Journal of Environmental Chemical Engineering 4 (2016) 3309–3321

Contents lists available at ScienceDirect

Journal of Environmental Chemical Engineering

journal homepage: www.elsevier.com/locate/jece

Removal of carbamazepine from municipal wastewater effluent using

optimally synthesized magnetic activated carbon: Adsorption and
sedimentation kinetic studies
Majid Baghdadi* , Elham Ghaffari, Behnoush Aminzadeh
Department of Environmental Engineering, Graduate Faculty of Environment, University of Tehran, P.O. Box: 1417853111, Tehran, Iran

A R T I C L E I N F O A B S T R A C T

Article history:
Received 19 April 2016 This study investigated adsorption of the carbamazepine (CBZ) from wastewater effluent to optimally
Received in revised form 25 June 2016 synthesized magnetic nanocomposite of activated carbon (AC). The adsorbent was characterized by SEM,
Accepted 29 June 2016 XRD, VSM, BET and FTIR analysis. Increase in the magnetite content up to mass ratio of 1:8 (Fe3O4: AC) had
Available online 30 June 2016 no adverse effect on the porous structure of adsorbent and decrease in adsorption capacity was only due
to increase in inactive mass fraction of adsorbent. However, at higher contents of magnetite, decrease in
Keywords: adsorption capacity may be due to blockage of porous structure of AC. Sedimentation data of MAC 1:8
Carbamazepine were well predicted by second order kinetic model with the rate constant of 8.9
103 NTU1 min 1.
Magnetic activated carbon
Response surface methodology was applied to optimize the treatment process. A significant quadratic
Sedimentation
model was achieved with the maximum removal efficiency of 93% within investigated intervals. By
Removal
Adsorption increasing of total dissolved solids (TDS); adsorption performance decreased slightly due to increase in
Wastewater effluent water cluster formation on carboxylic groups which can block adsorption on the carbonyl sites. The
equilibrium adsorption data were in agreement with the results predicted using Radke-Prausnitz,
Redlich-Peterson and Temkin models. The maximum capacity was found to be 182.9 mg g1. The
adsorption kinetic data were closely fitted to Elovich, intraparticle diffusion and pseudo-second-order
models. According to thermodynamic parameters, the adsorption process was found to be spontaneous
and exothermic in nature. The error functions were used to evaluate the agreement between the
experimental data and predictions using models.
ã 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Over the last decades, the occurrence of pharmaceuticals in
aquatic environments has recognized as one of the environmental
issues. Carbamazepine (CBZ) as an antiepileptic drug is among the
most frequently reported micropollutants in surface waters [1].
CBZ was observed at a concentration more than 10 times higher
than that of other micropollutants due to its high persistency [2].
The main source of pharmaceuticals in the aquatic environments
was found to be effluents of urban wastewater treatment plants
Abbreviations: CBZ, carbamazepine; AC, activated carbon; GAC, granular
activated carbon; PAC, powder activated carbon; MAC, magnetic activated carbon;
NAT-AC, Nitric acid treated- activated carbon; RSM, response surface methodology;
CCD, central composite design; RMSE, residual root-mean-square error; AARE,
average absolute relative error.
* Corresponding author.
E-mail addresses: m.baghdadi@ut.ac.ir (M. Baghdadi), elham.ghaffari@ut.ac.ir
(E. Ghaffari), bamin@ut.ac.ir (B. Aminzadeh).
(WWTPs) [3], because the urban WWTPs have not been designed
for micropollutants removal. CBZ was reported to be the most
persistent pharmaceutical which can be removed through
biological wastewater treatment on average by only 32.7%. The
highest removal of CBZ in WWTPs was reported as high as 64.5%
[4]. Therefore, discharge of effluents of urban WWTPs to
environment can lead to occurrence of micropollutants in the
aquatic environments, and subsequently in drinking water [5].
Therefore, finding the efficient post treatment is required to reduce
the release of micropollutants into surface waters.
In recent years, various treatment technologies have been
presented for removal of micropollutants from effluents of
WWTPs, including, coagulation-flocculation [6,7], chemical
oxidation [8,9] and adsorption [10–12]. Coagulation–flocculation
process is ineffective in removal of most micropollutants with the
removal efficiencies ranged from nil to 50%. The removal efficiency
of micropollutants with high octanol-water partition coefficient
(KOW) is as high as 50% [1]. The removal efficiency of 2% was
reported for CBZ (KOW: 2.54) using coagulation–flocculation

