Chemical Engineering Journal 225 (2013) 625–635

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal

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

Synthesis and characterization of anatase nanoadsorbent and applicationin removal of lead, copper and arsenic from water

1385-8947/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.cej.2013.03.106

⇑ Corresponding author. Tel.: +90 216 4839512; fax: +90 216 4839550.E-mail address: yyurum@sabanciuniv.edu (Y. Yürüm).

Züleyha Özlem Kocabas�-Ataklı, Yuda Yürüm ⇑Faculty of Engineering and Natural Sciences, Sabanci University, Tuzla, Istanbul 34956, Turkey

h i g h l i g h t s

� The anatase nanoparticles weresuccessfully synthesized by sol–gelmethod.� The kinetic data of lead, copper, and

arsenic followed the pseudo-second-order model.� The favorable Gibbs free energies of

lead and copper were entropicallycontrolled.� The surface oxygen-containing

functional groups were involved inthe adsorption.

g r a p h i c a l a b s t r a c t

a r t i c l e i n f o

Article history:Received 16 November 2012Received in revised form 15 March 2013Accepted 19 March 2013Available online 6 April 2013

Keywords:Anatase nanoparticlesLeadCopperArsenicKinetic parametersThermodynamics

a b s t r a c t

The existences of lead, copper, and arsenic in water supplies are great concern due to their potentialeffects on human health. This study demonstrates beneficial adsorptive eliminations of lead, copper,and arsenic from water using anatase nanoadsorbent produced by sol–gel method followed by calcina-tion at 400 �C.

The structure of the resulting sample and the surface species of anatase nanoadsorbent were deter-mined using X-ray diffraction (XRD), Raman spectroscopy and X-ray photoelectron spectrometer (XPS).The amounts of adsorbed lead and copper by the nanoadsorbent increased with increasing pH, whilethe adsorbed arsenic species were almost constant with the changes of pH values. The sorption kineticdata of lead, copper, and arsenic followed the pseudo-second-order model. The equilibrium data weremodeled with the help of Langmuir, Freundlich, Redlich–Peterson and Sips equations. Based on the valuesof correlation coefficients (R2) and the average percent error (APE), the Langmuir isotherm equation wasexcellently fitted to the experimental data of the lead adsorption with diminutive error while the adsorp-tions of copper and arsenic were described better with the Freundlich equation. The maximum experi-mental adsorption uptakes were 31.25 mg/g for lead, 23.74 mg/g for copper, and 16.98 mg/g forarsenic, respectively. From the thermodynamic calculations the favorable Gibbs energies were entropi-cally controlled in the case of lead and copper adsorptions while the arsenic uptake was thermodynam-ically favorable without any entropic or enthalpic limitations. This study is valuable for the qualityutilization of drinking water that includes arsenic, copper and lead contamination by using anatase nano-particles as a nanoadsorbent.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Lead, copper, and arsenic are well-known inorganic contami-nants in drinking water [1]. Arsenic is introduced into environmentby geochemical reactions, natural weathering reaction, miningactivities, industrial wastes and volcanic emissions [2]. However,

626 Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635

the primary sources of lead and copper are the corrosion of pipes,faucets, household plumbing systems, mining, smelting and bat-tery manufacturing [3,4]. Especially, the accumulation of arsenic,lead, and copper in the body may reach toxic levels. Long-termconsumption of even low levels of arsenic could be dangerous asit can result in lung, skin, liver, bladder, and kidney cancers [5,6].Similarly, lead poisoning in human causes severe harm to the kid-ney, nervous system, liver and brain. Moreover, chronic exposureto lead has been correlated with sterility, abortion, stillbirths andneo-natal deaths [7]. Although copper is one of the essential nutri-ents for good health, an excess of the metal can cause Wilson’s dis-ease [8]. The current allowable maximum contamination level(MCL) of lead, arsenic, and copper in drinking water as set by theUnited States Environmental Protection Agency (USEPA) are15 ppb (ppb), 10 ppb, and 1.3 ppm (ppm), respectively [9].

The most common methods, which have been proposed to re-move arsenic and heavy metal ions from an aqueous medium, con-tain solvent extraction, ultra-filtration, reverse osmosis, chemicalprecipitation, nanofiltration, ion exchange and adsorption [10–12]. Among these methods, adsorption is the most preferred meth-od since it is highly efficient technique with particularly low oper-ational cost and widely applicable range of adsorbents. Thematerials that are extensively used for the removal of arsenic, lead,and copper from water may comprise activated carbon, ion-ex-change resins, red mud, activated alumina, biomass, chitosan,and carbon nanotubes [13–18].

The use of nanoparticles in water treatment has received a sub-stantial interest due to their high specific surface area, mobility,and activity [19,20]. Remarkably, titanium dioxide (TiO2), whichhas many properties such as non-toxicity, relative cheapness andhydrophilicity, is becoming a new class of adsorbent material inthe removal of organic and inorganic pollutants from water[21,22]. Previous studies interpreted arsenic and heavy metalsadsorption onto commercial TiO2 containing both rutile and ana-tase crystalline forms of TiO2 [23]. Anatase crystalline form ofTiO2 is well known as a matter with a high photocatalytic activitydue to its strong oxidizing power and favorable band gap energy.As anatase is illuminated, electrons and holes are generated andvarious chemical redox reactions are initiated due to the migrationof those pairs to the anatase surface [24]. The electrons placed onthe surface can promote the hydroxyl formation at the interfacebetween surface and water [25]. Then, the formation of hydroxylgroups can increase the adsorption rate of metals from water.Moreover, the inherent capability of TiO2 to convert As(III) toAs(V) by photo-oxidation can provide an extra advantage [23].Considering these properties, anatase crystalline type of TiO2 canbe used as sorbent material for the removal of contaminants fromwater.

This paper presented a successful synthesis of anatase nanoad-sorbent, and demonstrated fast and effective Pb(II), Cu(II) andAs(III) removal from synthetically polluted water with the utiliza-tion of the nanoadsorbent. To the best of our knowledge, this is acomprehensive work about the assessment of the synthesizedphase pure anatase nanoadsorbent for the removal Pb(II), Cu(II)and As(III) considering kinetics, equilibrium and thermodynamicproperties.

