ww.sciencedirect.com

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 4

Available online at w

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

Superparamagnetic magnesium ferritenanoadsorbent for effective arsenic (III, V)removal and easy magnetic separation

Wenshu Tang a, Yu Su a, Qi Li a,*, Shian Gao a, Jian Ku Shang a,b

aEnvironment Functional Materials Division, Shenyang National Laboratory for Materials Science, Institute of

Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang, Liaoning Province 110016, PR ChinabDepartment of Materials Science and Engineering, University of Illinois at Urbana-Champaign, Urbana,

IL 61801, USA

a r t i c l e i n f o

Article history:

Received 1 February 2013

Received in revised form

14 April 2013

Accepted 15 April 2013

Available online 24 April 2013

Keywords:

Arsenite/arsenate adsorption

Magnesium ferrite nanoadsorbent

Superparamagnetic

Magnetic separation

* Corresponding author. Tel.: þ86 24 8397802E-mail addresses: qili@imr.ac.cn, qiliuiuc

0043-1354/$ e see front matter ª 2013 Elsevhttp://dx.doi.org/10.1016/j.watres.2013.04.023

a b s t r a c t

By doping a proper amount of Mg2þ (w10%) into a-Fe2O3 during a solvent thermal process,

ultrafine magnesium ferrite (Mg0.27Fe2.50O4) nanocrystallites were successfully synthesized

with the assistance of in situ self-formed NaCl “cage” to confine their crystal growth. Their

ultrafine size (average size of w3.7 nm) and relatively low Mg-content conferred on them a

superparamagnetic behavior with a high saturation magnetization (32.9 emu/g). The ul-

trafine Mg0.27Fe2.50O4 nanoadsorbent had a high specific surface area of w438.2 m2/g, and

demonstrated a superior arsenic removal performance on both As(III) and As(V) at near

neutral pH condition. Its adsorption capacities on As(III) and As(V) were found to be no less

than 127.4 mg/g and 83.2 mg/g, respectively. Its arsenic adsorption mechanism was found

to follow the inner-sphere complex mechanism, and abundant hydroxyl groups on its

surface played the major role in its superior arsenic adsorption performance. It could be

easily separated from treated water bodies with magnetic separation, and could be easily

regenerated and reused while maintaining a high arsenic removal efficiency. This novel

superparamagnetic magnesium ferrite nanoadsorbent may offer a simple single step

adsorption treatment option to remove arsenic contamination from water without the

pre-/post-treatment requirement for current industrial practice.

ª 2013 Elsevier Ltd. All rights reserved.

1. Introduction and Suzuki, 2002; Brown and Ross, 2002). In order to mini-

Arsenic, a relatively scarce but ubiquitous element, is of

serious health concern due to its toxicity and carcinoge-

nicity. Long-term exposure to arsenic contaminated water

could cause skin, lung, bladder, and kidney cancers as well

as pigmentation changes, skin thickening (hyperkeratosis),

neurological disorders, muscular weakness, loss of appetite,

nausea, cardiovascular and cerebrovascular disease, dia-

betes mellitus, and adverse reproductive outcomes (Mandal

8; fax: þ86 24 23971215.@gmail.com (Q. Li).ier Ltd. All rights reserve

mize its health risk, the World Health Organization (WHO)

set a new guideline limit of 0.01 mg/L in drinking water to

replace the previous 0.05 mg/L limit in 1996 for arsenic

(WHO, 1996), and this new guideline limit has been adopted

by many countries. For example, the European Union

required the compliance with this new guideline limit by

2007 (De Zuane, 2007), and this new guideline limit has also

been effective in the United States since January 2006

(USEPA, 2002).

d.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 4 3625

Because of its simplicity, potential for regeneration, and

sludge free operation, adsorption is considered to be one of

the most promising technologies for arsenic removal (Jang

et al., 2006). In recent years, synthesized metal oxide nano-

adsorbents, such as aluminum, iron, titanium, zirconium and

cupric based oxides, have been extensively studied for arsenic

adsorption, which demonstrated superior performance

because of their large surface areas from their nano-size and

preferred surface properties (Cao et al., 2007; Cui et al., 2012;

Dutta et al., 2004; Kim et al., 2004; Liu et al., 2010; Mohan and

Pittman, 2007; Pena et al., 2005, 2006). To implement the

stricter guideline limit of 0.01 mg/L arsenic in drinking

water, the development of more effective and cost-friendly

nanoadsorbents for arsenic removal from drinking water is

urgently needed. Furthermore, the separation of current

nanoadsorbents from treated water bodies still remains as a

challenge to their potential application in real water treat-

ment practice. The inefficient separation could cause their

dispersion into the aqueous environment, resulting in the

operation cost increase and potential damages to both natural

organisms and the environment.

Iron oxide-basedmaterials had been extensively studied as

the arsenic adsorbents because of their low costs, high sta-

bility, environmental friendliness, and strong affinity for

arsenic species (Gimenez et al., 2007). In our recent work, we

developed a process to synthesize ultrafine a-Fe2O3 nano-

crystallites of several nanometers by a room temperature

reaction followed with a solvent thermal process at low

temperature of 150 �C, which demonstrated a good adsorption

performance on both As(III) and As(V) species in water (Tang

et al., 2011a,b). However, their ultrafine size caused the diffi-

cult separation from treated water. It had been demonstrated

that magnetic separation could be a more efficient and se-

lective method to separate nanomaterials from aqueous

environment than conventional approaches of centrifugation

or filtration (Raven et al., 1998).

In this work, a superparamagnetic ultrafine magnesium

ferrite (Mg0.27Fe2.50O4) nanoadsorbent was created by doping

Mg2þ into ultrafine a-Fe2O3 nanocrystallite during the hydro-

thermal process. Similar to that demonstrated in Al-doped

iron oxide arsenic adsorbent in our previous work (Li et al.,

2011), the specific surface area of Mg0.27Fe2.50O4 largely

increased due to the Mg-doping, subsequently enhancing its

arsenic adsorption performance on both As(III) and As(V)

species compared with the ultrafine a-Fe2O3 nanoadsorbent

we developed before. Furthermore, a proper amount of w10%

Mg-doping into iron oxide crystal lattice caused a crystal

structure change from the rch a-Fe2O3 phase to a single

magnesium ferrite phase with cubic spinel structure, which

endowed the ultrafine Mg0.27Fe2.50O4 nanoadsorbent a super-

paramagnetic behavior with a high saturation magnetization.

