Sa

WS

a

ARRA

KSNMIA

1

irtflathtoHcr

ts[ina

a

1d

Chemical Engineering Journal 171 (2011) 431– 438

Contents lists available at ScienceDirect

Chemical Engineering Journal

j ourna l ho mepage: www.elsev ier .com/ locate /ce j

orption of norfloxacin by lotus stalk-based activated carbon and iron-dopedctivated alumina: Mechanisms, isotherms and kinetics

eifeng Liu, Jian Zhang ∗, Chenglu Zhang, Liang Renhandong Key Laboratory of Water Pollution Control and Resource Reuse, School of Environmental Science and Engineering, Shandong University, Jinan 250100, China

r t i c l e i n f o

rticle history:eceived 10 February 2011eceived in revised form 29 March 2011ccepted 30 March 2011

eywords:

a b s t r a c t

Two low-cost adsorbents, iron-doped activated alumina (Al2O3/Fe) and lotus stalk-based activated car-bon (LAC) were employed to remove norfloxacin from aqueous solutions. Sorption of norfloxacin toboth Al2O3/Fe and LAC showed a strong pH-dependent behavior. The maximum sorption capacity(21.58 �mol/g and 922.70 �mol/g) occurred at pH 6.5 and 5.5, respectively for Al2O3/Fe and LAC, which isnear the pHpzc of the sorbent. While the equilibria adsorption isotherm data on LAC fit well to the Langmuir

orptionorfloxacinechanism

ron-doped activated aluminactivated carbon

equation, both Langmuir and Freundlich models correlated the isotherm data on Al2O3/Fe quite well. Thesorption kinetics of both sorbents followed the pseudo-second order model. Several possible mechanismsfor the adsorption systems were proposed. For the sorption on Al2O3/Fe, surface complexation and cationbridging were dominant mechanisms responsible for norfloxacin removal, while hydrophobic interac-tion, cation exchange and �-electron-donor–acceptor interaction were likely important mechanisms forthe sorption on LAC.

. Introduction

The presence of antibiotics in aquatic environments is rais-ng an emerging concern recent years, regarding their potentialisks to human and ecological health. Norfloxacin (NOR) is a syn-hetic, broad-spectrum antibacterial compound that belongs touoroquinolone class organism, which presents high antibacterialctivity against both Gram-negative and Gram-positive bacteriahrough inhibition of DNA gyrase [1]. It has been extensively used inuman and veterinary treatments as well as in aquaculture applica-ion, many of which cannot be completely metabolized in humansr animals, and are eventually discharged into the environment [2].owever, traditional wastewater or water treatment proceduresannot completely remove this compound and it may present aisk to human health through contaminated drinking water.

Removal of norfloxacin residue from aquatic environment isherefore considered important and serves as an attractive casetudy. Technologies available for this purpose include adsorption3], photodegradation [4] and chlorine oxidation [5]. Adsorption

s considered a simple and effective method to remove contami-ants within water and wastewater. Till present time, only smallmount of researches focused on antibiotics’ removal from aque-

Abbreviations: LAC, lotus stalk-based activated carbon; Al2O3/Fe, iron-dopedctivated alumina; NOR, norfloxacin.∗ Corresponding author. Tel.: +86 531 88363015; fax: +86 531 88364513.

E-mail address: zhangjian00@sdu.edu.cn (J. Zhang).

385-8947/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.cej.2011.03.099

© 2011 Elsevier B.V. All rights reserved.

ous solutions using alternative adsorbents. Previous research onNOR sorption by different environmental matrices included sil-ica, alumina [3], soils [6], and carbon nanotubes [7], but limitedinformation is known about its adsorption by activated car-bon. Activated carbon is the most widely used adsorbent todaybecause of its extended surface area, well developed porosityand high adsorption capacity. However, its feasibility for largescale wastewater application is restricted due to the economicdefect, resulting mainly from the high-cost precursors, such ascoal, wood, and coconut shells. Therefore, developing economi-cally available activated carbon or alternative adsorbent whichis inexpensive and comes in abundant amount naturally is ofinterest.

Activated alumina (�-Al2O3) is a commercially available adsor-bent that has been extensively used to remove various watercontaminants. To our knowledge, there have been very few inves-tigations on the sorption of norfloxacin with oxide minerals.Lorphensri et al. [3] has reported the adsorption of norfloxacinby �-Al2O3, but such research is still limited in literature bynow. Impregnation with metal ions has proven to be an effec-tive way to improve the sorption ability of activated aluminafor certain substances [8,9]. The impregnates can change thenature and surface properties of �-alumina, on which not onlyadsorption, but also chemical reaction and catalytic decomposition

of different adsorbates may take place [8]. To our best knowl-edge, however, no research has been conducted on the removalof NOR by iron modified alumina presently. Consequently, it isattractive to understand the sorption behavior and mechanisms

4 eering

fr

owtrrstape

2

2

Bt3acMtgt

2

ad“prowitlaib�a

aipaiiBbca

2

2oh

32 W. Liu et al. / Chemical Engin

or the uptake of NOR by this effective and inexpensive mate-ial.

The objectives of this study were to investigate the performancef iron-doped activated alumina for adsorption of norfloxacin fromastewater. The performance of Al2O3/Fe is also compared with

hat of a low-cost activated carbon prepared from a hydrophytesesidue, lotus stalk. The NOR sorption behavior on these two mate-ials has never been reported in literature. Detailed adsorptiontudies were conducted as a function of pH to obtain insight intohe sorption performances and mechanisms. Sorption isothermsnd kinetics were also investigated. The results from this study willrovide useful information for potential NOR removal as well as itsnvironmental risk assessment.

