lable at ScienceDirect

Journal of Environmental Management 130 (2013) 166e175

Contents lists avai

Journal of Environmental Management

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

Adsorption of Direct Blue 53 dye from aqueous solutions by multi-walled carbon nanotubes and activated carbon

Lizie D.T. Prola a, Fernando M. Machado b, c, Carlos P. Bergmann b, Felipe E. de Souza a,Caline R. Gally a, Eder C. Lima a, *, Matthew A. Adebayo a, Silvio L.P. Dias a, Tatiana Calvete d

a Institute of Chemistry, Federal University of Rio Grande do Sul (UFRGS), Av. Bento Gonçalves 9500, Postal Box 15003, 91501-970 Porto Alegre, RS, Brazilb Department of Material, Federal University of Rio Grande do Sul (UFRGS), Av. Osvaldo Aranha 99, Laboratory 705C, 90035-190 Porto Alegre, RS, Brazilc Área of Technological Sciences, (UNIFRA), R. dos Andradas 1614, 97010-032 Santa Maria, Brazild Universitary Center La Salle (UNILASALLE), Av. Victor Barreto 2288, 92010-000 Canoas, RS, Brazil

a r t i c l e i n f o

Article history:Received 27 May 2013Received in revised form6 August 2013Accepted 1 September 2013Available online

Keywords:Carbon nanotubesAdsorptionNonlinear isotherm fittingDirect Blue 53

* Corresponding author. Tel.: þ55 51 3308 7175; faE-mail addresses: eder.lima@ufrgs.br, profederlima

0301-4797/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.jenvman.2013.09.003

a b s t r a c t

Multi-walled carbon nanotubes (MWCNT) and powder activated carbon (PAC) were used as adsorbentsfor adsorption of Direct Blue 53 dye (DB-53) from aqueous solutions. The adsorbents were characterisedusing Raman spectroscopy, N2 adsorption/desorption isotherms, and scanning and transmission electronmicroscopy. The effects of initial pH, contact time and temperature on adsorption capacity of the ad-sorbents were investigated. At pH 2.0, optimum adsorption of the dye was achieved by both adsorbents.Equilibrium contact times of 3 and 4 h were achieved by MWCNT and PAC adsorbents, respectively. Thegeneral order kinetic model provided the best fit of the experimental data compared to pseudo-firstorder and pseudo-second order kinetic adsorption models. For DB-53 dye, the equilibrium data (298e323 K) were best fitted to the Sips isotherm model. The maximum sorption capacity for adsorption ofthe dye occurred at 323 K, with the values of 409.4 and 135.2 mg g�1 for MWCNT and PAC, respectively.Studies of adsorption/desorption were conducted and the results showed that DB-53 loaded MWCNTcould be regenerated (97.85%) using a mixture 50% acetone þ 50% of 3 mol L�1 NaOH. Simulated dyehouse effluents were used to evaluate the application of the adsorbents for effluent treatment (removalof 99.87% and 97.00% for MWCNT and PAC, respectively, were recorded).

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Many industries such as textile, feedstuffs, paper, leather, cos-metics, among others use dyes for colouring their final products(Alencar et al., 2012; Royer et al., 2010), and consequently producelarge amounts of dye-containing effluents. The presence of dyecontaining waters can adversely affect the aquatic environment byimpeding light penetration, precluding photosynthesis of aquaticflora (Cardoso et al., 2011, 2012). Moreover, most of the dyes cancause allergy, dermatitis, skin irritation and can also provoke cancerand cell mutation (de Lima et al., 2007). Synthetic dyes belong to aclass of organic compound with a complex aromatic molecularstructure that can provide bright and firm colour to industrialisedproducts. However, the complex aromatic molecular structures ofdyes make them more stable and difficult to biodegrade (Prolaet al., 2013). Therefore, effluents containing dyes require

x: þ55 51 3308 7304.@gmail.com (E.C. Lima).

All rights reserved.

treatment before being released into the environment (Cardosoet al., 2012; Dotto et al., 2012; Ai and Jiang, 2012).

One of the unitary operations mostly used for the removal ofsynthetic dyes from industrial effluents is the adsorption (Royeret al., 2010) due to its simplicity and high efficiency as well as theavailability of a wide range of adsorbents (Calvete et al., 2009;Konicki et al., 2012). This process transfers the contaminant fromthe effluent to a solid phase, which significantly decreases thebioavailability of the hazardous species to living organisms (Unur,2013; Machado et al., 2011). This decrease could be attributed tothe fact that the toxicity of dye dissolved in water is much higherthan the amount loaded in a solid surface (de Lima et al., 2007). Thedecontaminated effluent can then be released to the environment(Machado et al., 2012) or alternatively the water can be reutilised insome industrial processes that do not require water of high purity.Subsequently, the adsorbent can be regenerated or stored in a dryplace with no direct contact with the environment (Alencar et al.,2012; Machado et al., 2011, 2012).

Different adsorbents have been used for the removal of dyesfrom aqueous solutions (Alencar et al., 2012; Calvete et al., 2009;

L.D.T. Prola et al. / Journal of Environmental Management 130 (2013) 166e175 167

Cardoso et al., 2011, 2012). Carbon nanotubes (CNTs) happen to beamong the adsorbents that have been employed for the successfulremoval of dyes from aqueous effluents (Chatterjee et al., 2011;Mishra et al., 2010; Wang et al., 2012). CNTs are attractive alter-native adsorbents for the removal of dye contaminants fromaqueous effluents because they possess large specific surface area,small size as well as hollow and layered structures, giving rise toadsorbents with much higher sorption capacity when comparedwith ordinary adsorbents (Machado et al., 2011, 2012).

In the present work, adsorption studies using multi-walledcarbon nanotubes (MWCNT) were compared with powder acti-vated carbon (PAC). MWCNT and PAC were used as adsorbents forthe removal of Direct Blue 53 (DB-53) textile dye from aqueoussolutions. This dye is largely used for dyeing textiles in the Braziliancloth industry.

2. Materials and methods

2.1. Solutions

Deionised water was used throughout the experiments for so-lution preparation. The textile dye C.I. Direct Blue 53 (DB-53; C.I.23860; CAS 314-13-6; C34H24N6O14S4Na4; 960.81 g mol�1) wasfurnished by Vetec (Rio de Janeiro, Brazil) at 85% of purity (seeSupplementary Fig. 1). The dye was used without further purifica-tion. It should be pointed out that no colour change of the dye wasobserved when it was immersed in aqueous solution of pH rangingfrom 2.0 to 9.0. The DB-53 is an azo compound, which has foursulphonate groups. These groups remain negatively charged inhighly acidic solutions because their pKa values are lower than zero(Calvete et al., 2009). The stock solutionwas prepared by dissolvingthe dye in distilled water to the concentration of 5.00 g L�1.Working solutions were obtained by diluting the dye stock solu-tions to the required concentrations. To adjust the pH solutions,0.10 mol L�1 sodium hydroxide or 0.10 mol L�1 hydrochloric acidsolutions were used. The pH of the solutions was measured using aSchott Lab 850 set pH meter.

2.2. Adsorbents

MWCNTs with purity of 95% were prepared by catalytic chem-ical vapour deposition (CCVD). This method of synthesis has beendescribed previously (Machado et al., 2011). Vetec (Rio de Janeiro,Brazil) supplied the PAC (325e400 mesh). Both adsorbents wereused without further purification.