2. Materials and methods

2.1. Synthesis of anatase nanoadsorbent

Anatase nanoadsorbent was synthesized by sol–gel method atlow temperature using titanium tetraisopropoxide (TTIP) (C12H28-

O4Ti, 97%). A metal organic solution was prepared by mixing TTIP(10 mL) with 30 mL of 2-propanol (C3H8O, 99%) in a 250 mL flask

and was kept stirring for 5 min under air. Secondly, the mixtureof distilled water and 2-propanol was prepared separately with avolume ratio of 1:2. Afterwards, the metal organic solution wasslowly added to the vigorous stirred second solution. After comple-tion of the addition, the mixture was continued to mix for 2 h at40 �C. The sample was dried for several hours at 100 �C and an-nealed at 400 �C for 3 h.

2.2. Characterization

X- ray diffraction (XRD) measurement was done with a Bruke-raxs advance powder diffractometer fitted with a Siemens X-raygun and equipped with Brukeraxs Diffrac PLUS software. The sam-ple was swept from 2h = 10� through to 70�. The X-ray generatorwas set to 40 kV at 40 mA. Raman spectrum was obtained froman InVia Reflex Raman microscope and spectrometer using a532 nm diode laser excitation. Leo Supra 35VP field emission scan-ning electron microscope, Leo 32 and electron dispersive spec-trometer software was used for images and analysis. Imagingwas generally done at 2–5 keV accelerating voltage, using the sec-ondary electron and in lens imaging techniques. The XPS measure-ments were performed using a Thermo K-alpha X-rayphotoelectron spectrometer with a monochromated Al Ka radia-tion and a combined low-energy electron/ion flood gun for chargeneutralization. The binding energy scale for final calibration wascorrected by the C1s peak to 285.0 eV.

2.3. Batch adsorption experiments

The stock solutions containing 1000 mg/L of each lead and cop-per were prepared separately by dissolving 1.599 g Pb(NO3)2

(99.9%) and 2.9515 g Cu(NO3)2 (99.9%) in 1 L of deionized water.On the other hand, a standard arsenic solution (1000 mg/L) wasprepared by dissolving 1.320 g As(III) (As2O3, 99.9%) in 10 mL 5%(w/v) NaOH and making up to 1 L with deionized water. Batchadsorption experiments were carried out in 100 mL flasks, eachcontaining 30 mL Pb(II), Cu(II) or As(III) ions solutions with appro-priate concentrations. After the addition of 0.015 g of anatasenanoadsorbent each, the bottles were shaken in a controlled incu-bator shaker at 150-rpm and 25 �C for 24 h. The kinetic experimentwere carried out at different time intervals (5 min to 36 h) in aninitial metal ion concentrations of 10 mg/L Pb(II), Cu(II), and As(III)in the solution at pH 6.0. Several initial concentrations rangingfrom 0.1 to 20 mg/L initial Pb(II), Cu(II), and As(III) of homogenousaqueous solutions were used for equilibrium studies at fixed pH6.0 for 12 h. The effect of pH on the adsorptions were studied byusing separately prepared 10 mg/L of Pb(II), Cu(II), and As(III) solu-tions in a pH range of 3.0–9.0 at 25 �C. In the effect of temperatureon Pb(II), Cu(II), and As(III) adsorptions, the temperatures werecontrolled to 298 K, 308 K and 318 K, respectively. At the end ofeach experiment the solution was separated from the solid adsor-bent by using 0.45 lm PVDF membrane filter. Pb(II), Cu(II), andAs(III) concentrations of the solutions were measured with a Var-ian, Vista-Pro CCD simultaneous inductively coupled plasma ICP-OES spectrophotometer. Samples before and after adsorptionexperiments were analyzed to obtain residual adsorbate ion con-centrations. The blank experiments including different adsorbateconcentrations were performed in the absence of the adsorbentto verify that no chemical precipitation occurred in the batchadsorption studies. In order to test the reproducibility, the experi-ments were carried in triplicate and average results were reported.

Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635 627

3. Results and discussion

3.1. Characterization of the nanoadsorbent

Fig. 1a showed that XRD pattern of the anatase nanoadsorbentobtained by sol–gel synthesis. The characteristic diffraction peakswere indexed at 2h of 25.2� (101), 37.8� (004), 48.0� (200),54.8� (211), and 62.6� (220) planes of anatase phase of TiO2,respectively [26]. No other irrelevant peaks were observed indicat-ing the sample was pure in anatase phase. The average size of theanatase crystallite was calculated by applying the Debye–Scherrerformula [27] to the main anatase (101) peak in XRD spectrum andthe average crystallite size was found to be 7.97 nm.

Raman spectroscopy provides a rapid way of obtaining the sur-face crystal structure of the TiO2. The Raman spectrum of theadsorbent material used in here was shown in Fig. 1b, with thecharacteristic Raman lines giving evidence for the crystalline typeof anatase rather than rutile and brookite. Anatase has six Ramanactive modes (A1g + 2B1g + 3Eg) and the Raman lines observed at152.0 cm�1, 399.0 cm�1, 516.0 cm�1, and 635.0 cm�1 assigned tothe Eg, B1g, A1g, and Eg vibrational modes of anatase, respectively[22,27]. The conclusion obtained from Raman spectra of samplewas consistent with the XRD result indicated in Fig. 1a.

The SEM (Fig. 1c and d) micrographs clearly indicated sphericalmorphology. The anatase nanoadsorbent was composed of spheri-cal particles that were 20–60 nm in diameter. The observed parti-

Fig. 1. (a) XRD pattern of (b) Raman spectra of (c) SEM image (secondary electron detect

cle sizes in SEM were found to be larger than the crystallite sizecalculated from the XRD line broadening due to agglomeration ofthe primary particles after the calcination. Similar finding was alsofound in previous works [28].

XPS, a highly surface selective technique, can provide quantita-tive chemical information as well as oxidation and structural envi-ronment of elements except hydrogen and helium. A typical widescan XPS survey spectrum of the anatase nanoadsorbent wasshown in Fig. 2a and the distinct photoelectron peaks were ob-served for Ti2p and O1s peaks of the titanium dioxide. The totalsurface concentrations of Ti, O and C were 30.3, 62.3, and7.4 at.%, respectively. The titanium to oxygen ratio was 0.49:1,which is close to that expected from the stoichiometry of TiO2. AC1s peak was seen due to the deposition of adventitious hydrocar-bon on the surface of sample from the pumps evacuating the anal-ysis chamber [22]. The Ti2p and O1s narrow scans were illustratedin Fig. 2b and c. The Ti2p1/2 and Ti2p3/2 peaks were observed at458.7 and 464.4 eV with a split of 5.7 eV between the doublets,indicating that Ti exists in the Ti4+ form [29] The O1s XPS spectrademonstrated at least two O chemical states according to the bind-ing energy range of 530.0–531.2 eV. The main contribution wasattributed to Ti–O in the nanoadsorbent. The other minor peakwas assigned to the OH groups presented on the surface of nanoad-sorbent. The difference between the binding energies of the as-signed oxide (O2�) and hydroxyl (OH) species was 1.2 eV, whichis close to reported differences of 1.5–1.9 eV in the literature [30].

or) and (d) SEM image of (in lens detector) the synthesized anatase nanoadsorbent.