. Materials and methods

.1. Chemicals

Norfloxacin (99.8%, standard grade) was purchased from Fuchiiological Technology Co. (China) and used as received. Impor-ant physicochemical properties of norfloxacin are [10,11] MW:19.331 g/mol; aqueous solubility: 161,000 mg/L at pH 5, 400 mg/Lt pH 7, and 910 mg/L at pH 9; and pKa: 6.22 and 8.51. The commer-ial �-alumina support, which was supplied by Wufeng Aluminumagnesium Technology Co., Zibo, China, had stated minimum puri-

ies of 99%. Other chemicals used in this study were of analyticalrade. Distilled water was used throughout for solution prepara-ions.

.2. Sorbent preparation and characterization

Lotus stalk-based activated carbon (LAC) was fabricatedccording to the method described elsewhere [12]. The iron-oped activated alumina (Al2O3/Fe) was prepared by anequilibrium–deposition–filtration” method, to obtain a high dis-ersion of iron deposited phase on alumina support. In detail, aseceived activated alumina was crushed and sieved into particlesf 100–140 mesh prior to use. 2 g of the �-Al2O3 support was mixedith 400 mL of 10−3 M Fe(NO3)3 solution, then shaken at 200 rpm

n a temperature controlled shaker (SHZ-88, Shanghai) at roomemperature. After 20 h, which is sufficient for the maximum ironoading, the solid mass was separated from the solvent by a vacuumir pump, and received a gentle wash with distilled water. The ironmpregnated �-Al2O3 was then dried at 105 ◦C for 12 h followedy calcination in air at 350 ◦C for 2 h. The Fe loading amount on-Al2O3 was 0.97 wt% as detected by inductively coupled plasmatomic emission spectroscopy (ICP-AES).

Surface area and pore characteristics of adsorbents (Al2O3/Fend LAC) were determined from nitrogen adsorption/desorptionsotherms at 77 K using a surface area analyzer (Quantachrome Cor-oration, USA). The surface chemistries of Al2O3/Fe and LAC beforend after NOR adsorption were detected using a Fourier transformnfrared radiation (FTIR) spectrometer (Fourier-380 FT-IR, Amer-ca), where the spectra were recorded from 400 to 4000 cm−1.oehm titration method [13] was employed to determine the num-er of surface functional groups on LAC. The pH at the point of zeroharge (pHpzc) of Al2O3/Fe and LAC particles were estimated from

batch equilibrium method described by Babic et al. [14].

.3. Sorption experiments

Batch equilibrium sorption experiments were performed at1 ± 1 ◦C. Synthetic norfloxacin solution with initial concentrationf 100 �M was prepared freshly within 1 day of use. This relativelyigh concentration value was chosen to represent the condition of

Journal 171 (2011) 431– 438

wastewater directly near a contamination source, e.g. pharmaceu-tical companies. Experimentally, a weighed quantity of sorbents(0.01 g for LAC and 0.4 g for Al2O3/Fe) were dispersed in 100 mLNOR solution at desired initial concentration, pH, and ionic strength(10 mM NaCl). Samples were equilibrated by shaking in a waterbath at 200 rpm in dark for 24 h (LAC) or 16 h (Al2O3/Fe). Afterequilibration, the samples were filtered and the residual NOR con-centration was analysed by an UV–vis spectrophotometer (UV-754,Shanghai) at the maximum absorption wavelength of 273 nm. Solu-tion pH was adjusted to the required value with 0.1 M HCl and 0.1 MNaOH. Control experiments without adsorbents were conducted inthe same manner to account for possible norfloxacin losses as sorp-tion to glass tubes and other reactions in solution, although littlesolute loss was detected under our experimental conditions.

The solid–water sorption coefficient, Kd (L/m2), was employedin this study, because the solid and aqueous-phase distribution ofa compound determines its mobility, bioavailability, and suscep-tibility to transformation reactions [6]. Kd is defined as Kd = CS/Cw,where Cw (�mol/L) is the equilibrium aqueous NOR concentration.CS (�mol/m2) is the equilibrium sorbed concentration, which canbe calculated as

CS = (C0 − Cw)VMSA

(1)

where C0 (�mol/L) is the control NOR concentration at the end ofthe experiment; V (L) is the volume of solution; Ms (g) is the massof solid sorbent, and A (m2/g) is the sorbent specific surface area.Area-normalized concentrations were computed because interac-tions of norfloxacin with the sorbents were anticipated to occur byan adsorptive mechanism.

The effect of solution pH on norfloxacin removal by the sorbentswas tested in pH range of 3.5–10.5. Experiments were conducted ina similar manner as described above, except that the sorbent dosewas 0.007 g/100 mL for LAC and 0.3 g/100 mL for Al2O3/Fe to con-trol the NOR removal percent between 40% and 90%. For Al2O3/Fe,blank flasks without NOR solute were also performed to test thedissolution amount of Al3+ and Fe3+ in distilled water. DissolvedAl3+ and Fe3+ concentrations in the solution were analysed by ICP-AES. Adsorption isotherms were conducted at three pH levels withinitial NOR concentrations ranging from 80 to 500 �mol/L.

