
lem (LP). Moreover, the proposed method may be used for decision making rel-
ated to energy integration strategy of an industrial zone. In addition, the pro-
posed method was applied on a case study. The results revealed that saving of
fossil fuel could be achieved.
Keywords: heat recovery, energy efficiency, heat integration, LP formul-ation.
Global industrial growth has triggered energy
consumption levels by the industrial sectors [1].
Energy intensive processes rely on usage of fossil
fuels to provide their energy requirements. Increased
fossil fuel consumption leads to undesirable increase
in greenhouse gas (GHG) emissions, causing climate
changes [2]. There are two basic concepts to dec-
rease the use of fossil fuels: replacing the fossil fuels
with renewable energy or increasing the energy effi-
ciency of the process [3]. In order to meet the ind-
ustrial requirements while reducing emissions of GHG,
the efficiency of production processes needs to be
improved. Also, new technologies are developed to
Correspondence: M. Kijevčanin, Faculty of Technology and
Metallurgy, University of Belgrade, Karnegijeva 4, 11120 Bel-
grade, Serbia. E-mail: [email protected] Paper received: 22 June, 2015 Paper revised: 2 February, 2016 Paper accepted: 4 March, 2016
target process integration and process intensification
[4]. Moreover, process inefficiencies cause substan-
tial heat and energy loss to the environment. Bendig
et al. classified energy loss as avoidable and unavoid-
able heat loss, and waste heat was defined as avoid-
able heat loss [5]. Heat recovery of waste heat is
considered as a very promising strategy for enhanc-
ing the overall energy efficiency in a process [6].
Recovery and reuse of waste heat can be applied at
the process level as well as at the plant level [7].
Recent studies applied waste heat recovery and
reuse strategy on various levels - industries, industrial
zones and continuous casting process [8–10]. In com-
plex systems, which consist of multiple processing
plants, each plant has its own independent operating
and maintenance schedules, which sets difficulties
and limitations during integration process. Also, an
important factor to consider is the distance between a
M.M. ZARIĆ et al.: TARGETING HEAT RECOVERY… Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017)
74
heat sink and a heat source, including heat loss due
to transport.
Process level heat recovery is usually being
referred to as a direct heat recovery, where heat rec-
overy is performed between hot and cold process
streams. This is not always achievable, due to many
practical implementation issues such as: control-
lability, possibility of generating hazards, different
operating scenarios, long distance between process
streams, etc. Plant level heat recovery helps over-
come the shortcomings of process level heat recovery
and is often defined as indirect heat recovery. This
type of integration is performed via intermediate fluids
(such as steam, hot oil, flue gas, etc.), and provides
advantages such as operational flexibility, control-
lability, as well as avoidance of hazards. However,
comparing the two levels of heat recovery, during
indirect heat recovery temperature driving forces are
reduced, which often results in lower heat recovery
and less energy saved [11]. Wang et al. considered
both direct and indirect heat integration, as well as
combination of both direct and indirect heat recovery
involving the features of both giving more design
options [12]. Recently, Miah et al. have maximized
the heat recovery of diverse production lines by com-
bining the direct and indirect heat exchange from
zonal to factory level [13].
The Pinch analysis (PA) method introduced by
Linnhoff et al. is one of the most commonly used
methods that can estimate possible heat recovery
within an individual process [14]. PA methods are
based on thermodynamic principles to determine
maximum heat recovery potential, and hence can be
used to construct efficient designs for heat exchanger
network (HEX) [15-20]. The main disadvantage of this
method is that it does not allow options for forbidden
or preferred matches between process streams.
Moreover, the total site analysis (TSA) method has
been introduced to improve heat recovery within a
given plant [21]. TSA applies energy integration
between multiple processes to enable maximum indi-
rect heat recovery potential. Processes within a plant
are considered to be supplied by a common utility
system, which provides the required heat and power.
Dhole and Linnhoff introduced TSA to establish tar-
gets for heat recovery by integrating processes and
optimizing the quantity of utility used in plants [21].
Practical implementation of TSA as an energy conser-
vation concept has been improved in previous studies
[22-24]. Chew et al. pointed out that TSA methods
should be extended to design, operability, reliability/
/availability, maintenance, regulatory policy, as well
as economics of utility systems [25]. Further studies
investigated total site heat integration considering
pressure drops and utilizing the TS Heat Integration
profiles for assessing the process modifications to
decrease capital costs [26,27]. Improvement of TSA
method has been investigated by Varbanov et al.,
who implemented a modification to enable the use of
different minimum temperature differences [28]. New
graphical approaches are proposed to present better
clarity for the quantitative interaction between the pro-
cess and utility system targets [29]. TSA methods
mainly rely on graphical techniques that cannot pro-
vide precise estimations [3,28,30]. Moreover, steam
superheating considerations are overlooked, often
causing heat recovery potential to be overestimated
[31]. Mathematical programming techniques have
been developed to overcome the aforementioned
drawbacks TSA in identifying optimal HEXs designs.
Papoulias and Grossman proposed a mathematical
programing technique that identifies an option for for-
bidden matches, whereas Becker and Marachel deve-
loped a mathematical programming model by adding
an option for intermediate heat transfer units [32,33].
Liew et al. developed an extended methodology TS
Heat Integration in a steam system that considers the
water sensible heat (boiler feed water preheating and
steam superheating during steam generation) using a
systematic numerical tool [34]. Several reviews of rel-
evant publications on heat exchange synthesis and
process integration have been published [35,36].
Recent studies have introduced an optimal design
approach and multi-objective optimization methods of
cogeneration systems based on exergo-economic
and exergo-environmental parameters [37,38].
The goal of this study is to develop an approach
to maximize energy efficiency of an industrial zone
through waste heat recovery and reuse, via indirect
heat integration. Previous work done by Stijepovic
and Linke proposed a method targeting maximal
waste heat recovery and reuse across decentralized
utility system [9]. This method is based on a study of
an industrial zone consisting of independent plants
operating multiple processes. It is considered that
each plant is served by an independent utility system
and each plant has been optimized for energy effi-
ciency. In this study, a new approach is developed to
target minimum energy requirements in an industrial
zone supplied by a centralized utility system instead
of decentralized utility system. To reveal potential
waste heat streams, the concept of exergy was used,
a method described by Stijepovic and Linke [9]. The
transshipment model is adopted to estimate maximal
heat recovery from recognized waste heat streams.
The study is based on targeting maximum waste heat
M.M. ZARIĆ et al.: TARGETING HEAT RECOVERY… Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017)
75
recovery potentials in a centralized utility system prior
to the design of optimal network. Furthermore, the
proposed approach may assist in the decision making
process regarding the retrofitting strategy for utility
system configuration in an industrial zone. For
example, it can reveal whether introducing a central-
ized utility system between multiple plants is justified.
Problem definition
A plant usually comprises several processes,
where a common utility system provides overall heat
for all processes. In this study, an industrial zone
consists of multiple independent plants and it is con-
sidered that all plants are served by one centralized
utility system, which provides required heat and
energy for all plants, instead of each plant having a
separate utility system. Both initial and target tempe-
ratures, heat capacities and heat loads are specified
for each hot and cold process stream. Centralized
utility system uses fossil fuel to generate very high
pressure steam (VHP), where thermodynamic state is
reduced to required utilities pressures and tempera-
tures by let-down stations. Depending on the require-
ments of the process, different types of high pressure
(HP), medium pressure (MP), and low pressure (LP)
utilities are generated. Heat demands for required
utilities define the overall consumption of fossil fuel.
Generally speaking, two types of process
streams exist: 1) hot process streams have to be
cooled down to a specified temperature, and 2) cold
process streams have to be heated to a specified
temperature. Hot process streams are cooled down
using cold process streams. Similarly, cold process
streams are heated up using hot process streams.
Any hot or cold process steams which are unable to
reach their specified temperature solely via heat
exchange must be cooled down or heated up addi-
tionally by using external cold or hot utilities, respect-
ively. Any excess heat that is released to cold utilities
may then be used to provide heat for another plant
within the industrial zone. Therefore, excess heat can
be used to generate utilities, which can then be used
as a heat source in another plant within the industrial
zone [31].
Process streams that eject excess heat into cold
utilities can be identified as a potential heat source,
as shown in Figure 1. Cold utilities mainly use air or
cooling water to cool down a hot process stream.
Heat released to cold utilities is considered waste
heat (Figure 1a), which can later be used as a heat
source to generate utilities subsequently referred to
as “recoverable utility” (Figure 1b). The role of intro-
duced recoverable utilities in heat integration is to
transfer heat from a process where excess heat is
identified to a process with heat deficit. The generat-
ed recoverable utility replaces required hot utilities,
either totally or partially, leading to decrease in overall
industrial zone heat demands set prior to heat integ-
ration.
Figure 2 illustrates a utility system with indirect
heat integration between processes within an indus-
trial zone. The utility system generates VHP steam
that is converted by let-down stations to hot utilities
HP steam, MP steam, and LP steam. Recoverable
utilities are generated using excess heat from hot
process streams and are directed to hot utility steam
headers, where they are linked with the specified
steam from let-down station. This leads to decrease
in demands of generating VHP as well as the usage
of fossil fuels.
Model formulation
The proposed model is depicted in the heat cas-
cade diagram in Figure 3. Hot streams represent heat
sources, and recoverable utilities represent heat sinks
during heat exchange which is carried out in tempe-
rature intervals k. Temperature intervals account for
the thermodynamic constrains that control heat trans-
fer in order to guarantee feasible heat transfer con-
ditions in each interval. This has been ensured by
partitioning the entire temperature range into small
temperature intervals, which are defined by initial and
final temperature of present streams. The entire
range of temperature values is set in decreasing
order in the cascade diagram. There are k tempera-
ture intervals for k+1 values of temperature, each
represented by a separate block of heat exchange
between sources (hot process streams) and sinks
Figure 1. a) Hot process stream cooled by cold utility; b) hot process stream generating recoverable utility.
M.M. ZARIĆ et al.: TARGETING HEAT RECOVERY… Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017)
76
(recoverable utilities). Index sets, parameters and
variables that are required by the problem are defined
in the nomenclature.
Each hot process stream with an excess heat
content at temperature interval, k, Qi,k, can exchange
heat content with any recoverable utility, j, that is pre-
sent in that particular temperature interval, k. The
transferred heat from hot process stream i to recover-
able utility j in temperature interval k is represented by
Qi,,j,k. Part of the heat content of hot process stream
that has not been exchanged in the temperature inter-
val k, is transferred to the next, lower temperature
interval, k+1, as a heat residual, Ri,k (Figure 3).
The proposed approach is defined as follows:
VHPminOF m (1)
Figure 2. Utility system with indirect heat integration.
Figure 3. Heat cascade diagram.
M.M. ZARIĆ et al.: TARGETING HEAT RECOVERY… Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017)
77
Set to constrains:
, , 1 , , ,
1
JH
i k i k i j k i k
j
R R Q Q , ,i Hk k TI (2)
, , ,
1
IC
i j k j k
i
Q Q , ,j Ck k TI (3)
req 0jm m , req ,RHU j Ck (4)
1 1
REQ JVHP
req j
req j
m m m (5)
, , ,, , 0i k i j k jR Q m (6)
,0 0iR (7)
The defined objective function (OF) minimizes
the mass flow rate of generated VHP steam, mVHP,
defined by Eq. (1). The heat balance for one tem-
perature interval, k, is described in Figure 4.
Figure 4. Heat flows during one temperature interval.
Each temperature interval k has two inputs, heat
content from all hot stream (
,1
IHi k
i
Q ), and residual
heat from the previous temperature interval
(
, 11
I
i ki
R ). Moreover, each temperature interval has
two outputs: heat transferred to recoverable utility
(
,1
JCj k
j
Q ), and surplus heat, known as heat residual
(
,1
I
i ki
R , Eq. (2)) [39].
During heat transfer, each recoverable utility
undergoes a phase change. As a result, each rec-
overable utility has a different heat capacity value
throughout the temperature range. The heat transfer-
red to recoverable utility, Qj,k, depends on the heat
capacity at that particular temperature interval (Figure
3). Heat demands of recoverable utility, Qj,k, is equi-
valent to the sum of the transferred heat from hot
streams,
, ,1
I
i j ki
Q , to particular recoverable utility, j,
which is defined as one of the constrains in Eq. (3).
After generating recoverable utilities, j, (mass
flow rate, mj), they are set to replace the specific
required hot utility either totally or partially, req (mass
flow rate, mreq) (Figure 2). Therefore, mass flow rate
of recoverable utility, mj must be less than or equal to
the mass flow rate of corresponding required utilities,
mreq. This condition is set as one of constrains, and it
is defined by Eq. (4). All required hot utilities, req, are
supplied by VHP steam, which is generated in com-
mon utility system using fossil fuels. Hence, gener-
ated recoverable utilities, j, replace the required hot
utilities, req and consequently reduce demands for
generated VHP steam, mVHP, in common utility (Eq. (5)).
Known parameters are set of heat sources,
,Hi kQ , i Hk , at each temperature interval and the
mass flow of each required hot utility, mreq:
, , 1( )Hi k p i k kQ C , ,i Hk k TI (8)
, ,
, , ,
/ ( ( )
( )),
req req vap req sat req
pG req uh req sat req
m Q H
c req RHU (9)
Optimized variables are mass flow rates of rec-
overable utilities, mj, j Ck , which define the heat
content, ,Cj kQ , represented in Eq. (10):
, , 1( )Cj k j p j k k kQ m c , ,j Ck k TI (10)
As aforementioned, recoverable utilities undergo
a phase change during the heat transfer therefore
phase state and specific heat capacity depends on
the specified temperature interval. This is represented
in Eq. (11) by three options for each of the phase
state: liquid phase (L), vaporization (VAP) and super-
heated state (SS):
,
, ,
,
,
, 1
, 1
L
VAP
SS
in satp j k k
sat satp j k p j k k
sat outp j k k
c T T
c c T T
c T T
(11)
,j Ck k TI
where
, / Tp j k vap jc H , 1T K (12)
During the vaporization, saturation temperature,
Tsat, is constant, but in this model formulation it is
approximated that the vaporization is happening
throughout 1 K. The values are obtained from the
standard thermodynamic tables [40]. Each specific
heat capacity value, cp, is calculated as an average
value of the two values: at the initial and final tempe-
rature of the phase state.
M.M. ZARIĆ et al.: TARGETING HEAT RECOVERY… Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017)
78
Case study
The proposed model has been applied to an
industrial case study [9]. An industrial zone consists
of four independent petrochemical plants, where each
plant consists of one or more processes. Each plant is
served by an independent utility system and each
plant has been optimized for energy efficiency. The
methodology to develop optimal heat recovery in
decentralized system is presented in previous study
[9]. In this study, the industrial zone is considered to
be served by one centralized utility system, which
provides required heat for all processes, instead of
each plant having independent utility system.
The fossil fuel used in the utility system is nat-
ural gas. Combustion of natural gas generates VHP
steam, which is expanded by a let-down station to
lower pressure and temperature: HP, MP or LP
steams, in order to satisfy the requirements set by
processes in each plant.
Data acquisition
For required optimization, three sets of data are
necessary. The first set of data represents hot pro-
cess streams and excess heat that can be reused.
The hot streams are defined by the Stijepovic and
Linke method using the concept of exergy is applied
[9]. There are seven hot process streams that are
recognized as potential heat sources and their pro-
perties are summarized in Table 1.
Table 1. Data for hot process streams
Stream number Plant number Tin / C Tout / °C H / kW
1 1 230 60 30000
2 1 200 55 20000
3 1 55 40 10000
4 2 200 60 20000
5 3 330 60 25000
6 3 300 70 20000
7 4 180 60 12000
The second set of data represents required
utility usage in each plant. The third set of data rep-
resents properties for recoverable utilities, which are
generated in common utility via heat exchange with
hot process streams. For each of the nine required
utilities, nine recoverable utilities are introduced in the
system. The second and third set of data are sum-
marized in Table 2. All represented utilities only use
steam at different levels.
The approach assumes that all process plants in
an industrial zone are linked through the central utility
system. Hot process streams (Table 1) generate rec-
overable utilities (Table 2) in a common utility. They
are then transported to a specific plant to replace the
specified required utilities either totally or partially
(Table 2).
In order for recoverable utility to replace the
required utility, conditions like temperature and pres-
sure must match. Pressure of recoverable utility must
be the same as the pressure of required utility and the
target temperature, i.e., outlet temperature, of rec-
overable utility must match the temperature of
required utility header. In order for heat transfer to be
feasible the minimum allowable temperature differ-
ence, Tmin, must be introduced for recoverable
utilities: 30 C for HP and MP steams, and 15 C for
LP steams.
For illustration purposes, the data used are
imaginary but comparable to data observed in exist-
ing production plants.
RESULTS AND DISCUSSION
Equations (1)-(12) form linear problem that is
solved using LINGO software [41]. The objective
function is to minimize VHP steam generated in the
centralized utility system through waste heat recovery
and reuse via indirect heat integration. Variables that
are optimized are mass flow rates of recoverable util-
ities.
Table 2. Data for required utilities
Str. No. Plant No. Required utilities Recoverable utilities
Tin / C Tout / C (Tsat) Heat required, kW Mass flow rate, kg/s Tsat / C Tin / C Tout / C
1 1 280 240 23000 12.09 240 108 280
2 1 240 200 29000 14.14 200 108 240
3 2 320 240 20000 9.88 240 108 320
4 2 250 220 39000 20.02 220 108 250
5 2 200 170 21000 9.89 170 108 200
6 2 150 120 28000 12.37 120 108 150
7 4 220 190 19000 9.23 190 108 220
8 4 150 130 25000 11.27 130 108 150
9 4 120 108 23000 10.18 108 108 120
M.M. ZARIĆ et al.: TARGETING HEAT RECOVERY… Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017)
79
Figure 5 represents indirect heat integration for
this case study. As the depicted utility system burns
natural gas, it generates VHP steam, which is red-
uced to required HP steam, MP steam and LP steam.
Recoverable utilities are generated through heat
exchange from hot process streams, as it is shown in
Figure 5. Generated recoverable utilities obtained by
optimization are four utilities with lowest saturated
temperatures: utilities 5, 6, 8 and 9. Recoverable util-
ities are directed to steam headers, where they are
linked to the corresponding required utilities. The rep-
lacement of required utilities decreases demands of
total VHP steam generated in utility system, as well
as decreases consumption of natural gas.
The obtained results of optimized variables,
mass flow rate of recoverable utilities, mj, are pre-
sented in the Table 3. Results of the optimization
show that all generated recoverable utilities are low
Figure 5. Case study system with indirect heat integration.
Table 3. Comparing mass flow rates of required utilities before and after heat integration
Utility Tsat / C Tout / C Mass flow rate of recoverable utility,
mj / kg s-1
Mass flow rate of required utility before
heat integration, kg/s
Mass flow rate of required utility
after heat integration, kg/s
1 240 280 0 12.09 12.09
2 200 240 0 14.14 14.14
3 240 320 0 9.88 9.88
4 220 250 0 20.02 20.02
5 170 200 1.16 9.89 8.73
6 120 150 12.37 12.37 0
7 190 220 0 9.23 9.23
8 130 150 11.27 11.27 0
9 108 120 10.18 10.18 0
Total – – 34.98 109.07 74.09
M.M. ZARIĆ et al.: TARGETING HEAT RECOVERY… Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017)
80
grade utilities: utilities 5, 6, 8 and 9. Three out of four
generated utilities are totally generated with targeted
mass flow rate. These results are expected, since the
objective function is set to minimize the quantity of
used VHP steam as heat resource.
Additionally, in Table 3 are represented data for
the mass flow rates of required utilities before heat
integration and after heat integration. The mass flow
rate of each required utility after heat integration rep-
resents difference between mass flow rate of required
utility before heat integration and mass flow rate of
corresponding recoverable utility, which is set to rep-
lace required utility. The requested total mass flow
rate by required utilities before the heat integration is
109.07 kg/s. After the heat integration, total mass flow
rate is reduced to 74.09 kg/s, which is 32.07% less
than before the integration. Therefore, demands for
generating VHP steam in utility system is reduced by
32.07%.
Heat determined by required utilities supplied by
the utility system, before and after heat integration is
presented in Table 4. As it can be observed, the opti-
mization results show that 78.463 kW heat can be
recovered via heat integration, which represents
34.56% of heat supplied by steam from centralized
utility system.
Less demands for generating VHP steam, after
the heat integration, leads to less demands for
combustion of natural gas in the centralized utility
system. The consumed natural gas is compared
before and after heat integration in order to evaluate
the amount of natural gas that can be saved through
heat integration in this case study. Natural gas
consumption is defined by next equation:
VHP , , VAP
, ng ng
( ( )
( ))
pL j sat j in
pG j out sat
m c H
c m LHV (13)
where mVHP is total mass flow rate of VHP steam that
is generated from burning natural gas (kg s-1), mng
mass of natural gas (kg s-1), and LHV is low heat
value for natural gas (kJ kg-1). Comparing the amount
of natural gas consumed before (7.89 kg s-1) and after
heat integration (5.36 kg s-1), the saved amount of
natural gas during the heat integration is 2.53 kg s-1.
CONCLUSION
A method for waste heat recovery and reuse via
heat integration is developed in order to increase
energy efficiency and decrease the use of fossil fuels.
Industrial zones usually consist of multiple independ-
ent plants, where each plant is supplied by an inde-
pendent utility system, as a decentralized system. In
this study, a method is applied to target minimum
energy requirements where an industrial zone is sup-
plied by a centralized utility system instead of decen-
tralized utility system. The proposed method is based
on linear programming problem (LP). It was tested out
on a case study and the results indicate that fossil fuel
savings are achieved, and energy efficiency of an
industrial zone is increased by recovering and reusing
waste heat via indirect heat integration. This approach
can be used in the decision making process in retro-
fitting strategy for utility system configuration in an
industrial zone.
Acknowledgments
This work was supported by Ministry of Edu-
cation, Science and Technology Development, Rep-
ublic of Serbia Project no. OI172063.
Nomenclature
Indices
i - hot stream
j - recoverable utility
k - temperature interval
req - required hot utility
pc - phase change of recoverable utility
Sets
Hk = {i | hot stream i supplies heat at interval k, i =
= 1,…,I}
Ck = {j | recoverable utility j demands heat at interval
k, j = 1,...,J}
TI = {k | temperature interval, k = 1,…,K}
RHU = {req | required hot utility, req = 1,…,REQ}
Table 4. Comparing data of required utilities before and after heat integration, heat, kW
Utility No. Heat required before the heat integration Heat required after the heat integration Heat recovered via heat integration
5 21000 18536 2464
6 28000 0 28000
8 25000 0 25000
9 23000 0 23000
M.M. ZARIĆ et al.: TARGETING HEAT RECOVERY… Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017)
81
Parameters
,Hi kQ - heat content of hot stream i at temperature
interval k, kW
reqQ - heat content of required hot utilities req, kW
mreq - mass flow rate for required utility, kgs-1
Cp,i - heat capacity flowrate of the hot process
streams, i, kJ K-1
cpG,req - specific heat capacity of the gas phase, G, for
required utility, req, kJkg-1 K-1
Hvap,req - latent heat of vaporization for required utility
stream, req, kJ kg-1
θuh,req, θsat,req - temperature of utility header and sat-
uration temperature of required utility, K
Variables
mj - mass flow rate for recoverable utility, kg s-1
,Cj kQ - heat content demanded by recoverable utility j
at temperature interval k, kW
, ,i j kQ - heat exchanged between hot stream i and
recoverable utility j at temperature interval k, kW
,i kR - heat residual of hot stream i at temperature
interval k, kW
θk , θk-1 - inlet and outlet temperatures of the each
temperature interval, k, K
cp,j,k, - specific heat capacity of recoverable utility j, kJ
kg-1 K-1
mVHP - mass flow rate of generated VHP steam, kg s-1
REFERENCES
[1] Core Writing Team, R.K. Pachauri, L.A. Meyer (Eds.),
IPCC, 2014: Climate Change 2014: Synthesis Report.
Contribution of Working Groups I, II and III to the Fifth
Assessment Report of the Intergovernmental Panel on
Climate Change, IPCC, Geneva, Switzerland, pp. 151
[2] C. Mandil, IEA. Tracking industrial energy efficiency and
CO2 emissions, https://www.iea.org/publications/freepub-
lications/publication/tracking_emissions.pdf (accessed
30. March 2014)
[3] S. Perry, J. Klemeš, I. Bulatov, Energy 33 (2008) p. 1489-
–1497
[4] J.J. Klemes, P.S. Verbanov, Clean Technol. Environ.
Policy 15 (2013) 417-422
[5] M. Bendig, F. Marechal, D. Favrat, Appl. Therm. Eng. 61
(2013) 134-142
[6] S. Ahmad, D.C.W. Hui, Comput. Chem. Eng. 15 (1991)
809-832
[7] A.K. Kralj, P. Glavič, M. Krajnc, Appl. Therm. Eng. 22
(2002) 1259-1269
[8] G.Y. Ma, J.J. Cai, W.W. Zeng, H. Dong, Energy Procedia
14 (2012) 1022–1028
[9] M.Z. Stijepovic, P. Linke, Energy 36 (2011) 4019–4031
[10] A. Villar, J. Parrondo, J.J. Arribas, Clean Technol.
Environ. Policy 17 (2015) 555–562
[11] S. Bandyopadhyay, J. Varghese, V. Bansal. Appl. Therm.
Eng. 30 (2010) 6–14
[12] Y. Wang, C. Chang, X. Feng, Energy 90 (2015) 56-67
[13] J.H. Miah, A. Griffiths, R. McNeill, I. Poonaji, R. Martin, A.
Yang, S. Morse, Appl. Energy 132 (2014) 451-464
[14] B. Linnhoff, Chem. Eng. Res. Des. 71 (1993) 503-522
[15] J. Klemeš, F. Friedler, I. Bulatov, P. Verbanov, Sustain-
ability in the process industry – Integration and optimiz-
ation, McGraw-Hill Education, New York, 2010, pp. 45–
-105
[16] I.C. Kemp, Pinch analysis and process integration,
Butterworth-Heinermann, Oxford, 2007, pp. 1-157
[17] U.V. Shenoy, Heat Exchanger Network Synthesis: Pro-
cesses Optimization by Energy and Resource Analysis,
Gulf Professional Publishing, Houston, TX, 1995, pp.
117–161
[18] M.M. El-Halwagi, Process Integration, Elsevier: Academic
Press, San Diego, CA, 2006, pp. 231-252
[19] B. Linnhoff, D.W. Townsend, D. Boland, G.F. Hewitt,
B.E.A. Thomas, A.R. Guy, R.H. Marsland, A user guide
on process integration for the efficient use of energy,
IChemE, Rugby, 1982, p. 247
[20] S.R.W. Alwi, Z.A. Manan, Chem. Eng. J. 162 (2010) 106-
–121.
[21] V.R. Dhole, B. Linnhoff, Comput. Chem. Eng. 17 (1993)
S101-S109
[22] R. Hackl, S. Harvey, Chem. Eng. Trans. 39 (2014) 103–
–108
[23] K.H. Chew, J.J. Klemeš, S.R. Wan Alvi, Z.A. Manan,
Appl. Therm. Eng. 78 (2015) 731–739
[24] R. Hackl, E. Andersson, S. Harvey, Energy 36 (2011)
4609–4615
[25] K.H. Chew, J.J. Klemes, S.R. Wan Alwi, Z.A. Manan,
Appl. Therm. Eng. 61 (2013) 17-25
[26] K.H. Chew, J.J. Klemeš, S.R.W. Alwi, Z.A. Manan, A.P.
Reverberi, Energies 8 (2015) 1114–1137
[27] K.H. Chew, J.J. Klemeš, S.R.W. Alwi, Z.A. Manan, Appl.
Therm. Eng. 89 (2015) 1023-1032
[28] P.S. Varbanov, Z. Fodor, J.J. Klemeš, Energy 44 (2012)
20–28
[29] L. Sun, S. Doyle, R. Smith, Clean Technol. Environ.
Policy 16 (2014) 1235-1243
[30] J. Klemeš, V.R. Dhole, K. Raissi, S.J. Perry, L. Puigjaner,
Appl. Therm. Eng. 17 (1997) 992-1003
[31] R. Smith, Chemical process design and integration, John
Wiley & Sons Ltd., Chichester, 2005, p. 489
[32] S.A. Papoulias, I.E. Grossman, Comp. and Chem. Eng. 7
(1983) 707-721
[33] H. Becker, F. Maréchal, Comput. Chem. Eng. 37 (2012)
104-118
[34] P.Y. Liew, S.R.W. Alwi, J.S. Lim, P.S. Varbanov, J.J. Kle-
meš, Z.A. Manan, J. Cleaner Prod. 77 (2014) 94-104
[35] K.C. Furman, N.V. Sahinidis, Ind. Eng. Chem. Res. 41
(2002) 2338-2370
[36] J.J. Klemeš, P.S. Varbanov, Z. Kravanja, Chem. Eng.
Res. Des. 91 (2013) 2037–2053
M.M. ZARIĆ et al.: TARGETING HEAT RECOVERY… Chem. Ind. Chem. Eng. Q. 23 (1) 7382 (2017)
82
[37] P. Navid, M.H.K. Manesh, A.M.B. Marigorta, Clean
Technol. Environ. Policy 16 (2014) 1045-1065
[38] H. Janalizadeh, M.H.K. Manesh, M. Amidpour, Clean
Technol. Environ. Policy 17 (2015) 103-117
[39] L.T. Biegler, I.E. Grossmann, A.W. Westerberg, System-
atic Methods of Chemical Process Design, Prentice Hall,
Upper Saddle River, NJ, 1997, pp. 527-563
[40] C.F. Beaton, Heat Exchanger Design Handbook, http://
//thermopedia.com/content/1150Begal (accessed 2 Nov
2013)
[41] LINDO System Inc., LINGO software.
MILANA M. ZARIĆ1
MIRKO STIJEPOVIC1
PATRICK LINKE2
JASNA STAJIĆ-TROŠIĆ1
BRANKO BUGARSKI3
MIRJANA KIJEVČANIN3
1Institut za hemiju, tehnologiju i
metalurgiju, Univerzitet u Beogradu,
Njegoševa 12, 11000 Beograd, Srbija 2Department of Chemical Engineering,
Texas A&M University at Qatar, P.O.
Box 23874, Education City, Doha,
Qatar 3Tehnološko-metalurški fakultet,
Univerzitet u Beogradu, Karnegijeva 4,
11120 Beograd, Srbija
NAUČNI RAD
REKUPERACIJA TOPLOTE U INDUSTRIJSKOJ ZONI
Sa ciljem da se smanji upotreba fosilnih goriva u industrijksim sektorima, a da zahtevi pro-
cesa proizvodnje budu zadovoljeni, razvijaju se novi pristupi toplotne integracije i reku-
peracije toplote. Cilj ove studije je razvijanje pristupa koji će omogućiti povećanje energet-
ske efikasnosti u industrijskim zonama rekeperacijom otpadne toplote putem indirektne
integracije. Uobičajeno je da se industrijska zona sastoji od više nezavisnih postrojenja,
kao decentralizovani sistem, gde je svako postrojenje obezbeđen nezavisinim sistemom
pomoćnih fluida. U ovoj studiji, razvijen je novi pristup, gde se minimalizuju energetski zah-
tevi i gde se industrijska zona obezbeđuje centralizovanim sistemom pomoćnih fluida,
umesto decentralizovanog sistema. Ovaj pristup pretpostavlja da su sva procesna postro-
jenja u industrijskoj zoni povezana kroz centralizovani sistem pomoćnih fluida. Predloženi
metod je formulisan kao problem linearnog programiranja (LP). Pored toga, ovaj postupak
može se koristiti tokom odlučivanja o strategiji energetske integracije industrijskih zona.
