You are on page 1of 9

Wat. Res. Vol. 35, No. 8, pp.

20492057, 2001
# 2001 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
0043-1354/01/$ - see front matter PII: S0043-1354(00)00467-X
ADSORPTION OF ARSENITE AND ARSENATE WITHIN
ACTIVATED ALUMINA GRAINS: EQUILIBRIUM AND
KINETICS
TSAIR-FUH LIN* and JUN-KUN WU
Department of Environmental Engineering, National Cheng Kung University, Tainan City 70101,
Taiwan, ROC
(First received 29 November 1999; accepted in revised form 13 September 2000)
Abstract}Equilibrium and kinetic adsorption of tri-valent (arsenite) and penta-valent (arsenate) arsenic
to activated alumina is elucidated. The properties of activated alumina, including porosity, specic surface
area, and skeleton density were rst measured. A batch reactor with temperature control was employed to
determine both adsorption capacity and adsorption kinetics for arsenite and arsenate to activated-alumina
grains. The Freundlich and Langmuir isotherm equations were then used to describe the partitioning
behavior for the system at dierent pH. A pore diusion model, coupled with the observed Freundlich or
Langmuir isotherm equations, was used to interpret an observed experimental adsorption kinetic curve for
arsenite at one specic condition. The model was found to t with the experimental data fairly well, and
pore diusion coecients can be extracted. The model, incorporated with the interpreted pore diusion
coecient, was then employed to predict the experimental data for arsenite and arsenate at various
conditions, including dierent initial arsenic concentrations, grain sizes of activated alumina, and system
pHs. The model predictions were found to describe the experimental data fairly well, even though the
tested conditions substantially diered from one another. The agreement among the models and
experimental data indicated that the adsorption and diusion of arsenate and arsenite can be simulated by
the proposed model. # 2001 Elsevier Science Ltd. All rights reserved
Key words}activated alumina, adsorption, arsenate, arsenic, arsenite, diusion
INTRODUCTION
The chronic toxicity of arsenic from drinking water
has been well documented (WHO, 1996), including
blackfoot disease observed in the south western coast
of Taiwan in 19301950 (Chen et al., 1994). Much of
the ground water in both north eastern and south-
western coast area of Taiwan contains high concen-
tration of arsenic. Although all the major drinking
water treatment plants in the area use surface water
as their source water, ground water with less arsenic
content, typically 0.020.03 mg/L and up to 0.3 mg/L,
in the area is still used in many small water treatment
plants (Lin, 1998, 1999). In all the treatment plants
using ground water, traditional treatment processes,
mostly aeration, pre-chlorination, sedimentation,
ltration, and chlorination are used. Field samples
showed that 070% of arsenic removal (typically less
than 1020%) were found in those plants (Lin, 1998).
Under this kind of treatment eciency, the nished
water would not be able to comply with the proposed
National Drinking Water Quality Standard of
Taiwan for arsenic, 0.01 mg/L, in December 2000
(ROCEPA, 1998). Therefore, better treatment pro-
cesses, such as enhanced coagulation, activated
alumina (AA) adsorption and membrane technolo-
gies are needed to reduce arsenic concentration in the
nished water.
Among the treatment processes appropriate for the
removal of arsenic, activated-alumina adsorption is
considered to be less expensive than the membrane
separation, and is more versatile than the ion
exchange process (Chen et al., 1999; Frey et al.,
2000). Activated alumina has strong selectivity to
arsenate ion (Cliord, 1990). This technique was used
in point-of-use water treatment devices (Fox and
Sorg, 1987; Fox, 1989; Hathaway and Rubel, 1987;
Cliord, 1990) and is a potential process in small
water treatment plants (Chen et al., 1999; Frey et al.,
2000) for the removal of arsenic. Engineering
estimation for the adsorption capacity of arsenate
and empty bed contact time (EBCT) for column
design have been suggested (Cliord, 1990).
Although equilibrium capacity and column break-
through behavior of arsenate and arsenite in AA
were studied thoroughly, the kinetics of arsenate and
arsenite within AA grains was investigated to a much
less extent.
*Author to whom all correspondence should be addressed.
