724

effective arsenate removal would require pH values between 7 and 11, chabazite be-
tween 4 and 5, and clinoptilolite between 3 and 7. Arsenite removal by iron-modied
activated carbon would require pH values between 7 and 11, chabazite between 7 and
10, and clinoptilolite between 4 and 11. Increasing temperature improved adsorption
performance for activated carbon and the zeolites. Increasing ionic strength improved
performance of iron-treated activated carbon and zeolites.
Adsorbents; Arsenate; Arsenite; Activated carbon; Zeolites; Chabazite;
Clinoptilolite.
INTRODUCTION
Heightened awareness of arsenic toxicity and regulatory changes has prompted
considerable research efforts toward developing methods for arsenic removal
from drinking water.
[1]
Naturally occurring arsenic contaminates groundwa-
ter in many countries including Argentina, Australia, Chile, China, Hungary,
India, Mexico, Peru, Taiwan, Thailand, and the United States.
[2]
In Bangladesh,
millions of people are suffering fromarsenic-related diseases such as skin, lung,
bladder, and kidney cancer due to heavy dependence of rural inhabitants on
groundwater for drinking, and among them about 20,000 people die each year
according to the UN Development Program. An estimated 30 to 35 million
Bangladesh people are exposed to arsenic concentrations above 50 g l
1
(World
Health Organization [WHO] guideline value prior to 1993), and the number of
people exposed to more than 10 g l
1
(current WHOguideline value) is 46 to 57
million.
[3]
Within the United States, high groundwater arsenic concentrations
(greater than 10 g l
1
) have been documented in Maine, New Hampshire,
Michigan, Minnesota, Oklahoma, South Dakota, and Wisconsin.
[4,5]
Arsenic occurs in the environment in several oxidation states (3, 0, +3,
+5). In natural water, inorganic arsenic is mostly found as trivalent arsenite
or pentavalent arsenate. In surface water under oxidizing conditions, arsen-
ate predominates while in anoxic water under reducing conditions, arsenite
becomes stable. At near neutral pH, the predominant species are H
2
AsO

4
and
HAsO
2
4
for arsenate, and uncharged H
3
AsO
3
for arsenite. Arsenic toxicity is
dependent on its chemical form. Inorganic arsenic combines with sulfhydryl
groups in proteins and is accumulated in human bodies. Since arsenite has a
higher afnity for protein than arsenate, arsenite is more toxic. Arsenate can
be reduced to arsenite by the activity of glutathione and results in the same
toxicity. However, since not all of the arsenate can be converted to arsenite, the
toxicity of arsenate is less than arsenite.
Several techniques effectively lower arsenic concentrations inaqueous solu-
tions: coagulation/precipitation, reverse osmosis, ion exchange, and adsorption.
Coagulation and softening with metal ions such as aluminum and ferric salts
require use of large-scale facilities for implementing water treatment. Reverse
725
osmosis requires the use of membranes, which are expensive to maintain and
replace, and ion exchange uses costly resins. Coagulation, reverse osmosis, and
ion exchange require the treatment of reject stream for the ultimate disposal
of arsenic contaminants. Adsorption of arsenic in both oxidation states from
the aqueous phase to the solid phase onto sediments has been found to occur.
Properties that make an adsorbent effective in removing certain species from
water are high surface area and an afnity for a desired sorbate. Characteris-
tics of the sorbate such as molecular size, polarity, water solubility, and shape
also play an important role in how well it sorbs to a reactive medium. Arsenate
is well adsorbed onto the surface of minerals. The adsorption of arsenate by
iron oxides plays an important role in preventing widespread arsenic toxicity
in nature. Arsenite is a neutral molecule at near neutral pH and is therefore
less adsorbed on most mineral surfaces than arsenate. Thus, arsenite is gener-
ally more mobile than arsenate and may be more likely to contaminate water
supplies.
[3]
Adsorbent materials studied for arsenic removal include activated alu-
mina, yash, pyrite nes, manganese greensand,
[6]
amino-functionalized meso-
porous silicas,
[7]
aluminum loaded Shirasu-zeolite,
[8]
clinoptilolite, and other
zeolites.
[911]
Iron oxides have been widely used as sorbents to remove contam-
inants from wastewater. Removal has been attributed to ion exchange, specic
adsorption to surface hydroxyl groups, or coprecipitation.
[12]
Research using
iron as an adsorbent or adsorbent improvement material includes: Fe(ClO
4
)
2
coated activated carbon,
[13]
iron oxidecoated sand and ferrihydrite,
[1416]
zero-valent iron,
[1,1723]
iron-coated spent catalyst,
[24]
goethite,
[25,26]
ferric
chloride,
[27,28]
and iron oxides and oxyhydroxides.
[2931]
Arsenic removal ef-
ciencies greater than 95% have been achieved with zero-valent iron.
[20]
Factors for removal of inorganic arsenic species from aqueous solutions are
thought to be the afnity between arsenic and different iron species, amount
of iron, reaction time, adsorbent surface area, pH, and positive electrostatic
charge.
Activated carbon and zeolites are low-cost and have high sorption capaci-
ties due to their structural formation yielding high surface area per unit mass.
Porous materials have found a variety of applications including catalysis, l-
tration membranes, sensors, and gas storage materials.
[32]
The goal of this research was to simultaneously determine the differences in
arsenate and arsenite sorption capacities for iron-treated activated carbon and
zeolites. Two iron treatments were investigated as well as the effects of varied
pH, temperature, and ionic strength on adsorption effectiveness. A potential
benet of this study is the identication of a low-cost, arsenic selective sorbent
for the removal of arsenic species fromaqueous solution. This research supports
worldwide research efforts to obtain drinking water with arsenic levels below
10 g l
1
. This work can be used to select economical and easy-to-use point-of-
use systems for small communities and water utilities.
726
MATERIALS AND METHODS
Materials
The activated carbon (Table 1) used in this study is randomly stacked micro-
crystallines of graphite produced from coal. The activated carbon was obtained
from Calgon Carbon Corporation.
Zeolites are naturally occurring, porous, hydrated aluminosilicates with ex-
changeable cations such as sodium and potassium. Chabazite and clinoptilolite
(volcanic in origin, mined in Arizona, and obtained from GSA Resources, Inc.)
were studied due to their availability at a very low cost. Linde type A (5A;
sodium calcium aluminosilicate) and Faujasite (13X; sodium potassium alu-
minosilicate) are synthetic versions of naturally occurring zeolites with well-
documented behavior. 5A and 13X were obtained from Aldrich.
10.0 g of eachadsorbent were placed in125 ml Nalgene bottles with50.00 ml
of 0.5 M FeSO
4
solution prepared with N
2
purged distilled water (to remove
oxygen present in the water, thus preventing oxidation of ferrous sulfate to
ferric sulfate).
Table 1: Adsorbent characterization.
Pore Surface
diameter area
Adsorbent Mesh Description (