The morphology of the MWCNT adsorbent was determinedusing scanning electronmicroscopy (SEM) using a JEOLmicroscope,model JSM 6060 (Tokio, Japan) and transmission electron micro-scopy (TEM) using a JEOL microscope, model JEM 2010 (Tokio,Japan).

Raman spectroscopy measurements were done using a 20�objective lens with a 632.8 nm excitation laser line and a laserpower ofw5mW, via a Bruker spectrometer, model SENTERRA. Thespectra were obtained with a resolution of 4 cm�1 and scans werein the range of 100e3500 cm�1.

The point of zero charge (pHpzc) of the adsorbent was deter-mined by adding 20.00 mL of 0.050 mol L�1 NaCl, with a previouslyadjusted initial pH (pHi values of the solutions were adjusted from2.0 to 10.0with 0.10mol L�1 of HCl or NaOH) to several 50.0 mL flat-bottom Falcon tubes. Each Falcon tube contains 30.0 mg of theadsorbent, which was securely capped immediately, and thensuspensions were shaken in an acclimatized shaker at 298 K(Oxylab, São Leopoldo, Brazil), and allowed to equilibrate for 48 h.The suspensions were then centrifuged at 14,000 rpm for 10 min toseparate the adsorbent from the aqueous solution. The initial pH

(pHi) of the solutions were accurately measured using the solutionsthat have no contact with the solid adsorbent; and the final pH(pHf) values of the supernatant after the contact with the solid wererecorded. The value of pHpzc is the point where the curve of DpH(pHf�pHi) versus pHi crosses the line equal to zero (Calvete et al.,2009).

2.3. Adsorption studies

The batch adsorption studies for evaluation of the ability ofMWCNT and PAC adsorbents to remove DB-53 dye from aqueoussolutions were carried out in triplicate, using the batch adsorptionmethod. For these experiments, 30.0 mg of adsorbent were placedin 50 mL flat-bottom Falcon tubes containing 20.0 mL of dye so-lution (80.00e1000.0 mg L�1), which were agitated for an appro-priate time (0.0833e24.00 h) using an acclimatized shaker (Oxylab,São Leopoldo, Brazil) at temperatures ranging from 298 to 323 K.Dye solutions with pH ranging from 2.0 to 9.0 were used. Subse-quently, in order to separate the adsorbents from the aqueous so-lutions, the flasks were centrifuged at 10,000 rpm for 5 min using aUnicen M Herolab centrifuge (Stuttgart, Germany), and aliquots of1e10 mL of the supernatant were properly diluted with an aqueoussolution fixed at pH 2.0.

The final concentrations of the dyes remaining in the solutionwere determined using visible spectrophotometry using a T90þUVeVIS spectrophotometer (PG Instruments, London, UnitedKingdom) fittedwith quartz optical cells. Absorbancemeasurementswere made at 607 nm, the maximum wavelength of DB-53 dye.

The amount of dye adsorbed and the percentage of the dyeremoved by the adsorbents were calculated by applying Eqs. 1 and2, respectively:

q ¼�C0 � Cf

�m

$V (1)

% Removal ¼ 100$

�C0 � Cf

�C0

(2)

in which q is the amount of dye adsorbed by the adsorbent(mg g�1), Co is the initial dye concentration placed in contact withthe adsorbent (mg L�1), Cf is the dye concentration (mg L�1) afterthe batch adsorption procedure,m is the mass of adsorbent (g) andV is the volume of dye solution (L).

The experiments of desorptionwere carried out according to theprocedure: a 100.0mg L�1 of DB-53 dyewas shakenwith 30.0mg ofMWCNT and PAC adsorbents for 1 h; afterwards, the loadedadsorbent was filtered in 0.2 mm cellulose acetate and it was firstlywashed with water for removing non-adsorbed dyes. Then, the dyeadsorbed on the adsorbent was agitated with 20.0 mL of: ethanol;methanol; NaOH aqueous solutions (1.0e7.0mol L�1); acetone (10e50%)þwater (90e50%), and acetone (10e50%)þ 3MNaOH for 15e60 min. The desorbed dyes were separated and estimated asdescribed above.

2.4. Quality assurance and statistical evaluation of the kinetic andisotherm parameters

To establish the accuracy, reliability and reproducibility of thecollected data, all the batch adsorption measurements were per-formed in triplicate and the relative standard deviation of allmeasurements was lower than 5% (Barbosa et al., 1999). Blank testswere run in parallel and were corrected when necessary (Limaet al., 2003).

L.D.T. Prola et al. / Journal of Environmental Management 130 (2013) 166e175168

All dye solutions were stored in glass flasks, which were cleanedby soaking in 1.4 mol L�1 HNO3 for 24 h (Vaghetti et al., 2003),rinsing five times with de-ionized water, drying and storing themin a flow-hood.

For analytical calibration, standard solutions with concentra-tions ranging from 5.00 to 120.0 mg L�1 of the dyes were used, inparallel with a blank solution of water adjusted to pH 2.0. The linearanalytical calibration of the curve was done using the UVWinsoftware of the T90 þ PG Instruments spectrophotometer. Thedetection limit of the method, obtained with a signal/noise ratio of3 (Lima et al., 1998a), was 0.16 mg L�1 of DB-53. All the analyticalmeasurements were performed in triplicate, and the precision ofthe standards was better than 3% (n ¼ 3). In order to verify theaccuracy of the DB-53 dye sample solutions during spectrophoto-metric measurements, standards containing dyes at 50.0 mg L�1

were employed as a quality control at every five determinations(Lima et al., 1998b).

The kinetic and equilibrium models were fitted by employing anonlinear method, with successive interactions calculatedemploying the LevenbergeMarquardt method; interactions werealso calculated using the Simplex method, based on the nonlinearfitting facilities of the software Microcal Origin 7.0. In addition, themodels were evaluated by using a determination coefficient (R2), anadjusted determination coefficient (R2adj), as well as by an errorfunction (Ferror) (Prola et al., 2013), which measured the differencesin the amount of dye taken up by the adsorbent as predicted by themodels and the actual q measured experimentally. R2, R2adj andFerror are given below, in Eqs. (3)e(5), respectively:

R2 ¼

0BBB@Pni

�qi;exp � qi;exp

�2 �Pni

�qi;exp � qi;model

�2Pni

�qi;exp � qi;exp

�21CCCA (3)

R2adj ¼ 1��1� R2

�$

�np � 1np � p

�(4)

Ferror %ð Þ ¼ 100�

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1

np � p

� �$X

i

n qi;exp � qi;model

qi;exp

!2vuut (5)

where qi, model represents each value of q predicted by the fittedmodel, qi, exp represents each value of q measured experimentally,qexp is the average of q experimentally measured, np is the numberof experiments performed, and p is the number of parameters ofthe fitted model (Prola et al., 2013).

2.5. Kinetic adsorption models

The kinetic equations used in this work are:Pseudo-first order (Eq. (6)); Pseudo second-order (Eq. (7));

General order kinetic model (Eq. (8)); and intra-particle diffusionmodel (Eq. (9)).

qt ¼ qe½1� expð � k1$tÞ� (6)

qt ¼ qe � qe½k2ðqeÞ$t þ 1� (7)

qt ¼ qe � qehkNðqeÞn�1$t$ðn� 1Þ þ 1

i1=1�n(8)

qt ¼ kidffiffit

pþ C (9)

For further details of these models, please see Supplementarymaterial (Alencar et al., 2012; Cardoso et al., 2012; Ho, 2006; Liuand Liu, 2008; Liu and Shen, 2008; Weber and Morris, 1963).

