Environmental Science and Pollution Research (2019) 26:24143–24161

https://doi.org/10.1007/s11356-019-05547-7

RESEARCH ARTICLE

Arsenic removal by copper-impregnated natural mineral tufa part II:

a kinetics and column adsorption study
Krstimir Pantić 1 & Zoran J. Bajić 2 & Zlate S. Veličković 2 & Jovica Z. Nešić 3 & Maja B. Đolić 1 & Nataša Z. Tomić 4 &
Aleksandar D. Marinković 1

Received: 17 February 2019 / Accepted: 21 May 2019 / Published online: 21 June 2019
# Springer-Verlag GmbH Germany, part of Springer Nature 2019

Abstract
This batch and column kinetics study of arsenic removal utilized copper-impregnated natural mineral tufa (T–Cu(A–C)) under
three ranges of particle size. Non-competitive kinetic data fitted by the Weber–Morris model and the single resistance mass
transfer model, i.e., mass transfer coefficient kfa and diffusion coefficient (Deff) determination, defined intra-particle diffusion as
the dominating rate controlling step. Kinetic activation parameters, derived from pseudo-second-order rate constants, showed
low dependence on adsorbent geometry/morphology and porosity, while the diffusivity of the pores was significant to removal
efficacy. The results of competitive arsenic adsorption in a multi-component system of phosphate, chromate, or silicate showed
effective arsenic removal using T–Cu adsorbents. The high adsorption rate—pseudo-second-order constants in the range 0.509–
0.789 g mg−1 min−1 for As(V) and 0.304–0.532 g mg1 min1 for As(III)—justified further application T–Cu(A–C) in a flow
system. The fixed-bed column adsorption data was fitted using empirical Bohart–Adams, Yoon–Nelson, Thomas, and dose–
response models to indicate capacities and breakthrough time dependence on arsenic influent concentration and the flow rate.
Pore surface diffusion modeling (PSDM), following bed-column testing, further determined adsorbent capacities and mass
transport under applied hydraulic loading rates.

Keywords Arsenite . Arsenate . Copper . Column study . Adsorption . Modeling

Introduction occur via all three principal exposures: inhalation, consumption

Considering that arsenic in the environment is omnipresent to a
degree, human exposure to it is inevitable. Potential risk can
Responsible editor: Tito Roberto Cadaval Jr
Electronic supplementary material The online version of this article
(https://doi.org/10.1007/s11356-019-05547-7) contains supplementary
material, which is available to authorized users.
* Maja B. Đolić
mdjolic@tmf.bg.ac.rs
1
Faculty of Technology and Metallurgy, University of Belgrade,
Karnegijeva 4, Belgrade 11120, Serbia
2
Military Academy, 33 General Pavle Jurišić–Šturm St.,
Belgrade, Serbia
3
Military Technical Institute, 1 Ratka Resanovića St., Belgrade 11000,
Serbia
4
Innovation Centre of the Faculty of Technology and Metallurgy in
Belgrade doo, Karnegijeva 4, Belgrade 11120, Serbia
of food and water, and dermal adsorption. However, the most
common and major source of arsenic is through contaminated
ground water. Such contamination has led to the issuing of new
criteria in assessing the quality of drinking water, stipulating that
the maximum permissible concentration (MPC) of arsenic was
reduced from 50 to 10 μg dm−3(Sahiner et al. 2015).
Arsenic in drinking water predominantly exists as arsenite
(As(III)) or arsenate (As(V)) of which the former is more toxic
than the latter; in fact, As(III) is two order of magnitude more
toxic than methylated arsenic compounds (Morrison et al.
1989). The presence of arsenic in groundwater originates from
the dissolution of minerals that naturally occurs in geological
formations or the leaching of metal ores, agricultural pesti-
cides, or wood preservatives. Worldwide, many wells exceed
this minimum stipulated value, especially in Southeast Asia,
North and South America, and some parts of Southeastern
Europe, such as the Balkan Peninsula (Hungary, Romania,
Serbia, Croatia, and Montenegro) (Bhattacharya et al. 2007;
Choong et al. 2007; Habuda-Stanić et al. 2007). Coagulation
and precipitation (Bordoloi et al. 2013; Lakshmanan et al.
24144
2008; Mohora et al. 2012; Pal et al. 2007), nanofiltration and
reverse osmosis (Gholami et al. 2006; Röhricht et al. 2009;
Shih 2005; Xu et al. 2013), ion-exchange, and adsorption
(Baig et al. 2015; Bajić et al. 2016; Fu et al. 2014; Miller
et al. 2000; Mohan and Pittman 2007; Saleh et al. 2011;
Vaiano et al. 2014; Wang et al. 2014) are just a few of the
techniques currently employed to remove arsenic. Due to its
simple separation after treatment and high efficacy of pollut-
ant removal, adsorption is generally considered to be the most
suitable method (Mohan and Pittman 2007).
Extensive efforts have also been made worldwide to devel-
op selective adsorbents for arsenic removal by using biologi-
cal materials, inorganic oxides, activated carbons, polymer
resins, industrial by-products/wastes, as well as soils, clays,
and sands (Mohan and Pittman 2007). Regardless of the re-
sults achieved, there are still many disadvantages of the adsor-
bents that have undergone studies. The design of adsorbents
for the removal of arsenic from real water that naturally incor-
porate safe disposal methods of the spent adsorbent still needs
research and development.
It has been reported (Saleh et al. 2011) that copper oxide
nanoparticles possess a high affinity for the adsorption of both
As(III) and As(V) (inorganic arsenic species—named iAs). The
same study establishes that CuO nanoparticles are able to effec-
tively remove arsenic over a wide pH range and the presence of
naturally occurring ions found in water (silicate and phosphate)
does not hinder the removal process. Moreover, CuO is advan-
tageous compared with similar adsorbents due to the fact that pH
adjustments and/or oxidation of As(III) to As(V) are not required
as a pre-treatment step prior to the arsenic removal. The effec-
tiveness of CuO for arsenic removal may be attributable to its
high point of a zero charge (9.4 ± 0.4) (Vaiano et al. 2014). In
accordance with that, the use of bigar (tufa) as support for mod-
ification with copper-oxide to produce high-performance adsor-
bents is additionally stimulated by their wide availability and low
cost (Bundschuh et al. 2008; International Committee on Natural
Zeolites 2000; Martinson and Reddy 2009). Sorption is the most
used method for wastewater treatment contaminated with arse-
nic. The significant value in overall technology cost relates to
sorbent replacement after saturation, influencing operational cost
of the sorbent exchange/regeneration and the exhausted sorbent
disposal. In comparison with most commercial adsorbents used
for arsenic removal, obtained by deposition/impregnation of
nanoparticles inside/outside a functionalized spherical-like
polymer/inorganic support structure, the application of T–Cu of-
fers many advantageous in a batch system (Bajić et al. 2016).
Despite these positive aspects of adsorption performance of T–
Cu in a batch system (Bajić et al. 2016), the issue of its possible
practical application in a flow system still remains open.
A dynamic packed-bed column system is a valuable tech-
nology applicable for small-scale simultaneous removal of
water pollutants (Agency 2005). Metal (hydr)oxides/oxides
are selective and effective adsorbent used in a flow system
Environ Sci Pollut Res (2019) 26:24143–24161
for arsenic removal at trace contaminant levels (K.
Hristovski et al. 2008a; Mohan and Pittman 2007).
Evaluation of media performance and breakthrough curve
could be performed using different methods of modeling.
The simplified multi-parameter fitting of the column adsorp-
tion data could be performed using Yoon–Nelson (Yoon and
Nelson 1984), Bohart–Adams (Bohart and Adams 1920),
Thomas (Thomas 1944), Clark (Clark 1987), modified
dose–response (MDR) (Song et al. 2011), and Wolborska
model (Thomas 1944). Also, arsenic removal capability could
be analyzed from Freundlich isotherm fitting of adsorption
data and simulation of the breakthrough curve using the pore
surface diffusion model (PSDM) (Mertz et al. 1999). In gen-
eral, results of short bed column testing (SBC) from PSDM
modeling, without pilot scale media testing, provide a good
evaluation of capacity and feasibility/applicability of studied
media (Crittenden et al. 1986; Hand et al. 1997).
A comprehensive study on adsorption performances of ar-
senic removal with the T–Cu adsorbent in a fixed-bed column,
using theoretical and experimental methods, was applied. To
address current environmental challenges, nano-enabled T–
Cu(A–C) hybrid adsorbents were recently synthesized (Bajić
et al. 2016) and their removal performances were assessed in a
batch condition. As a continuation, the main objectives there-
of have been directed toward the evaluation of a performance
of the T–Cu adsorbent in relation to (1) the kinetic aspects of
the adsorption process, (2) activation parameters, (3) evaluat-
ing the limiting step of adsorption and mass transfer coeffi-
cients, (4) modeling arsenic adsorption using empirical
methods, and 5) the breakthrough curves and capacity/
adsorption rate determination.
Materials and methods
Materials
Natural material, known as tufa, was collected in the area of
Temska region, Pirot City, Republic of Serbia. Tufa is a ma-
terial of natural origin that occurs when limestone containing
carbonate is exposed and precipitates in an ambient tempera-
ture water. The tufa composition is mostly formed by CaO
(42.02%), SiO2 (22.57%), Al2O3 (6.70%), Fe2O3 (1.49%),
MgO (2.0%), and others minor components. T–Cu samples
were prepared according to the previously published literature
(Bajić et al. 2016), sieved, and three fractions were obtained:
0–0.02, 0.02–0.25, and 0.25–1.50 mm for the samples of T–
Cu(A–C), respectively.
Equilibrium adsorption experiments
Adsorbents synthesis, i.e., T–Cu(A–C), adsorption experi-
ments, and results were presented in previous paper (Bajić
Environ Sci Pollut Res (2019) 26:24143–24161
et al. 2016). The adsorbent capacity was calculated according
to (Eq. 1):
C i −C f
q¼ V ð1Þ
m following a continuous inflow of the feed solution. In the
where q is adsorption capacity in mg g−1, Ci and Cf are the
initial and final arsenic concentrations in mg L−1 (μg L−1),
respectively, V is volume of the solution in liters, and m is
mass of adsorbent in grams.
Adsorption equilibrium was analyzed using the Freundlich
adsorption isotherm model (Eq. 2).
q ¼ K C e =n
1
ð2Þ
where q is adsorption capacity, K is the Freundlich adsorption
capacity parameter, Ce is the equilibrium concentration of ad-
sorbate in solution, and 1/n is the Freundlich adsorption inten-
sity parameter. Theoretical aspect of the Freundlich isotherm
model is presented in more detail in Supplementary informa-
tion. The pH values at the point of zero charge (pHPZC) of the
samples, i.e., the pH above which the total surface of the
samples is negatively charged, were measured using the pH
drift method (Milonjić et al. 2007).
Kinetic adsorption experiments
As a continuation of previous work, the effect of time in the
batch system was studied by ultrasonic treatment/mixing
method of the appropriate quantity of adsorbent and As(III)
or As(V) solutions. Time-dependent adsorption studies in a
batch system were performed using 100 mg dm−3 of adsorbent
at pH 6.0 ± 0.1, and initial concentrations Ci = 0.10 and
0.5 mg dm−3 of both As(III) and As(V) species. Samples for
analysis were taken at 1, 3, 5, 10, 15, 30, 45, 60, and 90 min,
and filtered through a Millipore 0.22-μm membrane filter
(Bedford, MA, USA) before analysis. Comparative sample
was left for 24 h to confirm extent of adsorption. Three tem-
peratures were selected: 293, 303, and 313 K, and kinetic
experiments was performed at 100, 250, and 400 rpm (ω)
using impeller, made from poly(styrene), with no sharp edges.
The diameter of impeller, dimp, was 0.055 m. Competitive
kinetic study was performed analogously to interference free
model system using 0.1 mg dm−3 phosphate, chromate, and
silicate.
A column study was carried out to evaluate the potential
application of copper-coated tufa for arsenic removal on lab-
oratory and industrial scale. A glass tube of 0.8 × 15 cm (d ×
H) connected to a sintered filter and a
poly(tetrafluoroethylene) (PTFE) valve was packed with T–
Cu(A–C) adsorbents in order to conduct the downflow col-
umn experiment. A layer of sand was introduced before the
addition of adsorption material to the column bed to prevent
an even flow. An adapter was attached to the top of the column
24145
to allow the PTFE tubing to be fitted. Prior to the column
being attached, deionized water was pumped through the tub-
ing to remove any impurities. After the addition of deionized
water, a vacuum was applied to remove trapped air bubbles
course of adsorption experiments, effluents were collected
and analyzed using the ICP-MS technique. The results were
then used for the breakthrough point determination. The flow
rate, Q, the empty bed volume (EBV), pH, and the mass of T–
Cu(A–C) (m) were adjusted to obtain an optimal empty bed
contact time (EBCT). The flow rate of the feed solution was
varied, Q = 1.0–10 cm3 min−1, in order to find the optimal
EBCT value. The fixed-bed column conditions were as fol-
lows: m(T–Cu(A–C)) ≈ 5.5 g and EBV = 6.0 cm3. All experi-
ments were performed at 298 K. The feed water with concen-
tration of Ci = 0.989, 4.986, and 9.985 mg dm−3 for As(III),
and Ci = 0.983, 4.995, and 9.983 mg dm−3 for As(V) were
used in a flow system. The water was passed through the
hybrid adsorbent bed using a peristaltic pump (Ismatec,
Switzerland). The residence time, EBCT, was calculated ac-
cording to EBCT = H/γ, where H is the bed depth (12 cm) and
γ is the linear flow rate (cm min−1).
Characterization method
Transmission electron microscopy (TEM) and high-resolution
TEM (HRTEM) analyses were performed on a Jeol 2100F
(200 kV, Cs-corrected condenser, GIF Tridiem imaging filter)
electron microscope. The arsenic concentrations in the solu-
tions after the adsorption and kinetic experiments were ana-
lyzed by inductively coupled plasma mass spectrometry (ICP-
MS), using an Agilent 7500ce ICP-MS system (Waldbronn,
Germany) equipped with an octopole collision/reaction cell,
Agilent 7500 ICP-MS ChemStation software, a MicroMist
nebulizer, and a Peltier cooled (2.0 °C) quartz Scott-type dou-
ble pass spray chamber. Standard optimization procedures and
criteria specified in the manufacturer’s manual were followed.
ICP-MS detection limit was 0.030 μg L−1 and the relative
standard deviation (RSD) of all arsenic species investigated
was between 1.3–5.1%.
Adsorbent porosity was evaluated by pycnometer analysis
(Sontheimer et al. 1988). Considering cylindrical structure of
the pores, the average pore diameter, rpore, was calculated
from the surface area–pore volume ratio (K. D. Hristovski
et al. 2008; Sontheimer et al. 1988) (Eq. 3):
2 Aad
¼ ð3Þ
rpore V ad
where rpore is the average pore radius (m); Aad is the surface
area of the adsorbent (m2); and Vad is the pore volume of the
adsorbent (m3). The macroporous nature of T–Cu(A–C) was
confirmed by the calculations of rpore using data from previous
24146
study (Bajić et al. 2016). The obtained rpore values for T–
Cu(A–C) of102, 119, and 122 nm, respectively, were lower
than ones calculated from nitrogen adsorption isotherms. Both
results clearly showed macroporous structure of used adsor-
bents (UPAC prescribed rpore > 50 nm as macropore) (Sing
et al. 1985).
Full-scale system modeling using validated pore
surface diffusion model
Time-dependent arsenic uptake was defined from experimen-
tal data as well as breakthrough curve prediction using a pore
surface diffusion model (PSDM), i.e., the AdDesignS™ soft-
ware, developed by Michigan Technological University
(Mertz et al. 1999). After performing batch adsorption test,
and fitting data using the Freundlich isotherm model, determi-
nation of breakthrough point is possible after model verifica-
tion in a short bed column testing (SBC). PSDM is a model
applicative for description of fixed-bed column system ac-
counting both surface and pore diffusion, and giving good
prediction of breakthrough performance. All of these charac-
teristics define possible applicability of adsorbent, i.e., the
feasibility of adsorbent media (Athanasaki et al. 2015;
Crittenden et al. 1986; Hand et al. 1997; Hristovski et al.
2008b; Magnuson and Speth 2005; Yan et al. 2013).
Additional information on applied methodology is given in
Supplementary material.
Response surface methodology used for prediction
of the capacity and breakthrough time determination
Response surface methodology (RSM) method applied for de-
termination of the adsorbent capacity and breakthrough time de-
termination is given in Supplementary material (Table S1). The
use of RSM is in compliance with principles of environmental
protection due to significant reduction of the number of experi-
ments, and therefore waste generation was reduced. Full detail on
adsorbents characterization, description of adsorption and kinetic
experiments, methodology applied for evaluation of the quality
of experimental data fitting, and error functions used (section S2)
are given in Supplementary material.
Results and discussion
Material characterization
The performance of the hybrid adsorbent of copper-
impregnated mineral tufa (T–Cu) indicates their applicability
for efficient arsenic removal (Bajić et al. 2016). The precipi-
tation of the reduced copper species at a level of 4% loaded
copper (≈ 5% calculated as CuO) created a nano-scaled depos-
it that exhibited adsorption performance and adherence to the
Environ Sci Pollut Res (2019) 26:24143–24161
tufa surface. Arsenic removal in a batch system utilizing T–Cu
adsorbents has recently been established (Bajić et al. 2016),
whose applicability in a dynamic system of arsenic removal
has been applied in the study presented here. Increased ad-
sorption capacities in the range of 4.65 to 67.83 mg g−1 for
As(III) and 6.84 to 104.62 mg g−1 for As(V) were found for
T–Cu(A) compared with the unmodified tufa. Similar results
were found for T–Cu(B) and T–Cu(C) (Bajić et al. 2016). The
electrostatic attraction of the negatively charged arsenic spe-
cies and the positive adsorbent surface as well as the specific
arsenate adsorption (i.e., the monodentate/bidentate surface
complexes formation) largely contributed to the overall ad-
sorption process (Markovski et al. 2014). As a continuation
of batch-mode testing (Bajić et al. 2016), a high-resolution
transmission electron microscopy (HRTEM) was employed
to investigate the internal structure of the T–Cu(A) adsorbent
prior to and following arsenic removal, Fig. 1.
According to the diffraction fringes (Fig. 1a), the crystals in
T–Cu(A) material were randomly distributed, most of which
consisted of interlayer spacing d = 0.24 nm where the crystallite
size ≤ 10 nm corresponded to the mixed crystalline structure of
CaO/SiO2/Fe2O3 (Geckeler and Nishide 2009; Geng et al. 2017;
Gorobinskii et al. 2005). Several small grains were also found
containing interlayer spacing d = 0.19 nm of a crystallite size of
~ 3 nm which may indicate the presence of copper oxide (Zhang
et al. 2018). After the arsenic adsorption, the crystal structure was
not able to be easily distinguished, Fig. 1b. The size of the crystal
grains (generally larger than 10 nm) and the interlayer spacing
were increased, which may suggest a crystallization reaction be-
tween the calcium and arsenate (Lei et al. 2018).
Adsorption kinetics and activation parameters
In conjunction with the results of a previous study (Bajić et al.
2016) strongly suggesting the potential practical use of T–
Cu(A–C) adsorbents, a fixed-bed column study was carried
out to consider the possible application of such adsorbents in a
water purification system. As a first step, the adsorption rate
was evaluated according to rate constant and rate-controlling
steps (i.e., mass transfer and chemical reaction processes). The
selection of pH 6 in the adsorption study was based on three
factors: sorption capacity, exhausted adsorbent stability (dis-
solution), and techno-economic indicators. A set of kinetic
models was used to fit the experimental data: the pseudo-
first order (Lagergren), the pseudo-second order (PSO) of
the Ho–McKay model (Eqs. S3 and S4), the Roginsky–
Zeldovich–Elovich equation, and the second-order rate equa-
tion, as well as adsorption diffusion models, the homogeneous
solid diffusion model, the parabolic or Weber–Morris model,
the Dunwald–Wagner, and the double exponential model (Foo
and Hameed 2010; Maliyekkal et al. 2009; Sahiner et al. 2015;
Yadanaparthi et al. 2009). The time-dependent removal of
arsenic species by the T–Cu(A) adsorbent (Fig. 2) showed
Environ Sci Pollut Res (2019) 26:24143–24161 24147