A) (m
2
g
1
) Si:Al
Activated Carbon (Calgon Carbon Corporation)
Filtrasorb (300) 8 30 Bituminous coal 22000 1200 N/A
Chabazite (GSA Resources, Inc.)
CABSORB-ZS500RW 8 20 Natural sodium
chabazite
(herschelite)
4.3 520 1.5 to 4.0
Clinoptilolite (GSA Resources, Inc.)
CABSORB-ZS403H 8 12 Natural clinoptilolite;
hydrous sodium
aluminosilicate
(sodium and
potassium in
exchangeable
cation positions)
4.0 40 4.0 to 5.5
Molecular Sieves (Aldrich)
5A 12 Sodium calcium
aluminosilicate
5.4 520 1.0
13X 12 Sodium potassium
aluminosilicate
7.4 730 2.2 to 2.6
727
10.0 g of eachadsorbent were placed in125 ml Nalgene bottles with25.00 ml
of 0.5 M Na
2
SO
3
and gently swirled for 30 s, then 25.00 ml of 0.5 M FeSO
4
solution prepared with N
2
purged distilled water were added.
Bottles were placed on a reciprocating shaker at 125 rpm for 48 h at
constant temperature (23.0 1

C). Following decanting of supernatant, solids

were washed with distilled water until there was no detection of greenish color
in the ltrate and no detection of sulfate, suction ltered, and dried in a vacuum
oven at 120

C for 24 h.
Adsorbent Study
0.1000 g of each type of treated adsorbent was placed in 50 ml Nalgene
bottles with 50.00 ml of 50 g l
1
arsenate solution (Na
2
HAsO
4
7H
2
O). Bot-
tles were placed on a reciprocating shaker at 125 rpm for 48 h at constant
temperature (23.0 1

C).
0.1000 g of each type of treated adsorbent was placed in 50 ml Nalgene
bottles with 50.00 ml of 50 g l
1
arsenite solution (NaAsO
2
) prepared with
N
2
purged distilled water and 3.0000 g l
1
NaHSO
3
(to maintainarsenic in its
reduced form by deoxygenating the water with sulte as the reducing agent).
Bottles were placed on a reciprocating shaker at 125 rpm for 24 h at constant
temperature (23.0 1

C).
Arsenic concentrations were analyzed by a Spectra AA 220 Graphite Fur-
nace Atomic Adsorption spectrometer (GFAAS) (Varian Australia Pty Ltd.,
Mulgrave, Australia) with a Varian Graphite Tube Analyzer (GTA) 110.
Calibration curves were completed before each sample run with 0, 5, 10, 25,
and 50 g l
1
solutions and %RSD < 5 for triplicate sample sets.
Isotherm Study
The isotherm study was performed using different weights of adsorbents
(0.0250, 0.0500, 0.0750, 0.1500, or 0.2500 g) to which a constant volume
(50.00 ml) of solution (50 g l
1
arsenate or arsenite) was applied in 50 ml
Nalgene bottles. Bottles were agitated at 125 rpm while allowed to achieve
equilibrium (48 h) at constant temperature (23.0 1