Thus, no magnetic attraction existed when there was no

external magnetic field applied during the water treatment,

which is beneficial to its better dispersion and the subsequent

better contact efficiency with arsenic species in water. After

the water treatment, however, the external magnetic field

applied could induce its easy magnetic separation from

treated water bodies. Then, the ultrafine Mg0.27Fe2.50O4 nano-

adsorbent could be easily recovered by NaOH washing and

reused for arsenic removal. To our best knowledge, no report

was available in literature on the superior arsenic removal

performance andmechanism study of the superparamagnetic

magnesium ferrite nanoadsorbent. With further develop-

ment, this novel superparamagnetic magnesium ferrite

nanoadsorbent may offer a simple single step adsorption

treatment option to remove arsenic contamination from

water without the pre-/post-treatment requirement for cur-

rent industrial practice.

2. Experimental

2.1. Chemicals and materials

All the chemicals used are of analytical reagent grade.

Anhydrous ferric chloride (FeCl3, 99.0 wt%, Chemical Reagent

Co., Ltd, Beijing, P.R. China) and anhydrous magnesium

chloride (MgCl2, 99.0 wt%, Shenyang Guo Yao Technology,

Shenyang, P.R. China) were used as the raw materials. So-

dium hydrate (NaOH, 98 wt%, Tianjin Damao Chemical Re-

agents Development Center, Tianjin, P.R. China) was used as

the precipitation agent, and anhydrous ethanol (C2H5OH,

�99.0%, Yili Chemical Reagent Co., Ltd, Beijing, P.R. China)

was used as the solvent. Concentrated hydrochloric acid

(HCl, 32e38%, Tianda Chemical Reagents Factory, Tianjin,

P.R. China) was used to stabilize the arsenic species after

water treatment. Sodium chloride (NaCl, 99 wt%, Tianjin

Damao Chemical Reagents Development Center, Tianjin, P.

R. China) was used as supporting electrolyte in the electro-

phoretic mobility (EM) study.

2.2. Nanoadsorbent synthesis

The typical synthesis process of the ultrafine magnesium

ferrite nanoadsorbent included a room-temperature reaction

followed with a solvent thermal treatment, similar as that of

ultrafine a-Fe2O3 nanocrystallites we reported previously

(Tang et al., 2011a,b). First, a mixture solution of 0.09 M FeCl3and 0.01 MMgCl2 was obtained by dissolving proper amounts

of FeCl3 and MgCl2 into 70 mL ethanol. Then, 10 mL NaOH

ethanol solution (2.03 M) was added into themixture solution

to obtain the fixed molar ratio of NaOH:FeCl3 of 3:1 and

NaOH:MgCl2 of 2:1 at room temperature, and the reaction

lasted for 1 h under ultrasonic oscillation. Yellow precipitates

were produced during the room temperature reaction, which

were composed of amorphous (Fe, Mg)x(OH)y nanoparticles

confined in NaCl “cage” in situ formed during the reaction.

Next, the suspension was sealed in a Teflon-lined autoclave,

placed in an oven pre-heated at 150 �C for 2 h, and then

cooled to room temperature naturally. During the solvent

thermal treatment, the thermal decomposition of (Fe,

Mg)x(OH)y occurred, and magnesium ferrite nanocrystallites

were produced. During this process, the NaCl “cage” still

existed and could confine the growth of magnesium ferrite

nanocrystallites. Then, the obtained precipitation was

washed with distilled water under ultrasonic oscillation for

three times to remove NaCl. After being dried at 80 �C for 12 h,

the ultrafine magnesium ferrite nanoadsorbent was

obtained.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 43626

2.3. Nanoadsorbent characterization

A D/MAX-2004 X-ray powder diffractometer (Rigaku Corpora-

tion, Tokyo, Japan) was used to analyze the sample’s crystal

structure.Transmissionelectronmicroscopy (TEM)wasusedfor

the morphology observation of nanocrystallites on a JEM 2100

transmission electron microscope (JEOL Corporation, Tokyo,

Japan) operated at 200 kV. The surface area and pore volume of

the sample were measured by N2 adsorptionedesorption

isotherm with an Autosorb-1 Series Surface Area and Pore Size

Analyzers (Quantachrome Instruments, Boynton Beach, FL,

U.S.A.). Prior to the experiment, the sample was dehydrated at

130 �C for 12 h. The relative pressure (P/P0) during 0.1018e0.3056

was used to determine the BET surface area. The pore-size dis-

tribution (PSD) was calculated using the desorption branches of

the N2 adsorption isotherm and the BarretteJoynereHalenda

(BJH) formula (Cheng et al., 1997). The average size and size

polydispersity of thesenanoparticlesdispersed inDIwaterwere

determined by dynamic light scattering with a Malvern Nano

ZS90 Zetasizer (Malvern Instruments Ltd., Malvern, Worcester-

shire, UK). The instrument was calibrated with standard latex

nanoparticles (Malvern Instruments Ltd., Malvern, Worcester-

shire, UK). The FTIR spectra of sampleswere carriedout at room

temperature in transmission mode by Fourier transform

infrared spectroscopy (FTIR, Bruker TENSOR 27, MCT detector)

with the resolution of 4 cm�1. The element composition of the

sample was determined by an inductive coupled plasmamass-

spectrometer (Perkin Elmer-SCIEX ELAN DRCe ICPeMS, Nor-

walk, U.S.A.). Zeta-potential of samples at different pHs was

measured by electrophoretic spectroscopy (JS84H, Shanghai

Zhongchen Digital Instrument Co., Ltd., Shanghai, P.R. China).

The zeta-potential was determined by mixing 100 mg samples

with 200 mL 0.01 M KNO3 solution with pH values adjusted be-

tween1and10byadding0.1MHNO3or0.1MKOH.Themagnetic

properties of samples at room temperaturewere examinedona

Quantum Design MPMS-XL superconducting quantum inter-

ference device (SQUID, Quantum Design, Inc., San Diego, CA,

U.S.A.). The semi-quantitative surface chemical composition

data of samples were examined by X-ray photoelectron spec-

troscopy (XPS) using an ESCALAB250 X-ray photoelectron

spectrometer (Thermo Fisher Scientific Inc., Waltham, MA,

U.S.A.) with an Al K anode (1486.6 eV photon energy, 0.05 eV

photon energy resolution, 300W).

