Water Research 37 (2003) 164–176

Arsenic removal using a polymeric/inorganic hybrid sorbent

Matthew J. DeMarco, Arup K. SenGupta*, John E. Greenleaf
Department of Civil and Environmental Engineering, Lehigh Univeristy, 13 E. Packer Avenue, Bethlehem, PA 18015, USA

Received 1 September 2001; received in revised form 1 May 2002; accepted 6 June 2002

Abstract

A fixed-bed sorption process can be very effective in removing trace concentrations of arsenic from contaminated
groundwater provided: the sorbent is very selective toward both As(III) and As(V) species; the influent and treated
water do not warrant any additional pre- or post- treatment; pH and composition of the raw water with respect to other
electrolytes remain unchanged besides arsenic removal, and the sorbent is durable with excellent attrition resistance
properties. In addition, the sorbent should be amenable to efficient regeneration for multiple reuse. This study reports
the results of an extensive investigation pertaining to arsenic removal properties of a polymeric/inorganic hybrid
sorbent. Each hybrid sorbent particle is essentially a spherical macroporous cation exchanger bead within which
agglomerates of nanoscale hydrated Fe oxide (HFO) particles have been uniformly and irreversibly dispersed using a
simple chemical–thermal treatment. The new sorbent, referred to as hybrid ion exchanger or HIX, combines excellent
mechanical and hydraulic properties of spherical polymeric beads with selective As(III) and As(V) sorption properties
of HFO nanoparticles at circum-neutral pH. Comparison of the results of fixed-bed column runs between the new
sorbent and the polymeric anion exchanger confirmed that both As(V) and As(III) were removed very selectively with
HIX. Equally important, no pH adjustment, pre- or post-treatment was warranted. Besides the absence of arsenic, the
treated water composition was identical to that of influent water. HIX was amenable to efficient in situ regeneration
with caustic soda and could subsequently be brought into service following a short rinse with carbon dioxide sparged
water. During fixed-bed column runs, intraparticle diffusion was identified as the primary rate-limiting step for both
As(III) and As(V) sorption. Repeated use of the same HIX particles during various laboratory investigations provided
strong evidence that the new sorbent possesses excellent attrition resistance properties and retains its arsenic removal
capacity over cycles. r 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Arsenic; Sorbent; Ion exchange; Toxic; Drinking water

1. Introduction thousand (4000) water supply systems serving an

In recent years, the presence of dissolved arsenic in
contaminated groundwater has emerged as a major
concern on a global scale [1,2]. According to the
estimate of the United States Environmental Protection
Agency (USEPA), the newly promulgated 10 mg/L
arsenic maximum contaminant level (MCL) in drinking
water would require corrective action for over four
*Corresponding author. Fax: +1-610-758-6405.
E-mail address: arup.sengupta@lehigh.edu
(A.K. SenGupta).
approximate population of 20 million. A vast majority
of these systems use groundwaters. A world away, over
70 million people in Bangladesh and in other regions of
Indian subcontinent are routinely exposed to arsenic
poisoning through drinking groundwater [2–4]. Natural
geochemical contamination through soil leaching is the
primary contributor of dissolved arsenic in ground-
waters around the world.
Removal of dissolved arsenic from natural ground-
water essentially involves selective separation of As(V)
and As(III) oxyanions or oxyacids from water. The
chemistry of As(III) and As(V) compounds (also

0043-1354/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 3 - 1 3 5 4 ( 0 2 ) 0 0 2 3 8 - 5
 M.J. DeMarco et al. / Water Research 37 (2003) 164–176
Table 1
Oxyacids and conjugate anions of As(V) and As(III)
Parent oxyacids pKa values Predominant
dissolved species
at pH 6.0
As(V): H3AsO4 pKa1 ¼ 2:2
pKa2 ¼ 6:98
pKa3 ¼ 11:6
As(III): HAsO2 pKa1 ¼ 9:2
referred to as arsenites and arsenates) and their relative
distribution are influenced by pH and the redox
condition. Predominance diagrams (i.e., pe-pH plots)
to this effect are available in the open literature [5,6]).
Under atmospheric or more oxidizing environment, the
predominant species is As(V), which, in the pH range of
6–9, exists predominantly as deprotonated oxyanions,
namely, H2AsO
2