C).
Comparison with Langmuir (Eqs. [1][3]) and Freundlich (Eqs. [4][5])
isotherm models was made using the following equations:
= ( )/(1 + )
[33]
(1)
= (( ) ) (2)
728
where
= mg of adsorbate per g of adsorbent at equilibrium (equilibrium uptake)
= mg of solute adsorbed per g of adsorbent
= Langmuir constant (liter of adsorbent per mg of adsorbate)
= equilibrium concentration of adsorbate in solution (mg l
1
)
= adsorbate initial concentration (mg l
1
)
= volume of solution (l)
= amount of adsorbent used (mg)
/ = 1/( ) +(1/ ) (3)
=
1/
(4)
where
= mg of adsorbate per g of adsorbent at equilibrium
= Freundlich constant
= a Freundlich constant, which is always greater than one
= concentration of adsorbate in solution (mg l
1
)
ln = ln +(1/ )(ln ) (5)
Results of experimental data were compared to the linear formof the isotherms,
Eqs. (3) and (5), to determine which model most accurately described adsorption
by the adsorbent.
Qualitative Analysis
The interaction before and after exposure to arsenic of the adsorbents was
completed using dry adsorbent samples obtained as described above. Fisher
Scientic IR Grade KBr, P-227, Lot 991470A was used to pelletize adsor-
bents at a 1:10, adsorbent to KBr, ratio. Infrared spectra were measured using
PerkinElmer Model 1600, Fourier transmission infrared (FTIR) spectropho-
tometer (PerkinElmer, Norwalk, CT). X-ray powder diffraction (XRD) analysis
was performed on chabazite, clinoptilolite, 13X, and 5A to conrm the crystal
structure and mineral identity of the zeolites. XRD patterns were obtained on
a Siemens diffractometer equipped with a rotating anode and Cu-K radiation
( = 0.15418 nm).
Kinetic Study
The kinetic study was performed on the treated zeolites and activated car-
bon using 50.00 ml of solution with an initial arsenate or arsenite concentration
729
of 50 g l
1
and 0.1000 g of adsorbent in 50 ml Nalgene bottles. The kinetic
study was not performed with iron-treated 13X and 5A due to their arsenic ad-
sorptive performance and increased cost over the natural zeolites and activated
carbon. Bottles were agitated at 125 rpm at constant temperature (23.0 1

C).
Samples were withdrawn for analysis at periods of 1, 2, 5, 10, 15, 20, 30, 60,
and 90 min, and 2, 3, 6, 12, 24, and 48 h.
The integrated rate law models how reactant and product concentrations
vary with time. The linearized integrated rate law for a rst-order reaction
[Eq. [6]) is shown:
log[ ] = ( /2.303)

+log[ ]
[34]
(6)
where
[ ] = concentration of at any time
[ ] = original concentration of at some initial time
= rate constant
pH Study
Measurements for the effects of varied pH on adsorbent performance used
0.1000 g of each adsorbent (iron-treated zeolites and activated carbon) added to
50.00 ml of 50 g l
1
arsenate or arsenite solution. The varied pHsolutions were
prepared by adding dilute NaOH or HNO
3
dropwise to achieve pH values of 2.0
through 12.0. Bottles were shaken at 125 rpm for 48 h at constant temperature
(23.0 1

C). R
Fe
modied adsorbents were not included in the pH study due
to the similar arsenic adsorptive results with Fe modied adsorbents in earlier
studies.
Temperature Study
Measurements for the effects of increased temperature on adsorbent per-
formance used 0.1000 g adsorbent (iron-treated zeolites and activated carbon)
added to 50.00 ml of 50 g l
1
arsenate or arsenite solution. Bottles were shaken
at a shake speed of 5 for 48 h at 23.0 1

C or 35.0 1

C in American Shaking
Water Bath(Model #YB-531, Japan). R
Fe
modied adsorbents were not included
in the pH study due to the similar arsenic adsorptive results with Fe modied
adsorbents in earlier studies.
Ionic Strength Study
Measurements for the effects of ionic strength on adsorbent performance
used 0.1000 g of each adsorbent (iron-treated zeolites and activated carbon)
added to 50.00 ml of 50 g l
1
arsenate or arsenite solution. Solutions contained
varied amounts of KNO
3
(0 M, 0.01 M, or 0.1 M). Bottles were shakenat 125 rpm
730
for 48 h at constant temperature (23.0 1

C). R
Fe
modied adsorbents were not
included in the pH study due to the similar arsenic adsorptive results with Fe
modied adsorbents in earlier studies.
Desorption Study
Dry adsorbent samples previously exposed to arsenate or arsenite were ob-
tained using a Buchner funnel to collect adsorbents on lter paper. Filter paper
and adsorbents were air-dried for 48 h. The desorption study used 0.0500 g of
air-dried adsorbent added to 50.00 ml of DI water (N
2
purged in the case of
arsenite) adjusted to varying pH (2.0 to 12.0) using HNO
3
or NaOH. Bottles
were shaken at 125 rpm for 48 h at constant temperature (23.0 1