Fig. 2. XPS spectra of (a) survey scan, (b) Ti2p, and (c) O1s of anatase nanoadsor-bent surface at pH 6.0.

Fig. 3. Kinetic modeling of sorption of (a) As(III), (b) Pb(II), and (c) Cu(II) ontosynthesized anatase nanoparticles (temperature = 25 �C, pH = 6.0, S:L = 500 mg/L).

628 Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635

3.2. Kinetic studies and mathematical modeling

The kinetic profiles of Pb(II), Cu(II), and As(III) adsorption at pH6.0 were illustrated in Fig. 3. It was noted that at the beginning of

the experiment the adsorption percentages of arsenic and heavymetal ions increased remarkably and then reached equilibrium.As can be seen in Fig. 3a, there was a substantial increase in the

Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635 629

As(III) adsorption percentage when contact time was increasedfrom 5 to 180 min and the maximum As(III) uptake was obtainedat 480 min of contact. On the other hand, Fig. 3b and c indicatedthe rapid adsorptions of Pb(II) and Cu(II) ions within the first5 min of contact time because of the high affinity to the sorptionsites. The fast adsorption trend at initial stage is mostly observedphenomenon in the removal of contaminants from water and itis most likely caused reactive chemical groups readily presentedon the surface [31]. After initial fast reactions of adsorbate ionswith the nanoparticles, the slower reactions were observed espe-cially in the cases of removal of Cu(II) and As(III). Such biphasicbehavior was likely resulted from the variety of reactive sorptionsites and/or the diffusion of adsorbate ions on the surface for seek-ing available sites [32].

In order to investigate the rate-controlling step for the sorptionprocesses, four kinetic models were studied by applying the exper-imental sorption data to the pseudo-first-order equation (Eq. (1))[33], the pseudo-second-order equation (Eq. (2)) [34], the Elovichequation (Eq. (3)) [35] and intraparticle diffusion (Eq. (4)) [36].The linear mathematical forms of kinetics equations used areshown below:

logðqe � qtÞ ¼ logðqeÞ �k1

2:303t ð1Þ

tqt¼ 1

k2q2eþ t

qeð2Þ

qt ¼1b

lnðabÞ þ 1b

t ð3Þ

qt ¼ kintt0:5 þ C ð4Þ

where qt is the amount of adsorbed contaminant (mg/g) at time t, qe

is the maximum adsorption capacity (mg/g) for the pseudo-first-or-der adsorption and pseudo-second-order adsorption, k1 is the pseu-do-first-order rate constant for the adsorption process (1/min), k2 isthe pseudo-second-order rate constant (g/mg min), constant a (mg/(g min)) is the initial adsorption rate, b (g/mg) is related to surfacecoverage, and kint is the constant for the particle diffusion rate (mg/g min1/2). The fitting ability of kinetic and adsorption models wereanalyzed according to the average percent error (APE) given inbelow:

APE ¼Pn

i¼1qexp�qcalc

qexp

N� 100 ð5Þ

where N is the number of data points. The subscripts ‘‘exp’’ and‘‘calc’’ show the experimental and calculated values. Error function

Table 1The kinetic sorption modelling parameters of Pb, Cu, and As on the anatase nanoparticles

Model Parameters

Pseudo-first-order k1 (1/min)qe1 (mg/g)R2

APE

Pseudo-second-order k2 (g/mg min)qe2 (mg/g)R2

APE

Elovich equation b (g/mg)a (mg/g min)R2

APE

Intraparticle diffusion kint (mg/g min1/2)R2

APE

was employed in this study to find out the most suitable isothermmodel to represent the experimental data. The APE minimizes thefractional error distribution across the entire concentration range[37]. Small values of APE represent better fits of the model to thedata. The sorption modeling parameters and APE values were sum-marized in Table 1.

The pseuodo-second-order model appeared to be the better-fit-ting model for the removal of Pb(II), Cu(II), and As(III) because ithas the highest R2 (0.999, 0.999, 0.998) and lower APE (2.605,1.139, 5.178) values. Good agreements with the pseudo-second-or-der model suggested that chemical adsorption was the rate-limit-ing step [34] and the adsorption rates were proportional to thesquare of the number of sites on the adsorbent surface. On theother hand, the higher R2 values found in the Elovich model forCu(II) and As(III) indicated that the adsorbing surface was heterog-onous and therefore exhibited different activation energies forchemisorption. The constant a is related to the rate of chemisorp-tion [37] and higher value for a was obtained for Cu(II) indicating afast kinetic process. For comparison purposes, kinetic data werealso evaluated using the intraparticle diffusion equation. Theintraparticle diffusion can be considered as the rate-determiningstep, if the linearized curve passes through the origin (0,0).Although, the obtained R2 values were comparably high with thepseudo-second-order model for all three absorbates, the resultsof fitting did not pass through the origin (0,0) interpreting thatintraparticle diffusion was not the rate-determining step in theadsorption processes. Such a finding was similar to that made inprevious works on adsorption [38]. Moreover, due to the apparentlack of linear behavior, the pseudo-first-order, which considersspecifically single class of sorbing sites, was not suitable to de-scribe the kinetic profiles. According to the results in Table 1 ob-tained from the above kinetic studies and based on the R2 andAPE values considering the entire contact time for the three con-taminants, it can be concluded that the pseudo-second-order andElovich models best explained the Cu(II) and As(III) adsorptionkinetics, whereas the pseudo-second-order model described thePb(II) adsorption kinetics. Furthermore, the experimental dataindicated better agreement for models that describe adsorptionto heterogeneous surfaces except for Pb(II) adsorption.

3.3. Adsorption isotherms

The Pb(II), Cu(II), and As(III) sorption studies onto the nanopar-ticles were investigated by varying the initial sorbate concentra-tions ranging from 0.1 to 20 mg/L at pH 6.0. The nonlinear Pb(II),Cu(II), and As(III) adsorption isotherms were given for the anatasenanoparticles in Fig. 4. The maximum adsorption capacities were

.