2.4. Batch arsenic adsorption experiments

All the arsenic adsorption experiments were carried out at

w25 �C. During the arsenic removal experiment, the arsenic

solutions were stirred mechanically at 300 rpm by an electric

stirrer (JJ-1, Shanghai Pudong Physical Optical Instrument Fac-

tory Shanghai, P.R. China) to disperse the adsorbent to ensure a

good contact with arsenic contaminations. After recovering the

adsorbent magnetically, one drop of concentrated HCl was

added into the clear solution to avoid the potential oxidation of

As(III) to As(V). As (III) and As (V) concentrations of the aqueous

solutions were determined by an atomic fluorescence spec-

trophotometer (AFS-9800, Beijing Ke Chuang Hai Guang In-

strument Inc., Beijing, P.R. China) with the valence analysis

function. Detailed information on experiment conditions could

be found in the Supplementary Data.

3. Results and discussion

3.1. Crystal structure, composition, and morphology

Fig. 1a shows the XRD pattern of the nanoadsorbent sample.

A crystallized magnesium ferrite phase with cubic spinel

structure was observed, which was very close to the cubic

spinel structure of MgFe2O4 (JCPDS NO. 36-0398) (Candeia

et al., 2006). The bulk composition of the nanoadsorbent

sample was obtained by ICPeMS, and the sample’s chemical

formula was determined as Mg0.27Fe2.50O4, identical to its

surface composition determined by XPS analysis. Thus, the

nanoadsorbent should have a single magnesium ferrite

phase and could not contain the mixture of crystallized iron

oxides (like g-Fe2O3 or Fe3O4) with amorphous Mg(OH)2/

MgO, although g-Fe2O3 or Fe3O4 had similar XRD diffraction

patterns and magnetic behaviors as that of magnesium

ferrite. The Mg/(Fe þ Mg) molar percentage of Mg0.27Fe2.50O4

was determined at w9.7%, just slightly lower than the

starting composition of 10% for raw materials. The insert

image in Fig. 1a shows the diffraction peak shift of

Mg0.27Fe2.50O4 toward a slightly higher diffraction angle,

compared with MgFe2O4. More diffraction peak position

shift data could be found in Table S1 in the Supplementary

Data. This diffraction peak position shift could be attributed

to the different Mg-contents between Mg0.27Fe2.50O4 and

MgFe2O4. Mg2þ has a larger radius of 0.71 �A, compared with

that of Fe3þ (0.63 �A) (Nakagomi et al., 2009). The smaller

amount of Mg-content in Mg0.27Fe2.50O4 resulted in the

decrease of its lattice spacing, compared with that of

MgFe2O4. According to the Bragg equation as demonstrated

in Eq. (1):

2d sin q ¼ l (1)

where d is the lattice spacing, l is the average wavelength of

the X-ray radiation, and q is the diffracting angle. Thus, the

smaller lattice spacing of Mg0.27Fe2.50O4 could cause the in-

crease of 2q, compared with that of MgFe2O4.

Fig. 1b shows the TEM observation of the Mg0.27Fe2.50O4

nanoadsorbent. It is clear that the Mg0.27Fe2.50O4 nano-

adsorbent consisted of ultrafine nanocrystallites. The insert

image of Fig. 1b demonstrates that these ultrafine nano-

crystallites had a narrow size distribution as determined by

the professional image processing and analysis software of

Image-Pro Plus 5.0 on over 10 TEM images by the quantitative

statistical method. By Gaussian fitting, their average size was

determined at w3.7 nm, which was smaller than that of ul-

trafine a-Fe2O3 nanocrystallites without Mg-doping (w4.8 nm)

(Tang et al., 2011a,b). Similar to the synthesis of ultrafine

a-Fe2O3 nanocrystallites, the in situ self-formed NaCl “cage”

was effective in confining the growth of Mg0.27Fe2.50O4 nano-

crystallites. Fig. 1c shows the high resolution TEM (HRTEM)

image of Mg0.27Fe2.50O4 nanocrystallites. The insert SAD

pattern demonstrated that these nanocrystallites had a high

degree of crystallinity, which was in agreement with XRD

analysis result. For example, a set of lattice planes with the

d-spacing at w0.2957 nm could be easily identified on one

nanocrystallite, which corresponded to the (220) plane of

magnesium ferrite.

Fig. 1 e (a) X-ray diffraction pattern, (b) TEM image, and (c) HRTEM image of the Mg0.27Fe2.50O4 nanoadsorbent.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 4 3627

3.2. Surface properties

The BET specific surface area of ultrafine Mg0.27Fe2.50O4

nanocrystallites was determined at w438.2 m2/g, about 2.7

times as that of ultrafine a-Fe2O3 nanocrystallites (w162 m2/g)

synthesized with a similar process (Tang et al., 2011a,b).

Compared with ultrafine a-Fe2O3 nanocrystallites (w4.8 nm),

their average particle size (w3.7 nm) only decreased w23%.

Thus, the particle size decrease was not the only reason

for such a large BET specific surface area increase of

Mg0.27Fe2.50O4 nanocrystallites. The substitution of heavy Fe3þ

by lighter Mg2þ could induce internal and surface defects/

pores, which could then further increase the surface area.

Pore size distribution analysis demonstrated that most pores

were mesoporous. The average pore size was w3.89 nm,

which should reflect the inter-particle porosity. The pore

volume was determined to be w0.648 cm3/g. Thus, these

Mg0.27Fe2.50O4 nanocrystallites had a relatively larger surface

area and pore volume, which were beneficial for their arsenic

adsorption capability. Dynamic light scattering results

demonstrated that these nanoparticles formed aggregates in

water. The average size of these aggregates was determined at

w630 nm and their size rangewas fromw450 nm tow820 nm.