C).
Statistical Analysis
Comparison between adsorbents and other parameters was completed by
single and two-factor ANOVA. Percent relative standard deviations were com-
puted for all replicate samples.
RESULTS AND DISCUSSION
Adsorbent Study
Iron-treated activated carbon and chabazite showed the most promise as
low-cost As(V) and As(III) adsorbents. Iron-treated activated carbon removed
approximately 60% of arsenate and arsenite (Fig. 1). Iron-treated chabazite
and 5A removed approximately 50% of arsenate and 30% of arsenite while
iron-treated clinoptilolite and 13X were ineffective in both arsenate and ar-
senite removal. The higher surface area for chabazite, compared to clinop-
tilolite, may provide more accessibility for ion exchange. These results were
consistent with Abdel-Fattah, Ansari, and Voice
[35]
in which iron-treated ac-
tivated carbon and chabazite were capable of reducing As(V) levels to below
10 g l
1
as well as reducing As(III) levels. Up to 95% of As(V) and As(III)
has been reported removed using clinoptilolite, but the adsorption process re-
quired 30 days and a 1:5 solid to solution ratio.
[10]
There was no signicant
difference in adsorptive capacity between the Fe modied and the R
Fe
modied
adsorbents.
Isotherm Study
The theoretical adsorption capacity of an adsorbent can be calculated with
an adsorption isotherm.
[36]
The Langmuir and Freundlich models used for
this study are the most common adsorption isotherm models used for aqueous
732
Table 2: Langmuir and Freundlich models parameters and the corresponding
correlation coefcients.
Langmuir model
Contaminant/Adsorbent
2
Arsenate
Activated Carbon, Fe modied 23.583 35.34 0.999
Activated Carbon, R
Fe
modied 1.588 78.74 0.933
Clinoptilolite, Fe modied 0.289 30.21 0.980
Clinoptilolite, R
Fe
modied 0.110 38.02 0.896
5A, R
Fe
modied 0.164 40.98 0.828
Freundlich model
2
Arsenate
Chabazite, Fe modied 4775.293 5.432 0.909
Chabazite, R
Fe
modied 27.810 1.727 0.785
5A, Fe modied 38.450 1.002 0.941
Arsenite
Activated Carbon, R
Fe
modied 181.970 1.606 0.906
Chabazite, Fe modied 363.078 1.700 0.931
Chabazite, R
Fe
modied 56.247 1.570 0.936
Clinoptilolite, Fe modied 14.191 1.427 0.895
Clinoptilolite, R
Fe
modied 11.863 1.417 0.867
5A, R
Fe
modied 0.636 0.770 0.827
the Langmuir constant ( ) represents the degree of sorption afnity the ad-
sorbate has to the adsorbent. A much stronger afnity for arsenate adsorption
by Fe modied adsorbents, especially activated carbon, is indicated by a
over twenty times greater for Fe modied activated carbon than
Fe
modied
activated carbon and two times greater for Fe modied clinoptilolite than R
Fe
modied clinoptilolite. Deliyanni et al.
[12]
reported correlation with the exper-
imental data and the Langmuir model for arsenate adsorption on akaganeite
(-FeO(OH)). The maximum sorption capacity was found to be 79 mg arsen-
ate per g of R
Fe
modied activated carbon. Although lower than the 120 mg
As(V) per g of akaganeite reported,
[12]
all adsorbents showing the best corre-
lation with the Langmuir model had higher maximum sorption capacity than
previously reported (25 mg As(V) per g of goethite).
[25]
and values for
arsenate adsorption on iron-treated clinoptilolite were comparable to values ob-
tained by Elizalde-Gonzalez
[911]
using clinoptilolite from Mexico. A strong t
to the Langmuir model for the adsorption of As(V) and As(III) on Fe(III)-loaded
chelating resin (45 mg Fe g
1
) has been reported.
[38]
Adsorption on nonuniform sites, either preexisting in the different adsorp-
tion sites or caused by repulsive forces between adsorbed atoms or molecules,
is the basis of the Freundlich model. Arsenate adsorption by iron-treated
chabazite and arsenite adsorption by iron-treated chabazite, clinoptilolite, and
activated carbon best t the Freundlich model (Table 2, Fig. 2). The Freundlich
733
Figure 2: Adsorption isotherms of arsenic ions by iron-treated activated carbon ( & ),
chabazite ( & ), clinoptilolite ( & ), 5A ( & ), and 13X ( & ).
model assumes an innite supply of unreacted adsorbent sites and tends to rep-
resent heterogeneous materials better than other models. The higher the Fre-
undlich constant ( ), the higher the potential is to characterize the adsorbent
as more reactive, although the constant tends to be site and adsorbent specic.
734
The constant is an associated empirical constant that is dependent on the
heterogeneity of sorbing sites.
[39]
For values greater than 1, interactions are
binding at sorbing sites, whereas values less than 1 suggest the model does
not approximate the binding interactions for the process. Iron-treated chabazite
has a larger value for both arsenate and arsenite removal than iron-treated
clinoptilolite indicating more adsorption at equilibriumfor chabazite. and
values for arsenite adsorption on iron-treated clinoptilolite were comparable to
values obtained using clinoptilolite from Mexico.
[11]
Arsenate and arsenite ad-
sorption on iron oxide-coated sand and ferrihydrite
[14]
and As(V) adsorption on
aluminum-loaded Shirasu-zeolite
[8]
were reported as best tting the Freundlich
isotherm.
In conclusion, linear sorption characteristics were observed to be con-
sistent with the Langmuir-type plot. Previous studies have suggested that
isotherms relevant to anion sorption are typically pseudo-Langmuirian at
all sorbate/sorbent ratios, indicating one dominant type of binding site (lig-
and exchange/complexation).
[40]
Sorption of cations are best modeled follow-
ing the Freundlich equation as they are said to correspond for two-sites
adsorption.
Qualitative Analysis
Fourier transmission infrared (FTIR) spectrophotometer results, both be-
fore and after iron treatment and exposure to arsenate and arsenite, show
several frequency shifts (Table 3). Oscillations of the zeolite framework in the
2001200 cm
1
range have been classied as being either structure-sensitive
oscillations of external tetrahedral bonds or structure-insensitive oscillations
of individual tetrahedral.
[41]
Bands were ascribed to three modes: antisym-
metric stretches (9501250 cm
1
), symmetric stretches (650720 cm
1
), and
deformations (490500 cm
1
). Vibrations of the Si-O(Al) bond caused by inter-