4 or HAsO4 . Under mildly reducing
conditions, As(III) is thermodynamically stable and
exists predominantly as H3AsO3 (or HAsO2) at pH
below 9.0. Dissolved As(III) and As(V) compounds have
been found to be simultaneously present in many
contaminated groundwaters [7,8]. Table 1 shows the
major As(V) and As(III) species at slightly above- and
below-neutral pHs and also highlights their ligand
characteristics, i.e., electron donating ability. It is worth
mentioning that at low dosages (i.e., less than 500 mg/L
of water), As(V) compounds are readily reduced to
As(III) inside human bodies [9]. Contrary to belief in
certain quarters, chronic toxicity caused by ingestion of
arsenic-contaminated water is thus unaffected by the
relative distribution of As(III) and As(V) compounds,
i.e., a mere conversion of As(III) to As(V) does not
lower the toxicity of contaminated groundwaters [10].
A number of arsenic removal technologies, namely,
activated alumina sorption, polymeric anion exchange,
sorption by iron oxide coated sand (IOCS) particles,
enhanced coagulation with alum or ferric chloride
dosage, ferric chloride coagulation followed by micro-
filtration, pressurized granulated iron particles, iron
oxide doped alginate, manganese dioxide coated sand,
polymeric ligand exchange and zero-valent iron have
been tried in the laboratory and/or field-scale testing for
removals of trace arsenic [11–25]. As diverse as the
equipment configuration and operational protocol for
the above-mentioned technologies appear to be, ion
exchange and Lewis acid–base interactions are the
underlying arsenic removal mechanisms in every in-
stance. The technical details of the foregoing processes
165
Predominant Sorption interaction
dissolved species
at pH 8.0
As(V) species or arsenates can
undergo Coulombic (ion
exchange) as well as Lewis acid–
base interaction
As(III) species or arsenites can
undergo only Lewis acid–base
interaction
and their relative advantages and disadvantages are
recorded in the open literature [11–25] and being
avoided here for the sake of brevity.
1.1. Desirable attributes of an arsenic removal process
It is universally recognized that a fixed-bed sorption
process is operationally simple, requires virtually no
start-up time and is forgiving toward fluctuations in feed
compositions. However, in order for the fixed-bed
process to be viable and economically competitive, the
sorbent must exhibit high selectivity toward the target
contaminant, be durable and amenable to efficient
regeneration and reuse. Ideally, the removal of the
target contaminant should not cause major changes in
pH or in the composition of the influent water. Two
fixed-bed processes for removing arsenic from ground-
water were recently tried at pilot-scale studies in
Albuquerque, N.M. [18]. They are ion exchange (IX)
and activated alumina (AA). For arsenic-laden ground-
waters with low sulfate content, ion exchange has the
potential to be a viable process but pH and alkalinity of
the product water during pilot-scale operation in
Albuquerque decreased significantly during the initial
period of the column run for tens of bed volumes, thus
necessitating post-treatment pH adjustment. Also, poly-
meric anion exchangers prefer sulfate over arsenate.
Arsenic removal capacity is, therefore, greatly reduced
in the presence of high concentrations of competing
sulfate ions. For activated alumina sorption, pH
adjustment through acid dosage was required to achieve
high arsenic removal capacity. Such operational com-
plexities and concurrent expenses tend to compromise
with the desirable attributes of fixed-bed processes.
It is worth mentioning that As(III) was altogether absent
in Albuquerque groundwater. Arsenic-contaminated
groundwaters, in general, contain both As(III) and
As(V) dissolved species in varying proportions [7,8].
Neither polymeric anion exchanger nor activated
166 M.J. DeMarco et al. / Water Research 37 (2003) 164–176
alumina is effective for As(III) removal unless As(III)
has been pre-oxidized by chlorine, manganese dioxide,
ozone or other innocuous oxidizing agents [20,26,27].
Introducing a pre-oxidation step again poses opera-
tional complexity and diminishes the overall viability of
the fixed-bed process. Collectively, these shortcomings
and the disposal of arsenic-containing spent regenerant
made activated alumina and ion exchange fixed-bed
processes less viable in comparison with coagulation/
micro filtration process in Albuquerque [18].
The broad objective of this study is to present a new
hybrid polymeric/inorganic sorbent that: is compatible
with fixed-bed column processes with excellent mechan-
ical strength and attrition resistance properties; is
selective toward both As(III) and As(V) species; does
not warrant any pre- or post-treatment such as pH
adjustment or oxidation; does not alter the water quality
besides arsenic removal; does not produce fines during
lengthy column runs; and lastly, is amendable to efficient
regeneration and reuse.
1.2. Polymeric/inorganic sorbent: underlying concept,
synthesis and premise of the study
Both amorphous and crystalline hydrated Fe oxide
(HFO) show strong sorption affinity toward both
As(III) and As(V) oxyacids and oxyanions through