http://dx.doi.org/10.1016/j.jece.2016.06.034
2213-3437/ã 2016 Elsevier Ltd. All rights reserved.
3310 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321
process [6]. The formation of by-products is the main disadvantage
of chemical oxidation processes [13], which may increase the
toxicity of treated wastewater. In order to remove the oxidation by-
products, sand filtration or activated carbon (AC) filtration are
required. The adsorption process is more effective in micro-
pollutants removal in comparison with coagulation–flocculation
process [14]. Several adsorbents have been recently reported for
removal of micropollutants from wastewaters, including carbon
nanotube [15], graphene oxide [16] and functionalized materials
[17]. However, theses adsorbents have a low potential for large-
scale applications. Activated carbon is one of the well-known
adsorbent which has great specific surface area and pore structure.
Adsorption by granular activated carbon (GAC) is the most
commonly used process which can be operated in large-scale
for water and wastewater treatment. However, powder activated
carbon (PAC) has been considered as a more efficient adsorbent
compared to GAC for removal of persistent organic micropollutants
[18] and inorganic pollutants [19]. However, the large scale
application of PAC has been limited in wastewater treatment due to
small particle size and subsequently difficulty in separation of its
particles from effluent after adsorption process [20,21]. In order to
overcome the mentioned disadvantage, magnetic nanocomposite
of AC has been proposed instead of AC which can be separated
easily after treatment by a magnetic separator. Magnetic activated
carbon (MAC) is mainly prepared by a two-step method in which
activated carbon is combined with magnetic nanoparticles.
Chemical co-precipitation iron salts at the presence of AC has
also been reported [22]. The loading of nanoparticles on the AC can
affect the porous structure of AC and subsequently its adsorption
capacity.
In this research, optimally synthesized magnetic nanocompo-
site of AC was used in order to overcome the limitation of large
scale application of PAC in micropollutants removal from effluent
of WWTPs. The effect of magnetite content of nanocomposite on its
adsorption capacity was investigated. The sedimentation kinetic of
magnetic nanocomposite was compared with that of AC. Central
composite design (CCD) was applied to optimize the CBZ removal
from effluent of urban wastewater. The effect of adsorbent dosage,
CBZ concentration, contact time and TDS were studied. The
thermodynamics, equilibrium and kinetics of adsorption process
were comprehensively investigated using several models. The
error functions were used to evaluate the extent of agreement
between the experimental data and predictions using models.
2. Materials and methods
2.1. Materials
Commercial powder activated carbon was supplied from Loba
Chemie (India). Nitric acid (65% wt), ammonia solution (25% wt)
and iron salts (FeCl24H2O and FeCl36H2O) were purchased from
Merck (Darmstadt, Germany) and used for preparation of the
adsorbent. Initial pH of solutions was adjusted using NaOH and HCl
which were also purchased from Merck. CBZ with a purity of 99.9%
was obtained from Sigma-Aldrich (St. Louis, MO). Stock solution of
CBZ was prepared at a concentration of 4000 mg L1 in acetonitrile
and was stored in dark at 10  C and used within 1 month. All
obtained reagents were of extra pure grade and were used without
further purification. Deionized water was used throughout the
whole experiments.
2.2. Wastewater effluent sample
Treated wastewater samples were taken from the effluent of
secondary sedimentation tank of Ekbatan wastewater treatment
plant (Tehran, Iran). The analysis of obtained sample is presented
in Table 1. Although the CBZ can only be found in municipal
wastewater in the range from a few ng L1 to tens of mg L1, in this
study to investigate the performance of prepared adsorbent, the
concentration of CBZ in the wastewater was kept higher than
2 mg L1.
2.3. Preparation of MAC
In this work, magnetite nanoparticles were first synthesized
and then mixed with nitric acid treated activated carbon (NAT-AC).
Nitric acid treatment can decrease the point of zero charge of AC
(pHpzc = 2). Magnetite nanoparticles have a high enough point of
zero charge (pHpzc = 6) to support a positively charged surface in
acidic medium. Therefore, magnetite nanoparticles and NAT-AC
attract each other at pH range of 2–6 due to the opposite charges.
Nitric acid treatment was performed using the method reported by
Jafari et al. [23]. Briefly, 40 g of AC was slowly added to 200 mL of
HNO3 (65% wt). The mixture was constantly stirred for 3 h at 80  C.
NAT-AC was filtered, rinsed several times with deionized water and
finally dried in an oven at 50  C for 24 h. Magnetite nanoparticles
were synthesized by co-precipitation of Fe2+ and Fe3+ ions in the
absence of atmospheric oxygen. For this purpose, 2.92 g of
FeCl36H2O and 1.05 g of FeCl24H2O were dissolved in 300-mL
deoxygenated water and then vigorously mixed at 80  C under the
nitrogen flow, followed by adding the 80 mL of ammonia (25% wt)
dropwise to the solution within 30 min. The colloidal magnetite
nanoparticles were separated by a neodymium magnet (4
4
6
cm) and rinsed with 100 mL of deoxygenized water. Magnetite
nanoparticles (0.3 g) and NAT-AC were dispersed in 500 mL of
deoxygenated water in different mass ratios of magnetite to AC
(1:2, 1:4 and 1:8), followed by pH adjustment to 4 and stirring
under N2 flow at room temperature for 1 h. The magnetic
nanocomposite was easily separated by a magnet, dried at 50  C
overnight, and finally, dried at 110  C for 4 h. In order to remove the
iron ions adsorbed on the surface of nanocomposite, the MAC was
washed with 500 mL of 0.2 mol L1 HCl, followed by washing with
deionized water and drying at room temperature.
2.4. Analytical methods
CBZ concentration was determined by HPLC (Agilent 1100)
equipped with a C-18 column (5 mm, 4.6
250 mm) and a UV
detector at the wavelength of 286 nm in the isocratic mode. A
mixture of acetonitrile/water (60:40 v/v) was used as the mobile
phase and was delivered at 1.0 mL min1. Samples were filtered on
the PTEF filter (0.2 mm) and then were injected through a 20 mL-
injector loop. Chemical oxygen demand (COD) was determined
using a UV/VIS spectrophotometer (HACH, DR 5000, USA)
according to standard method 5220 D (Closed Reflux, Colorimetric
Method) [24]. Total suspended solids (TDS) were measured
according to standard methods 2540 D [24]. pH measurements
were performed using a pH meter (Metrohm 691, Switzerland).
Table 1
The analysis of wastewater effluent*.
Parameter Unit Quantity
pH – 6.65
TDS mg L1 693
EC mg L1 989
COD mg L1 29
*Taken from the effluent of secondary sedimentation tank of Ekbatan wastewater
treatment plant (Tehran, Iran).
 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321 3311

2.5. Characterization of adsorbent investigated using Analysis of Variance (ANOVA) and the coeffi-
cient of R2. The experimental design matrix and the predicted
The X-ray diffraction (XRD) pattern of Fe3O4 nanoparticles was results are presented in Table 3. Software package of Design-Expert
recorded on an X-ray diffractometer using Cu Ka radiation. (X'Pert 10 trial version (Minneapolis, USA) was employed to design the
PRO MPD PANalytical Company). The magnetization measure- experiments and to analyze the obtained data.
ments for magnetite nanoparticles were carried out using a
vibration sample magnetometer (VSM) with an applied magnetic 2.7. Error analysis
field of 10 KOe. FTIR spectra of PAC, MAC and NAT-AC were
recorded on a FT-IR spectrometer (Bruker, Tensor 27, Germany) The residual root-mean-square error (RMSE), average absolute
with KBr pellets in the range of 4000–400 cm1. Surface area relative error (AARE) and the chi-square test (x2), were used to
analysis was performed by surface area analyzer (PHS1020, China) evaluate the agreement between the experimental data and
using nitrogen adsorption–desorption method. In order to remove predictions using models. The related equations are as follow:
moisture, the samples were purged with N2 gas at 150  C for 2 h. vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u n  2
The morphological studies were conducted by field emission u 1 X
RMES ¼ t qe;exp  qe;cal ð5Þ
scanning electron microscope (FESEM, Hitachi, S4160, Japan). n  p i¼1 i

2.6. Experimental design and data analysis

" #
A central composite design (CCD) was used to determine the 1X N
ðqe;cal  qe;exp Þ
AARE ¼ j j ð6Þ
optimal conditions for the critical factors. Preliminary experiments N i¼1 qe;exp
i
were carried out to select the effective parameters on the removal
efficiency of CBZ by MAC. In this study, Adsorbent dosage, CBZ
2 2 3
concentration and contact time were selected as the independent
X 6 qe;exp  qe;cal 7
n
variables and labeled as A, B and C, respectively. The levels of the x ¼ 4
2
5 ð7Þ
selected variables are presented in Table 2. i¼1
qe;cal
i
The real values of the independent variables (Xi) were coded to
zi according to Eq. (1) where qe,exp is the experimental results, qe,cal is the calculated
values using the model, n is the number of experimental data, and
Xi  Xo
zi ¼ ð1Þ p is the number of model parameters. RMSE and AARE should be as
DX i
small as possible. When more than one model is acceptable
where zi is the dimensionless value of an independent variable, Xi statistically then the chi-square (x2) test is required to find the
represents the real value of the independent variable, Xo is the real best-fit model. The lowest value indicates the best model [26].
value of the independent variable at the center point, and DXi is the
step change [25]. 3. Results and discussion
The number of experiments is calculated using Eq. (2), where k
and n are the number of factors and center runs, respectively (k = 3, 3.1. Characterization of MAC
n = 6).
The XRD patterns of the Fe3O4 nanoparticles and MAC 1:8 are
2k ðfactorial pointsÞ þ 2kðaxial pointÞ þ nðcenter pointsÞ ð2Þ presented in Fig. 1(a) and (b), respectively. The diffraction peaks at
As a consequence, a total of 20 experiments were carried out, 30.2 (220), 35.5 (311), 43.2 (400), 53.5 (422), 57.2 (511), 62.7
consisting of 8 factorial points, 6 axial points and 6 replicates. The (440) can be related to the standard XRD pattern of cubic
value of a was calculated by the following equation: magnetite and maghemite. The magnetization curves of Fe3O4

a ¼ 24
k
ð3Þ Table 3
RSM design and the observed and predicted values.
Therefore, a is equal to (2) = 1.68 according to (Eq. (3)). The
3/4