2.4. Desorption experiments

Immediately after the equilibrium sorption experiments asdescribed above, the adsorbent was filtered and gently washed withdistilled water to remove any unadsorbed NOR molecular trappedbetween the adsorbent particles. Next, a desorption treatment wasapplied by adding 100 mL of either (i) 0.01 M NaCl, (ii) 1 M NaCl,(iii) 0.2 M MgCl2, (iv) methanol, (v) 0.1 M NaOH or (vi) 0.25 M EDTA(only for Al2O3/Fe) to the solid. Desorption solutions were shaken indark for 24 h (LAC) or 16 h (Al2O3/Fe) before analysis for dissolvednorfloxacin mass.

3. Results and discussion

3.1. Physical properties of adsorbent

Different structure characteristics of the two samples are evi-dent as shown from the pore size distributions in Fig. 1. Thepores of LAC mostly locate in the range of micropores and smallmesopores with average pore width of 3.41 nm, indicating a micro-

mesoporous structure in activated carbon. While Al2O3/Fe showsa typically mesoporous structure (most pores 3–8 nm, averagepore width 5.33 nm). The calculated BET surface area of LAC is1289.1 m2/g, in which the micropore and external surface areas are

W. Liu et al. / Chemical Engineering Journal 171 (2011) 431– 438 433

2016128400.0

0.1

0.2

0.3

0.4

0.5Vo

lum

e (c

m3 /n

m/g

)Vo

lum

e (c

m3 /n

m/g

)

Pore whith (nm)

a

2016128400.00

0.02

0.04

0.06

0.08

0.10

0.12

Pore whith (nm)

b

5a

3

cftBgnnesaoc1

s3Ttg

tt

5001000150020002500300035004000

a

Tran

smitt

ance

%

Wavenumbers (cm-1)

Wavenumbers (cm-1)

LAC

LAC-NOR

1452.7 1260.3

803.1746.4

1558.71181.8

1071.7

5001000150020002500300035004000

Al2O3/Fe-NOR

Tran

smitt

ance

%

b

Al2O3/Fe

3441.4

1636.4

550.5

Fig. 1. Pore size distributions for LAC (a) and Al2O3/Fe (b).

09.4 m2/g and 779.7 m2/g, respectively. Al2O3/Fe has a BET surfacerea of 140.6 m2/g with no microspores.

.2. Chemical properties of adsorbent

Surface oxygen-containing functional groups within activatedarbon determine its surface acidity/basicity, and act as a crucialactor affecting its adsorption capacity and selectivity. The exis-ence of several acidic and basic groups on LAC can be verified fromoehm titration. Specifically, it contains 0.749 mmol/g of carboxylicroup, 0.405 mmol/g of lactonic group and 0.567 mmol/g of phe-olic group, thus giving a total acidity of 1.721 mmol/g. The totalumber of basic functional groups was 0.894 mmol/g. The pres-nce of oxygen functionalities can also be confirmed by the FTIRpectra of LAC. As shown in Fig. 2a, the spectra band observedt 1558.7 cm−1 corresponds to the stretching vibrations of C O,riginated from carboxyl, lacton or basic functional groups (e.g.hromene and pyrone structures). Peaks around 1181.8 cm−1 and071.7 cm−1 can be assigned to C–O bonding [15].

Surface functional groups of Al2O3/Fe as determined by FTIRpectra are depicted in Fig. 2b. The very broad peak centered at441.4 cm−1 is ascribed to be O–H group in Al–O–H or Fe–O–H [16].he peak around 1636.4 cm−1 comes from H–O–H bending whilehe peak around 550.5 cm−1 can be recognized to Al–O or Fe–O

roups [17].

The point of zero charge of adsorbent (pHpzc) is a pH point wherehe amount of negative charges on a sorbent surface just equalshe amount of positive charges [16]. When solution pH < pHpzc,

Fig. 2. FTIR spectra of LAC (a) and Al2O3/Fe (b) before and after norfloxacin adsorp-tion.

the adsorbent surface will be protonated by the excess H+ ions,thereby acquiring a positive charge. When solution pH > pHpzc, theadsorbent surface will be deprotonated by the OH− ions and thusacquiring a negative charge. The plots to determine the point ofzero charges for Al2O3/Fe and LAC are shown in Fig. S1. The pHpzc

for Al2O3/Fe and LAC were found to be 6.53 and 5.48, respectively.The relatively low pHpzc for LAC is probably caused by the largeamount of acidic functional groups on its surface.

3.3. Effect of pH and sorption mechanisms

In adsorption of norfloxacin, solution pH is considered to be themost important factor affecting the sorption performance, as it canalter the charge and species of norfloxacin as well as the sorbentsurface properties. To evaluate the influence of pH on adsorption,studies of norfloxacin adsorption onto Al2O3/Fe and LAC were con-ducted within pH range of 3.5–10.5.