ligand exchange in the coordination spheres of structur-
al Fe atoms. Recent investigations using extended X-ray
absorption fine structure (EXAFS) spectroscopy con-
firmed that As(III) and As(V) species are selectively
bound to the oxide surface through formation of inner
sphere complexes [28,29]). However, the current process
of syntheses, although straightforward, produces only
very fine submicron particles which are unusable in fixed
beds because of excessive pressure drops and poor
mechanical strength. In Germany, Jekel and co-workers
recently developed pressurized granular ferric hydroxide
material and tested it for high As(V) removal in fixed
beds [12]. Mechanical strengths and attrition resistance
properties of these granular particles are, however, much
weaker compared to polymeric ion exchanger beads.
After one column run, the materials were disposed of
without regeneration. Iron oxide doped alginate gels
were tested by Min and Hering in various batch
sorption/desorption studies for arsenic removal [14].
These gels have the potential for usage in fixed-bed
columns provided their chemical stability and hydraulic
property (i.e., pressure drop in a column setting) are
appropriately enhanced.
During the last 3 years, we have perfected a simple
chemical–thermal technique to produce a hybrid (poly-
meric/inorganic) sorbent which is selective towards both
As(III) and As(V) compounds and at the same time
compatible with fixed-bed column operation. The hybrid
sorbent consists of spherical macroporous polymeric
Table 2
Salient properties of polymeric cation exchangers
Resin Purolite C-145
Structure (repeating unit)
Functional group Sulfonic acid
Matrix Polystyrene
Capacity (meq/g resin) 4.1 (Dry)
2.5 (Wet)
Manufacturer Purolite Inc., Philadelphia
cation exchanger beads within which submicron HFO
particles have been uniformly and irreversibly dispersed.
The new sorbent, referred to as hybrid ion exchanger or
HIX, combines excellent mechanical strengths and
hydraulic properties of spherical polymeric beads with
selective As(III) and As(V) removal capacities of HFO
microparticles over a wide range of pH. Purolite C-145, a
commercially available macroporous cation exchanger
with polystyrene matrix and sulfonic acid functional
group, was used as the parent cation exchanger. Table 2
provides salient properties of the cation exchanger.
However, no endorsement is implied; similar exchangers
are available also from other manufacturers. The
preparation of HIX consisted of the following three steps:
Step 1: Loading of Fe(III) onto the sulfonic acid sites
of the cation exchanger by passing 4% FeCl3
solution at an approximate pH of 2.0;
Step 2: Desorption of Fe(III) and simultaneous pre-
cipitation of Fe(III) hydroxides within the gel
and pore phase of the exchanger through
passage of a solution containing both NaCl
and NaOH, each at 5% w/v concentration;
Step 3: Rinsing and washing with a 50/50 ethanol–
water solution followed by a mild thermal
treatment (50–601C) for 60 min.
Fig. 1 depicts the major steps of the process. Step 2
was repeated for the second time to achieve greater
Fe(III) loading. Experimental observations suggest that
at the conclusion of Step 3 both amorphous and
crystalline phases were present and submicron or nano-
scale HFO particles coalesced to form agglomerates.
Use of ethanol lowered the dielectric constant of water
 M.J. DeMarco et al. / Water Research 37 (2003) 164–176
Fig. 1. Illustration of a three-step procedure to disperse both crystalline and amorphous HFO microparticles inside the spherical
polymer beads.
and supposedly enhanced the agglomeration of sub-
micron particles through suppression of surface charges.
HFO agglomerates were irreversibly encapsulated with-
in the spherical exchanger beads; turbulence and
mechanical stirring did not result in any loss of HFO
particles whatsoever. Constancy of the sorption capacity
data, as presented later, also tends to confirm this
observation. High concentration of sulfonic acid func-
tional groups allowed high and fairly uniform loading of
HFO particles (approximately 9–12% of Fe by mass)
within the polymeric beads. Since, HFO particles reside
within the macropores of the cation exchanger beads,
they are easily accessible to As(V) anions or arsenates,
i.e., Donnan exclusion effect does not impede As(V)
sorption onto HFO particles.
Fig. 2A shows HIX particles; note that the spherical
geometry is retained after processing. Fig. 2B shows the
scanning electron microphotograph (SEM) of a sliced
parent polymer bead. A number of macropores in the
sizes of 20–300 nm can be readily observed. Fig. 2C
shows a sliced HIX particle; both crystalline and
amorphous HFO nanoparticles are present on the
interior surface. A series of SEM pictures taken for
HIX particles and close physical observation suggest
that HFO agglomerates are accessible to dissolved
solutes through a network of pores. Results of
desorption experiments presented later tend to confirm
167
the same. Both protonated and electrically neutral iron
oxides selectively sorb As(V) and As(III) species through
Lewis acid–base interaction, i.e., formation of inner
sphere complexes. On the contrary, competing Cl
,