mathematical relationship between three independent variables Run Factors Removal efficiency (%)
and the removal efficiency can be described by the following CBZ(mg L1) MAC (mg L1) Time (min) Actual Predicted
quadratic equation: A B C
1 2.00 175.0 5.50 82.7 83.8
%Removal ¼ bo þ b1 A þ b2 B þ b3 C þ b11 A2 þ b22 B2 þ b33 C 2 2 11.00 175.0 5.50 65.0 67.0
þ b12 AB þ b13 AC þ b23 BC ð4Þ 3 11.00 300.0 5.50 84.0 82.9
4 11.00 175.0 5.50 65.0 67.0
where, bo is the constant coefficient, b1 , b2 and b3 are the linear 5 11.00 50.0 5.50 33.9 36.5
coefficients, b11 , b22 and b33 are the quadratic coefficients, b12 , b13 6 16.35 100.7 2.82 44.4 43.1
and b23 are the interaction coefficients and A, B and C are the 7 16.35 249.3 8.18 75.1 76.8
8 11.00 175.0 1.00 58.4 60.3
values of the variables. The reliability of the fitted model was 9 11.00 175.0 5.50 69.3 67.0
10 11.00 175.0 5.50 66.6 67.0
11 11.00 175.0 5.50 67.0 67.0
Table 2
12 16.35 249.3 2.82 72.3 71.3
Experimental range and levels of the independent variables.
13 5.65 100.7 8.18 66.2 66.1
Variables Factor Unit Levels 14 5.65 249.3 2.82 78.2 78.2
15 5.65 100.7 2.82 58.5 55.7
a 1 0 +1 +a 16 5.65 249.3 8.18 92.9 93.1
CBZ A mg L 1
2.00 5.65 11.00 16.35 20.00 17 16.35 100.7 8.18 45.2 44.1
MAC B mg L1 50.0 100.0 11.7 249.3 300 18 11.00 175.0 5.50 68.3 67.0
Time C min 1.00 2.82 5.50 8.18 10.00 19 11.00 175.0 10.00 75.0 73.6
20 20.00 175.0 5.50 58.9 59.4
3312 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321
Fig. 1. XRD spectrum of the magnetite nanoparticles (a) and MAC 1:8 (b); Magnetization curve of the magnetite nanoparticles(c) and magnetic nanocomposites (d).
nanoparticles and nanocomposites with different mass ratios are
shown in Fig. 1(c) and (d), respectively. The saturation magnet-
izations of Fe3O4 nanoparticles and MAC with the mass ratio of 1:8
(Magnetite: AC) were found to be 60 and 5.06 emu g1, respective-
ly, showing high enough magnetization to be separated easily by
an external magnetic field. As it is shown in Fig. 1(d), the saturation
magnetization increased as the magnetite content of nano-
composite increased. No hysteresis loop was observed for MAC
due to presence of single-domain magnetic nanoparticles on the
surface of AC.
The microstructures of AC, NAT-AC, MAC 1:8 and Fe3O4
nanoparticles were investigated using SEM (Fig. 2). It is clear
from Fig. 2(a) and (b) that the surface of NAT-AC is rougher than
that of AC (Fig. 2(c) and (d)). On the other hand, the large
macropores were observed in NAT-AC due to acid treatment. The
average particle size of Fe3O4 nanoparticles was less than 50 nm
(Fig. 2(e)). The magnetic nanoparticles were observed on the
surface of AC (Fig. 2(f)). In order to test the stability of magnetic
nanoparticles immobilized on the surface of AC, an aqueous
suspension of magnetic nanocomposite was subjected to ultra-
sonic waves for 5 min and then placed on the magnet. The obtained
clear solution after sedimentation of nanocomposite particles
confirmed the strong bonding of nanoparticles to the surface of AC.
The specific surface area of AC and MAC 1:8 were determined using
BET. Results showed that the specific surface area of MAC with
mass ratio of 1:8 (1241 m2 g1) was slightly lower than that of AC
(1378 m2 g1) which may be due to blockage of porous structure of
AC by oxygen-containing functional groups and magnetite nano-
particles.
The FT-IR spectrum of AC, NAT-AC and MAC are presented in
Fig. 3. As can be seen, FT-IR spectrum of MAC contains a new
absorbance band at 640 cm1 which corresponds to Fe–O bond,
confirming the presence of magnetite nanoparticles on the surface
of AC [27]. The stretching vibration of  OH group was found at
3440 cm1 [28]. The C¼O stretching vibration appears at
1680 cm1, indicating the presence of carboxyl acid groups in
the structure of AC. The band observed at 1440 cm1 can be related
to carboxyl-carbonates structures [29]. The band appeared at
1100 cm1 is as a result of stretching vibration of C
O bond [30]. It
is clear from comparison of FT-IR spectrum of NAT-AC with that of
AC that NAT-AC exhibited stronger absorbance bands at 3440 and
1690 cm1 which can be due to higher carboxylic content. On the
other hand, the MAC showed stronger absorbance band at 3440 nm
than NAT-AC which may be as a result of  OH group of magnetite
nanoparticles adsorbed on the surface of AC.
3.2. Effect of nitric acid treatment and magnetite content on the
adsorption capacity
Adsorption capacities of AC, NAT-AC and magnetic nano-
composites with different mass ratios were compared in Fig. 4. As
can be seen in Fig. 4(a), the adsorption capacity of AC was higher
 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321 3313

Fig. 2. SEM images of (a) NAT-AC
40.0 K, (b) NAT-AC
1.00 K (c) AC
40.0 K, (d) AC
1.00 K, (e) Fe3O4 nanoparticles
20.0 K (f) MAC 1:8
40.0 K.

(275 mg L1) than that of NAT-AC (210 mg L1). Nitric acid
treatment can lead to destruction of porous structure and change
in surface structure of AC. On the other hand, carboxylic functional
groups which were incorporated in AC matrix as a result of nitric
acid treatment can increase the hydrophilicity of AC and
subsequently decrease in affinity of hydrophobic pollutants toward
the adsorbent.
It is clear from Fig. 4(b) that the adsorption capacity of
nanocomposite was decreased by increasing its magnetite content.
It may be due to blockage of some pores of NAT-AC or increase in
inactive mass of adsorbent (Fe3O4). In order to find the reason of
decrease in adsorption capacity, the adsorption capacities were
expressed based on mass of carbon matrix of nanocomposite and
results were shown in Fig. 4(a). Increase of magnetite content up to
mass ratio of 1:8, had no adverse effect on the corrected adsorption
capacity of NAT-AC. Therefore the reduction in adsorption capacity
is due to increase in inactive mass of adsorbent which was mainly
loaded on the outer surface of NAT-AC. As the mass ratio increased
up to 1:4, the corrected adsorption capacity decreased as much as
14.6% which is due to penetration of nanoparticles in pores of NAT-
AC and blockage of porous structure. The much increase in the
magnetite content of nanocomposite had no effect on the corrected
adsorption capacity of adsorbent. The nanocomposite with the
mass ratio of 1:8 had enough magnetization characteristics that
could be separated easily by a magnet after treatment process.
Therefore, it was chosen as a suitable adsorbent for removal of CBZ
from wastewater effluent.
3.3. Statistical analysis
The following quadratic model was obtained using CCD which
correlates the removal efficiency (Y) as a function of initial
3314 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321

concentration of CBZ (A), adsorbent dosage (B) and contact time

(C).