nal deformation (symmetric and antisymmetric) are revealed at the following
ranges: 650720, 750820, and 9001250 cm
1
. Bands within these ranges
were evident in this study: 650720 cm
1
(720 cm
1
for chabazite, 675 cm
1
for 5A, and 675 cm
1
for 13X); 750820 cm
1
(760 cm
1
for chabazite, 790
cm
1
for clinoptilolite, and 755 cm
1
for 13X); and 9501250 cm
1
(1040 cm
1
for chabazite, 1055 and 1200 cm
1
for clinoptilolite, and 975 cm
1
for 13X).
Oscillations of chains of aluminosilicate oxygen tetrahedrals are revealed at
550650 cm
1
.
[41]
Bands within this range were evident in this study at 520
and 640 cm
1
for chabazite, 605 cm
1
for clinoptilolite, 550 cm
1
for 5A, 565
cm
1
for 13X. In the region of deformation vibrations, one mid-1600 cm
1
band
is observed for all zeolites due to water bound in various ways in the lat-
tice. The band was observed in this study for all the zeolites.
[41]
Weak bands
were reported at 600 and 796 cm
1
and assigned to the external tetrahe-
dral vibrations for Mexican clinoptilolite.
[11]
A strong band at 10001100 cm
1
736
reect the effect of the heavier iron ions on vibration frequencies, and the fact
that stronger bonds usually vibrate faster than weaker bonds.
[42]
In this study, chabazite was characterized by absorption frequencies 520,
640, 720, 760, and 1040 cm
1
. For both Fe modied and R
Fe
modied chabazite,
the bands at 760 cm
1
shifted to higher wave numbers. For R
Fe
modied
chabazite, the bands at 640 cm
1
and 1040 cm
1
also shifted to higher wave
numbers. With the addition of arsenate to both iron treatments, the band at
760 cm
1
nearly returned to the untreated wave number, the band at 720
cm
1
shifted to a slightly lower wave number, and the band at 1040 cm
1
shifted to a slightly higher wave number. With the addition of arsenite, the
band at 760 cm
1
nearly returned to the untreated wave number for Fe
modied chabazite and shifted to the highest wave number for R
Fe
modied
chabazite.
Clinoptilolite was characterized by absorption frequencies 605, 790, 1055,
and 1200 cm
1
. For bothFe modied and R
Fe
modied clinoptilolite, the bands at
1060 cm
1
shifted to higher wave numbers. In addition, the bands at 790 cm
1
and 1200 cm
1
shifted to lower wave numbers for Fe modied clinoptilolite,
while the bands at 605 cm
1
and 1200 cm
1
shifted to higher wave numbers
for R
Fe
modied clinoptilolite. With the addition of arsenate to both iron treat-
ments, the bands at 790 cm
1
shifted to higher wave numbers than untreated
or iron-treated clinoptilolite. The addition of arsenate to Fe modied clinoptilo-
lite resulted in the bands at 1200 cm
1
, 1055 cm
1
, and 790 cm
1
shifting to
higher wave numbers than for untreated clinoptilolite or Fe modied clinoptilo-
lite not exposed to arsenate. The addition of arsenite to both iron treatments
resulted in the bands at 1200 cm
1
shifting to wave numbers equivalent to that
of untreated clinoptilolite. The addition of arsenite to both iron treatments
also resulted in the bands at 790 cm
1
shifting to lower wave numbers than
untreated or iron-treated clinoptilolite; this shift was nearly 12 cm
1
for R
Fe
-
treated clinoptilolite exposed to arsenite. Additionally, the exposure to arsenite
for Fe modied clinoptilolite resulted inthe band at 1055 cm
1
shifting to higher
wave numbers than untreated or Fe modied clinoptilolite by nearly 12 cm
1
.
5A was characterized by absorption frequencies 550, 675, 1010, and
1470 cm
1
. For both Fe and R
Fe
modied 5A, the bands at 675 cm
1
shifted
to wave numbers nearly 14 cm
1
lower than untreated 5A and a shoulder ap-
peared (at 1135 cm
1
for Fe modied 5A and 1145 cm
1
for R
Fe
modied 5A).
Fe modied 5A also showed a shift to a higher wave number at the 1010 cm
1
band and a shift to a lower wave number at the 550 cm
1
band. R
Fe
modied 5A
showed a shift to a lower wave number at the 1010 cm
1
band. With the addi-
tion of arsenate or arsenite, the shoulder disappeared for both iron treatments.
Fe modied 5A returned to near untreated wave numbers at the 1010 cm
1
and
550 cm
1
bands with exposure to arsenate and arsenite; however, the 675 cm
1
bands remained lower than for untreated 5A. R
Fe
modied 5A showed a further
shift to lower wave numbers with exposure to arsenate at the 1010 cm
1
band;
737
exposure to arsenite for R
Fe
modied 5A did not result in a signicant shift
for this band. Exposure to arsenate for R
Fe
modied 5A nearly returned the
675 cm
1
band to the untreated wave number but shifted the 550 cm
1
band
to a lower wave number. Exposure to arsenite for R
Fe
modied 5A shifted the
675 cm
1
band to the highest wave number for this adsorbent and shifted the
550 cm
1
band to a the lowest wave number for this adsorbent.
13X was characterized by absorption frequencies 565, 675, 755, and
975 cm
1
. Iron treatments of 13X result in a signicant shift to higher wave
numbers for the 975 cm
1
bands; exposure to arsenate or arsenite maintains
this shift. Iron treatments of 13X result in a shift to lower wave numbers for
the 565 cm
1
bands; exposure to arsenate or arsenite negates this shift. Iron
treatments of 13X also result in a small shift to lower wave numbers for the
755 cm
1
bands. Further exposure of Fe modied 13X to arsenate maintains
this shift, while exposure to arsenite results in a signicant shift to a higher
wave number. Further exposure of R
Fe
modied 13X to arsenate negates the
shift (the wave number returns to the untreated value), while exposure to ar-
senite results in a marked shift to a lower wave number. 13X exhibited a shift
to increasingly higher wave numbers for the 675 cm
1
band between untreated
13X, Fe modication, exposure to arsenate, and exposure to arsenite. Similar
shifts in the 675 cm
1
bands were not evident with R
Fe
modied 13X.
Oncomparing, the FTIRabsorptionfrequencies inthe range 7581068 cm
1
for the iron-modied zeolites exposed to arsenate and arsenite were shifted,
and distinct from iron arsenate (arsenite) salts, due to surface complexation
between iron sites and arsenate and arsenite species rather than solid phase
precipitation.
[43]
Furthermore, the XRD patterns for adsorbents exposed to ar-
senate and arsenite had no new peaks at 22