SO2
4 and HCO3 ions form only outer sphere complexes
with only weak affinities toward HFO.
A series of laboratory investigations were carried out
to evaluate the pertinent properties of HIX for its long-
term usage in fixed-bed systems for arsenic removal.
Specific objectives of this communication are to present
experimental evidences pertaining to: As(V) and As(III)
sorption behaviors during column runs at slightly
above-neutral pH representative of many groundwaters;
impact on the overall treated water quality; sorption
kinetics; and the efficiency of regeneration and reusa-
bility of the new sorbent material. For comparison, one
commercial strong base anion exchanger with quatern-
ary ammonium functional groups was also included in
the study.
2. Experimental procedures
2.1. Column run, regeneration and rinsing
Fixed-bed column runs were carried out using glass
columns (11 mm in diameter), constant-flow stainless
168 M.J. DeMarco et al. / Water Research 37 (2003) 164–176
Fig. 2. (A) Photographs (10  ) of spherical HIX beads; (B)
scanning electron microphotograph (40,000  ) of a sliced
parent polymer bead; (C) scanning electron microphotograph
(40,000  ) of a sliced HIX bead.
steel pumps and an ISCO fraction collector. The ratio of
column diameter to exchanger bead diameter was
approximately 20:1; earlier work on chromate and
phosphate removals with similar set-ups showed no
premature leakage due to wall effects under identical
conditions [30,31]. The superficial liquid velocity (SLV)
and the empty bed contact time (EBCT) were recorded
for each column run. Deionized water and analytical
grade reagents were used for preparing feed solutions.
For As(III) column runs, nitrogen gas was continu-
ously sparged to guarantee no As(III) was oxidized to
As(V). Independent analyses of feed samples at various
stages of the column run confirmed the same. A blank
test where As(III) feed solution was kept open to
laboratory atmosphere showed only partial (20%)
oxidation to As(V) after 2 weeks.
Exhausted HIX beads following lengthy column runs
were regenerated using 10% w/v sodium hydroxide
solution. EBCT during regeneration was greater (at least
6.0 min) than that used during the sorption cycle.
Following regeneration, the bed was rinsed with carbon
dioxide sparged treated water for about ten to fifteen
bed volumes.
2.2. Kinetic test
Kinetic test for the sorption of As(V) onto HIX was
performed using a set-up shown in Fig. 3. HIX beads
were loaded within the cell of the stirrer, and the stirrer
speed could be adjusted and read directly from a digital
output. When the stirrer assembly rotates, the centrifu-
gal action produces a rapid circulating flow of solution
entering the cage at the bottom and leaving through the
radial holes of the caging. Thus, a vigorous hydro-
dynamic condition surrounding the ion-exchanger beads
is maintained; diffusional resistances in the liquid film
around HIX beads are essentially absent under this
condition. Underlying principles of this apparatus are
identical to the one developed originally by Kitchener
and described by Helfferich [32]. At different time
intervals, small volumes (less than 5 ml) of samples were
collected from the solution and analyzed. Arsenic
concentration in the HIX phase was calculated from
the mass balance. The solution pH was 7.2–7.5 during
the course of the kinetic test and sulfate and bicarbonate
were present as competing ions.
2.3. Chemical analyses
Chloride and sulfate were analyzed using Dionex Ion
Chromatograph (Model DX-120 IC). Bicarbonate or
inorganic carbon was determined using a Shimadzu
Carbon Analyzer (Model 5050A). Particle size analysis
of the HIX beads was carried out using a Beckman/
Coulter LS 100Q Particle Counter with a binocular
optical system.
Total dissolved arsenic analyses of samples were
carried out using a Perkin Elmer atomic absorption
spectrophotometer (AAS) with graphite furnace acces-
sories (Model SIMAA 6000) and an electrodeless
discharge lamp (EDL). Filtration of samples through
0.4 mm filters and subsequent analyses confirmed that
arsenic during fixed-bed column run was present only in
 M.J. DeMarco et al. / Water Research 37 (2003) 164–176
Fig. 3. A schematic of the batch kinetic test apparatus and reactor-stirrer assembly.
the dissolved state. The stock solutions for calibration
were prepared using analytical grade arsenite (NaAsO2)
and arsenate (KH2AsO4). Analysis was performed at a
wavelength of 193.7 nm and each sample injection
included a palladium matrix modifier. Furnace condi-
tions included a 1101C injection temperature with a 30 s
hold time, a 12001C pretreatment temperature with a
20 s hold time, and a 23001C atomization temperature
for 3 s. For quality control, arsenic concentrations of
stock solutions were intermittently checked against an
arsenic standard solution purchased from Fisher Scien-
tific Inc. For analyses of As(V) and As(III), they were
first separated at pH near 4.0 using a strong-base anion
exchange resin in accordance with the technique
developed by Ficklin [33] and Clifford [34]. Total arsenic
was determined from the original sample, As(III) from
the sample collected at the exit of the anion exchange
column and As(V) by difference. The technique was
validated by using samples of known As(III) and As(V)
concentrations.
3. Results and discussion
3.1. Fixed-bed column runs
3.1.1. As(V) in the feed
Fig. 4 shows complete effluent histories during a fixed-
bed column run using HIX. The influent composition
and the hydrodynamic conditions during the column
run, i.e., EBCT and SLV are also provided in Fig. 4.
169
Influent pH was slightly above neutral and no pH
adjustment was made. The following observations are
noteworthy: First, other anions, namely, sulfate and
chloride broke through almost immediately after the
start of the column run while 10 mg/L As breakthrough
was observed after about 4000 bed volumes. Mass-
balance calculations confirmed that in the absence
of any kinetic limitations, HIX may treat well over
eight thousand bed volumes of contaminated water
prior to arsenic breakthrough and the complete As
effluent history in Fig. 4 was in agreement with that.
Second, the effluent pH of 7.2 during the entire column
run was essentially equal to the influent pH. Thus,
besides the absence of dissolved arsenic, the influent
composition and pH remained essentially unchanged
following the passage of contaminated water through
HIX column.
Fig. 5 shows comparison of effluent histories of As(V)
during separate column runs with two different sor-
bents, namely, a strong-base polymeric anion exchanger
(IRA-900 from Rohm and Haas Co., Philadelphia, PA,
USA) and the HIX. While complete arsenic break-
through for IRA-900 started at less than 300 bed
volumes, HIX was removing arsenic well after 5000 bed
volumes. Also, arsenic breakthrough from IRA-900
underwent chromatographic elution, i.e., the concentra-
tion of arsenic at the exit of the column became
significantly greater than its influent concentration. This
phenomenon is attributed to anion exchanger’s greater
sulfate selectivity over As(V) oxyanions and similar
observations were also made by other investigators.
170 M.J. DeMarco et al. / Water Research 37 (2003) 164–176