Y ¼ 37:16  2:34
A þ 0:2764
B þ 1:99
C þ 0:00358

AB  0:1641
AC þ 0:00566
BC þ 0:0574
A2
 0:00046B2 þ 0:0272C 2 ð8Þ

The analysis of the variance (ANOVA) was applied to investigate

the validation of models and significance of variables (Table 4). The
terms with p-value < 0.05 were considered statistically signifi-
cance (A, B, C, AC, A2 and B2). The significant degree of interactive
factors followed the sequence: AC > AB > BC. The model F-value of
94.58 showed that the model was significant. The p-value of lack of
fit (LOF) was not significant relative to the pure error, suggesting
that the model fitted well the experimental data. The predicted R2
was in reasonable agreement with the adjusted R2, which confirm
the accuracy of the model. The experimental and predicted values
for removal efficiency (%) of CBZ are presented in Table 3 which
indicates a good agreement between the predicted removal
efficiencies by the model and the experimental data. Therefore,
Fig. 3. FTIR spectra for AC, NAT-AC and MAC.

Fig. 4. (a) Adsorption capacities of AC, NAT-AC and magnetic nanocomposites with different mass ratios based on mass of carbon matrix, (b) Adsorption capacities of
magnetic nanocomposites with different mass ratios based on mass of nonocomposite.CBZ: 20 mg L1, Adsorbent 50 mg L1, Contact time: 30 min.

Table 4
ANOVA for quadratic model.

Factor SSa dFb MSc F-value p-value

Model 3746.06 9 416.23 94.58 <0.0001
A-CBZ concentration 715.15 1 715.15 162.51 <0.0001
B- MAC dosage 2600.62 1 2600.62 590.96 <0.0001
C- Contact time 212.87 1 212.87 48.37 <0.0001
AB 16.25 1 16.25 3.69 0.0836
AC 44.18 1 44.18 10.04 0.0100
BC 10.13 1 10.13 2.30 0.1603
A2 38.95 1 38.95 8.85 0.0139
B2 93.39 1 93.39 21.22 0.0010
C2 0.54 1 0.54 0.12 0.7323
Residual 44.01 10 4.40
Lack of Fit 28.97 5 5.79 1.93 0.2444
Pure Error 15.03 5 3.01
Cor Total 3790.07 19

R2 = 0.9882, Adj R2 = 0.9797, Pred R2 = 0.9443.

a
Sum of square.
b
Degree of freedom.
c
Mean square.
Fig. 5. Perturbation plots for CBZ removal by MAC.
 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321 3315

the proposed model can be applied to predict the removal 3.5. Effect of independent variables and their interactions on the
efficiency of CBZ using MAC. removal efficiency

3.4. Comparative effects of variables on the removal efficiency
By using the perturbation plot, the effect of all factors on the
removal efficiency can be seen and compared at a point in the
design space on the response. It was clear from the perturbation
plot presented in Fig. 5, that all factors have significant effect on the
removal efficiency. The removal efficiency was increased by
increasing the contact time and adsorbent dosage. However, a
decrease in removal efficiency was observed as the concentration
of CBZ increased. As can be seen, the adsorbent dosage and CBZ
concentration were more effective on the removal efficiency than
contact time.
Fig. 6 shows the response surface plots for CBZ removals
resulting from Eq. (8). The response were plotted as a function of
two factors varying in the factorial part of the design (1  zi  + 1),
while setting the time and MAC to factorial point of +1 and CBZ to
factorial point of 1. The effect of MAC dosage and initial
concentration of CBZ on the removal efficiency of CBZ is shown
in Fig. 6(a). The removal efficiency increased with increasing the
adsorbent dosage and increasing the CBZ concentration. A
maximum removal efficiency of 92% was obtained at the adsorbent
dosage of 249.3 mg L1 and CBZ concentration of 5.65 mg L1. It is
clear from Fig. 6(b) that there is a significant interaction between
time and CBZ concentration. At the high concentration of CBZ, the

Fig. 6. The 3D response surface plots for interactive effect of (a) Adsorbent dosage and CBZ concentration; Time at the level of +1, (b) CBZ concentration and time; Adsorbent
dosage at the level of +1 (c) adsorbent dosage and time; CBZ concentration at the level of 1.
3316 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321

Table 5 adsorb in the initial time of mixing. Although the adsorption rate is
The composition of stock solution of TDS with concentration of 1000 mg L1.
high, however the removal efficiency is low due to small amount of
Cation mg L1 Anion mg L1 active sites available.
Na+ 125.7 HCO3 339.2 The optimal values of the independent variables for removal of
K+ 24.7 Cl 82.0 CBZ from the municipal wastewater effluent were determined
Ca2+ 203.4 NO3 39.5 from response surface analysis. The maximum removal efficiency
Mg2+ SO42
50.6 135.1
of 93% was obtained at the initial CBZ concentration of 5.88 mg L1,
adsorbent dosage of 248.5 mg L1 and contact time of 8.17 min.
After optimization of the response, confirming experiment was
carried out at the optimum point to evaluate the validity of model.
The actual value (95.2%) located within the 95%- prediction
intervals (88.99–97.46%), confirming that the model can well
predict the corresponding responses.

3.6. Effect of TDS

The effect of TDS on the adsorption of CBZ by MAC was studied

Fig. 7. Effect of TDS on the adsorbtion capacity of MAC 1:8. Adsorbent: 50 mg L1,
CBZ: 20 mg L1, Temperature: 298 K, Contact time: 30 min.
adsorption takes place fast due to high driving force of CBZ towards
the active sites of adsorbent. However, at the low concentration of
CBZ the removal efficiency is much affected by contact time. The
effect of MAC dosage and contact time on the removal efficiency of
CBZ is shown in Fig. 6(c). By increasing the CBZ/MAC ratio, the
driving force increases. As a consequence, the most of CBZ can
by measuring the adsorbed CBZ from the solutions containing
different amount of dissolved solids in the range of 0–1000 mg L1.
A stock solution of TDS with concentration of 1000 mg L1 was
prepared by dissolving the appropriate amounts of related solids to
obtain the composition presented in Table 5.
As can be seen in Fig. 7, by increasing TDS, adsorption
performance decreased slightly which can be due to increase in
water adsorption on the surface of AC. AC can adsorb carbamaze-
pine through donor-acceptor complex formation between surface
carbonyl groups and aromatic rings of carbamazepine. Water
cluster formation on carboxylic groups is suspected to block
adsorption on the carbonyl sites. Water adsorption is expected to
increase with increasing salt concentration because of ion
exchange between H+ and alkali ions at the surface functional
groups (carboxylic, phenolic and hydroxylic) [31]. The removal
efficiency of CBZ from wastewater effluent (41.7% at CBZ: 10 mg L1,
TDS: 693 mg L1) is lower than that from deionized water (61.1% at
CBZ: 10 mg L1, TDS: 0.0 mg L1) which may be attributed to
interference effect of TDS content of effluent wastewater.

Table 6
Isotherm parameters obtained with linear regression and error values for adsorption of CBZ onto MAC 1:8 at 298 K.