Štaviše, predloženi metod je primenjen na studiju slučaja. Rezultati pokazuju da je moguće
ostvarenje uštede fosilnih goriva.
Ključne reči: rekuperacija toplote, energetska efikasnost, toplotna integracija, LP
programiranje.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2016) CI&CEQ
83
SANDRA RAQUEL KUNST1
LILIAN VANESSA ROSSA
BELTRAMI2
MARIELEN LONGHY1
HENRIQUE RIBEIRO
PIAGGIO CARDOSO2
TIAGO LEMOS MENEZES2
CÉLIA de FRAGA MALFATTI2
1Programa de Pós Graduação em
Engenharia de Processos e
Tecnologias (PGEPROTEC),
University of Caxias do Sul (UCS),
Caxias do Sul, RS – Brasil 2Laboratório de Pesquisa em
Corrosão (LAPEC), Federal
University of Rio Grande do Sul
(UFRGS), Porto Alegre, RS - Brasil
SCIENTIFIC PAPER
UDC 669.715:678:544:543.42
https://doi.org/10.2298/CICEQ150725013K
EFFECT OF DIISODECYL ADIPATE CONCENTRATION IN HYBRID FILMS APPLIED TO TINPLATE
Article Highlights
• Tinplate protection from corrosion
• Flexible hybrid film obtained by sol-gel process
• Electrochemical impedance spectroscopy tests
Abstract
Siloxane hybrid films are fragile and have low mechanical strength due to their
vitreous material properties. Hence, a new formulation incorporating a plasticizer
agent was developed in order to increase the layer thickness of a uniform and
homogeneous hybrid film on tinplate, and to provide flexibility to the polymeric
matrix. Tinplate sheets were coated with a hybrid film obtained from a sol-gel
process, constituted by t+he addition of the following alkoxide precursors:
3-(trimethoxysilyl) propyl methacrylate and tetraethoxysilane with 0.01 mol L-1
cerium nitrate addition. The influence of the diisodecyl adipate plasticizer con-
centration was evaluated. The films were characterized by scanning electron
microscopy, profilometry, open circuit potential monitoring, polarization curves
and electrochemical impedance spectroscopy. The results showed that all films
with diisodecyl adipate had higher electrochemical performance compared to
uncoated tinplate. However, the film with the 2% plasticizer concentrations had
the best performance in the electrochemical tests, although it had thinner layer.
Keywords: hybrid film, diisodecyl adipate, tinplate, electrochemical behaviour.
The sol-gel process is a widely researched and
used technology [1-5], as a result of the obtained
surface protection properties, the simplicity of the
formulation process and its economic viability. Among
its advantages, the following can be mentioned: the
stoichiometry is easy to control and adjust [6]; it is
possible to tailor a film with high purity and with a uni-
form distribution of the components [7]; the process
can be carried out under normal pressure and low
temperatures [8-12]. Furthermore, the sol-gel process
has been used for decades to obtain a high number of
hybrid materials, using inorganic and polymeric pre-
cursors [13,14].
Correspondence: S.R. Kunst, Programa de Pós Graduação em
Engenharia de Processos e Tecnologias (PGEPROTEC), Uni-
versity of Caxias do Sul (UCS), Caxias do Sul, RS – Brasil. E-mail: [email protected] Paper received: 25 July, 2015 Paper revised: 6 November, 2015 Paper accepted: 11 March, 2016
The sol-gel process consists of the hydrolysis
and condensation of alkoxide precursors with metal-
oxane, in order to obtain a three-dimensional siloxane
network. After deposition of the film on the substrate
by means of appropriate techniques, the film is
exposed to air at the beginning of the condensation
reaction. After a few minutes of drying, a network (gel)
is formed on the substrate [15,16]. The formed net-
work is a hybrid (organic and inorganic) [16]. Water
molecules are eliminated through sintering (densific-
ation) at an appropriate temperature, and a compact
hybrid layer is formed. Then, the film is submitted to a
final thermal treatment for structure control [17].
Hybrid films are used to coat a variety of metallic
substrates to protect them from corrosion [18,19].
Amongst these, tinplate can be mentioned. Tinplates
are used in food and non-food packaging applications
[20]. For good quality, the presence of tin oxides on
the surface and the condition of the passivation layer
are very relevant. The presence of such oxides can
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
84
alter the sheet appearance, weldability and capability
of being coated with organic films [21]. Currently,
many surface treatments for packaging are based on
chromates, as they offer excellent corrosion resist-
ance, but their application process is highly toxic.
Thus, non-toxic pre-treatment alternatives have been
developed recently to replace the chromating process
[22,23]. Hybrid films obtained by the sol-gel process
are one of those alternatives and the industrialization
of the sol-gel process involves work with real sub-
strates [16,24-26].
Research has been performed on the tinplate
coating with hybrid films. Kunst et al. [27] studied the
coating the tinplate with mono and bilayer of hybrid
films modified with polyethylene glycol (PEG). These
hybrid films were obtained from precursors 3-(trimeth-
oxysilyl)propylmethacrylate (TMSM) and tetraethoxy-
silane (TEOS) with the addition of cerium nitrate in a
concentration of 0.01 M. The hybrid films were
applied in single and double layers and cured at dif-
ferent temperatures (60 and 90 C). The hybrid film
with monolayer and cured at 90 C showed a more
compact, uniform, less porous layer and better elec-
trochemical impedance results.
Kunst et al. [28] studied the effect of the con-
centration of tetraethoxysilane (TEOS) on the pro-
tective properties of the film on tinplate substrate. The
tinplate was coated with a hybrid film obtained from a
sol-gel method constituted of the alkoxide precursors
3-(trimethoxysilyl)propylmethacrylate (TMSM) and
poly(methyl methacrylate) (PMMA) and different con-
centrations of tetraethoxysilane (TEOS). The hydro-
lysis of these films was performed at a pH value of 3
using acetic acid as a catalyst. All the studied films
have shown good performance as to corrosion resist-
ance on tinplate, but the film with ratio TEOS:MPTS
of 1 showed the best electrochemical results.
Kunst et al. [29] studied protective coatings as
hybrid films composed by different acids are studied
to improve the barrier effect against corrosion. The
hybrid films deposited on a tinplate from a sol made
up of the alkoxide precursors 3-(trimethoxysilyl)pro-
pylmethacrylate (TMSM), tetraethoxysilane (TEOS)
and poly(methylmethacrylate) (PMMA) with benzoyl
peroxide (BPO). The hybrid sols were prepared by
mixing water with three different acids: acetic
(CH3COOH), hydrochloric (HCl) and nitric acid (HNO3).
The results demonstrate that the hybrid film obtained
by acetic acid addition exhibiting the greatest
improvement the protective properties of the tinplate.
Malfatti et al. [30] showed that the addition of
Ce3+ in a hybrid films confers additional active cor-
rosion protection through a self-healing behavior that
can be verified by the reduction in corrosion rate com-
pared to hybrid films without Ce3+ addition. Thus,
considering the industrial potential for application of
the coating, it is expected that the Ce3+ should dec-
rease the corrosion process in case of damage to the
barrier layer.
In order to improve the barrier effect and coat
the tinplate uniformly and homogeneously, it is desir-
able to obtain coatings with higher layer thickness.
For this purpose, there are two possible approaches:
one is to increase the number of layers up to a limit
number to avoid delamination [31]. An alternative
approach is to increase the sol viscosity; this para-
meter can be modified either by varying the tempe-
rature to alter the hydrolysis and condensation react-
ion kinetics during the siloxane film formation, or by
introducing a plasticizer agent to modify the intrinsic
properties of the gel. The latter was chosen in this
study, due to the fact that it is easily adjustable [15].
The purpose of this work was to coat tinplate
with hybrid film, obtained from a sol-gel process. This
film was constituted by the alkoxide precursors 3-(tri-
methoxysilyl)propyl methacrylate (TMSM) and tetra-
ethoxysilane (TEOS) and was added different diiso-
decyl adipate plasticizer concentrations (0.5, 1, 2 and
4%). The films were applied using a by dip-coating
process and were cured at 60 C for 20 min. From
this study, an ideal concentration of diisodecyl adipate
plasticizer was determined for the training of a hybrid
film with protective properties superior to other films
studied.
MATERIALS AND METHODS
Surface preparation
Tinplate coupons with dimensions of 20 mm40
mm were obtained from industrial sheet, rinsed with
acetone and dried. The tinplate used in this study has
the maximum percentage composition of: 0.06C,
0.2Mn, 0.02P, 0.02S, 0.005N and 0.06Al. The aver-
age thickness of the tinplate used was 0.245 mm. The
used tinplate had yield strength ranging from 210 to
310 MPa, tensile strength of 290 to 410 MPa, hard-
ness of 51 to 59 HR 30 T D and tin coating of 3.0 to
2.0 g m-2. The plate samples were degreased with a
10 min immersion in neutral detergent at 70 C,
washed and dried.
Hybrid films elaboration
Hydrolysis reactions were conducted with the
silane precursors 3-(trimethoxysilyl)propylmethac-
rylate (TMSM, C10H20SiO5) and tetraethoxysilane
(TEOS, C8H20SiO4), with 0.01 mol L-1 cerium nitrate
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
85
addition, and water and ethanol were used as sol-
vents. The TEOS/TMSM/H2O/EtOH proportions were,
respectively, 25/4/16/55 wt.%. A diisodecyl adipate
plasticizer was added to the sol formulation in four dif-
ferent concentrations (0.5, 1, 2 and 4 wt.%); a sample
without plasticizer addition also was analyzed. The
hydrolysis time was of 24 h at 25 C. A dip-coating
process was conducted to coat the substrate with the
hydrolyzed hybrid solution, with a removal rate of 10
cm min-1 and a residence time of 5 min. After the dip-
-coating process, the hybrid film pre-treated sub-
strates were thermally oven-cured at 60±2 C for 20
min.
Table 1 presents the description of the studied
samples, and the entire protocol of the sol preparation
and coatings is detailed in the flow chart presented in
Figure 1.
Experimental techniques
Morphological characterization was performed
using a JEOL 6060 scanning electron microscope
(SEM) at an acceleration voltage of 20 kV. The
samples were observed from top surface view and in
a cross-section to determine the layer thickness. The
surface micro-roughness was evaluated in a contact
profilometer (PRO500 3D). The wettability of the hyb-
rid films was evaluated by contact angle measure-
ments through the sessile drop method in equipment
developed by the Laboratory of Corrosion Research
(LAPEC) at UFRGS. The contact angle was deter-
mined using image analysis software.
The corrosion performance of the coatings was
evaluated by open circuit potential (OCP) monitoring,
polarization curves and electrochemical impedance
spectroscopy (EIS) measurements in a 0.05 mol L-1
NaCl solution. All electrochemical tests were per-
formed in triplicate, in an environment temperature
and without aeration. This concentration is sufficiently
high to activate corrosion in a relatively short expo-
sure time but is low enough to enable the effects of
the plasticizer to be determined. Kozhukharov et al.
[32] also used 0.05 mol L-1 NaCl to ensure a suf-
ficiently low concentration to allow for the observation
of corrosion inhibitor effects. A three-electrode cell
was used to perform the evaluations, with a platinum
wire as the counter-electrode and a saturated calomel
electrode (SCE) as the reference electrode. The area
of the working electrode was 0.626 cm2. The polariz-
ation curves were collected at a scan rate of 1 mV s-1
and potential intervals between 200 mV (below OCP)
and 400 mV (above OCP).
From the extrapolation of Tafel slopes at the
polarization curves were determined the corrosion
potential (Ecorr), corrosion current (icorr), polarization
resistance (Rp) and the protection efficiency (PE) for
samples studied. PE of the coatings was determined
for the equation, where icorr and corr*i are the corrosion
current densities obtained for uncoated and coated
substrates, respectively:
Table 1. Description of the samples
Sample Description
F1A60M Tinplate coated with hybrid film with 0.5% diisodecyl adipate plasticizer addition
F2A60M Tinplate coated with hybrid film with 1% diisodecyl adipate plasticizer addition
F3A60M Tinplate coated with hybrid film with 2% diisodecyl adipate plasticizer addition
F4A60M Tinplate coated with hybrid film with 4% diisodecyl adipate plasticizer addition
F5A60M Tinplate coated with hybrid film without plasticizer addition
Fl Uncoated tinplate
Figure 1. Schematic representation of coating elaboration.
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
86
corr corr
corr
* 1 00
i iPE
i
For the EIS measurements, the samples were
available up to 96 h and the systems were monitored
by open circuit potential (OCP) for 1 h prior to each
test. The amplitude of the EIS perturbation signal was
a sinusoidal 10 mV (rms) signal, and the frequency
range was from 100 kHz to 10 mHz using a Nova®
frequency response analyzer and a Autolab PGSTAT
30 potentiostat. The results obtained were fitted using
electrical equivalent circuits (EEC) using the Nova®
software. The consistency of the experimental data
was verified with the Kramers-Kronig transform
(KKT), and data that did not match were discarded.
RESULTS AND DISCUSSION
Morphological characterization
SEM micrographs were performed for the hybrid
films with plasticizer addition F1A60M (0.5% plas-
ticizer), F2A60M (1% plasticizer), F3A60M (2% plas-
ticizer) and F4A60M (4% plasticizer), and the hybrid
film without plasticizer addition F5A60M before the
electrochemical tests. They analyses are shown in
Figure 2.
The micrographs show that the hybrid films with
less diisodecyl adipate plasticizer concentration
F1A60M and F2A60M (0.5 and 1%) were the only
samples to show cracks and the F4A60M sample
shows discontinuities in the film surface.
The presence of cracks in the films with a low
concentration of plasticizer was due to silane pre-
cursors TMSM and TEOS that tend to form a compact
tridimensional network and, due the presence of the
plasticizer, the latter is encapsulated in the network
due to the interaction forces (i.e., hydrogen bonds,
van der Waals forces or covalent bonds). Thus,
branch mobility was limited, due to the absorption of
mechanical energy. This produces a more rigid and
fragile hybrid film. The phenomenon is known as anti-
plasticizing [33]. The presence of discontinuities in the
film with the highest concentration of plasticizer can
be associated with incompatibilities between sol and
plasticizer. According to Grossman et al. [34], adipate
plasticizers typically are used from C7 to C10 and
heterogeneities can rise when the plasticizer is added
in higher concentrations for a higher number of car-
bons. As TEOS is a C8 and TMSM is a C10 molecule,
this explanation accounts for the discontinuities obs-
erved in Figure 2 (F4A60M).
The increase in thickness of the films with the
addition of diisodecyl adipate could be due the plas-
ticizer, which enters into polymeric chains, thus inc-
reasing their free volume. Adipates are external plas-
ticizers; i.e., they are additives that interact physically
with the polymer. There can be weak attractive force
between the polymeric matrix of the film and the plas-
ticizer, through weak force as hydrogen bonds and/or
van der Waals forces, though adipates do not react
chemically with the silane precursor radical. Due to its
smaller molecular size compared with the polymeric
matrix of the film, diisodecyl adipate promotes an
increase in mobility of the film, flexibility of the mole-
cules, attributed to the generated increase of free vol-
ume, enabling an increase in the layer thickness for
those systems.
The thickness of the hybrid film layer was
determined from a cross-section SEM image analysis
(Figure 2) and these results are shown in Table 2. As
can be observed, all hybrid films with diisodecyl adi-
pate addition (F1A60M, F2A60M, F3A60M and
F4A60M) showed a significant layer thickness inc-
rease, compared to the sample without plasticizer
addition (F5A60M). The F1A60M and F2A60M hybrid
films had the higher layer thickness values; however,
they also exhibited cracks (Figure 2).
However in this study, it was observed that the
hybrid films with low plasticizer concentration pre-
sented higher layer thickness values, but also exhi-
bited cracks (Figure 2). This was due to the fact that,
despite of the adipate plasticizers being external plas-
ticizers, i.e., that they only interact physically with the
polymer; this additive is esterificated by the adipic
acid after esterification, and hence they pertain a
reactive group. A problem that can rise with such
reactive groups is that they can react with the poly-
meric matrix of the coating, making the film molecular
size bigger, thereby reducing its flexibility [35], and
promoting crack formation.
The results of the contact angle measurements
for the different systems studied are shown in Table
2. According to the results, all of the hybrid films
showed hydrophilic behavior and there were no sig-
nificant differences among the contact angle values
for the systems. The samples were hydrophilic in nat-
ure and there was not a significant difference among
the contact angle values for the systems, because the
diisodecyl adipate plasticizer is not involved in the
silane hydrolysis and condensation reactions. The
predominant interactions between the plasticizer and
the polymeric matrix are hydrogen bonds or van der
Waals forces, which are weak interactions and do not
interfere in film formation.
Silane precursor-based hybrid films show hydro-
phobic behavior when they are not sufficiently cross-
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
87
Figure 2. SEM images and layer thickness of the hybrid films.
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
88
Table 2. Layer thickness, contact angle and surface roughness values for the hybrid films studied
Sample Layer thickness, µm Contact angle, Surface roughness
Ra / µm Rms / µm Peak-to-peak, µm
F1A60M 3.93±0.31 72±1.09 0.37±0.09 0.47±0.08 4.41±0.16
F2A60M 3.21±0.36 77±0.72 0.36±0.11 0.48±0.10 3.91±0.13
F3A60M 1.96±0.18 76±0.87 0.31±0.08 0.39±0.09 4.11±0.11
F4A60M 2.14±0.24 66±0.60 0.34±0.09 0.43±0.08 2.57±0.09
F5A60M 0.63±0.11 78±1.50 0.41±0.12 0.50±0.11 3.53±0.12
Fl – 73±1.90 0.43±0.10 0.51±0.10 2.40±0.11
-linked. During the curing process, siloxane bonds
form a network that hinders water penetration, con-
sequently enhancing the hydrophobic character.
Hence, the contact angle for a highly cross-linked film
is about 90. Neither silane precursors hydrolysis nor
cross-linking (polycondensation) are completed during
curing at 60 C [36]. Subsequently, therefore, non-
-hydrolyzed ester and hydrophilic –OH groups are pre-
sent in the hybrid films structures and favor water
absorption [37].
All of the studied hybrid film samples contained
Ce in their formulation and in accordance with a lite-
rature survey performed by Palomino et al. [38], the
results suggest that this behavior was due to the joint
action of two factors: i) improved barrier properties
imparted to the silane film due to the incorporation of
silica particles, which would block preferential path-
ways for electrolyte penetration, ii) the higher degree
of polymerization and increased thickness of the sil-
ane layers, due to the presence of cerium ions. The
direct influence of Ce can be observed in the results
of contact angle for F5A60M sample (without plasti-
cizer) where the presence of Ce promoted the form-
ation of a more hydrophobic film than was the case
with the other films.
Roughness values are summarized in Table 2,
where Ra stands for arithmetic average, Rms for aver-
age square roughness, and Ry for maximum rough-
ness or peak to trough size. There was no evident
consistent variation in the roughness values, due to
the low resolution capability of the technique (i.e., a
variation of only micrometers). However, all of the
hybrid films with plasticizer addition showed better
roughness values compared to the F5A60M sample
(without adipate) and to uncoated tinplate.
Electrochemical characterization
Open circuit potential (OCP) monitoring was
conducted in a 0.05 mol L-1 NaCl solution to verify the
potential variation with time. These results are shown
in Figure 3a. Also, polarization curves for all the sys-
tems are presented in Figure 3b. From Tafel slope
extrapolations (Table 3), the corrosion current density
(icorr), corrosion potential (Ecorr) and polarization resist-
ance (Rp) values were determined.
Open circuit potential values (Figure 3a) for 1 h
of immersion indicate that all hybrid films had poten-
tials shifted upwards in relation to the uncoated tin-
plate (Fl), in the following order: F1A60M (-390 mV) >
F4A60M (-440 mV) > F5A60M (-448 mV) > F2A60M
(-450 mV) > F3A60M (-468 mV) > Fl (-509 mV); that
is, the hybrid films presented less active potentials
and promoted the formation of a barrier between the
substrate and the medium.
Furthermore, the potential for the F1A60M sys-
tem shifted in the positive direction (–390 mV) com-
Figure 3. a) Open circuit potential and b) polarization curves in a 0.05 mol L-1 NaCl solution for all studied hybrid films
and for tinplate substrate.
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
89
pared to the uncoated tinplate sample Fl (–509 mV);
i.e., this difference in potential value results from a
barrier effect from the hybrid film [39]. Moreover, the
equilibrium potential that was observed for the tinplate
substrate after 1 h immersion (–509 mV vs. SCE) is
associated with the formation of a Sn(II)-oxide/hydro-
xide layer [40,41].
With regard to the polarization curves (Figure
3b), it is observed that all of the hybrid films (F1A60M,
F2A60M, F3A60M, F4A60M and F5A60M) showed a
decrease (two orders of magnitude) in the corrosion
current density (icorr) in comparison with the uncoated
tinplate sample, which denotes the protective action
of those films. The polarization curves also showed
an increase by almost one or two orders of magnitude
in terms of the polarization resistance value (Table 3).
Analyzing the hybrid films only, the siloxane-adi-
pate film obtained with a 2% concentration of diiso-
decyl adipate, which corresponds to the sample
F3A60M (icorr = 0.914 mA m-2 and Rp = 73.2 Ω m2)
showed the best performance of the studied hybrid
films, yet only marginally so.
From the polarization curves it can be observed
that the tinplate substrate showed higher values of
current density (icorr = 47.1 mA m-2) and a lower pola-
rization resistance (Rp = 0.554 m2) compared to the
studied hybrid films.
From polarization tests showed that all of the
hybrid films exhibited better protective performance
than the sample without a hybrid film. This occurs due
to the covalent bonds of the organic and inorganic
precursors, demonstrating the synergistic effect of the
TMSM and TEOS precursors’ presence in the hybrid
film.
Of the hybrid films, the sample with 2% plas-
ticizer performed better, due to the presence of weak
bond strength between the plasticizer and the silane
precursors radicals and the increased layer thickness.
However, the film was more rigid and fragile [35].
The poor performance of uncoated tinplate
could be related to the presence of tin oxide/hydro-
xide that partially passivates the metallic surface. The
passivity breakdown at –430 mV vs. SCE is related to
pitting attack that is induced by the chloride effect
[42]. Galic [43] suggests that the adsorption of chlo-
ride ions at the oxide-electrolyte interface leads to the
formation of a tin-oxychloride film, with less protective
behavior.
Nyquist and Bode plots obtained by electroche-
mical impedance spectroscopy tests for 24 and 96 h
of immersion in 0.05 mol L-1 NaCl solution for all the
studied systems and for tinplate are shown in Figure
4. Table 4 shows the RHF (resistance for high fre-
quency) and RCP (resistance caused by corrosion pro-
cess).
The impedance results obtained from the
Nyquist plot (Figure 4) show higher resistance values
for the sample F3A60M (RHF = 472 k cm2) after 24 h
of immersion. This is due to the fact that these coat-
ings had no cracks or discontinuities, as was obs-
erved on the SEM images (Figure 2). However, after
96 h of immersion the resistance for the system
F3A60M (RHF = 312 k cm2), the resistance dim-
inished by a factor of two. Furthermore, the F3A60M
sample showed a higher phase angle (around 68 -
Figure 4) and a higher impedance modulus (around
log |Z| = 5.25 - Figure 4) values, after 96 h of immer-
sion.
On the other hand, the F1A60M sample, which
had a high layer thickness and had the addition of
0.05% plasticizer showed lower resistance values
after 24 (RHF = 92.71 k cm2) and 96 h (RHF = 4.31 k
cm2) of immersion. These reveal the fragility of this
coating, with cracks that allow permeation of the elec-
trolyte through the film and towards the substrate.
The EIS tests results indicated that the good
electrochemical performance with 2% of diisodecyl
adipate plasticizer occurs because the plasticizer con-
centration is the most suitable to increase the free
volume, allowing the movement of the polymeric chain
and giving flexibility to the organic radicals bound to
the silicon atom. Also, plasticizer inclusion avoids
interactions between the organic radicals in the silane
precursors, enhances the mobility of the chain ext-
remities and improves the flexibility of the system.
Table 3. Obtained data from Tafel extrapolation
Sample icorr / mA m-2 Ecorr / mV Rp / m² PE / %
F1A60M 0.566 -417 72.5 98.80
F2A60M 0.574 -464 56.0 98.78
F3A60M 0.914 -504 73.2 98.06
F4A60M 0.778 -451 42.1 98.35
F5A60M 0.570 -497 68.0 98.79
Fl 47.1 -581 0.554 -
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
90
EIS tests also revealed the poor performance of
the hybrid film with less plasticizer concentration.
Although the formulation of these systems has pro-
moted an increase in the layer thickness (Figure 2
and Table 2) due to the plasticizer addition, a weak
and porous structure resulted, due to the formation of
intertwined molecules liked only weak bonds (hydro-
gen bonds). Moreover, it restricted the mobility of
small branches in the chain, causing cracks on the
films, which contribute to the poor corrosion resist-
ance of these samples and their consequent inability
to resist long periods of immersion.
It was observed that after 96 h of immersion the
hybrid films showed a decrease in their protective
action, which reduced their efficiency in the corrosive
environment. This behavior was due the permeability
of the film, which occurred during the test and red-
uced the barrier properties of the film [44].
Figure 4. Nyquist and Bode plots obtained for hybrid film coated and uncoated tinplate in NaCl 0.05 mol L-1 solution for:
a) 24 and b) 96 h of immersion.
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
91
Equivalent circuit models can be used to explain
the electrochemical impedance data obtained from
the EIS tests. These models, which use a com-
bination of resistance, capacitance and other elec-
trical elements to simulate the coating response, can
be helpful in providing a clear understanding of the
response of the electrochemical system [45]. In this
work, two equivalent electrical circuit models were
used (Figure 5). In several circuits, the capacitance
was substituted by a CPE in order to accommodate
the non-ideality of the systems. In these circuits (Fig-
ure 5a), Re represents the electrolyte resistance, RHF
represents the film resistance and CPEHF represent
film capacitance for high frequency. The same equi-
valent circuit model (Figure 5a) is proposed for the
electrochemical behavior simulation for all immersion
times studied (1, 24, 48, 72 and 96 h) for the
F2A60M, F3A60M and F4A60M samples. This beha-
vior was observed by the other authors [46,47], indi-
cating that the hybrid films retard the corrosion pro-
cesses on the tin plate surface.
Table 4 presents the equivalent electrical para-
meter obtained by fitting the equivalent circuit from
the EIS test data, obtained for the F1A60M, F2A60M,
F3A60M, F4A60M and F5A60M hybrid films after 96
h of immersion in a 0.05 mol L-1 NaCl solution. The
percentage errors shown in brackets in Table 4 show
that the errors involved in the fitting procedure were
less than 10% (less than 5% in most cases).
The other equivalent circuit model (Figure 5b)
was proposed for the electrochemical behavior sim-
ulation at all immersion times studied (1, 24, 48, 72
and 96 h) for the F1A60M and F5A60M samples
[40,41]. This new equivalent circuit model was emp-
loyed because the other circuit equivalent (Figure 5a)
does not allow approximation between simulation and
real data. In these circuits (Figure 5b), Re represents
the electrolyte resistance, RHF represents the film
resistance and CPEHF represent film capacitance for
high frequency, RCP and CPECP represent the resist-
ance and a constant phase element indicating the
diffusion of electrolyte through the film, initiating the
Figure 5. Equivalent electrical circuits for EIS data fitting (a,b) and evolution of the hybrid films resistance for F1A60M, F2A60M,
F3A60M, F4A60M and F5A60M in a 0.05 mol L-1 NaCl solution with the immersion time (c).
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
92
corrosion process [17,41,48,49]. The same equivalent
circuit model (Figure 5b) was proposed for the elec-
trochemical behavior simulation at all immersion times
studied (1, 24, 48, 72 and 96 h) for the F1A60M and
F5A60M samples. This model is consistent with the
protective action lower of these systems. The system
resistance values are found to be around (RHF = 4.31
k cm2 for F1A60M and RHF = 8.51 k cm2 for
F5A60M) after the 96 h of immersion (Table 4).
Table 4 shows that the sample F3A60M had
lower capacitance values than the other analyzed
samples until the end of the test (CPE = 11.89 µF cm–2),
denoting the best protective barrier properties of the
hybrid film. Furthermore, there was an increase in the
coating capacitance of the F1A60M sample after 1
hour of immersion (CPE = 3.29 µF cm–2, Table 4) and
after 24 h of immersion (CPE = 3.69 µF cm–2, Table 4).
For the samples F2A60M, F4A60M and
F5A60M, a slight increase capacitance values (Table
4) was observed in the mean 48 h for 96 h immersion.
After 96 h of immersion, another small significant inc-
rease in the capacitance of the film was observed.
The film capacitance always tends to increase
with immersion time as result of electrolyte uptake.
This behavior is due to a significant increase of the
dielectric constant of the coating, which is influenced
strongly by electrolyte penetration into the coating.
Moreover, chi-square (2) was introduced in Table 4
and the observe values were around 10–3, similar to
results obtained for Sakai et al. (2012) [40].
Figure 5c shows the evolution of the coating
properties (i.e., the resistance) as a function of
immersion time. The F3A60M (RHF = 867 k cm2)
sample exhibited the highest resistances for 1 h
immersion. Furthermore, the F3A60M sample exhi-
bited the highest resistances at all immersion times.
The resistance all of the coatings decreases slowly
over an immersion time of 96 hours , which reflected
the stability of the coating [18].
The evolution of a coating resistance is a major
characteristic of the barrier properties of a protective
layer. Generally, the resistance values decreased
during the first hours of immersion, due to the dev-
elopment of conductive pathways inside the film. The
Table 4. Electrical elements fitted values for the samples up to 96 hours of immersion in 0.05 mol L-1 NaCl solution. The error percent-
age associated with each parameter value is given in parenthesis
Time, h Re / cm2 RHF / cm2 CPEHF-Q / µF cm–2 CPEHF-n RCP / cm2 CPECP-Q / nF cm–2 CPECP-n 2103
F1A60M
1 202 (5.7) 92.71 (3.4) 3.29 (5.1) 0.73 (5.8) 134 (5.6) 9.77 (5.7) 0.51 (6.2) 0.79
24 184 (2.7) 69.5 (3.9) 3.69 (4.9) 0.44 (1.4) 89.7 (5.2) 11.4 (2.8) 0.73 (1.5) 1.32
48 167 (2.1) 14.53 (3.5) 4.01 (1.3) 0.41 (3.4) 69.7 (1.2) 13.3 (1.2) 0.69 (0.6) 0.87
96 132 (3.1) 4.31 (5.1) 4.47 (4.2) 0.39 (1.9) 41.8 (1.4) 14.9 (0.4) 0.64 (0.5) 0.95
F2A60M
1 234 (1.7) 357 (2.3) 3.09 (1.8) 0.76 (0.6) - - - 3.17
24 201 (0.8) 124 (1.9) 3.45 (1.4) 0.72 (0.5) - - - 4.71
48 189 (0.6) 64.7 (1.9) 3.73 (1.7) 0.69 (0.2) - - - 2.98
96 166 (0.8) 31.1 (1.2) 3.99 (1.1) 0.66 (0.3) - - - 3.47
F3A60M
1 293 (4.1) 867 (2.5) 0.11 (2.1) 0.84 (1.4) - - - 2.93
24 271 (1.9) 472 (3.7) 0.51 (1.4) 0.81 (0.6) - - - 3.56
48 243 (1.7) 347 (3.9) 1.31 (1.9) 0.79 (0.3) - - - 3.02
96 229 (2.1) 312 (4.2) 2.73 (2.4) 0.76 (0.7) - - - 3.93
F4A60M
1 241 (3.1) 768 (5.7) 2.81 (3.7) 0.78 (0.6) - - - 1.07
24 211 (1.3) 179 (2.7) 2.93 (2.9) 0.73 (0.3) - - - 1.15
48 197 (0.6) 139 (0.8) 3.12 (0.9) 0.71 (0.4) - - - 0.79
96 171 (1.5) 64.7 (3.3) 3.64 (2.3) 0.70 (0.3) - - - 0.69
F5A60M
1 207 (2.2) 63.7 (4.9) 3.61 (4.7) 0.69 (5.4) 912 (4.8) 9.33 (3.2) 0.71 (3.9) 5.31
24 159 (3.5) 52.37 (4.4) 3.94 (5.8) 0.64 (3.8) 287 (3.1) 11.26 (2.7) 0.76 (3.9) 4.39
48 137 (3.1) 24.74 (5.2) 4.42 (5.4) 0.60 (6.6) 146 (4.4) 12.88 (4.1) 0.71 (4.1) 3.27
96 112 (2.9) 8.51 (3.5) 4.89 (2.6) 0.54 (2.2) 61.9 (2.7) 13.86 (2.1) 0.63 (3.4) 2.94
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
93
high resistance of the sample with 2% plasticizers
should be due to the increasing the ability of polymer
chains to slide over one another; that is to say, the
chain movement enhances the flexibility of the org-
anic radicals that are bound to the silicone atoms.