and 54

indexed for iron arsenate

(arsenite) precipitates, which also indicates surface complexation.
[12]
Kinetic Study
Iron-treated zeolites 13X or 5A were not included in further studies due to
their adsorptive performance and higher cost compared to the natural zeolites
and activated carbon. Adsorption due solely to electrostatic forces is very rapid,
in the order of milliseconds
[44]
to seconds.
[45]
Adsorption of arsenic in this study
was in the order of hours, which may indicate a specic adsorption or surface
complexation as indicated in the FTIR measurements. Iron-treated activated
carbon, chabazite, and clinoptilolite showed an initial drop in arsenate con-
centration followed by a gradual decline over the remaining two days. Arsenite
removal for the three iron-treated adsorbents showed a gradual decline over the
entire 48-h period. These results are consistent with the literature: akaganeite
(-FeO(OH)) required an order of hours for maximum removal,
[12]
ferrihydrite
required an order of hours for removal,
[16]
and natural clinoptilolite required
24 h to show signicant adsorption of arsenic.
[9]
Sorption on zeolites may be
739
Figure 3: Remaining arsenate concentration versus time by iron-treated activated carbon
( ), chabazite ( ), and clinoptilolite ( ).
(pH 1214).
[9]
Therefore adsorption of arsenic species is highly dependent on
pH.
Due to insignicant differences in adsorptive capacity between the Fe modi-
ed and the R
Fe
modied adsorbents, R
Fe
modied adsorbents were not included
in the remaining studies. The inuence on adsorption by pH was studied for
the pH interval 2.012.0 (Fig. 6). Fe modied activated carbon most effectively
adsorbed arsenate in the pH interval 7.011.0 with nearly 90% removal at pH
10.0; chabazite in the pH interval 4.05.0, and clinoptilolite in the pH interval
3.07.0. Fe modied activated carbon most effectively adsorbed arsenite in the
pH interval 7.011.0 with nearly 60% removal at pH 10.0; chabazite in the pH
interval 7.010.0 with nearly 50%removal, and clinoptilolite in the pHinterval
740
Figure 4: Remaining arsenite concentration versus time by iron-treated activated carbon
( ), chabazite ( ), and clinoptilolite ( ).
4.011.0. There is a signicant difference in adsorption due to the inuence of
pH ( 0.001, = 0.001, = 33). pH dependence of As(V) and As(III) adsorp-
tion was demonstrated for goethite, lepidocrocite, and hematite,
[46]
arsenite and
arsenate removal by clinoptilolite was reported in the pH interval of 4 to 11,
[9]
and high As(III) adsorption on basic yttrium carbonate was reported in the
pH interval 9.810.5 with As(V) removal in the pH interval 7.59.0.
[47]
High-
est As(V) removal by goethite was reported in the pH interval 36.
[26]
High
As(III) removal using ferrihydrite was reported at pH 7.5 and greater,
[16]
and
increased As(III) adsorption with increasing pH was reported using activated
carbon derived from coconut husk.
[48]
741
Figure 5: Arsenic solution concentration versus time for arsenate and arsenite ion sorption
by iron-treated activated carbon ( & ), chabazite ( & ), and clinoptilolite ( & ).
It can then be presumed that the following processes are involved in ad-
sorption at different pH ranges:
Z(A) [Fe OH]
+
+(ads)HOAs(OH)
2
for As(III) at pH < 7 (7)
Z(A) Fe O