Fig. 1 HRTEM analysis of

structure properties of T–Cu(A).
a) Neat, b) after As(V) removal,
c) enlarged part of a), and d)
enlarged part of b)

the removal rate to be 99% for As(V) and 73% for As(III) for where: h is the Planck constant, K is the Boltzmann constant,
30 min. The kinetic data, analyzed according to linear and and R the gas constant (8.314 kJ mol−1 K−1). The obtained
non-linear least-squares methods (Kaygusuz et al. 2015), were results are given in Table 1.
satisfactorily correlated to the PSO equation as charted by
regression coefficients. The results of non-linear fitting are
given in Table 1.
Rate constants determined at 293, 303, and 313 K were
used for the correlation of logarithmic values of the rate con-
stants (Table 1) vs. the reciprocal temperature. The linearized
form of the Arrhenius equation (lnk = −(Ea/R)1/T + ln A) was
used to calculate the activation energy, Ea(Ho 2006). In this
equation, the parameter k is the rate constant, A is the pre-
exponential factor or frequency factor, R is the gas constant
(8.314 J mol−1 K−1), and T is the absolute temperature in K
(293, 303, and 313 K). The activation energy was further used
for the calculation of the entropy of activation, ΔS# (Eq. 4):
ΔS # 25 ¼ ðln½k2 h=KT þ E a =RTÞ R ð4Þ

The enthalpy of activation, ΔH#(Eq. 5):

ΔH # 25 ¼ E a −RT ð5Þ

and the Gibbs energy of activation, ΔG#(Eq. 6):

ΔG# 25 ¼ ΔH # 25 −ΔS # 25 T ð6Þ Fig. 2 Adsorption kinetics of arsenate onto T–Cu(A) adsorbent at pH 6.0
for both As(V) and As(III); Ci = 0.10 mg dm−3; T = 298 K
24148 Environ Sci Pollut Res (2019) 26:24143–24161

Table 1 Kinetic data obtained using PSO equation and the calculated activation parameters for As(III) and As(V) adsorption onto T–Cu(A–C) (pH = 6,
m/V = 100 mg dm−3, Ci = 0.10 mg dm−3, ω = 100 rpm)

T– Temp. (K) k2 (g mg−1 min−1) Ea* (kJ mol−1) ΔH25# (kJ mol−1) ΔG25# (kJ mol−1) TΔS25# (kJ mol−1)
Cu

As(III) A 293 0.509 ± 0.021 16.42 13.92 − 19.80 33.72

303 0.522 ± 0.022
313 0.532 ± 0.022
B 293 0.363 ± 0.018 17.12 14.62 − 25.25 39.88
303 0.444 ± 0.022
313 0.496 ± 0.028
C 293 0.304 ± 0.014 17.34 14.82 − 26.67 41.49
303 0.382 ± 0.017
313 0.479 ± 0.022
As(V) A 293 0.702 ± 0.035 17.11 14.62 − 26.88 41.51
303 0.744 ± 0.037
313 0.789 ± 0.039
B 293 0.682 ± 0.029 18.23 15.72 − 37.28 53.01
303 0.699 ± 0.030
313 0.712 ± 0.031
C 293 0.509 ± 0.017 19.15 16.62 − 45.94 62.56
303 0.514 ± 0.016
313 0.539 ± 0.015