2.6. Equilibrium models

Langmuir (Eq. (10)), Freundlich (Eq. (11)) and Sips (Eq. (13)) arethe equilibrium equations used in this work.

qe ¼ Qmax$KL$Ce1þ KL$Ce

(10)

qe ¼ KF$C1=nFe (11)

qe ¼ Qmax$ðKS$CeÞ1=nS

1þ ðKS$CeÞ1=nS(12)

For further details of these models, please see Supplementarymaterial (Langmuir, 1918; Freundlich, 1906; Sips, 1948).

2.7. Simulated dye-house effluent

One synthetic dye-house effluent containing five representativetextile dyes usually used for colouring fibres and their corre-sponding auxiliary chemicals was prepared at pH 2.0, using amixture of different dyes most often applied in the textile fibreindustry. According to the practical information obtained from adye-house, typically 10e60% (Hessel et al., 2007) of synthetic dyesand 100% of the dye bath auxiliaries remain in the spent dye bath,and its composition undergoes a 5e30-fold dilution during thesubsequent washing and rinsing stages (Machado et al., 2011, 2012;Calvete et al., 2009; Cardoso et al., 2012). The concentrations of thedyes and auxiliary chemicals selected to imitate an exhausted dyebath are given in Supplementary Table 1 (Machado et al., 2011,2012; Calvete et al., 2009; Cardoso et al., 2012).

3. Results and discussion

3.1. Characterisation of the adsorbents

The textural properties of the MWCNT and PAC adsorbents arepresented in Supplementary Table 2. The average pore diameter ofMWCNT adsorbent is relatively large, when compared with theactivated carbon; this could be attributed to the aggregated pores ofMWCNT (Upadhyayula et al., 2009). Carbon nanotubes formaggregated pores due to the entanglement of tens and hundreds ofindividual tubes that are adhered to each other because of van derWaals forces of attraction (Upadhyayula et al., 2009). Theseaggregated pores have the dimensions of a mesopore or higher(Upadhyayula et al., 2009). The maximum diagonal length of theDB-53 dye is 2.04 nm (see Supplementary Fig. 1). The ratios ofaverage pore diameter of the MWCNT and PAC adsorbents to themaximum diagonal length of the dye are 3.7 and 1.7, respectively.Therefore, the mesopores of MWCNT could accommodate up to 3molecules of DB-53 and the PAC adsorbent could only accommo-date one dye molecule.

Scanning and transmission electron microscopy(Supplementary Fig. 2) shows the morphological structure of the ofthe MWCNT adsorbent. The SEM micrograph (SupplementaryFig. 2A) suggests that MWCNT could be expanded whenimmersed in aqueous solution, because this adsorbent is anentanglement of nanotubes (Machado et al., 2011). TEM image(Supplementary Fig. 2B) shows that the outer diameters of the

1 2 3 4 5 6 7 8 9 1080

85

90

95

100

105lavo

meR

%

initial pH

MWCNTCAC

pHPZCMWCNT

pHPZCPAC

A

B

C

-+

+-

1 2 3 4 5 6 7 8 9 10-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

Hp

initial pH

pHpzc= 7.00

1 2 3 4 5 6 7 8 9 10

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Hp

initial pH

pHpzc = 7.80

Fig. 1. (A) Effect of pH on the adsorption of DB-53 dye on MWCNT and PAC. Conditions:Co 300.0 and 100.0 mg L�1 of dye solution, for MWCNT and PAC, respectively; mass ofadsorbent of 30.0 mg; the temperature was fixed at 298 K. Time of contact between thedye and adsorbate was fixed in 4.0 h. (B) Point of zero charge of MWCNT. (C) Point ofzero charge of PAC.

L.D.T. Prola et al. / Journal of Environmental Management 130 (2013) 166e175 169

MWCNTs are in the range of 3e40 nm and, also provides evidenceof the ‘‘bamboo-like’’ structure of MWCNTs.

The Supplementary Fig. 3 displays the Raman spectra of theMWCNT before the adsorption and loaded with the dye DR-53. Thediamond (or ‘‘defect’’) mode (D) at about 1316 cm�1 for MWCNTpristine, induced by sp3 electronic states (considered to be defectsin the planar sp2 graphitic structure) (Machado et al., 2012), wasvisualised. The peaks near 1583 cm�1 for MWCNT pristine, are the

so-called G band, which are related to the graphite E2g symmetry ofthe interlayer mode. This mode reflects the structural integrity ofsp2-hybridised carbon atoms of the nanotubes. In addition, a G0-band occurs at 2641 cm�1 for MWCNT pristine, which is the secondharmonic of D-band. The Raman spectra of the MWCNT loadedwith the dye DR-53 shows a shift in position of D-band, G-band andG0-band at 1324 cm�1, 1594 cm�1 and 2673 cm�1, respectively. Theinteraction between DB-53 dye molecule and surface of adsorbentcan induce the shift of the characteristic MWCNT bands to higher ofRaman shift values due to the increase in the elastic constant of theharmonic oscillator of the dye-adsorbed MWCNT þ DB-53 (Mishraet al., 2010). The van der Waals attraction between the dye and thegraphite sheets of CNT may increase the energy needed for vibra-tions to occur, which is reflected in the higher frequency of Ramanpeaks (Mishra et al., 2010). Moreover, in the Supplementary Fig. 3,for the MWCNT þ DB-53, a N]N stretching band is recognized at1440 cm�1 (Zhang and Silva, 2010). The stretch of N]N of dyemolecules accounts for the manifestation of strong Raman band at1440 cm�1 (Zhang and Silva, 2010), therefore, this band wasmonitored to indicate the presence of DB-53 dye molecules, sug-gesting the great quantity this molecules on carbon nanotubes. Thepeaks observed in the region from 404 to 830 cm�1 are relatedprobably to the DB-53 dye.

3.2. Effects of pH on adsorption

One of the most important variables that affect the sorptioncapacity of the adsorbents during the adsorption studies is the pHof adsorbate solution (Calvete et al., 2009; Cardoso et al., 2011,2012). Different adsorbates may present different optimal pH forbeing adsorbed depending of the chosen adsorbent. The effects ofinitial pH on the percentage removal of DB-53 dye using MWCNTand PAC adsorbents were evaluated within the pH range of 2.0 and9.0 (Fig. 1). For both adsorbents, the percentage of dye removalslightly decreased from pH 2.0 up to 9.0. The decrease in the per-centage values of dye removal when the pH was varied from 2.0 to9.0 was 5.56% for MWCNT adsorbent and 5.42% for PAC adsorbent.

The pHPZC values obtained for MWCNTand PAC adsorbents were7.00 and 7.80, respectively. For pH values lower than pHpzc, theadsorbent presents a positive surface charge (Calvete et al., 2009;Cardoso et al., 2011). The dissolved DB-53 dye is negativelycharged in aqueous solution, because it possesses four sulphonategroups (Calvete et al., 2009). The adsorption of this dye takes placewhen the adsorbents have a positive surface charge. Electrostaticinteractions occur for MWCNT and PAC (at pH <7.00 and 7.80,respectively). However, when the pH value is much lower thanpHpzc, the surface of the adsorbent becomes more positive (Alencaret al., 2012; Calvete et al., 2009; Cardoso et al., 2012). This phe-nomenon explains the high adsorption capacity of both adsorbentsfor DB-53 dye at pH 2.0. In order to continue the adsorption studies,the initial pH of the dye solution was fixed at 2.0. Under this con-dition, after the adsorption experiments, the final pH of dye solu-tion did not change significantly (<3%).