Pb(II) Cu(II) As(III)

0.001 0.001 0.0031.198 8.356 3.1190.539 0.493 0.64698.17 97.98 64.29

0.021 0.027 0.00515.79 12.53 9.9800.999 0.999 0.9982.605 1.139 5.178

2.306 1.840 0.9114506 1289 13.160.917 0.986 0.9390.002 0.104 0.452

22.04 27.528 55.200.986 0.919 0.9900.001 0.080 0.069

Fig. 4. Adsorption isotherm of (a) As(III), (b) Pb(II), and C) Cu(II) adsorbed bysynthesized anatase nanoparticles (initial concentration = 0.3–20 mg/L, tempera-ture = 25 �C, pH = 6.0, S:L = 500 mg/L).

Table 2Langmuir, Freundlich, Redlich–Peterson, and Sips, isotherm parameters for Pb(II),Cu(II), and As(III) removal on the anatase nanoparticles.

Isotherm parameters Pb(II) Cu(II) As(III)

qexp (mg/g) 31.25 23.74 16.95

Langmuirqmon (mg/g) 31.05 23.09 16.83b (L/mg) 13.42 1.968 3.644R2 0.999 0.929 0.986APE 0.696 36.98 39.62

FreundlichKF (mg/g) 30.17 9.948 8.7291/n 0.539 0.379 0.354R2 0.836 0.989 0.951APE 26.39 1.139 5.393

Redlich–PetersonKRP (L/g) 90.09 3.966 2.490

aRP ðL=mgÞbRP25.31 0.280 277.1

bRP 0.876 0.331 0.743R2 0.995 0.968 0.981APE 0.701 63.75 7.051

SipsKLF (L/g) 31.94 25.71 21.99aLF (L/mg) 9.859 1.192 1.244nLF 0.950 0.592 0.588R2 0.999 0.987 0.965APE 0.014 3.377 25.43

630 Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635

found 31.25 mg/g for Pb(II), 23.74 mg/g for Cu(II), and 16.95 mg/gfor As(III). In order to investigate adsorption performance, Lang-

muir [39], Freundlich [40], Redlich–Peterson [41], and Sips [42],(Langmuir–Freundlich) isotherm models were utilized. The calcu-lated isotherm parameters of all four models were summarizedin Table 2. Langmuir model assumes that single specie of the sor-bate adsorbs on specific homogenous adsorption sites within thesorbent by forming monolayer coverage. However, Freundlichmodel is derived considering a theory of the heterogeneousadsorption sites and interactions [39,40]. Langmuir (Eq. (6)) andFreundlich (Eq. (7)) isotherms can be expressed by the followingequations:

qe ¼qmonbCe

1þ bCeð6Þ

qe ¼ KFC1=ne ð7Þ

where qe is the solute amount adsorbed per unit weight of adsor-bent (mg/g), Ce is liquid phase sorbate concentration in equilibrium,qmon is the monolayer adsorption capacity (mg/g), KF indicates theadsorbent’s relative adsorption capacity (mg/g), 1/n is the constantrepresenting adsorption intensity and b is the adsorption equilib-rium constant (L/mg) related to adsorption enthalpy(b / exp�DH/RT).

The Redlich–Peterson isotherm model, which includes featuresof both the Langmuir and Freundlich equations, can be described asfollows:

qe ¼KRPCe

1þ aRPCbRPe

ð8Þ

where KRP is Redlich–Peterson isotherm constant (L/g), aRP is Red-lich–Peterson isotherm constant ðL=mgÞbRP and bRP is the exponentwhich lies between 0 and 1. The bRP parameter (dimensionless) rep-resents the heterogeneity of the binding surface. If bRP tends tounity Redlich–Peterson model results in Langmuir model. WhenbRP tends to zero Redlich–Peterson model represents Henry law[41]. The Sips isotherm model, which is combined form of Langmuirand Freundlich equations, possesses a finite saturation limit whenthe initial sorbate concentration is sufficiently high. The nonlinearform of Sips sorption isotherm is:

Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635 631

qe ¼KLFCnLF

e

1þ ðaLFCeÞnLFð9Þ

where KLF, aLF, and nLF are the Sips isotherm constant. The differencebetween this equation and Langmuir equation is the additional nLF

parameter in the Sips equation. If nLF parameter is unity, the equa-tion effectively reduces to the Langmuir equation appropriate forideal surfaces. On the other hand, the deviation further away fromunity indicates heterogeneity of the system.

Considering the APE error percentages and R2 values, theadsorption isotherm behavior of Pb(II) was better fitted to Sips iso-therm, followed by Langmuir isotherm. The calculated nLF valuewas found to close to unity, therefore, the Langmuir model suc-cessfully explained the adsorption of Pb(II) ions onto the nanopar-ticles. Additionally, the suitability of Langmuir isotherm model tothe Pb(II) sorption data was also supported by the obtained dimen-sionless parameter value in Redlich–Peterson, which was also closeto unity. On the other hand, the deviation of nLF parameter furtheraway from unity indicates heterogeneity of the system. This wasthe case of adsorption of Cu(II), which fitted better to the Freund-lich model, followed by the Sips model. However, the system het-erogeneity might be resulted from the sorbate or sorbent or acombination of both. Moreover it can be seen that As(III) fitted bet-ter to the Langmuir isotherm model followed by Redlich–Petersonisotherm model according to R2 values. However, Langmuir modelwith the high APE error reduced the applicability of model to theexperimental data. Therefore, the lower APE error in betweenexperimental and calculated data was required to obtain theappropriate evaluation for the adsorption of As(III). As can be seenin Table 2, the Freundlich and Redlich–Peterson isotherm modelsindicated a good agreement according to APE error percentages.The Redlich–Peterson isotherm equation reduced to Freundlichisotherm equation since the calculated aRP value for As(III) adsorp-tion was large enough. It can be concluded that the Langmuir mod-el was more suitable for Pb(II) uptake supported with lower APEand higher R2 value, in contrast the higher correlation betweenexperimental and calculated adsorption data of Cu(II) and As(III)were observed in Freundlich isotherm model. Moreover, the KF val-ues achieved from the Freundlich model were 9.948 mg/g for Cu(II)and 8.729 mg/g for As(III), suggesting higher binding affinity ofCu(II) as compared to As(III).