Tel.: +886-6-236-4455; fax: +886-6-275-2790; e-mail:
tin@mail.ncku.edu.tw
2049
Cliord (1990) pointed out that the alumina
adsorption kinetics is slow, and 2 days were needed
for the adsorption of arsenate onto 2848 mesh AA
grains to reach half of equilibrium. However, the
time required for surface adsorption of arsenate and
arsenite onto metal hydroxides is known to be
relatively fast. For example, only a few hours were
needed to reach equilibrium for the adsorption of
arsenate and arsenite onto ferrihydrite (Raven et al.,
1998), and less than 1 h was needed for more than
90% of arsenate adsorption onto amorphous alumi-
num hydroxide (Anderson et al., 1976). Fuller et al.
(1993) suggested that the slow kinetics may be
attributed to the diusion of arsenic species into the
hydroxide aggregate spheres. In a column study,
Cliord and Lin (1991) indicated that the break-
through curves for column tests of dierent mesh size
of AA had a dramatic dierence. The coarse mesh
grade AA only treated 2/3 of bed volumes of water
before reaching 50 mg/L in the euent compared to
that of the ne grade one. It is very likely that a
broader breakthrough curve (or a longer mass
transfer zone) for the coarse grade AA grains was
caused by the longer transport path (larger grain
diameter) for arsenate ions. As a result of that, the
adsorption capacity at 50 mg/L of euent As
concentration is smaller for the coarse grade AA
grains.
Although diusion of arsenic species within metal
hydroxide grains has been linked to the slow
adsorption, very little information related to the
description of both adsorption equilibrium and
kinetics of arsenate and arsenite within activated
alumina grains is present. A clear understanding of
adsorption kinetics of arsenic within AA grains will
ensure a better prediction of transport behavior of
arsenic in columns. In this study, equilibrium and
kinetic adsorption of tri-valent (arsenite) and penta-
valent (arsenate) arsenic to activated alumina are
elucidated. Experiments were conducted to determine
AA properties, adsorption capacity and adsorption
kinetics for arsenite and arsenate to AA grains, and
models were used to interpret the observed experi-
mental adsorption kinetic curves at various condi-
tions. Based on the results, pore diusion coecients
and tortuosity factors of arsenate and arsenite within
AA grains were interpreted.
THEORY
The diusion and adsorption model employed in
this study is modied from an earlier eort developed
for gas-phase transport of volatile organic com-
pounds (VOCs) within activated carbon grains by
Lin et al. (1996). In the model, activated-alumina
(AA) grains are considered as spherical and porous,
with constant diameter and with adsorption sites
uniformly distributed throughout the grains. The
arsenic and arsenate ions are assumed to be
transported from bulk aqueous phase to the internal
surface of AA grains by diusion, and only two mass
transfer processes, external lm diusion and intra-
granular pore diusion, are considered as rate limited
steps. At any time, local equilibrium between
aqueous and adsorbed phases is maintained within
the pores.
Considering the mass balance of arsenic and
arsenate ions, a transient equation describing trans-
port of the ions within AA grains (equation (1)) and
the initial and boundary conditions in a batch
adsorption system (equation (2)) can be expressed as
e
p
@C
@t
1 e
p
r
p
@q
@t

e
p
r
2
@
@r
r
2
D
p
@C
@r

1
where e
p
is the grain porosity (g/cm
3
), r
p
the skeleton
density of the AA grains (g/cm
3
), r the radial
coordinate of the grain (cm), D
p
the pore diusivity
of solutes through the intragranular pore space (cm
2
/
s), C the aqueous concentration within the pore
(g/cm
3
), q the solid phase concentration of arsenic
(g/g), and t time scale (s);
C0 r a; t 0 0
@C
@r
r 0; t 0
e
p
D
p ra
j
@C
@r
k
f
C
b
C
s

2
where k
f
is the mass transfer rate constant at the
external water lm (cm/s), C
b
the concentration in
bulk aqueous phase (g/cm
3
), C
s
the concentration at
the external surface of AA grain (g/cm
3
), and a the
radius of AA grain (cm).
In the model, q is assumed to be in equilibrium
with C in the adjacent pore water with the isotherm
being the same as the one obtained from a bulk
experiment. The equations often used to describe the
relation between q and C include Langmuir isotherm
(equation (3)) and Freundlich isotherms (equation
(4)).
q
q
m
bC
1 q
m
C
3
where q
m
is the adsorption capacity of monolayer
(g/g), and b the empirical constant (cm
3
/g);
q kC
1=n
4
where k is the empirical constant ((g/cm
3
)/(g/cm
3
)
n
),
and n the empirical constant (dimensionless).
Considering together with the mass balance of
arsenate or arsenite in the system, equation (1) may
be combined with equation (2) and either equations
(3) or (4), for solving C(r) and C
b
at dierent time
using appropriate numerical methods. The numerical
scheme employed in this study is modied from a
numerical code, BATCH, developed by Tien (1994).