3.3. Kinetic studies

Adsorption kinetic studies are important in the treatment ofaqueous effluents because they provide valuable information onthe mechanism of the adsorption process (Prola et al., 2013; Royeret al., 2010).

It is important to point out that the initial DB-53 concentrationemployed during the kinetic studies was 300.0 mg L�1. This choiceis attributed to the fact that using PAC as adsorbent for removal ofDB-53 from aqueous solution, the useful initial dye concentrationsranged from 80 to 360 mg L�1. For concentrations lower than

Table 1Kinetic parameters for DB-53 dye removal using MWCNT and PAC as adsorbents.Conditions: pH 2.0; adsorbent mass 30.0 mg, initial DB-53 dye concentration300 mg L�1.

MWCNT PAC

Pseudo-first orderk1 (h�1) 2.860 1.503qe (mg g�1) 116.4 88.02h0 (mg g�1 h�1) 332.7 132.3R2 adj 0.9995 0.9981Ferror (%) 1.318 8.241Pseudo-second orderk2 (g mg�1 h�1) 3.184.10�2 1.684.10�2

qe (mg g�1) 126.7 102.0h0 (mg g�1 h�1) 510.9 107.1R2 adj 0.9863 0.9793Ferror (%) 7.812 18.17General orderkn [h�1.(g mg�1)n�1] 1.855 2.415qe (mg g�1) 116.8 87.37n 1.101 0.8842h0 (mg g�1 h�1) 350.8 125.7R2 adj 0.9999 0.9990Ferror (%) 0.4327 6.664Intraparticlekid,2 (mg g�1 h�0.5)a 26.27 13.29

a Second stage.

L.D.T. Prola et al. / Journal of Environmental Management 130 (2013) 166e175170

80mg L�1, the removal of dye is practically 100%, being not possibleto measure the dye in the final concentration, therefore, invalid-ating the kinetic studies. Initial dye concentrations higher than360mg L�1 are also inappropriate, since the plateau of saturation ofPAC adsorbent has already been attained. On the other hand, usingMWCNT as adsorbent for DB-53 dye, the useful initial dye con-centrations for adsorption studies ranged from 280.0 to1000.0 mg L�1. For concentrations lower than 280.0 mg L�1, theremoval of the dye is quantitative (y100%), being the final con-centration of the dye, which is below the detection limit of spec-trophotometric measurement. The superior limit of initialconcentration of DB-53 dye is 1000 mg L�1, where the saturation ofthe MWCNT is already obtained. Considering just the lower limit ofinitial concentration of DB-53, it will be inadequate to concludethat the sorption capacity of MWCNT is superior to that of PAC.

In an attempt to describe the adsorption kinetics of DB-53 dyeusing the MWCNT and PAC adsorbents, three kinetic models weretested as shown in Fig. 2. The kinetic parameters for the kineticmodels are listed in Table 1. In order to compare different kineticmodels, the Ferror of each individual model was divided by the Ferrorof minimum value (Ferror ratio). The general order kinetic modelgave least Ferror values. The pseudo-first order kinetic model gaveFerror ratio values of 3.0 (MWCNT), and 1.2 (PAC). In addition, for thepseudo-second order model, the Ferror ratio values of 18.1(MWCNT), and 2.7 (PAC). The lower the error function, the lowerthe difference of the q calculated by the model from the experi-mentally measured q (Prola et al., 2013). It should be pointed outthat the Ferror utilised in this work takes into account the number offitted parameters (p term in Eq. (5)). It has been reported in theliterature (El-Khaiary et al., 2010; El-Khaiary andMalash, 2011) thatirrespective of the number of parameters present in a nonlinear

2

4

6

8

10

q t(mgg-1 )

0 1 2 3 4 5 60

20

40

60

80

100

120

Experimental pointsPseudo-first orderPseudo-second orderGeneral-orderq t

(mgg-1 )

Time (h)

0

20

40

60

80

100

120

q t(mgg-1 )

0 1 2 3 4 5 6

0

20

40

60

80

100

Experimental pointsPseudo-first orderPseudo-second orderGeneral-orderq t

(mgg-1 )

Time (h)

Fig. 2. Kinetic adsorption curves for DB-53 uptake at 298 K on MWCNT and PAC.

equation, it gives the best fitting results. For this reason, thenumber of fitting parameters should be considered in the calcula-tion of Ferror. In the same vein, it was verified that the qe valuesfound in the general order kinetic model were closer to theexperimental qe values when compared with all other kineticmodels. These results indicate that the general order kinetic model

0.0 0.5 1.0 1.5 2.0 2.50

0

0

0

0

0

Time0.5 (h0.5)

MWCNT

PAC

0.0 0.5 1.0 1.5 2.0 2.5

Time0.5 (h0.5)

Conditions: Co 300.0 mg L�1; pH was fixed at 2.0; adsorbent mass 30.0 mg.

Table 3Maximum sorption capacities of different adsorbents used to remove different dyes.

Adsorbent Dye adsorbate Qmax (mg g�1) Ref.

Mangifera indica (mango) seeds Victazol Orange 3R 51.2 (Alencar et al., 2012)Mangifera indica (mango) seeds (protonated) Victazol Orange 3R 71.6 (Alencar et al., 2012Jatropha curcas shell Reactive Red 120 40.94 (Prola et al., 2013)Jatropha curcas shell (treated by non-thermal plasma) Reactive Red 120 65.63 (Prola et al., 2013)Multi-wall-carbon nanotube (MWCNT) Congo Red 140.1 (Mishra et al., 2010)Multi-wall-carbon nanotube (MWCNT) Reactive green HE4BD 151.9 (Mishra et al., 2010)Multi-wall-carbon nanotube (MWCNT) Golden yellow MR 141.6 (Mishra et al., 2010)Spirulina platensis microalgae Reactive red 120 482.2 (Cardoso et al., 2012)Commercial activated carbon (PAC) Reactive red 120 267.2 (Cardoso et al., 2012)Single-walled carbon nanotubes (SWCNT) Reactive Blue 4 567.7 (Machado et al., 2012)Multi wall carbon nanotube (MWCNT) Reactive Blue 4 502.5 (Machado et al., 2012)Chitosan hydrogel beads coated on multiwall carbon nanotube Congo Red 200 (Bilgili, 2006)Octosilicate Na-RUB-18 Reactive Black 5 76.8 (Royer et al., 2010)Powered activatec carbon (PAC) Reactive Red M-2BE 260.7 (Machado et al., 2011)Multi wall carbon nanotube (MWCNT) Reactive Red M-2BE 335.7 (Machado et al., 2011)Commercial activated carbon (PAC) Direct Blue 53 135.2 This workMulti wall carbon nanotube (MWCNT) Direct Blue 53 409.4 This work

L.D.T. Prola et al. / Journal of Environmental Management 130 (2013) 166e175 171

should best explain the adsorption process of DB-53 dye usingMWCNT and PAC adsorbents.