Fig. 2a shows the FTIR spectrum of ultrafine Mg0.27Fe2.50O4

nanocrystallites. Compared with that of a-Fe2O3 nano-

crystallites, two new absorption bands at w593 cm�1 and

442 cm�1 occurred in the FTIR spectrum of Mg0.27Fe2.50O4

nanocrystallites, which corresponded to the vibration of

tetrahedral and octahedral complexes, respectively. These

two absorption bands were the indication of the formation of

magnesium ferrite with cubic spinel structure (Pradeep et al.,

2008). Both FTIR spectra had evident HOH stretching

(w3398 cm�1) and bending (w1618 cm�1) vibrations of

physically adsorbed H2O, and the deformation (w1387 cm�1)

and bending (1047 cm�1) vibrations of hydroxyl groups on

metal oxides (Sun et al., 2009; Keiser et al., 1982). The signal

intensities of these vibrations of Mg0.27Fe2.50O4 nano-

crystallites were much stronger than that of a-Fe2O3 nano-

crystallites (see the transmittance data of these FTIR

adsorption bands in Table S2 in the Supplementary Data),

which indicated that more hydroxyl groups existed on their

surface. Fig. 2b compares the zeta potential of ultrafine

Mg0.27Fe2.50O4 nanocrystallites with that of ultrafine a-Fe2O3

nanocrystallites. The isoelectric point (IEP) of Mg0.27Fe2.50O4

nanocrystallites was w pH 5.2, lower than that of a-Fe2O3

nanocrystallites (wpH 6.1). Ultrafine Mg0.27Fe2.50O4 nano-

crystallites were more negatively charged near neutral pH

than a-Fe2O3 nanocrystallites, which indicated that more

surface hydroxyl groups existed on their surface and was in

agreement with the FTIR comparison result. Thus, the larger

specific surface area and the existence of more surface hy-

droxyl groups on ultrafine Mg0.27Fe2.50O4 nanocrystallites

indicated that they could possess an even better arsenic

removal effect by adsorption than a-Fe2O3 nanocrystallites.

3.3. Magnetic properties of magnesium ferritenanocrystallites

The large crystal structure change from the rhomb-centered

hexagonal (rch) a-Fe2O3 phase to the cubic spinel magnesium

ferrite structure caused a dramatic change on their magnetic

properties. Fig. 3 compares the magnetic field-dependent be-

haviors of ultrafineMg0.27Fe2.50O4 nanocrystallites and ultrafine

a-Fe2O3 nanocrystallites. Because of their ultrafine particle size

(w3.7 nm), ultrafine Mg0.27Fe2.50O4 nanocrystallites demon-

strated the typical superparamagnetic behavior with zero

Fig. 2 e (a) IR spectra and (b) zeta potential curves of ultrafine Mg0.27Fe2.50O4 and a-Fe2O3 nanocrystallites.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 43628

remanence and zero coercivity (Yang et al., 2009; Qin et al.,

2009; Skumryev et al., 2003), while ultrafine a-Fe2O3 nano-

crystallites had a weak ferromagnetic behavior with a narrow

magnetic hysteresis loop. The saturation magnetization, Ms,

could be obtained by extrapolating the graph of M vs. 1/H to

1/H / 0 (for H > 10 kOe). Ultrafine Mg0.27Fe2.50O4 nano-

crystallites had a highMs value ofw32.9 emu/g, far higher than

that of ultrafine a-Fe2O3 nanocrystallites (w4.5 emu/g). The

superparamagnetic behavior of ultrafine Mg0.27Fe2.50O4 nano-

crystallites could enhance their dispersion in arsenic solution

because no magnetic attraction existed when there was no

external magnetic field, while their high saturation magneti-

zation could make their magnetic separation feasible after the

water treatment.

3.4. Adsorption kinetic studies on As(III) and As(V)adsorption by magnesium ferrite nanocrystallites

Fig. 4a and b show the kinetics of As(III) and As(V) adsorption

onto the ultrafine Mg0.27Fe2.50O4 nanoadsorbent at different

sample loadings in the lab-prepared water samples at neutral

condition (pHw 7.0), respectively.With the increase of sample

loading amount, the equilibrium arsenic concentration in the

treated water samples gradually decreased. With just 0.01 g/L

sample loading, the equilibrium As(III) concentration dropped

from the initial 0.097 mg/L to 0.003 mg/L after the treatment

Fig. 3 e Magnetic field-dependent behaviors of ultrafine

Mg0.27Fe2.50O4 and a-Fe2O3 nanocrystallites.

(Fig. 4a), far below the USEPA limit for arsenic in drinking

water (0.01 mg/L). Similarly, the ultrafine Mg0.27Fe2.50O4

nanoadsorbent demonstrated a strong adsorption effect on

As(V). With just 0.01 g/L sample loading, the equilibrium As(V)

concentration dropped from the initial 0.101 mg/L to zero

(Fig. 4b), representing a 100% removal. Thus, a single step

As(III)/As(V) removal process is feasible by the ultrafine

Mg0.27Fe2.50O4 nanoadsorbent without pre-treatment (oxida-

tion/pH adjustment) and post-treatment pH adjustment.

The adsorption kinetic experimental results could be best

fitted into a pseudo-second-order rate kinetic model (Ho and

McKay, 1999), which had been widely used to describe metal

ion adsorption on different adsorbents (Ho and McKay, 1999;

Martinez et al., 2006). The integrated pseudo-second-order

rate expression could be described by Eq. (2):

t=qt ¼ t=qe þ 1=�Kadq

2e

�(2)

where qe and qt are the amount (mg/g) of arsenic adsorbed at

equilibrium and at time t, respectively, and Kad is the rate

constant of adsorption (g/(mg min)). The As(III) and As(V)

adsorption experimental data fitting results were shown in

Figure S1a and S1b of Supplementary Data, respectively, and

the kinetics parameters obtained in fitting the experimental

data were summarized in Table S3 of Supplementary Data.

The closeness of the square of the correlation coefficients r (r2)

to 1 indicated that the pseudo-second-order rate kinetic

model fitted the experimental data accurately. The rate con-

stant of adsorption (Kad) increased steadily with the increase

of the adsorbent loading for both arsenic species. Kad for the

adsorption of As(V) was higher than that for As(III) under the

similar experimental conditions, indicating that the ultrafine

Mg0.27Fe2.50O4 nanoadsorbent had a faster removal effect on

As(V) than on As(III). Similar result was observed for the

adsorption of As(III) and As(V) onto ferrihydrite surface (Raven

et al., 1998). The accurate fitting of the kinetic experimental

data to the pseudo-second-order rate kinetic model indicated

that the rate-limiting step may be a chemical sorption

involving valence forces through sharing or exchange of

electrons between sorbent and sorbate (Skumryev et al., 2003).

3.5. Equilibrium isotherm studies on As(III) and As(V)adsorption by magnesium ferrite nanocrystallites

The arsenic adsorption capacity of the ultrafine Mg0.27Fe2.50O4

nanoadsorbent at near neutral condition (pH w 7.0) was

Fig. 4 e Arsenic adsorption kinetics in lab-prepared water samples on the ultrafine Mg0.27Fe2.50O4 nanoadsorbent: (a) with

initial As(III) concentration of w0.097 mg/L, (b) with initial As(V) concentration of w0.101 mg/L.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 4 3629

investigated by the equilibrium adsorption isotherm study.