Fig. 4. A complete normalized effluent history of sulfate, chloride, arsenic and pH during a fixed-bed column run with HIX: EBCT,
empty bed contact time; SLV, superficial liquid velocity.

Fig. 5. Comparison of As(V) effluent histories between a strong base anion exchanger (IRA-900) and HIX under otherwise identical
conditions.

3.1.2. As(III) in the feed
Fig. 6 provides As(III) effluent histories of two
separate column runs using the same feed composition
with an As(III) concentration of 100 mg/L. For one run,
the anion exchanger IRA-900 was used in the fixed-bed
column while HIX was the sorbent during the second
run. At near-neutral pH, As(III) is non-ionized (i.e.,
HAsO2 or H3AsO3) and, therefore, IRA-900 was unable
to remove As(III) as shown in Fig. 6. In comparison,
As(III) was removed over a long period of time by HIX.
Total dissolved arsenic breakthrough after 2000 bed
volumes was less than 10% of its influent concentration.
Nitrogen was continuously sparged in the influent
storage tank to eliminate any possible As(III) oxidation
to As(V). Intermittent analyses of the influent according
to the protocol prescribed by Ficklin [33] and Clifford
[34] confirmed that As(V) was altogether absent in the
feed. For HIX run with As(III), treated water composi-
tion with respect to pH and other electrolytes remained
essentially the same as the influent. During all the fixed-
bed column runs, hydraulic characteristic, i.e., pressure
drop across HIX columns was essentially identical to
 M.J. DeMarco et al. / Water Research 37 (2003) 164–176 171

Fig. 6. Comparison of As(III) effluent histories between an anion exchanger (IRA-900) and HIX under otherwise identical conditions.

that in anion exchange columns. No generation of fines

or a sudden decrease in flow rate was observed for
month-long column runs with HIX.

3.2. Sorption kinetics: mechanism and effective

intraparticle diffusivity

HIX column runs were subjected to interruption tests

during the time periods marked within circles in Figs. 5
and 6. As the dissolved arsenic gradually started exiting
from the column, the influent flow was deliberately
discontinued for 24 h. When the flow was resumed, the
effluent arsenic concentration dropped significantly as
can be seen from the two figures. Following the passage
of several hundred bed volumes of influent solution after
the restart, arsenic at the exit of the column reached the
concentration prior to interruption. For an intraparticle
diffusion controlled sorption process, the concentration
Fig. 7. Illustration of arsenic concentration gradients within a
gradient within the sorbent particle serves as the driving single porous HIX bead for intraparticle diffusion controlled
force and governs the overall rate. With the progress of sorption.
the column run, this concentration gradient attenuates.
The interruption allows the sorbed arsenic to evenly
spread out within the spherical bead. As a result, the A single spherical HIX bead may be viewed as an
concentration gradient, and thus the uptake rate ensemble of tiny HFO agglomerates with an intercon-
immediately after the column restart is greater than nected network of pores. Fig. 7 illustrates solid-phase
the uptake rate prior to the interruption. In other words, arsenic concentration gradients at a given time, t, for a
a faster uptake immediately after the interruption and single HIX particle at radius R. In order to determine
consequent drop in aqueous-phase exit concentration of the intraparticle pore diffusivity for As(V), a finite
the solute is confirmatory evidence in support of volume batch kinetic test was carried out using the set-
intraparticle diffusion as the primary rate-limiting step. up shown in Fig. 3. The stirrer assembly was rotated at
Several previous studies also confirmed that kinetics in 1500 revolutions per minute. It was confirmed indepen-
selective sorption processes is governed by intraparticle dently that the liquid-phase film diffusion was not the
diffusion [35,36]. rate-limiting step under this experimental condition. At
172 M.J. DeMarco et al. / Water Research 37 (2003) 164–176

Fig. 8. Results of batch kinetic test data and computed (solid lines) best-fit effective intraparticle diffusivity: (A) plot of aqueous-phase
As(V) concentration versus time; (B) plot of fractional uptake versus time.