Taking into account that the general order kinetic equation hasdifferent orders (n) when the concentration of the adsorbate ischanged (see Table 1), it is difficult to compare the kinetic param-eters of the model. Therefore, it is useful to use the initial sorptionrate h0 (Ho, 2006) to evaluate the kinetics of a given model, usingEq. (13):

h0 ¼ kN$qne (13)

where h0 is the initial sorption rate (mg g�1 h�1), kN is the rateconstant [h�1 (g mg�1)n�1], qe is the amount adsorbed at equilib-rium (mg g�1) and n is the order of the kinetic model. It should bestressed that when n ¼ 2, this equation is the same as the initialsorption rate early reported in the literature (Ho, 2006). Taking intoaccount that the kinetic data were better fitted with the generalorder kinetic model, more confident of initial sorption rates (h0)were obtained with the general order kinetic model. In addition, itis surprising that the h0 values obtained for MWCNT adsorbent arealways much higher than the values obtained for PAC adsorbent,onewould expect that the kinetics of adsorption of DB-53 would befaster with MWCNT compared to that of PAC.

Since the kinetic results fit very well to the general order ki-netic model for DB-53 dye using MWCNT and PAC adsorbents(Table 1 and Fig. 2), the intra-particle diffusion model (Weber and

Table 2Isotherm parameters for DB-53 adsorption, using MWCNTand PAC adsorbents. ConditionsPAC, respectively.

Temperature (K) MWCNT

298 303 308 313 318

LangmuirQmax (mg g-1) 334.8 338.5 339.2 336.7 332.8KL (L mg-1) 1.619 1.848 2.322 2.872 3.86R2adj 0.9827 0.9992 0.9847 0.9676 0.9572Ferror 7.521 1.147 6.145 7.456 7.124FreudlichKF (mg g-1 (mg L�1)�1/nF) 220.7 228.2 227.5 234.9 243.8nF 10.17 10.36 9.868 10.73 11.97R2adj 0.8350 0.9324 0.9508 0.9759 0.9871Ferror 24.65 11.02 11.11 6.547 4.319SipsQmax (mg g-1) 326.4 341.7 357.8 374.7 393.2Kg (L mg-1) 1.507 1.895 2.360 2.908 3.593nS 0.6194 1.124 1.594 2.164 2.815R2adj 0.9998 0.9999 0.9998 0.9999 0.9999Ferror 0.4539 0.2747 0.3755 0.2842 0.2473

Morris, 1963) was used to verify the influence of mass transferresistance on the binding of RB-4 dye to the adsorbents (Table 3and Fig. 4). The intra-particle diffusion constant, kid(mg g�1 h�0.5), can be obtained from the slope of the plot of qt(uptake at any time, mg g�1) versus the square root of time. Thisfigure shows the plots of qt versus t1/2, with multi-linearity for DB-53 dye using the MWCNT and PAC adsorbents. These results implythat the adsorption process involves more than one single kineticstage (or adsorption rate) (Alencar et al., 2012). For both adsor-bents, the adsorption process occurred in three stages, whichcould be attributed to each linear portion as shown in Fig. 2. Thefirst linear portion was attributed to the diffusional process of thedye to the adsorbent surface (Alencar et al., 2012); hence, it wasthe fastest sorption stage. The second portion was ascribed tointra-particle diffusion, a delayed process (Alencar et al., 2012).The third stage may be regarded as the diffusion through smallerpores, which is followed by the establishment of equilibrium(Alencar et al., 2012).

As shown in Fig. 2, the minimum contact times of DB-53 dye toreach equilibrium are about 2 and 3 h for the MWCNT adsorbentand PAC adsorbent, respectively. For subsequent experimentalwork, the contact times for the removal of DB-53 dye were fixed at3 and 4 h for MWCNT adsorbent and PAC adsorbent, respectively.The increment in the contact time utilised in this work was toguarantee that DB-53 dye equilibrium would be attained even athigher adsorbate concentrations (Calvete et al., 2009).

: adsorbent mass 30.0 mg; pH fixed at 2.0, contact time 3 and for 4 h for MWCNTand

PAC

323 298 303 308 313 318 323

332.4 90.67 88.36 87.14 93.14 97.29 98.113.980 0.3787 0.5390 0.7841 0.9125 0.9712 1.7470.9608 0.9995 0.9844 0.9695 0.9718 0.9744 0.96656.522 1.098 6.119 7.281 6.277 5.253 5.839

249.1 44.63 45.60 49.95 55.83 62.95 67.8813.21 6.062 6.402 7.474 8.091 9.563 10.910.9946 0.9505 0.9738 0.9878 0.9906 0.9957 0.99722.522 11.72 8.293 4.676 3.820 2.319 1.713

409.4 92.00 98.65 105.6 114.3 124.3 135.24.348 0.3752 0.4349 0.5072 0.5940 0.7015 0.81083.517 1.085 1.644 2.247 2.438 2.963 3.8740.9998 0.9999 0.9998 0.9999 0.9999 0.9998 0.99970.3342 0.4347 0.4818 0.4010 0.2865 0.4581 0.5066

0 20 40 60 80 100 120 1400

50

100

150

200

250

300

350

400

Experimental pointsLangmuirFreundlichLiuq e

ggm(

1-)

Ce (mg L-1)

0 20 40 60 80 100 1200

20

40

60

80

100

120

Ce (mg L-1)

Experimental pointsLangmuirFreundlichLiu

q eggm(

1-)

A

B

Fig. 3. Isotherm curves of DB-53 adsorption at 323 K on (A) MWCNT and (B) PAC.Conditions: pH was fixed at 2.0; the adsorbent mass was fixed at 30 mg; the contacttime was fixed at 3 and 4 h, for MWCNT and PAC, respectively.

300 400 500 600 700 800

0.00

0.15

0.30

0.45

0.60

0.75

0.90

ecnabrosbA

Wavelength (nm)

Effluent A diluted 15-fold before treatmentEffluent A undiluted after treatment with MWCNTEffluent A undiluted after treatment with PAC

300 400 500 600 700 800

0.0

0.2

0.4

0.6

0.8

1.0

1.2Effluent B diluted 15-fold before treatmentEffluent B undiluted after treatment with MWCNTEffluent B undiluted after treatment with PAC

ecnabrosbA

Wavelength (nm)

A

B

Fig. 4. UVeVIS spectra of simulated dye effluents before and after adsorption treat-ment with MWCNT and PAC. The temperature was fixed at 298 K and pH 2.0. A-Effluent A; B-Efflluent B. For effluent composition see Table 1.

L.D.T. Prola et al. / Journal of Environmental Management 130 (2013) 166e175172

It was observed that the kinetics of sorption of DB-53 dye onMWCNT adsorbent were faster than those obtained using PACadsorbent. Considering the initial sorption rate (h0) of DB-53 dyeadsorbed by MWCNT and PAC adsorbents (see Table 1), as obtainedby the general order kinetic model, it was observed that h0 ofMWCNTadsorbent is 2.79 times higher than h0 of PAC adsorbent. Inthe sameway, the value of kid,2 of MWCNT is 1.98 times higher thanthe corresponding value of PAC adsorbent. The differences in ki-netic constants (h0 and kid,2) are responsible for the 1 h difference inthe minimum equilibrium time for adsorption of DB-53 ontoMWCNT surface in relation to PAC adsorbent.