Tsair-Fuh Lin and Jun-Kun Wu 2050
More detailed information of the code can be found
in Tien (1994).
MATERIALS AND METHODS
Preparation of activated alumina
Granular activated alumina used in this study was
amorphous and is commercially available (Macherey-Nagel,
Germany). Before experiments, activated alumina samples
were oven-dried at 105 58C for 24 h, and then stored in a
desiccator for further analysis and experiments. To better
simulate the experimental results, three narrow size ranges
of sample were used in all the experiments, including 80100
mesh (0.1770.149 mm), 140150 mesh (0.1050.104 mm),
and 200230 mesh (0.0740.062 mm).
Properties of activated alumina
Specic surface was measured, as it can strongly inuence
arsenic-sorption behavior. Skeleton density and porosity
were also determined as these were required to interpret the
kinetic data. Surface area was measured by low-temperature
nitrogen adsorption and the data were interpreted using the
BET equation (Brunauer et al., 1938). The nitrogen
adsorption tests were carried out in a Micrometritics
Surface Area Analyzer at 77 K (Flow Sorb II 2300,
Micrometritics, USA). Skeleton density was measured based
on the water displacement method proposed in Black
(1965). Although the method was designed for soil
originally, experimental results indicated that the method
produced almost equivalent results as those from helium
pycnometer for sandy aquifer materials (Ball et al., 1990). In
estimating the porosity of AA, about 100 oven-dried AA
grains for each size range were rst measured for their
weight and number. Since all the three size ranges were so
narrow and the grains look very spherical, grain volume for
each size range may be estimated by assuming that all the
activated alumina grains are spherical with identical
diameter equal to the geometric mean for each size range.
The porosity (e
p
) of an AA grain can then be interpreted
using the equation
e
p
1
r
s
r
p
5
where r
p
is the skeleton density of AA grain (g/cm
3
) and r
s
the grain density of AA grain (g/cm
3
).
Adsorption experiments
Na
2
HAsO
4
7H
2
O (KR Grade, Aldrich, USA) and
NaAsO
2
(GR Grade, Sigma, USA) were used to represent
arsenate and arsenite, respectively. The equilibrium con-
centrations for both anionic species were controlled at 0.02
12 mg-total arsenic/L. Adsorption experiments of arsenite
and arsenate to activated alumina were conducted using 0.1-
L polyethylene (PE) vials placed within a temperature-
controlled reciprocating shaker (TLC-10 and SB-70, Wis-
dom Corp., Taiwan). In each experiment, the solution
prepared at a predetermined arsenite or arsenate concentra-
tion using de-ion water was rst lled into the vial. About
0.10.5 g of oven-dried activated-alumina grains were then
placed into the vial. The pH in each vial was adjusted using
hydrochloric acid (GR grade, Merck, Germany) and
sodium hydroxide (GR Grade, Ridel-de Haen, Germany)
within 0.1 of predetermined pH. In the system, the
temperature was controlled at 25 0.58C, and the recipro-
cating speed was at 150 rpm for all experiments. The
changes of pH values through all the experiments conducted
were monitored for each run, and were all found to be
within 0.15 of the initial values. Preliminary experiments
indicated the equilibrium of adsorption can be established
within 40 and 170 h for arsenite and arsenate systems,
respectively. During the experiment, one of the vials was
taken to measure the residual aqueous concentration of
arsenic at a predetermined time. Through a mass balance
calculation, the adsorbed amount of arsenite and arsenate
on the activated alumina was then calculated. Based on the
standard method suggested by APHA et al. (1996), an
atomic absorption spectrometer (5100, Perkin Elmer, USA)
equipped with a manual hydride generator (MSH-10, Perkin
Elmer, USA) was employed to determined the arsenic
concentration in the solution. All the samples were ltrated
using 0.45 mm lters (Advantec MFS, Japan) before
analysis. The detection limit for this study was 0.3 mg/L of
arsenic, while analysis of the duplicates found was all within
5% of errors. To assure that the interaction between the
PE vials and arsenic species is negligible and the sampling
procedure is appropriate, blank experiments were also
conducted. Both arsenite and arsenate remained nearly
constant (98100% of original concentrations) during the
time scale of this study, suggesting that the interaction of the
two arsenic species to PE vials can be neglected.