A results of the series of kinetic experiments (Table 1)
showed two general trends: higher values of k2 for As(V)
removal and similar trends of temperature dependence of k2
versus T for both As(V) and As(III). The activation energy
(Ea) is the quantum energy necessary for the adsorption pro-
cess to overcome the energy barrier as determined by the
slowest adsorption step. Similar values of Ea likewise indicate
energy demand related to the rate-limiting adsorption step.
Differences in k are mainly governed by both adsorbent ge-
ometry and porosity (i.e., the adsorption rate as controlled by
diffusional processes). The physisorption processes, for Ea <
40 kJ mol−1, indicate a dominance of diffusion-controlled
sorption processes in porous media (Inglezakis and Zorpas
2012).The obtained Ea values, in the range 8–22 kJ mol−1,
indicate that intra-particular diffusion was at a rate-
controlling step (Haring 1942; Arrhenius 1889).
The effect of the particle size/diameter, which is the effect
of shear stress on the boundary surface of the size-dependent
particles in a stirred system, indicates a limited influence of Ea
and ΔH# as well as a somewhat higher influence on the ΔG#
and ΔS#. The creation of multiple chemical and physical
bonds in structurally arranged arsenic species on adsorbent
surfaces contributes to the ΔG25# and ΔS25# value change.
The higher values of ΔG25# vs. temperature increase for all
adsorbents also might indicate a higher feasibility and sponta-
neity. The higher feasibility of both As(V) and As(III) removal
demonstrates a decreased rate of diffusional limitation and
high chemisorption contribution.
By and large, the fast arsenic uptake was achieved using T–
Cu(A–C) adsorbents (Table 2). The rapid attainment of the
equilibrium was catalyzed by the high adsorbent surface af-
finity to arsenic species, and the specific distribution of the
pore network by comparison of the textural properties of the
T–Cu(A) support material (the surface area/pore volume/av-
erage pore diameter of 15.54/0.074/84.47 with respect to T–
Cu(A) of 68.47/0.336/191.89) (Bajić et al. 2016) points to
significant differences in the porosity of basic (tufa) and
copper-based deposited materials. The exceptional structural
properties/porosity distribution of the deposited material of its
core shell structure with a thin layer of copper-based coating
contains a beneficial outer surface concentrated porosity that
contributes to the lower traveling path of the analyte and, thus,
the low diffusional limitation. Accordingly, the preliminary
study on the influences of the agitation speed, ω = 250 and
400 rpm to the adsorption kinetics showed a negligible in-
crease of rate constants (k2) with respect to film diffusion;
e.g., an increase of 15.4% for system As(V)/T–Cu(A) at
400 rpm. These results show that the external diffusional re-
sistance is due to the boundary layer’s film thickness lowering
the change with respect to the agitation rate.
Analysis of diffusional processes
Fitting the kinetic data to the PSO equation provides an overall
understanding of the time-dependent adsorption process. The
obtained kinetic law/parameters on arsenic removal
Environ Sci Pollut Res (2019) 26:24143–24161 24149

Table 2 The rate constants for diffusion kinetic modeling of the D–W, HSDM, and W–M models in the adsorption of As(V) and As(III) onto T–Cu(A–
C)

As(V) As(III)

Kinetic model Constants T–Cu(A) T–Cu(B) T–Cu(C) T–Cu(A) T–Cu(B) T–Cu(C)

D–W K (min−1) 0.0268 0.0235 0.0222 0.024 0.0256 0.0191
R2 0.676 0.667 0.672 0.698 0.678 0.535
HSDM Ds (m2 s−1) 2.86E−11 2.62E−11 2.52E−11 2.69E−11 2.77E−11 2.10E−11
R2 0.669 0.664 0.700 0.697 0.672 0.532
−1 -0.5
W–M (step 1) kp1 (mg g min ) 0.152 0.148 0.122 0.146 0.141 0.135
C (mg g−1) 0.418 0.293 0.468 0.279 0.336 0.263
R2 0.998 0.997 0.986 0.967 0.995 0.997
W–M (step 2) kp2 (mg g−1 min-0.5) 0.0015 0.0075 0.0037 0.0205 0.0122 0.0004
C (mg g−1) 1.067 1.076 1.077 0.989 0.991 0.996
R2 0.974 0.974 0.993 0.933 0.989 0.920

performance is not able to describe mass transport phenome- transfer is only due to surface diffusion and that pore volume
na. The kinetics of the process become important when diffu- diffusion resistance is negligible (Foo and Hameed 2010;
sional limitations in a batch and fixed packed-bed configura- Zhang et al. 2009). The results of the D–W and HSDM fitting
tions are considered. In order to evaluate a rate-limiting step in of the kinetic data are given in Table 2. A simplified analysis
the overall process, intra-particle diffusion models were used of the contribution of the diffusional process is obtainable
(Arrhenius 1889; Bajić et al. 2016; Glasston et al. 1941). using the Weber–Morris Eq. (S12) (W–M). The results obtain-
Adsorption usually occurs through several consecutive ed following the W–M model show two linear plots (Fig. 3),
steps: external mass transfer (diffusion through the bulk of with rapid kinetics in the first phase and a slow attainment of
the liquid), diffusion across the boundary layer surrounding equilibrium in the second.
the particle (film diffusion), diffusional transport within the Due to the concentration gradient, the adsorbate travels
internal structure of the adsorbent (intra-particle diffusion), from the bulk, via axial convection and the Brownian diffu-
and adsorption on the solid surface. In general, overall mass sion, to the particle surface. The high slope of the correlation
transport may be controlled by more than one step at different line indicates a rapid external mass transfer from the bulk
stages of the adsorption process. The first two steps are high- solution to the most available external adsorbent surface and
rated processes while pore diffusion and adsorption are able to easily available macropore surface. Better kinetics, i.e., more
be controlling factors of the overall process. rapid saturation of external and easily available surface ad-
Dunwald–Wagner (Dünwald and Wagner 1934) (D–W) sorption sites, was obtained for the smaller particle of T–
derived an equation for solid-state reaction analysis based on Cu(A) adsorbent (Table 2) due to developed surface area
the Fick’s second law for diffusion into or out of a sphere. A
plot of the logarithm (1 − (q/qt)2) ~ t (Eqs. S6– S7) gave a
constant K as the measure of the rate-limiting step.
A typical intra-particle diffusion model is the homoge-
neous solid diffusion model (HSDM), descriptive of mass
transfers in amorphous and homogeneous spheres (as present-
ed by Eqs. S8).The HSDM model defines a three-step process:
(1) the adsorbate diffuses through a liquid film layer surround-
ing the adsorbent particle; (2) the adsorbate adsorbs from the
liquid phase onto the other surface of the adsorbent particle;
and (3) the adsorbate diffuses along the inner surface of the
adsorbent particles until it reaches adsorption sites (Crittenden
and Weber 1978; Najm et al. 1991). Crank and Skelland
(Skelland 1974) solved the differential equation Eq. (S8) giv-
ing Eq. (S9-S11) (detail presented in Supplementary material).
Generally, HSDM predicts the diffusion of a molecule from Fig. 3 An intra-particle diffusion model of the kinetics for As(III) and
the external surface of the adsorbent particle through the pore As(V) onto T–Cu(A) adsorbent (Ci = 0.1 mg dm−3, m/V = 100 mg dm−3,
surface to the adsorption site and assumes that internal mass pH = 6, T = 298 K)
24150
(Bajić et al. 2016), and the preferable ratio of the number of
surface active sites to the adsorbent mass. The correlation lines
do not pass through the origin, suggesting a contribution of
other processes to the overall mass transfer resistance (the
contribution of film diffusion).
The saturation of the external surface was followed by the
solute transport inside the particle. The solute travels into the
particle interior via pore diffusion, whereby the final step is
established by an adsorption/desorption equilibrium. The sec-
ond part of the correlation line indicates a high dependence of
the adsorption on the adsorbent porosity (the pore network
geometry/structure) (Foo and Hameed 2010; Zhang et al.
2009). The time-dependent reductions of mass transfer that
occurs due to intra-particle diffusion resistance slow down
the adsorbate transport to saturation point as the establishment
of adsorption/desorption equilibrium is achieved. The results
of the W–M fitting of kinetic data are given in Table 2.
The results of the D–W and HSDM fittings are not statisti-
cally valid, thereby indicating a lower dependence to the ad-
sorbent properties. However, corresponding differences were
noted with respect to the analytes. The results of the W–M
correlations indicate a dependence on diffusional processes as
a function of the nanoparticle size and adsorbent textural prop-
erties where there was a decreasing adsorption rate in the first
phase (kp1 in Table 2) followed by a decrease in the surface for
T–Cu(A) and T–Cu(C), from 65.47 to 46.89 m2 g−1, respec-
tively. The differences in the surface area of the adsorbents T–
Cu(A–C), mesopore volume (0.336 vs. 0.285 cm3), and
mesopore diameter (191.96 vs. 189.22 nm) (Bajić et al.
2016), respectively, are an additional factor governing varia-
tions in the kp1 and kp2 values of the adsorption process
(Table 2).
In order to resolve the extent of diffusional processes, the
external and internal mass transfer contribution to overall
mass transfer resistance were analyzed according to the kinetic
data through the single-resistance mass transfer model. When
the solution underwent intensive mixing in contact with the
adsorbent surface—from 100 to 400 rpm—it was found that
the suppression of the boundary layer would take place.
Consequently, an appropriate decrease of the boundary layer’s
resistance is to be expected (Pillewan et al. 2011). In terms of
efficiency of the process, the influence of the mixing rate on
the mass transfer of T–Cu(A–C) particles on the kinetics was
further analyzed. In order to evaluate range of external mass
transfer, the methodology used for kinetic modeling of mono-
valent ion adsorption onto a zeolite (Myneni et al. 1998) was
performed according to (Eq. 7):
 
qt
ln 1− ¼ k f at ð7Þ
qe
where qt (mmol g−1) refers to the adsorption at corresponding
time incrementation and qe (mmol g−1) is the adsorbed amount
Environ Sci Pollut Res (2019) 26:24143–24161
at equilibrium, while kfa (min−1) is the volumetric external
mass transfer coefficient. An analogous methodology was
used for modeling arsenic adsorption onto the T–Cu.
The mass transfer coefficients kf were obtained according
to dimensional analysis using dimensionless quantitative
criteria such as Sherwood (Sh), Eqs. (8) and (9), Reynolds
(Re), Eq. (S22), Schmidt number (Sc), Eq, (S23), which are
all valid for an isolated sphere (Boyd et al. 1947):
Sh ¼ 2 þ 1:58 Re0:4 Sc1=3 ð8Þ
Sh ¼ 2 þ 1:21Re1=2 Sc1=3 ð9Þ
where Sh = kf·dp/Dm; Sc = ν /Dm; Re = υ·dimp/ν (Re = ρlυ·dimp/
μ) the peripheral velocity is given by υ = dimpN as the velocity
of the fluid (m s−1) v = dimpN. Whereas dp(m) is the particle
diameter (values given in footnote of the Table 3), ν (m2 s−1) is
the fluid kinematic viscosity (1.004 × 10−6 at 20 °C), dimp is
the impeller diameter (m), Dm (m2 s−1) is the diffusivity of
arsenate in water (Dm = 9.05 × 10−10 m2 s−1), and N is the
agitation speed (rpm). Eqs. (8) and (9) are applicable at fol-
lowing constraints: 0 ≤ Sc ≤ 250; 0.001 ≤ Re < 5.8 and 0 ≤
Sc ≤ 250; 0 ≤ Re < 200, respectively.
Diffusion inside the pores could also slowdown the analyte
transport; as thus, they should be evaluated. A pore diffusion
model was developed by Boyd for linear isotherms.
Reichenberg (Reichenberg 1953), applied Fourier transforma-
tion and integration to the original equation, and obtained the
following approximation:
 