3.4. Equilibrium studies

Adsorption isotherm describes the relationship between theamount of adsorbate adsorbed by the adsorbent (qe) and theadsorbate concentration remaining in the solution after the systemhas attained the equilibrium state (Ce) at constant temperature. Theadsorption parameters of the equilibrium models often providesome insights into the adsorption mechanism, the surface proper-ties and affinity of the adsorbent by the adsorbate. In this work, theLangmuir (Langmuir, 1918), the Freundlich (Freundlich, 1906) andthe Sips (Sips, 1948) isotherm models were tested.

The isotherms of adsorption were carried out between 298 and323 K with DB-53 dye on the adsorbents using the best experi-mental conditions previously described (see Fig. 3 and Table 2). In

order to compare the isothermmodels, the Ferror of each model wasdivided by the Ferror of the minimum value (Ferror ratio). Sips equi-librium model has the least Ferror values at all the six temperaturesstudied (Table 2), this implies that the q fit by the Sips isothermmodel was close to the q measured experimentally. The Langmuirisotherm model, with Ferror ratio values ranging from 4.2 to 28.8(MWCNT) and from 2.5 to 21.9 (PAC), did not suitably fit ourexperimental data. In the same way, the Freundlich isotherm hasFerror ratio ranging from 7.5 to 54.3 (MWCNT) and from 3.4 to 27.0(PAC).

The maximum amounts of DB-53 uptake were 409.4 and135.2 mg g�1 for MWCNT and PAC, respectively. These valuesindicate that these adsorbents are good for removal of DB-53 dyefrom aqueous solutions.

Taking into consideration that it is difficult to compare thesorption capacities of the dyes with diverse adsorbents, since thedyes reported in the literature are given as commercial dyes, andalso considering the fact that the same dye could have differentcommercial trade names depending on its manufacturer, Table 3(Alencar et al., 2012; Prola et al., 2013; Royer et al., 2010; Mishraet al., 2010; Cardoso et al., 2012; Machado et al., 2011, 2012;Bilgili, 2006) presented a comparison of sorption capacities ofdifferent dyes adsorbed by different adsorbents. The values wereobtained at the best experimental conditions of each work. Basedon the data in Table 3, both MWCNT and PAC adsorbents have goodsorption capacities when compared with the adsorption ofdifferent dyes on different adsorbents, and can be alternatively

Table 4Desorption of DB-53 dye loaded on MWCNT and PAC adsorbents. Conditions foradsorption: initial DB-53 concentration 100 mg L�1; mass of adsorbent 30.0 mg, pH2.0; time of contact 1 h.

Conditions for desorption % desorption

MWCNT PAC

Ethanol 9.53 6.82Methanol 10.96 9.701.0 mol L�1 NaOH (aqueous solution) 10.45 12.272.0 mol L�1 NaOH (aqueous solution) 11.63 14.823.0 mol L�1 NaOH (aqueous solution) 18.54 16.364.0 mol L�1 NaOH (aqueous solution) 17.36 14.875.0 mol L�1 NaOH (aqueous solution) 18.03 16.326.0 mol L�1 NaOH (aqueous solution) 17.36 14.367.0 mol L�1 NaOH (aqueous solution) 12.36 11.6510% Acetone þ 90% water 35.56 20.5220% Acetone þ 80% water 45.65 21.3630% Acetone þ 70% water 55.62 22.7440% Acetone þ 60% water 75.65 25.6550% Acetone þ 50% water 86.36 27.3610% acetone þ 90% 3 M NaOH 86.65 28.6320% acetone þ 80% 3 M NaOH 92.56 30.2130% acetone þ 70% 3 M NaOH 95.56 31.3240% acetone þ 60% 3 M NaOH 97.63 33.4750% acetone þ 50% 3 M NaOH 97.85 31.57

L.D.T. Prola et al. / Journal of Environmental Management 130 (2013) 166e175 173

considered for the removal of direct dyes from aqueous effluents. Itshould be noted that the maximum amount of DB-53 dye adsorbedon MWCNT was on average 3.31 times higher than the corre-sponding value obtained for PAC. The textural characteristics ofMWCNT discussed in Section 3.1 justify this difference.

3.5. Thermodynamic studies and mechanism of adsorption

Thermodynamic parameters related to the adsorption process,i.e. Gibb’s free energy change (DG�, kJ mol�1), enthalpy change(DH�, kJ mol�1) and entropy change (DS�, J mol�1 K�1) weredetermined by the following equations:

DG� ¼ DH

� � TDS�

(14)

DG� ¼ �RT$LnðKÞ (15)

The combination of Eqs (14) and (15) gives.

LnðKÞ ¼ DS�

R� DH

�

R� 1T

(16)

where R is the universal gas constant (8.314 J K�1 mol�1), T is theabsolute temperature (Kelvin) and K represents the equilibriumadsorption constants of the isotherm fits. It has been reported inthe literature that different adsorption equilibrium constants (K)were obtained from different isothermmodels (Alencar et al., 2012;Calvete et al., 2009; Cardoso et al., 2012; Gupta et al., 2011;Machado et al., 2012; Leechart et al., 2009; Liu and Liu, 2008).Thermodynamic parameters of adsorption can be estimated fromKS, Sips equilibrium constant (Calvete et al., 2009).

DH� and DS� values can be calculated from the slope andintercept of the linear plot of Ln(K) versus 1/T.

The thermodynamic results are depicted in SupplementaryTable 3. The R2 values of the linear fit are 0.9999 and 0.9988 forMWCNT and PAC, respectively, indicating that the values ofenthalpy and entropy calculated for both adsorbents are in order. Inaddition, the magnitude of enthalpy was consistent with the elec-trostatic interaction of an adsorbent with an adsorbate (Sun andWang, 2010). The type of interaction can be classified, to a certainextent, by the magnitude of enthalpy change. Physical sorption,such as hydrogen bonding, is usually lower than 40 kJ mol�1 (Sun

and Wang, 2010). Enthalpy changes (DH�) indicate that theadsorption followed an endothermic process. Negative values of DGsignify that the DB-53 dye adsorption by MWCNT and PAC adsor-bents were spontaneous and favourable processes at all theexperimental temperatures. The positive values of DS� confirmed ahigh preference of DB-53 molecules for the surface of MWCNT andPAC, and suggested the possibility of some structural changes orreadjustments in the dye-carbon sorption complex (Asouhidouet al., 2009).

3.6. Desorption experiments and mechanism of adsorption

In order to check the reuse of the MWCNT and PAC adsorbentsfor the adsorption of DB-53 dye, desorption experiments werecarried-out. The eluents such as, ethanol; methanol; NaOH aqueoussolutions (1.0e3.0 mol L�1); acetone (10e50%) þ water (90e50%),and acetone (10e50%) þ 3 M NaOH were tested for regeneration ofthe loaded adsorbent (see Table 4). For both adsorbents, the eluentssuch as ethanol, methanol and NaOH solutions were not efficientfor the dye desorption from MWCNT and PAC adsorbents(desorption<19%). On the other hand, for MWCNT, mixtures of 50%acetone þ 50% water achieved more than 86% recovery of DB-53dye loaded on the adsorbent. Similarly, recoveries of more than97% were obtained using 40% acetone þ 60% of 3 M NaOH and 50%acetone þ 50% of 3 M NaOH. Otherwise, for PAC adsorbent, therecoveries were always < 34% for all eluents presented on Table 4,as previously reported (Machado et al., 2011).