3.4. Adsorption thermodynamics

In order to analyze the effect of temperature, experiments werecarried out at temperatures of 25, 35, and 45 �C and the followingequations were used to calculate the thermodynamic parameters:

Kc ¼qe

Ceð10Þ

DG ¼ �RT ln Kc ð11Þ

ln Kc ¼DSR� DH

RTð12Þ

where Kc is an equilibrium constant between the adsorbed equilib-rium concentration and the aqueous equilibrium concentration (l/

Table 3Thermodynamic parameters for Pb(II), Cu(II), and As(III) sorption on the anatase nanopart

Temperature (�C) Pb(II) Cu(II)

25 35 45 25

DG (kJ/mol) �14.4 �17.9 �21.5 �6.56DH (kJ/mol) 91.76DS (kJ/mol K) 0.356

g). Furthermore, DH (kJ/mol) and DS (kJ/mol K), which are the var-iation of enthalpy and entropy, were calculated (Table 3) from theslope and the intercept of lnKC versus 1/T graphs.

It has been stated that the value of DG found below 8 kJ/mol isconsistent with a physical adsorption involving electrostatic inter-action between adsorption sites and the adsorbate ion. Moreover,it has been noted that if the value of DG is in between 8 and16 kJ/mol, the adsorption proceeds via a chemical ion exchangemechanism. The more negative DG values imply chemisorptionmechanism by forming stable complexes on the adsorbent surface[43]. According to the results presented in Table 3, the values of DGfor Pb(II) were found to range from �14.4 to �21.5 kJ/mol suggest-ing chemical adsorption as major mechanism responsible for theadsorption process. In the case of Cu(II) adsorption, the most prob-able type of reaction would be ion-exchange according to the val-ues of DG. Furthermore, the calculated DS values were positive andthe favorable Gibbs energies were entropically controlled since theadsorptions of Pb(II) and Cu(II) were endothermic signifying thedecrease in uptake capacity by increasing the temperature. Dueto increase in the randomness of the system at the adsorbate–adsorbent interface, the positive values for DS were observed forPb(II) and Cu(II) together with the structural change during theadsorption process. In the case of As(III), physical adsorption mightbe the main mechanism since the calculated DG values were in therange of electrostatic interactions. Especially, the negative DH va-lue of As(III) might be suggested that the physical adsorptionwas the leading mechanism of the adsorption.

The negative DG values were obtained for Cu(II), Pb(II), andAs(III) adsorptions on the adsorbent by confirming the feasibilityof the adsorption processes and the spontaneous nature of adsorp-tions. Moreover, the numerical values of 1/n which is calculatedfrom the linearized form of Freundlich model was found between0 and 1, indicated beneficial adsorptions [20]. In addition to that,from the adsorption equilibrium constant in Langmuir model, thedimensionless separation factor (RL), which was the indicative ofthe isotherm shape, could be obtained by using the followingequation:

RL ¼1

1þ bCoð13Þ

where Co is the initial absorbate concentration (mg/L). The value ofRL demonstrates the nature of adsorption as unfavorable (RL > 1),linear (RL = 1), favorable (0 < RL < 1) and irreversible (RL = 0). TheRL were found to be 0.427 to 0.0037 for Pb(II), 0.8356 to 0.0248for Cu(II) and 0.4777 to 0.01353 respectively for As(III) indicatingirreversible adsorptions of the sorbates on the anatase nanoadsor-bent with the increase of initial sorbate concentration.

3.5. Effect of solution pH

The uptakes of Pb(II), Cu(II), and As(III) onto the anatase nano-particles for an initial concentrations of 10 mg/L within a pH rangeof 3.0–9.0 were shown in Fig. 5a. The pH is a significant parameterfor determining the adsorption between the adsorbent and aque-ous interface. At pH 3.0, the adsorption percentages of the nanoad-sorbent to Pb(II), Cu(II), and As(III) were 29.73%, 1.15%, and 51.76%,

icles.

As(III)

35 45 25 35 45

�9.54 �12.5 �4.63 �4.08 �3.5319.36 �21.200.117 �0.055

Fig. 5. (a) Adsorption of 10 mg/L Pb(II), Cu(II), and As(III) as a function of pH by500 mg/L synthesized anatase nanoparticles 25 �C, and (b) pHfinal versus pHinitial

during the adsorption.

632 Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635

respectively. From pH 3.0 to 8.0, the uptake percentages of con-taminants increased, 97.41% of Pb(II), 62.57% of Cu(II) and 49.29%of As(III) were removed at pH 6.0. Similar findings were also foundin literature [17,20,25]. Compare to Pb(II) and Cu(II), the effect ofpH on As(III) sorption on the anatase nanoadsorbent was nearlynegligible in the pH range of 3.0–8.0 [23].

According to surface complex formation theory (SCF), the in-crease in the pH value leads to the decrease in the competition be-tween proton and metal ions for the adsorption sites [44]. At acidicpH conditions surface of the anatase nanoparticles was mostly cov-ered with positively charged surface groups, leading to more cou-lombic repulsion of the metal ions. Additionally, at lower pH valuesheavy metal ions possibly found less chance to form a complexwith the adsorbent material due to the competition with hydrogenand heavy metal ions for the sorption sites. At higher pH values,the surface of adsorbent was covered with more negative chargesand more sorption sites were available to attract the heavy metalions. However, Pb(II) and Cu(II) ions could begin to precipitate asPb(OH)2 and Cu(OH)2 at pH > 6.0. In the case of As(III) adsorption,the amount of negatively charged As(III) increased when the pHvalue was shifted to higher pH values [23]. Therefore, at higherpH value was not suitable for efficient As(III) adsorption onto thenanoparticles. At pH values from 3.0 to 8.0, the neutral As(III)

species mostly existed and the interaction of neutral species withsome hydroxyl groups presented on the surface of the nanoparticlecould be observed [45].

Variations of pHf against pHi of batch experiments for Pb(II),Cu(II), and As(III) sorptions were illustrated Fig. 5b. For As(III)adsorption, it has been seen that the solution pHf was greater thanthe pHi in the range of 3.0–8.0. However, after that, the final pH ofthe solution was decreased during sorption of As(III) by anatasenanoadsorbent. This was likely due to the uptake of H+ ions bythe solid surface at pHi ranging from 3.0 to 8.0. In addition to that,the lowering of pHf than pHi can be ascribed by the releasing H+

ions. In the case of Pb(II) and Cu(II) adsorption, the pHf were higherthan pHi until pH 5.0. However, the decrease in final pH valueswere observed due to liberating H+ ions after the reaction betweenthe heavy metal ions with anatase nanoadsorbent.