3.3.1. Mechanisms of NOR sorption to Al2O3/Fe

The extent of norfloxacin sorption onto Al2O3/Fe as a func-

tion of pH is presented in Fig. 3a. Strong adsorption occurredin pH range of 5.5–7.5 with the maximum adsorption extent(21.58 �mol/g, removal rate 90.14%) at pH ∼6.5, which is near the

434 W. Liu et al. / Chemical Engineering

111098765430.000

0.004

0.008

0.012

0.016

0.020

0

20

40

60

80

100

AnionZwitterionCation

Kd

(L/m

2 )K

d (L

/m2 )

pH

a

Spec

ious

frac

tion

(%)

111098765430.00

0.04

0.08

0.12

0.16

0.20

0

20

40

60

80

100

AnionZwitterion

Spec

ious

frac

tion

(%)

pH

Cation

b

Fig. 3. pH-sorption profile of norfloxacin onto Al2O3/Fe (a) and LAC (b)(C = 100 �mol/L, I = 0.01 M NaCl, sorbent dose = 3 g/L for Al O /Fe and 0.07 g/L forLf

pda[bNb

solubility of oxide minerals through the ligand-promoted dissolu-

Fo

0 2 3

AC). The dashed lines represent the fraction of cationic, zwitterionic and anionicorms of norfloxacin.

Hpzc of Al2O3/Fe. Above and below this rang, rapid and gradualecrease of the NOR adsorption was observed. A similar trend waslso found during the sorption of norfloxacin to aluminum oxides

3]. The weak sorption at extremely low and high pH ranges cane attributed to the great electrostatic repulsive force betweenOR and the mineral surface, i.e. both norfloxacin and Al2O3/Feecame predominantly positively charged at low pH (<4.5) and

ig. 4. Schematic diagrams for surface complexation (a) and cation bridging (b) interaction the edge of Al2O3/Fe surface.

Journal 171 (2011) 431– 438

negatively charged at high pH (>8.5), thus could repel each other.The adsorption mechanisms of fluoroquinolone antimicrobials tosoils [18] and oxide minerals [3,19] have been investigated byseveral researchers. Generally, primary sorption interactions ofantimicrobials with model sorbent phases have been documentedto be cation exchange [18], surface complexation [19] and cationbridging [20]. Based on the experimental results and previousresearches, several possible adsorption mechanisms of norfloxacinonto Al2O3/Fe can be proposed in this study.

The first mechanism is surface complexation mechanism. It isknown that norfloxacin is capable of bonding with multivalentmetal ions such as Al3+, Fe3+ and Mn2+ to form strong com-plexes [21,22]. Surface complexation has been well establishedfor the sorption of fluoroquinolone compounds to metal oxides[19]. It is reasonable to hypothesis that stable complexation formedbetween norfloxacin and Al2O3/Fe surface during the sorption. Con-sidering the structures of Al/Fe–norfloxacin complexes, a possibleNOR–Al2O3/Fe chelating interaction is hypothesized to occur in thisstudy [23]. Norfloxacin is likely to form an energetically favorablemononuclear bidentate complex (i.e., a six-membered ring) withthe Fe and Al atom on the Al2O3/Fe surface through the keto oxy-gen and one of the O’s of the carboxylate. The proposed surfacecomplexation structure is depicted in Fig. 4a.

Cation bridging is another important mechanism probablyinvolved in the sorption of NOR to Al2O3/Fe. Because of the pres-ence of free H+ in the solution, Al2O3/Fe may be protonated withextra H+ bonding to the surface. In addition, the edge Al/Fe atomsin Al2O3/Fe mineral are ready to be protonated in forms of Al3+

and/or Fe3+, some of which bond in the diffuse interlamellar space.These edge sites cations (H+, Al3+ and Fe3+) can intercalate withthe anionic COO− group of NOR to form a ‘cation bridge’. Thiscation bridge interaction is generally very stable in aqueous envi-ronments [6,20]. Moreover, Al3+ and Fe3+ cations are considered tobe more effective than H+ in promoting adsorption of norfloxacinon Al2O3/Fe surface due to their stronger electric charges, providinga stronger electrostatic attraction for the anionic carboxyl moietyof norfloxacin. The cation bridge model of adsorption of NOR toAl2O3/Fe surface is presented in Fig. 4b. The same mechanism hasalso been proposed in the sorption of quinolone acid derivatives toclay mineral surfaces by Nowara et al. [20]. It is fairly well-knownthat sorption of chelating compounds will significantly increase the

tion mechanism [24]. However, the solubility of Al2O3/Fe was notnotably promoted by contacting with NOR as compared to distilledwater at pH 4.5–8.5 when intensive adsorption occurred (Fig. 5).

ns between norfloxacin and Al2O3/Fe surface. Mn+ represents H+, Al3+ or Fe3+ ions

W. Liu et al. / Chemical Engineering

111098765430

1

2

3

4

5

6

7

111098765430.0

0.1

0.2

0.3

0.4

Solu

able

Al c

once

ntra

tion

(mg/

L)

pH

no NOR NOR

Fp

Tsd

smapioNtriMpmNnMnehsTflmFsinf

3

tmsoptarc

ig. 5. Solubility of Al and Fe (inset) from Al2O3/Fe sample as a function of pH in theresence and absence of norfloxacin.

his may be attributed to the cation bridging interaction that ‘con-umed’ Al3+ and Fe3+ cations and hindered their dissolution andiffusion into the bulk solution.