Isotherm Equation Parameters

Langmuir Ce
¼
1
þ
Ce
ð9Þ qm: mg L1 182.9
qe K L qm qm

K L : L mg1 0.743
Freundlich 1 1=nF 0.3121
lnqe ¼ lnK F þ lnC e ð10Þ
nF
K F : (mg g1)(L mg1)1/n 74.10
Temkin qe ¼ B1 lnA þ B1 lnC e ð11Þ B1 : mg g1 28.74
A : L mg1 21.00
Dubinin-Radushkevich lnqe ¼ lnqs  be2 ð12Þ qs : mg g1 144.11
 
e ¼ RTln 1 þ
1
ð13Þ b: mol2 J2 5.56
108
Ce
1
E ¼ pffiffiffiffiffiffiffi ð14Þ E : kJ mol1 3.00
2b
 
Redlich-Peterson Ce g 0.8212
ln K R  1 ¼ glnC e þ lnaR ð15Þ
qe

KR: L g1 450

aR: (L mg1)g 4.19
Radke-Prausnitz Ce 1 Cm
e
m: 0.8328
¼ þ ð16Þ
qe K RP qm qm

K RP : L mg1 4.56
qm : mg g1 104.4

qm : maximum adsorption capacity reflected a complete monolayer (mg g1), K L : Langmuir constant or adsorption equilibrium constant (L mg1), 1=n: isotherm
constant indicates the empirical parameter, K F : isotherm constant indicates the capacity parameter (mg g1)(L mg1)1/n, KR: (L g1), aR (L mg1) and g: Redlich-
Peterson constants, qs : adsorption capacity (mg g1), b: constant of Dubinin-Radushkevich (mol2 J2), E : related to free energy (kJ mol1), K RP : Radke-Prausnitz
constant (L mg1), R: 8.314 gas constant (J mol1 K1), T: absolute temperature (K).
 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321 3317

Table 7 models [35] can be applied either in homogeneous or heteroge-

Performance of different isotherms for adsorption of CBZ onto MAC 1:8 at 298 K.
neous systems [36]. Redlich-Peterson model approaches the
Models R2 RMSE AARE x2 Langmuir and Freundlich models at low and high concentrations,
Langmuir 0.9963 11.77 0.1141 21.45 respectively. Since, these models contain three parameters; a trial
Freundlich 0.9778 10.99 0.0689 4.44 and error procedure has to be used to find the best fit to
Redlich-Peterson 0.9994 3.80 0.0251 0.57 experimental data using maximization the coefficient of determi-
Dubinin-Radushkevich 0.8629 28.02 0.1659 29.48
nation (R2) and minimization of the sum of squared errors.
Temkin 0.9943 4.28 0.0377 1.10
Radke-Prausnitz 0.9996 3.48 0.0219 0.44
Adsorption isotherm was investigated at constant dose of
adsorbent (50 mg L1) at temperature of 298 K. The samples were
stirred for 30 min and after separation of adsorbent, the
3.7. Equilibrium studies concentration of carbamazepine was determined. The equilibrium
adsorption of CBZ by MAC was investigated through mentioned
The adsorption isotherm indicates the relation between the isotherms and the results are presented in Table 6. The
equilibrium concentration of pollutant in solution (Ce) and in performance of different models was evaluated by different error
adsorbent (qe). Various isotherm equations such as Langmuir, functions and the results are shown in Table 7. According to RMSE,
Freundlich, Dubinin–Radushkevich, Redlich-Peterson, Radke- AARE and x2 values, Radke-Prausnitz, Redlich-Peterson and
Prausnitz and Temkin were used to describe the equilibrium Temkin models exhibited the lower errors than the others for
characteristics of adsorption process. The linear forms of these MAC. The fitting degree followed the sequence:
models are presented in Table 6. The Langmuir model considers the Radke-Prausnitz > Redlich-Peterson > Temkin > Freundlich >
monolayer adsorption onto identical active sites of adsorbent Langmuir > Dubinin-Radushkevich.
without lateral interaction and steric hindrance between the The experimental and corresponding calculated data using
adsorbed molecules [32]. The Langmuir model is able to estimate adsorption models are shown in Fig. 8, which confirm the results of
the maximum adsorption capacity of adsorbent corresponding to error analysis.
monolayer adsorption on the surface of adsorbent. However, The value of 1/nF was found to be 0.31. When the value of 1/nF
Freundlich isotherm is based on multi-layer adsorption process goes toward zero, the surface of adsorbent is more heterogeneous
[33]. Temkin isotherm assumes that the heat of adsorption of all [37]. The equilibrium adsorption constant is described by
the molecules in the layer decreases linearly with coverage due to Langmuir constant (KL, L mg 1) which is as high as 0.743 L mg1.
adsorbent–adsorbate interaction. The adsorption is characterized The greater Langmuir constant, the higher affinity of adsorbate
by a uniform distribution of binding energies [34]. According to towards adsorbent. Although Radke-Prausnitz model can predict
Dubinin-Radushkevich isotherm, the adsorption occurs on a the results well, it is not successful in estimation of maximum
heterogeneous surface with steric hindrance between adsorbed adsorption capacity. Therefore, the maximum capacity of Radke-
and incoming particles. Redlich-Peterson and Radke-Prausnitz Prausnitz (104.4 mg g1) was not in agreement with the

Fig. 8. The equilibrium adsorption isotherms for adsorption of CBZ onto MAC 1:8 (a), Zoomed section of equilibrium adsorption isotherms at low concentrations (b) Residual
plots of investigated models (c). Adsorbent: 50 mg L1, Contact time: 30 min, Temperature: 298 K.
3318 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321

Table 8
Comparison of monolayer equilibrium capacity for CBZ onto other different Adsorbents.

Adsorbent qm (mg g1) Reference

Regenerable granular carbon nanotubes 87.2 mg g1 (369.5 mmol g1) [11]
Amberlite XAD-7 acrylic-ester-resin 97 [38]
Molecularly imprinted polymer 86 [39]
single-walled carbon nanotubes (SWCNTs) 130 [15]
Mesoporous silica SBA-15 0.16 [40]
Agricultural waste rice straw 40 [41]
AC 274 Present study
MAC 1:2 105 Present study
MAC 1:4 125 Present study
MAC 1:8 182.9 Present study

Table 9
e;exp : 121:6; qe;exp : 120:7; qe;exp : 116:0.
Adsorption kinetic parameters calculated by various kinetic models for adsorption of CBZ onto MAC 1:8 at different temperatures, q273K 288K 303K

Model Parameters Temperature (K)

273 288 303

1
Pseudo-first-order k1 : min 0.0544 0.0829 0.0970
k1 t
logðqe  qt Þ ¼ logðqe Þ  2:303 (17) qe (calc): mg g1 27.08 20.19 15.30
Pseudo-second-order k2: g mg1 min1 0.0087 0.0147 0.0216
 
qe (calc): mg g1 122.2 121.5 116.8
q ¼ k q2 þ q t(18)
t 1 1
t 2 e e

Intraparticle diffusion kdif : mg g1 min1/2 4.79 2.86 2.5

qt ¼ kdif t þ C(19)
1=2 C: 88.76 101.34 99.70
Liquid film diffusion kf d : min1 0.0544 0.0829 0.0970
lnð1  FÞ ¼ kf d t(20)
Elovich b: mg g1 min1 0.1137 0.1999 0.2207
qt ¼ b1 lnðabÞ þ b1 lnðtÞ(21) a: g mg1 1.75
105 2.88
109 1.46
1010

k1 : Rate constant of pseudo-first order adsorption (min1), k2 : second-order rate constant of adsorption (g mg1 min1), qe (calc): equilibrium capacity (mg g1),
kdif : rate constant of intraparticle diffusion (mg g1min1/2), C: intercept of intraparticle diffusion, kf d :Liquid film diffusion rate coefficients (min1), a :
(mg g1 min1) and b : (g mg1) are the initial adsorption rate of the Elovich equation and the desorption constant related to the extent of surface coverage and
activation energy constant for chemisorption.