Figure 6 shows images for all of the hybrid films
after 96 h of the electrochemical impedance tests.
Hybrid films with higher plasticizer concentrations
(F3A60M, Figure 6c and F4A60M, Figure 6d) were
observed to have developed less corrosion products
after the electrochemical tests, confirming the impe-
dance results.
Red corrosion products were observed on the
electrode surface at the end of the experiments, indi-
cating the formation of iron oxides. For hybrid films
without plasticizer (Figure 6e) and with lower plasti-
cizer concentrations F1A60M (Figure 6a) and F2A60M
(Figure 6b), abundant and more localized red cor-
rosion products were observed on their surfaces,
which were considered to be iron oxides.
CONCLUSION
The obtained results are sufficient to state that
all hybrid film coatings shifted potentials indicating
less activity, and diminished the corrosion current
densities (icorr) in relation to the uncoated tinplate (Fl),
evidencing the protection provided by the films. The
hybrid film with 2% of diisodecyl adipate showed the
best performance in the electrochemical impedance
spectroscopy tests.
Films with concentrations lower than 2% plasti-
cizer were thicker, brittle and permeable, which dec-
reases the effect barrier. Films with concentrations
greater than 2% plasticizer featured lower thickness,
irregularities and cracks in the surface, which dec-
reases its efficiency as a barrier film.
Therefore, the addition of 2% of diisodecyl adi-
pate on the silane film improved the morphological
and electrochemical properties of the film, when com-
pared to the film without plasticizer. These results
indicate that this film is a viable alternative to protect
the tinplate against corrosion.
Acknowledgments
This work was done with the financial support of
CAPES, a Brazilian government agency for the
human resources development. Authors would also
like to acknowledge the financial support of CNPq
and FAPERGS, and the Centre for Electron Micro-
scopy of the Federal University of Rio Grande do Sul
for the SEM analysis.
REFERENCES
[1] S. Gnanam, V. Rajendran, J. Sol-Gel Sci. Technol. 58
(2011) 62–69
[2] J. Zhang, P. Lan, J. Li, H. Xu, Q. Wang, X. Zhang, L.
Zheng, Y. Lu, N. Dai, W. Song, J. Sol-Gel Sci. Technol.
71 (2014) 267–275
[3] M.A. Aegerter, R. Almeida, A. Soutar, K. Tadanaga, H.
Yang, T. Watanabe, J. Sol-Gel Sci. Technol. 47 (2008)
203–236
[4] C. Sanchez, P. Belleville, M. Popall, L. Nicole, Chem.
Soc. Rev. 40 (2011) 696–753
Figure 6. SEM images of the samples: a) F1A60M, b) F2A60M, c) F3A60M, d) F4A60M, e) F5A60M and f) Fl after 96 h of immersion
in NaCl 0.05 mol L-1.
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
94
[5] L.L. Blanc, E. Campazzi, P. Savigne, US 8715405 B2 -
PCT/FR2006/050493, 2006
[6] C. Sanchez, B. Julián, P. Belleville, M. Popall, J. Mater.
Chem. 15 (2005) 3559–3592
[7] C.J. Brinker, Sol-gel science: the physics and chemistry
of sol-gel processing, Academic Press, Boston, MA, 1990
[8] V.R. Capelossi, M. Poelman, I. Recloux, R.P.B. Her-
nandez, H.G. de Melo, M. G. Olivier, Electrochim. Acta
124 (2014) 69–79
[9] R.B. Figueira, C.J.R. Silva, E.V. Pereira, J Coat. Technol.
Res 12 (2015) 1–35
[10] K.A. Yasakau, J. Carneiro, M.L. Zheludkevich, M.G.S.
Ferreira, Surface Coat. Technol. 246 (2014) 6–16
[11] L. Li, J. He, J. Lei, W. Xu, X. Jing, X. Ou, S. Wu, N. Li, S.
Zhang, Surface Coat. Technol. 279 (2015) 72–78
[12] S. Bera, T.K. Rout, G. Udayabhanu, R. Narayan, Prog.
Org. Coat. 88 (2015) 293–303
[13] C. Oh, C. Do Ki, J. Young Chang, S.-G. Oh, Mater. Lett.
59 (2005) 929–933
[14] G.R. Salazar-Banda, K.I.B. Eguiluz, A.J. Motheo, S.A.S.
Machado, Mater. Res. 16 (2013) 1315–1324
[15] J. Martin, B. Hosticka, C. Lattimer, P. Norris, J. Non-Crys.
Solids 285 (2001) 222–229
[16] D. Wang, G.P. Bierwagen, Prog. Org. Coat. 64 (2009)
327–338
[17] D. Xia, S. Song, J. Wang, H. Bi, Y. Jiang, Z. Han, Trans-
act. Nonferrous Met. Soc. China 22 (2012) 717–724
[18] S.S. Pathak, A.S. Khanna, Prog. Org. Coat. 62 (2008)
409–416
[19] E. Certhoux, F. Ansart, V. Turq, J.P. Bonino, J.M. Sob-
rino, J. Garcia, J. Garcia, Prog. Org. Coat. 76 (2013) 165–
–172
[20] E. Zumelzu, C. Cabezas, Mater. Charact. 34 (1995) 143–
–148
[21] D.H. Xia, J.H. Wang, S.Z. Song, B. Zhong, Z.W. Han,
Adv. Mater. Res. 233–235 (2011) 1747–1751
[22] X. Huang, N. Li, H. Wang, H. Sun, S. Sun, J. Zheng, Thin
Solid Films 516 (2008) 1037–1043
[23] A. Nincević Grassino, Z. Grabarić, A. Pezzani, G. Fasa-
naro, A. Lo Voi, Food Chem. Toxicol. 47 (2009) 1556–1561
[24] W.E.G. Hansal, S. Hansal, M. Pölzler, A. Kornherr, G.
Zifferer, G.E. Nauer, Surf. Coat. Techol. 200 (2006) 3056–
–3063
[25] H. Hasannejad, M. Aliofkhazraei, A. Shanaghi, T. Shar-
abi, A.R. Sabour, Thin Solid Films 517 (2009) 4792–4799
[26] D. Álvarez, A. Collazo, C. Pérez, Prog. Org. Coat. 76
(2013) 1819–1826
[27] S.R. Kunst, L.V.R. Beltrami, H.R.P. Cardoso, M.R.O.
Veja, E.K.K. Baldin, T.L. Menezes, C. de F. Malfatti,
Mater. Res. 17 (2014) 1071–1081
[28] S.R. Kunst, G.A. Ludwig, H.R.P. Cardoso, J.A. Santana,
V.H.V. Sarmento, C. de F. Malfatti, Mater. Res.h 17
(2014) 75–81
[29] S.R. Kunst, L.V.R. Beltrami, H.R.P. Cardoso, J.A.
Santana, V.H.V. Sarmento, I.L. Müller, C. de F. Malfatti,
Mater. Res. 18 (2015) 151–163
[30] C.F. Malfatti, T.L. Menezes, C. Radtke, J. Esteban, F.
Ansart, J.P. Bonino, Mater. Corr. 63 (2012) 819–827
[31] C. Merlatti, F.X. Perrin, E. Aragon, A. Margaillan, Prog.
Org. Coat. 61 (2008) 53–62
[32] S. Kozhukharov, V. Kozhukharov, M. Schem, M. Aslan,
M. Wittmar, A. Wittmar, M. Veith, Prog. Org. Coat. 73
(2012) 95–103
[33] G. Wypych, Handbook of plasticizers, 2nd ed., ChemTec
Pub, Toronto, 2012
[34] R.F. Grossman, J.T.L. Jr., Polymer Modifiers and Addi-
tives, CRC Press, New York, 2000
[35] J. Edenbaum, Plastics additives and modifiers handbook,
Van Nostrand Reinhold, New York, 1992
[36] S.R. Kunst, H.R.P. Cardoso, C.T. Oliveira, C.I. da S.
Filho, V.H.V. Sarmento, T.L. Menezes, I.L. Müller, C. de
F. Malfatti, Int. J. Electrochem. Sci. 8 (2013) 11984–12004
[37] W.J. van Ooij, D. Zhu, M. Stacy, A. Seth, T. Mugada, J.
Gandhi, P. Puomi, Tsinghua Sci. Technol. 10 (2005) 639–
664
[38] L.M. Palomino, P.H. Suegama, I.V. Aoki, M. Fatima
Montemor, H. G. De Melo, Corros. Sci. 50 (2008) 1258–
1266
[39] C.R. Tomachuk, C.I. Elsner, A.R.D. Sarli, O.B. Ferraz, J.
Coat. Technol. Res. 7 (2010) 493–502
[40] R.T. Sakai, F.M.D. L. da Cruz, H.G. de Melo, A.V.
Benedetti, C.V. Santilli, P.H. Suegama, Prog. Org. Coat.
74 (2012) 288–301
[41] P.H. Suegama, V.H.V. Sarmento, M.F. Montemor, A.V.
Benedetti, H.G. de Melo, I.V. Aoki, C.V. Santilli, Electro-
chim. Acta 55 (2010) 5100–5109
[42] G. Bombara, N. Azzerri, G. Baudo, Corros. Sci. 10 (1970)
847–856
[43] K. Galić, M. Pavić, N. Ciković, Corros. Sci. 36 (1994) 785–
–795
[44] R. Zandi Zand, K. Verbeken, A. Adriaens, Int. J. Electro-
chem. Sci. 7 (2012) 9592–9608
[45] M.E.P. de Souza, E. Ariza, M. Ballester, I.V.P. Yoshida,
L.A. Rocha, C.M. de A. Freire, Mater. Res. 9 (2006) 59–
–64
[46] J.B. Bajat, I. Milošev, Ž. Jovanović, R.M. Jančić-Heine-
mann, M. Dimitrijević, V.B. Mišković-Stanković, Corros.
Sci. 52 (2010) 1060–1069
[47] P.C.R. Varma, J. Colreavy, J. Cassidy, M. Oubaha, C.
McDonagh, B. Duffy, Thin Solid Films 518 (2010) 5753–
–5761
[48] S.V. Lamaka, H.B. Xue, N.N.A.H. Meis, A.C.C. Esteves,
M.G.S. Ferreira, Prog. Org. Coat. 80 (2015) 98–105
[49] J. Li, L. Ecco, M. Fedel, V. Ermini, G. Delmas, J. Pan,
Progr. Org. Coat. 87 (2015) 179–188.
S.R. KUNST et al.: EFFECT OF DIISODECYL ADIPATE CONCENTRATION… Chem. Ind. Chem. Eng. Q. 23 (1) 8395 (2017)
95
SANDRA RAQUEL KUNST1
LILIAN VANESSA ROSSA
BELTRAMI2
MARIELEN LONGHY1
HENRIQUE RIBEIRO PIAGGIO
CARDOSO2
TIAGO LEMOS MENEZES2
CÉLIA de FRAGA MALFATTI2
1Programa de Pós Graduação em
Engenharia de Processos e
Tecnologias (PGEPROTEC),
University of Caxias do Sul (UCS),
Caxias do Sul, RS – Brasil 2Laboratório de Pesquisa em Corrosão
(LAPEC), Federal University of Rio
Grande do Sul (UFRGS), Porto Alegre,
RS - Brasil
NAUČNI RAD
UTICAJ KONCENTRACIJE DIIZODECIL-ADIPATA U HIBRIDNIM FILMOVIMA NANETIH NA LIMU
Hibridni filmovi na bazi siloksana su krhki i male mehaničke čvrstoce zbog njihovih osobina
staklastog materijala. Stoga je razvijena jedna nova formulacija koja uključuje plastifikator
u cilju povecanja debljine nanosa uniformnog i homogenog hibridnog filma na limu, i
obezbeđenja fleksibilnosti polimernog matriksa. Limene ploče su obložene hibridnim
filmom dobijenim sol-gel postupkom, uz dodatak alkoksidnih prekursora 3-(trimetoksisilil)
propil metakrilat i tetraetokisilan, kao i 0,01 mol/dm3 cerijum nitrata. Ocenjivan je uticaj
koncentracije plastifikatora di-isodecil adipata. Filmovi su okarakterisani skenirajućim
elektronskim mikroskopom, profilometrijom, merenjem potencijala otvorenog kola,
polarizacionim merenjima i spektroskopijom elektrokemijske impedancije. Rezultati su
pokazali da su svi filmovi sa di-isodecil adipatom imali vecu elektrokemijsku performansu u
odnosu na nepremazan lim. Međutim, film sa koncentracijom plastifikatora od 2 % je imao
najbolju performansu u elektrohemijskim testovima, iako je imao manju debljinu.
Ključne reči: hibridni film, diizodecil-adipat, beli lim, elektrohemijsko ponašanje.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017) CI&CEQ
97
MILOVAN JANKOVIĆ
SNEŽANA
SINADINOVIĆ-FIŠER
OLGA GOVEDARICA
JELENA PAVLIČEVIĆ
JAROSLAVA
BUDINSKI-SIMENDIĆ
University of Novi Sad, Faculty of
Technology, Novi Sad, Serbia
SCIENTIFIC PAPER
UDC 665.335.2:66.094.3:544.4
https://doi.org/10.2298/CICEQ150702014J
KINETICS OF SOYBEAN OIL EPOXIDATION WITH PERACETIC ACID FORMED IN SITU IN THE PRESENCE OF AN ION EXCHANGE RESIN: PSEUDO-HOMOGENEOUS MODEL
Article Highlights
• In situ epoxidation of soybean oil in the presence of an ion exchange resin is studied
• Occurrence of the reactions during the incremental addition of the reactant is considered
• Pseudo-homogeneous kinetic model is applied
• Temperature dependency of kinetic parameters was determined
• The kinetic model fits well the experimental data
Abstract
A kinetic model was proposed for the epoxidation of vegetable oils with peracetic
acid formed in situ from acetic acid and hydrogen peroxide in the presence of an
acidic ion exchange resin as a catalyst. The model is pseudo-homogeneous with
respect to the catalyst. Besides the main reactions of peracetic acid and epoxy
ring formation, the model takes into account the side reaction of epoxy ring
opening with acetic acid. The partitioning of acetic acid and peracetic acid
between the aqueous and organic phases and the change in the phases’ vol-
umes during the process were considered. The temperature dependency of the
apparent reaction rate coefficients is described by a reparameterized Arrhenius
equation. The constants in the proposed model were estimated by fitting the
experimental data obtained for the epoxidations of soybean oil conducted under
defined reaction conditions. The highest epoxy yield of 87.73% was obtained at
338 K when the mole ratio of oil unsaturation:acetic acid:hydrogen peroxide was
1:0.5:1.35 and when the amount of the catalyst Amberlite IR-120H was 4.04
wt.% of oil. Compared to the other reported pseudo-homogeneous models, the
model proposed in this study better correlates the change of double bond and
epoxy group contents during the epoxidation process.
Keywords: soybean oil, epoxidation, peracetic acid, ion exchange resin, kinetics.
The epoxidation of soybean oil is commercially
important since the obtained epoxide is used as poly-
mer stabilizer and plasticizer, paint and coating com-
ponent, and lubricant. It is also an intermediate for the
production of glycols, alkanolamines, polyols and
polymers [1,2]. Besides performic acid, peracetic acid
is a common oxidizing agent for the epoxidation of
vegetable oils [3,4]. Because of instability and safety
Correspondence: S. Sinadinović-Fišer, University of Novi Sad,
Faculty of Technology, Bul. cara Lazara 1, 21000 Novi Sad,
Serbia. E-mail: [email protected] Paper received: 2 July, 2015 Paper revised: 24 February, 2016 Paper accepted: 11 March, 2016
issues, peracetic acid is usually produced in situ
through the acid-catalyzed peroxidation of acetic acid
with hydrogen peroxide in an aqueous solution [5].
The epoxidation of vegetable oil double bonds with in
situ generated peracetic acid also involves an un-
catalyzed reaction of epoxy ring formation in the org-
anic phase, and a few side reactions of acid-cata-
lyzed epoxy ring opening. Some soluble mineral
acids, like sulphuric acid [1,5-8], and acidic ion
exchange resins can be applied as catalysts for this
process [3,9-20]. The stability of epoxy ring and, thus,
the selectivity of the process, are higher when an ion
exchange resin is used as the catalyst compared to
the mineral acid [21]. Among the sulphonated ion
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
98
exchange resins investigated, the highest conversion
and selectivity, which were almost constant for five
reuses, were obtained with Amberlite IR-120H [20].
The volume of this resin increases about 43% in
water and about 8% in glacial acetic acid [22]. The
swelling occurs because of the absorption and ads-
orption of solvent molecules on the sulphonated
groups inside the catalyst pores [23,24].
Since the epoxidation of vegetable oils is sig-
nificantly influenced by the reaction conditions, such
as molar ratio of reactants, type of the catalyst, cat-
alyst concentration, temperature, stirring speed, and
presence of the solvent, it is necessary to establish
an appropriate kinetic model for its optimization. In
some studies, the effect of reaction conditions on the
in situ epoxidation of vegetable oils has been inves-
tigated [9,10,18,19], whereas in other studies, the pro-
cess kinetics was also considered [1,3-6,11-17,25-33].
When the ion exchange resin is used as the cat-
alyst for in situ generation of peracetic acid, the react-
ion system for the epoxidation of vegetable oils is
three-phase liquid-liquid-solid (organic-aqueous-solid).
For rigorous mathematical modeling of such a sys-
tem, the intrinsic kinetics, the mass diffusion, and the
partitioning of the components between the phases
have to be considered. Up to the present, this react-
ion system has been described with the pseudo-
-homogeneous [3,16,17] or pseudo-two-phase (liquid-
–solid) [11-15,33] models. Both types of models took
into consideration the main reactions of peracetic acid
and epoxy compound formation, as well as some of
the side reactions of epoxy ring opening that may
occur during the process. The proposed pseudo-
homogeneous models were developed by assuming
that the ion exchange resin is dissolved in the react-
ion mixture. The pseudo-two-phase models were
established by applying Eley-Rideal and Langmuir-
–Hinshelwood-Hougen-Watson approaches to the
reaction of peracetic acid formation. Although a few
authors investigated the partitioning of acetic acid
between the aqueous and organic phases separately
from the epoxidation process [34-36], this pheno-
menon was not considered in the reported kinetic
models.
Because of the aforementioned, the objective of
this study was to develop a kinetic model for the
epoxidation of vegetable oils with peracetic acid
formed in situ in the presence of an acidic ion
exchange resin which takes into consideration the
partitioning of acetic acid and peracetic acid between
the organic and aqueous phases, as well as the
change in the volume of these two phases. The
pseudo-homogeneity of the catalyst with respect to
the aqueous phase is assumed. Besides the main
reactions, the model describes the epoxy ring open-
ing reaction with acetic acid. The model parameters
were determined by fitting the experimental data
obtained for the epoxidation of soybean oil. The pro-
posed model was compared with the pseudo-homo-
geneous model reported in the literature.
EXPERIMENTAL
Materials
Soybean oil was kindly provided by Dijamant
(Zrenjanin, R. Serbia). The acid form of sulphonated
polystyrene-type ion exchange resin Amberlite IR-
120H from Rohm&Hass Co. (Philadelphia, PA, USA)
was used as the catalyst. Glacial acetic acid, 30%
aqueous hydrogen peroxide solution and hydrobromic
acid were purchased from J.T. Baker (Deventer,
Netherlands). Alfapanon (Novi Sad, Serbia) was a
supplier of the aqueous solutions of sodium hydroxide
(0.1 N) and sodium thiosulfate (0.1 N). Iodine (p.a.)
and bromine (p.a.) were purchased from Centrohem
(Stara Pazova, R. Serbia). Potassium iodide (extra
pure), chloroform (min 98.5%), benzene (min 99.8%),
potassium hydrogen phthalate (min 99.0%) were
bought from LachNer (Neratovice, Czech Republic).
Hydrogen bromide solution (33.0%) in acetic acid and
crystal violet were purchased from Sigma-Aldrich (St.
Louis, MO, USA).
Epoxidation procedure
The epoxidation of soybean oil in bulk was car-
ried out with peracetic acid formed in situ according to
the method reported in the literature [15]. Soybean oil
with an initial iodine number (IN0) of 128.62, which
corresponds to 0.5067 mol of double bond per 100 g
of oil, was used for the syntheses. The molar ratio of
soybean oil unsaturation:acetic acid:hydrogen per-
oxide was approximately 1:0.5:1.35 for all runs. The
oil (approximately 150 g) was mixed with an appro-
priate mass of glacial acetic acid before being poured
into a 1000 mL three-neck glass reactor equipped
with a magnetic stirrer, a thermometer, a reflux con-
denser and an addition funnel. The reactor was
placed in a water bath. Amberlite IR-120H ion
exchange resin was introduced to the reactor before
the reactants. The amount of catalyst was 1.98–
–7.86 wt.% of oil. Subsequently, the 30% aqueous
hydrogen peroxide solution was added drop-wise to
the reaction mixture at a constant rate within 40 to 65
min. During the addition, the temperature of the mix-
ture was maintained at 323 or 338 K with fluctuation
of less than ±1 K. Further, where applicable, the tem-
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
99
perature was increased at a uniform rate to the
desired level and maintained at 323, 338, 348 or 353
K within ±1 K. The temperature of the reaction mix-
ture was controlled by changing the water bath tem-
perature. The fine dispersion of oil in the reaction
mixture was achieved with the uniform agitation using
a PTFE-coated cylindrical stirring bar (35 mm6 mm)
under constant stirring speeds of 900, 1000 or 1100
rpm. The beginning of the addition of hydrogen per-
oxide solution was considered to be the “zero time” of
the process. The progress of epoxidation was fol-
lowed by withdrawing 8 mL samples of the reaction
mixture at defined time intervals. The first sample was
taken immediately after the completion of the hydro-
gen peroxide addition, whereas the second imme-
diately after raising the reaction temperature, where
applicable. The others were collected at about 30, 60
or 120 min intervals. After cooling and centrifugation
of the sample, the separated organic phase was
washed with water (323 K) until pH 7. Water was eva-
porated from the sample at 333 K under the vacuum.
The evaporation lasted a minimum of 1 h. The
samples were then analyzed to determine the iodine
number (IN) and epoxy oxygen content (EO). The
infrared analysis of the samples was also provided.
Analyses
The iodine number and epoxy oxygen content
were measured in triplicate according to the Hanus
method and the standard HBr-acetic acid method,
respectively [37]. Using attenuated total reflectance
(ATR) FT-IR spectroscopy, the spectra were recorded
by Thermo Finnigan’s Nicolet 5700 FTIR spectro-
meter in the range of 4000-400 cm-1 by accumulating
32 scans at a resolution of 4 cm-1.
RESULTS AND DISCUSSION
Soybean oil epoxidation
Nine epoxidation runs were conducted using
soybean oil and peracetic acid formed in situ from
acetic acid and 30% hydrogen peroxide aqueous
solution in the presence of Amberlite IR-120H. The
mole ratio of soybean oil unsaturation:acetic acid:hyd-
rogen peroxide was approximately 1:0.5:1.35 for all
runs. The reaction conditions and some of the results
are summarized in Table 1. Residual iodine number
(IN), conversion of double bond (X), relative epoxy
yield (REY) and selectivity (SE) are presented only for
the content of epoxy oxygen (EOt) reached in each
run after defined period of time (t).
The disappearance of double bonds and form-
ation of epoxy groups during the runs were monitored
on the basis of the FT-IR spectrum of the samples, as
illustrated in Figure 1. The changes in the intensities
of epoxy group doublet band (with maxima at 823 and
845 cm–1) and double bond band (at 3007 cm-1) were
qualitatively analyzed.
The progress of the epoxidation was quantified
by determining the iodine number (IN) and epoxy
oxygen content (EO) for all samples withdrawn during
the run. The values are presented as points in Figures
2-4.
There are detailed discussions in our previous
studies [15,16] and in other studies [3,9–14,17] on the
Table 1. Reaction conditions and the residual iodine number (IN), conversion of double bond (X), relative epoxy yield (REY) and sel-
ectivity (SE) for content of epoxy oxygen (EOt) reached after reaction time (t) in each run of the epoxidation of soybean oil (SO)(a) in bulk
with peracetic acid formed in situ from 30% hydrogen peroxide aqueous solution (aqHP) and acetic acid (A) in the presence of Amberlite
IR-120H as the catalyst when the molar ratio of soybean oil unsaturation:acetic acid:hydrogen peroxide was approximately 1:0.5:1.35; initial iodine number IN0 = 128.62 corresponds to theoretical epoxy oxygen content (EOth) of 7.50% where EOth = 100{(IN0/2AI)/
/[100+(IN0/2AI)AO]}AO
Run Measured mass, g Amberlitea
wt.%
Stirring speed
rpm
Temperature, K t
min
EOt
% INb
Xc)
%
REYd
% SEe
SO A aqHP H2O2 addition Reaction
1 150.00 22.82 116.83 3.92 [4.21] 1000 323 323 630 4.65 44.2 65.64 62.00 0.94
2 150.00 22.82 116.83 3.92 [4.21] 1000 323 338 645 6.25 7.45 94.21 83.33 0.88
3 150.00 22.82 116.83 3.92 [4.21] 1000 323 348 635 6.22 1.73 98.65 82.93 0.84
4 150.00 22.82 116.83 3.92 [4.21] 1000 323 353 470 6.08 4.45 96.54 81.06 0.84
5 145.44 21.66 115.90 4.04 [4.27] 1100 338 338 625 6.58 1.99 98.45 87.73 0.89
6 150.23 22.81 109.46 3.91 [4.45] 1000 338 338 480 6.45 7.22 94.39 86.00 0.91
7 145.50 21.91 117.02 4.04 [4.23] 900 338 338 615 6.40 2.71 97.89 85.33 0.87
8 148.32 21.84 112.56 1.98 [2.19] 1100 323 353 590 6.37 2.87 97.77 84.93 0.87
9 149.82 21.57 114.53 7.86 [8.65] 1100 323 323 630 5.51 22.63 82.41 73.46 0.89
aCatalyst concentration is expressed in percentage of soybean oil weight; value given in square brackets is percentage of catalyst in respect to acetic acid
and 30% hydrogen peroxide weight; bresidual iodine number, IN; cdouble bond conversion, X = 100(IN0-IN)/IN0; drelative epoxy yield, REY = 100EOt/EOth; eselectivity, SE = EOtIN0/[EOth(IN0-IN)]
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
100
Figure 1. FT-IR spectra of soybean oil and samples taken after 55, 175 and 625 min of soybean oil epoxidation in bulk with peracetic
acid formed in situ from acetic acid and hydrogen peroxide in the presence of amberlite IR-120H in the amount of 4.04 wt.% of oil at 338
K and 1100 rpm, when the mole ratio of double bond in oil:acetic acid:hydrogen peroxide was approximately 1:0.5:1.35.
Figure 2. Time dependency of experimentally determined (points) and calculated (curves) iodine number (IN) and epoxy oxygen content
(EO) for the epoxidation of soybean oil in bulk with peracetic acid generated in situ in the presence of Amberlite IR-120H in the amount
of approximately 4.0 wt.% of oil at 338 K, when the stirring speeds (S) were 900 and 1100 rpm and when the mole ratio of double bond
in oil:acetic acid:30 wt.% hydrogen peroxide was approximately 1:0.5:1.35.
Figure 3. Time dependency of experimentally determined (points) and calculated (curves) iodine number (IN) and epoxy oxygen content
(EO) for the epoxidation of soybean oil in bulk with peracetic acid generated in situ in the presence of Amberlite IR-120H in the amount
of 1.98, 3.92 and 7.86 wt.% of oil at 323 (Runs 1and 9) or 353 K (Run 8), when the mole ratio of double bond in oil:acetic acid:30 wt.%
hydrogen peroxide was approximately 1:0.5:1.35.
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
101
Figure 4. Time dependency of experimentally determined (points) and calculated (curves): a) iodine number (IN) and b) epoxy oxygen
content (EO), for the epoxidation of soybean oil in bulk with peracetic acid generated in situ in the presence of Amberlite IR-120H in the
amount of 3.92 wt.% of oil at 323, 338, 348 and 353 K and 1000 rpm, when the mole ratio of double bond in oil:acetic acid:30 wt.%
hydrogen peroxide was approximately 1:0.5:1.35.
influence of the reaction conditions on the conversion
of double bond, relative epoxy yield, and selectivity
for the in situ epoxidation of vegetable oils with pera-
cetic acid in the presence of the ion exchange resin.
This study discusses the results obtained for the in
situ epoxidation of soybean oil.
The stoichiometric ratio of hydrogen peroxide to
vegetable oil unsaturation is 1:1. However, this oxy-
gen agent is usually applied in excess for the epoxid-
ation of vegetable oils [3,9-18]. The mole ratio of
hydrogen peroxide to double bond was studied in the
range from 0.8 to 3 for various oils [3,9-18]. The
significant increase in the rate of epoxy ring formation
was observed with an increase in molar ratio from 0.8
to 1.5 [9,12-14] or 2 [10,11]. With a further increase in
molar ratio, there was no appreciable decrease of the
residual iodine number or change of the epoxy oxy-
gen content [9-14]. Also, when a mole ratio higher
than 1.5 was applied, a higher rate of the epoxy ring
opening reactions was observed [3,9,10]. Addition-
ally, the excess of hydrogen peroxide solution in the
reaction system decreases the concentration of acetic
acid as the other reactant in the reaction of peracetic
acid formation. Hence, a hydrogen peroxide to veget-
able oil unsaturation molar ratio of about 1.35 was
chosen for this study.
The stirring speed influences the external mass
transfer in the multiphase reaction systems. To red-
uce the mass transfer resistance, adequate mixing
has to be achieved. Under a particular stirring speed,
the intensity of mixing also depends upon the type
and design of a stirrer [18]. At higher stirring speeds,
the mass transfer resistance is lower. Consequently,
the rate of the epoxidation should be higher. How-
ever, no significant influence on the double bond con-
version and relative epoxy yield was observed when
the stirring speed changed from 900 to 1100 rpm
under the conditions applied in this study (Figure 2).
The reported amounts of catalyst for the epoxid-
ation of vegetable oils vary from 1.28 to 25 wt.% of oil
[3,9-18]. The quantity of Amberlite IR-120H used in
this study was 1.98–7.86 wt.% of oil. To ensure good
mixing of the reaction mixture, such low amounts of
the catalyst were chosen. As concluded in most
papers, when a higher amount of the catalyst was
used, an increase in the double bond conversion and
epoxy yield was observed. At the same time, the
selectivity of the process was reduced with an inc-
rease in the catalyst load (Table 1, runs 1 and 9). This
was the consequence of the promotion of acid-cat-
alyzed epoxy ring opening reactions. Figure 3 shows
the change in iodine number and epoxy oxygen con-
tent over time for the epoxidations of soybean oil per-
formed under different catalyst concentrations. It
should be noted that the reaction temperature of the
run with the lowest catalyst loading was higher than
the temperature for the other two runs.