The desorption behaviour is in agreement with the pH studies aswell as magnitude of enthalpy changes described above. The DB-53dye at pH 2.0 is attracted electrostatically by the MWCNT. Thisinteraction was disrupted by addition of NaOH solution. However,besides this electrostatic interaction, there are also some in-teractions between the aromatic groups present on the MWCNTwith the aromatic rings of the dye. Acetone was used to break thisinteraction, and this significantly improved the elution efficiency(achieved up to 97.85% with the mixture 50% acetone þ 50% of3.0 mol L�1 NaOH). For PAC adsorbent, the elution efficiency waslower, about 34% for all the eluents tested, indicating that theactivated carbon could not be reutilized for adsorption purposes.On the other hand, for MWCNT eluted with the mixture 50%acetone þ 50% of 3.0 mol L�1 NaOH was reutilized for the adsorp-tion of the DB-53 dye, attained a sorption efficiency of about 94% inthe second cycle, 90% in the third cycle and 88% at the fourth cycleof adsorption/desorption when compared with the first cycle ofadsorption/desorption. Therefore, the use of MWCNT for dyeadsorption is economically viable since it allows its regeneration. Inaddition, there is a tendency for reduction in prices of CNT withtheir industrial production growth in the last years (Agboola et al.,2007).

A mechanism for the adsorption of DB-53 on MWCNT is pro-posed (see Supplementary Fig. 4). In the first step, the adsorbentsare immersed in a solution with pH < pHpzc (pH 2.0), so the func-tional groups OH, CaO, COOH (Prola et al., 2013) of the adsorbentare protonated (see Fig. 3). This step is fast. The second stage is theelectrostatic attraction of the negatively charged dyes by thepositively charged surface of adsorbents at pH 2.0. This stage shouldbe the rate-controlling step. The enthalpy of adsorption is consis-tent with the electrostatic attraction that exists between the dye(negatively charged) and the adsorbent (with a positive superficialcharge, since pH < pHpzc).

3.7. Treatment of a simulated dye house effluent

In order to verify the efficiency of MWCNT and PAC as adsor-bents for the removal of dyes from textile effluents, two simulated

L.D.T. Prola et al. / Journal of Environmental Management 130 (2013) 166e175174

dyehouse effluents were prepared (see Supplementary Table 1).The UVeVIS spectra of the untreated effluents at pH 2.0 aftertreatment with MWCNT and PAC were recorded from 350 to800 nm (Fig. 4). The area under the absorption bands from 350 to800 nm was utilised to monitor the percentage of dye mixtureremoved from the simulated dye effluents. The MWCNT and PACadsorbents removed 99.87% and 97.00% of the effluent A, respec-tively, and 99.31% and 96.77% of effluent B, respectively (Fig. 4).These results indicate the potentiality of using MWCNT and PAC asexcellent adsorbents for the treatment of industrial effluentscontaminated with dyes. It is to be emphasised that the composi-tion of these simulated effluents is of higher dye concentration thanthe concentration that is usually found in real textile effluents(Hessel et al., 2007), which reinforces the applicability ofMWCNT and PAC as potential adsorbents for real textile effluentstreatment.

4. Conclusions

The DB-53 dye interacted with the MWCNT and PAC adsorbentsat the solid/liquid interface when suspended in water. The generalorder kinetic model best described the kinetic of adsorption. Theequilibrium isotherms of the DB-53 dye were best fit with the Sipsisotherm model. The maximum adsorption capacities were 409.4and 135.2 mg g�1 for MWCNT and PAC, respectively. The enthalpy(DH�) of adsorption indicated that adsorption was an endothermicprocess and the magnitude of enthalpy was consistent with anelectrostatic interaction of the adsorbent with the adsorbate. TheMWCNT material loaded with DB-53 dye could be regeneratedefficiently using a mixture 50% acetone þ 50% of 3 mol L�1 NaOH.Based on our data, it is possible to propose a mechanism of in-teractions of the DB-53 dye with the MWCNT adsorbent. In the firststep, the MWCNT is immersed in acidic solution (pH 2.0), so thefunctional groups of the adsorbent are protonated. This step is fast.The second stage is the electrostatic attraction of the negativelycharged dye by the positively charged surface of MWCNT at pH 2.0.This stage should be the rate-controlling step. For treatment ofsimulated industrial textile effluents, the adsorbents showed verygood performances for removing at least 99.87% (MWCNT) and97.00% (PAC) of the mixture of dyes. These results show clearly thatMWCNT and PAC could be used in the treatment of real effluents.MWCNT could be regenerated and used in several cycles ofadsorption/desorption, however, PAC attain regeneration below34%, increasing the cost of the real wastewater treatment. However,it is still expected that the price of production of CNTs decrease inthe next years (Machado et al., 2011, 2012), since the price of CNT isthe main factor that determines its application in large-scalewastewater treatment.

Acknowledgements

The authors are grateful to The National Council for Scientificand Technological Development (CNPq, Brazil), to The Coordinationof Improvement of Higher Education Personnel (CAPES, Brazil), andto The Academy of Sciences for Developing World (TWAS, Italy) forfinancial support and fellowships. We are also grateful to Center ofElectron Microscopy (CME-UFRGS) for the use of the SEM and TEMmicroscope.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jenvman.2013.09.003.

References

Agboola, A.E., Pike, R.W., Hertwig, T.A., Lou, H.H., 2007. Conceptual design of carbonnanotube processes. Clean Techn. Environ. Policy 9, 289e311.

Ai, L., Jiang, J., 2012. Removal of methylene blue from aqueous solution with self-assembled cylindrical grapheneecarbon nanotube hybrid. Chem. Eng. J. 192,156e163.

Alencar, W.S., Lima, E.C., Royer, B., dos Santos, B.D., Calvete, T., da Silva, E.A.,Alves, C.N., 2012. Application of aqai stalks as biosorbents for the removal of thedye Procion Blue MX-R from aqueous solution. Sep. Sci. Technol. 47, 513e526.

Asouhidou, D.D., Triantafyllidis, K.S., Lazaridis, N.K., Matis, K.A., Kim, S.S.,Pinnavaia, T.J., 2009. Sorption of reactive dyes from aqueous solutions by or-dered hexagonal and disordered mesoporous carbons. Microporous Meso-porous Mater. 117, 257e267.

Barbosa Jr., F., Krug, F.J., Lima, E.C., 1999. On-line coupling of Electrochemical pre-concentration in tungsten coil electrothermal atomic absorption spectrometryfor determination of lead in natural waters. Spectrochim. Acta B 54, 1155e1166.

Bilgili, M.S., 2006. Adsorption of 4-chlorophenol from aqueous solutions by xad-4resin: Isotherm, kinetic, and thermodynamic analysis. J. Hazard. Mater. 137,157e164.

Calvete, T., Lima, E.C., Cardoso, N.F., Dias, S.L.P., Pavan, F.A., 2009. Application ofcarbon adsorbents prepared from the Brazilian-pine fruit shell for removal ofProcion Red MX 3B from aqueous solution - kinetic, equilibrium, and thermo-dynamic studies. Chem. Eng. J. 155, 627e636.

Cardoso, N.F., Lima, E.C., Pinto, I.S., Amavisca, C.V., Royer, B., Pinto, R.B., Alencar, W.S.,Pereira, S.F.P., 2011. Application of cupuassu shell as biosorbent for the removalof textile dyes from aqueous solution. J. Environ. Manage 92, 1237e1247.