FTIR spectra of Al2O3/Fe after equilibrating with norfloxacinhowed little difference from that of virgin Al2O3/Fe (Fig. 2b). Thisay be due to the low NOR sorption amount on Al2O3/Fe (922.7

nd 21.5 �mol/g for LAC and Al2O3/Fe, respectively at optimalH), causing minimal changes on the IR signals. Desorption stud-

es were conducted to obtain further insight into the mechanismf interactions between norfloxacin and the adsorbents (Table 1).aCl solutions were employed to respect the importance of elec-

rostatic interaction. Less than 8% of the sorbed norfloxacin wasecovered by NaCl solutions of various strengths, implying a lessmportant role of electrostatic interactions in the sorption process.

agnesium(II) cations are known to form solution phase com-lexes with norfloxacin [25]. Only 8.42% of the sorbed norfloxacinass was desorbed by 0.2 M MgCl2, suggesting that attachment ofOR molecules with the Al2O3/Fe surface is very strong and can-ot be easily dissociated by the complexation of NOR with Mg2+.ethanol was used to determine the role of hydrophobicity in

orfloxacin sorption. Less than 2% of sorbed norfloxacin was recov-red by a methanol wash, indicating a negligible contribution ofydrophobic interaction to the adsorption. As much as 83.1% of theorbed norfloxacin was desorbed by the presence of NaOH solution.his implied that the dominant mechanism in the sorption of nor-oxacin on Al2O3/Fe should be strong chemical interactions, whichay originate from surface complexation and/or cation bridging.

inally, EDTA, a strong chelating agent capable of forming innerphere complexes with iron/alumina oxides, was used to compet-tively displace sorbed norfloxacin. More than 60% of the sorbedorfloxacin was recovered by EDTA addition, providing evidence

or the surface complexation mechanism.

.3.2. Mechanisms of NOR sorption to LACFig. 3b shows the sorption coefficients as a function of pH for

he adsorption of NOR onto LAC. The adsorption achieved its maxi-um (922.70 �mol/g, removal rate 95.10%) at pH 5.5 and decreased

ignificantly when pH was higher or lower than this range. Thebserved sorption behavior can be attributed to a combination ofH-dependent speciation of norfloxacin and surface charge charac-

eristics of LAC. Norfloxacin has two proton-binding sites (carboxylnd piperazinyl group) with reported pKa values of 6.22 and 8.51,espectively. With these two pKa values, norfloxacin can exist inationic form, zwitterionic form or anionic form depending on the

Journal 171 (2011) 431– 438 435

solution pH (see Fig. S2). As can be seen from the speciation (Fig. 3),at pH below 6.2 the cationic form of NOR is predominant. At pH6.2–8.5 the three forms coexist with the zwitterionic form beingdominant. The anionic form dominates at pH higher than 8.5. Thesurface of LAC was positively charged at pH lower than pHpzc (5.48)and negatively charged at pH higher than pHpzc. Thus, electro-static interaction between charged NOR and charged LAC surface isexpected to occur in the sorption process. At pH < 4.5 and pH > 8.5,most NOR molecules had the same sign of charge as LAC and couldrepel each other, causing the sorption greatly depressed.

As shown in Fig. 3b, at pH 4.5–6.0 where the NOR sorptionwas strong, cationic NOR was the dominant specie in the solu-tion (98%–62%). Especially, the cationic form account for 83.98%of the total molecule when the maximum sorption occurred at pH∼5.5. This suggested that cationic NOR acted as a vital contributorto the overall norfloxacin sorption onto LAC. Accordingly, cationexchange is proposed to be an important mechanism participat-ing in the adsorption process. In principle, the protons binding toLAC surface due to the extra H+ in bulk solution were exchangedby the positively charged piperazinyl group within norfloxacin. Thisexchange mechanism can be confirmed by the fact that the solutionpH increased after NOR sorption at pH 4.5–5.5, but the increasingextent was much lower than that of distilled water with LAC. Itis possible that cation exchange released partial protons from LACsurface to the bulk solution. Note that at pH 5.5–6.5, part of theLAC surface was still protonated so cation exchange could also takeplace. Cation exchange has been proposed to be an important mech-anism in the sorption of fluoroquinolone compounds to soils [6,18].The schematic diagram for this mechanism is presented in Fig. S3.

Activated carbon is an organic semiconductor with delocalized� electrons on its surfaces, thus shows electron–donor properties[26]. The �-electron-donor–acceptor (EDA) interaction is, there-fore, considered to participate in the sorption of NOR onto LAC.Phenol groups with an unshared pair of � electrons on the oxy-gen atom, as well as deprotonated carboxyl groups on the carbonsurface can act as electron-donors [26]. On the other hand, the ben-zene ring on NOR can function as a �-electron-acceptor due to thestrong electron-withdrawing ability of the fluorine group. Based onthe EDA theory, a �-donor compound and a �-acceptor compoundare ready to interact with each other and the interaction is generallystrong.

Another possible mechanism is hydrophobic interactionbetween the carbon surface and NOR molecule. The basic graphiticstructure of activated carbon provides hydrophobic sites, whichcan interact with hydrophobic molecules. Although NOR is mosthydrophobic at pH between its two pKa values, where the sorp-tion extent was not the maximum, hydrophobic interaction can stilltook place during the sorption process. As such, the contribution ofhydrophobic interaction to NOR sorption could not be excluded,despite the lack of substantial evidence for this process.

FTIR spectra of LAC after adsorption confirmed the presenceof norfloxacin on LAC surface as shown in Fig. 2a. New peaks at1452.7, 803.1 and 746.39 cm−1 in the FTIR spectrum of LAC afterNOR sorption occurred, all of which can be assigned to bendingmodes of aromatic rings, indicating the attachment of benzene ringof NOR onto LAC surface. This serves an evidence for the EDA inter-action between LAC and norfloxacin molecules. Additional peak at1260.3 cm−1 can also be detected, which came from the asymmet-ric C–N stretching, suggesting the involvement of piperazinyl groupof NOR in the interaction with LAC. This result is in keeping withthe cation exchange mechanism proposed above.