The temperature of the process should be
adjusted to increase the rate of the peracetic acid
formation and epoxidation reaction, but not to com-
promise the stability of the epoxy ring. For the epoxid-
ations of different vegetable oils, the temperatures
varied from 303 to 363 K [3,9-17]. In order to deter-
mine the dependency of the Arrhenius type model
parameters from temperature, the epoxidations of
soybean oil were carried out at 323, 338, 348 and 353
K in this study. The reaction temperatures were the
same or higher than the temperature at which the
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
102
hydrogen peroxide solution was added into the react-
ion mixture. The addition was performed gradually to
avoid an increase in temperature due to the exo-
thermic effect of the reaction of peracetic acid form-
ation. An increase in the reaction rates with an inc-
rease in the reaction temperature was observed
(Figure 4a and b). Consequently, the reaction time for
achieving the same epoxy yield was shortened.
The highest yield of epoxide of 87.73% with
almost complete conversion of double bonds of
98.45% was achieved after 625 min, when the tem-
perature of the hydrogen peroxide solution addition
was the same as desired reaction temperature of 338
K. The amount of the catalyst was only 4.04 wt.% of
oil. The selectivity of double bond conversion to
epoxide of 0.89 and the residual unsaturation of epox-
idized soybean oil of 1.99 imply good quality of the
obtained product (Table 1).
Kinetic model
The epoxidation of soybean oil involves an acid-
catalyzed reaction of the peracetic acid (P) and water
(W) formation from acetic acid (A) and hydrogen
peroxide (H):
(1)
followed by an un-catalyzed reaction of the double
bond (D) conversion into the epoxy ring (E):
(2)
Among a few acid-catalyzed side reactions of
epoxy ring cleavage (Eqs. (3a)-(3e)), the most likely is
the reaction with acetic acid that leads to the form-
ation of hydroxy acetate (HA), Eq. (3a) [38]:
(3)
In this three-phase organic-aqueous-solid react-
ion system, the catalyst Amberlite IR-120H swells due
to its sorption affinity towards the polar components. It
can be assumed that the swelling of the resin is com-
pleted already at the beginning and that the swelling
degree is constant during the epoxidation. Because of
the preferential sorption of some components over
the others, the concentrations of the components dif-
fer in the aqueous bulk phase and inside the catalyst
pores [23,24]. The proceeding of the reactions causes
the mass transfer phenomena across the organic-
aqueous interface, across the aqueous-solid inter-
face, and through the resin pores. In Figure 5a, the
system phases with the possible reactions and mass
transfer of the components across the reaction sys-
tem are schematically presented.
Acetic acid and hydrogen peroxide diffuse from
the aqueous bulk phase to the catalyst external sur-
face and further into the catalyst pores, in both cases,
to reach the active sites where they react. The pro-
ducts of the reaction, peracetic acid and water, mig-
rate to the aqueous phase. As a consequence of the
reaction, a concentration gradient of the components
exists inside the catalyst pores. From the aqueous
phase, the peracetic acid diffuses into the organic
bulk phase where it epoxidizes the triglyceride double
bond. In the epoxidation reaction, regenerated acetic
acid diffuses to the aqueous phase. A fraction of the
acid reacts with the epoxy ring, however. The epoxid-
ized triglycerides diffuse from the organic phase to
the aqueous phase to reach the catalyst. Due to steric
hindrance, the epoxidized triglycerides cannot enter
the catalyst pores. However, they react with acetic
acid or hydrogen peroxide at acidic active sites avail-
able at the external surface of the catalyst [39]. Tri-
glycerides with opened epoxy rings diffuse into the
aqueous and further to the organic phase. The epoxy
ring opening reaction also occurs at the organic-aque-
ous interface according to some authors [25,29].
Since a mathematical description of the three-
-phase reaction system for the epoxidation of veget-
able oils is complex, the kinetic models developed in
other studies are based on a simplified reaction
scheme, a reduced number of phases, and/or neg-
lected transport phenomena [11-17,26,33]. The model
proposed in this study is established by assuming that
only reactions (1), (2) and (3a) occur during the epox-
idation. Some authors reported that the effect of
internal mass transfer is negligible in the reaction sys-
tem for the epoxidation of vegetable oils with pera-
cetic acid formed in situ [11-14]. The same was con-
cluded for the peracetic acid formation investigated
apart from the epoxidation process [30]. Therefore,
the preferential sorption and internal mass transfer
inside the resin pores were neglected in this study.
This enabled using a pseudo-homogeneous model to
describe the kinetics of the reaction of peracetic acid
formation.
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
103
To develop the mathematical model for the
epoxidation of vegetable oils, the material balances
for components in the reaction system were derived
from the equation for semi-batch, i.e., feed-batch
reactor [40]:
,
1
d
d
NRj
j j i i
i
NF V r
t (4)
where Nj (mol) is the number of moles of component j;
t (min) indicates the reaction time; Fj (mol/min) is the
molar flow of component j; V (L) indicates the volume;
NR is the total number of reactions; j,i is the stoi-
chiometric coefficient of component j in the reaction i;
and ri is the rate of reaction i.
It is known from the literature that the solubility
of hydrogen peroxide in the organic phase of the
epoxidation reaction system can be neglected [39].
The amount of water in the organic phase of the
epoxidized soybean oil-acetic acid-water system
ranges from 1.62 to 3.04 wt.% at 323-353 K [36],
whereas it is not present in the organic phase of the
soybean oil-acetic acid-water system [34,35]. How-
ever, for the sake of simplicity, it was assumed that
the water is present only in the aqueous phase,
whereas the triglycerides are present only in the org-
anic phase of the investigated reaction system. Thus,
the number of moles of the components is: aqH HN N ,
aqW WN N , o
D DN N , oE EN N and o
HA HAN N . Since
acetic acid and peracetic acid are partitioned between
the organic and aqueous phases, the mass transfer
and concentrations of these components in the par-
ticular phase should be taken into consideration [5]:
aq
aqaq aq aq o aq oA
A, L,A A A A
1
d
d
NR
i i
i
NV r k a C K C V
t (5)
o
oo o o o aq oA
A, L,A A A A
1
d
d
NR
i i
i
NV r k a C K C V
t (6)
aq
aqaq aq aq aq o oP
P, L,P P P P
1
d
d
NR
i i
i
NV r k a K C C V
t (7)
o
oo o o aq o oP
P, L,P P P P
1
d
d
NR
i i
i
NV r k a K C C V
t (8)
where superscripts aq and o denote the aqueous and
organic phases, respectively; kL,A and kL,P (m/min) are
Figure 5. The reaction system for the epoxidation of soybean oil with peracetic acid formed in situ from acetic acid and hydrogen
peroxide in the presence of Amberlite IR-120H: a) reactions and mass diffusion and b) in this work considered reactions
and partitioning of the components.
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
104
the mass transfer coefficients for acetic acid and
peracetic acid, respectively; a (m2/m3) indicates the
interface area; oAC , aq
AC and oPC , aq
PC (mol/L) are the
concentrations of acetic acid and peracetic acid in a
particular phase, respectively; and KA and KP indicate
the partition coefficients for acetic acid and peracetic
acid, respectively.
The partition coefficients of acetic acid and per-
acetic acid are defined as follows:
o o oA A
A aq aq aqA A
/
/
C N VK
C N V (9)
o o oP P
P aq aq aqP P
/
/
C N VK
C N V (10)
When the number of moles of acetic acid in the
organic phase is expressed from Eq. (9) and substi-
tuted in the material balance equation for acetic acid:
aq oA A AN N N (11)
the following equation is derived:
aq o
aq A AA A aq
K N VN N
V (12)
From Eq. (12), the number of moles of acetic
acid in the aqueous phase may be expressed as a
function of the total number of moles of acetic acid in
the system:
aqaq AA aq o
A
V NN
V K V (13)
Therefore, the concentration of acetic acid in the
aqueous phase is:
aq AA aq o
A
NC
V K V (14)
Likewise, the number of moles and concentra-
tion of acetic acid in the organic phase are as follows:
oo A AA aq o
A
K V NN
V K V (15)
o A AA aq o
A
K NC
V K V (16)
Analogously, the concentrations of peracetic
acid in the aqueous and organic phases can be
derived and expressed as follows:
aq PP aq o
P
NC
V K V (17)
o P PP aq o
P
K NC
V K V (18)
To reduce the external mass transfer resistance
between the organic and aqueous phases, as well as
between the aqueous and solid phases, effective mix-
ing was ensured by vigorous stirring in this study. By
investigating the influence of the stirring speed on the
kinetics of the epoxidation process, it was found that
the mass transfer is faster than the reaction kinetics.
Thus, the mass transfer resistance can be neglected
and the terms related to the mass transfer can be
omitted from the kinetic model [4]. The material bal-
ance equation for acetic acid, on the basis of the Eqs.
(5), (6) and (11), becomes:
aq o
aq oA A A
aq aq aq o o oA, A,
1 1
NR NR
i i i i
i i
dN dN dN
dt dt dt
V r V r
(19)
and, likewise, the material balance equation for per-
acetic acid is as follows:
aq o
aq oP P P
aq aq aq o o oP, P,
1 1
NR NR
i i i i
i i
dN dN dN
dt dt dt
V r V r
(20)
Now, the mathematical model that describes the
reaction system for the epoxidation of vegetable oils
is derived as:
3
aq aq aqaq aq aq aqH P W
1 A HH O11
d
d
N C Ck C C C V
t K (21)
aq aqH H
H
1
d d
d d
N NF
t t (22)
aqo o o o o oA H
2 P D 3 E A
1
d d
d d
nN Nk C C V k C C V
t t (23)
aqo o oP H
2 P D
1
d d
d d
N Nk C C V
t t (24)
aq aqW H
W
1
d d
d d
N NF
t t (25)
o
o o oD2 P D
d
d
Nk C C V
t (26)
o
o o o o o oE2 P D 3 E A
d
d
nNk C C V k C C V
t (27)
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
105
o
o o oHA3 E A
d
d
nNk C C V
t (28)
where k1 (mol-2min-1), k2 (mol-1min-1) and k3 (mol-n min-1)
are the rate coefficients for the reactions (1), (2) and
(3a); and n is the order of epoxy ring opening reaction
with respect to acetic acid which was assumed to be
the first or the second. The reactions and the parti-
tioning of the components considered in the model
proposed in this study are shown in Figure 5b.
Since the model is pseudo-homogeneous with
respect to the aqueous phase, the concentration of
hydronium ions ( +3H O
C ) corresponds to the complete
dissociation of resin sulphonated groups in the aque-
ous phase and it is expressed as:
3
aq saqH O
mCC
V (29)
where m (g) is the mass of the catalyst; and Cs (mol/g
catalyst) indicates the concentration of active catalyst
sites (sulphonated groups).
After substituting the expressions (29), (14) and
(16)-(18) in the model Eqs. (21)–(28) and expressing
the concentrations of components as the ratio of the
number of moles and the phase volume, the model
equations become functions of the total number of
moles of the components:
H
1
s 1 A H P Waq aq o aq o
A 1 P
d
d
( )
N
t
mC k N N N N
V V K V K V K V
(30)
H HH
1
d d
d d
N NF
t t (31)
A H
1
P P D A A2 3 Eaq o aq o
P A
d d
d d
n
N N
t t
K N N K Nk k N
V K V V K V
(32)
P H P P D2 aq o
P1
d d
d d
N N K N Nk
t t V K V (33)
W HW
1
d d
d d
N NF
t t (34)
D P P D2 aq o
P
d
d
N K N Nk
t V K V (35)
E P P D A A2 3 Eaq o aq o
P A
d
d
nN K N N K N
k k Nt V K V V K V
(36)
HA A A3 E aq o
A
d
d
nN K N
k Nt V K V
(37)
Instead of using the number of moles of com-
ponent j (Nj) in the model Eqs. (30)-(37), the amount
of component j in the reaction system can be expres-
sed as the number of moles of component j per 100 g
of oil ([j]).
For the regression of the experimental data by
the proposed model, some approximations were
applied as follows.
The volumes of the organic and aqueous phases
were accepted as the sums of the components’
volumes. For this purpose, the influence of tempe-
rature on the densities of both phases was neglected.
The volume of the aqueous phase may be expressed
per 100 g of oil as:
aq aq aq aq aqW H A Pv v v v v (38)
The volume of each component in the aqueous
phase per 100 g of oil was expressed as the ratio of
mass and density of the component. Further, the
mass was defined via the component’s molecular
mass and its amount, or its concentration in the
aqueous phase:
aq aq aq aqaq W H A A P P
W H A P
[W]M [H]M M MC v C vv (39)
where W , H , A and P (g/L) are densities of
water, hydrogen peroxide, acetic acid and peracetic
acid, respectively. By substituting the concentrations
of acetic acid and peracetic acid, expressed from
Eqs. (14) and (17), into Eq. (39), the volume of the
aqueous phase becomes:
The volume of the organic phase per 100 g of oil
is calculated from the material balance of the initial
double bonds, which undergo epoxidation and further
the acetylation, as follows:
aq aqaq W H A P
o aq o aqW H A A P P
[W]M [H]M [A] M [P] M
( ) ( )
v vv
K v v K v v (40)
o o o oo 0 0A A P P
oA P o 0
100[D] (100 76.054[D] )[HA] (100 16[D] )[E]M M
[D]
C v C vv (41)
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
106
where [D]0 (mol/100 g of oil) is the initial amount of
double bond in the oil; and oo (g/L) is the density of
the oil in the organic phase. Further, density of the oil
in the organic phase was calculated by the linear
interpolation of densities of soybean oil ( SO ) and
epoxidized soybean oil ( ESO ) for the particular double
bond and epoxy group concentrations in the oil:
oo SO ESO SO
ESO
EO
EO (42)
where EOESO (wt.%) is the epoxy oxygen content in
epoxidized soybean oil. By substituting the concentra-
tions of acetic acid and peracetic acid in the organic
phase, expressed on the basis of Eqs. (16) and (18),
into the Eq. (41), the volume of the organic phase
becomes:
The system of two nonlinear Eqs. (40) and (43),
with two unknowns, vaq and vo, may be solved by
some of numerical methods, such as the Newtonian
method of simultaneous solving the system of non-
linear equations. However, in this case, the sequential
determination of the unknowns’ values with a two
loops algorithm was applied to simplify the calcul-
ation. The value of vo was determined in the outer
loop, whereas the value of vaq was determined in the
inner loop for the current value of vo. In both loops,
the modified Newtonian method for nonlinear equa-
tions was applied. The calculation of the phases' vol-
ume must be run at each step of integration of the
differential Eqs. (30)-(37).
The amounts of acetic acid and peracetic acid in
one phase were calculated on the basis of the parti-
tion coefficients and known amounts of components
in the other phase. The partition coefficient for acetic
acid at temperature T was approximated by double
linear interpolation of the values of the partition coef-
ficient for acetic acid for the soybean oil-acetic acid-
–water and epoxidized soybean oil-acetic acid-water
systems:
A,SO, 1 A,SO,
A,SO A,SO,
1
p pp p
p p
K KK K T T
T T (44)
A,ESO,p 1 A,ESO,
A,ESO A,ESO,
1
pp p
p p
K KK K T T
T T (45)
A,SO A,ESO
A
D E
D E
K KK (46)
where KA,SO, KA,SO, p and KA,SO,p+1 are the partition
coefficients for acetic acid between the soybean oil
and water at the temperatures T, Tp and Tp+1, res-
pectively; KA,ESO, KA,ESO,p and KA,ESO,p+1 are the parti-
tion coefficients for acetic acid between the epox-
idized soybean oil and water at the temperatures T,
Tp and Tp+1, respectively; and Tp and Tp+1 are the tem-
peratures given in Table 2 with the closest values to
the T. The values of the partition coefficients for these
three-component systems were calculated on the
basis of the experimental data given in the literature
[34,36]. Further, it was assumed that the value of the
partition coefficient for peracetic acid is always 2.5
times higher than the value of the partition coefficient
for acetic acid [5].
Table 2. Values of the partition coefficient for acetic acid in the
systems soybean oil-acetic acid-water (SO-A-W) and epoxid-
ized soybean oil-acetic acid-water (ESO-A-W) calculated on the
basis of experimental data reported the literature [34,36]
T / K Partition coefficient for acetic acid
SO-A-W ESO-A-W
323 0.04255 0.1647
338 0.04568 0.1620
353 0.04867 0.1860
For the temperature dependency of the chem-
ical equilibrium constant for peracetic acid formation
from acetic acid and hydrogen peroxide in an aque-
ous solution (K1), the expression reported in the
literature was applied [41]:
1
6 2
exp(12.2324ln 0.0229913
9.70452 10 3045.76 / 72.8758)
K T T
T T (47)
where T (K) is the temperature.
A drop-wise addition of the hydrogen peroxide
solution to the reaction mixture is approximated with
continuous flows of hydrogen peroxide (FH) and water
(FW), both in (mol/(min·100 g oil)):
1 1HS H H HS HS
H
HS
0
m w M t t tF
t t (48)
1 1HS H W HS HS
W
HS
(1 )
0
m w M t t tF
t t (49)
o oo 0 0A P
oaq aq
o 0o oA P
PA
100[D] (100 76.054[D] )[HA] (100 16[D] )[E][A] M [P] M
[D]
v vv
V vv v
K K
(43)
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
107
where mHS (g) is the mass of the hydrogen peroxide
aqueous solution; wH is the mass fraction of hydrogen
peroxide in the solution; MH and MW (g/mol) are mol-
ecular masses of hydrogen peroxide and water; and
tHS (min) is the duration of hydrogen peroxide solution
addition.
The temperature dependency of the rate coef-
ficient ki for the reaction i is expressed with repara-
meterized form of the Arrhenius equation [42]:
a,E
,0
1 1exp
i
i i
a
kk k
R T T (50)
where ki,0 and a,Eik are the modified Arrhenius
equation constants related to the frequency coef-
ficient and the activation energy, respectively; R
(8.3143 J/(mol K)) is the universal gas constant; and
Ta (K) indicates an average temperature of experi-
ments.
Kinetic parameters
Prior to fitting the experimental data with the
proposed model, it is necessary to define the time
dependency of the reaction temperature. In this study,
the hydrogen peroxide aqueous solution addition to
the reaction mixture was isothermal. The increase in
temperature to the desired reaction temperature was
approximated as linear with reaction time. Further, the
epoxidation was run isothermally. These temperature
changes are described mathematically as follows:
where THS (K) is the temperature of the hydrogen
peroxide solution addition; Tr (K) is the temperature of
the reaction; and tTI (min) is the period of the
temperature increase.
The constants of the reparameterized Arrhenius
equation, ki,0 and a,Eik , were determined by fitting the
values of double bond [D] and epoxy group [E]
amount changes with reaction time (t) for the soybean
oil epoxidations run under different reaction condi-
tions. The Marquardt method was used to fit the data
[43]. The following objective function (F) was mini-
mized:
NSNRN 2 2calc exp calc exp
, , , ,1 1
F D D E Ek
k l k l k l k lk l
(52)
where NRN indicates the number of runs and NSk is
the number of samples in run k; [D] and [E] are
determined as [D] = IN/[2AI] and [E] = EO/[100AO],
where AI is the atomic mass of iodine and AO is the
atomic mass of oxygen; and superscripts calc and
exp indicate the calculated and experimentally deter-
mined value, respectively.
The model’s system of differential Eqs. (30)-(37)
was integrated by applying a fourth order Runge-
–Kutta method.
The model parameters, i.e., the rate coefficients
ki for the investigated reactions were calculated when
the average temperature of the experiments (Ta) was
accepted as 338 K and assuming the first and the
second order of the epoxy ring opening reaction with
respect to acetic acid. The results are presented in
Table 3 together with the values of the least sum of
squares (objective function F) and average absolute
deviations for amounts of double bond (AAD[D]) and
epoxy group (AAD[E]).
The least sum of squares is lower (0.06519)
when the order of the side reaction was assumed as
the first. The temperature dependencies of the kinetic
parameters for such assumption are determined as
follows:
1
83959.13 1 1exp 10.15365s
a
k CR T T
(53)
2
1425.418 1 1exp 2.564719
a
kR T T
(54)
3
6873.391 1 1exp 6.143859
a
kR T T
(55)
Note that the rate coefficients are not intrinsic
but apparent, since the effect of the mass transfer
resistance on the kinetics was not taken into con-
sideration. Also, the rate coefficient for peracetic acid
formation is expressed in combination with the con-
centration of the active catalyst sites (Cs). An increase
of all rate coefficients with an increase in temperature
was obtained. Since all calculated reactant and pro-
duct amounts have positive values, the viability of the
developed kinetic model is confirmed. Due to lower
average absolute deviation of 0.01258 than 0.01530,
it can be concluded that the model fits the time vari-
ation of double bond amount slightly better than
epoxy group amount, respectively (Table 3). On the
basis of double bond and epoxy group amounts
HS HS
HS r HS HS TI TI HS, HS TI
HS TIr
( )( ) / ( )
( )
k
T t t
T T T T t t t t t t t t
T t t t
(51)
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
108
estimated with developed model, the iodine number
and epoxy oxygen content, respectively, were calcul-
ated. In Figures 2-4 are shown changes of the iodine
number and epoxy oxygen content over time for the
epoxidations of soybean oil with peracetic acid gener-
ated in situ in the presence of Amberlite IR-120H. The
proposed model fits the corresponding experimental
data reasonably well.
Comparison of the proposed model with the
pseudo-homogeneous model reported in the literature
The kinetic model proposed in this study is
established assuming the pseudo-homogeneity of the
catalyst with respect to the aqueous phase. Unlike the
other pseudo-homogeneous models found in the
literature [16,31,44], it takes into consideration the
partitioning of acetic acid and peracetic acid between
the organic and aqueous phases, as well as the
changing of the phases’ volumes during the process.
The model also takes into account the occurrence of
the reactions during the incremental addition of the
oxidizing agent solution and defined temperature
changes during the epoxidation process. These phen-
omena were already considered in the pseudo-homo-
geneous model developed for the in situ epoxidation
of castor oil [16]. In order to compare the latter
pseudo-homogeneous model with the model pro-
posed in this study, the experimental data obtained
for the epoxidation of soybean oil under defined react-
ion conditions (Table 1) were fitted. In both models,
the beginning of the addition of hydrogen peroxide
solution to the reaction mixture was considered as the
zero reaction time. Also, the variation of the chemical
equilibrium constant for peracetic acid formation with
temperature was defined with the same expression.
Further, the temperature dependencies of the kinetic
parameters were defined with the same Arrhenius
type model. The same objective function F was applied
for both regressions. The least sum of squares of
0.06519 obtained for the model developed in this
study is more than 44% lower than those of 0.11800
obtained for the reported model when the second
order of the epoxy ring opening reaction with respect
to acetic acid was assumed (Table 3). This confirmed
that the improved model better describes the reaction
system for the epoxidation of soybean oil conducted
under the investigated conditions. Since it was shown
in the literature [16] that the model developed for the
epoxidation of castor oil fits the experimental data
better than the two pseudo-homogeneous kinetic
models proposed by other authors [31,44], it can be
concluded that the model proposed in this study cor-
relates the change of double bond and epoxy group
contents during the in situ epoxidation of vegetable
oils better than the other reported pseudo-homo-
geneous models.
Table 3. Statistical values of the model parameters determination when the models proposed in this study and in the literature [16] were
applied for the in situ epoxidation of soybean oil. The order of the epoxy ring opening reaction with respect to acetic acid was 1 or 2. The
constants of the reparameterized Arrhenius equation for compared models are given
Parameter
This study Reference [16]
Order of the epoxy ring opening reaction, n
1 2 1 2
Error
F 0.06519 0.06764 0.12250 0.11800
AAD[D]a 0.01258 0.01275 0.01999 0.01953
AAD[E]b 0.01530 0.01556 0.01937 0.01879
Constants of the reparameterized Arrhenius equation
(k1Cs)0 -10.15365 -10.20897 -5.171507 -5.172997
(k1Cs)Ea 83959.13 82829.17 79854.73 79039.30
k2,0 -2.564719 -2.501162 -1.332284 -1.282917
k2,Ea 1425.418 -1040.259 -11635.34 -13083.98
k3,0 -6.143859 -4.765239 -5.737784 -4.010654
k3,Ea 6873.391 300.7545 -3704.474 -3174.165
aAverage absolute deviation for double bond amount:
NSNRNcalc exp
[D] NRN , ,1 1
1
1AAD D D
NS
k
k l k lk l
kk
baverage absolute deviation for epoxy group amount:
NSNRNcalc exp
[E] NRN , ,1 1
1
1AAD E E
NS
k
k l k lk l
kk
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
109
CONCLUSION
The epoxidations of soybean oil with peracetic
acid formed in situ from acetic acid and hydrogen
peroxide in the presence of an acidic ion exchange
resin as the catalyst were carried out under different
temperatures of incremental oxidizing agent addition,
reaction temperatures, catalyst concentrations and
stirring speeds. It was determined that the highest
epoxide yield was achieved at the reaction tempe-
rature of 338 K after 625 min when approximately 0.5
mol of glacial acetic acid and 1.35 mol of 30%
aqueous hydrogen peroxide solution per mole of
double bond of soybean oil were used. The epox-
idation was catalyzed with Amberlite IR-120H in the
amount of only 4.04 wt.% of oil. A kinetic model was
developed assuming the pseudo-homogeneity of the
catalyst with respect to the aqueous phase. The pro-
posed model took the partitioning of acetic acid and
peracetic acid between the organic and aqueous
phases into consideration, as well as the changing of
the phases’ volumes during the process. The tempe-
rature dependency of the kinetic parameters was det-
ermined and an increase in all reaction rate coef-
ficients with an increase in temperature was obtained.
A comparison was made between the proposed
model and the pseudo-homogeneous model reported
in the literature. The model developed in this study fits
the experimental data obtained for the in situ epox-
idations of soybean oil run under the investigated
conditions better than the kinetic model taken for the
comparison.
Acknowledgements
This work is part of the Project No. III45022
financially supported by the Ministry of Education,
Science and Technology Development of the Rep-
ublic of Serbia. The authors thank the German Aca-
demic Exchange Service (DAAD) and Institute of
Environmental Research (INFU), FR Germany, for
instrument donations.
Nomenclature
A-acetic acid
AI-atomic mass of iodine
AO-atomic mass of oxygen
AAD-average absolute deviation
a-interphase area (m2/m3)
+3H O
C -concentration of hydronium ions (mol/L)
Cj-concentration of component j (mol/L)
Cs-concentration of catalytically active sites (mol/g
catalyst)
D-double bond
E-epoxy group, i.e., ring
EO-epoxy oxygen content (wt.%)
EOESO-epoxy oxygen content in epoxidized soybean
oil (wt.%)
ESO-epoxidized soybean oil
F-objective function
Fj- molar flow of component j (mol/min)
H-hydrogen peroxide
HA-hydroxy acetate
IN-iodine number (g iodine/100 g oil)
IN0-initial iodine number (g iodine/100 g oil)
[j]-amount of component j (mol/100 g oil)
K1-chemical equilibrium constant for peracetic acid
formation
Kj-partition coefficient for component j
KA,SO-partition coefficient for acetic acid in the system
soybean oil-acetic acid-water
KA,ESO-partition coefficient for acetic acid in the system
epoxidized soybean oil-acetic acid-water
ki-rate coefficient for reaction i
ki,0-constant of reparameterized Arrhenius equation
related to the frequency coefficient
a,Eik -constant of reparameterized Arrhenius equation
related to the activation energy
kL,j-mass transfer coefficient for component j (m/min)
Mj-molecular mass of component j (g/mol)
m-mass of the catalyst (g)
mHS-mass of the hydrogen peroxide aqueous solution
(g)
Nj-number of moles of component j (mol)
NR-total number of reactions
NRN-number of runs
NSk-number of samples in run k
P-peracetic acid
R-universal gas constant (J/mol·K)
ri-rate of reaction i
REY-relative epoxy yield (%)
S-stirring speed (rpm)
SE-selectivity
SO-soybean oil
T-temperature (K)
Ta-average temperature of runs (K)
t-reaction time (min)
tHS-period of hydrogen peroxide aqueous solution
addition (min)
tTI-period of temperature increase (min)
V-volume (L)
v-volume (L/100 g oil)
W-water
wH-mass fraction of hydrogen peroxide in its aqueous
solution
X-conversion of double bond (%)
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
110
Greek letters
αj,i-stoichiometric coefficient of component j in the
reaction i
j -density of component j (g/L)
oo -density of oil in the oil phase (g/L)
SO -density of soybean oil (g/L)
ESO -density of epoxidized soybean oil (g/L)
Subscript
0-initial value
m-maximum value
th-theoretical value
Superscript
aq-aqueous phase
calc-calculated value
exp-experimentally determined value
n-order of reaction
o-organic phase
REFERENCES
[1] C. Cai, H. Dai, R. Chen, C. Su, X. Xu, S. Zhang, L. Yang,
Eur. J. Lipid Sci. Technol. 110 (2008) 341-346
[2] Z.S. Petrović, Polym. Rev. 48 (2008) 109-155
[3] Z.S. Petrović, A. Zlatanić, C.C. Lava, S. Sinadinović-
Fišer, Eur. J. Lipid Sci. Technol. 104 (2002) 293-299
[4] J. Zheng, J. Wärnå, T. Salmi, F. Burel, B. Taouk, S.
Leveneur, AIChE J., doi: 10.1002/aic.15037
[5] B. Rangarajan, A. Havey, E. Grulke, P. Dean Culnan, J.
Am. Oil Chem. Soc. 72 (1995) 1161-1169
[6] S. Dinda, A.V. Patwardhan, V.V. Goud, N.C. Pradhan,
Bioresour. Technol. 99 (2008) 3737-3744
[7] V.V. Goud, A.V. Patwardhan, N.C. Pradhan, Bioresour.
Technol. 97 (2006) 1365-1371
[8] V.V. Goud, N.C. Pradhan, A.V. Patwardhan, J. Am. Oil
Chem. Soc. 83 (2006) 635-640
[9] R. Mungroo, N. Pradhan, V. Goud, A. Dalai, J. Am. Oil
Chem. Soc. 85 (2008) 887-896
[10] R. Mungroo, V. Goud, S.N. Naik, A. Dalai, Eur. J. Lipid
Sci. Technol. 113 (2011) 768-774
[11] S. Dinda, V.V. Goud, A.V. Patwardhan, N.C. Pradhan,
Asia-Pac. J. Chem. Eng. 6 (2011) 870-878
[12] V.V. Goud, A.V. Patwardhan, S. Dinda, N. Pradhan,
Chem. Eng. Sci. 62 (2007) 4065-4076
[13] V.V. Goud, A.V. Patwardhan, S. Dinda, N. Pradhan, Eur.
J. Lipid Sci. Technol. 109 (2007) 575-584
[14] V.V. Goud, A.V. Patwardhan, N. Pradhan, Ind. Eng.
Chem. Res. 46 (2007) 3078-3085
[15] S. Sinadinović-Fišer, M. Janković, Z.Petrović, J. Am. Oil
Chem. Soc. 78 (2001)725-731
[16] S. Sinadinović-Fišer, M. Janković, O. Borota, Chem. Eng.
Process. 62 (2012) 106-113
[17] V.V. Goud, A.V. Patwardhan, N.C. Pradhan, Solid State
Sci. Technol. 14 (2006) 62-68
[18] E. Monono, D. Hagenson, D. Wiesenborn, Ind. Crop
Prod. 67 (2015) 364-372
[19] E. Monono, D. Webster, D. Wiesenborn, Ind. Crop Prod.