any time, t; the mass balance of As(V) between the liquid

and the exchanger phase gives
V ðCAs;0
CAs;t Þ ¼ mqAs;t ; ð1Þ
where V and m represent the solution volume and the
mass of HIX. CAs;0 ; CAs;t and qAs;t represent initial
aqueous-phase arsenic concentration, arsenic concentra-
tion at time t; and arsenic uptake by HIX at time t;
respectively. At equilibrium, i.e., after infinite time
V ðCAs;0
CAs;N Þ ¼ mqAs;N : ð2Þ
Normalized fractional loading or uptake (F ) by the
exchanger during the batch kinetic test changes with
time and is defined as
qAs;t
F¼ : ð3Þ
qAs;N Fig. 9. Dissolved arsenic concentration profile during regen-
eration with caustic soda.
Fig. 8A shows the decrease in the aqueous-phase As(V)
concentration with time and Fig. 8B shows the F versus
time plot for the batch kinetic test. Pertinent experi-
mental conditions are also included in the same figures. and b can be calculated from its relationship with the
Dissolved arsenic concentration after 72 h was taken as final fractional uptake
the equilibrium value. For finite solution volume system
and appropriate initial and boundary conditions, frac- qAs;N 1
¼ : ð6Þ
tional uptake rates are available in the open literature VCAs;0 1 þ o
[37] and written as
The solid lines in Figs. 8A and B represent the model
qAs;t XN
6oðo þ 1Þexpð
D% eff b2n t=R2 Þ predictions of the kinetic test results and the best-fit
F¼ ¼1
ð4Þ
qAs;N n¼1 9 þ 9o þ b2n o2 effective intraparticle diffusivity, Deff ; was computed to
bn are the non-zero roots of be 5.5  10
11 cm2/s. The magnitude is comparable to
other highly preferred solutes for selective sorbents
3bn
tan bn ¼ ð5Þ studied earlier [35,42] and is representative of high
3 þ ob2n sorption affinity. This effective intraparticle diffusivity
 M.J. DeMarco et al. / Water Research 37 (2003) 164–176
value can be used in modeling arsenic effluent history
during fixed-bed column runs [38,39].
3.3. Regeneration, rinsing and reuse
In order for the sorption process to be viable, the
hybrid exchanger (HIX) has to be amenable to efficient
regeneration and reuse. To this end, a portion of the
exhausted HIX after the lengthy run in Fig. 5 was
regenerated using 10% NaOH. Concentration profiles of
arsenic during the desorption process and other salient
hydrodynamic parameters are provided in Fig. 9. Note
that in less than eight bed volumes, almost the entire
amount of As(V) was completely desorbed from the bed.
Mass balance calculations confirmed that over 90%
arsenic was recovered during the regeneration process.
Similar high efficiency of regeneration was also obtained
for HIX column used for As(III) removal. The observa-
tion that arsenic desorption was completed in less
than ten bed volumes demonstrates that sorption
sites of HFO microparticles are easily accessible
through the network of pores, i.e., pore blockage
and consequent increase in the tortuosity of dissolved
solutes did not result from the dispersion of submicron
HFO particles within the porous beads. At high alkaline
pH, HFO sorption sites are all deprotonated and
negatively charged; so are all arsenite and arsenate
species. Donnan co-ion exclusion effect is thus very
predominant under this condition resulting in efficient
desorption.
173
Following regeneration, the bed was reconditioned
using carbon dioxide sparged treated water to bring
down pH from alkaline to near-neutral condition.
Fig. 10 shows pH versus bed volume plots for three
successive rinsing processes following separate column
regenerations. In less than eight bed volumes, pH was
lowered to neutral value. During sodium hydroxide
regeneration for approximately 15 bed volumes, dis-
solved iron concentration in the spent regenerant ranged
from 0.6-0.8 mg/L. However, total iron content of HIX
remained unaltered due to dissolution of such trace
amounts. Following rinsing, HIX bed was essentially
ready for the next sorption cycle and no pH adjustment
is further necessary. Regeneration with caustic soda and
rinsing with CO2-sparged water essentially involve
deprotonation and protonation of the surface sorption
sites of HFO particles as shown below.
The modified rinsing process avoided use of strong
hydrochloric acid or sulfuric acid altogether in the
process. During the 3-year period of lengthy laboratory
investigation, the same HIX particles were used repeat-
edly following sorption, desorption and rinsing steps; no
noticeable deterioration was observed. In order to
further validate that HIX retains its arsenic removal
capacity over long-term usage, separate mini-column
tests at two different pHs were carried out for two
successive cycles using the same HIX particles. Fig. 11
shows the results of the study; note that the As(V)
removal capacity remained practically unchanged be-
tween these two successive cycles.
174 M.J. DeMarco et al. / Water Research 37 (2003) 164–176
Fig. 10. Change in effluent pH during rinsing with CO2 sparged
treated water following regeneration.
Fig. 11. Arsenic removal capacities of HIX during two
successive cycles at two different pHs.
3.4. Field experience with HIX and comparison with
activated alumina (AA)
In order to examine the performance of HIX under