3.6. Proposed binding mechanism

The binding mechanisms of Pb(II), Cu(II), and As(III) to the ana-tase nanoparticles were mainly depended on the type of activesites on the surface of adsorbent and the concentrations of the pro-duced sorbate species at measured pH value. In literature, the pointof zero charge (pzc) for the anatase nanoparticles was approxi-mately found in between 5.9 and 6.9 [46]. The anatase nanoparti-cles are composed of the mixture of positively charged TiOHþ2 andneutral species of TiOH0 when pH is below. However, the primaryspecies of the nanoparticles include the neutral species of TiOH0

and the negatively charged TiO� when pH is above pHpzc [23,47].Therefore, the surface of nanoparticles mostly covered with theneutral TiOH0 and negatively charged TiO� at pH 6.0. Bonatoet al. presented experimental study in order to determine the ac-tive lead species in aqueous solution as a function of pH. In aque-ous solution at pH < 10, the main lead species were Pb2+, PbOH+

and Pb3ðOHÞ2þ4 . Additionally, the anionic PbðOHÞ�3 was only foundunder alkaline conditions. They observed that Pb(OH)2(aq.) wasthe prevalent molecule at pH 10, but due to the low solubility con-stant of Pb(OH)2(s) it precipitated at high lead concentration (whenthe total lead concentration was greater than 20.72 mg/L) [48].According to study conducted by Deng et al., the hydrolysis ofCu(II) to form Cu2ðOHÞ2þ2 and Cu(OH)+, Cu(OH)2, and Cu(OH)3� spe-cies occurred in dilute solution with increasing pH from 8 to 12when the total copper concentration was 100 mg/L). Additionally,CuðOHÞ24� was observed in the more alkaline solutions [49]. AtpH 6.0, the predominant specie of lead and copper was Me2+, fol-lowed by Me(OH)+ [46,48]. The stable complexes might easily formbetween the metal cations/hydroxyl metal cations and active sur-face sites. At pH 6.0–9.4, the available As(III) species were the com-bination of H3AsOo

3 and H2AsO�3 [45]. The negatively chargedsurface would not tend to attract the negatively charged As(III)ions, therefore the less favorable interactions were observed inthe case of As(III) adsorption.

Besides the influence of concentrations of absorbate species andactive sorbing sites on the sorption, the defined kinetic and adsorp-tion modeling could be supportive evidence in order to compre-hend the adsorption mechanism. Kinetic sorption analysisindicated that adsorption of Pb(II), Cu(II), and As(III) were betterdescribed by pseudo-second-order that assumes more than onebinding pathway. Essentially, the adsorbed amount of Pb(II) andCu(II) increased through the basic pH values and this supportedthe formation of surface complexation rather than electrostaticinteractions with surface sites [31]. In addition to that, from thecalculated dimensionless separation factor (RL) for Pb(II), the moreirreversible adsorptions were observed and this could be the indi-cation of chemical adsorption. In contrast, heterogeneity of the sur-face due to involvement of both strong and weaker binding sitesfor adsorption were favored for Cu(II) with a higher suitability to

Fig. 6. XPS spectra of O1s after (a) Pb(II), (b) Cu(II), (c) As(III) adsorption at pH 6.0.

Fig. 7. XPS (a) Pb 4f, (b) Cu 2p, (c) As 3d XPS core-level spectra on the anatasenanoadsorbent surface after adsorption at pH 6.0.

Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635 633

the Freundlich isotherm model [47]. Apart from that, according tothe obtained DG values for Cu(II), the removal of absorbate fromsolution with the nanoparticles was mainly governed by ion ex-change mechanism rather than physical adsorption, and chemicaladsorption might also contribute to the adsorption of Cu(II). Inthe case of As(III) adsorption, the uptake capacity was remain al-most constant as the pH value was changed from 3.0 to 8.0 and

the reason might be related to low affinity of the available As(III)species towards the active surface sites.

Fig. 8. The proposed reaction mechanism of Pb(II), Cu(II), and As(III) on the anatase nanoadsorbent surface.

634 Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635

To further probe into the removal mechanisms of Pb(II), Cu(II),and As(III) by the anatase nanoadsorbent, XPS analyses on thenanoadsorbent were carried out after adsorptions of arsenic andheavy metals. The O1s narrow scans were illustrated in Fig. 6.The spectra were divided into two peaks positioned at 530.0 and531.3 eV, which could be assigned to metal oxide (M–O) and hy-droxyl bonded to metal (M–OH), respectively [50]. After adsorp-tion, it was found that the area ratio of the peak at 530.0 eVassigned to M–O increased due to the formation of Ti–O–R (R:functional groups), Pb–O, Cu–O or As–O groups on the anatase sur-face. On the other hand, the peak area ratio of the oxygen at 531.3decreased after Pb(II), Cu(II), and As(III) adsorption, suggesting thathydroxyl groups created on the surface of anatase participated inthe adsorption.

Fig. 7a illustrated the Pb4f spectrum of the adsorbent afterPb(II) adsorption, and the to Pb4f7/2 and Pb4f5/2 binding energiesof 138.8 eV and 143.6 eV were attributed the PbO2 and PbO forms[48,50], respectively. Fig. 7b showed the XPS spectrum of Cu2p re-gion. The XPS spectrum of Cu2p3/2 indicates that there are twocomponents after deconvolution, attributed to Cu2O (932.5 eV)and CuO (934.3 eV) on the surface of anatase nanoadsorbent,respectively. Peaks at 941.3 and 943.8 eV were assigned to CuOsatellites [51]. The As 3d spectrum of adsorbent after As(III)adsorption was shown in Fig. 7c, and only one peak observed at44.6 eV ascribing to As(III) species verified its successful bindingto the adsorbent surface [51]. This result indicated that As(III)was not oxidized into As(V) during the sorption process. Accordingto the XPS analysis, the metal oxide and hydroxyl groups on theanatase nanoadsorbent surface were responsible for Pb(II), Cu(II),and As(III) sorption. Those groups on the surface of nanoadsorbentcould react with arsenic and heavy metals directly to form stableinner-sphere complexation and outer-sphere complexationthrough electrostatic binding reactions. Moreover, it was supposedthat Pb(II), Cu(II), and As(III) could adsorb onto the nanoadsorbentby forming surface complexes of monodentate and bidentate. Thepossible sorption mechanisms are illustrated in Fig. 8.