74 (2015) 987-997
[20] L. Rios, D. Echeverri, A. Franco, Appl. Catal., A 394
(2011) 132-137
[21] A. Campanella, B.A. Mandagarán, Eur. J. Lipid Sci.
Technol. 106 (2004) 524-530
[22] K. Dorfner, In: Ion Exchangers, K. Dorfner (Ed.), Walter
de Gruyter, New York, 1991, p. 366
[23] S. Leveneur, D. Yu. Murzin, T. Salmi, J-P. Mikkola, N.
Kumar, K. Eränen, L. Estel, Chem. Eng. J. 147 (2009)
323-329
[24] S. Leveneur, C.A. de Araujo Filho, L. Estel, T. Salmi, Ind.
Eng. Chem. Res. 51 (2012) 189-201
[25] E. Santacesaria, A. Renken, V. Russo, R. Turco, R. Tes-
ser, M. Di Serio, Ind. Eng. Chem. Res. 51 (2012) 8760-
8767
[26] M. Janković, S. Sinadinović-Fišer, Hem. Ind. 58 (2004)
569-576
[27] J. la Scala, R.P. Wool, J. Am. Oil Chem. Soc. 79 (2002)
373-378
[28] R. Mungroo, V. Goud, S. Narayan Naik, A. Dalai, Asia-
–Pac. J. Chem. Eng. 6 (2011) 14-22
[29] E. Santacesaria, R. Tesser, M. Di Serio, R. Turco, V.
Russo, D. Verde, Chem. Eng. J. 173 (2011) 198-209
[30] R.L. Musante, R.J. Grau, M.A. Baltanás, Appl. Catal., A
197 (2000) 165-173
[31] L.H. Gan, S.H. Goh, K.S. Ooi, J. Am. Oil Chem. Soc. 69
(1992) 347-351
[32] S. Leveneur, J. Zheng, B. Taouk, F. Burel, J. Wärnå, T.
Salmi, J. Taiwan Inst. Chem. Eng. 45 (2014) 1449-1458
[33] M. Janković, S. Sinadinović-Fišer, O. Govedarica, Ind.
Eng. Chem. Res. 53 (2014) 9357-9364
[34] S. Sinadinović-Fišer, M. Janković, J. Am. Oil Chem. Soc.
84 (2007) 669-674
[35] A. Campanella, B.A. Mandagarán, E.A. Campanella, J.
Am. Oil Chem. Soc. 86 (2009) 513-519
[36] M. Janković, S. Sinadinović-Fišer, J. Am. Oil Chem. Soc.
87 (2010) 591-600
[37] Standard Methods for the Analysis of Oils, Fats and Deri-
vatives, Blackwell Scientific Publications, London, 1987
[38] A. Campanella, M.A. Baltanás, Chem. Eng. J. 118 (2006)
141-152
[39] A. Campanella, B.A. Mandagarán, Chem. Eng. Process.
46 (2007) 210-221
[40] G.F. Froment, B.K. Bischoff, Chemical Reactor Analysis
and Design, John Wiley & Sons, New York, 1979
[41] M. Janković, S. Sinadinović-Fišer, J. Am. Oil Chem. Soc.
82 (2005) 301-303
[42] F.J. Dumez, G.F. Froment, Ind. Eng. Chem. Proces. Des.
Dev. 15 (1976) 291-301
[43] D.W. Marquardt, J. Soc. Ind. Appl. Math. 11 (1963) 431-
–441
[44] J. Wisniak, E. Navarrete, Ind. Eng. Chem. Prod. Res.
Dev. 9 (1970) 33-41.
M. JANKOVIĆ et al.: KINETICS OF SOYBEAN OIL EPOXIDATION… Chem. Ind. Chem. Eng. Q. 23 (1) 97111 (2017)
111
MILOVAN JANKOVIĆ
SNEŽANA
SINADINOVIĆ-FIŠER
OLGA GOVEDARICA
JELENA PAVLIČEVIĆ
JAROSLAVA
BUDINSKI-SIMENDIĆ
Univerzitet u Novom Sadu, Tehnološki
fakultet Novi Sad, Bulevar cara Lazara
1, 21000 Novi Sad, Srbija
NAUČNI RAD
KINETIKA EPOKSIDOVANJA SOJINOG ULJA PERSIRĆETNOM KISELINOM FORMIRANOM IN SITU U PRISUSTVU JONOIZMENJIVAČKE SMOLE: PSEUDO-HOMOGENI MODEL
Predložen je kinetički model epoksidovanja biljnih ulja persirćetnom kiselinom formiranom
in situ iz sirćetne kiseline i vodonik-peroksida u prisustvu jonoizmenjivačke smole kao
katalizatora. Model je pseudo-homogen u odnosu na katalizator. Pored osnovnih reakcija
formiranja persirćetne kiseline i epoksidne grupe, model opisuje i sporednu reakciju otva-
ranja epoksidne grupe sa sirćetnom kiselinom. U modelu se razmatraju i raspodela sir-
ćetne i persirćetne kiseline između vodene i organske faze sistema i promena zapremina
faza tokom odvijanja procesa epoksidovanja. Temperaturna zavisnost prividnih koeficije-
nata brzina reakcija je opisana reparametrizovanom Arrhenius jednačinom. Kinetički para-
metri predloženog modela su izračunati fitovanjem eksperimentalnih podataka dobijenih
tokom epoksidovanja sojinog ulja izvođenih pri definisanim reakcionim uslovima. Najveći
prinos epoksida od 87,73% je postignut pri temperaturi od 338 K kada je molski odnos
nezasićenost ulja:sirćetna kiselina:vodonik-peroksid iznosio 1:0.5:1.35 i kada je prime-
njena količina katalizatora Amberlite IR-120H bila 4,04 mas.% u odnosu na ulje. U pore-
đenju sa publikovanim pseudo-homogenim modelima, model predložen u ovom radu bolje
koreliše promene sadržaja dvostrukih veza i epoksidnih grupa tokom procesa epoksido-
vanja.
Ključne reči: sojino ulje, epoksidovanje, persirćetna kiselina, jonoizmenjivačka
smola, kinetika.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 113119 (2017) CI&CEQ
113
SALAH H. ALJBOUR1
SULTAN A. TARAWNEH2
ADNAN M. AL-HARAHSHEH1
1Department of Chemical
Engineering, College of
Engineering, Mutah University,
Karak, Jordan 2Department of Civil &
Environmental Engineering,
College of Engineering, Mutah
University, Karak, Jordan
SCIENTIFIC PAPER
UDC 669.1.054.8:666.94:62
https://doi.org/10.2298/CICEQ151002016A
EVALUATION OF THE USE OF STEELMAKING SLAG AS AN AGGREGATE IN CONCRETE MIX: A FACTORIAL DESIGN APPROACH
Article Highlights
• A factorial design methodology was applied to evaluate concrete mix production
• Steel slag was evaluated as an aggregate in concrete mix production
• Influential factors on the compressive strength were proposed
• Possible factor-interaction effects were examined
Abstract
Slag is investigated towards its potential use as an aggregate in concrete mix
production. Full factorial design methodology is applied to study the effect of two
process input variables, namely: slag as coarse aggregate and slag as medium
aggregate on the properties of concrete mix. Additionally, the interaction
between input variables is also examined. Incorporating steel slag aggregate in
the concrete mix affected its compressive strength. Enhanced compressive
strength concrete mix was obtained with 70 wt.% coarse slag aggregate and 70
wt.% medium slag aggregate. Under these proportions, the 28-days compres-
sive strength was higher than the 28-days compressive strength of a concrete
mix prepared from normal aggregate. Strong interaction effect exists between
slag aggregate size on the compressive strength at 7-days curing. Lower com-
pressive strength for the concrete mix might be obtained if improper proportions
of mixed medium and coarse slag aggregate were employed.
Keywords: steelmaking slag; concrete; factorial design; compressive strength.
The aggregates typically account for about 75%
of the concrete volume and play a substantial role in
different concrete properties such as workability,
strength, dimensional stability and durability. Conven-
tional concrete consists of sand as fine aggregate and
gravel, limestone or granite in various sizes and
shapes as coarse aggregate. There is a growing inte-
rest in using waste materials as alternative aggregate
materials and significant research is made on the use
of many different materials as aggregate substitutes
such as coal ash, blast furnace slag, and steel slag
aggregate. This type of use of a waste material can
solve problems of lack of aggregate in various con-
struction sites and reduce environmental problems
Correspondence: S.H. Aljbour, Department of Chemical Engine-
ering, College of Engineering, Mutah University, Karak, Jordan. E-mail: [email protected] Paper received: 2 October, 2015 Paper revised: 12 February, 2016 Paper accepted: 17 March, 2016
related to aggregate mining and waste disposal. The
use of waste aggregates can also reduce the cost of
the concrete production [1].
Jordan’s steel and iron industry began with the
establishment of the first local steel manufacturer,
Jordan Iron and Steel in 1965. Due to continuous and
rising need for steel, twelve local steel mills are now
established in Jordan with a total annual production
capacity of 1.2 million t [2]. Some of these factories
utilize imported semi processed iron in the form of
plates to be melted and processed in a later stage to
produce reinforced bars. The other type of factories
utilizes iron scrap for the production of reinforced
bars. It is estimated that the annual production of slag
from these factories is around 100-200 thousand
tons. Inventing new ways to reuse this accumulated
waste is the most pressing and daunting challenge
that face Jordan’s industrial sector.
The primary components of iron and steel slag
are limestone and some other materials in oxide form.
S.H. ALJBOUR et al.: EVALUATION OF THE USE OF STEELMAKING SLAG… Chem. Ind. Chem. Eng. Q. 23 (1) 113119 (2017)
114
In the case of steelmaking slag, the slag contains
metallic elements such as iron in oxide form; how-
ever, because refining time is short and the amount of
limestone contained is large, a portion of the lime-
stone auxiliary material may remain un-dissolved as
free CaO [3]. In general, several factors are affecting
physical and chemical properties of steel slag. These
factors include type of steel furnace in the steel-
making plant and the method of steel slag processing.
Various approaches can be followed for the utilization
of steel slag in cement and concrete applications.
One may think about using steel slag in the product-
ion of cement. In this approach, the steel slag is
mixed with limestone and clay as a raw material feed
to cement kiln. In this case, the slag must be clinkered
[4]. Another approach is the incorporation of steel
slag in cement and composite cements [5]. In addition
steel slag can be utilized as an aggregate material.
Several advantages of using slag aggregates in con-
crete mixes are gained such as: reliable quality, inc-
reased strength and does not generate alkali-aggre-
gate reactions. In addition, blast furnace slag fine
aggregate does not contain materials that may affect
the strength and durability of concrete, such as chlo-
rides, organic impurities clay and shells [5,6].
Extensive research has been conducted for the
application of steel slag in broad areas of construct-
ion. It is vital to quantify the benefits of using such
cheap material in concrete technology and concrete
asphalt pavement [6-14]. In this study, we evaluate
the beneficial use of steel slag obtained from Jordan’s
steel industry as an aggregate in concrete mix pro-
duction. Our investigation involves studying the effect
of using steel slag of different grades when combined
with normal aggregates by different ratios on impro-
ving the mechanical properties of hardened concrete.
The effect of medium slag aggregate and coarse slag
aggregate and the combination between them on the
mechanical quality of concrete mix is investigated by
applying a 22 full factorial design methodology. The
main objectives of this study are to identify the most
influential process operating conditions on the pro-
duction of concrete mix and the interaction effects
among variables on the compressive strength of the
produced concrete.
EXPERIMENTAL
Preparation of the steel slag-based aggregate
Steel slag was provided by local Jordanian fac-
tory in the form of boulders (size < 100 mm). The
factory produces steel products made from scrap
metal, recycled from used automobiles, plant equip-
ment, machinery, or byproducts of the manufacturing
and construction sectors by utilizing an electric arc
furnace (EAF) for scrap melting and a ladle furnace
for the precision control of chemistry and the puri-
fication of the liquid steel.
The majority of the steel slag contains free CaO
and MgO. Experiments must be performed to inves-
tigate the content of free CaO and MgO in the slag. In
general, the content of free CaO and MgO in EAF-
-slag is significantly lower than in basic oxygen fur-
nace steel slag (BOF slag) [1]. Slag pretreatment is
necessary to reduce the content of free CaO and
MgO in the slag. The ageing or weathering method
was followed to reduce the content of free CaO and
MgO in the EAF slag. It has been reported that that a
proper treatment aimed to stabilize slag by exposing
them to outdoor weather and regular spraying for at
least 90 days, may eliminate any subsequent exp-
ansive phenomenon, allowing a safe use of such slag
as aggregate in concrete production [15]. Prior to use,
the EAF slag was aged for a period of 6 months. The
air aging method was applied by leaving the EAF slag
out in an open area to enable weathering. The pre-
sence of free CaO and MgO in the slag does not
seem to represent a limit for the durability of concrete,
due to their stabilization in crystalline lattice [16].
The aged EAF slag aggregate was prepared by
crushing the boulders, followed by sorting the ground
slag by sieving. In this study, medium and coarse
EAF slag was used. The medium EAF slag aggregate
was obtained from the sieved material which passed
through the 12.5 mm sieve and retained on the 4.75
mm sieve. The coarse EAF slag aggregate was
obtained from the sieved material which passed
through the 19 mm sieve and retained on the 12.5
mm sieve. Physical and chemical properties of the
EAF slag aggregates are given in Table 1. The mech-
anical properties of the EAF slag aggregates were
conducted according to (ASTM C 33, 2003; ASTM C
138/C 138M, 2001; ASTM C 150, 2005).
Table 1. Mechanical properties of aggregate used in this study
Parameter Steel slag
aggregate
Natural
aggregate
Saturated density, kg/m3 2323 2020-2100
Apparent specific gravity 3.2 -
Abrasion value, % 19.4 20-24
Flakiness index 10.98 20-30
Elongation index 9.89 10
Crushing value, % 26.1 -
S.H. ALJBOUR et al.: EVALUATION OF THE USE OF STEELMAKING SLAG… Chem. Ind. Chem. Eng. Q. 23 (1) 113119 (2017)
115
Preparation of normal aggregate
The normal coarse, medium and fine aggregate
source for all mix designs was obtained from gravel
pits. The gravel was obtained from local area known
in Jordan as “Al-Ghoor”. The gravel was crushed then
sieved to obtain the desired size fraction. The fine
aggregate was obtained from sieved material which
passed through the 4.75 mm. The medium aggregate
was obtained from the sieved material which passed
through the 12.5 mm sieve and retained on the 4.75
mm sieve. The coarse aggregate was obtained from
the sieved material which passed through the 19 mm
sieve and retained on the 12.5 mm sieve.
Preparation of concrete mix
All concrete mixes were prepared by keeping
the water-to-cement ratio and fine aggregate-to-total
aggregate ratio constant at 0.6 and 0.35, respectively.
No additives were used for the preparation of con-
crete mix. The high water-to-cement ratio was applied
to ensure good workability conditions since no addi-
tives were used for concrete mix preparation and to
account for water absorption by aggregates. Cement
used in this study was Ordinary Portland Cement pro-
duced by a local Jordanian factory. The cement is
classified as CEM-1 42.50 N. The chemical compo-
sition of cement is as follows: 19.94% SiO3, 5.37%
Al2O3, 3.18% Fe2O3, 63.65% CaO, 2.59% MgO,
2.88% SO3, 0.82% K2O and 0.1% Na2O [17].
The materials were added to the concrete mix-
ture in the following order: coarse aggregate, medium
aggregate, fine aggregate and cement. The mixture
was mixed under dry condition for about 1 min, then
80 % of water was added. After 1.5 min of mixing, the
rest of water was added. Every batch of concrete mix
was mixed for a total time of 3 min. The concrete
mixes were cast in steel molds (150 mm150
mm150 mm) and compacted using a tamping rod.
One day after casting, the concrete samples were
removed from the mold and cured in a tank of water
at a temperature of 20 C for 3, 7 and 28 days. All
concrete mix are prepared by keeping the cement
content as 425 kg/m3, medium-to-total aggregate ratio
of 0.38 and coarse-to-total aggregate ratio of 0.6.
Table 2 shows the mix proportions for the mixes
applied in this study.
Characterization of samples
Sample characterization was carried out by
examining the compressive strength. The compres-
sive strength of concrete was determined according
to ASTM C-39.
Experimental factorial design and analysis
Design of experiment is a powerful tool that can
be used in a wide spectrum of experimental situ-
ations. Design of experiments allows for multiple input
factors to be studied and to determine their effect on a
desired process/design/quality output. When studying
multiple inputs at the same time, design of experiment
can identify important interactions that may missed
when experimenting with one variable at a time
(OVAT approach). All possible combinations between
process input variables can be investigated by con-
ducting full factorial design. The factorial design
methodology can be utilized to confirm possible input/
/output relationships and to develop a predictive
equation suitable for performing design simulations
with minimum time and cost.
In this research, EAF slag is incorporated as
aggregate during the production of concrete mix.
Most of studies concerning concrete containing slag
aggregate are conducted by adopting the OVAT
approach. In these studies, a given grade of con-
ventional aggregate are replaced partially or totally by
its counterpart slag aggregate with the remaining
factor held constants. The effect of such replacement
of either fine, medium or coarse aggregate is then
solely evaluated. This approach provides an estimate
of the effect of a single variable at a selected fixed
condition of the other variables. However, for such an
estimate to have general relevance it is necessary to
check whether the effect would be the same at other
settings of the other variables or not. Nevertheless,
studies on the effect of combined replacement of
Table 2. Mix design parameters for concrete mixes applied in this study (kg/m3)
Component Run-1 Run-2 Run-3 Run-4 Run-5
Cement 425 425 425 425 425
Water 255 255 255 255 255
Fine aggregate 510 510 510 510 510
Medium slag aggregate 138 415 138 415 0
Medium normal aggregate 415 138 415 138 553
Coarse slag aggregate 223 223 668 668 0
Coarse normal aggregate 668 668 223 223 891
S.H. ALJBOUR et al.: EVALUATION OF THE USE OF STEELMAKING SLAG… Chem. Ind. Chem. Eng. Q. 23 (1) 113119 (2017)
116
mixed grade of aggregate on the mechanical pro-
perties of concrete at the same time are rarely con-
ducted.
To achieve this goal, a full factorial design meth-
odology was followed to identify the main effects of
two processing factors on the mechanical properties
of concrete. The two factors studied were replace-
ment percentage of medium conventional aggregate
by medium EAF slag aggregate (X1), and replace-
ment percentage of coarse conventional aggregate by
coarse EAF slag aggregate (X2). Each of the two
factors was studied at two levels (Table 3). Therefore,
the arrangement and number of experiments is
considered to be 2x2 or 22 factorial design. The four
formulations are shown in Table 4 with variable levels
coded with plus and minus signs. The prepared con-
crete mixes were subjected to the following tests (res-
ponse output variables): 3-days compressive strength
(Y1), 7-days compressive strength (Y2), and 28-days
compressive strength (Y3). In addition, one concrete
mix that contained 100% normal aggregate was pre-
pared and tested for its 3, 7 and 28-days compressive
strength.
Table 3. Process input variables and levels of variables
Input variable Slag as medium
aggregate, wt.% (X1)
Slag as a coarse
aggregate, wt.% (X2)
Level Low (-) High (+) Low (-) High (+)
Condition 30 70 30 70
The main effect of each variable is computed
using the following equation [18]:
1 1E Y Y
where 1Y is average response for the high level of
the variable and 1Y is average response for the low
level of the variable.
The interaction ( A,BI ) between two process
parameters (say, A and B) can be computed using the
following equation [18,19]:
A,B A,B( 1) A,B( 1)
1
2I E E
where A,B( 1)E is the effect of factor “A” at high level of
factor “B” while A,B( 1)E is the effect of factor “A” at low
level of factor “B”.
RESULTS AND DISCUSSION
Main and interaction effects of process input variables
on compressive strength
The main effects of process variables on the 3, 7
and 28 day compressive strength have been studied.
Table 5 shows the calculated main effects based on
all experimental observations. Figure 1 shows the
main effect plots for the studied process variables.
The absolute value of the effect of EAF slag
aggregate grade on the 3-days compressive strength
was the same. Very slight effect was noticed for both
aggregate grades at any proportion on the 3-days
compressive strength. Concrete mix made of either
30 or 70 wt.% coarse EAF slag aggregate and con-
crete mix made of either 30 or 70 wt.% medium EAF
slag aggregate exhibited almost the same (21.9–23.4
MPa) 3-days compressive strength. In addition, con-
crete mix made from normal aggregate without incor-
porating any of the steel slag aggregate possessed
nearly the same 3-days compressive strength of 22.8
MPa. It seems that three days of curing was not
sufficient enough to produce strong adhesion between
the cement paste and the aggregate, thereby, giving
close value of compressive strength. To assess the
viability of effect of parameters and confirm the exist-
ence or absence of interaction effects between para-
meters, we constructed the interaction plot shown in
Figure 2.
Table 5. Full 22 factorial design analysis of process response
Term Response
Y1 Y1 Y3
Main Effect of X1 -1.6 -2.2 4.6
Main Effect of X2 -1.6 3.7 6.3
Interaction effect X1X2 0.5 3.7 -1.9
Figure 2 shows that there is week interaction
effect between parameters, that is, the main effect of
Table 4. Experimental design matrix and response of output variables
Run In put variables Output variables
X1 X2 Y1 / MPa Y2 / MPa Y3 / MPa Slump, cm
1 - - 24.4 31.1 34.6 0
2 + - 22.4 25.3 41.1 2
3 - + 22.4 31.1 42.8 1
4 + + 21.3 32.6 45.5 2
5 0% 0% 22.8 31.1 40.9 0
S.H. ALJBOUR et al.: EVALUATION OF THE USE OF STEELMAKING SLAG… Chem. Ind. Chem. Eng. Q. 23 (1) 113119 (2017)
117
coarse EAF slag aggregate proportion in the concrete
mix on the 3-days compressive strength is indepen-
dent on the proportion of medium EAF slag aggregate
in the same mix. Similarly, the main effect of medium
EAF slag aggregate proportion in the concrete mix on
the 3-days compressive strength is independent on
the proportion of coarse EAF slag aggregate in the
same mix.
Figure 2. Interaction effect plot for coarse and medium slag
aggregate on 3-days compressive strength.
The 7-days curing enhanced the compressive
strength of concrete mixes. The 7-days compressive
strength of concrete mixes containing EAF slag
aggregate was larger than 3-days compressive
strength by at least 30%. The same results was
noticed for concrete mix made from normal aggre-
gate without incorporating any of the EAF slag aggre-
gates. The effect of medium EAF slag aggregate on
the 7-days compressive strength was slightly higher
at 30 wt.% replacement compared to 70 wt.%
replacement. On the other hand, the effect of coarse
EAF slag aggregate on the 7-days compressive
strength was slightly higher at 70 wt.% replacement
compared to 30 wt.% replacement. This trend in
results must also be judged by confirming existence
or absence of interaction effect between EAF slag
aggregate grades. This is shown in the interaction
plot given in Figure 3.
Figure 3. Interaction effect plot for coarse and medium slag
aggregate on 7-days compressive strength.
The interaction between the process input vari-
ables on the 7-days of compressive strength repre-
sents an antagonistic-type interaction [19]. At 75 wt.%
medium slag proportion, increasing the coarse EAF
slag aggregate proportion in the concrete mix from 30
to 70 wt.% substantially increased the 7-days com-
pressive strength from 25.3 to 32.6 MPa. However, at
30 wt.% medium EAF slag proportion increasing the
coarse EAF slag aggregate proportion in the concrete
mix from 30 to 70 wt.% kept the 7-days compressive
strength unchanged at 31 MPa. This interaction
effect between parameters can hardly be observed by
following the OVAT approach of experimentation. The
interaction effect shown in Figure 3 shows that local
maximum of 7-days compressive strength could be
obtained at 70 wt.% medium EAF slag aggregate
proportions and 70 wt.% coarse EAF slag aggregate
proportions in the concrete mix.
Figure 1. Main effect plots for process input variables (coarse and medium slag aggregate) on compressive strength.
S.H. ALJBOUR et al.: EVALUATION OF THE USE OF STEELMAKING SLAG… Chem. Ind. Chem. Eng. Q. 23 (1) 113119 (2017)
118
The 28-days curing enhanced the compressive
strength of concrete mixes. The 28-days compressive
strength of concrete mixes containing EAF slag
aggregates was larger than 3-days compressive
strength by at least 62%. Concrete mix made from
normal aggregate exhibited 40.9 MPa of 28-days
compressive strength with an enhancement by appro-
ximately 80% in reference to the 3-days compressive
strength of the concrete mix made from normal aggre-
gate. The effect of medium EAF slag aggregate on
the 28-days compressive strength was higher at 70
wt.% replacement compared to 30 wt.% replacement.
Similarly, the effect of coarse EAF slag aggregate on
the 28-days compressive strength was noticeably
higher at 70 wt.% replacement compared to 30 wt.%
replacement. This trend in results must also be jud-
ged by confirming existence or absence of interaction
effect between slag aggregate grades. This is shown
in the interaction plot given in Figure 4.
Figure 4. Interaction effect plot for coarse and medium slag
aggregate on 28-days compressive strength.
Small interaction between the process input
variables on the 28-days of compressive strength was
noticed within the concrete mix proportions selected
in this study. At 70 wt.% medium EAF slag proportion,
increasing the coarse EAF slag aggregate proportion
in the concrete mix from 30 to 70 wt.% increased the
28-days compressive strength from 41.1 to 45.5 MPa.
In the same manner, at 30 wt.% medium EAF slag
proportion increasing the coarse EAF slag aggregate
proportion in the concrete mix from 30 to 70 wt.%
increased the 28-days compressive strength from 34.6
to 42.8 MPa. The interaction effect shown in Figure 4
shows that local maximum of 28-days compressive
strength could be obtained at 70 wt.% medium EAF
slag aggregate proportion and 70 wt.% coarse EAF
slag aggregate proportion in the concrete mix.
CONCLUSION
Concrete mixes were successfully prepared by
utilizing slag generated from the steelmaking industry.
A full factorial design analysis was effectively per-
formed to assess the most influential process oper-
ating conditions on the compressive strength of con-
crete mixes made from different mixed proportions of
EAF slag aggregates. The proportions of the course
and medium size EAF slag aggregate were prominent
process variables found to be affecting the 7-days
and 28-days compressive strength of a concrete mix.
Under 3-days curing, there was no appreciable
enhancement on the compressive strength of con-
crete mixes prepared from EAF slag aggregate at any
proportion in comparison with concrete mixes prepared
from normal aggregates. Both type of mixes pos-
sessed almost the same 3-days compressive strength.
Under 7-days curing, strong interaction effect for
the proportion of EAF slag aggregates was noticed on
the compressive strength. Concrete mixes comprising
lower coarse slag aggregate content and higher
medium EAF slag content exhibited lower compres-
sive strength compared with concrete mix prepared
from normal aggregate. Nevertheless, concrete mixes
comprising higher coarse slag aggregate content and
any proportion of medium EAF slag content exhibited
almost the same compressive strength of concrete
mix prepared from normal aggregate.
Under 28-days curing, there was noticeable
enhancement on the compressive strength of con-
crete mixes prepared from high proportions of EAF
slag aggregate in comparison with concrete mixes
prepared from normal aggregates.
The mechanical properties of concrete mixes
prepared in this study are satisfactory. However, it is
recommended that other important properties such as
durability and corrosion of EAF slag concrete are to
be investigated before mass use.
Acknowledgement
Lab engineers: Wafaa Suhaimat and Hussein
Sarayreh of the civil engineering department and Ali
Alzoubi of the Chemical Engineering Department at
Mutah University are acknowledged for laboratory
assistance.
REFERENCES
[1] J. de Brito, N. Saikia, Recycled Aggregate in Concrete,
Use of Industrial, Construction and Demolition Waste,
Springer-Verlag, London, 2013, pp. 23,34
[2] Jordan Steel Company, Global Research, Global Invest-
ment House, Kuwait, Jan 2009
S.H. ALJBOUR et al.: EVALUATION OF THE USE OF STEELMAKING SLAG… Chem. Ind. Chem. Eng. Q. 23 (1) 113119 (2017)
119
[3] Chemical characteristics of iron and steel slag, Nippon
Slag Association, http://www.slg.jp/e/slag/character.html
(accessed 1 Oct, 2015)
[4] A. Monshia, M.K. Asgarani, Cem. Concr. Res. 29 (1999)
1373-1377
[5] S. Kourounis, S. Tsivilis, P.E. Tsakiridis, G.D. Papadi-
mitriou, Z. Tsibouki, Cem. Concr. Res. 37 (2007) 815-822
[6] S.A. Tarawneh., E.S. Gharaibeh, F.M. Saraireh, Am. J.
Appl. Sci. 11 (2014) 700-706
[7] F.H. Al-Negheismish, A.I. Al-Sugair, R.Z. Al-Zaid, J. King
Saud Univ., Eng. Sci. 9 (1996) 39-55
[8] M. Maslehuddin, A.M. Sharif, M. Shameem, M. Ibrahim,
M.S. Barry, Constr. Build. Mater. 17 (2003) 105-112
[9] H.Y. Moon, J.H. Yoo, J. Korea Concr. Inst. 11 (1999) 101-
–111
[10] P.T. Sherwood, Alternative Materials in Road Cons-
truction, A Guide to the Use of Recycled and Secondary
Aggregates, Thomas Telford Publishing, London, 2001,
p. 97
[11] I.M. Asi, H.Y. Qasrawi, F.L. Shalabi, Can. J. Civ. Eng. 34
(2007) 902-911
[12] J.M. Manso, J.J. Gonzalez, J.A. Polanco, J. Mater. Civ.
Eng. 16 (2004) 639-645
[13] H. Beshr, A.A. Almusallam, M. Maslehuddin, Constr.
Build. Mater. 17 (2003) 97-103
[14] R. Alizadeh, M. Chini, P. Ghods, M. Hoseini, Sh. Mon-
tazer, M. Shekarchi, Utilization of Electric Arc Furnace
Slag as Aggregates in Concrete - Environmental Issue, in
Proceedings of the 6th CANMET/ACI International Con-
ference on Recent Advances in Concrete Technology,
Bucharest, Romania (June 2003), p. 451
[15] C. Pellegrino, V. Gaddo, Cem. Concr. Compos. 31 (2009)
663-671
[16] C. Pellegrino, P. Cavagnis, F. Faleschini, K. Brunelli,
Cem. Concr. Compos. 37 (2013) 232-240
[17] H. Al-Hamaideh, O. Maaitah, S. Mahadin, Electron. J.
Geotech. Eng. 15 (2010) 601-608
[18] G.E.P. Box, W.G. Hunter, J.S. Hunter, Statistics for Expe-
rimenters –Design, Innovation and Discovery, John Wiley
and Sons Inc., Hoboken, NJ, 2005, p. 180
[19] J. Antony, Design of Experiments for Engineers and Sci-
entists. Elsevier Science & Technology, Burlington, 2003,
p. 19.
SALAH H. ALJBOUR1
SULTAN A. TARAWNEH2
ADNAN M. AL-HARAHSHEH1
1Department of Chemical Engineering,
College of Engineering, Mutah
University, Karak, Jordan 2Department of Civil & Environmental
Engineering, College of Engineering,
Mutah University, Karak, Jordan
NAUČNI RAD
PROCENA UPOTREBE ŠLJAKE IZ PROIZVODNJE ČELIKA KAO AGREGATA U CEMENTNIM MEŠAVINAMA POMOĆU PUNOG FAKTORIJALNOG PLANA
U radu je analizirana upotreba šljake kao potencijalnog agregata u proizvodnji cementa.