Fig. 5a and b demonstrate the equilibrium adsorption isotherm

curves for As(III) and As(V), respectively. As demonstrated in

Fig. 5a, the As(III) adsorption data could be best fitted with the

Freundlich isotherm as given in Eq. (3):

qe ¼ KFCe1=n (3)

where qe is the amount (mg/g) of As(III) adsorbed at equilib-

rium, Ce is the equilibrium As(III) concentration (mg/L) in

water samples, and KF and n are the Freundlich constants of

adsorption. The parameters obtained in fitting the experi-

mental data were summarized in Table S4 of Supplementary

Data. The adsorption capacity of the ultrafine Mg0.27Fe2.50O4

nanoadsorbent for As(III) at the near neutral pH environment

should be larger than 127 mg/g, far higher than that of ultra-

fine a-Fe2O3 nanoparticles (w95 mg/g) (Tang et al., 2011a,b).

The As(V) adsorption data could be best fitted with the Lang-

muir isotherm as given in Eq. (4):

qe ¼ qmKLCe=ð1=KLCeÞ (4)

where qe is the amount (mg/g) of As(V) adsorbed at equilib-

rium, qm is the maximum As(V) adsorption capability amount

(mg/g), Ce is the equilibrium As(V) concentration (mg/L) in

water samples, and KL is the Langmuir constant of adsorption.

The parameters obtained in fitting the experimental datawere

also shown in Table S4 of Supplementary Data. Its maximum

Fig. 5 e The equilibrium arsenic adsorption isotherms on ultra

As(V) adsorption capability was determined at w83 mg/g,

nearly twice as high as that of ultrafine a-Fe2O3 nanoparticles

(Tang et al., 2011a,b).

Because of the very low arsenic guideline limit of 0.01 mg/L

in drinking water, the amount of arsenic that an adsorbent

could adsorb at low equilibrium concentration is more

important than its maximum adsorption capability to esti-

mate its performance for arsenic removal in drinking water.

The inset images in Fig. 5a and b demonstrate the equilibrium

As(III) and As(V) adsorption isotherm curves at near neutral

condition for low equilibrium arsenic concentrations,

respectively. Under such conditions, the amounts of As(III)

and As(V) adsorbed at equilibrium increased with the equi-

librium arsenic concentration increase at a linear relation-

ship, as described by Eq. (5):

qe ¼ KCe þ b (5)

where qe is the amount (mg/g) of arsenic adsorbed at equilib-

rium, Ce is the equilibrium arsenic concentration (mg/L) in

water samples, K (L/g) and b are the adsorption constants. At an

equilibrium arsenic concentration of 0.01 mg/L (the USEPA

standard for drinking water), the amounts of adsorbed As(III)

and As(V) were around 10.1 mg/g and 11.8 mg/g, respectively,

nearly 2.7 and 2.0 times as high as that of ultrafine a-Fe2O3

nanoparticles (Tang et al., 2011a,b). From the above analysis, it

is clear that the amount of As(V) adsorbed at equilibrium was

fine Mg0.27Fe2.50O4 nanoadsorbent: (a) As(III) and (b) As(V).

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 43630

slightly more than the amount of As(III) adsorbed at equilib-

rium by the ultrafine Mg0.27Fe2.50O4 nanoadsorbent under low

arsenic equilibrium concentration, while it was far less when

the equilibrium arsenic concentration was high. This observa-

tion could be attributed to the different surface charge condi-

tions of As(III) and As(V) species under the neutral pH

environment. Similar phenomena had been observed and

explained in details in our previous work on various arsenic

adsorbents (Cui et al., 2012; Li et al., 2011; Tang et al., 2011a,b).

The equilibrium isotherm adsorption studies demonstrated

that the ultrafine Mg0.27Fe2.50O4 nanoadsorbent had a superior

arsenic adsorption performance than ultrafine a-Fe2O3 nano-

particles we previously reported. Its arsenic adsorption capac-

ity wasmuch higher than that of various traditional adsorbents

(Jeong et al., 2007; Balaji et al., 2005; Dutta et al., 2004).

3.6. Adsorption mechanism studies

The arsenic adsorption mechanism on the ultrafine

Mg0.27Fe2.50O4 nanoadsorbent was investigated by both

macroscopic and microscopic techniques. Fig. 6a and b show

the removal efficiencies of As(III) and As(V) by the ultrafine

Mg0.27Fe2.50O4 nanoadsorbent with different ionic strengths,

respectively. The ionic strength in the solution was adjusted

by the addition of different amounts of NaCl. For both As(III)

and As(V), their removal efficiencies showed no obvious

change with the increase of the ionic strength from pH 1 to pH

13. As suggested by Goldberg and Johnston (2001), the

adsorption of arsenic species will decrease with the increase

of ionic strength if arsenic species form outer-sphere surface

complexes, while the adsorption of arsenic species will not

Fig. 6 e (a) and (b) Ionic strength effects on the arsenic removal

respectively. (c) Zeta-potential curves and (d) FTIR spectra of the

adsorption, respectively.

change or increase with the increase of ionic strength

if arsenic species form inner-sphere surface complexes.

Thus, both As(III) and As(V) adsorptions on the ultrafine

Mg0.27Fe2.50O4 nanoadsorbent followed the inner-sphere

complex mechanism.

The electrophoretic mobility measurement was also used

to examine the arsenic adsorption mechanism. Fig. 6c shows

the zeta potentials of Mg0.27Fe2.50O4, Mg0.27Fe2.50O4 after As(III)

adsorption (1 mg/L), and Mg0.27Fe2.50O4 after As(V) adsorption

(1 mg/L), respectively. The IEP of the ultrafine Mg0.27Fe2.50O4

nanoadsorbent decreased from w pH 5.2 to wpH 4.0 after

As(III) adsorption and to wpH 3.0 after As(V) adsorption. The

IEP of a metal oxide is determined by the protonation and

deprotonation of surface hydroxyl groups. It had been re-

ported that the formation of outer-sphere surface complexes

could not shift the IEP because there is no chemical reaction

between the adsorbate and the adsorbent surface that could

change the surface charge (Frimmel, 1993; Pena et al., 2005;

Sun et al., 2009). Therefore, the shift of IEP provided the evi-

dence of the formation of inner-sphere As(III)/As(V) anionic

charged surface complexes on the ultrafine Mg0.27Fe2.50O4

nanoadsorbent (Frimmel, 1993; Sun et al., 2009).