H
+
+(ads)

OAs(OH)
2
for As(III) at pH > 7 (8)
Z(A) Fe O

H
+
+(ads)

OAsO(OH)
2
for As(V) at pH 3 12 (9)
The adsorption of As(V) by iron-modied activated carbon, chabazite, and
clinoptilolite is of a wide pH range, which should be of advantage for practical
operation.
745
arsenic oxyanions.
[26]
Fe modied adsorbents showed enhancement in arsenite
adsorptive performance in the presence of KNO
3
, but as the concentration of
KNO
3
was further increased, arsenite absorption declined (but remained above
the control level). Coexisting solutes in Bangladesh waters have been reported
to adversely affect the removal of As(V) and As(III) by iron hydroxides.
[31]
Addi-
tional studies should assess the inuence of common groundwater anions such
as phosphate, silicate, carbonate, sulfate, chromate, molybdate, nitrate, and
borate on arsenic removal by iron-treated activated carbon and zeolites.
Desorption Study
Over the tested pH interval 2.012.0, iron-treated activated carbon,
chabazite, and clinoptilolite held arsenate and arsenite in the pH interval 3.0
10.0, with desorption occurring outside this range (Fig. 9). The results showthe
spent adsorbents can be effectively regenerated for further use with HNO
3
.
Proposed Mechanism
The removal of As(III) and As(V) can be explained by the respective spe-
ciation differences. As(V) in solution above pH 3 is present in anionic forms
(AsO
3
4
, HAsO
2
4
, H
2
AsO

4
) and therefore can be effectively removed by iron
hydroxide attached to the surface of the zeolites (activated carbon), which at
this pH range are present as cationic monomers (Fe(OH)
+
2
). On the other hand,
As(III) is present as ananionexclusively above pH9 (AsO
3
3
, HAsO
2
3
, H
2
AsO

3
),
whereas in the pH interval 69 only a small percentage of H
3
AsO
3
is dissoci-
ated. Thus for effective treatment of As(III) the solution pH must exceed 7.
The responsible mechanism for the removal of arsenic was adsorption on
iron-modied adsorbents, which refers to the formation of surface complexes
between soluble arsenic species and the surface hydroxyl groups, as arsenic
gets in contact with the iron sites.
[40,49]
Ferric arsenate or ferric arsenite is
produced.
Simplied equation of arsenate sorption on iron sites:
Z(A) FeOH+H
3
AsO
4
Z(A) Fe H
2
AsO
4
+H
2
O (10)
Simplied equation of arsenite sorption on iron sites:
Z(A) FeOH+H
3
AsO
3
Z(A) Fe H
2
AsO
3
+H
2
O (11)
In summary, the authors believe the mechanism for arsenic removal is surface
complexation, since the pH study showed very stable arsenic removal over a
wide pH range and desorption did not occur without extreme pH. If the mech-
anism were coulombic/electrostatic, the pH study would have shown very un-
stable results.
746
Figure 9: Effects of pH on arsenate desorption ( ) and arsenite desorption ( ) on iron
modied activated carbon, chabazite, and clinoptilolite.
CONCLUSIONS
Iron-treated activated carbon and chabazite showed promise as efcient low-
cost adsorbents for arsenate and arsenite removal. Using the same mass of
adsorbent, iron-treated clinoptilolite did not perform as well. XRD and FTIR
747
measurements indicated surface complexation between iron sites and arsenate
(arsenite) species rather than solid phase precipitates. Adsorption was signi-
cantly affected by pHwith adsorption generally highest for arsenate and arsen-
ite in the pH interval 7.011.0 for iron-treated activated carbon and zeolites.
Increasing temperature and ionic strength improved adsorption performance
for Fe modied activated carbon and zeolites.
REFERENCES
1. Melitas, N.; Wang, J.; Conklin, M.; ODay, P.; Farrell, J. Understanding soluble ar-
senate removal kinetics by zero-valent iron media. Environ. Sci. Technol. 2002, (9),
20742081.
2. WHO. Arsenic in drinking water. Fact sheet no. 210, 2001. http://www.who.int/inf-
fs/en/fact210.html.
3. BGS and DPHE. Arsenic contamination of groundwater in Bangladesh. In
, Vol. 2; Kinniburgh, D.G.; Smedley, P.L. Eds. BCSR WC/00/19 British Geological
Survey: Keyworth, 2001.
4. Ayotte, J.D.; Montgomery, D.L.; Flanagan, S.M.; Robinson, K.W. Arsenic in ground-
water in eastern New England: Occurrence, controls, and human health implications.
Environ. Sci. Technol. 2003, (10), 20752083.
5. Welch, A.H., Helsel, D.R.; Focazio, M.J.; Watkins, S.A. 1999. Arsenic in ground wa-
ter supplies of the United States. pp. 917. In ;
Chappell, W.R.; Abernathy, C.O.; Calderon, R.L., Eds. Elsevier Science: New York, 1999;
917.
6. Subramanian, K.S.; Viraraghavan, T.; Phommavong, T.; Tanjore, S. Manganese
greensand for removal of arsenic in drinking water. Wat. Qual. Res. J. Can. 1997, (3),
551561.
7. Yoshitake, H.; Yokoi, T.; Tatsumi, T. Adsorption behavior of arsenate at transition
metal cations captured by amino-functionalized mesoporous silicas. Chem. Mater. 2003,
(8), 17131721.
8. Xu, Y.H.; Nakajima, T.; Ohki, A. Adsorption and removal of arsenic(V) fromdrinking
water by aluminum-loaded Shirasu-zeolite. J. Hazard. Mater. 2002, (3), 275287.
9. Elizalde-Gonzalez, M.P.; Mattusch, J.; Einicke, W.-D.; Wennrich, R. Sorption on nat-
ural solids for arsenic removal. Chem. Eng. J. 2001, (13), 187195.
10. Elizalde-Gonzalez, M.P.; Mattusch, J.; Wennrich, R. Application of natural zeolites
for preconcentration of arsenic species in water samples. J. Environ. Monitor. 2001,
(1), 2226.
11. Elizalde-Gonzalez, M.P.; Mattusch, J.; Wennrich, R.; Morgenstern, P. Uptake of
arsenite and arsenate by clinoptilolite-rich tuffs. Micropor. Mesopor. Mater. 2001, ,
277286.
12. Deliyanni, E.A.; Bakoyannakis, D.N.; Zouboulis, A.I.; Matis, K.A. Sorption of As(V)
ions by akaganeite-type nanocrystals. Chemosphere 2003, , 155163.
13. Huang, C.P.; Vane, L.M. Enhancing As
5+
removal by a Fe
2+
-treated activated car-
bon. Res. J. Water Pollut. Cont. Fed. 1989, (9), 15961603.
14. Thirunavukkarasu, O.S.; Viraraghavan, T.; Subramanian, K.S. Removal of arsenic
in drinking water by iron oxide-coated sand and ferrihydritebatch studies. Wat. Qual.
Res. J. Can. 2001, (1), 5570.
749
33. Muhammad, N.; Parr, J.; Smith, M.D.; Wheatley, A.D. Adsorption of heavy metals
in slow sand lters.
. Islamabad, Pakistan, 1998; 346349.
34. McMurry, J.; Fay, R.C. , 2nd Ed.; Prentice-Hall: New Jersey, 1998; 472
482, 838-843.
35. Abdel-Fattah, T.M.; Ansari, A.Z.; Voice, T.C. Screening of low-cost adsorbents for
arsenic removal. American Chemical Society. Prepr. Extended Abstracts 2000, (1),
422423.
36. Tchobanoglous, G.; Burton, F.L.
, 3rd Ed. McGraw-Hill: New York, 1991; 318324.
37. Banat, F.; Al-Asheh, S.; Al-Rousan, D.A comparative study of copper and zinc ion
adsorption on to activated and non-activated date-pits. Adsorpt. Sci. Technol. 2002,
(4), 319335.
38. Matsunaga, H.; Yokoyama, T.; Eldridge, R.J.; Bolto, B.A. Adsorption characteristics
of arsenic(III) and arsenic(V) on iron(III)-loaded chelating resin having lysine-N