Primenjen je pun faktorijalni plan radi proučavanja uticaja dve ulazne procesne promen-
ljive, i to šljaka kao krupni agregat i šljaka kao srednji agregat na osobine cementne meša-
vine. Pored toga, ispitivana je interakcija između ulaznih promeljivih. Ubacivanje agregata
šljake iz proizvodnje čelika u cementnu mešavinu uticalo je na njenu pritisnu čvrstoću.
Povećana pritisne čvrstoće cementne mešavine sa 70 mas.% krupne šljake i sa 70 mas.%
srednje šljake. Pri ovim proporcijama, 28-dnevna pritisna čvrstoća dobijene cementne
mešavine je jača od 28-dnevne pritisne čvrstoće cementne mešavine pripremljene od nor-
malnih agregata. Postoji jaka interakcija između veličine agregata šljake i pritisne čvrstoće
posle sedmodnevnog očvršćavanja. Manja pritisna čvrstoća cementne mešavine mogla bi
biti dobijena ako bi se koristile agregati srednje i krupne šljake pomešani u neodgovara-
jućim proporcijama.
Ključne reči: bioetanol, pamuk, lignoceluloza, predtretman, hidroliza, fermentacija.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 121129 (2017) CI&CEQ
121
DUŠAN Lj. PETKOVIĆ
MILOŠ J. MADIĆ
GORAN M. RADENKOVIĆ
University of Niš, Faculty of Mechanical
Engineering, Niš, Serbia
SCIENTIFIC PAPER
UDC 669.14.018.8:546.175–323:51
https://doi.org/10.2298/CICEQ151127020P
THE EFFECTS OF PASSIVATION PARAMETERS ON PITTING POTENTIAL OF BIOMEDICAL STAINLESS STEEL
Article Highlights
• Corrosion resistance of 316LVM stainless steel was increased by passivation
• Multiple regression analysis and artificial neural network (ANN) were employed
• Only the ANN model provided a statistically accurate mathematical model
• Pitting potential is highly non-linearly dependent on the passivation parameters
• Nitric acid concentration has the strongest influence on the pitting potential
Abstract
Passivation is a chemical process in which the electrochemical condition of
passivity is gained on the surface of metal alloys. Biomedical AISI 316LVM
stainless steel (SS) can be passivized by means of nitric acid immersion in order
to improve a protective oxide layer on the surface and consequently increase
corrosion resistance of the SS in the physiological solutions. In this study, mul-
tiple regression analysis and artificial neural network (ANN) were employed for
mathematical modeling of the AISI 316LVM SS passivation process after immer-
sion in the nitric acid solution. The pitting potential, which represents the mea-
sure of pitting corrosion resistance, was chosen as the response, while the pas-
sivation parameters were nitric acid concentration, temperature and passivation
time. The comparison between experimental results and models predictions
showed that only the ANN model provided statistically accurate predictions with
a high coefficient of determination and a low mean relative error. Finally, based
on the derived ANN equation, the effects of the passivation parameters on pitting
potential were examined.
Keywords: stainless steel, nitric acid, passivation, multiple regression analysis, artificial neural networks.
AISI 316LVM is a vacuum melted stainless steel
(SS) widely used for biomedical applications. It has
high tensile strength and fatigue resistance, good
deformability, and relatively low price. Examples of its
biomedical applications include bone plates and
screws, hip and knee prosthesis, nails and pins, den-
tal prostheses as well as vascular and urological
stents [1]. The main disadvantages of this steel are
local corrosion susceptibility during prolonged contact
with human tissue, and release of metal ions [2].
Additionally, nickel is known as a strong immuno-
logical reaction medium and may cause various
Correspondence: D.Lj. Petković, University of Niš, Faculty of
Mechanical Engineering, A. Medvedeva 14, 18000 Niš, Serbia. E-mail: [email protected] Paper received: 27 November, 2015 Paper revised: 2 April, 2016 Paper accepted: 6 April, 2016
health problems [3]. Despite the above listed weak-
nesses, SS has the ability to spontaneously form a
stable self-protecting oxide layer (passive film) on its
surface in the reaction with air or most aqueous envi-
ronments. This film consists mostly of chromium
oxide (Cr2O3) and typically shows thicknesses of few
nanometers [4]. The presence of nonmetallic inc-
lusions on the material’s surface, such as sulfide inc-
lusions, represents a discontinuity of the passive film
and therefore a potential place of pitting corrosion
initiation [5].
Localized corrosion may cause an accidental
deterioration of the whole system with disastrous con-
sequences, while the total mass loss is insignificant
[6]. Corrosion of SS implants have two effects [7]:
first, the implant may become weak and the pre-
mature failure of the implant may happen; and sec-
D.Lj. PETKOVIĆ et al.: THE EFFECTS OF PASSIVATION PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 121129 (2017)
122
ond, the release of corrosion products from the imp-
lant can cause the tissue reaction.
Corrosion pits commonly start because of chem-
ical or physical heterogeneities at the surface, which
include dislocations, mechanical damage, inclusions,
or second phase particles [8]. The resistance of SS to
a pitting attack depends largely on the type of SS
used, and on the subsequent physicochemical pro-
perties of the protective passive oxide layer formed
on its surface [9]. There has been a constant attempt
by engineers and scientists to improve the surface-
related properties of biomedical materials to reduce
the failure of implants and leaching of ions due to
wear and corrosion. A number of research groups
have done extensive research on the improvement of
both general and pitting corrosion resistance of SS by
developing techniques for the modification of the mat-
erial’s surface and passive film. Further, pitting attack
resistance directly depends on the physicochemical
properties of the protective passive oxide layer formed
on the surface [10].
A beneficial effect of nitric acid solution on chro-
mium enrichment in the modified passive layer of SS
was reported in the literature [11-14]. Immersion in
nitric acid solutions is particularly effective in impro-
ving the pitting resistance of austenitic SS [15]. Also,
immersion in nitric acid removes sulphide inclusions,
eliminating the preferential sites for attack [16].
Mathematical modeling of the passivation pro-
cess based on the scientific principles allows one to
study and better understand this complex process.
Multiple regression analysis (MRA) and ANNs are two
important competitive data mining techniques widely
used for the development of predictive mathematical
models [17]. MRA is a conceptually simple method for
development of the functional relationships between
several independent (input) variables and one depen-
dent (output). ANNs are a computational tool, based
on the properties of biological neural systems, which
have been used successfully where conventional
computer systems have traditionally been slow and
inefficient. Both methodologies can be successfully
applied for different process modeling. However,
when compared to one another, different conclusions
can be drawn in certain cases [18].
In the literature there are few studies which are
aimed at modeling the passivation process of biomed-
ical material in nitric acid as well as in other fluids in
general. Masmoudi et al. [19] studied the passivation
process of CP Ti (commercially pure titanium) and
Ti6Al4V alloy by immersing in HNO3 solution. Their
main aim was to improve corrosion resistance of
tested materials after acid treatment. Mathematical
models were obtained by employing MRA, while opti-
mization was performed by applying the least square
method. Jiménez-Come et al. [20] presented an auto-
matic model based on artificial intelligence techniques
to predict pitting potential values. Their model was
aimed to compare pitting corrosion resistance of AISI
316L austenitic SS in different environmental condi-
tions without requiring the use of electrochemical tests.
They showed that the presented model provides an
automatic way to compare the pitting corrosion resist-
ance of austenitic stainless steel in different environ-
mental conditions. Petković et al. [13] analyzed the
possibilities for enhancing corrosion resistance of bio-
medical AISI 316LVM SS by immersing in nitric acid
solutions under different passivation conditions.
Namely, the effects of nitric acid solution concentra-
tion, temperature and passivation time on the pitting
potential, which was selected as a parameter for cor-
rosion resistance assessment, were investigated. A
total of 27 experimental trials were carried out
according to 33 full experimental design. A mathemat-
ical model was determined by using MRA, while opti-
mal values for passivation parameters were found by
means of genetic algorithm.
We decided to broaden the previous research in
order to obtain more precise and accurate experimen-
tal results by repeating the experiment twice more
and by applying ANNs for the purpose of mathemat-
ical modeling. Thus, in this paper, the pitting potential
value, as the response, was calculated as a mean of
the three pitting potential values measured for all 27
experiment trials (test). Moreover, three additional
measurements were carried out in order to determine
the pitting potential for non-passivized sample (con-
trol). Hence, there were a total of 84 experiment trials.
Compared to our previous study, the application of
the experiment designs with replications increases its
reliability significantly.
The aim of this study was to develop a mathe-
matical model relating passivation parameters with
pitting potential as the response. To evaluate the best
possible mathematical model, a statistical analysis of
the results was performed. Based on the conducted
statistical analysis, one can argue that MRA is not
able, on a satisfactory level, to accurately model the
underlying relationships between passivation para-
meters and pitting potential. For this reason, a mathe-
matical model of the passivation process was deve-
loped by using ANN in combination with a 33 full fac-
torial design with three replications. Finally, the ANN
model was compared with the MRA model to assess
the adequate methodology for further modeling of the
similar processes.
D.Lj. PETKOVIĆ et al.: THE EFFECTS OF PASSIVATION PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 121129 (2017)
123
EXPERIMENTAL
Biomedical SS
For this research, 81 test samples (3 per each of
27 experimental trials) and 3 control samples were
machined. The samples were cylindrical with the dia-
meter of 6 mm and height of 20 mm made of AISI
316LVM SS, containing Cr, Ni, Mo and Mn as main
alloy elements. Chemical composition of tested AISI
316LVM SS is in accordance to ISO 5832-1 [21].
Passivation process
Three input variables (X1: HNO3 concentration,
X2: temperature of passivation solution, and X3: passi-
vation time) were selected as passivation parameters.
The 33 full factorial design with three replications was
used. Real and coded values of the parameters and
their levels used in the experimentation are given in
previous published paper [13].
Prior to each test, the exposed surface of the
samples was wet ground with silicon carbide paper up
to 1200 grit and polished by using diamond paste with
grain size of up to 0.25 µm. Then, the samples were
rinsed with distilled water and washed with ethanol in
an ultrasonic cleaner. The passivation treatment was
performed by immersing samples in nitric acid sol-
utions. Lastly, the samples were rinsed in double dis-
tilled water and alcohol, respectively.
Electrochemical measurements
Electrochemical tests for each sample were per-
formed using a three-compartment cylindrical glass
cell equipped with a saturated calomel electrode
(SCE) as the reference electrode and a platinum foil
as the counter electrode. The average of pitting pot-
entials for three samples with the same treatment was
chosen as a measure of corrosion resistance. The
specimens were immersed 15 s before the start of the
potential rise and this time was set by the program.
The starting potential was –400 mV with a scan rate of
0.25 mV/s to anodic potential direction. The tests were
finished when the current density reached about 0.2
mA/cm2. The pitting potential (Ep) was chosen as a
measure of corrosion resistance and represented a
level of potential when the passive film broke down [22].
The electrochemical tests were conducted in
Hank’s solution, which is a simulated body fluid and
most frequently used for in vitro tests. During the
experiments, the temperature was maintained at
37±1 C (typical body temperature). The composition
and instruction for preparation of the Hank’s solution
are described elsewhere [23].
Mathematical models
MRA model
In this study, a second order polynomial was
selected for mathematical modeling of pitting potential
depending on the passivation parameters forms, as
follows:
p 0 1 1 2 2 3 3 12 1 2
2 2 213 1 3 23 2 3 11 1 22 2 33 3
E b b X b X b X b X X
b X X b X X b X b X b X (1)
where pE is the pitting potential (output), jX are
coded values of the parameters (input), 0b is the
model constant, jb is the first degree coefficient, jkb
are the cross-products coefficients and jjb are the
quadratic coefficients.
The regression coefficients, 0b , jb , jkb and
jjb , were estimated by the least squares method.
Values of regression coefficients and their statistical
significance were determined by using Minitab 15 sta-
tistical software package.
ANN model
Three neurons in the input layer (for each of the
passivation parameters), one neuron at the output
layer for pitting potential, and only one hidden layer
were used to define the ANN architecture [18,24]. The
number of hidden neurons was selected by consider-
ing the following: i) too few neurons in the hidden
layer can lead to under-fitting, i.e., inability to perform
appropriate function approximation, whereas too
many neurons can contribute to over-fitting [25],
which results in a lack of generalization capability of
the developed model; ii) the more hidden neurons,
the more expressive power of the ANN – however,
with the increase of the number of hidden neurons,
the number of unknown parameters (weights and
biases) to be estimated also increases; (ii) the upper
limit of the number of hidden neurons can be deter-
mined considering that the total number of unknown
parameters does not exceed the number of available
data for training process. As noted by Sha and
Edwards [26], although in the case where the number
of the connections to be fitted is larger than the num-
ber of available data for training, ANN can still be
trained, the case is mathematically undetermined.
Therefore, relatively small ANN architecture 3-5-1
was selected to model this passivation process.
Since it was assumed that a nonlinear relation-
ship exists between the passivation parameters and
pitting potential, the hyperbolic tangent sigmoid trans-
fer (activation) function was used in the hidden layer,
and linear transfer function was used in the output
layer. According to the selected transfer functions in
the input and output layer, all experimental data were
D.Lj. PETKOVIĆ et al.: THE EFFECTS OF PASSIVATION PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 121129 (2017)
124
normalized in the [1,1] range. The goal of the ANN
training process is to determine (near) optimal values
of weights and biases in the hidden and output layer,
previously initialized by the Nguyen-Widrow algo-
rithm, in order to minimize the mean squared error
between ANN predictions and experimental data. The
ANN was trained with gradient descent with mom-
entum by using 23 out of 27 sets of input/output expe-
rimental data and the rest was used for testing the
ANN’s generalization performance capability. Learn-
ing rate () and momentum (µ) were kept at 0.1 and
0.9, respectively. The training process was finished
after 7800 epochs, with the minimal achieved mean
squared error of 0.00548.
Statistical evaluation of developed models
Coefficient of determination R2 was used to
evaluate the performance of the developed models,
and indicate how well mathematical models fitted
experimental data [24]. In addition, for the estimation
of the prediction performance of the developed math-
ematical models, relative error, as one of the most
stringent criteria, was calculated by using the follow-
ing equation:
Relative error
Experimental value Predicted value100
Experimental value
(2)
The mean relative error (MRE) was also calcul-
ated.
RESULTS AND DISCUSSION
The results of modeling the passivation process
by using MRA and ANN are displayed and compared
in this section. In addition, the results are discussed
and analyzed.
The second order MRA model (full quadratic
regression model with interactions), relating passiv-
ation parameters and the pitting potential, was
obtained as:
p 1 2 3
2 21 2 2 3 1 2
1.50 0.086 0.0014 0.0334
0.077 0.056 0.227 0.130
E X X X
X X X X X X (3)
Based on the Table 1, it should be noted that
insignificant model terms X1X3 and X32 were elimin-
ated since they were highly correlated with other
variables.
The R2 value indicates that the passivation para-
meters explain 53.7 % of variance in pitting corrosion
potential. Apart from that, adjusted coefficient of det-
ermination and predicted coefficient of determination
(given in Table 1) are considerably smaller indicating
that the model is inadequate and over-fitted. Con-
sequently, the MRA model is not reliable enough to
describe the investigated relationship. Moreover,
analysis of variance (ANOVA), shown in Table 2, rev-
eals that F-ratio of 2.19 corresponds to confidence
level of 92.2%, which is lower than standard (95 or
99%).
Based on the conducted statistical analysis, one
can argue that MRA model is not able, on a satis-
factory level, to accurately model the underlying rel-
ationships between passivation parameters and pit-
ting potential. For these reasons, modeling of the pas-
sivation process was attempted by means of ANN.
By considering the data normalization, transfer
functions used in the hidden and output layer, and by
using the weights and biases from Table 3, one can
obtain a mathematical equation for pitting potential
calculation. After denormalization, the mathematical
model for pitting potential in terms of the passivation
parameters can be expressed by the following equation:
1 1p 2 22
20.36 1 1 0.77
1X W B
E W Be
(4)
Table 1. Regression coefficients of the MRA model; S = 0.174801; R2 = 53.7%; R2(adj.) = 29.2%; R2(pred.) = 0.0%
Coefficient Calculated coefficient value SE Coefficient T Probability density P
b0 1.5003 0.9300 16.13 0.000
b1 0.0856 0.0412 2.08 0.053
b2 0.0014 0.0414 0.03 0.973
b3 0.0334 0.0414 0.81 0.431
b12 0.0769 0.0498 1.54 0.141
b13 -0.0297 0.0498 -0.60 0.559
b23 -0.0558 0.0505 -1.11 0.284
b11 -0.2269 0.0781 -2.91 0.010
b22 -0.1300 0.0714 -1.82 0.086
b33 -0.0517 0.0714 -0.72 0.479
D.Lj. PETKOVIĆ et al.: THE EFFECTS OF PASSIVATION PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 121129 (2017)
125
where X is a column vector that contains the nor-
malized values of HNO3 concentration, temperature of
passivation solution and passivation time. Coefficient
of determination R2 for this model reveals that the
passivation parameters explain 81% of variance in
pitting corrosion potential, suggesting that the model
has a good fit. Therefore, the obtained ANN model is
better and more reliable than the MRA model. Details
related to the ANN model are given in Table 3.
Comparison of the models
In order to compare the models as well as effec-
tiveness of the passivation process, the measured
and predicted values of the pitting potential for all
experimental trials are listed in Table 4. Apart from
the results, relative errors of the models were calcul-
ated as well as standard deviation for measured data.
Moreover, the measured pitting potentials for the con-
trol sample are shown in Table 4. Maximal effect of
passivation was measured for passivation condition in
12th experimental trial.
At first, positive effect of the passivation on the
corrosion resistance is obvious according to mea-
sured pitting potential values. Standard deviation of
measured pitting potentials is about of 5% indicating
high reliability of the measured values. Then, taking
into consideration the coefficient of determination for
both models one can notice significantly higher value
for the ANN model. Additionally, MRE shows better
prediction performance of the ANN model since it pro-
duces two times less MRE than the MRA model. In
other words, ANN model is more suitable for the ana-
lysis process with a large non-linearity such as SS
passivation. Therefore, the influence of the passiv-
ation parameters on the corrosion resistance of the
SS is considered by using the ANN model only.
Effects of passivation parameters on pitting potential
The first part of the analysis is concerned with
the analysis of main effects of passivation parameters
on pitting potential. To this aim, Eq. (4) was plotted by
changing one passivation parameter at a time, while
keeping the other two constant at the center level
(Figure 1).
It is evident that the mathematical relationships,
presented graphically in Figure 1, are highly non-
linear. Quantitatively, based on the analysis of the
main effects, concentration is the most influential
parameter, followed by temperature and passivation
time as less influential, respectively. While passivat-
ion temperature and time are on central level, the
highest corrosion resistance is achieved when the
concentration is about 20%. For central levels of con-
centration and passivation time the highest corrosion
resistance is achieved when the temperature is
slightly lower than 30 C. Finally, when the concen-
tration and temperature are set on the central level,
the highest corrosion resistance is achieved when the
duration of the process is about 25 min.
In order to determine the interaction effects of
the passivation parameters on the pitting potential,
3-D surface plots were generated considering two
parameters at a time, while the third one was kept
constant at the center level (Figure 2).
From Figure 2 it can be observed that the pitting
potential is highly sensitive to the selected passiv-
ation parameters. It is also obvious that the effects of
the parameter are variable depending on their own
level, since there are significant interaction effects of
passivation parameters on the pitting potential. The
functional dependence between the pitting potential
and the passivation parameters is strongly nonlinear,
therefore the effect of a given parameter on the pitting
potential must be considered through the interaction
with the other parameters.
For instance, if the passivation time is set on the
central level (40 min), Figure 2a, and temperature on
the low level, an increase in concentration of the sol-
ution leads firstly to the pitting potential increase up to
some extreme value, which corresponds with the
middle level of the concentration. Further increase in
concentration leads to the reduction of the pitting pot-
ential. In this case, the pitting potential is very low
when the concentration is on the high level, while the
middle one provides a fairly high pitting potential.
When the temperature is set on the high level, with an
increase in concentration from the low level the pitting
potential firstly decreases, then increases, and the
nearby high level starts to impair.
When the temperature is set on the middle level
(Figure 2b) and concentration is on the low level, an
Таble 2. ANOVA results for the MRA model;DF - degree of freedom; SS - sum of squares; MS - mean square; F - value of Fisher’s distri-
bution; P - probability density
Source DF SS MS F P
Regression 9 0.60322 0.06702 2.19 0.078
Residual Error 17 0.51944 0.03056
Total 26 1.12267
D.Lj. PETKOVIĆ et al.: THE EFFECTS OF PASSIVATION PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 121129 (2017)
126
Table 3. The weights and biases of the developed ANN mode; W1: weights between input and hidden layer; W2: weights between hidden
and output layer; B1: biases of the hidden neurons; B2: bias of the output neuron
W1 W2 B1 B2
-0.26021 -2.5744 -0.78867 0.36462 2.1845 -1.2035
-1.2355 -2.2487 -1.317 -1.1196 1.3122
1.4764 1.5512 0.95885 -1.1753 -0.27368
-1.9939 0.97443 0.30408 -1.0401 -2.539
-1 -1.6398 -1.4029 -1.0435 -2.4896
Table 4. Comparative review of the measured pitting potential and predicted values for the pitting potential by means of MRA and ANN
models; important remark: shaded rows - testing data for ANN model performance
Exp. trial Passivation parameters Experimental MRA Model ANN Model
HNO3 concentration
%
Temperature
C
Passivation time
min
Ep
V
Standard deviation
%
Ep
V
Relative error
%
Ep
V
Relative error
%
Control - - - 0.68 4.04 - - - -
1 10 17 20 0.85 4.51 1.08 26.89 0.86 1.15
2 10 17 40 1.16 5.13 1.13 2.21 1.13 2.49
3 10 17 60 1.19 6.11 1.19 0.01 1.47 23.85
4 10 40 20 1.12 4.16 1.19 6.03 1.14 1.37
5 10 40 40 1.44 5.51 1.19 17.53 1.35 6.14
6 10 40 60 1.18 4.58 1.19 0.64 1.22 3.38
7 10 60 20 1.08 3.06 1.04 4.03 1.05 2.49
8 10 60 40 0.77 2.52 0.98 27.35 0.87 12.79
9 10 60 60 0.81 5.20 0.92 14.17 0.78 3.48
10 30 17 20 1.0 4.93 1.29 29.49 1.03 3.00
11 30 17 40 1.39 4.04 1.35 2.83 1.46 5.08
12 30 17 60 1.49 5.03 1.41 5.60 1.47 1.32
13 30 40 20 1.41 3.21 1.46 3.53 1.46 3.27
14 30 40 40 1.47 5.14 1.46 0.70 1.33 9.56
15 30 40 60 1.39 3.04 1.46 5.02 1.30 6.16
16 30 60 20 1.16 5.77 1.36 17.63 1.20 3.37
17 30 60 40 1.41 2.52 1.31 7.18 1.30 7.79
18 30 60 60 1.33 4.62 1.25 5.79 1.26 5.02
19 65 17 20 1.28 4.20 1.10 14.37 1.32 3.35
20 65 17 40 1.29 5.00 1.15 10.71 1.31 1.26
21 65 17 60 0.94 3.55 1.21 28.47 1.04 10.44
22 65 40 20 1.13 2.89 1.36 20.24 1.11 1.94
23 65 40 40 1.14 5.77 1.36 19.18 1.19 4.20
24 65 40 60 1.43 4.93 1.36 4.99 1.48 3.64
25 65 60 20 1.42 3.79 1.36 4.13 1.34 5.40
26 65 60 40 1.17 3.21 1.31 11.59 1.34 14.89
27 65 60 60 1.34 4.15 1.25 6.73 1.31 2.38
Mean relative error (MRE) 11.00 5.53
increase in passivation time leads firstly to the pitting
potential increase up to some extreme value and then
decrease. In this case, the pitting potential is very low
when the passivation time is on the low and high
level, while the middle one provides pretty high pitting
potential. When the concentration is set on the high
level, with increasing passivation time from the low
level the pitting potential firstly increases rashly, then
slightly decreases up to about 30 min, and then starts
to grow again up to the high level.
Based on Figure 2a and b, it can be concluded
that the highest pitting potentials correspond with a
combination of parameters concentration-temperature
and concentration-passivation time slightly below the
D.Lj. PETKOVIĆ et al.: THE EFFECTS OF PASSIVATION PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 121129 (2017)
127
Figure 1. Main effects of passivation parameters on pitting potential.
Figure 2. Interaction effects of passivation parameters on pitting potential.
D.Lj. PETKOVIĆ et al.: THE EFFECTS OF PASSIVATION PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 121129 (2017)
128
middle levels. As it can be seen in Figure 2c (where
concentration is on the middle level), the lowest pit-
ting potential predicted for temperature of 17 C and
passivation time 20 min; the highest pitting potentials
is predicted for the temperature and passivation time
slightly below the middle level.
CONCLUSION
Passivation by immersing in nitric acid solution
is an effective method to improve corrosion resistance
of biomedical SS. In biomedical SS passivation pro-
cess, MRA and ANN were introduced to model func-
tional relationship between pitting potential and pas-
sivation parameters such as nitric acid concentration,
temperature and passivation time. A non-linear func-
tional dependence between the passivation para-
meters and the pitting potential was determined.
Hence, the ANN model proved to be more suitable for
modeling the processes such as chemical passivation
of SS. Nitric acid concentration has maximum inf-
luence on the pitting potential followed by the tempe-
rature and passivation time. The best experimental
result was achieved by a combination of parameters:
HNO3 concentration – 30%; temperature – 17 C; pas-
sivation time – 60 min.
Acknowledgement
This paper is a result of the projects ON174004
and TR35034 supported by the Ministry of Education,
Science and Technological Development of the Rep-
ublic of Serbia.
REFERENCES
[1] Q. Chen, G.A. Thouas, Mat. Sci. Eng. Rep. 87 (2015) 1–57
[2] T. Hanawa, Mat. Sci. Eng., C 24 (2004) 745–752
[3] S.H. Giertz, B.L. Fernandes, C.R. Fernandes, C.L.C.
Franco, Mater. Res. 13 (2010) 35-39
[4] J.W. Schultze, M.M. Lohrengel, Electrochim. Acta 45
(2000) 2499–2513
[5] C.C. Shih, C.M. Shih, Y.Y. Su, L.H. Julie Su, M.S. Chang,
S.J. Lin, Corros. Sci. 46 (2004) 427-441
[6] S. Virtanen, I. Milošev, E. Gomez-Barrena, R.Trebše, J.
Salo, Y.T. Konttinen, Acta Biomater. 4 (2008) 468–476
[7] U.K. Mudali, T.M. Sridhar, B. Raj, Sãdhanã 28 (2003)
601–637
[8] R.M. Pidaparti, L. Fang, M.J. Palakal, Comp. Mater. Sci.
41 (2008) 255–265
[9] G. Manivasagam, D. Dhinasekaran, A. Rajamanickam,
Recent Pat. Corros. Sci. 2 (2010) 40-54
[10] A. Shahryari, S. Omanovic, Electrochem. Commun. 9
(2007) 76–82
[11] J.S. Noh, N.J. Laycock, W. Gao, D.B. Wells, Corros. Sci.
42 (2000) 2069-2084
[12] T. Hong, M. Nagumo, W.P. Jepson, Corros. Sci. 42
(2000) 289-298.
[13] D. Petković, G. Radenković, V. Blagojević, P. Živković, I.
Ćirić, Ann. Fac. Eng. Hunedoara 12 (2014) 151-154
[14] D. Petković, G. Radenković, in Proceedings of the 2nd
International Conference Mechanical Engineering in XXI
Century, Niš, Serbia, 2013, pp. 87-90
[15] M.A. Barbosa, A. Garrido, A. Campilho, I. Sutherland,
Corros. Sci. 32 (1991) 179-184
[16] K. Murugan, A. Ragupathy, V. Balasubramanian, K.
Sridhar, Surf. Coat. Tech. 247 (2014) 90–102
[17] J. Ding, H. Wang,K. Dai, Y. Zi, Z. Shi, Chem. Ind. Chem.
Eng. Q. 20 (2014) 29-38
[18] M. Madić, M. Radovanović, Int. J. Phys. Sci. 7 (2012)
2422 – 2430
[19] M. Masmoudi, D. Capek, R. Abdelhedi, F. El Halouani, M.
Wery, Surf. Coat. Tech. 200 (2006) 6651–6658
[20] M.J. Jiménez-Come, I.J. Turias, F.J. Trujillo, Electro-
chem. Commun. 35 (2013) 30–33
[21] ISO 5832-1: Implants for surgery - Metallic materials -
Part 1: Wrought stainless steel
[22] J. Xie, A.T. Alpas, D.O. Northwood, Mater. Charact. 48
(2002) 271–277
[23] M. Esfahanian, M Nikzad, G Najafpour, A.A. Ghoreyshi,
Chem. Ind. Chem. Eng. Q.19 (2013) 241−252
[24] K. Benyelloul, H. Aourag, Comp. Mater. Sci. 67 (2013)
353–358
[25] S.R. Karnik, V.N. Gaitonde, J. Campos Rubio, A. Esteves
Correia, A.M. Abrão, J. Paulo Davim, Mater. Design 29
(2008)1768–1776
[26] W. Sha, K.L. Edwards, Mater. Design 28 (2007) 1747–
–1752.
D.Lj. PETKOVIĆ et al.: THE EFFECTS OF PASSIVATION PARAMETERS… Chem. Ind. Chem. Eng. Q. 23 (1) 121129 (2017)
129
DUŠAN LJ. PETKOVIĆ
MILOŠ J. MADIĆ
GORAN M. RADENKOVIĆ
Univerzitet u Nišu, Mašinski fakultet,
Katedra za proizvodno-informacione
tehnologije, A. Medvedeva 14, 18000
Niš, Srbija
NAUČNI RAD
UTICAJ PARAMETARA PASIVIZACIJE NA PITING POTENCIJAL BIOMEDICINSKOG NERĐAJUĆEG ČELIKA
Pasivizacija predstavlja hemijski process u kome se površina metalnih legura dovodi u
stanje elektrohemijske pasivnosti. Biomedicinski nerđajući čelik AISI 316LVM može se
pasivizirati potapanjem u azotnu kiselinu, jer se time poboljšava zaštitini oksidni sloj;na taj
način se povećava i koroziona postojanost ovog materijala u fiziološkim rastvorima. U
ovom istraživanju, za matematičko modeliranje procesa pasivizacije korišćene su više-
struka regresiona analiza i veštačke neuronske mreže. Za izlazni (zavisni) parametar
modela izabran je piting potencijal, koji predstavlja meru korozione postojanosti. Kao para-
metri pasivizacije razmatrani su: koncentracija azotne kiseline, temperatura i vreme pasi-
vizacije. Upoređeni su eksperimentalni rezultati i rezultati modela. Pokazalo se, da jedino
model dobijen pomoću veštačkih neuronskih mreža ima statistički zadovoljavajuću tačnost
predikcije. Na kraju, na osnovu modela dobijenog pomoću veštačkih neuronskih mreža,
izvedena je analiza uticaja parametara pasivizacije na piting potencijal biomedicinskog
nerđajućeg čelika.