field conditions, two separate mini-column runs were
carried out in parallel using arsenic-contaminated
groundwater near a village bordering Bangladesh and
India. One column contained HIX while the other
contained granular activated alumina; all other condi-
tions including influent water composition remained
identical. The diameter of each column was 12 mm and
the bed height equal to 40 mm. The feedwater was
replaced every week and analyzed for total arsenic and
total iron. The average values of pH, alkalinity and
dissolved silica in the influent were 7.2, 180 mg/L as
CaCO3 and 10 mg/L as SiO2, respectively. The AA in the
second column is commercially available locally (Grade
AS-37, Oxide India Ltd., Durgapur, India) and cur-
rently being used as an arsenic-selective sorbent in over
two dozen well-head units in the region to treat arsenic-
contaminated groundwaters [40,41]. These lengthy
column runs were carried out through an active
collaboration between the Bengal Engineering College,
India, and Lehigh University under the auspices of
Water For People (WFP).
Total arsenic and dissolved iron concentration in the
influent varied significantly from week to week, as
depicted in Fig. 12. Nearly 50% removal of dissolved
iron was observed for both HIX and AA columns.
Fig. 12 provides a detailed comparison of arsenic
effluent histories between the two column runs. Note
that for nearly 5000 bed volumes, both of them removed
arsenic to a concentration less than 50 mg/L, the current
MCL for drinking water in Bangladesh and India. The
performance of HIX was marginally better than AA. It
is possible that the dissolved iron in groundwater might
have contributed toward a small percentage of arsenic
removal during the lengthy runs. Nevertheless, the
effectiveness of HIX, especially when compared with
activated alumina, is quite apparent from these column
runs. For both HIX and AA columns, there was an early
arsenic breakthrough caused primarily by the passage of
colloidal iron oxide particles containing sorbed arsenic
species.
4. Conclusions
In this study, an extensive laboratory investigation
was carried out to evaluate the performance of a new
reusable polymeric/inorganic sorbent material for selec-
tive arsenic removal. The new sorbent essentially
comprises spherical polymer beads inside which agglom-
erated microparticles of hydrated Fe(III) oxides have
been dispersed in gel and pore phases of the beads. The
sulfonic acid functional groups of the cation exchanger
beads allowed high (9–12% by mass) and fairly uniform
Fe(III) loading. While HFO microparticles offered high
sorption affinity toward dissolved arsenic species,
spherical polymer beads provided excellent hydraulic
characteristics in fixed beds. Mechanical agitation and
repeated column usage did not result in any loss of HFO
particles from the beads. Major conclusions can be
summarized as follows:
* New hybrid sorbent, HIX, is very selective to both
arsenites and arsenates and exhibits excellent arsenic
removal without any pH adjustment or pre-oxidation
of the influent.
* Besides the absence of arsenic, treated water quality
with respect to other anions and pH remain
essentially the same as the influent.
* Spherical HIX beads are extremely robust, durable
and ideally suited for fixed-bed column operation.
 M.J. DeMarco et al. / Water Research 37 (2003) 164–176
Fig. 12. Comparison of effluent histories of HIX and activated alumina column runs with contaminated groundwater in Indian
subcontinent.
During lengthy column runs and use of the same
materials for multiple number of cycles, no genera-
tion of fines or any significant increase in pressure
drop across the column is observed.
* Like other selective sorption processes, intraparticle
diffusion is the primary rate-limiting step for
dissolved arsenic removal by HIX particles. The
experimentally determined intraparticle diffusivity
(5.5  10
11 cm2/s) under representative conditions
is comparable to those observed for other selective
sorption processes.
* Following sorption, fixed-bed HIX columns can be
efficiently regenerated with 10% NaOH in less than
ten bed volumes. Subsequent rinsing with carbon
dioxide sparged treated water for less than ten bed
volumes is adequate to bring the column ready for
the sorption cycle.
* Total iron content of HIX following the complete
cycle of removal, regeneration and rinsing remained
practically unchanged. No noticeable change in
HIX’s arsenic removal capacity is observed during
successive cycles.
* Last but not the least, polymeric cation exchanger is
one of the most inexpensive, durable and widely
available sorbents throughout the world. Use of this
material as the primary substrate is likely to make
HIX costwise competitive with other arsenic removal
sorbents already available commercially.
Acknowledgements
Authors are thankful to A. Gupta of the Bengal
Engineering College, Howrah, India for his assistance in
carrying out the column run study in Fig. 12. The study
received partial financial support from the United States
175
Environmental Protection Agency through Grant No.
G9R11010.
References
[1] American WaterWorks Association (AWWA) ‘‘Arsenic