experimental data (182.9 mg g1). The Langmuir model can predict The adsorbent used in this study showed excellent adsorbent
well the high experimental results, as a consequence the predicted capacity. As it can be observed, the adsorption capacity of MAC 1:8
maximum capacity using Langmuir model was in accordance to (182.9 mg g1) is higher than that of other adsorbents.
experimental result. The obtained maximum capacity of MAC
using Langmuir model was 182.9 mg g1. Maximum adsorption 3.8. Kinetic studies
capacities of various adsorbents for CBZ were presented in Table 8.
Kinetics studies were carried out at different temperatures
Table 10 (273, 288 and 303 K) using pseudo-first-order, pseudo-second-
Performance of different kinetic model for adsorption of CBZ onto MAC 1:8 at order, intraparticle diffusion, liquid film diffusion and Elovich
different temperatures. models. The linear forms of investigated models and the obtained
Model Error Function Temperature (K) results are presented in Table 9. To determine the adsorption
kinetics, 100 mL of sample solution containing the MAC (50 mg L1)
273 288 303
and CBZ (20 mg L1) were stirred for various time intervals. The
Pseudo-first-order R2 0.9319 0.9657 0.9290
reliability of the kinetic models was determined by measuring the
RMSE 94.76 114.17 113.7
AARE 0.7801 0.9001 0.9180 residual root-mean-square error (RMSE), average absolute relative
x2 2723 15728 17823 error (AARE) and chi-square test (x2) and the results are shown in
Pseudo-second-order R2 0.9995 0.9999 1.0000 Table 10. Elovich, intraparticle diffusion and pseudo-second-order
RMSE 8.75 9.85 6.16 models exhibited good fit to experimental data and the fitting
AARE 0.0545 0.042 0.026
x2 7.64 8.54 3.13
degree followed the sequence: Elovich > intraparticle diffusion >
Intraparticle diffusion R2 0.8079 0.9105 0.8360
RMSE 6.005 2.30 2.85
AARE 0.0417 0.0159 0.0215 Table 11
x2 2.56 0.3327 0.5217 Thermodynamic parameters for adsorption of CBZ onto MAC 1:8 at initial CBZ
Liquid film diffusion R2 0.9319 0.9657 0.9290 concentration of 20 mg L1.
RMSE 63.92 62.53 58.43
AARE 0.4799 0.4029 0.3777 T (K) ln Ke DG (kJ mol1) DH (kJ mol1) DS (J mol1 K1)
x2 1911 1555 1256 274 2.62 5.98 2.80 12.45
Elovich R2 0.9499 0.9739 0.9586 288 2.67 6.40
RMSE 3.07 1.24 1.43 298 2.62 6.49
AARE 0.0223 0.0083 0.097 313 2.58 6.71
x2 0.723 0.1011 0.1321
 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321 3319

Table 12
Kinetic parameters for sedimentation of MAC 1:8 and PAC.

Kinetic model MAC 1:8 PAC

ka
R 2
RMSE AARE x2 Ka R2 RMSE AARE x2
Zero-Order 32.15 0.6861 12.62 0.6295 5027 43.51 0.9861 3.494 0.075 2.06
First-Order 0.147 0.8975 6.68 0.0683 4.20 0.0661 0.9790 2.13 0.040 0.70
Second-Order 0.0089 0.9984 0.90 0.0095 0.08 0.00011 0.9232 5.43 3.45 4.4
a
The units of sedimentation rate constants are NTU min–1, min–1 and NTU–1, min–1 for the zero-order, first-order and second-order models, respectively.

pseudo-second-order > liquid film diffusion > pseudo-first-order. the exothermic nature of the adsorption process. The positive value
Furthermore, the calculated adsorption capacities at different of DS was as a result of increasing in randomness at solid/solution
temperatures using pseudo-second-order model were close to the interface during adsorption. The spontaneous nature of adsorption
related experimental values. process was found from the negative value of DG .
The values of the pseudo-second-order rate constant, k2, found
to be 8.70
103, 1.47
102 and 2.16
102 g mg1 min1 for an 3.10. Sedimentation kinetic of adsorbent
initial concentration of 20 mg L1 at the temperatures 273, 288 and
303 K, respectively. As a consequence, the adsorption is faster at The batch experiments of sedimentation were performed
the higher temperatures. If the adsorption process follows the through placing a well dispersed suspension of MAC 1:8 and
intra-particle diffusion process, a plot of qt against t1/2 should be a PAC (200 mg L1) on the neodymium magnet (5
5
4 cm) and
straight line. The best fit straight lines did not pass through the turbidity were determined at different time intervals. The results
origin, indicating that the intra-particle diffusion is not the only (Fig. 9) showed that the sedimentation rate of MAC is considerable
rate determining step. However, other processes may also affect higher than that of PAC. The kinetic of sedimentation was analyzed
the rate of adsorption [42]. using the zero-order, first-order and second-order models and the
results are shown in Table 12. The corresponding values of x2and
3.9. Thermodynamic study RMSE confirm that the sedimentation process of MAC follows the
second order kinetic model. As can be seen in Fig. 9, the
The thermodynamic parameters, including the standard sedimentation rate of PAC does not follow any of the investigated
1 1
enthalpy (DHo ; Jmol ), standard free energy (DGo ; Jmol ) and models, because the PAC contains the particles with different sizes
1 and subsequently different sedimentation rates. However, the
standard entropy (DS ; Jmol o
K 1
) were obtained from the
fitting degree followed the sequence: first order > zero order >
following equation:
second order. The obtained sedimentation results are in accor-
qe dance to the results reported for sedimentation of magnetic sludge
Ke ¼ ð22Þ
Ce in ISSMI process [43]. The rate constants (k1 and k2) of
sedimentation process for MAC 1:8 are significantly higher than
that of PAC. On the other hand, the use of MAC instead of PAC can
DGo ¼ RTlnðK e Þ ð23Þ lead to significant decrease in sludge volume as a result of
magnetic force.