Ključne reči: nerđajući čelik, azotna kiselina, pasivizacija, višestruka regresiona
analiza, veštačke neuronske mreže.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 131139 (2017) CI&CEQ
131
MARIJA KODRIC
SANDRA STOJANOVIC
BRANKA MARKOVIC
DRAGAN DJORDJEVIC
University of Niš, Faculty of
Technology, Leskovac, Serbia
SCIENTIFIC PAPER
UDC 677.494.674.027.4:66.081.3:544
https://doi.org/10.2298/CICEQ151113022K
MODELLING OF POLYESTER FABRIC DYEING IN THE PRESENCE OF ULTRASONIC WAVES
Article Highlights
• Non-standard dyeing of PES fabric without carrier is possible by ultrasound
• Continual growth of the amount of exhaustion dye with mass of material is noted
• The Langmuir model shows precise description of experimental data
• Kinetic model of pseudo second-order adequately describes disperse dye – PES fab-
ric system
Abstract
In this paper, modelling of dyeing, i.e., adsorptive behaviour of disperse dyes on
polyester fibres under the influence of ultrasound has been considered with the
aim of getting data about binding mechanisms, as well as defining the conditions
of dyeing with additional energy source without the use of carriers, compounds
that increase permeability of the fibres and help dyeing. Dyeing adsorption was
conducted under different conditions, and the concentration of dyes, mass of the
substrate, recipes and time of dyeing were varied. It was established that ultra-
sound allows dyeing without carriers, and that the efficiency of dyeing depends
on the time of contact, initial concentration of the dye and the amount of abs-
orbent material. Continual growth of the amount of bound dye with the mass of
the absorbent was observed. Characteristic plots obtained from confirmed that
the Langmuir isotherm model ensures a precise description of polyester dyeing
by disperse dye. The dyeing kinetics was remarkably well described by pseudo
second-order in regards to the high functionality.
Keywords: adsorption, polyester, disperse dye, ultrasound, Langmuir isotherm, kinetics.
It is known that polyester (PES) belongs to a
group of synthetic fibres that possess active spots
where dye molecules can be adsorbed. PES contains
a great number of ester groups, as well as a certain
number of carboxylic groups, which are placed at the
ends of chains, such that during the dyeing of this
fibre hydrogen bonds with dye molecules will be est-
ablished. PES has significant hydrophobic character
and compact structure, so taking into consideration
this kind of fibres behaviour in the dyeing solution, it is
necessary to modify the usual dyeing process, i.e. to
increase the rates of dye diffusion in the fibres. Gen-
erally, the rates of dyeing could be increased by using
Correspondence: D. Djordjevic, University of Niš, Faculty of
Technology, Bulevar oslobodjenja 124, Leskovac, Serbia. E-mail: [email protected] Paper received: 13 November, 2015 Paper revised: 12 April, 2016 Paper accepted: 13 April, 2016
suitable dyes, changing the fibre structure, changing
the conditions of dyeing, etc. 1,2.
The rate of dye diffusion can be intensified by
increasing the permeability of the fibres, i.e., inc-
reasing the fibre swelling ability. This can be achieved
by adding simple organic compounds (with smaller
molecules than the dye molecules and with certain
affinity towards the fibre) in the dyeing solution. These
compounds are termed as carriers: due to their small
molecular size they quickly diffuse into the fibre, bind
to carboxylic groups and establish hydrogen bridges.
Carriers cannot be completely removed from the fib-
res by washing; therefore, they require special atten-
tion due to their toxic and dermatologic effects. On
the other hand, the presence of these substances on
the fibres adversely affects the fastness of many dyes
with light, and in certain cases they influence fibre
shrinking as well [3,4].
M. KODRIC et al.: MODELLING OF POLYESTER FABRIC DYEING… Chem. Ind. Chem. Eng. Q. 23 (1) 131139 (2017)
132
Ultrasound in the 20-100 kHz frequency range is
used to increase the chemical reaction rate and adv-
ance different physical processes, such as cleaning,
emulsion, extraction, etc. It can achieve the same or
even better results in comparison to already existing
techniques under less extreme conditions (e.g., lower
temperatures and high chemical concentrations of
reactants). Hence, the process of material dyeing
using ultrasound is very significant. Observed imp-
rovements in ultrasound processes of dyeing gen-
erally refer to the phenomenon of cavitation, but some
other mechanical influences can occur too, such as
dispersion and diffusion [5-7].
This paper strives to explain the ability of dye
adsorption into the fibres in aqueous environment
with usual auxiliaries according to standard recipes in
the presence of ultrasound waves, instead of carriers.
The purpose is to successfully perform the process of
dyeing of very hydrophobic and crystal fibres under
normal pressure and temperature conditions. Conse-
quently, if the dye exhaustion is higher, the amount of
dyed wastewater is lower and less harmful to the
environment. The inclination was to eliminate the
usual supplement – carriers, since it is known that
most of these supplements have negative impact on
human health and environment. Furthermore, the
goal was to explain the polyester dyeing process in
new circumstances using a series of experiments, as
well as modelling the system and kinetic parameters.
EXPERIMENTAL
As an adsorbent, 100% polyester fabric (poly-
ethylene terephthalate) has been used with the fol-
lowing characteristics: weave – twill ½S, warp and
weft fineness of 162 tex each, warp and weft density
of 30 and 22 cm-1 and surface mass of 195 g/m2. The
structure of the used disperse dye (adsorbate), C.I.
Disperse Blue 79 (DB79), is shown in Figure 1. It is a
large molecule, highly energetic dye with good fast-
ness in sublimation and wet treatment. It is suitable
for dyeing by exhausting and thermosol procedure, as
well as printing synthetic material.
Figure 1. Structure of the used disperse dye,
C.I. Disperse Blue 79.
The dyeing process (the method of exhaustion)
was performed with the dye bath heated at 60-70 C
with 1.5 g/dm3 of carrier (Icelan PSN, anionic, on the
basis of mixture of different aromatic esters and spe-
cial emulsifiers, Textilcolor AG, Switzerland) and 1
g/dm3 of dispersing agent (TC-Dispergator DTS, non-
-ionic, on the basis of polyglycolether-derivates, Tex-
tilcolor AG, Switzerland). The dye bath was adjusted
to pH 5 by acetic acid, and, after mixing, the disper-
sant substance was added, with continual heating to
95 C. The PES fabric sample was then added and
dyed for 60 min. After dyeing completion, the dyed
material was washed at 70-80 C.
Non-standard dyeing was performed under iden-
tical conditions, but in the absence of carrier. Instead,
an Elac Ultrasonic Laboratory Reactor URS 1000 was
used. The frequency of the applied ultrasound oscil-
lations was 140 kHz, while the power was 50 W.
The amount of PES fabric in the dyeing adsorp-
tion test was varied from 2 to 6 g, and the used dye
solution (constant volume of 0.1 dm3) were prepared
at concentrations of 50, 100, 200, 300 and 400
mg/dm3 in distilled water. The time of treatment, with
constant mixing, was 10, 20, 30, 40, 50 and 60 min.
The time of 60 min was taken as equilibrium no sig-
nificant changes in dyeing adsorption were observed
with further treatment.
For determining the concentration of dye in the
solution, a Cary 100 Conc UV-Vis Varian spectro-
photometer was used (absorption maximum at 550
nm).
The level of dye exhaustion (%) is calculated as
[8]:
0
0
100c c
c (1)
where c0 and c, mg/dm3, are the initial and final con-
centration of the solution of dye.
Langmuir isotherm was obtained from the fol-
lowing equation [9]:
ee
e max max
1 1cc
q Q b Q (2)
where ce, g/dm3, is the equilibrium concentration of
dye after finished adsorption; qe, g/kg, is the adsorbed
amount of adsorbate (dye) per mass unit of the
adsorbent (fabric); Qmax, g/kg, is the maximum
amount of adsorbate which can be bound on the
adsorbent; b, dm3/g, is the Langmuir constant.
The amount of adsorbed dye per mass unit of
fabric is calculated as [9]:
M. KODRIC et al.: MODELLING OF POLYESTER FABRIC DYEING… Chem. Ind. Chem. Eng. Q. 23 (1) 131139 (2017)
133
0 e
e
c c Vq
w (3)
where w, kg, is the mass of fabric and V, dm3, is the
volume of the solution from which the adsorption was
done.
Lagergren’s pseudo first-order equation is usu-
ally expressed in the linear form [10]:
1e elog log
2.303t
kq q q t (4)
where qe and qt, mg/g, are the capacities of the ads-
orption in equilibrium and after time t, respectively,
and k1, 1/min, is the rate constant of pseudo first-
-order adsorption.
The kinetic equation of the pseudo second-order
adsorption has the form [10]:
2
e2 e
1 1
t
tt
q qk q (5)
where k2, g/mgmin, is the rate constant of the sec-
ond-order adsorption.
RESULTS AND DISCUSSION
The chosen dye DB79 is not water-soluble. The
solubility depends on the chemical composition and
especially on the content of polar and non-polar func-
tional groups. As a typical donor-acceptor chromo-
gen, this dye has two nitro groups and one bromine
that pull on electrons, and, to a certain extent, elec-
tron-donor ethoxy group, whereas acetylamino and
diacetyloxyetilamino groups practically are not suit-
able donors and acceptors of electrons. In consider-
ation of interaction of the mentioned functional groups,
substantivity will occur towards the textile, which is in
this case a hydrophobic material.
The inclusion of dispersing agent in the dye bath
is a crucial factor in the application of disperse dyes.
The hydrophobic tails of the dispersing agent mole-
cules are inside a micelle which, as a consequence,
is able to solubilise the disperse dye molecules, thus
conferring a higher apparent solubility to the dye [11].
The dye transfers to the fibre from the micelles.
As micelles release their dye, they reform and dis-
solve more dye from the solid particles. Much of the
evidence that is available on the subject suggests that
in dyed polyester fibres the disperse dyes are present
chiefly in the monomolecular state. At the end of the
dyeing process, the dye that has been absorbed by
the fibre is in a state of dynamic equilibrium with the
dye that remains in the bath, and the fraction of the
latter that is in aqueous solution must be present in the
same state of aggregation as the dye in the fibre [11].
The influence of time or the length of contact
between adsorbate and adsorbent on adsorption –
dye exhaustion during ultrasound (without carrier), for
different initial concentration of disperse dye, is shown
by comparative plots in Figure 2. The continuity in
changes during time is present, i.e., the longer time
carries greater amount of adsorbed dye per mass unit
of adsorbent. At lower dye concentrations there is
greater dye exhaustion in contrast to higher dye con-
centrations. Realistically, however, higher amount of
dye adsorbed occurs at higher initial dye concentra-
tions.
In other words, for example, dye exhaustion is
72.48% for the concentration of 50 mg/dm3 and 10
min of adsorption, while dye exhaustion is 57.03% for
the concentration of 400 mg/dm3 and 10 min of ads-
orption.
Therefore, there is a greater exhaustion at low
concentration, but after calculating, real dye adsorp-
tion at a lower concentration is: 72.48%50 mg/dm3 =
Figure 2. The influence of adsorption time on the percentage of DB79 exhaustion (A - PES, 2 g; B - PES, 6 g).
M. KODRIC et al.: MODELLING OF POLYESTER FABRIC DYEING… Chem. Ind. Chem. Eng. Q. 23 (1) 131139 (2017)
134
= 36.24 mg of dye adsorbed on PES fabric; while at
higher concentration the amount is: 57.03%400
mg/dm3 = 228.12 mg of dye adsorbed on PES fabric;
Linear parts of the curve reflect diffusion in the
surface layer, whereas the parts of the plateau on the
curve respond to diffusion in pores. The diagram for
adsorption on 4 g of PES is not shown, because there
are no great differences in the appearance of the
curves.
A similar investigation by Olenka et al. used the
same dye (Disperse Blue 79) for PES dyeing [12].
They implemented the chemical modification of the
substrate, instead of using ultrasound, and PES dye-
ing was performed without carriers. The results of dye
exhaustion showed that the best dyeing conditions
were to treat the PES for 15 min at 85 C with N,N-
-dimethylacrylamide as a modifier, followed by a dye-
ing time of 30 min at 85 C. These conditions were
shown to be suitable for the dye. Photoacoustic spec-
troscopy allows the determination of the penetration
conditions at which the modified PES can absorb
more dye of original.
Also, disperse dye exhaustion was analyzed by
Choi and Kang [13]. They prepared six nano-disperse
dyes using corresponding O/W nanoemulsions in
order to dye PES (two type, regular- and micro-fib-
ers). Dye exhaustion using nano dyes resulted in low
exhaustion yields of 17-26% on regular polyester fiber
and 28-38% on ultramicrofiber polyester. The obs-
erved low exhaustion yields of nano-disperse dye can
be explained by the solubilization of dye particles into
surfactant micelles as well as the high stability of the
nanoemulsions, which might reduce the capacity of
dye uptake by the fibers. As commercial disperse
dyes exhibit exhaustion of 90-95%, these results were
extremely lower than conventional disperse dyes.
However, in the case of dyeing with nano-dyes pre-
pared on ultramicrofibers, it was observed that micro-
fiber site exhibited higher K/S values than those of
regular polyester site, in the range of 1.5-2.4%, which
was promising a possibility for higher efficiency on
ultramicrofibers.
Likewise, the potential of ultrasound as a means
of enhancing dyeing efficiency was evaluated by Lee
and Kim [14]. Changes in the particle sizes of a dis-
perse dye dispersion with ultrasound irradiation are
studied using the turbidity concept, and the effect of
particle size on the exhaustion rate is also inves-
tigated. Ultrasound irradiation of a dye dispersion red-
uces the particle sizes of disperse dyes, and the
exhaustion rate of the dyes on fibers is enhanced by
this reduced particle size by the ultrasound pretreat-
ment before dyeing experiments. These results sug-
gest that ultrasound is useful method of accelerating
the dyeing rate and increasing dyeing efficiency.
During the dyeing of Disperse Blue 56 with and
without ultrasound, as might be expected, increasing
the dyeing temperature increased the dye uptake and
dyeing rate regardless of ultrasound use [15]. Com-
paring the dyeing kinetics of Disperse Blue 56 with
and without ultrasound, authors did not see a signific-
ant increase in dye uptake and dyeing rate for Dis-
perse Blue 56 with ultrasound use. The effect of ultra-
sound on the dyeing behavior of Disperse Blue 56 on
PET fibers was almost non-existent, with little inf-
luence on dye absorption and dyeing rate. PET dye-
ing of Disperse Red 60 with ultrasound irradiation
showed a considerable increase not only in the dye-
ing rate but also in dye uptake. Comparing dyeing
with and without ultrasound, dye uptake was higher
by some 68 and 6%, respectively, for Disperse Red
60 when dyeing was conducted with ultrasound.
Compared with the results for Disperse Blue 56, this
demonstrates that ultrasonic energy can have a pro-
found effect on the dye uptake and dyeing rate of
more crystalline dyes like Disperse Red 60.
A comparison of dyeing with carrier with or
without ultrasound, and dyeing with ultrasound only,
are shown in Figure 3 with respect to their depen-
dency on exhaustion dye–time of dyeing. Only data for
dye concentration of 400 mg/dm3 and minimal and
maximal amount of PES fabric is shown, since very
similar behaviour is observed under other concentra-
tions and 4 g of remaining material mass. The shape
of the curves is very much alike for the given concen-
tration and 6 g of PES, whereas for minimal amount
of PES (2 g), the curves partly differ.
Figure 3. The influence of adsorption-dyeing time on DB79
exhaustion (400 mg/dm3) in different conditions.
M. KODRIC et al.: MODELLING OF POLYESTER FABRIC DYEING… Chem. Ind. Chem. Eng. Q. 23 (1) 131139 (2017)
135
The fact that stands out is that dyeing with ultra-
sound leads to more exhaustion and binding dye for
PES fibres in comparison to dyeing with carrier with-
out ultrasound. It is interesting that the presence of
ultrasound in the dye bath with all auxiliaries (includ-
ing carrier) gives better results, i.e. the biggest DB79
exhaustion on the fibre. This kind of behaviour prac-
tically is mapped at all dye concentrations and all
amounts of adsorbents. As the amount of adsorbent
is increased, dyeing with ultrasound without carrier
performs much better in comparison to dyeing with
carrier. Nonstandard dyeing with ultrasound, but with
carrier, gives the best results of dyeing in all cases. It
is associated with the fact that the present carrier,
having achieved the ability of swelling, opens the
structure of fibres and causes greater dye exhaustion
that is helped very much by the present ultrasound.
This proves the positive effect of ultrasound
waves on dyeing with and without carrier, which is
explained by dispersion (separation of big dye par-
ticles into smaller ones), then by degassing (ejection
of soluble or trapped air from the capillaries of the
fibres), expansion (speeding up the dye molecules to
penetrate into PES fibres) and intensive mixing of the
dye solution [16].
Aside from the above mentioned, greater dye
exhaustion can be observed due to the change in
PES fibre crystallinity of PES fibres changes during
the ultrasound processing, because of removal of oli-
gomer from the surface and surface layers, and the
cavitation of ultrasound dislocating macromolecules
among micro crystallites, thus enlarging the amor-
phous area [5].
In other words, in order to diffuse the molecules
of dye properly into the fibres, the free volume must
be formed inside the substrate (fibre). The applied
ultrasound helps the free volume to be created easily
inside the polymer via thermal moving of the chains of
molecules and dye molecules, which enter this free
area. At the same temperature, thermal moving of
molecule chains is directly related to the strength of
polymer substrate, i.e., the faster dye diffusion can be
achieved in softer and more flexible substrates of
polymers, which is partially enabled by using ultra-
sound waves [5].
Similarly, Saligram et al. found that sonication
during dyeing brought much more [17]. Namely, ultra-
sound used in the dyeing of polyester fibres with
disperse dyes at low temperature, and results com-
pared with those achievable in conventional dyeing at
the boil using a carrier. Dyeing was enhanced in pre-
sence of carrier and by pre-swelling the fibres,
although the results obtained were not generally as
good as those that can be obtained in conventional
high-temperature processes. Taking into account the
energy conservation aspect, ultrasound appears to be
a promising technique for dyeing.
Also, poly(trimethylene terephthalate) (PTT) fab-
ric, a new type of polyester fibre, was dyed with Dis-
perse Red FB by using ultrasonic power by Wang et
al. [18]. It was shown that the ultrasonic-assisted
dyeing could increase the depth of shade in PTT fab-
ric at a lower temperature. The ultrasonic energy can
disintegrate the large particles of oligomer on the sur-
face of the PTT fibre to smaller ones and slightly dec-
reases the crystallinity of the PTT fibre, which can
reduce the particle size of the disperse dye in the dye
solution as well. Moreover, the ultrasonic dyeing of
the PTT fabric with a swelling agent can enhance the
colour strength of the dyed fabric, thereby reducing
the dyeing time as well as saving energy. The effects
of ultrasound on the K/S values of the fabric, the fibre
structure and the disperse dye were investigated. The
results show that the ultrasonic power increases the
K/S values of the fabric, disintegrates the large par-
ticles of oligomer on the surface of poly(trimethylene
terephthalate) fibre into smaller ones and reduces the
particle size of the dye in solution. The K/S value of
poly(trimethylene terephthalate) fabric dyed using
ultrasound is much higher than that without ultra-
sound, especially at temperatures higher than 60 C.
In a similar manner, an attempt was made to
evaluate the possibility of using ultrasonic techniques
for effective low-temperature dyeing of polyester [19].
The ultrasonic dyeing depends strongly on a preswel-
ling process that would be both expensive and difficult
to carry out commercially (particularly in terms of
health and safety aspects). Performance also dep-
ends on the energy levels of the dyes used. The res-
ults obtained for the dyes of higher relative molecular
mass were much worse than those obtained when
using carrier at the boil (although raising the dyeing
temperature would be expected to provide notable
improvements in depth). Little advantage would be
gained over conventional dyeing methods, particularly
when carrier was incorporated in the dyebath. An
ultrasound dyeing unit with a lower frequency level
(around 26 kHz), to generate more pronounced cavit-
ation effects, may have given better results. However,
the unit chosen had a frequency rating that was both
readily available commercially (38 kHz) and could still
be considered operationally viable. The noise levels
associated with lower frequency units would be unac-
ceptable in commercial use.
Since it is being established that the dye
exhaustion on the polyester is sufficiently good with
M. KODRIC et al.: MODELLING OF POLYESTER FABRIC DYEING… Chem. Ind. Chem. Eng. Q. 23 (1) 131139 (2017)
136
ultrasound without carrier, the results of modelling of
dyeing – adsorption without carrier but under the inf-
luence of ultrasound are given as follows. The results
of changing the adsorbed amount of adsorbate on the
adsorbent during time, for different initial concentra-
tions of dye during dyeing with ultrasound (without
carrier) in relation to the mass of PES fabric, are
shown in Figure 4. The continuity in changes with
time can be observed, i.e., the longer the time, the
greater the amount of the adsorbed dye per mass unit
of adsorbent. The highest adsorption occurred at the
highest applied concentrations of dye.
Since the total surface area of fibres is higher
than the outside surface, the molecules of dye will
faster adsorb during dyeing than the present auxil-
iaries. Since the dynamic equilibrium of the solution
deranges because of that, the aggregates of dye will
dissolve into molecules and establish equilibrium
again in the solution. Adsorption will continue up to
the point when the equilibrium between dye concen-
tration in the solution and dye concentration in the
fibre is not reached. Since the molecules of dye have
the tendency to form aggregates in aqueous solution,
ultrasound energy causes degradation of dye aggre-
gates in the solution, decreasing the size of dye par-
ticles in dispersion, which is the first pre-condition for
better adsorption on the adsorbent [5-7].
Figure 5 represents the interpretation of Lang-
muir adsorptive isotherm for different amounts of
adsorbents, showing the dependency of parameters
(ce/qe) on the equilibrium dye concentration (ce) during
dyeing with ultrasound (without carriers). From the
slope and intercept of the fits, the values of Langmuir
constants (Qmax and b) were obtained, which are rel-
ated to the maximum amount of dye that can bind on
the fibre, and the free energy of adsorption, respect-
ively.
Figure 5. Langmuir adsorption isotherms for DB79 – PES
fabric system.
Figure 5 shows that adsorption curves are flat
and continual, which leads to saturation during dif-
ferent concentrations on the outer interphase of ads-
orbent material. Moderate dispersing of data is pre-
sent which indicates the adequacy of Langmuir iso-
therm for describing the adsorption equilibrium of the
examined systems. This kind of behaviour could be
explained by the assumption that the dye in the
beginning adsorbs on the outer surface of PES fabric,
and after reaching a certain level of saturation, it
enters the inner space of PES fibres, when it becomes
adsorbed by inner surfaces. After diffusing of dye into
pores of the fibres, diffused resistance increases,
which leads to a decrease in diffusion rate. By dec-
reasing the concentration of dye in the solution, the
diffusion rate becomes constantly lower so the dif-
fusion process reaches an equilibrium state, which
completely submits to the law of equilibrium adsorpt-
ion defined by Langmuir isotherm.
Figure 4. Adsorbed amount of dye DB79 on PES fabric during time of ultrasonic treatment (A - PES, 2 g; B - PES, 6 g).
M. KODRIC et al.: MODELLING OF POLYESTER FABRIC DYEING… Chem. Ind. Chem. Eng. Q. 23 (1) 131139 (2017)
137
Table 1 shows the analytical expressions of
Langmuir isotherm, values of Langmuir parameters
Qmax and b, and values of coefficient of determination
R2. The coefficient of determination is a relative mea-
sure of representability of regression line or measures
of usage of Langmuir model.
According to the results for coefficient determin-
ation from Table 1, high values (above 0.97) are
noted, which means that great percentage of square
sums of variables values that deviate from arithmetic
mean is explained by regression model. Therefore, in
the procedure of determining capacities of dye retain-
ing and affinities of adsorbent to dye, Qmax and b,
respectively, the Langmuir monolayer model fits the
experimental data reasonably well and it can be
acceptable for adsorption of disperse dye on PES
fabric.
The Langmuir model parameters are strongly
dependent on the amount of adsorbent: Qmax dec-
reases with the increasing of PES amount, whereas
the values of other constant, b, increase continually
with increasing the mass of the material. Higher
values of b obtained for PES – disperse dye system
mean stronger binding of dye.
In the study of Wang et al., the Nernst, Langmuir
and Freundlich isotherm models were employed in
the fitting of the experimental points using the soft-
ware Origin [18]. It can be seen that the coefficients of
determination (R2) of the Langmuir isotherm are the
highest in the three models, indicating that the Lang-
muir model is the most effective in simulating the ads-
orption isotherm of Disperse Red FB onto the PTT
fibre. In addition, the adsorption capacities (Qmax) of
the PTT fibre dyed in the presence of ultrasound are
all higher than that of the PTT fibre dyed without ultra-
sound at three temperatures. This proves the ultra-
sound-assisting effect on the dyeability of PTT fibre
with disperse dye.
Figure 6 shows diagrams with results related to
kinetic sorption of DB79 on PES fabric for the applied
mass of 2 g of adsorbent and different initial dye con-
centration during dyeing with ultrasound (without car-
rier). According to linear forms of pseudo first-order
model (Figure 6A), it can be concluded that the rates
of adsorption in given experimental conditions, does
not describe properly pseudo first-order model for the
whole period of sorption, in comparison to the model
of pseudo second-order (Figure 6B) which gives func-
tional straight line for all initial concentrations of dye.
Based on this, it can be said that, in this particular case,
the pseudo second-order model is more usable. Simi-
lar behaviour is observed for mass of material of 4
and 6 g, hence those results are not shown in this text.
Tables 2 and 3 show results of kinetic para-
meters of process of adsorption DB79 on PES fabric
(equilibrium rates constant of kinetics for pseudo first-
-order and pseudo second-order) for the used mass
of adsorbent, all initial concentration of dye, as well as
values for parameter q (calculated – qcal and experi-
mental – qexp).
Table 1. Analytical expressions of Langmuir isotherm with coefficient for DB79 – PES system
Adsorbent, g Analytical expression Langmuir’s parameters R2
(Langmuir’s equations) Qmax / mg g–1 b / dm3 mg–1
2 ce/qe = 9.635 + 0.078ce 12.67 0.008 0.973
4 ce/qe = 12.295 + 0.175ce 5.70 0.014 0.992
6 ce/qe = 9.057 + 0.180ce 5.55 0.020 0.997
Figure 6. Kinetic of sorption DB79 on 2 g of PES fabric (A - pseudo first-order; B - pseudo second-order).
M. KODRIC et al.: MODELLING OF POLYESTER FABRIC DYEING… Chem. Ind. Chem. Eng. Q. 23 (1) 131139 (2017)
138
Table 2. Kinetic parameters of process of adsorption of dye on
2 g of PES fabric (pseudo first-order)
Dye concentration
mg/dm3
qe,exp
mg/g
Pseudo first-order
k1 / min–1 qe,cal / mg g–1 R2
50 0.98 0.074 0.078 0.921
100 1.93 0.145 0.295 0.979
200 3.83 0.078 0.831 0,949
300 5.29 0.070 0.912 0.933
400 6.60 0.066 0.922 0.965
Table 3. Kinetic parameters of process of adsorption of dye on
2 g of PES fabric (pseudo second-order)
Dye concentration
mg/dm3
qe,exp
mg/g
Pseudo first-order
k2 / g mg–1 min–1 qe,cal / mg g–1 R2
50 0.98 3.63 0.98 0.999
100 1.93 1.63 1.95 0.999
200 3.83 0.32 3.86 0.998
300 5.29 0.28 5.32 0.998
400 6.60 0.28 6.61 0.999
Although the coefficient of determination R2 for
kinetic model of pseudo first-order is higher than 0.92,
completely different values are obtained for the cal-
culated parameter (qcal) in comparison to those for the
experimental parameter (qexp). Because of this, ads-
orption cannot be well described by the kinetic model
of pseudo first-order, because in many cases the
equation of pseudo first-order does not cover ade-
quately the whole range of contact time, which is con-
firmed by the results in Table 2.
In contrast, kinetic model of pseudo second-
order has in all cases R2≈1, Table 3, by which the
complete functionality is achieved and the model can
be completely used for describing processes of ads-
orption of dye on PES fabric. In addition, differences
between parameters qcal and qexp for this model are
insignificant, i.e. acceptable.
Concentration of dye decreases very fast during
initial absorption, before diffusion inside the particles
starts to control adsorption kinetics in all cases. Inc-
rease of contact time decreases the resistance of
borderline layers, and supported by ultrasound waves,
it intensifies the mobility of dye during the time of ads-
orption [20].
In addition, the paper of Carrion-Fité proposes a
dyeing process for polyester at low temperatures (40
C and below) with disperse dyes using a microemul-
sion prepared by ultrasonic agitation, composed of a
small proportion of organic solvent (alkyl halogen)
and phosphogliceride as the emulsifier [21]. The kin-
etics of this dyeing system are determined as a funct-
ion of temperature with various disperse dyes at dif-
ferent molecular weights. In general, dyes with lower
molecular weight have a faster dyeing rate and those
with a higher molecular weight have a lower rate.
Activation energies range from 20-40 kcal/mol; these
values are similar to those achieved in traditional dye-
ing with a carrier.
CONCLUSIONS
Modelling of disperse dye adsorption, i.e., dye-
ing of PES fabric was being tested under different
conditions without carrier in dye bath, although it is
well-known that dyeing of synthetic PES by disperse
dye is performed usually in the presence of this che-
mical and/or in conditions of high pressure and tem-
perature. Usually, the used carriers can show prob-
lems in usage for such dyed material or in production
itself during dyeing. Use of ultrasound could solve all
the drawbacks, having in mind health advantages and
economic savings.
Based on obtained experimental results, the fol-
lowing conclusions can be stated:
• non-standard dyeing of PES fabric without
carrier is possible by ultrasound in atmospheric con-
ditions;
• the longer the contact time, the higher the
amount of dye adsorbed on the material;
• the concentration of dye in solution dec-
reases with the duration of adsorption;
• the continuity of growth in the amount of
exhaustion dye with mass of material is noted, i.e.,
greater adsorption occurs in smaller proportion of bath;
• data obtained from teh Langmuir equation
shows that this model enables precise description of
experimental data;
The kinetic model of pseudo second-order ade-
quately describes disperse dye – PES fabric system.
REFERENCES
[1] T.K. Kim, Y.A. Son, Y.J. Lim, Dyes Pigm. 67 (2005) 229-
–234
[2] C.C. Lai, K.M. Chen, Text. Res. J. 78 (2008) 382-389
[3] Z. Ren, C. Qin, R.C. Tanga, G. Chen, J. Soc. Dyers
Colour. 128 (2012) 147–152
[4] T.S. Choi, Y. Shimizu, H. Shirai, K. Hamada, Dyes Pigm.
50 (2001) 55–65
[5] L. Wang, H.F. Zhao, J.X. Lin, J. Soc. Dyers Colour. 126
(2010) 243–248
[6] C. Udrescu, F. Ferrero, M. Periolatto, Ultrason. Sono-
chem. 21 (2014) 1477-1481
[7] S.A. Larik, A. Khatri, S. Ali, S.H. Kim, Ultrason. Sono-
chem. 24 (2015) 178-183
M. KODRIC et al.: MODELLING OF POLYESTER FABRIC DYEING… Chem. Ind. Chem. Eng. Q. 23 (1) 131139 (2017)
139
[8] S. Dhouib, A. Lallam, F. Sakli, Text. Res. J. 76 (2006)
271-280
[9] A.S. Ozcan, J. Supercrit. Fluids 35 (2005) 133–139
[10] R.G. Ovejero, J.R. Sanchez, J.B. Ovejero, J. Valldeperas,
M.J. Lis, Text. Res. J. 77 (2007) 804–809
[11] A. Ujhelyiova, E. Bolhova, J. Oravkinova, R. Tino, A.