Rule.’’ Mainstream. Vol. 45, No. 2, February 2001.
[2] Bagla P, Kaiser J. India’s spreading health crisis draws
global arsenic experts. Science 1996;274:174–5.
[3] Nickson R, et al. Arsenic poisoning of Bangladesh
groundwater. Nature 1998;395:338.
[4] Bearak D. New Bangladesh disaster: wells that pump
poison. The NewYork Times, November 10, 1998.
[5] Edwards M. Chemistry of arsenic removal during coagula-
tion and Fe–Mn oxidation. J Am Water Works Assoc
1994;76:64–78.
[6] Ferguson JF, Gavis J. A review of arsenic cycle in natural
waters. Water Res 1972;6:1259–74.
[7] Safiullah S, et al. Proceedings of the International
Conference on Arsenic in Bangladesh (Dhaka), February
1998, pp. 8–12
[8] Hering JG, Chiu van Q. Arsenic occurrence and speciation
in municipal ground-water-based supply system. J Environ
Eng ASCE 2000;126(5):471–4.
[9] Vahter M, Envall J. In vivo reduction of arsenate in mice
and rabbits. Environ Res 1983;32:14–24.
[10] Smith AH. Personal communication, 1999.
[11] Ghosh MM, Yuan JR. Adsorption of inorganic arsenic
and organioarsenicals on hydrous oxides. Environ Prog
1987;3(3):150–7.
[12] Driehaus W, Jekel M, Hildebrandt U. Granular ferric
hydroxide—a new adsorbent for the removal of arsenic
from natural water. J Water SRT Aqua 1998;47(1):30–5.
[13] McNeil LS, Edwards M. Soluble arsenic removal at water
treatment plants. J Am Water Works Assoc 1995;87(4):
105–14.
[14] Min JM, Hering J. Arsenate sorption by Fe(III)-doped
alginate gels. Water Res 1998;32(5):1544–52.
176 M.J. DeMarco et al. / Water Research 37 (2003) 164–176
[15] Ramana A, SenGupta AK. Removing selenium (IV)
and arsenic(V) oxyanions with tailored chelating
polymers. J Environ Eng Div ASCE 1992;118(5):
755–75.
[16] Sorg TJ, Logsdon GS. Treatment technology to meet
the interim primary drinking water regulations for
inorganics: part 2. J Am Water Works Assoc 1978;70(7):
379.
[17] Lackovic JA, Nikolaidis NP, Dobbs GM. Inorganic
arsenic removal by zero-valent iron. Environ Eng Sci
2000;17(1):29–39.
[18] Chwirka JD, Thomson BM, Stomp JM. Removing arsenic
from groundwater. J Am WaterWorks Assoc 2000;92(3):
79–88.
[19] Clifford D, Subramonian S, Sorg T. Removing dissolved
inorganic contaminants from water. Environ Sci Technol
1986;20:1072–80.
[20] Driehaus W, Seith R, Jekel M. Oxidation of arsenic(III)
with manganese oxides in water treatment. Water Res
1995;29(1):297–305.
[21] Kartinen EO, Martin CJ. An overview of arsenic removal
processes. Desalination 1995;103:79–88.
[22] Meng X, Bang S, Korfiatis GP. Effects of silicate, sulfate
and carbonate on arsenic removal by ferric chloride. Water
Res 2000;34(4):1255–61.
[23] Shen YS. Study of arsenic removal from drinking water. J
Am Water Works Assoc 1973;65(8):543.
[24] Torrens KD. Evaluating arsenic removal technologies.
Pollution Eng 1999; 25–8.
[25] Vagliasandi FGA, Benjamin MM. Arsenic removal in
fresh and NOM-preloaded ion exchange packed bed
adsorption reactors. Water Sci Technol 1998;38(6):337–43.
[26] Bajpai S, Chaudhury M. Removal of arsenic from
manganese dioxide coated sand. J Environ Eng ASCE
1999;125(8):782–4.
[27] Frank P, Clifford D. As(III) oxidation and removal from
drinking water. EPA Project Summary, Report No. EPA/
600/S2-86/021, Water Engineering Research Laboratory,
Environment Protection Agency, Office of Research and
Development, Cincinnati, OH, 1986.
[28] Manning B, Goldberg S. Adsorption and stability of
arsenic(III) at the clay mineral-water interface. Environ Sci
Technol 1997;31:2005–11.
[29] Manning BA, Fendorf SE, Goldberg S. Surface structures
and stability of As(III) on goethite: spectroscopic evidence
for inner-sphere complexes. Environ Sci Technol 1998;
32(16;):2383–8.
[30] SenGupta AK, Lim L. Modeling chromate ion-exchange
processes. AIChE J 1988;34:2019.
[31] Zhao D, SenGupta AK. Ultimate removal of phosphate
using a new class of anion exchanger. Water Res
1998;32(5):1613–25.
[32] Helfferich F. Ion exchange. New York: McGraw-Hill Co.,
1963.
[33] Ficklin W. Separation of As(III) and As(V) in ground-
waters by ion exchange. Talanta 1983;30(5):371.
[34] Clifford D, Ceber L, Chow S. As(III)/As(V) separation by
chlorid-form by ion-exchange resins. Proceedings of the
AWWA WQTC, Norfolk, VA, 1983.
[35] Li P, SenGupta AK. Intraparticle diffusion during
selective ion exchange with a macroporous exchanger.
React Func Polym 2000;44:273–87.
[36] Sontheimer H, Crittenden JC, Summers RS. Activated
carbon for water treatment, DVGW-Forschungstelle, 1988.
[37] Crank J. The mathematics of diffusion, 2nd ed.. Oxford,
UK: Oxford Science Publications, 1975.
[38] Suzuki M. Adsorption engineering. Amsterdam, NL:
Elsevier Science, 1990.
[39] Zhao D. Ph.D. dissertation. Department of Civil &
Environmental Engineering, Lehigh University, Bethle-
hem, PA, USA, 1997.
[40] Murphy M. Helping people help themselves. Am Water
Works Assoc J 2000;92(4):139–48.
[41] SenGupta AK, Gupta A, Deb A. Arsenic crisis in Indian
subcontinent. Water 21: IWA Magazine. December, 2001;
34–37.
[42] Li P, SenGupta AK. Intraparticle diffusion during
selective sorption of trace contaminants: the effect of gel
versus macroporous morphology. Environ Sci Technol
2000;34:5193–200.