,N

-
diacetic acid moiety. React. Funct. Polym. 1996, , 167174.
39. Calace, N.; Di Muro, A.; Nardi, E.; Petronio, B.M.; Pietroletti, M. Adsorption
isotherms for describing heavy-metal retention in paper mill sludges. Ind. Eng. Chem.
Res. 2002, , 54915497.
40. Dzombak, D.A.; Morel, F.M.M.
. Wiley: New York, 1990.
41. Tsitsishvili, G.V.; Andronikashvili, T.G.; Kirov, G.N.; Filizova, L.D. .
Ellis Horwood: New York, 1992.
42. Skoog, D.A.; Holler, F.J.; Nieman, T.A. , Fifth
Ed. Harcourt Brace and Co.: Orland, 1998.
43. Goldberg, S.; Johnson, C.T. Mechanisms of arsenic adsorption on amorphous ox-
ides evaluated using macroscopic measurements, vibrational spectroscopy, and surface
complexation modeling. J. Coll. Interf. Sci. 2001, , 204216.
44. Grossl, P.R.; Eick, M.; Sparks, D.L.; Goldberg, S.; Ainsworth, C.C. Arsenate and
chromate retention mechanisms on goethite. 2. Kinetic evaluation using a pressure-
jump relaxation technique. Environ. Sci. Technol. 1997, , 321326.
45. Pierce, M.L.; Moore, C.B. Adsorption of arsenite and arsenate on amorphous iron
hydroxide. Water Res. 1982, (7), 12471253.
46. Bowell, R.J. Sorption of arsenic by iron oxides and oxyhydroxides in soils. Appl.
Geochem. 1994, (3), 279286.
47. Wasay, S.A.; Haron, M.J.; Uchiumi, A.; Tokunaga, S. Removal of arsenite and ar-
senate ions from aqueous solution by basic yttrium carbonate. Water Res. 1996, (5),
11431148.
48. Manju, G.N.; Raji, C.; Anirudhan, T.S. Evaluation of coconut husk carbon for the
removal of arsenic from water. Water Res. 1998, (10), 30623070.
49. Edwards, M. Chemistry of arsenic: Removal during coagulation and Fe-Mn oxida-
tion. J. AWWA 1994, (9), 6478.