 DH o DS o
lnðK e Þ ¼ þ ð24Þ
RT R
where R is the universal gas constant (8.314 J mol1 K1), T is the
absolute temperature (K) and Ke is the thermodynamic equilibri-
um constant of the adsorption. The calculated values of DH , DS
and DG are given in Table 11. The negative value of DH showed
Fig. 9. Sedimentation curves of MAC 1:8 and PAC, Sample volume: 1L, MAC:
200 mg L1, PAC: 200 mg L1.
4. Conclusion
In this research, magnetic activated carbon was proposed as an
efficient adsorbent for removal of CBZ from the effluent of
municipal wastewater treatment plant. The effect of magnetite
content of nanocomposite on the adsorption capacity was
investigated. Decrease in adsorption capacity by increasing the
magnetite content up to mass ratio of 1:8 was due to increase in
inactive mass of adsorbent. However, reduction of adsorption
capacity of magnetic nanocomposite with the mass ratio of 1:4 was
as a result of blockage of porous structure and increasing the
inactive mass of adsorbent. The nanocomposite with the mass ratio
of 1:8 had enough magnetization characteristics that could be
separated easily by a magnet after treatment process. Therefore, it
was chosen as a suitable adsorbent for removal of CBZ from urban
wastewater effluent. A full factorial central composite design was
applied to optimize the treatment process. A significant quadratic
model was achieved with a significant interaction between CBZ
concentration and contact time. The maximum removal efficiency
of 93% was obtained within the investigated intervals. By
increasing of TDS; adsorption performance decreased slightly
due to increase in water cluster formation on carboxylic groups
which can block adsorption on the carbonyl sites. The equilibrium
adsorption data were in agreement with the results predicted
using Radke-Prausnitz, Redlich-Peterson and Temkin models. The
3320 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321
sequence of fitting degree followed the sequence: Radke-
Prausnitz > Redlich-Peterson > Temkin > Freundlich > Langmuir >
Dubinin-Radushkevich. The maximum adsorption capacity was
found to be 182.9 mg g1. The adsorption kinetic data were closely
fitted to Elovich, intraparticle diffusion and Pseudo-second-order
models and the fitting degree followed the sequence: Elovich >
intraparticle diffusion > pseudo-second-order > liquid film diffu-
sion > pseudo-first-order. According to thermodynamic param-
eters, the adsorption process was found to be spontaneous and
exothermic in nature. Sedimentation data of MAC 1:8 were well
predicted by second order kinetic model with the rate constant of
8.9
103 NTU 1 min 1.
Acknowledgement
The authors wish to acknowledge the Nanotechnology
Research Center of Graduate Faculty of Environment, University
of Tehran.
References
[1] Y. Luo, W. Guo, H.H. Ngo, L.D. Nghiem, F.I. Hai, J. Zhang, et al., A review on the
occurrence of micropollutants in the aquatic environment and their fate and
removal during wastewater treatment, Sci. Total Environ. 473–474 (2014) 619–
641, doi:http://dx.doi.org/10.1016/j.scitotenv.2013.12.065.
[2] S. Kleywegt, V. Pileggi, P. Yang, C. Hao, X. Zhao, C. Rocks, et al., Pharmaceuticals,
hormones and bisphenol A in untreated source and finished drinking water in
Ontario, Canada—occurrence and treatment efficiency, Sci. Total Environ. 409
(2011) 1481–1488, doi:http://dx.doi.org/10.1016/j.scitotenv.2011.01.010.
[3] B.F. da Silva, A. Jelic, R. López-Serna, A.A. Mozeto, M. Petrovic, D. Barceló,
Occurrence and distribution of pharmaceuticals in surface water, suspended
solids and sediments of the Ebro river basin, Spain, Chemosphere 85 (2011)
1331–1339, doi:http://dx.doi.org/10.1016/j.chemosphere.2011.07.051.
[4] K. Choi, Y. Kim, J. Park, C.K. Park, M. Kim, H.S. Kim, et al., Seasonal variations of
several pharmaceutical residues in surface water and sewage treatment plants
of Han River, Korea, Sci. Total Environ. 405 (2008) 120–128, doi:http://dx.doi.
org/10.1016/j.scitotenv.2008.06.038.
[5] K. Fent, A.A. Weston, D. Caminada, Ecotoxicology of human pharmaceuticals,
Aquat. Toxicol. 76 (2006) 122–159, doi:http://dx.doi.org/10.1016/j.aqua-
tox.2005.09.009.
[6] V. Matamoros, V. Salvadó, Evaluation of a coagulation/flocculation-lamellar
clarifier and filtration-UV-chlorination reactor for removing emerging con-
taminants at full-scale wastewater treatment plants in Spain, J. Environ.
Manage. 117 (2013) 96–102, doi:http://dx.doi.org/10.1016/j.jenv-
man.2012.12.021.
[7] S. Suarez, J.M. Lema, F. Omil, Pre-treatment of hospital wastewater by coag-
ulation-flocculation and flotation, Bioresour. Technol. 100 (2009) 2138–2146,
doi:http://dx.doi.org/10.1016/j.biortech.2008.11.015.
[8] T.E. Doll, F.H. Frimmel, Photocatalytic degradation of carbamazepine, clofibric
acid and iomeprol with P25 and Hombikat UV100 in the presence of natural
organic matter (NOM) and other organic water constituents, Water Res. 39
(2005) 403–411.
[9] D. Gerrity, S. Gamage, J.C. Holady, D.B. Mawhinney, O. Quiñones, R.A. Tren-
holm, et al., Pilot-scale evaluation of ozone and biological activated carbon for
trace organic contaminant mitigation and disinfection, Water Res. 45 (2011)
2155–2165, doi:http://dx.doi.org/10.1016/j.watres.2010.12.031.
[10] Z. Yu, S. Peldszus, P.M. Huck, Adsorption characteristics of selected pharma-
ceuticals and an endocrine disrupting compound-Naproxen, carbamazepine
and nonylphenol-on activated carbon, Water Res. 42 (2008) 2873–2882, doi:
http://dx.doi.org/10.1016/j.watres.2008.02.020.
[11] D. Shan, S. Deng, T. Zhao, G. Yu, J. Winglee, M.R. Wiesner, Preparation of
regenerable granular carbon nanotubes by a simple heating-filtration method
for efficient removal of typical pharmaceuticals, Chem. Eng. J. 294 (2016) 353–
361, doi:http://dx.doi.org/10.1016/j.cej.2016.02.118.
[12] W.A. Cabrera-Lafaurie, F.R. Román, A.J. Hernández-Maldonado, Removal of
salicylic acid and carbamazepine from aqueous solution with Y-zeolites
modified with extraframework transition metal and surfactant cations:
equilibrium and fixed-bed adsorption, J. Environ. Chem. Eng. 2 (2014) 899–
906, doi:http://dx.doi.org/10.1016/j.jece.2014.02.008.
[13] U. von Gunten, Ozonation of drinking water: part I. Oxidation kinetics and
product formation, Water Res. 37 (2003) 1443–1467, doi:http://dx.doi.org/
10.1016/S0043-1354(02)00457-8.
[14] K.-J. Choi, S.-G. Kim, S.-H. Kim, Removal of antibiotics by coagulation and
granular activated carbon filtration, J. Hazard. Mater. 151 (2008) 38–43, doi:
http://dx.doi.org/10.1016/j.jhazmat.2007.05.059.
[15] I. Lerman, Y. Chen, B. Xing, B. Chefetz, Adsorption of carbamazepine by carbon
nanotubes: effects of DOM introduction and competition with phenanthrene
and bisphenol A, Environ. Pollut. 182 (2013) 169–176, doi:http://dx.doi.org/
10.1016/j.envpol.2013.07.010.