It had been reported that substitution of hydroxyl groups

(eOH) groups by arsenic species played the key role in their

adsorption mechanism (Goldberg and Johnston, 2001). So the

arsenic adsorption mechanism was further investigated with

the microscopic technique of FTIR spectroscopy. Fig. 6d

demonstrates the FTIR spectra of Mg0.27Fe2.50O4 after As(III)

and As(V) adsorption, respectively. Compared with that of

Mg0.27Fe2.50O4 without arsenic adsorption (see Fig. 2a), the

MeOH bending vibration peak (1047 cm�1) mostly

efficiency of As(III) and As(V) under different pHs,

ultrafine Mg0.27Fe2.50O4 nanoadsorbent after As(III) or As(V)

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 4 3631

disappeared, and the intensities of the MeOH deformation

vibration peak (1387 cm�1) and the two peaks from physically

adsorbed H2O at 1618 cm�1 and 3698 cm�1 decreased. This

observation indicated that the replacement of eOH and H2O

occurred at during the arsenic adsorption. For As(III) adsorp-

tion, a new peak at 762 cm�1 appeared which corresponded to

the stretching vibration of uncomplexed As(III)eO bond (Sun

et al., 2009). As demonstrated by Pena et al. (2006) on the

arsenic adsorption by TiO2, the uncomplexed As(III)eO bond

of dissolved arsenite species was at 650 cm�1. The red shift of

As(III)eO observed here was caused by the decrease of the

strength of the uncomplexed AseO bond because of the for-

mation of AseOeM bond (Frimmel, 1993; Sun et al., 2009),

indicating the formation of inner-sphere complexes on the

nanoadsorbent surface after As(III) adsorption. For As(V), a

similar result was observed. A newpeak at 823 cm�1 appeared,

which corresponded to n(AseOeM) (Frimmel, 1993). The for-

mation of AseOeM bond indicated that the As(V) adsorption

on the nanoadsorbent surface also followed the inner-sphere

complex mechanism.

XPS analysis had been used in literature to study surface

interactions between adsorbates and adsorbents in the

adsorption process (Zhang et al., 2003; Deliyanni et al., 2006;

Lim et al., 2009; Martinson and Reddy, 2009). The interactions

between arsenic species and the functional groups (eOH) on

the ultrafine Mg0.27Fe2.50O4 nanoadsorbent were analyzed by

examining their signal intensity change. Fig. 7aec demon-

strate the O 1s XPS peak scan over ultrafine Mg0.27Fe2.50O4

nanoadsorbent, and ultrafine Mg0.27Fe2.50O4 nanoadsorbent

after As(III) and As(V) adsorption, respectively. The O 1s XPS

peak spectrum could be best fitted by the combination of three

Fig. 7 e (a)e(c) The surface O 1s spectra of the ultrafine Mg0.27Fe

adsorption, respectively. (d) The surface O 1s spectra of ultrafin

peaks with the center position at 530.1 eV, 531.2 eV, and

532.5 eV, which could be assigned to metal oxide (MeO), hy-

droxyl bonded to metal (MeOH), and adsorbed H2O, respec-

tively (Deng and Ting, 2005; Goh et al., 2009; Mamindy-Pajany

et al., 2009). After As(III) and As(V) adsorption, the area ratio of

MeOH peak decreased sharply from 35.4% to 15.7% and 18.5%,

respectively, indicating the decrease of the amounts of hy-

droxyl group on the adsorbent surface during the arsenic

adsorption. This observation was in agreement with the FTIR

analysis result on the formationofAseOeMbond (Fig. 6d). Fig. 7d

shows the O 1s XPS peak scan over ultrafine a-Fe2O3 nano-

particles. ItsarearatioofMeOHpeakwasonly18.9%, far less than

that of ultrafine Mg0.27Fe2.50O4 nanoadsorbent. This observation

was in agreement with the FTIR analysis on the hydroxyl group

signal (Fig. 2a), and indicated that the ultrafine Mg0.27Fe2.50O4

nanoadsorbent should have superior arsenic adsorption

performance than the ultrafine a-Fe2O3 nanoparticles.

3.7. Arsenic removal from natural water samples

The ultrafine Mg0.27Fe2.50O4 nanoadsorbent demonstrated a

good arsenic removal performance in natural water samples

from Lake Yangzonghai in the Yunnan Province of China even

when large amounts of competing ions were present (see its

water quality data in Table S5 of Supplementary Data). A

much better arsenic removal performance than that of the

ultrafine a-Fe2O3 nanoparticles was observed for it (see

Fig. 8a). The initial As(III) and As(V) concentrations were found

at w0.044 mg/L and w0.071 mg/L, respectively, and the water

sample’s pH value was w7.0. With a low material loading at

0.02 g/L, about 98% arsenic contaminationwas removed by the

2.50O4 nanoadsorbent before and after As(III) or As(V)

e a-Fe2O3 nanoadsorbent.

Fig. 8 e (a) Arsenic removal percentages on natural water samples from Lake Yangzonghai by the ultrafine Mg0.27Fe2.50O4

nanoadsorbent, compared with that by the a-Fe2O3 nanoadsorbent (material loading at 0.02 g/L). (b) Images of ultrafine

Mg0.27Fe2.50O4 and a-Fe2O3 nanoadsorbents dispersed in water w/o external magnetic field (NbFeB: a kind of rare earth

permanent magnet purchased from Shenyang Guo Yao Technology, Shenyang, P.R. China). (c) The arsenic removal

percentage of the regenerated ultrafine Mg0.27Fe2.50O4 nanoadsorbent on natural water samples from Lake Yangzonghai for

5 times.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 43632

ultrafine Mg0.27Fe2.50O4 nanoadsorbent, while only 28%

arsenic contamination was removed by the ultrafine a-Fe2O3

nanoparticles. Thus, the ultrafine Mg0.27Fe2.50O4 nano-

adsorbent could effectively remove both As(III) and As(V) from

natural water samples without pre-treatment (oxidation/pH

adjustment) and post-treatment pH adjustment to meet the

USEPA standard for arsenic in drinking water.