qt 6 4Deff π2 q
ln 1− ¼ ln 2 − t; for t > 0:85 ð10Þ
qe π d 2p qe
Assuming for spherical particles, Deff (m2 min−1) is the
effective diffusion coefficient and dp (m) the mean particle
diameter. The following variables were adopted: qt/qe = F as
the fractional attainment of equilibrium and B = (4 Deff π2)/
(dp2). It was found that all the experimental data for sorption
onto T–Cu(A–C) for t > 5 min met the criteria qt/qe > 0.85,
thus allowing for the application of Eq. (10). The calculated
parameters are provided in Table 3.
By increasing the agitation speed from 100 to 400 rpm, the
k f values for T–Cu(A) increased from 8.01 to 9.16 ×
10 −3 m min −1 for As(III), and from 1.03 to 1.51 ×
10−2 m min−1 for As(V) (Table 3). Similar trend of results
was found for T–Cu(B and C) adsorbents. These slight incre-
ments in kf changes indicate that the resistance due to external
film diffusion was of a low dependence to the mixing rate. The
overall mass transfer resistance in the first interval also is in
relation to the transportation/diffusional properties of the ion
inside the pore close to the adsorbent surface. The kf value was
inversely proportional to the film thickness. The low effect of
Environ Sci Pollut Res (2019) 26:24143–24161 24151

Table 3 The calculated mass transfer coefficients kfa, Deff and B for As(V) on T–Cu(A–C)

Adsorbent N (rpm) dp* (mm) kfa** (min−1) Re Sh kf (m min−1) Deff × 1011 (m2 min−1) B

T–Cu(A) 100 ≤ 0.020 0.0270 0.199 3.770 0.01032 5.475 0.1081

250 0.0265 0.498 4.798 0.01314 4.332 0.0855
400 0.0226 0.797 5.540 0.01516 3.740 0.0738
T–Cu(B) 100 0.020–0.250 0.0205 1.494 4.729 0.00984 31.20 0.0821
250 0.0189 3.735 6.316 0.01313 27.61 0.0727
400 0.0161 5.976 7.459 0.01551 24.14 0.0635
T–Cu(C) 100 0.250–1.500 0.0202 6.225 5.497 0.01125 127.9 0.0808
250 0.0178 15.563 7.530 0.01541 104.2 0.0658
400 0.0158 24.900 8.995 0.01841 95.36 0.0602
*
T–Cu(A–C) particle diameter range (mean particle diameter (dP = 6.40 ± 2.2 × 10−6 m for T–Cu(A), 64.6 ± 14.8 × 10−6 m for T–Cu(B), and 780 ±
187.6 × 10−6 m for T-Cu(C); ** Eq. (8) was used for T–Cu(A) and T–Cu(B) (100 and 250 rpm)

the mixing rate on the kfа value changes indicates negligible
resistance of the surrounding external fluid film for appropri-
ate adsorbent, while differences between adsorbent them-
selves are moderate to low. Similar results have been previ-
ously published in a study of mass transfer rate of Cu2+ re-
moval using hydroxyl apatite (Šljivić Ivanović et al. 2013).
The credibility of such a conclusion is confirmed by calcu-
lating qt for t > 15 min using Eq. (7) and calculated kf a values
(Table 3). For t = 20 min, the value of qt = 0.78 mg g−1 and
1.04 mg g−1 for arsenite and arsenate, respectively, (Fig. 2),
while the calculated values were significantly higher, indicat-
ing that expected film diffusion inside the pores becomes a
significant factor for t > 15 min. The boundary film resistance
proved to be low throughout the adsorption. Low values and
changes were found compared the adsorbents used.
The correlation of the external (volumetric) mass transfer
coefficient and agitation speed yielded the following equa-
tions:

of diffusion coefficient. Their mobility depends on
for AsðIIIÞ : k f ¼ 2:01  10–3  N 0:25 R2 ¼ 0:991 ð11Þ


for AsðVÞ : k f ¼ 3:93  10–3  N 0:22 R2 ¼ 0:992 ð12Þ
The kinetic data obtained by interpolation is provided in
Fig. 4, displaying a linear correlation of lnkf vs. lnN.
A linear correlation of lnkf vs. lnN (Fig. 4), obtained from
Eqs. (11) and (12), can be used to predict the kf in relation to
agitation speed in order to avoid additional experimental
work.
The evaluation of internal diffusional resistance was
preformed according to Eq. (10) from linear correlations ln(1
– qt/qe) versus to obtain slope B from which derive Deff
(Table 3). The values of Deff were found to be in the range
from 5.475 to 3.740 × 10−11 m2 min−1 for T–Cu(A), while
taking into calculation agitation speed changes from 100 to
400 rpm. These values are almost sixfold higher for T–Cu(B)
and 24-fold for T–Cu(C). Negligible contribution of Knudsen
diffusion (fourth order lower value) in relation to Deff is due to
short mean free path and large pore diameter (rpore in the range
102–122 nm) which contribute to infrequent collision with the
pore wall. In general, it could be assumed that Fickian diffu-
sion controls effective flux of adsorbate inside the pore in the
direction of decreasing concentration gradient.
Lower values and the similar trend of Deff changes were
found for As(III) removal. The higher intra-particular diffu-
sional resistance in T–Cu(A) particles, i.e., lower Deff values
for arsenate (Table 3), was affected by textural properties and
adsorbent porosity structure. In general, the extent of diffusion
of solute is governed by molecular size/geometry, charge (de-
gree of ionization), and charge distribution in the molecule
(degree of polarization). The pH-dependent speciation of ar-
senite and arsenate (Figs. S1) indicates that concentrations of
neutral and depronotated forms, i.e., charged oxyanions, due
to different diffusional capabilities influence the overall value
association/interactions with surrounding water molecules
(hydration shell) consisting of more or less arranged water
molecules. Moreover, the solute diffusional property
(Table 3) in bulk throughout the surface film is mainly depen-
dent on the mixing rate and particle size. Unexpectedly, an
increase in the effective diffusivities versus the particle size
was also noted (Table 3), wherein similar results were reported
to correspond to textural properties of studied adsorbents (Ma
et al. 1996).
A decreasing trend of B on both the agitation speed and
particle diameter was recorded (Table 3). A similar trend of
decreasing B values in the range of 12–28% for T–Cu(A–C)
when the initial metal concentration was increased from Ci =
0.10 to 0.50 mg dm−3. Such phenomena can be attributed to
the counter direction of the diffusion of the surface-released
cations/hydrogen (Jia et al. 2005). The calculated B value was
24152
Fig. 4 The relation between lnkf and the agitation speed lnN
significantly higher than that of kfa (Table 3), indicating the
existence of resistance of film diffusion in the initial period (t
< 15 min), while diffusion inside the pores contributed in the
initial and governed diffusional processes in the subsequent
adsorption period. It seems that low-slope plateau at t >
15 min relates to diffusional transport inside the lowest pore
diameter (micropore).The low dependence of parameter B on
the agitation speed (Table 3) reflects the significance of the
turbulent flow (shear stress at particle surface) which affects
traveling performances of ions through boundary layer and
pore system. This result is similar to those obtained by the
W–M fitting.
Competitive kinetic study
Apart from the attainment of the capacity and the adsorption
rate, the T–Cu adsorbents must also possess a high affinity/
selectivity in accordance with the common pollutant to be
adsorbed. Competitive adsorption experiments (Tables 4 and
5) were therefore carried out in order to evaluate the efficiency
of arsenic removal. The kinetic data obtained in the competi-
tive study of both As(V) and As(III) removal in the presence
of a single interfering ion, silicate, phosphate, and chromate
were successfully processed using a PSO equation. Therefore,
the results of the PSO modeling from t/qt − t are given in
Tables 4 and 5.
The results from Tables 4 and 5 indicate corresponding
relationships between the kinetic results in combination to
the adsorbent grain size and interfering ion present. Single
and shows similar behavior to phosphate. The most effective
adsorption occurs in the pKa (Veličković et al. 2012), strongly
confirming concurrency of phosphate versus arsenate for the
Environ Sci Pollut Res (2019) 26:24143–24161
competitive As(V) and As(III) adsorption showed a limited
influence of silicate, 11.4% decrease of k2, 19.0% for chro-
mate, and 30.3% for silicate in relation to the T–Cu(A) adsor-
bent. A similar trend was found for As(III): 4.13% for silicate,
14.2% for chromate, and 20.9% for phosphate. Phosphate had
the most significant interfering effect as a consequence of its
similar chemistry with arsenic and similar bonding type on the
adsorbent surface. A similar trend was observed for T–Cu(B
and C) adsorbents (Tables 4 and 5).
The results from Tables 4 and 5 indicate effects of
a couple of factors: the geometry/porosity of adsor-
bents, the chemical structure/properties of interfering
ions, and surface properties of T–Cu(A–C) adsorbents
as well as type of adsorption/interactions to competi-
tive kinetic adsorption. The similarity of chemical
properties of interfering ions, mostly phosphate,
causes the highest detrimental effect when carrying
out arsenic removal (speciation diagram given in
Fig. S1). The acid dissociation constants of phospho-
ric acid (pKa 1 2.16, pKa 2 7.21, and pKa 3 12.32) in-
dicate that the mono/divalent phosphate ion (H 2PO 4−/
HPO 42−) are the dominant competitive species at pH
~ 6 (Awual et al. 2011). It was likewise found for
chromate, where HCrO 4 − /CrO 4 2 − ions participate
while silicate H 4SiO 4 dominates at pH < 8. The speci-
ation diagrams of silicate, phosphate, and chromate
are given in Fig. S2.
The ionization of three protonic arsenic acids as presented
by successive ionization steps is
same adsorption sites. The speciation diagram of As(V) and
As(III) is provided in Fig. S1. The negatively charged As(V)
species participate in a varied electrostatic/complexation
Environ Sci Pollut Res (2019) 26:24143–24161 24153

Table 4 Results of the PSO

modeling of As(V) removal in the Non- Competitive condition
presence of interfering ions competitive