As shown in Table 1, less than 9% of the sorbed norfloxacin was

recovered from LAC by NaCl solutions of various strengths. Thisimplied that the sorption mechanism was stronger than nonspe-cific electrostatic interactions for NOR. The amount of desorbednorfloxacin by MgCl2 increased to about 21.2% of the sorbed mass,

436 W. Liu et al. / Chemical Engineering Journal 171 (2011) 431– 438

Table 1Percentage of sorbed norfloxacin mass desorbed by various treatments.

Adsorbent 0.01 M NaCl 1 M NaCl 0.2 M MgCl2 Methanol 0.1 M NaOH 0.25 M EDTA

Al2O3/Fe 7.21% 4.72% 8.42% 1.86% 83.10% 65.82%LAC 2.26% 8.26% 21.16% 13.39% 60.81% –

“–” denotes treatments not tested.

Table 2Langmuir and Freundlich isotherm parameters for adsorption of norfloxacin on Al2O3/Fe and LAC.

Adsorbent pH Langmuir isotherm Freundlich isotherm

Qm (�mol/g) KL (L/�mol) R2 Kf (�mol/g (L/�mol)1/n) 1/n R2

Al2O3/Fe 4.5 62.8931 0.0171 0.9817 3.9279 0.4656 0.98776.5 102.0408 0.0195 0.9931 6.0599 0.5156 0.99079.5 64.1026 0.01321 0.9869 5.8580 0.3960 0.9931

000

ptwbpttes

3

aiLvnT[w

Q

Q

wr(esd

tsrvts

TK

LAC 4.5 1250.0000 0.1905

5.5 1428.5714 0.1250

9.5 1111.1111 0.0714

resumably due to competing NOR complexation with Mg2+ inhe aqueous phase. About 13.4% of the sorbed norfloxacin massas recovered by methanol treatment, suggesting that hydropho-

ic interaction between LAC and norfloxacin molecules did tooklace but not a major mechanism. Finally, as much as 60.8% ofhe sorbed norfloxacin was desorbed by 0.1 M NaOH, indicatinghat strong chemical interactions, which may come from cationxchange and/or EDA interaction, are dominant mechanisms in theorption norfloxacin on LAC.

.4. Sorption isotherms

Sorption isotherm is critical to evaluate the sorption capacity ofdsorbents as well as to understand the nature of sorbate–sorbentnteractions. The adsorption isotherms of NOR onto Al2O3/Fe andAC were examined. The initial experiments were conducted at pHalues of pH 4.5, 6.5, 9.5 for Al2O3/Fe and 4.5, 5.5, 9.5 for LAC, span-ing the sorbate–sorbent charge states for these two adsorbents.wo commonly used models, the Langmuir [27] and Freundlich28] equations were employed to correlated the experimental data,hich can be expressed respectively as

e = QmKLCe

1 + KLCe(2)

e = KfC1/ne (3)

here Qe (�mol/g) is the amount of NOR adsorbed at equilib-ium; Qm (�mol/g) is the maximum NOR adsorption capacity; KLL/�mol) is the Langmuir constant that is related to the apparentnergy of adsorption; Kf (�mol/g(L/�mol)1/n) is the Freundlich con-tant indicating the relative adsorption capacity, and n is a constantepicting the sorption intensity.

As-fitted Langmuir and Freundlich parameters for the adsorp-ion of norfloxacin onto Al2O3/Fe and LAC are listed in Table 2. Ashown in Table 2, both the Langmuir and Freundlich models cor-

elated the sorption isotherms on Al2O3/Fe quite well, with all R2

alues higher than 0.98. The Langmuir equation is derived fromhe assumption of monolayer adsorption on specific homogenousites, while the Freundlich model represents adsorption on het-

able 3inetic parameters of the pseudo-first and second-order models for adsorption of norflox

Adsorbent Qexp (�mol/g) Pseudo-first-order model

k1(1/h) Qcal (�mol/g) R2

Al2O3/Fe 22.9515 0.4312 19.7608 0.9913

LAC 954.0353 0.2312 734.7277 0.9715

.9994 683.4818 0.1102 0.9386

.9996 624.7178 0.1459 0.9060

.9974 414.0141 0.1761 0.9095

erogeneous surfaces. The good application of both models impliedthat besides adsorption onto specific sorption sites through chem-ical interactions such as surface complexation and cation bridging,sorbate–sorbate interaction may also play an important role in thesorption process by Al2O3/Fe. This is reasonable because NOR solu-tion under environmental pH is a complex mixture of cationic,zwitterionic and anionic forms, the interactions between whichwould be complicated and nonspecific interactions such as elec-trostatic repulsion may be significant. The maximum NOR sorptionoccurred at pH 6.5 according to the Langmuir parameter Qm andFreundlich parameter Kf. Values of 1/n were less than 1, suggestingthat the sorption of NOR onto Al2O3/Fe was a favorable process.

On the other hand, the sorption isotherms of NOR on LAC can bedescribed better by the Langmuir model than the Freundlich modelfor all pH levels, due to the high correlation coefficient values of theformer (R2 > 0.997). This result implied that monolayer sorption ofNOR on LAC probably occurred. It could be that cation exchangeand/or EDA interaction dominated the process, and the sorptionamount reached the maximum when the available sites were sat-urated with NOR molecules. The Freundlich constant 1/n were lessthan 1 indicating a favorable adsorption process [3].