4. Conclusion

The synthesized anatase nanoparticles indicated greater poten-tial for the removal of Pb(II) and Cu(II) from drinking water. Addi-tionally, the nanoparticles were able to reduce As(III)concentrations below the MCL requirements for drinking waterup to moderate initial concentrations. The amount of Pb(II) and

Cu(II) ions removed from water was strongly influenced by thepH and the type of reactive adsorbate species and active sites onthe surface of adsorbent. In the case of As(III) adsorption, the up-take percentages were almost same upon entire pH value. Basedon the R2 and APE values, the pseudo-second-order model de-scribed the Pb(II), Cu(II), and As(III) adsorption kinetics. Langmuirmodel fitted well for Pb(II) uptake supported with lower APE andhigher R2 value, on the other hand the higher relationship betweenexperimental and calculated adsorption data for Cu(II) and As(III)were observed in Freundlich isotherm model. Based on experimen-tal data, the maximum adsorption capacity for Cu(II), Pb(II), andAs(III) ions were 31.25 mg/g, 23.74 mg/g, and 16.95 mg/g respec-tively. Thermodynamic parameters indicated that the calculatedDS values were positive and the favorable Gibbs energies wereentropically controlled in the adsorptions of Pb(II) and Cu(II), whilethe As(III) adsorption was thermodynamically favorable withoutany limitations. XPS analyses revealed that the surface oxygen-containing functional groups including hydroxyl groups were in-volved in the adsorption process. Thus, the whole Pb(II), Cu(II),and As(III) sorption studies on the nanoparticles indicated thatthe synthesized nanoparticles can be effectively used in the treat-ment of water polluted with the arsenic and heavy metal ions.

Acknowledgements

We greatly acknowledge the support of Prof. Ismail Çakmak andICP Laboratory head technician Mr. Veli Bayır of the Biological Sci-ences and Bioengineering Program at Sabanci University for pro-viding us the measurement facilities of inductively coupledplasma (ICP) spectrometer.

References

[1] R.L. Calderon, The epidemiology of chemical contaminants of drinking water,Food Chem. Toxicol. 38 (2000) S13–S20.

[2] A. Naranjo-Pulido, H. Romero-Schmidt, L. Mendez-Rodriguez, B. Acosta-Vargas,A. Ortega-Rubio, Soil arsenic contamination in the Cape Region, BCS, Mexico, J.Environ. Biol. 23 (2002) 347–352.

[3] R.A. Isaac, L. Gil, A.N. Cooperman, K. Hulme, B. Eddy, M. Ruiz, K. Jacobson, C.Larson, O.C. Pancorbo, Corrosion in drinking water distribution systems: amajor contributor of copper and lead to wastewaters and effluents, Environ.Sci. Technol. 31 (1997) 3198–3203.

[4] V. Milu, J. Leroy, C. Peiffert, Water contamination downstream from a coppermine in the Apuseni Mountains, Romania, Environ. Geol. 42 (2002) 773–778.

[5] B. Benham, E.J. Ling, B. Wright, K. Haering, Virginia Household Water QualityProgram: Corrosive Household Water, Blacksburg, Virginia, 2011.

[6] A. Chatterjee, D. Das, B.K. Mandal, T.R. Chowdhury, G. Samanta, D. Chakraborti,Arsenic in groundwater of six districts of West Bengal, India: the biggest

Z. Özlem Kocabas�-Ataklı, Y. Yürüm / Chemical Engineering Journal 225 (2013) 625–635 635

arsenic calamity in the world. Part I. Arsenic species in drinking water andurine of the affected people, Analyst 120 (1995) 643.

[7] R.A. Goyer, I.J. Chisolon, Lead, in: Metallic Contamination and Human Health,Academic Press, New York/London, 1972.

[8] L. Leggio, G. Gasbarrin, G. Addolorato, Wilson’s disease, Lancet 369 (2007) 902.[9] National Primary Drinking Water Regulations, EPA Report No. EPA 816-F-09-

0004, Environmental Protection Agency, Washington, DC, May, 2009.[10] A. Maiti, S. DasGupta, J.K. Basu, S. De, Batch and column study: adsorption of

arsenate using untreated laterite as adsorbent, Ind. Eng. Chem. Res. 47 (2008)1620–1629.

[11] P.R. Kumar, S. Chaudhari, K.C. Khilar, S.P. Mahajan, Removal of arsenic fromwater by electrocoagulation, Chemosphere 55 (2004) 1245–1252.

[12] A. Figoli, A. Cassano, A. Criscuoli, M.S.I. Mozumder, M.T. Uddin, M.A. Islam, E.Drioli, Influence of operating parameters on the arsenic removal bynanofiltration, Water Res. 44 (2010) 97–104.

[13] A. Jusoh, L.S. Shiung, N. Ali, M.J.M.M. Noor, A simulation study of the removalefficiency of granular activated carbon on cadmium and lead, Desalination 206(2007) 9–16.

[14] Y. Kim, C. Kim, I. Choi, S. Rengaraj, J. Yi, Arsenic removal using mesoporousalumina prepared via a templating method, Environ. Sci. Technol. 38 (2004)924–931.

[15] Y.H. Li, F.Q. Liu, B. Xia, Q.J. Du, P. Zhang, D.C. Wang, Z.H. Wang, Y.Z. Xia,Removal of copper from aqueous solution by carbon nanotube/calciumalginate composites, J. Hazard. Mater. 177 (2010) 876–880.

[16] Z.O. Kocabas, Y. Yürüm, Kinetic modeling of arsenic removal from water byferric ion loaded red mud, Separ. Sci. Technol. 46 (2011) 2380–2390.

[17] M.F. Sawalha, J.R. Peralta-Videa, M. Duarte-Gardea, J.L. Gardea-Torresday,Removal of copper, lead and zinc from contaminated water by saltbushbiomass: analysis of the optimum binding, stripping, and binding mechanism,Bioresource Technol. 99 (2008) 4438–4444.

[18] J.S. Yamani, S.M. Miller, M.L. Spaulding, J.B. Zimmerman, Enhanced arsenicremoval using mixed metal oxide impregnated chitosan beads, Water Res. 46(2012) 4427–4434.

[19] Y. Jin, F. Liu, M. Tong, Y. Hou, Removal of arsenate bycetyltrimethylammonium bromide modified magnetic nanoparticles, J.Hazard. Mater. 227–228 (2012) 461–468.

[20] F. Ge, M.M. Li, H. Ye, B.X. Zhao, Effective removal of heavy metal ions Cd2+,Zn2+, Pb2+, Cu2+ from aqueous solution by polymer-modified magneticnanoparticles, J. Hazard. Mater. 211–212 (2012) 366–372.

[21] O. Carp, C.L. Huisman, A. Reller, Photoinduced reactivity of titanium dioxide,Prog. Solid State Chem. 32 (2004) 33–177.

[22] J.S. Dalton, P.A. Janes, N.G. Jones, J.A. Nicholson, K.R. Hallam, G.C. Allen,Photocatalytic oxidation of NOx gases using TiO2: a surface spectroscopicapproach, Environ. Pollut. 120 (2002) 415–422.