RESULTS AND DISCUSSION
Properties of activated alumina
The properties of AA used in this study were listed
in Table 1 for the specic surface area, skeleton
density and interpreted porosity. It is clearly to see
that both specic surface area and skeleton density of
AA for all the three size ranges are almost identical.
However, the interpreted porosity for the three size
ranges were dierent, especially for the smallest size
grains. The AA grains at smallest size were found to
have only 17% of intragranular pores compared to
2629% for the other two samples, even though the
surface area for the three samples are similar.
Although pore size distribution of the samples was
not measured, it is reasonable to assume that the
intragranular pores of the 200230 mesh AA grains
are smallest compared to the other two samples.
Equilibrium of adsorption
Equilibrium adsorption of arsenate and arsenite
were each conducted at ve or six dierent pHs,
Table 1. Properties of activated alumina
Size (mesh) 80100 140150 200230
Size range (mm) 0.1770.149 0.1050.104 0.0740.062
Geometric mean (mm) 0.162 0.1045 0.0677
Skeleton density (kg/m
3
) 3580 3580 3580
Specic area (m
2
/g) 118 115 116
Porosity (%) 29.3 25.7 17.2
Adsorption of arsenite on activated alumina 2051
ranging from 2.5 to 12. At each pH condition, four
initial concentrations, between 0.79 and 4.90 mg/L
for arsenite and between 2.85 and 11.5 mg/L for
arsenate, were controlled to obtain adsorption
isotherms. The adsorption data were tted with
Freundlich and Langmuir isotherm equations and
are summarized in Table 2. As shown in the table, all
the nonlinear regression coecients (R
2
) for dierent
conditions were larger than 0.93, indicating that both
Langmuir and Freundlich isotherms successfully
describe the partition behavior between water and
AA surface for arsenite and arsenate.
The uptake of arsenite and arsenate in AA is
illustrated in Figs 1 and 2. It is clear to see that the
uptake of arsenite is much less than that of arsenate
for AA in most pH conditions. Under most pH
conditions, arsenate is present in negatively ionic
form and arsenite is in non-ionic form. Since point of
zero charge (pH
pzc
) for dierent type of alumina is
around 8.49.1 (Stumm and Morgan, 1981; Bowers
and Huang, 1985), surface of AA is positively
charged until pH5pH
pzc
. The anionic species,
arsenate, would thus have stronger interaction
(specic binding) with AA surface and have higher
uptake. The results are similar to column studies of
arsenic removal from ground water by Cliord
(1990), who concluded that breakthrough time of
As(V) in AA column is much longer than that of
As(III). This is an important reason to oxidize
As(III) before applying AA to remove arsenic from
ground water.
The uptake of arsenite (Fig. 1(a)) increases as pH
increases until pH7, and then decreases as pH
increases, similar to that reported by Yuan et al.
(1987). Under the condition that pH is far less than
9.2, the non-ionic H
3
AsO
3
is the dominant species
(Xu et al., 1991), and van der waal force between the
solute and the alumina surface is expected. As pH
increases toward 7, the anionic species, H
2
AsO
3

,
would increase and thus more arsenic uptake to
alumina is expected due to more specic binding. At
pH larger than 9, the uptake of arsenite drops rapidly
due to the repulsion from alumina surface.
Adsorption of arsenate to activated alumina (Fig.
(2a)) shows a dierent pattern with that of arsenite.
Table 2. Isotherm parameters for arsenate and arsenite onto activated alumina
Langmuir isotherm equation Freundlich isotherm equation
pH
a
q
m
(mg/g) b (L/mg) R
2
K ((mg/cm
3
)/(mg/cm
3
)
-n
) n (dimensionless) R
2
Arsenite
3.3 0.77 2.29 0.987 0.47 2.18 0.971
4.9 1.42 1.44 0.985 0.85 1.55 0.990
6.1 1.69 2.92 0.985 1.68 1.48 0.986
6.9 3.48 0.59 0.989 1.21 1.53 0.998
9.2 1.49 1.53 0.961 0.93 1.64 0.983
11.9 1.25 0.16 0.991 0.17 1.24 0.992
Arsenate
2.6 12.34 49.59 0.968 18.50 3.3 0.936
5.2 15.90 10.03 0.970 22.77 2.05 0.959
7.2 9.93 8.25 0.996 8.36 3.69 0.980
9.5 4.90 0.25 0.933 1.07 1.97 0.948
11.9 0.48 0.39 0.992 0.18 3.07 0.993
a
The pHs were all controlled within 0.1 of the reported values.