[16] N. Cai, P. Larese-Casanova, Sorption of carbamazepine by commercial gra-
phene oxides: a comparative study with granular activated carbon and
multiwalled carbon nanotubes, J. Colloid Interface Sci. 426 (2014) 152–161,
doi:http://dx.doi.org/10.1016/j.jcis.2014.03.038.
[17] N. Suriyanon, P. Punyapalakul, C. Ngamcharussrivichai, Mechanistic study of
diclofenac and carbamazepine adsorption on functionalized silica-based
porous materials, Chem. Eng. J. 214 (2013) 208–218, doi:http://dx.doi.org/
10.1016/j.cej.2012.10.052.
[18] M. Boehler, B. Zwickenpflug, J. Hollender, T. Ternes, A. Joss, H. Siegrist, Removal
of micropollutants in municipal wastewater treatment plants by powder-
activated carbon, Water Sci. Technol. 66 (2012) 2115–2121, doi:http://dx.doi.
org/10.2166/wst.2012.353.
[19] T. Depci, Comparison of activated carbon and iron impregnated activated
carbon derived from Gölbaşı lignite to remove cyanide from water, Chem. Eng.
J. 181–182 (2012) 467–478, doi:http://dx.doi.org/10.1016/j.cej.2011.12.003.
[20] B. Kakavandi, A. Jonidi, R. Rezaei, S. Nasseri, A. Ameri, A. Esrafily, Synthesis and
properties of Fe3O4-activated carbon magnetic nanoparticles for removal of
aniline from aqueous solution: equilibrium, kinetic and thermodynamic
studies, Iran. J. Environ. Health Sci. Eng. 10 (2013) 19, doi:http://dx.doi.org/
10.1186/1735-2746-10-19.
[21] R. Fu, Y. Liu, Z. Lou, Z. Wang, S.A. Baig, X. Xu, Adsorptive removal of Pb(II) by
magnetic activated carbon incorporated with amino groups from aqueous
solutions, J. Taiwan Inst. Chem. Eng. 62 (2016) 247–258, doi:http://dx.doi.org/
10.1016/j.jtice.2016.02.012.
[22] S. Zhang, L. Tao, Y. Zhang, Z. Wang, G. Gou, M. Jiang, et al., The role and
mechanism of K2CO3 and Fe3O4 in the preparation of magnetic peanut shell
based activated carbon, Powder Technol. 295 (2016) 152–160, doi:http://dx.
doi.org/10.1016/j.powtec.2016.03.034.
[23] A. Jafari Kang, M. Baghdadi, A. Pardakhti, Removal of cadmium and lead from
aqueous solutions by magnetic acid-treated activated carbon nanocomposite,
Desalin. Water Treat. 3994 (2015) 1–17, doi:http://dx.doi.org/10.1080/
19443994.2015.1095123.
[24] E.W. Rice, L. Bridgewater, A.P.H. Association, A.W.W. Association, W.E. Fed-
eration, Standard Methods for the Examination of Water and Wastewater,
American Public Health Association, 2012.
[25] E. Radaei, M.R. Alavi Moghaddam, M. Arami, Removal of reactive blue 19 from
aqueous solution by pomegranate residual-based activated carbon: optimi-
zation by response surface methodology, J. Environ. Heal. Sci. Eng. 12 (2014)
65, doi:http://dx.doi.org/10.1186/2052-336x-12-65.
[26] T. Mitra, B. Singha, N. Bar, S.K. Das, Removal of Pb(II) ions from aqueous
solution using water hyacinth root by fixed-bed column and ANN modeling, J.
Hazard. Mater. 273 (2014) 94–103, doi:http://dx.doi.org/10.1016/j.jhaz-
mat.2014.03.025.
[27] D. Mohan, A. Sarswat, V.K. Singh, M. Alexandre-Franco, C.U. Pittman Jr., De-
velopment of magnetic activated carbon from almond shells for trinitrophenol
removal from water, Chem. Eng. J. 172 (2011) 1111–1125, doi:http://dx.doi.org/
10.1016/j.cej.2011.06.054.
[28] M. Abbas, S. Kaddour, M. Trari, Kinetic and equilibrium studies of cobalt
adsorption on apricot stone activated carbon, J. Ind. Eng. Chem. 20 (2014) 745–
751, doi:http://dx.doi.org/10.1016/j.jiec.2013.06.030.
[29] H. Liu, Q. Gao, P. Dai, J. Zhang, C. Zhang, N. Bao, Preparation and characteri-
zation of activated carbon from lotus stalk with guanidine phosphate acti-
vation: sorption of Cd(II), J. Anal. Appl. Pyrolysis 102 (2013) 7–15, doi:http://
dx.doi.org/10.1016/j.jaap.2013.04.010.
[30] H.A. Omar, H. Moloukhia, Use of activated carbon in removal of some radi-
oisotopes from their waste solutions, J. Hazard. Mater. 157 (2008) 242–246,
doi:http://dx.doi.org/10.1016/j.jhazmat.2007.12.108.
[31] H.A. Arafat, M. Franz, N.G. Pinto, Effect of salt on the mechanism of adsorption
of aromatics on activated carbon, Langmuir (1999) 5997–6003.
[32] M. Dog an, M. Alkan, Y. Onganer, Adsorption of methylene blue from aqueous
solution onto perlite, Water. Air. Soil Pollut. 120 (2000) 229–248.
[33] A. Kurniawan, H. Sutiono, N. Indraswati, S. Ismadji, Removal of basic dyes in
binary system by adsorption using rarasaponin-bentonite: revisited of ex-
tended Langmuir model, Chem. Eng. J. 189–190 (2012) 264–274, doi:http://dx.
doi.org/10.1016/j.cej.2012.02.070.
[34] B.H. Hameed, Spent tea leaves: a new non-conventional and low-cost ad-
sorbent for removal of basic dye from aqueous solutions, J. Hazard. Mater. 161
(2009) 753–759, doi:http://dx.doi.org/10.1016/j.jhazmat.2008.04.019.
[35] K.Y. Foo, B.H. Hameed, Insights into the modeling of adsorption isotherm
systems, Chem. Eng. J. 156 (2010) 2–10, doi:http://dx.doi.org/10.1016/j.
cej.2009.09.013.
[36] D. Mohan, K.P. Singh, G. Singh, K. Kumar, Removal of dyes from wastewater
using flyash, a low-cost adsorbent, Ind. Eng. Chem. Res. 41 (2002) 3688–3695.
[37] S.-L. Chan, Y.P. Tan, A.H. Abdullah, S.-T. Ong, Equilibrium, kinetic and ther-
modynamic studies of a new potential biosorbent for the removal of Basic Blue
3 and Congo Red dyes: pineapple (Ananas comosus) plant stem, J. Taiwan Inst.
Chem. Eng. 000 (2016) 1–10, doi:http://dx.doi.org/10.1016/j.jtice.2016.01.010.
[38] J.R. Dom??nguez, T. Gonz??lez, P. Palo, E.M. Cuerda-Correa, Removal of
common pharmaceuticals present in surface waters by Amberlite XAD-7
acrylic-ester-resin: influence of pH and presence of other drugs, Desalination
269 (2011) 231–238, doi:http://dx.doi.org/10.1016/j.desal.2010.10.065.
[39] C. Dai, S.-U. Geissen, Y. Zhang, Y. Zhang, X. Zhou, Performance evaluation and
application of molecularly imprinted polymer for separation of carbamaze-
pine in aqueous solution, J. Hazard. Mater. 184 (2010) 156–163, doi:http://dx.
doi.org/10.1016/j.jhazmat.2010.08.018.
 M. Baghdadi et al. / Journal of Environmental Chemical Engineering 4 (2016) 3309–3321 3321

[40] T.X. Bui, H. Choi, Adsorptive removal of selected pharmaceuticals by meso-
porous silica SBA-15, J. Hazard. Mater. 168 (2009) 602–608, doi:http://dx.doi.
org/10.1016/j.jhazmat.2009.02.072.
[41] Z. Liu, X. Zhou, X. Chen, C. Dai, J. Zhang, Y. Zhang, Biosorption of clofibric acid
and carbamazepine in aqueous solution by agricultural waste rice straw, J.
Environ. Sci. (China). 25 (2013) 2384–2395, doi:http://dx.doi.org/10.1016/
S1001-0742(12)60324-6.
[42] S.D. Khattri, M.K. Singh, Colour Removal from Synthetic Dye Wastewater Using
a Bioadsorbent, Water. Air. Soil Pollut. 120(n.d.) 283–294. doi:10.1023/
A:1005207803041.
[43] S. Moharramzadeh, M. Baghdadi, In situ sludge magnetic impregnation (ISSMI)
as an efficient technology for enhancement of sludge sedimentation: removal
of methylene blue using nitric acid treated graphene oxide as a test process, J.
Environ. Chem. Eng. 4 (2016) 2090–2102.