Marcincin, Dyes Pigm. 72 (2007) 212-216
[12] L. Olenka, E.N. da Silva, W.L.F. dos Santos, A.F. Rubira,
E.C. Muniz, A.N. Medina, M.L. Baessoa, A.C. Bento,
Analyst 127 (2002) 310–314
[13] J.-H. Choi, M.-J. Kang, Fibers Polym. 7 (2006) 169-173
[14] K.W. Lee, J.P. Kim, Textile Res. J. 71 (2001) 395-398
[15] K.W. Lee, Y.S. Chung, J.P. Kim, Textile Res. J. 73 (2003)
755-761
[16] M.A. Iskender, B. Becerir, A. Koruyucu, Text. Res. J. 75
(2005) 462-465
[17] A.N. Saligram, S.R. Shukla, M. Mathur, J. Soc. Dyers
Colour. 109 (1993) 263–266
[18] L. Wang, H.F. Zhao, J.X. Lin, J. Soc. Dyers Colour. 126
(2010) 243–248
[19] W.Y.W. Ahmad, M. Lomas, J. Soc. Dyers Colour. 112
(1996) 245-248
[20] M.V. Kraan, M.V. Cid, G.F. Woerlee, W.J. Veugelers,
G.J. Witkamp, Text. Res. J. 77 (2007) 550-558
[21] F.J. Carrion-Fité, Text. Res. J. 65 (1995) 362-368.
MARIJA KODRIĆ
SANDRA STOJANOVIĆ
BRANKA MARKOVIĆ
DRAGAN ĐORĐEVIĆ
Univerzitet u Nišu, Tehnološki fakultet,
Bulevar oslobođenja 124, Leskovac,
Srbija
NAUČNI RAD
MODELOVANJE BOJENJA POLIESTARSKE TKANINE U PRISUSTVU ULTRAZVUČNIH TALASA
U radu je razmotreno modelovanje bojenja, odnosno, adsorpciono ponašanje disperzne
boje na poliestru (bojenje) pri delovanju ultrazvuka, sa ciljem dobijanja podataka o meha-
nizmu vezivanja boje i definisanju uslova procesa bojenja ovog sintetičkog vlakna uz
dodatni izvor energije bez primene kerijera, jedinjenja koja povećavaju permeabilnost vla-
kana i pomažu bojenje. Bojenje-adsorpcija je vođena pod različitim uslovima, varirana je
koncentracija boje, masa supstrata, receptura i vreme bojenja. Utvrđeno je da ultrazvuk
dozvoljava bojenje bez kerijera a efikasnost bojenja zavisi od vremena kontakta, početne
koncentracije boje i količine adsorbenta - tkanine. Postoji kontinuitet rasta količine vezane
boje sa masom adsorbenta. Karakteristični prikazi dobijeni iz Langmuir-ove izoterme potvr-
dili su da ovaj model obezbeđuje dovoljno precizan opis bojenja poliestra disperznom
bojom. Kinetika bojenja odlično je protumačena pseudo II redom s obzirom na visoku funk-
cionalnost.
Ključne reči: adsorpcija, poliestar, disperzna boja, ultrazvuk, Langmuir izoterma,
kinetika.
Chemical Industry & Chemical Engineering Quarterly
Available on line at
Association of the Chemical Engineers of Serbia AChE
www.ache.org.rs/CICEQ
Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017) CI&CEQ
141
AISHI ZHU
SHANSHAN LIU
KANFENG WU
CHUAN REN
MAOQIAN XU
School of Biological and Chemical
Engineering, Zhejiang University of
Science and Technology,
Hangzhou, China
SCIENTIFIC PAPER
UDC 633.17:577.114.4:66.061:66
https://doi.org/10.2298/CICEQ151011026Z
COMPARING OF HOT WATER AND ACID EXTRACTION OF POLYSACCHARIDES FROM PROSO MILLET
Article Highlights
• Polysaccharides were extracted from proso millet with hot water and acid solution
• Response surface methodology was used and a model was set up
• The best extraction methods were obtained
• The polysaccharides yield of acid extraction was significantly higher than hot water
extraction
Abstract
The extraction of polysaccharides from proso millet was investigated experi-
mentally using hot water and acid aqueous solution. Response surface method-
ology, based on a three-level, three- or four-variable Box-Behnken design for hot
water extraction or acid extraction, respectively, was employed to obtain the best
possible combination of acid concentration, liquid-solid ratio, extraction time, and
extraction temperature for maximum polysaccharides yield. The obtained experi-
mental data were fitted to a second-order polynomial equation and analyzed by
appropriate statistical methods. The corresponding optimum extraction condi-
tions of each method were obtained. Under the optimum conditions, the experi-
mental yield was well in close agreement with the predicted value by the model.
The results showed that the polysaccharides yield of acid extraction was 42.13
mg g-1, significantly higher than 20.07 mg g-1 of the yield of hot water extraction,
the obtained equation could be used to predict the extraction experimental
results.
Keywords: proso millet, polysaccharides, hot water extraction, acid ext-raction, response surface methodology.
The proso millet (Panicum miliaceum L.) is a
warm-season grass with a short growing season and
low moisture requirement that is capable of producing
food or feed where other grain crops would fail. It is
the a widely planted species of millet; is extensively
cultivated for its grain in the arid areas of China, India,
Russia, Ukraine, the Middle East, Turkey and Rom-
ania. The seeds as grain are small (2–3 mm) and rich
in starch, protein, fat, dietary fiber, vitamin and trace
elements. It has many functions beneficial to human
health, such as the prevention and management of
diabetes mellitus [1], callus and shoots regeneration
Correspondence: A. Zhu, School of Biological and Chemical
Engineering, Zhejiang University of Science and Technology,
318 Liuhe Road, Hangzhou 310023, China. E-mail: [email protected] Paper received: 11 October, 2015 Paper revised: 2 January, 2016 Paper accepted: 26 April, 2016
from protoplasts of proso millet [2]. Polysaccharides
have unique biological properties such as anti-oxid-
ative [3], anti-viral [4] and anti-complementary acti-
vities [5], as well as chemical and physical properties
[6]. They have important effects in the process of
growth and development for living organisms [7].
Response surface methodology (RSM) is an
effective statistical technique for optimizing complex
processes. It is widely used in optimizing the extract-
ion process variables [8,9].
Kim et al. [1] studied the inhibitory effects of
ethanol extracts from proso millet on α-glucosidase
and α-amylase activities, Heyser [2] studied the callus
and shoot regeneration from protoplasts of proso
millet, Yañez et al. [10] studied some chemical and
physical properties of proso millet starch. However,
there is no report about polysaccharides extraction
from proso millet, especially extraction with acid sol-
A. ZHU et al.: COMPARING OF HOT WATER AND ACID EXTRACTION… Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017)
142
ution. In this article, we report on the extraction of
polysaccharides from proso millet by hot water and
acid solution. Polysaccharides extracted with the
acidic aqueous solution can be made for a pure poly-
saccharide, greatly improve the yield of polysacchar-
ide, and the activity of polysaccharide is higher [11],
high acid concentration can accelerate the degrad-
ation of polysaccharides [8]. Based on the results of
the single factor investigation, once extracted and stir-
ring rate of 150 rpm in the extraction process, RSM
was designed to help to optimize extraction variables
of acid concentration (hydrochloric acid aqueous sol-
ution, mol L-1), liquid-solid ratio (water or acid aque-
ous solution volume with proso millet mass, mL g-1),
extraction time (h) and extraction temperature (C)
and to systematically analyzed the influence of the
variables.
MATERIALS AND METHODS
Materials and extraction of polysaccharides
The decorticated grain proso millet was obtained
from Donghua rice industry limited company, Jinzhou,
Liaoning, China, 2014 harvest. The size of proso mil-
let particle was about 2–3 mm. Before the extraction
experiment, the sample was placed first in electro-
thermal blast oven (DHG-9123A, Precision Experi-
mental Facilities Limited Company, Shanghai, China)
to dry at 60.0 C to constant mass. Five grams of dry
proso millet, weighed with an electronic balance
(BS124S Sartorius Instrument System Limited Com-
pany, Beijing, China) and mixed with different volume
of water or different concentration acid solution was
extracted in a 500-ml three-neck flask. The three-
-neck flask was soaked in a thermostatic water bath
(DK-S24 Jing Hong Experimental Facilities Limited
Company, Shanghai, China) which controlled the
needed extraction temperature, with stirring at 150
rpm with an electric mixing paddle for a given time
during the entire extraction process. After extraction,
the mash was quickly put in cold bath, cooled down to
room temperature and then vacuum-filtered. The fil-
trate was concentrated in a rotary evaporator (RE52CS
Yarong Biochemistry Instrument Plant, Shanghai,
China) to 20% of the initial volume at 60.0 C under
vacuum. The obtained solution was mixed with four
volumes of dehydrated ethanol (ethanol final concen-
tration, 80%) to obtain the precipitate. Then, the sus-
pension was centrifuged using centrifuge (800B
Shanghai Anting Scientific Instrument Factory, Shan-
ghai, China) at 4000 rpm for 15 min, the precipitate
was collected as extract and washed three times with
dehydrated ethanol. The extract was dried at 50 C in
electrothermal blast oven until constant mass and
weighed. The concentration of polysaccharides con-
tent in the extract was examined with a spectrophoto-
meter (722E Shanghai Spectrum Instruments Co.,
LTD, Shanghai, China). The proso millet was ext-
racted once and the extraction procedure was rep-
eated twice. All reagents were of analytical grade.
Experimental design
The Box-Behnken Design was applied to deter-
mine the best combination of extraction variables for
the yield of proso millet polysaccharides. A three-
level, three-variable experimental design was carried
out to hot water extraction, and three extraction vari-
ables considered for this research were: liquid-solid
ratio (mL g-1, X1), extraction time (h, X2), and extract-
ion temperature (C, X3). A three-level, four-variable
experimental design was used to acid extraction, and
four extraction variables considered for these
researches were: acid concentration (mol L-1, X1),
liquid-solid ratio (mL g-1, X2), extraction time (h, X3),
and extraction temperature (C, X4) [8]. The reason-
able range of variables was obtained through the
single factor investigation. The polysaccharides ext-
raction yield was as the dependent variable. The
complete experiment scheme was consists of 17 or
29 experimental points (including five replicates of the
center point) for three-level, three-variable experi-
mental design or three-level, four-variable experimen-
tal design, respectively. A quadratic polynomial model
was used to fit the obtained experimental data and
the regression coefficients were obtained through sta-
tistical analyses. The mathematic expression of the
quadratic polynomial model was seen Eq. (1) [8,12-
–14]:
4 4 3 4
20
1 1 1 1i i ii i ij i j
i i i j j
Y X X X X (1)
where Y is the measured response (polysaccharides
yield) of each experiment; β0, βi, βii, βij are constant
regression coefficients of the model; X is the code
levels of independent variables; XiXj and Xi2 represent
the interaction and quadratic terms of independent
variables, respectively.
Analysis of samples
The phenol-sulphuric acid colorimetric method
at 485 nm was used to determine the polysaccharides
content [12,13]. In this method the glucose was used
as standard, the polysaccharides yield (mg g-1) was
calculated using Eq. (2):
1000XV
YM
(2)
A. ZHU et al.: COMPARING OF HOT WATER AND ACID EXTRACTION… Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017)
143
where Y is yield of polysaccharides, mg g-1; V is the
total volume of the coarse polysaccharides dissolved
after the constant volume, L; M is dosage of raw
material millet, g; X is concentration of polysacchar-
ides in solution, mg L-1. X was calculated using Eq.
(3):
0.0255 0.00006X A (3)
where A is the absorbance of measured solution at
485 nm.
Statistical analyses
The Design Expert Software (version 8.0.5.0,
Stat-Ease Inc., Minneapolis, MN) was used to mul-
tiple nonlinear regressions of the responses obtained
from each design experimental. The fitting degree of
between the experimental data and the equation was
inspected using the coefficient of determination R2,
F-test and p-value were used for checking the sig-
nificance of the regression coefficient. p-Values below
0.05 were regarded as statistically significant.
RESULTS AND DISCUSSION
Hot water extraction
Model fitting. The factors and their levels of hot
water extraction were chosen on the basis of single-
factor experiments (Table 1). The experiment design
and the yields are shown in Table 1. The regression
analysis results are shown in Table 2.
Table 1. Experimental design and results of hot water extraction
Test number Liquid-solid ratio Extraction time Extraction temperature
Y / mg g-1 X1 / mL g-1 X2 / h X3 / C
1 20:1 1.0 65 7.85±0.22
2 20:1 1.0 75 16.11±0.33
3 20:1 2.0 65 6.20±0.29
4 20:1 2.0 75 17.91±0.33
5 15:1 1.5 65 4.14±0.21
6 15:1 1.5 75 17.19±0.25
7 25:1 1.5 65 7.63±0.27
8 25:1 1.5 75 15.24±0.29
9 15:1 1.0 70 16.52±0.34
10 15:1 2.0 70 14.88±0.31
11 25:1 1.0 70 14.75±0.27
12 25:1 2.0 70 17.94±0.21
13 20:1 1.5 70 19.82±0.22
14 20:1 1.5 70 19.20±0.22
15 20:1 1.5 70 19.46±0.22
16 20:1 1.5 70 19.34±0.22
17 20:1 1.5 70 19.27±0.22
Table 2. Analysis of variance for regression equation to hot water extraction; R2 = 0.9987, Adj. R2 = 0.9970; **extremely significant;
*significant
Source Sum of squares df Mean square F-value Prob > F Significance
Model 423.54 9 47.06 597.64 < 0.0001 **
X1 1.00 1 1.00 12.71 0.0092 **
X2 0.36 1 0.36 4.59 0.0694 –
X3 206.35 1 206.35 2620.53 < 0.0001 **
X1X2 5.83 1 5.83 74.07 < 0.0001 **
X1X3 7.40 1 7.40 93.96 < 0.0001 **
X2X3 2.98 1 2.98 37.79 0.0005 **
X12 20.04 1 20.04 254.47 < 0.0001 **
X22 6.21 1 6.21 78.81 < 0.0001 **
X32 161.15 1 161.15 2046.50 < 0.0001 **
A. ZHU et al.: COMPARING OF HOT WATER AND ACID EXTRACTION… Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017)
144
Table 2. Continued
Source Sum of squares df Mean square F-value Prob > F Significance
Residual 0.55 7 0.079 –
Lack of fit 0.31 3 0.10 1.74 0.2962 –
Pure error 0.24 4 0.060 –
Cor. total 424.09 16 –
The response surface plots graphs are shown in
Figure 1. The response value Y (i.e., the proso millet
polysaccharides yield) can be expressed by Eq. (4) in
terms of actual values:
1 2 3
1 2 1 3 2 3
2 2 21 2 3
1337.566 6.645 18.817 36.231
0.483 0.054 0.345
0.087 4.856 0.247
Y X X X
X X X X X X
X X X
(4)
The results of the analysis of variance, good-
ness-of-fit and the adequacy of the models are sum-
marized in Table 2. The value of probability (p) was
less than 0.05, which indicates that the selected
factors and their ranges have significant influence on
the yield of polysaccharides. The residual analysis
was then performed to check the adequacy of the
developed model and determine whether the approx-
imating model would give poor or misleading results.
Figure 2 shows the residual and the influence plots
for the experimental data [9]. The predicted values
obtained are quite close to the experimental values,
and the points of all predicted and experimental
response values fall very close to the 45 line (Figure
2a), indicating that the model developed is successful
in capturing the correlation between the process vari-
ables on the response. Figure 2b shows the normal
probability plot of residuals for response is normally
distributed, as they lie reasonably close on a straight
line and shows no deviation of the variance. The
goodness of fit of the model is analyzed by construct-
ing the internally studentized residuals versus experi-
mental runs and shows that all the data points lay
within the limits (Figure 2c). Since the Cook’s dis-
tance values are in the determined range (Figure 2d),
there is no strong evidence of influential observations
in experimental data. The above results indicate a
good adequate agreement between BBD experimen-
tal data and the model could be better predicted yield
of polysaccharides. Linear term of X3 (extraction tem-
perature, p < 0.0001) showed the largest effect on
polysaccharide yield, followed by linear term of X1
(liquid-solid ratio, p = 0.0092 < 0.05) , all interaction
terms and all quadratic terms (p < 0.0001) were also
extremely significant. Linear term of X2 (extraction
time, p = 0.0694) was however not significant (p >
> 0.05). The p-value of model was less than 0.0001
and Adj R2 was 0.9970 which would give a better fit to
the mathematical model (Eq. (4)).
Figure 1. a) Response surface plot of Y = f(X1,X2) (X3 = 70.0
C); b) response surface plot of Y = f(X1,X3) (X2 = 1.5 h);
c) Response surface plot of Y = f(X2,X3) (X1 = 20.0 mL g-1);
X1: liquid-solid ratio (mL g-1), X2: extraction time (h), X3:
extraction temperature (C), Y: yield (mg g-1).
A. ZHU et al.: COMPARING OF HOT WATER AND ACID EXTRACTION… Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017)
145
Verification of the model. The regression model
predicted the optimum extraction conditions of proso
millet polysaccharides, which were liquid-solid ratio
19.77:1 mL g-1, extraction time 1.85 h, extraction tem-
perature 71.79 C. Under this optimum extraction
conditions, the polysaccharides yield was predicted
for 20.17 mg g-1.
The experiment was repeatedly carried out three
times for verifying the prediction from the model at
liquid-solid ratio 19.8:1 mL g-1, extraction time 1.8 h
and extraction temperature 71.8 C. The average
value of practical polysaccharides yield was 20.07±
±0.21 mg g-1(n = 3), the relative error was -0.5%
compared with predicted yield of 20.17 mg g-1. The
analysis results proved that the quadratic polynomial
model was suitable for expressing the optimization
results and the satisfaction and accuracy of Eq. (4)
was high.
Acid extraction
Model fitting. The factors and their levels of acid
extraction were chosen on the basis of single-factor
experiments (Table 3). The experiment design pro-
posal and the yields are shown in Table 3. The reg-
ression analysis results are shown in Table 4.
Table 3. Experimental design and results of acid extraction
Test number Acid concentration Liquid-solid ratio Extraction time Extraction temperature
Y / mg g-1 X1 / mol L-1 X2 / mL g-1 X3 / h X4 / C
1 3.0 20:1 1.5 70 30.48±0.53
2 2.5 25:1 1.5 60 12.08±0.26
3 2.0 25:1 1.5 70 22.67±0.35
4 2.5 25:1 1.0 70 22.94±0.45
Figure 2. Diagnostic plots for the model adequacy of hot water extraction; a) predicted vs. actual, b) normal plot of residuals, c) residuals
vs. run and d) Cook’s distance.
A. ZHU et al.: COMPARING OF HOT WATER AND ACID EXTRACTION… Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017)
146
Table 3. Continued
Test number Acid concentration Liquid-solid ratio Extraction time Extraction temperature
Y / mg g-1 X1 / mol L-1 X2 / mL g-1 X3 / h X4 / C
5 3.0 20:1 2.0 70 26.86±0.58
6 2.5 20:1 1.0 80 31.00±0.54
7 2.5 15:1 2.0 70 25.64±0.42
8 2.5 15:1 1.5 60 12.88±0.26
9 2.0 20:1 1.5 60 10.24±0.36
10 2.0 20:1 1.0 70 17.88±0.28
11 3.0 20:1 1.5 60 26.87±0.37
12 3.0 20:1 1.5 80 31.88±0.48
13 2.0 20:1 1.5 80 35.50±0.63
14 2.5 25:1 2.0 70 27.50±0.53
15 2.5 25:1 1.5 80 34.99±0.57
16 2.5 20:1 2.0 80 32.09±0.42
17 2.5 15:1 1.0 70 25.58±0.34
18 2.0 20:1 2.0 70 29.97±0.38
19 2.5 20:1 2.0 60 14.98±0.29
20 3.0 25:1 1.5 70 30.48±0.39
21 3.0 15:1 1.5 70 27.27±0.47
22 2.5 15:1 1.5 80 31.61±0.50
23 2.0 15:1 1.5 70 23.55±0.41
24 2.5 20:1 1.0 60 15.39±0.45
25 2.5 20:1 1.5 70 40.67±1.16
26 2.5 20:1 1.5 70 37.59±1.16
27 2.5 20:1 1.5 70 38.25±1.16
28 2.5 20:1 1.5 70 38.52±1.16
29 2.5 20:1 1.5 70 40.07±1.16
Table 4. Analysis of variance for regression equation to acid extraction; R2 = 0.9641, Adj. R2 = 0.9281; **extremely significant;
*significant
Source Sum of squares df Mean square F-value Prob > F Significance
Model 2011.80 14 143.70 26.82 <0.0001 **
X1 77.36 1 77.36 14.44 0.0020 **
X2 0.099 1 0.099 0.019 0.8936 –
X3 17.07 1 17.07 3.19 0.0960 –
X4 913.18 1 913.18 170.42 <0.0001 **
X1X2 4.20 1 4.20 0.78 0.3911 –
X1X3 37.06 1 37.06 6.92 0.0198 *
X1X4 102.48 1 102.48 19.12 0.0006 **
X2X3 14.06 1 14.06 2.62 0.1276 –
X2 X4 4.44 1 4.44 0.83 0.3782 –
X3 X4 0.56 1 0.56 0.11 0.7504 –
X12 213.43 1 213.43 39.83 <0.0001 **
X22 316.17 1 316.17 59.00 <0.0001 **
X32 320.45 1 320.45 59.80 <0.0001 **
X42 448.60 1 448.60 83.72 <0.0001 **
Residual 75.02 14 5.36 – – –
Lack of fit 68.32 10 6.83 4.08 0.0938 –
Pure error 6.70 4 1.67 – – –
Cor. total 2086.82 28 – – – –
A. ZHU et al.: COMPARING OF HOT WATER AND ACID EXTRACTION… Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017)
147
The response surface plots are shown in Figure
3. The response value Y can be expressed by the
following second order polynomial equation in terms
of actual values:
1 2
3 4 1 2
1 3 1 4 2 3
22 4 3 4 1
2 2 22 3 4
906.997 200.730 7.565
96.915 14.512 0.410
12.175 1.012 0.750
0.021 0.075 22.944
0.279 28.115 0.083
Y X X
X X X X
X X X X X X
X X X X X
X X X
(5)
The results of the analysis of variance, good-
ness-of-fit and the adequacy of the models are sum-
marized in Table 4. Figure 4 shows the residual and
the influence plots for the experimental data, it is
similar to Figure 2. The linear term of X1 (acid con-
centration, p = 0.002) and X4 (extraction temperature,
p < 0.0001) had an extremely significant effect on
polysaccharide yield. Each quadratic term (p <
< 0.0001) and the interaction term of X1 and X4 (p =
= 0.0006) were also extremely significant. The inter-
Figure 3. Response surface plots of: a) Y = f(X1,X2) (X3 = 1.5 h, X4 = 70.0 C); b) Y = f(X1,X3) (X2 = 20 mL g-1, X4 = 70.0 C);
c) Y = f(X1,X4) (X2 = 20 mL g-1, X3 = 1.5 h); d) Y = f(X2,X3) (X1 = 2.5 mol L-1, X4 = 70.0 C); e) Y = f(X2,X4) (X1 = 2.5 mol L-1, X3 = 1.5 h);
f) Y = f(X3,X4) (X1 = 2.5 mol L-1, X2 = 20 mL g-1); X1: acid concentration (mol L-1), X2: liquid-solid ratio (mL g-1), X3: extraction time (h),
X4: extraction temperature (C), Y: yield (mg g-1).
A. ZHU et al.: COMPARING OF HOT WATER AND ACID EXTRACTION… Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017)
148
action term of X1 and X3 (acid concentration and ext-
raction time, p = 0.0198 < 0.05) was significant. The
other terms were however not significant (p > 0.05).
The p-value of model was less than 0.0001 and Adj.
R2 was 0.9281 which would give a better fit to the
mathematical model (Eq. (5)).
Verification of the model. The optimum con-
ditions of acid extraction predicted by regression
model were acid concentration 2.48 mol L-1, liquid-
-solid ratio 20.3:1 mL g-1, extraction time 1.56 h and
extraction temperature 75.45 C. Under these opti-
mum extraction conditions, the polysaccharides yield
was predicted for 41.41 mg g-1.
The experiment was repeated three times at
acid concentration 2.48 mol L-1, liquid-solid ratio
20.3:1 mL g-1, extraction time 1.56 h and extraction
temperature 75.5 C. The average value of practical
yield of polysaccharide was 42.13±0.15 mg g-1(n=3),
the relative error was 1.74% compared with predicted
yield of 41.41 mg g-1. The analysis results confirmed
that the second order polynomial equation was
adequate for reflecting the expected optimization and
the Eq. (5) was satisfactory and accurate.
Effect of acid on yield of polysaccharides
The experimental results showed that the poly-
saccharides yield of acid extraction was significantly
higher than that of hot water extraction. Extraction
yield was increased 109.9%. This is because the acid
can help eliminate the physical and chemical effect
between cell wall of polymer molecules, make more
polysaccharides to dissolve from cells into solution in
the same extraction time, and thus the yield of poly-
saccharides is increased. However, high acid concen-
tration would reduce the polysaccharides yield because
of the destruction of the structure of polysaccharide
caused by the acid catalyzed hydrolysis [8].
Effect of other factors on yield of polysaccharides
The obtained experiment data indicated that
liquid-solid ratio, extraction time and extraction tem-
perature of each extraction method were similar, and
their trends of influence on polysaccharides yield in
Figure 4. Diagnostic plots for the model adequacy of acid extraction; a) predicted vs. actual, b) normal plot of residuals, c) residuals vs.
run and d) Cook’s distance.
A. ZHU et al.: COMPARING OF HOT WATER AND ACID EXTRACTION… Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017)
149
each extraction process were also similar. The yield
raised with increasing liquid-solid ratio because the
polysaccharides concentration difference between
sample and solution was increased to result in the
raised of mass transfer driving force. However, if the
liquid-solid ratio was too high, it would lead to inc-
reased operating costs, and the increased of amount
of acid in the extract system would lead to the des-
truction of the structure of polysaccharide caused by
the acid catalyzed hydrolysis. These results are in
good agreement with previous observations [8,13-15].
The extraction time had a similar effect on yield as
liquid-solid ratio. The yield first increased and then
decreased with increasing of the extraction time. This
phenomenon could be explained by that the proso
millet cell-wall was broken, the liquid was infiltrated
into the dried sample, the polysaccharides in sample
was dissolved and subsequently diffused out from the
sample to exterior solvent, which needed a long time,
but the polysaccharides dissolved in solution would
be partly degradation because they remained in the
solution for a long time [13,16]. The extraction yield
was found to increase with increase of extraction tem-
perature, and then decreased after a peak. The rea-
son was that the increased system temperature res-
ulted in the decreased of solvent viscosity to enhance
the solvent and solute diffusivity within suspension
solution system, which raised the polysaccharides
solubility. But the polysaccharide could be degraded
under high temperature, so the yield would be dec-
reased [13,15]. There were no obvious differences
between Figures 2 and 4, both shown that no obvious
patterns were found in the analysis of model and
indicated the accuracy of the developed model.
CONCLUSION
The performance of the extraction of polysac-
charides from proso millet was studied with a sta-
tistical method based on the response surface meth-
odology in order to identify and quantify the variables
that may maximize the yield of polysaccharides. The
second order polynomial model had higher correlation
for experiment data and could be better used for
optimizing proso millet polysaccharides extraction
technology. The optimum conditions of hot water ext-
raction were liquid-solid ratio 19.8:1 mL g-1, extraction
time 1.8 h and extraction temperature 71.8 C, and
the yield was 20.07±0.21 mg g-1 (n = 3). The optimum
conditions of acid extraction were acid concentration
2.48 mol L-1, liquid-solid ratio 20.3:1 mL g-1, extraction
time 1.56 h and extraction temperature 75.5 C, the
polysaccharides yield was 42.13±0.15 mg g-1 (n = 3).
The acid solution extraction of polysaccharides was
better than hot water extraction; the yield of acid sol-
ution extraction was higher by 109.9% than of hot
water extraction. The optimal extraction conditions of
hot water extraction and acid extraction were deter-
mined, and under the optimum conditions, the prac-
tical yield was agreed closely with the predicted yield
value.
REFERENCES
[1] J.S. Kim, T.K. Hyun, M.J. Kim, Food Chem. 124 (2011)
1647–1651
[2] J.W. Heyser, Z. Pflanzenphysiol. 113 (1984) 293–299
[3] S.P. Li, K.J. Zhao, Z.N. Ji, Z.H. Song, T.T.X. Dong, C.K.
Lo, J.K.H. Cheung, S.Q. Zhu, K.W.K. Tsim, Life Sci. 73
(2003) 2503–2513
[4] H.X. Xu, S.H.S. Lee, S.F. Lee,, R.L. White, J. Blay,
Antivir. Res. 44 (1999) 43–54
[5] A.B. Samuelsen, I. Lund, J.M. Djahromi, B.S. Paulsen,
J.K. Wold, S.H. Knutsen, Carbohydr. Polym. 38 (1999)
133–143
[6] I.A. Schepetkin, M.T. Quinn, Int. Immunopharmacol. 6
(2006) 317–333
[7] B. Yang, Y.M. Jiang, R. Wang, M.M. Zhao, J. Sun, Food
Chem. 112 (2009) 428–431
[8] C.Y. Gan, N.H. Abdul Manaf, A.A. Latiff, Carbohydr.
Polym. 79 (2010) 825–831
[9] J. Prakash Maran, S. Manikandan, K. Thirugnanasam-
bandham, C. Vigna Nivetha, R. Dinesh, Carbohydr.
Polym. 92 (2013) 604–611
[10] G.A. Yañez, C.E. Walker, L.A. Nelson, J. Cereal Sci. 13
(1991) 299–305
[11] H. Yaich, H. Garna, S. Besbes, M. Paquot, C. Blecker, H.
Attia, Food Hydrocolloids 31 (2013) 375–382
[12] D.H. Luo, Carbohydr. Polym. 90 (2012) 284–288
[13] A.S. Zhu, J.N. Ye, F.N. Yan, Chem. Ind. Chem. Eng. Q.
20 (2014) 579−585
[14] X.Z. Zhang, F.P. Yin, C.H. Liu, X.W. Xu, J. Northeast
Agric. Univ. 18 (2011) 79–86
[15] C.J. Ren, L. Gao, C.M. Wang, M.C. Wu, J.F. Chen,
Trans. CSAE 24 (2008) 251–254
[16] Y. Chen, M.Y. Xie, X.F. Gong, J. Food Eng. 81 (2007)
162–170.
A. ZHU et al.: COMPARING OF HOT WATER AND ACID EXTRACTION… Chem. Ind. Chem. Eng. Q. 23 (1) 141150 (2017)
150
AISHI ZHU
SHANSHAN LIU
KANFENG WU
CHUAN REN
MAOQIAN XU
School of Biological and Chemical
Engineering, Zhejiang University of
Science and Technology, Hangzhou,
China
NAUČNI RAD
POREĐENJE EKSTRAKCIJA POLOSAHARIDA IZ PROSA TOPLOM VODOM I KISELIM RASTVOROM
Ispitivana je ekstrakcija polisaharida iz prosa pomocu tople vode i kiseli vodeni rastvor. Za
određivanje najbolje moguće kombinacije koncentracije kiseline, odnosa tečno-čvrsto, vre-
mena i temperature ekstrakcije koja obezbeđuje ostvarivanje najvećeg prinosa polisaha-
rida ekstrakcijom sa toplom vodom, odnosno kiselim rastvorom, korišćena je metodologija
odgovora površine, Box–Behnken dizajnu sa tri nivo i 3, odnosno 4 faktora. Eksperimen-
talni podaci su fitovani polinomnom jednačinom drugog reda i analizirani odgovarajucim
statističkim metodama, pri čemu su određeni odgovarajuci optimalni uslovi obadve eks-
trakcione metode. Pod optimalnim uslovima, eksperimentalni prinos se dobro slaže sa
vrednostima koje su izračunate modelom. Prinos polisaharida dobijenih kiselom ekstrak-
cijom iznosi 42,13 mg·g-1 i veći je od prinosa ostvarenog ekstrakcijom pomoću tople vode
(20,07 mg·g-1). Obe kvadratne jednačine se mogu koristiti za predviđanje rezultata eks-
trakcija.
Ključne reči: proso, polisaharidi, ekstrakcija toplom vodom, kisela ekstrakcija,
metodologija odgovora površine.