3.8. Magnetic separation and adsorbent regeneration/reuse

As demonstrated in Fig. 8b, the ultrafine Mg0.27Fe2.50O4 nano-

adsorbent could be easily separated from treatedwater bodies

at the presence of external magnetic field within 5 min, while

the ultrafine a-Fe2O3 nanoparticles were still dispersed well in

water under the same magnetic field. Thus, magnetic sepa-

ration could be a feasible way to remove the ultrafine

Mg0.27Fe2.50O4 nanoadsorbent from treated water, which is

more efficient than conventional approaches of centrifugation

or filtration. After the separation from water, arsenic species

adsorbed on the ultrafine Mg0.27Fe2.50O4 nanoadsorbent could

be desorbed by washing with 2 M NaOH solution. Over 90%

arsenic species could be desorbed, and the regenerated ul-

trafine Mg0.27Fe2.50O4 nanoadsorbent could be reused for

arsenic removal. Fig. 8c demonstrates the arsenic removal

percentage of the regenerated ultrafine Mg0.27Fe2.50O4 nano-

adsorbent on natural water samples from Lake Yangzonghai

for 5 times. After being regenerated for four times, the re-

generated ultrafine Mg0.27Fe2.50O4 nanoadsorbent could still

keep w70e80% of its arsenic removal capability, which is

desirable for its potential applications in real practice to

reduce the operation cost.

4. Conclusions

In summary, we successfully synthesized a super-

paramagnetic ultrafine magnesium ferrite (Mg0.27Fe2.50O4)

nanoadsorbent in a simple room-temperature precipitation

reaction followed with a hydrothermal process at a low tem-

perature of 150 �C. The in situ self-formed NaCl “cage”

confined the growth of Mg0.27Fe2.50O4 nanocrystallites to an

ultrafine size of w3.7 nm, which endowed a super-

paramagnetic behavior to these Mg0.27Fe2.50O4 nano-

crystallites. The ultrafine Mg0.27Fe2.50O4 nanoadsorbent had a

large BET specific surface area of w438.2 m2/g, and demon-

strated a superior arsenic adsorption performance on both

As(III) and As(V). Its adsorption capacities on As(III) and As(V)

were found to be no less than 127.4 mg/g and 83.2 mg/g,

respectively. The amounts of adsorbed As(III) and As(V)

reached 10.1 mg/g and 11.8 mg/g, respectively, at an equilib-

rium arsenic concentration of just 0.01 mg/L (the USEPA

standard for drinking water). The arsenic adsorption on the

ultrafine Mg0.27Fe2.50O4 nanoadsorbent followed the inner-

sphere complex mechanism, and abundant surface hydroxyl

groups played themajor role in its superior arsenic adsorption

performance. It could effectively remove arsenic contamina-

tion in natural water samples with the presence of large

amounts of competing ions, could be readily separated by

external magnetic field, and could be regenerated and reused

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 4 3633

for arsenic removal. With further development, it may offer a

simple and effective single step treatment option to treat

arsenic contaminated water without the pre-/post-treatment

requirement for current industrial practice.

Acknowledgments

This study was supported by the National Natural Science

Foundation of China (Grant No. 51102246), the Knowledge

Innovation Programof Chinese Academy of Sciences (Grant No.

Y0N5711171), the Knowledge Innovation Programof Institute of

Metal Research, Chinese Academy of Sciences (Grant No.

Y0N5A111A1), and the Youth Innovation Promotion Associa-

tion, Chinese Academy of Sciences (Grant No. Y2N5711171).

Appendix A. Supplementary data

Supplementary data related to this article can be found, in the

onlineversion,athttp://dx.doi.org/10.1016/j.watres.2013.04.023.

r e f e r e n c e s

Balaji, T.T., Yokoyama, T., Matsunaga, H., 2005. Adsorption andremoval of As(V) and As(111) using Zr-loaded lysine diaceticacid chelating resin. Chemosphere 59 (8), 1169e1174.

Brown, K.G., Ross, G.L., 2002. Arsenic, drinking water, and health: aposition paper of the American Council on Science and Health.Regulatory Toxicology and Pharmacology 36 (2), 162e174.

Candeia, R.A., Souza, M.A.F., Bernardi, M.I.B., Maestrelli, S.C.,Santos, I.M.G., Souza, A.G., Longo, E., 2006. MgFe2O4 pigmentobtained at low temperature. Materials Research Bulletin 41(1), 183e190.

Cao, A.M., Monnell, J.D., Matranga, C., Wu, J.M., Cao, L.L., Gao, D.,2007. Hierarchical nanostructured copper oxide and itsapplication in arsenic removal. Journal of Physical ChemistryC 111 (50), 18624e18628.

Cheng, C.F., Zhou, W.Z., Park, D.H., Klinowski, J., Hargreaves, M.,Gladden, L.F., 1997. Controlling the channel diameter of themesoporous molecular sieve MCM-41. Journal of the ChemicalSociety e Faraday Transactions 93 (2), 359e363.

Cui, H., Li, Q., Gao, S.A., Shang, J.K., 2012. Strong adsorption ofarsenic species by amorphous zirconium oxide nanoparticles.Journal of Industrial and Engineering Chemistry 18 (4),1418e1427.

De Zuane, J., 2007. Appendix A-III: European Drinking WaterDirectives. Handbook of Drinking Water Quality. John Wiley &Sons, Inc., pp. 535e539.

Deliyanni, E.A., Nalbandian, L., Matis, K.A., 2006. Adsorptiveremoval of arsenites by a nanocrystalline hybrid surfactant-akaganeite sorbent. Journal of Colloid and Interface Science302 (2), 458e466.

Deng, S.B., Ting, Y.P., 2005. Polyethylenimine-modified fungalbiomass as a high-capacity biosorbent for Cr(VI) anions:sorption capacity and uptake mechanisms. EnvironmentalScience & Technology 39 (21), 8490e8496.

Dutta, P.K., Ray, A.K., Sharma, V.K., Millero, F.J., 2004. Adsorptionof arsenate and arsenite on titanium dioxide suspensions.Journal of Colloid and Interface Science 278 (2), 270e275.

Frimmel, F.H., 1993. Chemistry of the solidewater interface.Processes at the mineral-water and particleewater interface

in natural systems. VonW. Stumm.Wiley, Chichester, 1992. X,428 S., Broschur 32.50£ e ISBN 0-471-57672-7. AngewandteChemie 105 (5), 800.

Gimenez, J., Martinez, M., De Pablo, J., Rovira, M., Duro, L., 2007.Arsenic sorption onto natural hematite, magnetite, andgoethite. Journal of Hazardous Materials 141 (3), 575e580.

Goh, K.H., Lim, T.T., Dong, Z., 2009. Enhanced arsenic removal byhydrothermally treated nanocrystalline Mg/Al layered doublehydroxide with nitrate intercalation. Environmental Science &Technology 43 (7), 2537e2543.