SiO44-* CrO42-* PO43-*

T–Cu(A) qe** 1.087 ± 0.036 0.974 ± 0.031 0.859 ± 0.027 0.769 ± 0.032
k2*** 0.702 ± 0.035 0.622 ± 0.0312 0.568 ± 0.0322 0.489 ± 0.0286
2
R 0.988 0.987 0.990 0.986
T–Cu(B) qe 0.968 ± 0.022 0.907 ± 0.051 0.815 ± 0.032 0.728 ± 0.025
k2 0.682 ± 0.029 0.616 ± 0.024 0.574 ± 0.027 0.502 ± 0.028
R2 0.995 0.989 0.987 0.996
T–Cu(C) qe 0.937 ± 0.019 0.897 ± 0.019 0.796 ± 0.033 0.672 ± 0.022
k2 0.569 ± 0.027 0.516 ± 0.022 0.465 ± 0.021 0.416 ± 0.021
R2 0.987 0.991 0.985 0.994
*
Ci = 0.1 mg L−1 , m/V = 100 mg L−1 , t = 90 min; ** mg g−1 ; *** g mg−1 min−1

interaction (attraction/repulsion) with the surface/ion charges
(Bajić et al. 2016). The positively charged surface of the T–Cu
adsorbents at pH < pHPZC (8.1) disparately attracts negatively
charged arsenate and phosphate ions. The higher flux of the
arsenate anionic specie (H2AsO4−) toward the adsorbent sur-
face reflects in higher rate constants. The lower adsorption
capacity of the phosphate at a lower pH (in the range 5–7)
offers a competitive benefit with respect to the arsenate. The
sulfate does not act as competitive specie in the operational
conditions used.
For the purposes of further investigation, adsorption
experiments were also carried out in the presence of
sodium chloride and sodium nitrate (0.5 and
2.0 mg dm−3). At an initial T–Cu(A) concentration of
100 mg dm −3 of 97% of As(V), 71% of As(III) in
presence 0.5, 94% of As(V), and 70% of As(III)in the
presence of 2 mg dm−3 of sodium chloride were re-
moved after 30 min. Similar results were obtained for
sodium nitrate. The presence of these salts in the course
of ion-binding contributes to the compensation of the
negative surface charges conglomerated in the course
of the adsorption (Grossl et al. 1997).Such similarities
in the arsenate uptake versus the increase of ionic
strength indicate the formation of inner-sphere com-
plexes in the main adsorption mechanism (Bajić et al.
2016).
Fast arsenic uptake was obtained and total arsenic
concentration was below 10 μg dm −3, which is the
WHO limit prescribed for arsenic in drinking water.
These results demonstrate the effectiveness of T–Cu-
based adsorbents (Tables 4 and 5), not only for interfer-
ing ion-free models in test water, but also for common
interfering ions present in real water as well. The results
obtained indicate a high potential for porous T–Cu that
contains a high affinity with respect to the arsenic when
applied in a dynamic system of pollutant removal.
Comparison of kinetic/adsorption data of T–Cu(A–C)
adsorbents
The adsorbent performance of T–Cu(A–C) may be deduced
when comparing with the adsorption characteristics of other
copper-based adsorbents (Escudero et al. 2009; Goswami
et al. 2012; Malana et al. 2011; Pillewan et al. 2011; Ranjan
et al. 2009; Reddy et al. 2013).Differences in the experimental
conditions allow only a more general qualitative overview of
the adsorption and kinetic data related to arsenate adsorption
given in the literature. Nevertheless, a short comparative anal-
ysis of up-to-date results for copper-based adsorbents, as pro-
vided in Table 6, with respect to the data on sorption kinetics
(Tables 1 and 2) and maximum adsorption capacity (Bajić
et al. 2016) of the T–Cu(A–C) adsorbents was carried out.
The time-dependent adsorption process on copper-based
nanoparticles is a multi-step process controlled by boundary
layer diffusion in first fast step and intra-particle diffusion in
the subsequent two; its mass transport capabilities correspond
to its mesopores and micropores (Table 6). Copper-based
nanoparticles and copper oxide–incorporated mesoporous
alumina (Table 6) are adsorbents of a significantly larger sur-
face area and lower pore diameter wherein intra-particle dif-
fusion contributes more to the overall resistance to adsorbate
mass transport. These findings shed light on how the adsorp-
tion mechanism may be attributed to the textural parameters
and the nature pollutant.
Adsorption isotherms parameters for As(III) and As(V)
removal by T–Cu(A–C), obtained in a previous study (Bajić
et al. 2016), are presented in Tables S2-S4. Comparison of qe
values shows that copper(II) oxide nanoparticles (from 0.101
to 1.01 mg g−1 with increasing Ci from 100 to 1000 μg dm−3),
and opposite trend of decreasing adsorption rate of k2, from
1.37 to 0.03, was presented (Goswami et al. 2012; Ranjan
et al. 2009). Similarly, low values of adsorption capacity were
presented for nanoaluminium doped manganese CuFe2O4
24154 Environ Sci Pollut Res (2019) 26:24143–24161

Table 5 Results of the PSO

modeling of As(III) removal in Non- Competitive condition
the presence of interfering ions competitive

SiO44-* CrO42-* PO43-*

T–Cu(A) qe** 0.808 ± 0.031 0.785 ± 0.022 0.736 ± 0.031 0.682 ± 0.029
k2*** 0.508 ± 0.0212 0.487 ± 0.0204 0.436 ± 0.0214 0.402 ± 0.0278
2
R 0.989 0.988 0.992 0.989
T–Cu(B) qe 0.741 ± 0.046 0.703 ± 0.023 0.686 ± 0.029 0.651 ± 0.032
k2 0.363 ± 0.0178 0.322 ± 0.0188 0.298 ± 0.0221 0.268 ± 0.0175
R2 0.997 0.990 0.988 0.993
T–Cu(C) qe 0.706 ± 0.022 0.682 ± 0.023 0.635 ± 0.021 0.601 ± 0.016
k2 0.304 ± 0.0137 0.289 ± 0.0122 0.275 ± 0.0122 0.256 ± 0.0112
R2 0.989 0.994 0.989 0.995
*
Ci = 0.1 mg L−1 , m/V = 100 mg L−1 , t = 90 min; ** mg g−1 ; *** g mg−1 min−1

polymer composite (Malana et al. 2011), CuO/mesoporous
alumina composite (Pillewan et al. 2011), and CuO nanopar-
ticles (Reddy et al. 2013).
Waste metal (hydr)-oxide showed high Langmuir adsorp-
tion capacity of 77.4 and 26.8 mg g−1 with respect to As(III)
and As(V), at very high initial As(III) and As(V) ion concen-
tration (5.00 and 500 mg dm−3 respectively) (Escudero et al.
2009)compared with the synthesized T–Cu(A) adsorbent
(67.83 and 104.62 mg g−1, respectively). Also, agricultural
waste (rice polish) reaches very high adsorption capacity
(139 and 147 mg g−1) at lower initial arsenic concentration
(0.1–1.0 mg L−1), but at low adsorption rate (Ranjan et al.
2009). Comparing maximum adsorption capacities for heavy
metal ions/oxyanions removal of cooper- and waste material-
based adsorbents in literature indicated that T–Cu showed
good adsorption performances at applied operational/
experimental conditions. Column adsorption testing was sub-
sequently performed to obtain relevant data on possible real-
istic application of studied materials.
Adsorption study in a the fixed-bed column
Recently, T–Cu adsorbents were tested for arsenic removal in
a batch system (Bajić et al. 2016). Advantageous characteris-
tics of the T–Cu adsorbent such as good mechanical proper-
ties, high adsorption capacity, and an effective reusability
make them to be promising candidate for the use in a flow
adsorption system.
In general, modeling of the breakthrough curve demands a
comprehensive mathematical calculation which includes dif-
ferent diffusion transport phenomena, such an axial disper-
sion, an external film resistance, and intra-particle diffusion.
The inclusion of all these operational factors needs time-
consuming procedure which will produce output results of
less or more acceptable level of the uncertainty. Thus, simpli-
fication of the modeling procedure may be achieved by
approximation assuming system idealization: regular change
of linear driving force and local equilibrium which provide
simple description of adsorption process. In that way, avoid-
ance of differential equation which describes incrementally
any stage of adsorption could be achieved.
The column capacity, qexp (mg g−1), for a given feed con-
centration and flow rate could be calculated from the experi-
mental data, i.e., the area under breakthrough curve (Sarkar
et al. 2007; Song et al. 2011). The complex procedure of
mathematical calculation in determination of adsorbent capac-
ity could be solved by using different commercial software.
An attractive alternative is a simplified modeling approach
used for the prediction of the dynamic behavior of the column
which does not consider the prediction of the breakthrough
curves. The multi-parameter fitting of the experimental points
using appropriate models gives results which approximately
describe the adsorbent performances.
Wealth of theoretical models dealing on mass transfer phe-
nomena in a bed column system is used to model the adsorp-
tion system under different operating conditions. In order to
overcome difficulties of mathematical modeling of the adsorp-
tion process in a fixed-bed system (Alberti et al. 2012), sim-
plified semi-empirical models were applied: Yoon–Nelson
(Yoon and Nelson 1984), Bohart–Adams (Bohart and
Adams 1920), Thomas (Thomas 1944), Clark (Clark 1987),
modified dose–response (MDR) (Song et al. 2011; Yan et al.
2001), and Wolborska model (Thomas 1944). The empirical
equations used for column system modeling are presented in
Supplementary Materials (Eqs. S13-S21).The study of the ad-
sorbents’ performances in a fixed-bed column experiment
needed additional optimization of various operational param-
eters: flow rate, adsorbate concentration, temperature, etc. An
assessment of the breakthrough performance of the T–Cu(A–
C) adsorbent was conducted using model water with Ci =
0.983, 4.995 and 9.983 mg dm−3 of As(V), and Ci = 0.989,
4.986 and 9.985 mg dm−3 of As(III), adjusted at pH ~ 6.0 and
Environ Sci Pollut Res (2019) 26:24143–24161
passed through the column at controlled flow rate: Q = 1–
10 cm3 min−1(step 1 cm3 min−1). The effluents were collected
at defined time intervals and used, after pH adjustment, for
determination of arsenic using ICP-MS. Adsorption capacity
of T–Cu(A–C)was calculated according to the breakthrough
point, i.e., the number of bed volumes until effluent water
attain arsenic concentration higher than 10 μg dm−3(the max-
imum contaminant level (MCL)) in effluent water.
Thus, results of T–Cu(A) fixed-bed column performance,
obtained by non-linear fitting of the column adsorption data,
following the Bohart–Adams, Yoon–Nelson, Thomas, Clark,
MDR, and Wolborska models, using Eqs. (S13)–(S21), are
shown in Tables 7 and 8.
It was found that the optimal flow rate is 2.26 cm3 min−1
for the T–Cu(A), 1.90 cm3 min−1 for the T–Cu(B), and
1.50 cm3 min−1 for the T–Cu(C) adsorbent. Example of dy-
namic flow experiments used for calculation of adsorbent ca-
pacity for T–Cu(A) is given in Fig. 5.
Analogous studies were performed at Q = 3 cm3 min−1,
Ci = 0.983, 4.995 and 9.983 mg dm−3 of As(V) and Ci =
0.989, 4.986 and 9.985 mg dm − 3 for As(III) (F =
6 cm min−1, EBCT = 2 min). Modeling of experimental data
using the Thomas models gave 40.23, 36.44, and
32.88 mg g − 1 for As(III) and 72.54, 66.86, and
58.78 mg g−1 for As(V), respectively, related to T–Cu(B).
Additional increase of Q = 4.0 cm3 min−1 (F = 8 cm min−1,
EBCT = 1.5 min) for T–Cu(B) gave following results: 37.44,
32.29, and 29.45 mg g−1 for As(III), and 66.56, 60.33, and
54.22 mg g−1 for As(V). Preliminary study performed for T–
Cu(C) gave lower capacity for 15–32% which indicates de-
veloped surface with more available adsorption sites.
Obtained results clearly indicate that an increase of the flow
rate causes decrease of the removal capacity. At higher flow
rate, the external mass transfer plays dominant role and thus
rapid saturation of the easily available sites could be achieved.
The short residence time in the column, i.e., the fast directional
mass transport and slow diffusion processes inside the pores,
causes saturation of external and mesopore surface.
Decreasing the flow rate, i.e., increasing EBCT, had the oppo-
site effect on adsorption capacity. The commonly acceptable
EBCT value for real application should be in the range 8–
15 min. Adjusting the flow bed system with T–Cu(B), poten-
tially applicable adsorbents in a real system for water purifi-
cation, using 0.8 × 22 cm column at 0.85 cm3 min−1 (EBCT ≈
13 min; F = 0.85 cm min −1 ; C i = 0.989 for As(V) and
0.983 mg dm−3 for As(III)) revealed an increase in the adsorp-
tion capacities to 57.26 mg g−1 for As(III) and 89.98 mg g−1
for As(V) (comparing with results from Tables 7 and 8), ob-
tained using the Thomas model, respectively. The longer bed
depth at the same velocity means more functional active sites
for adsorption, and thus higher removal efficiency (higher
exhaustion time) was achieved. The obtained capacity in-
crease is in appropriate relation to EBCT change, and thus
24155
selection of operational column parameters will be subject of
additional techno-economic analysis/calculation which
should justify decision on the operational and capital cost.
The obtained results showed that adjusting of the lower
EBCT affects the reduction of the operational and capital costs,
contributing to increased regeneration frequency with con-
comitant slight decrease of adsorption capacity.
Two important steps in a successful operational application
of an efficient flow system both material and process design
should be solved. Material properties were defined in a previ-
ous paper (Bajić et al. 2016), while in this work, the most
important task was related to adjustment of water flow and
adsorbate concentration. Determination of breakthrough point
is an important factor as it defines media capacity and time
necessary for regeneration. The effects of different process
variables, i.e., flow rate and influent arsenic concentration
(Table S1) on the selected system response (breakthrough
time) in column mode were explored considering the most
important process variables using the RSM method. The re-
sults indicate a wide range of breakthrough time (from 8.3 to
69.4 days) for the predicted responses, signifying that the se-
lected responses are strongly related to the considered vari-
ables. The predicted values’ breakthrough time was calculated
with the two variables, and presented with the help of contour
plots (the projection of the response surface as a 3D plane) as
on Fig. 6. The ANOVA for response surface quadratic model
for breakthrough time is listed in Tables S5 and S6, which
exemplified the statistical significance of the RSM quadratic
model for breakthrough time with F-value of 108.86, p value
< 0.0001, lack of fit 1.25, R2 was 0.9917, adjusted R2 was
0.9826, and predicted R2 was 0.9127. Furthermore, Fig. 6d
exhibits good agreement between the actual and predicted
values of breakthrough time. Accordingly, the statistical va-
lidity of the model was confirmed.
The predicted results are found to be in good agree-
ment with experimental results (Fig. 6d). From Fig.
6b, c, a general trend could be observed: the break-
through time increases with the decrease in flow rate
and increase in Ci. These results indicate a well-known
fact that lower flow rate provides better/efficient
adsorbate–adsorbent interaction (higher residence time)
and thus better performance of column due to efficient
use of adsorption sites located in internal pore network
(exhaustion of the most internal adsorption sites). By
analysis of overall interrelation between breakthrough
time and flow rate and Ci for increase of adsorbent size,
i.e., T–Cu(A–C), it could be observed that increase of
particle diameter affects decrease of time at which
breakthrough time was achieved. Thus, Fig. 6c depicts
the appearance of lower breakthrough times for T–
Cu(C) with respect to T–Cu(A and B) adsorbents (Fig.
6a, b). Along with lower number of functional adsorp-
tion sites, faster adsorption saturation was influenced by
24156 Environ Sci Pollut Res (2019) 26:24143–24161