3.5. Sorption kinetics

The kinetics for adsorption of norfloxacin onto Al2O3/Fe andLAC were also examined. The experimental data were fitted by thepseudo-first-order model Eq. (4) and pseudo-second-order modelEq. (5) [29]:

ln(Qe − Qt) = ln Qe − k1t (4)

t

Qt= 1

k2Q 2e

+ t

Qe(5)

where Qe and Qt (�mol/g) are the amounts of NOR adsorbed atequilibrium and at time t (h); k1(1/h) and k2 (g/(�molh)) are thepseudo-first-order and pseudo-second-order rate constants.

acin on Al2O3/Fe and LAC.

Pseudo-second-order model

�q (%) k2 (g/(�molh)) Qcal (�mol/g) R2 �q (%)

38.742 0.0427 24.1546 0.9990 13.02655.119 0.0008 1000.0000 0.9990 12.891

W. Liu et al. / Chemical Engineering

201510500

5

10

15

20

25

t (h)

a

3025201510500

200

400

600

800

1000

Qt (

umol

/g)

Qt (

umol

/g)

t (h)

b

Fig. 6. Sorption kinetics of norfloxacin onto Al2O3/Fe (a) and LAC (b) and mod-elA

ts

�

wcmba

wair(TcprfttmmAa

[

[

[

[

[

[

ling using the pseudo-first-order (dashed lines) and pseudo-second-order (solidines) equations (C0 = 100 �mol/L, I = 0.01 M NaCl, pH ∼6.6, sorbent dose = 4 g/L forl2O3/Fe and 0.1 g/L for LAC).

To determine the validity of each model for the adsorption sys-em, the correlation coefficients (R2), together with a normalizedtandard deviation �q (%) were employed. �q is defined as

q(%) = 100

√∑[(Qexp − Qcal)/Qexp]2

N − 1(6)

here Qexp and Qcal (�mol/g) are the experimental and model-alculated NOR uptake amounts; N is the number of measurementsade. It is clear that a smaller �q value means little deviation

etween the experimental data and that calculated from the modelnd vice versa.

Adsorption kinetics of norfloxacin on Al2O3/Fe and LAC alongith the correlated results using the first and second order models

re presented in Fig. 6, and the as fitted parameters are summarizedn Table 3. For both sorbents, the pseudo-second-order model cor-elates the kinetic data quite well with high correlation coefficientsR2 0.999) and good agreements between Qcal and Qexp (�q < 15%).he pseudo-second-order model assumes that the sorption rate isontrolled by chemical interaction and the sorption capacity is pro-ortional to the number of active sites on the sorbent [29]. Thisesult is in keeping with the sorption mechanisms proposed aboveor Al2O3/Fe and LAC. It can also be seen from Fig. 6 and Table 3hat the equilibrium NOR sorption amounts on LAC was about 40imes of that on Al2O3/Fe, which may be largely attributed to the

uch higher specific surface area of the former that can provideore contacting area and specific sorption sites for NOR. Note thatl2O3/Fe is still an effective adsorbent for norfloxacin due to itsdvantages such as abundance and low cost.

[

[

Journal 171 (2011) 431– 438 437

4. Conclusions

The present study shows that norfloxacin can be effectivelyremoved by low-cost adsorbents iron-doped activated alumina(Al2O3/Fe) and lotus stalk-based activated carbon (LAC). The max-imum sorption capacity occurred near the pHpzc of the sorbents.Adsorption isotherms on LAC fit well to the Langmuir equa-tion, while both Langmuir and Freundlich models correlated theisotherm data on Al2O3/Fe quite well. The sorption kinetics of bothsorbents followed the pseudo-second order model. While com-plexation and cation bridging appears dominant mechanisms fornorfloxacin sorption on Al2O3/Fe, hydrophobic interaction, cationexchange and �-electron-donor–acceptor interaction were likelyimportant mechanisms for the sorption on LAC.

Acknowledgements

The work was supported by the National Water Special Project(No. 2009ZX07210-009-04), the National Natural Science Foun-dation of China (No. 21007032) and the Independent InnovationFoundation of Shandong University (No. 2009JQ009).

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.cej.2011.03.099.

References

[1] R. Hirsch, T.A. Ternes, K. Haberer, K.L. Kratz, Occurrence of antibiotics in theenvironment, Sci. Total Environ. 225 (1999) 109–118.

[2] E.M. Golet, A.C. Alder, W. Giger, Environmental exposure and risk assessment offluoroquinolone antibacterial agents in wastewater and river water of the GattValley watershed, Switzerland, Environ. Sci. Technol. 36 (2002) 3645–3651.

[3] O. Lorphensri, J. Intravijit, D.A. Sabatini, T.C.G. Kibbey, K. Osathaphan, C. Sai-wan, Sorption of acetaminophen, 17-ethynyl estradiol, nalidixic acid, andnorfloxacin to silica, alumina, and a hydrophobic medium, Water Res. 40 (2006)1481–1491.

[4] T. Paul, P.L. Miller, T.J. Strathman, Visible-light-mediated TiO2 photocataly-sis of fluoroquinolone antibacterial agents, Environ. Sci. Technol. 41 (2007)4720–4727.