Cardoso, N.F., Lima, E.C., Royer, B., Bach, M.V., Dotto, G.L., Pinto, L.A.A., Calvete, T.,2012. Comparison of Spirulina platensis microalgae and commercial activatedcarbon as adsorbents for the removal of Reactive Red 120 dye from aqueouseffluents. J. Hazard. Mater. 241e242, 146e153.

Chatterjee, S., Chatterjee, T., Lim, S.R., Woo, S.H., 2011. Effect of the addition mode ofcarbon nanotubes for the production of chitosan hydrogel coreeshell beads onadsorption of Congo red from aqueous solution. Bioresour. Technol. 102, 4402e4409.

de Lima, R.O.A., Bazo, A.P., Salvadori, D.M.F., Rech, C.M., Oliveira, D.P.,Umbuzeiro, G.A., 2007. Mutagenic and carcinogenic potential of a textile azodye processing plant effluent that impacts a drinking water source. Mutat. Res.Genet. Toxicol. Environ. Mutagen 626, 53e60.

Dotto, G.L., Lima, E.C., Pinto, L.A.A., 2012. Biosorption of food dyes onto Spirulinaplatensis nanoparticles: equilibrium isotherm and thermodynamic analysis.Bioresour. Technol. 103, 123e130.

El-Khaiary, M.I., Malash, G.F., 2011. Common data analysis errors in batch adsorp-tion studies. Hydrometallurgy 105, 314e320.

El-Khaiary, M.I., Malash, G.F., Ho, Y.S., 2010. On the use of linearized pseudo-second-order kinetic equations for modeling adsorption systems. Desalination 257, 93e101.

Freundlich, H., 1906. Adsorption in solution. Phys. Chem. Soc. 40, 1361e1368.Gupta, V.K., Gupta, B., Rastogi, A., Agarwal, S., Nayak, A., 2011. A comparative

investigation on adsorption performances of mesoporous activated carbonprepared from waste rubber tire and activated carbon for a hazardous azo dye-Acid Blue 113. J. Hazard. Mater. 186, 891e901.

Hessel, C., Allegre, C., Maisseu, M., Charbit, F., Moulin, P., 2007. Guidelines andlegislation for dye house effluents. J. Environ. Manage 83, 171e180.

Ho, Y.S., 2006. Review of second-order models for adsorption systems. J. Hazard.Mater. 136, 681e689.

Konicki, W., Pe1ech, I., Mijowska, E., Jasi�nska, I., 2012. Adsorption of anionicdye Direct Red 23 onto magnetic multi-walled carbon nanotubes-Fe3Cnanocomposite: kinetics, equilibrium and thermodynamics. Chem. Eng. J.210, 87e95.

Langmuir, I., 1918. The adsorption of gases on plane surfaces of glass, mica andplatinum. J. Am. Chem. Soc. 40, 1361e1403.

Leechart, P., Nakbanpote, W., Thiravetyan, P., 2009. Application of ‘waste’ wood-shaving bottom ash for adsorption of azo reactive dye. J. Environ. Manage 90,912e920.

Lima, E.C., Brasil, J.L., Santos, A.H.D.P., 2003. Evaluation of Rh, Ir, Ru, W-Rh, W-Ir, andW-Ru as permanent modifiers for the determination of lead in ashes, coals,sediments, sludges, soils, and freshwaters by electrothermal atomic absorptionspectrometry. Anal. Chim. Acta 484, 233e242.

Lima, E.C., Fenga, P.G., Romero, J.R., de Giovani, W.F., 1998a. Electrochemicalbehaviour of [Ru(4,4’-Me2bpy)2(PPh3)(H2O)](ClO4)2 in homogeneous solutionand incorporated into carbon paste electrodes. Application to oxidation ofbenzylic compounds. Polyhedron 17, 313e318.

Lima, E.C., Krug, F.J., Nobrega, J.A., Nogueira, A.R.A., 1998b. Determination of ytter-bium in animal faeces by tungsten coil electrothermal atomic absorptionspectrometry. Talanta 47, 613e623.

Liu, Y., Liu, Y.J., 2008. Review e biosorption isotherms, kinetics and thermody-namics. Sep. Purif. Technol. 61, 229e242.

Liu, Y., Shen, L., 2008. A general rate law equation for biosorption. Biochem. Eng. J.38, 390e394.

Machado, F.M., Bergmann, C.P., Fernandes, T.H.M., Lima, E.C., Royer, B., Calvete, T.,Fagan, S.B., 2011. Adsorption of Reactive Red M-2BE dye fromwater solutions bymulti-walled carbon nanotubes and activated carbon. J. Hazard. Mater. 192,1122e1131.

L.D.T. Prola et al. / Journal of Environmental Management 130 (2013) 166e175 175

Machado, F.M., Bergmann, C.P., Lima, E.C., Royer, B., de Souza, F.E., Jauris, I.M.,Calvete, T., Fagan, S.B., 2012. Adsorption of Reactive Blue 4 dye from watersolutions by carbon nanotubes: experiment and theory. Phys. Chem. Chem.Phys. 14, 11139e11153.

Mishra, A.K., Arockiadoss, T., Ramaprabhu, S., 2010. Study of removal of azo dye byfunctionalized multi walled carbon nanotubes. Chem. Eng. J. 162, 1026e1034.

Prola, L.D.T., Acayanka, E., Lima, E.C., Umpierres, C.S., Vaghetti, J.C.P., Santos, W.O.,Laminsi, S., Djifon, P., 2013. Comparison of Jatropha curcas shells in natural formand treated by non-thermal plasma as biosorbents for removal of Reactive Red120 textile dye from aqueous solution. Ind. Crop Prod. 46, 328e340.

Royer, B., Cardoso, N.F., Lima, E.C., Macedo, T.R., Airoldi, C., 2010. A useful organo-functionalized layered silicate for textile dye removal. J. Hazard. Mater. 181,366e374.

Sips, R., 1948. On the structure of a catalyst surface. J. Chem. Phys. 16, 490e495.Sun, C.L., Wang, C.S., 2010. Estimation on the intramolecular hydrogen-bonding

energies in proteins and peptides by the analytic potential energy function.J. Mol. Struct. 956, 38e43.

Unur, E., 2013. Functional nanoporous carbons from hydrothermally treatedbiomass for environmental purification. Microporous Mesoporous Mater. 168,92e101.

Upadhyayula, V.K.K., Deng, S., Mitchell, M.C., Smith, G.B., 2009. Application of car-bon nanotube technology for removal of contaminants in drinking water: areview. Sci. Total Environ. 408, 1e13.

Vaghetti, J.C.P., Zat, M., Bentes, K.R.S., Ferreira, L.S., Benvenutti, E.V., Lima, E.C., 2003.4-Phenylenediaminepropylsilica xerogel as a sorbent for copper determinationin waters by slurry-sampling ETAAS. J. Anal. Atom. Spectrom. 18, 376e380.

Wang, S., Ng, C.W., Wang, W., Li, Q., Hao, Z., 2012. Synergistic and competitiveadsorption of organic dyes on multiwalled carbon nanotubes. Chem. Eng. J. 197,34e40.

Weber- Jr., W.J., Morris, J.C., 1963. Kinetics of adsorption on carbon from solution.J. Sanit. Eng. Div. Am. Soc. Civil Eng. 89, 31e59.

Zhang, W., Silva, S.R.P., 2010. Raman and FT-IR studies on dye-assisted dispersionand flocculation of single walled carbon nanotubes. Spectrochim Acta A 77,175e178.