Table 6 The kinetic parameters of the PSO fitting for both As(III) and As(V) adsorption on disparate Cu-modified sorbents

Adsorbent qm* Surface area Ci (μg dm−3) As qe** (mg g−1) k2 (g mg−1 min−1) Ref.
(mg g−1) (m2 g−1) species

Copper(II) oxide 1.086 52.11 100 As(III) 0.101 1.37 (Goswami et al. 2012)
nanoparticles 200 As(III) 0.203 0.41
500 As(III) 0.513 0.09
1000 As(III) 1.01 0.03
Nano aluminum doped 0.053 20–200 As(III) 0.0288 0.015 (Malana et al. 2011)
manganese CuFe2O4
polymer composite
CuO/mesoporous Alumina 2.16 189.25 1000 As(III) 0.191 0.011 (Pillewan et al. 2011)
2.02 As(V) 0.556 0.4047
CuO nanoparticles 86.51 100 Total As 0.0240 (4.0 g L−1) 36.18 (Reddy et al. 2013)
0.420 (0.2 g L−1) 0.26
Waste metal (hydr)-oxide 77.4 168 ± 6 5000–5·105 As(III) 7.2 ± 0.2 0.023 ± 0.005 (Escudero et al. 2009)
26.8 As(V) 11.9 ± 0.2 0.018 ± 0.003
Agricultural waste 139 452.0 100–1000 As(III) 58.82 9.27 × 10−4 (Ranjan et al. 2009)
(rice polish) 147 As(V) 73.52 6.88 × 10−4
T–Cu(A) 65.47 100–10,000 As(III) 67.83 0.509 ± 0.021 This study
As(V) 104.62 0.702 ± 0.035

*Maximum adsorption capacity calculated according to the Langmuir model; ** adsorption capacity obtained by calculation from equilibrium data

lower porosity, i.e., lower limitation of diffusional trans-
port (considering extent of developed pore structure).
According to results obtained, higher adsorption rate at
higher capacity could be obtained using the T–Cu(A)
adsorbent, and wide range of operational condition
could be applied for all adsorbents. In general, addition-
al techno-economic analysis should contribute to
narrowing optimal operational system performance to
adjust/optimize technology which could be applicable
in a real condition.

Table 7 Results of the fixed-bed column data fitting using different models for As(V) removal onto T–Cu(A) (EBCT = 2.67 min, Z* = 12 cm, F** =
4.5 cm min−1, Q = 2.26 cm3 min−1, pH 6.0 ± 0.2)

Ci mg dm−3 0.983 4.995 9.983

Bohart–Adams model kBA dm3 mg−1 min−1 0.04349 0.04201 0.04422
N0 mg dm−3 53.93 37.11 35.76
R2 0.9999 0.9999 0.9999
Yoon–Nelson model kYN min−1 0.0435 0.2101 0.4422
θcal Min 143.8 19.79 9.538
θexp Min 138.6 21.34 8.610
R2 0.9999 0.9999 0.9999
Thomas model kTh dm3 mg−1 min−1 0.0435 0.0440 0.0441
q0 mg g−1 86.29 75.40 57.24
R2 0.999 0.999 0.999
Clark model A 8.193 1.747 1.472
r min−1 0.02631 0.1333 0.2960
n 0.651 0.782 0.854
R2 0.9999 0.9999 0.9999
MDR model a 6.199 4.235 4.399
b mg g−1 28.20 19.11 18.54
R2 0.9999 0.9999 0.9999
Wolborska model β min−1 0.7817 0.5198 0.5272
N0 mg dm−3 53.93 37.11 35.76
R2 0.9999 0.9999 0.9999

*Z = H; **F ≡ γ (section 2.3)

Environ Sci Pollut Res (2019) 26:24143–24161 24157

Table 8 Results of the fixed-bed column data fitting using different packed-bed column with T–Cu media (laboratory scale) was
models for As(III) removal onto T–Cu(A) (EBCT = 2.67 min, Z* =
tested and results were processed using the pore surface dif-
12 cm, F** = 4.5 cm min−1,Q = 2.26 cm3 min−1, pH 6.0 ± 0.2)
fusion model (PSDM). In that way, prediction of a short bed
Ci mg dm−3 0.989 4.986 9.985 column performance was performed in order to obtain valid
Bohart–Adams kBA dm3 mg−1 min−1 0.0369 0.0463 0.0479 results without continuation on experimental work on pilot
model N0 mg dm−3 48.39 32.97 32.43 scale. Calculations of the external mass transport coefficient
R2 0.998 0.997 0.999 were based on the Gnielinski correlation with the following
Yoon–Nelson model kYN min−1 0.0369 0.2317 0.4789 constraints: Re × Sc > 500; 0.6 ≤ Sc ≤ 104; 1 ≤ Re < 100; 0.26
θcal min 129.0 17.59 8.648 < ε < 0.935 (Weber et al. 1988). Mathematical equations used
θexp min 131.2 19.11 7.940 for PSDM modeling are given in Supplementary Information
R2 0.999 0.999 0.999 (section S3.7; Eqs. S22-S27).
−1 −1
Thomas model kTh 3
dm mg min 0.0451 0.0476 0.0509 The medium porosity of T–Cu(A) with 4% loaded copper
q0 mg g−1 55.40 48.66 39.88 (≈ 5% calculated as CuO) showed the following adsorption
R2 0.998 0.999 0.998 parameters, calculated according to the Freundlich isotherm
Clark model A 2.763 1.547 0.2120 model: adsorption capacity parameter of K ≈ 87.7 and 32.5
r min−1 0.0235 0.1502 0.3334 (μg g−1)(L μg−1)1/n and the Freundlich intensity parameter
n 0.761 0.809 0.981 of 1/n ≈ 0.62 and 0.65 for As(V) and As(III), respectively.
R2 0.999 0.997 0.998 The evaluated tortuosity was 7.5 thus resulting in the pore
MDR model a 4.839 4.151 4.278 diffusion coefficient (Dp) (Eq. (S25)) of 4.4 × 10−11 m2 s−1. The
b mg g−1 25.02 16.92 16.73 relative importance of internal and external mass transport resis-
R2 0.999 0.999 0.999 tance was evaluated by calculating the pore Biot number (BiP)
Wolborska model β min−1 3.216 0.5094 0.5177 (Eq. S27), which was found to be 4.7, 47.8, and 578.1 for T–
N0 mg dm−3 48.39 32.97 32.43 Cu(A–C), respectively. Weber et al. (Weber et al. 1988) defined
R2 0.999 0.999 0.999 that for BiP > 20, the intra-particle diffusion controls the overall
mass transport of the system. Also, the calculated pore diffusion
*Z = H; **F ≡ γ (section 2.3) coefficient is significantly smaller than the free diffusivity of
arsenic in the water. Diffusivity of neutral form of arsenous acid
is (11.6 ± 0.10) × 10−6 cm2 s−1 (at pH < 7), monovalent arsenate
PSDM modeling of the column adsorptions results ions (8.12 ± 0.10) × 10−6 cm2/s (at pH 4.75), and divalent arse-
nate ions (7.27 ± 0.03) × 10−6 cm2 s−1 (at pH 9.2) showed small
Attainment of short mass transfer zones in a fixed-bed flow change and decrease of diffusion coefficients versus ionization
system is a goal necessary to obtain effective arsenic removal state comparing with Dp, suggesting that the pore diffusion de-
under high hydraulic loading rates and short empty bed con- termines the intensity of mass transfer processes. The obtained
tact times. Adsorption experiments in a small point-of-use results are presented on Fig. 7.

Fig. 5 Arsenic breakthrough curves for the T–Cu(A) adsorbent with model water by varying feed concentration for As(V) (a), and As(III) (b); Q =
2.26 cm3 min−1; EBCT = 2.67 min)
24158 Environ Sci Pollut Res (2019) 26:24143–24161

Fig. 6 Response surface plots for breakthrough time versus Ci(mg dm−3) and flow rate Q (cm−3 min) (a–c) and inter-correlation of predicted versus
experimental values

As obtained in Fig. 7, the number of bed volumes that can
be treated until the maximum contaminant level (MCL) of
10 μg dm−3 is directly related to BV numbers. The PSDM
predicted that 24,700 BV of As(V) solution can be treated at
EBCT of 2.5 min (C/Ci ≈ 0.01). This number increases to
37,900 and 52,700 with increase of the EBCT to 5 and
10 min, respectively. Similar to that, 14,100, 24,000, and
39,500 BV of As(III) solution can be treated at EBCT of 2.5,
5, and 10 min, respectively. Obtained results showed that
more BV can be treated at higher EBCT. In general, without
experimental testing at pilot scale, the PSDM modeling was
useful in providing a realistic media’s capacity under applied
hydraulic loading rates. Moreover, the maximum number of
bed volumes treated (BVMAX) can be estimated using Eq.