[23] G. Jegadeesan, S.R. Al-abed, V. Sundaram, H. Choi, K.G. Scheckel, D.D.Dionysiou, Arsenic sorption on TiO2 nanoparticles: size and crystallinityeffects, Water Res. 44 (2010) 965–973.

[24] J.C. Lee, M.S. Kim, B.W. Kim, Removal of paraquat dissolved in a photoreactorwith TiO2 immobilized on the glass-tubes of UV lamps, Water Res. 36 (2002)1776–1782.

[25] M.S. Kim, K.M. Hong, J.G. Chung, Removal of Cu(II) from aqueous solutions byadsorption process with anatase-type titanium dioxide, Water Res. 37 (2003)3524–3529.

[26] P.C.A. Alberius, K.L. Frindell, R.C. Hayward, E.J. Kramer, G.D. Stucky, B.F.Chmelka, General predictive syntheses of cubic, hexagonal, and lamellar silicaand titania mesostructured thin films, Chem. Mater. 14 (2002) 3284–3294.

[27] A.K.L. Sajjad, S. Shamaila, B. Tian, F. Chen, J. Zhang, One step activation of WOx/TiO2 nanocomposites with enhanced photocatalytic activity, Appl. Catal. B:Environ. 91 (2009) 397–405.

[28] A.K. Khanra, L.C. Pathak, S.K. Mishra, M.M. Godkhindi, Effect of NaCl on thesynthesis of TiB2 powder by a self-propagating high-temperature synthesistechnique, Mater. Lett. (2004) 733–738.

[29] J. Hu, Shipley HJ, Evaluation of desorption of Pb (II), Cu (II) and Zn (II) fromtitanium dioxide nanoparticles, Sci. Total Environ. 431 (2012) 209–220.

[30] E. Bedri, R.A. Hunsicker, G.W. Simmsons, E.D. Sudol, V.L. Dimonie, M.S. El-Asser, XPS and FTIR surface characterization of TiO2 particles used in polymerencapsulation, Langmuir 17 (2001) 2664–2669.

[31] J. Li, J. Hu, G. Sheng, G. Zhao, Q. Huang, Effect of pH, ionic strength, foreign ionsand temperature on the adsorption of Cu(II) from aqueous solution to GMZbentonite, Colloids Surf. A 349 (2009) 195–201.

[32] K.P. Raven, A. Jain, R.H. Loeppert, Arsenite arsenate adsorption on ferrihydrite:kinetics, equilibrium, and adsorption envelopes, Environ. Sci. Technol. 32(1998) 344–349.

[33] S. Lagergren, Zur theorie der sogenannten adsorption geloster stoffe kungligasvenska vetenskapsakademiens, Handlingar 24 (1898) 1–39.

[34] Y.S. Ho, G. Mckay, Pseudo-second order model for sorption processes, ProcessBiochem. 34 (1999) 451–465.

[35] S.H. Chien, W.R. Clayton, Application of Elovich equation to the kinetics ofphosphate release and sorption in soils, Soil Sci. Soc. Am. J. 44 (1980) 265–268.

[36] W.J. Weber Jr., J.C. Morris, Kinetics of adsorption on carbon from solution, J.Sani. Eng. Div. 89 (1963) 31–39.

[37] A.B. Pérez-Marín, V. Meseguer Zapata, J.F. Ortuño, M. Aguilar, J. Sáez, M.Lloréns, Removal of cadmium from aqueous solutions by adsorption ontoorange waste, J. Hazard. Mater. B 139 (2007) 122–131.

[38] L. Abramian, H. El-Rassy, Adsorption kinetics and thermodynamics of azo-dyeOrange II onto highly porous titania aerogel, Chem. Eng. J. 150 (2009) 403–410.

[39] I. Langmuir, The adsorption of gases on plane surfaces of glass, mica andplatinum, J. Am. Chem. Soc. 40 (1918) 1361–1403.

[40] G. McKay, Use of Adsorbents for the Removal of Pollutants from Wastewaters,CRC Press, Boca Raton, 1995.

[41] O. Redlich, D.L. Peterson, A useful adsorption isotherm, J. Phys. Chem. 63(1959) 1024.

[42] R. Sip, Combined form of Langmuir and Freundlich equations, J. Chem. Phys. 16(1948) 490.

[43] G.J. Copello, L.E. Diaz, V. Campo Dall’ Orto, Adsorption of Cd(II) and Pb(II) ontoa one step-synthesized polyampholyte: Kinetics and equilibrium studies, J.Hazard. Mater. 217-218 (2012) 374–381.

[44] D.A. Dzombak, F.M.M. Morel, Surface complexation modeling, in: HydrousFerric Oxide, Wiley, New York, 1990.

[45] V.M. Boddu, K. Abburi, J.L. Talbott, E.D. Smith, R. Haasch, Removal of arsenic(III) and arsenic (V) from aqueous medium using chitosan-coated biosorbent,Water Res. 42 (2008) 633–642.

[46] D. Vu, Z. Li, H. Zhang, W. Wang, Z. Wang, X. Xu, B. Dong, C. Wang, Adsorption ofCu(II) from aqueous solution by anatase mesoporous TiO2 nanofibers preparedvia electrospinning, J. Colloid Interface Sci. 367 (2012) 429–435.

[47] P.K. Dutta, A.K. Ray, V.K. Sharma, F.J. Millero, Adsorption of arsenate andarsenite on titanium dioxide suspensions, J. Colloid Interface Sci. 278 (2004)270–275.

[48] M. Bonato, K.V. Ragnarsdottir, G.C. Allen, Removal of uranium(VI), lead(II) atthe surface of TiO2 nanotubes studied by X-ray photoelectron spectroscopy,Water Air Soil Pollut. 223 (2012) 3845–3857.

[49] S. Deng, R. Bai, J.P. Chen, Behaviors and mechanisms of copper adsorption onhydrolyzed polyacrylonitrile fibers, J. Colloid Interface Sci. 260 (2003) 265–272.

[50] Y. Ren, N. Yan, J. Feng, J. Ma, Q. Wen, N. Li, Q. Dong, Adsorption mechanism ofcopper and lead ions onto grapheme nanosheet/d-MnO2, Mater. Chem. Phys.136 (2012) 538–544.

[51] X.Y. Yu, R.X. Xu, C. Gao, T. Luo, Y. Jia, J.H. Liu, X.J. Huang, Novel 3D hierarchicalcotton-candy-like CuO: surfactant-free solvothermal synthesis and applicationin As(III) removal, ACS Appl. Mater. Interfaces 4 (2012) 1954–1962.