Goldberg, S., Johnston, C.T., 2001. Mechanisms of arsenicadsorption on amorphous oxides evaluated usingmacroscopic measurements, vibrational spectroscopy, andsurface complexation modeling. Journal of Colloid andInterface Science 234 (1), 204e216.

Ho, Y.S., McKay, G., 1999. Pseudo-second order model for sorptionprocesses. Process Biochemistry 34 (5), 451e465.

Jang, M., Min, S.H., Kim, T.H., Park, J.K., 2006. Removal of arseniteand arsenate using hydrous ferric oxide incorporated intonaturally occurring porous diatomite. Environmental Science& Technology 40 (5), 1636e1643.

Jeong, Y., Fan, M., Singh, S., Chuang, C.L., Saha, B., vanLeeuwen, H., 2007. Evaluation of iron oxide and aluminumoxide as potential arsenic(V) adsorbents. ChemicalEngineering and Processing 46 (10), 1030e1039.

Keiser, J.T.,Brown,C.W.,Heidersbach,R.H.,1982.Theelectrochemicalreductionof rustfilmsonweathering steel surfaces. Journal of theElectrochemical Society 129 (12), 2686e2689.

Kim, Y.H., Kim, C.M., Choi, I.H., Rengaraj, S., Yi, J.H., 2004. Arsenicremoval using mesoporous alumina prepared via a templatingmethod. Environmental Science & Technology 38 (3), 924e931.

Li, R.H., Li, Q., Gao, S.A., Shang, J.K., 2011. Enhanced arseniteadsorption onto Litchi-like Al-doped iron oxides. Journal ofthe American Ceramic Society 94 (2), 584e591.

Lim, S.F., Zheng, Y.M., Chen, J.P., 2009. Organic arsenic adsorptiononto a magnetic sorbent. Langmuir 25 (9), 4973e4978.

Liu, Z., Zhang, F.S., Sasai, R., 2010. Arsenate removal from waterusing Fe3O4-loaded activated carbon prepared from wastebiomass. Chemical Engineering Journal 160 (1), 57e62.

Mamindy-Pajany, Y., Hurel, C., Marmier, N., Romeo, M., 2009.Arsenic adsorption onto hematite and goethite. ComptesRendus Chimie 12 (8), 876e881.

Mandal, B.K., Suzuki, K.T., 2002. Arsenic round the world: areview. Talanta 58 (1), 201e235.

Martinez, M., Miralles, N., Hidalgo, S., Fiol, N., Villaescusa, I.,Poch, J., 2006. Removal of lead(II) and cadmium(II) fromaqueous solutions using grape stalk waste. Journal ofHazardous Materials 133 (1e3), 203e211.

Martinson, C.A., Reddy, K.J., 2009. Adsorption of arsenic(III) andarsenic(V) by cupric oxide nanoparticles. Journal of Colloidand Interface Science 336 (2), 406e411.

Mohan, D., Pittman, C.U., 2007. Arsenic removal from water/wastewater using adsorbents e a critical review. Journal ofHazardous Materials 142 (1e2), 1e53.

Nakagomi, F., da Silva, S.W., Garg, V.K., Oliveira, A.C.,Morais, P.C., Franco, A., 2009. Influence of the Mg-content onthe cation distribution in cubic MgxFe(3�x)O(4) nanoparticles.Journal of Solid State Chemistry 182 (9), 2423e2429.

Pena, M.E., Meng, X.G., Korfiatis, G.P., Jing, C.Y., 2006. Adsorptionmechanism of arsenic on nanocrystalline titanium dioxide.Environmental Science & Technology 40 (4), 1257e1262.

Pena, M.E., Korfiatis, G.P., Patel, M., Lippincott, L., Meng, X.G.,2005. Adsorption of As(V) and As(III) by nanocrystallinetitanium dioxide. Water Research 39 (11), 2327e2337.

Pradeep, A., Priyadharsini, P., Chandrasekaran, G., 2008. Solegelroute of synthesis of nanoparticles of MgFe(2)O(4) and XRD,FTIR and VSM study. Journal of Magnetism and MagneticMaterials 320 (21), 2774e2779.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 3 6 2 4e3 6 3 43634

Qin, J., Asempah, I., Laurent, S., Fornara, A., Muller, R.N.,Muhammed, M., 2009. Injectable superparamagnetic ferrogelsfor controlled release of hydrophobic drugs. AdvancedMaterials 21 (13), 1354e1357.

Raven, K.P., Jain, A., Loeppert, R.H., 1998. Arsenite and arsenateadsorption on ferrihydrite: kinetics, equilibrium, and adsorptionenvelopes. Environmental Science&Technology 32 (3), 344e349.

Skumryev, V., Stoyanov, S., Zhang, Y., Hadjipanayis, G.,Givord, D., Nogues, J., 2003. Beating the superparamagneticlimit with exchange bias. Nature 423 (6942), 850e853.

Sun, X.F., Hu, C., Qu, J., 2009. Adsorption and removal of arseniteon ordered mesoporous Fe-modified ZrO(2). Desalination andWater Treatment 8 (1e3), 139e145.

Tang, W.S., Li, Q., Gao, S.A., Shang, J.K., 2011a. Arsenic (III, V)removal from aqueous solution by ultrafine alpha-Fe2O3

nanoparticles synthesized from solvent thermal method.Journal of Hazardous Materials 192 (1), 131e138.

Tang, W.S., Li, Q., Gao, S.A., Shang, J.K., 2011b. Ultrafine alpha-Fe2O3 nanoparticles grown in confinement of in situ self-formed “cage” and their superior adsorption performance onarsenic(III). Journal of Nanoparticle Research 13 (6),2641e2651.

USEPA, 2002. EPA re-establishes lower arsenic limit for drinkingwater: new standard fully in effect by 2006. CA: a CancerJournal for Clinicians 52 (1), 2e3.

WHO, 1996. Guidelines for drinking-water quality. Health Criteriaand Other Supporting Information, vol. 2.

Yang, X., Zhang, X., Ma, Y., Huang, Y., Wang, Y., Chen, Y., 2009.Superparamagnetic graphene oxideeFe3O4 nanoparticleshybrid for controlled targeted drug carriers. Journal ofMaterials Chemistry 19 (18), 2710e2714.

Zhang, Y., Yang, M., Huang, X., 2003. Arsenic(V) removal with aCe(IV)-doped iron oxide adsorbent. Chemosphere 51 (9),945e952.