Fig. 7 Experimental data and PSDM prediction of a full-scale system packed with T–Cu(A) media in arsenic-only water at 20 °C (Ci~1 mg dm−3, pH 6.0
± 0.2)
Environ Sci Pollut Res (2019) 26:24143–24161
(S28) as suggested by Sontheimer et al. The maximum num-
ber of bed volumes that could be treated was estimated to ~
95170BV for As(V) at Ci = 0.983 mg dm−3. In that way, with-
out performing experimental work on pilot scale obtained,
results gave useful information necessary for decision about
material applicability in a real water purification processes.
Future challenges
These results present modeling of column-type processes used
to describe the dynamic adsorption process and the estimation
of the adsorbent usage. The adsorption performances of T–Cu
adsorbents, comparable/higher than the most scientifically
presented/commercial materials, observed for synthesized/
well-characterized/applicable copper oxides based adsorbents
could be recognized as contribution of two factors: (i) use of
the tufa natural materials with high affinity toward reduced
copper species and (ii) development of controllable method
for precipitation of copper nanomaterials. The intention of the
presented study was devoted on the work to reach deeper
understanding of basic principles related to influences of the
media properties, and morphology/geometry on adsorption
performances of T–Cu adsorbents. Obviously, crystal packing
of precipitated copper oxide species, exposed surface, and
crystal faces showed extraordinary effects on adsorbent prop-
erties. According to obtained results, good adsorption perfor-
mances of the T–Cu adsorbents were related to controllable
copper species precipitation which causes beneficial facet
evolution, resulting to high affinity, good kinetic, and efficient
removal of As(V) and As(III) with T–Cu. It offers possibility
for repeated use in the processes of natural water remediation
even in the presence of competing ions.
In a previous study (Bajić et al. 2016), techno-economic
analysis of T–Cu adsorbent indicated favorable properties/
cost of T–Cu(A) for application in a real water purification
system. Therefore, in the present study, high-capacity and
good-affinity adsorbents used for arsenate removal from nat-
ural water were synthesized. Efficient arsenic removal was
achieved even at high concentrations of interfering ions. The
high removal capacity with effective kinetic and predicted
applicability in a flow system gave a positive circular indicator
in a comparison with commercial adsorbents. The obtained
results indicate that T–Cu could be an acceptable alternative
to commercial expensive adsorbents currently in use.
Concluding remarks
Through experimentation, this study has found a highly effi-
cient, low-cost adsorbent for arsenic removal as being obtain-
able by the precipitation of reduced copper species onto natu-
rally occurring mineral tufa that may be readily found in
Southern Serbia in high quantities. A comprehensive kinetic
24159
study was carried out, including an evaluation of general rate
kinetic law, diffusional equations, and the contribution of ex-
ternal and internal diffusion to an overall removal. The
pseudo-second-order and the Weber–Morris kinetic models
yielded the best fitting of the kinetic data. Intra-particle diffu-
sional processes were indicated to be the main controlling
factors of the exceptionally fast adsorption processes.
The high correlation between qexp and qcal indicates the
Bohart–Adams, Yoon–Nelson, and Thomas models preferably
fit the column adsorption data, which is manifested through the
simplicity of the application and consistency in predicting their
breakthrough curves and adsorption capacity. The experimental
kinetic results showed that the tested (T–Cu(A–C)) adsorbents
were efficient and reusable for arsenic removal from natural
water in the presence of competitive phosphate, chromate,
and silicate ions in batch operational modes. Thereby, the main
outcomes of this study were achieved toward the creation of a
simple and productive valorization of T–Cu adsorbents as well
as the use of RSM methods to reduce the total number of
experiments. These experiments have all been conducted in
accordance with environmental protection policies in material
science as to define the future development of new efficient
adsorbents for wastewater treatment.
Funding information The authors acknowledge receiving financial sup-
port from the Ministry of Education, Science and Technological
Developments of the Republic of Serbia, Project No. III45019,
III43009, OI172057, and University of Defense, Republic of Serbia, pro-
ject VA-TT/4/16-18.
References
Agency EP (2005) Treatment technologies for arsenic removal.
Washington DC
Alberti G, Amendola V, Pesavento M, Biesuz R (2012) Beyond the syn-
thesis of novel solid phases: review on modelling of sorption phe-
nomena. Coord Chem Rev 256:28–45. https://doi.org/10.1016/j.ccr.
2011.08.022
Arrhenius S (1889) Über die Dissociationswärme und den Einfluss der
Temperatur auf den Dissociationsgrad der Elektrolyte. Z Phys Chem
4. https://doi.org/10.1515/zpch-1889-0108
Athanasaki G, Sherrill L, Hristovski KD (2015) The pore surface diffu-
sion model as a tool for rapid screening of novel nanomaterial-
enhanced hybrid ion-exchange media. Environ Sci Water Res
Technol 1:448–456. https://doi.org/10.1039/C5EW00108K
Awual MR, El-Safty SA, Jyo A (2011) Removal of trace arsenic(V) and
phosphate from water by a highly selective ligand exchange adsor-
bent. J Environ Sci 23:1947–1954. https://doi.org/10.1016/S1001-
0742(10)60645-6
Baig SA, Sheng T, Hu Y, Xu J, Xu X (2015) Arsenic removal from
natural water using low cost granulated adsorbents: a review.
CLEAN - Soil, Air, Water 43:13–26. https://doi.org/10.1002/clen.
201200466
Bajić ZJ, Veličković ZS, Djokić VR, Perić-Grujić AA, Ersen O,
Uskoković PS, Marinković AD (2016) Adsorption study of arsenic
removal by novel hybrid copper impregnated tufa adsorbents in a
batch system. CLEAN - Soil, Air, Water 44:1477–1488. https://doi.
org/10.1002/clen.201500765
24160
Bhattacharya P, Mukherjee AB, Loeppert R, Bundschuh J, Zevenhoven R
(2007) Arsenic in soil and groundwater environment, volume 9, 1st
edn. Elsevier Inc, New York
Bohart GS, Adams EQ (1920) Some aspects of the behavior of charcoal
with respect to chlorine. J Am Chem Soc 42:523–544. https://doi.
org/10.1021/ja01448a018
Bordoloi S, Nath SK, Gogoi S, Dutta RK (2013) Arsenic and iron remov-
al from groundwater by oxidation–coagulation at optimized pH:
laboratory and field studies. J Hazard Mater 260:618–626. https://
doi.org/10.1016/j.jhazmat.2013.06.017
Boyd GE, Adamson AW, Myers LS (1947) The exchange adsorption of
ions from aqueous solutions by organic zeolites. II. Kinetics 1. J Am
Chem Soc 69:2836–2848. https://doi.org/10.1021/ja01203a066
Bundschuh J, Armienta MA, Birkle P, Bhattacharya P, Matschullat J,
Mukherjee AB (2008) Natural arsenic in Groundwaters of Latin
America. CRC Press, Taylor & Francis Group, London
Choong TSY, Chuah TG, Robiah Y, Gregory Koay FL, Azni I (2007)
Arsenic toxicity, health hazards and removal techniques from water:
an overview. Desalination 217:139–166. https://doi.org/10.1016/j.
desal.2007.01.015
Clark RM (1987) Evaluating the cost and performance of field-scale
granular activated carbon systems. Environ Sci Technol 21:573–
580. https://doi.org/10.1021/es00160a008
Crittenden J, Weber W (1978) Predictive model for design of fixed-bed
adsorbers: single-component model verification. J Environ Eng Div
104:433–443
Crittenden JC, Hutzler NJ, Geyer DG, Oravitz JL, Friedman G (1986)
Transport of organic compounds with saturated groundwater flow:
model development and parameter sensitivity. Water Resour Res 22:
271–284. https://doi.org/10.1029/WR022i003p00271
Dünwald H, Wagner C (1934) Methodik der Messung von
Diffusiongeschwindigkeiten bei Losungsvorgangen von Gasen in
festen Phasen (Measurement of Diffusion Rate in the Process of
Dissolving Gases in Solid Phases). Z Phys Chem B 24:53–58
Escudero C, Fiol N, Villaescusa I, Bollinger J-C (2009) Arsenic removal
by a waste metal (hydr)oxide entrapped into calcium alginate beads.
J Hazard Mater 164:533–541. https://doi.org/10.1016/j.jhazmat.
2008.08.042
Foo KY, Hameed BH (2010) Insights into the modeling of adsorption
isotherm systems. Chem Eng J 156:2–10. https://doi.org/10.1016/j.
cej.2009.09.013
Fu F, Dionysiou DD, Liu H (2014) The use of zero-valent iron for
groundwater remediation and wastewater treatment: a review. J
Hazard Mater 267:194–205. https://doi.org/10.1016/j.jhazmat.
2013.12.062
Geckeler KE, Nishide H (eds) (2009) Advanced nanomaterials. Wiley-
VCH Verlag GmbH & Co. KGaA, Weinheim. https://doi.org/10.
1002/9783527628940
Geng G, Myers RJ, Qomi MJA, Monteiro PJM (2017) Densification of
the interlayer spacing governs the nanomechanical properties of
calcium-silicate-hydrate. Sci Rep 7:10986. https://doi.org/10.1038/
s41598-017-11146-8
Gholami MM, Mokhtari MA, Aameri A, Alizadeh Fard MR (2006)
Application of reverse osmosis technology for arsenic removal from
drinking water. Desalination 200:725–727. https://doi.org/10.1016/
j.desal.2006.03.504
Glasston S, Laidler KJ, Eyring H (1941) The theory of rate processes.
McGraw-Hill, New York
Gorobinskii LV, Efimova NN, Fattakhova ZT, Korchak VN (2005)
Preparation of CaO-CaCO3 pillared clay. Inorg Mater 41:203–205.
https://doi.org/10.1007/s10789-005-0045-9
Goswami A, Raul PK, Purkait MK (2012) Arsenic adsorption using cop-
per (II) oxide nanoparticles. Chem Eng Res Des 90:1387–1396.
https://doi.org/10.1016/j.cherd.2011.12.006
Grossl PR, Eick M, Sparks DL, Goldberg S, Ainsworth CC (1997)
Kinetic evaluation using a pressure-jump relaxation technique.
Environ Sci Pollut Res (2019) 26:24143–24161
Environ Sci Technol 31:321–326. https://doi.org/10.1021/
es950654l
Habuda-Stanić M, Kuleš M, Kalajdžić B, Romić Ž (2007) Quality of
groundwater in eastern Croatia. The problem of arsenic pollution.
Desalination 210:157–162. https://doi.org/10.1016/j.desal.2006.05.
040
Hand DW, Crittenden JC, Hokanson DR, Bulloch JL (1997) Predicting
the performance of fixed-bed granular activated carbon adsorbers.
Water Sci Technol 35. https://doi.org/10.1016/S0273-1223(97)
00136-4
Haring MM (1942) The theory of rate processes (Glasstone, Samuel;
Laidler, Keith J.; Eyring, Henry). J Chem Educ 19:249. https://doi.