[5] M.C. Dodd, A.D. Shah, U.V. Gunten, C. Huang, Interactions of fluoroquinoloneantibacterial agents with aqueous chlorine: reaction kinetics, mechanisms, andtransformation pathways, Environ. Sci. Technol. 39 (2005) 7065–7076.

[6] R.A. Figueroa-Diva, D. Vasudevan, A.A. MacKay, Trends in soil sorptioncoefficients within common antimicrobial families, Chemosphere 79 (2010)786–793.

[7] Z. Wang, X. Yu, B. Pan, B. Xing, Norfloxacin sorption and its thermodynamics onsurface-modified carbon nanotubes, Environ. Sci. Technol. 44 (2010) 978–984.

[8] A.K. Khattak, K. Mahmood, M. Afzal, M. Saleem, R. Qadeer, Thermodynamicstudies of methanol adsorption on metal impregnated �-alumina samples,Colloids Surf. A 236 (2004) 103–110.

[9] S.S. Tripathy, J.L. Bersillon, K. Gopal, Removal of fluoride from drinking waterby adsorption onto alum-impregnated activated alumina, Sep. Purif. Technol.50 (2006) 310–317.

10] D.L. Ross, C.M. Riley, Aqueous solubilities of some various substituted quinoloneantimicrobials, Int. J. Pharm. 63 (1990) 237–250.

11] K. Takacs-Novák, M. Józan, I. Hermecz, G. Szász, Lipophilicity of antibacterialfluoroquinolones, Int. J. Pharm. 79 (1992) 89–96.

12] L. Huang, Y. Sun, T. Yang, L. Li, Adsorption behavior of Ni (II) on lotus stalksderived active carbon by phosphoric acid activation, Desalination 268 (2011)12–19.

13] H.P. Boehm, Some aspects of the surface chemistry of carbon blacks and othercarbons, Carbon 32 (1994) 759–769.

14] B.M. Babic, S.K. Milonjic, M.J. Polovina, B.V. Kaludierovic, Point of zero chargeand intrinsic equilibrium constants of activated carbon cloth, Carbon 37 (1999)477–481.

15] E.K. Putra, R. Pranowo, J. Sunarso, N. Indraswati, S. Ismadji, Performance ofactivated carbon and bentonite for adsorption of amoxicillin from wastewater:mechanisms, isotherms and kinetics, Water Res. 43 (2009) 2419–2430.

16] S . Kubilay, R. Gürkan, A. Savran, T. S ahan, Removal of Cu(II), Zn(II) and Co(II)ions from aqueous solutions by adsorption onto natural bentonite, Adsorption13 (2007) 41–51.

17] E. Finocchio, G. Busca, Characterization and hydrocarbon oxidation activity ofcoprecipitated mixed oxides Mn3O4/Al2O3, Catal. Today 70 (2001) 213–225.

4 eering

[

[

[

[

[

[

[

[

[

[inum, J. Am. Chem. Soc. 40 (1918) 1361–1403.

[28] H. Freundlich, W. Heller, The Adsorption of cis- and trans-Azobenzene, J. Am.

38 W. Liu et al. / Chemical Engin

18] A.A. MacKay, D.E. Seremet, Probe compounds to quantify cation exchange andcomplexation interactions of ciprofloxacin with soils, Environ. Sci. Technol. 42(2008) 8270–8276.

19] C. Gu, K.G. Karthikeyan, Sorption of the antimicrobial ciprofloxacin to alu-minum and iron hydrous oxides, Environ. Sci. Technol. 39 (2005) 9166–9173.

20] A. Nowara, J. Burhenne, M. Spiteller, Binding of fluoroquinolone carboxylic acidderivatives to clay minerals, J. Agric. Food. Chem. 45 (1997) 1459–1463.

21] J. Al-Mustafa, B. Tashtoush, Iron(II) and iron(III) perchlorate complexes ofciprofloxacin and norfloxacin, J. Coord. Chem. 56 (2003) 113–124.

22] P.T. Djurdjevic, M. Jelikic-Stankov, D. Stankov, Fluorescence reaction and com-plexation equilibria between norfloxacin and aluminium(III) ion in chloride

medium, Anal. Chim. Acta 300 (1995) 253–259.

23] A.I. Drakopoulos, P.C. Ioannou, Spectrofluorimetric study of the acid–base equi-libria and complexation behavior of the fluoroquinolone antibiotics ofloxacin,norfloxacin, ciprofloxacin and pefloxacin in aqueous solution, Anal. Chim. Acta354 (1997) 197–204.

[

Journal 171 (2011) 431– 438

24] E. Molis, O. Barrès, H. Marchand, E. Sauzéat, B. Humbert, F. Thoms, Initialsteps of ligand promoted dissolution of gibbsite, Colloids Surf. A 163 (2000)283–292.

25] I. Turel, The interactions of metal ions with quinolone antibacterial agents,Coord. Chem. Rev. 232 (2002) 27–47.

26] E.D. Lavrinenko-Ometsinskaya, K.A. Kazdobin, V.V. Strelko, Quantum-chemicalevaluation of electron–donor properties of ordinary and modified carbon sor-bents, Theor. Exp. Chem. 25 (1989) 666–670.

27] I. Langmuir, The adsorption of gases on plane surfaces of glass, mica and plat-

Chem. Soc. 61 (1939) 2228–2230.29] Y.S. Ho, J.C.Y. Ng, G. McKay, Kinetics of pollutant sorption by biosorbents:

review, Sep. Purif. Rev. 29 (2000) 189–232.