org/10.1021/ed019p249.1
Ho Y (2006) Review of second-order models for adsorption systems. J
Hazard Mater 136:681–689. https://doi.org/10.1016/j.jhazmat.2005.
12.043
Hristovski KD, Westerhoff PK, Crittenden JC, Olson LW (2008)
Arsenate removal by nanostructured ZrO 2 spheres. Environ Sci
Technol 42:3786–3790. https://doi.org/10.1021/es702952p
Hristovski K, Westerhoff P, Crittenden J (2008a) An approach for evalu-
ating nanomaterials for use as packed bed adsorber media: a case
study of arsenate removal by titanate nanofibers. J Hazard Mater
156:604–611. https://doi.org/10.1016/j.jhazmat.2007.12.073
Inglezakis VJ, Zorpas AA (2012) Heat of adsorption, adsorption energy
and activation energy in adsorption and ion exchange systems.
Desalin Water Treat 39:149–157. https://doi.org/10.1080/
19443994.2012.669169
International Committee on Natural Zeolites (2000) Natural zeolites for
the third millenium. A. De Frede, Napoli
Jia Y, Wang R, Fane A (2005) Atrazine adsorption from aqueous solution
using powdered activated carbon—improved mass transfer by air
bubbling agitation. Chem Eng J. https://doi.org/10.1016/j.cej.2005.
10.014
Kaygusuz H, Uzaşçı S, Erim FB (2015) Removal of fluoride from aque-
ous solution using aluminum alginate beads. CLEAN - Soil, Air,
Water 43:724–730. https://doi.org/10.1002/clen.201300632
Lakshmanan D, Clifford D, Samanta G (2008) Arsenic removal by coag-
ulation with aluminium, iron, titanium and zirconium. J Am Water
Works Assoc 100:76–79
Lei J, Peng B, Liang Y-J, Min X-B, Chai L-Y, Ke Y, You Y (2018) Effects
of anions on calcium arsenate crystalline structure and arsenic sta-
bility. Hydrometallurgy 177:123–131. https://doi.org/10.1016/j.
hydromet.2018.03.007
Ma Z, Whitley RD, Wang N-HL (1996) Pore and surface diffusion in
multicomponent adsorption and liquid chromatography systems.
AICHE J 42:1244–1262. https://doi.org/10.1002/aic.690420507
Magnuson ML, Speth TF (2005) Quantitative structure−property rela-
tionships for enhancing predictions of synthetic organic chemical
removal from drinking water by granular activated carbon.
Environ Sci Technol 39:7706–7711. https://doi.org/10.1021/
es0508018
Malana MA, Qureshi RB, Ashiq MN (2011) Adsorption studies of arse-
nic on nano aluminium doped manganese copper ferrite polymer
(MA, VA, AA) composite: kinetics and mechanism. Chem Eng J
172:721–727. https://doi.org/10.1016/j.cej.2011.06.041
Maliyekkal SM, Philip L, Pradeep T (2009) As(III) removal from drink-
ing water using manganese oxide-coated-alumina: performance
evaluation and mechanistic details of surface binding. Chem Eng J
153:101–107. https://doi.org/10.1016/j.cej.2009.06.026
Markovski JS, Marković DD, Dokić VR, Mitrić M, Ristić MD, Onjia AE,
Marinković AD (2014) Arsenate adsorption on waste eggshell mod-
ified by goethite, α-MnO2 and goethite/α-MnO2. Chem Eng J 237:
430–442. https://doi.org/10.1016/j.cej.2013.10.031
Martinson CA, Reddy KJ (2009) Adsorption of arsenic(III) and
arsenic(V) by cupric oxide nanoparticles. J Colloid Interface Sci
336:406–411. https://doi.org/10.1016/j.jcis.2009.04.075
Environ Sci Pollut Res (2019) 26:24143–24161
Mertz KA, Gobin F, Hand DW, Hokanson DR, Crittenden JC (1999)
Manual: adsorption design software for windows (Ad-DesignS)
Miller GP, Norman DI, Frisch PL (2000) A comment on arsenic species
separation using ion exchange. Water Res 34:1397–1400. https://
doi.org/10.1016/S0043-1354(99)00257-2
Milonjić SK, Čerović LS, Čokeša DM, Zec S (2007) The influence of
cationic impurities in silica on its crystallization and point of zero
charge. J Colloid Interface Sci 309:155–159. https://doi.org/10.
1016/j.jcis.2006.12.033
Mohan D, Pittman CU (2007) Arsenic removal from water/wastewater
using adsorbents—a critical review. J Hazard Mater 142:1–53.
https://doi.org/10.1016/j.jhazmat.2007.01.006
Mohora E, Rončević S, Dalmacija B, Agbaba J, Watson M, Karlović E,
Dalmacija M (2012) Removal of natural organic matter and arsenic
from water by electrocoagulation/flotation continuous flow reactor.
J Hazard Mater 235–236:257–264. https://doi.org/10.1016/j.
jhazmat.2012.07.056
Morrison GMP, Bately GE, Florence TM (1989) Metal speciation and
toxicity. Chem Br 25:791–795
Myneni SCB, Traina SJ, Waychunas GA, Logan TJ (1998) Vibrational
spectroscopy of functional group chemistry and arsenate coordina-
tion in ettringite. Geochim Cosmochim Acta 62:3499–3514. https://
doi.org/10.1016/S0016-7037(98)00221-X
Najm IN, Snoeyink VL, Galvin TL (1991) Control of Organic
Compounds with Powdered Activated Carbon. AWWA Research/
Association, Denver
Pal P, Ahammad SZ, Pattanayak A, Bhattacharya P (2007) Removal of
arsenic from drinking water by chemical precipitation – a modeling
and simulation study of the physical-chemical processes. Water
Environ Res 79:357–366. https://doi.org/10.2175/
106143006X111754
Pillewan P, Mukherjee S, Roychowdhury T, Das S, Bansiwal A, Rayalu S
(2011) Removal of as(III) and as(V) from water by copper oxide
incorporated mesoporous alumina. J Hazard Mater 186:367–375.
https://doi.org/10.1016/j.jhazmat.2010.11.008
Ranjan D, Talat M, Hasan SH (2009) Biosorption of arsenic from aque-
ous solution using agricultural residue ‘rice polish.’. J Hazard Mater
166:1050–1059. https://doi.org/10.1016/j.jhazmat.2008.12.013
Reddy KJ, McDonald KJ, King H (2013) A novel arsenic removal pro-
cess for water using cupric oxide nanoparticles. J Colloid Interface
Sci 397:96–102. https://doi.org/10.1016/j.jcis.2013.01.041
Reichenberg D (1953) Properties of ion-exchange resins in relation to
their structure. III. Kinetics of exchange. J Am Chem Soc 75:589–
597. https://doi.org/10.1021/ja01099a022
Röhricht M, Krisam J, Weise U, Kraus UR, Düring R-A (2009)
Elimination of carbamazepine, diclofenac and naproxen from treat-
ed wastewater by nanofiltration. CLEAN - Soil, Air, Water 37:638–
641. https://doi.org/10.1002/clen.200900040
Sahiner N, Demirci S, Sahiner M, Yilmaz S, Al-Lohedan H (2015) The
use of superporous p(3-acrylamidopropyl)trimethyl ammonium
chloride cryogels for removal of toxic arsenate anions. J Environ
Manag 152:66–74. https://doi.org/10.1016/j.jenvman.2015.01.023
Saleh TA, Agarwal S, Gupta VK (2011) Synthesis of MWCNT/MnO2
and their application for simultaneous oxidation of arsenite and
sorption of arsenate. Appl Catal B Environ 106:46–53. https://doi.
org/10.1016/j.apcatb.2011.05.003
Sarkar S, Blaney LM, Gupta A, Ghosh D, SenGupta AK (2007) Use of
ArsenXnp, a hybrid anion exchanger, for arsenic removal in remote
villages in the Indian subcontinent. React Funct Polym 67:1599–
1611. https://doi.org/10.1016/j.reactfunctpolym.2007.07.047
Shih M-C (2005) An overview of arsenic removal by pressure-
drivenmembrane processes. Desalination 172:85–97. https://doi.
org/10.1016/j.desal.2004.07.031
Sing KSW, Everett DH, Haul RAW, Moscou L, Pierotti RA, Rouquérol J,
Siemieniewska T (1985) Reporting physisorption data for gas/solid
24161
systems with special reference to the determination of surface area
and porosity (recommendations 1984). Pure Appl Chem 57:603–
619. https://doi.org/10.1351/pac198557040603
Skelland AHP (1974) Diffusional mass transfer. Wiley, New York
Šljivić Ivanović M, Smičiklas I, Pejanović S (2013) Analysis and com-
parison of mass transfer phenomena related to Cu2+ sorption by
hydroxyapatite and zeolite. Chem Eng J 223:833–843. https://doi.
org/10.1016/j.cej.2013.03.034
Song J, Zou W, Bian Y, Su F, Han R (2011) Adsorption characteristics of
methylene blue by peanut husk in batch and column modes.
Desalination 265:119–125. https://doi.org/10.1016/j.desal.2010.07.
041
Sontheimer H, Crittenden J, Summers S (1988) Activated carbon for
water treatment. DVGW-Forschungsstelle, Engler-Bunte Institut,
Karlsruhe
Thomas HC (1944) Heterogeneous ion exchange in a flowing system. J
Am Chem Soc 66:1664–1666. https://doi.org/10.1021/ja01238a017
Vaiano V, Iervolino G, Sannino D, Rizzo L, Sarno G, Farina A (2014)
Enhanced photocatalytic oxidation of arsenite to arsenate in water
solutions by a new catalyst based on MoOx supported on TiO2.
Appl Catal B Environ 160–161:247–253. https://doi.org/10.1016/j.
apcatb.2014.05.034
Veličković Z, Vuković GD, Marinković AD, Moldovan M-S, Perić-
Grujić AA, Uskoković PS, Ristić MĐ (2012) Adsorption of arsenate
on iron(III) oxide coated ethylenediamine functionalized multiwall
carbon nanotubes. Chem Eng J 181–182:174–181. https://doi.org/
10.1016/j.cej.2011.11.052
Wang C, Luo H, Zhang Z, Wu Y, Zhang J, Chen S (2014) Removal of
as(III) and as(V) from aqueous solutions using nanoscale zero valent
iron-reduced graphite oxide modified composites. J Hazard Mater
268:124–131. https://doi.org/10.1016/j.jhazmat.2014.01.009
Weber WI, Sontheimer H, Crittenden JC, Summers S (1988) Activated
carbon for water treatment, 2nd edn. DVGW-Forschungsstelle,
Engler-Bunte Institut, Universitat Karlsruhe, Karlsruhe
Xu P, Capito M, Cath TY (2013) Selective removal of arsenic and mono-
valent ions from brackish water reverse osmosis concentrate. J
Hazard Mater 260:885–891. https://doi.org/10.1016/j.jhazmat.
2013.06.038
Yadanaparthi SKR, Graybill D, von Wandruszka R (2009) Adsorbents for
the removal of arsenic, cadmium, and lead from contaminated wa-
ters. J Hazard Mater 171:1–15. https://doi.org/10.1016/j.jhazmat.
2009.05.103
Yan G, Viraraghavan T, Chen M (2001) A new model for heavy metal
removal in a biosorption column. Adsorpt Sci Technol 19:25–43.
https://doi.org/10.1260/0263617011493953
Yan D, Gang DD, Zhang N, Lin L (2013) Adsorptive selenite removal
using iron-coated GAC: modeling selenite breakthrough with the
pore surface diffusion model. J Environ Eng 139:213–219. https://
doi.org/10.1061/(ASCE)EE.1943-7870.0000633
Yoon YH, Nelson JH (1984) Application of gas adsorption kinetics I. A
theoretical model for respirator cartridge service life. Am Ind Hyg
Assoc J 45:509–516. https://doi.org/10.1080/15298668491400197
Zhang G, Liu H, Liu R, Qu J (2009) Adsorption behavior and mechanism
of arsenate at Fe–Mn binary oxide/water interface. J Hazard Mater
168:820–825. https://doi.org/10.1016/j.jhazmat.2009.02.137
Zhang Y, Zhou Y, Peng C, Shi J, Wang Q, He L, Shi L (2018) Enhanced
activity and stability of copper oxide/γ-alumina catalyst in catalytic
wet-air oxidation: critical roles of cerium incorporation. Appl Surf
Sci 436:981–988. https://doi.org/10.1016/j.apsusc.2017.12.036
Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.