Fig. 1. Adsorption of arsenite onto activated alumina, (a)
adsorption capacity at dierent initial concentrations and
dierent pH, (b) dissociation of arsenite at dierent pH (Xu
et al., 1991) and (c) surface charge of g-alumina (Bowers
and Huang, 1985).
Tsair-Fuh Lin and Jun-Kun Wu 2052
Uptake of arsenate to activated alumina remains
almost constant for pH less than 6, and then dropped
signicantly for higher pH conditions. For pH less
than 6, the surface of AA is predominantly positively
charged and the major arsenic species is H
2
AsO
4

(Xu
et al., 1991). Therefore, specic binding is expected
for the adsorption process. As pH increases, the
portion of positively charged surface sites on AA
decreases, causing the reduction of adsorption
uptake.
Transport within activated-alumina grains
Typical adsorption kinetic data for arsenite onto
80100 mesh activated alumina grains are shown in
Fig. 3. About 40 h were needed to reach equilibrium.
Similar results for arsenite and arsenate to dierent
size of AA grains were also observed, except that the
time for arsenate to reach equilibrium is about 170 h.
This time scale is similar to that reported by Cliord
(1990), who found that 2 days were needed for the
adsorption of arsenate onto 2848 mesh of AA grains
to reach half of equilibrium.
The pore diusion model proposed was used to
interpret the data from the adsorption kinetic
experiments. In the models, either Langmuir (equa-
tion (3)) or Freundlich (equation (4)) isotherm was
incorporated into the diusion model to describe the
partition behavior of arsenite and arsenate between
pore water and intragranular pore surface of AA
grains. The equilibrium constants for the model input
were assumed to be the same as extracted from the
bulk-phase equilibrium experiments shown in Table
2. It is noted that only one adjustable parameter,
pore diusion coecient (D
p
), was used to t the
models to the experimental data, while no adjustable
parameter was used in the following predictions.
Besides the equilibrium constants and D
p
, all other
input parameters were either obtained from experi-
mental measurement (see Table 1), including porosity
(e
p
), skeleton density (r
p
), and grain diameter (a) or
from theoretical calculation (mass transfer rate
constant at the external water lm). The mass
transfer rate constant at the external water lm, k
f
,
is estimated by the RanzMarshall correlation for
single particles modied by Wakao and Funazkri
(1978) as
Sh 2:0 0:6Sc
1=3
Re
1=2
6
where Sh=a k
f
/D
b
is the Sherwood number (dimen-
sionless), Sc=v/D
b
is the Schmidt number (dimen-
sionless), Re=ua/v is the particle Reynold number
(dimensionless), D
b
is the diusion coecient of
arsenate or arsenite solute in water (cm
2
/s), u is the
velocity of water ow (cm/s), and v is the kinematic
viscosity (cm
2
/s).
The model was rst tted to the adsorption kinetic
data of arsenite onto 80100 mesh AA grains at
pH=8.8 and initial concentration=5.6 mg/L. As
illustrated in Fig. 3, the experimental data is well
described by the diusion model incorporated with
either Langmuir or Freundlich isotherm. The ex-
tracted pore-diusion coecients, D
p
, for the models
were found to be 1 10
7
and 2 10
7
cm
2
/s for
Fig. 2. Adsorption of arsenate onto activated alumina, (a)
adsorption capacity at dierent initial concentrations and
dierent pH, (b) dissociation of arsenate at dierent pH (Xu
et al., 1991) and (c) surface charge of g-alumina (Bowers
and Huang, 1985).
Fig. 3. Adsorption kinetic data and tted models of arsenite
(initial concentration=5.6 mg/L, pH=8.8) onto 80100
mesh activated alumina.
Adsorption of arsenite on activated alumina 2053
Langmuir and Freundlich isotherm cases, respec-
tively. Due to tortuous pathway of diusion and pore
constrictions, this D
p
is certainly smaller than
diusion coecient of solutes in aqueous system,
which is around 1 10
5
cm
2
/s.
The extracted pore diusion coecients were
further used to predict the adsorption kinetics for
various conditions, including dierent initial solute
concentrations, pH, and solutes (arsenate and
arsenite), and dierent sizes of AA grains. Figure 4
illustrates comparison of model predictions and
experimental data for dierent initial arsenite con-
centrations at two AA grain sizes. Although slight
over-estimation at initial stage, the models capture
the main feature of the experimental data under
dierent initial concentrations for both 80100 and
140150 mesh of AA grains (Fig. 4(a) and (b)).
However, when attempting to predict the adsorption
of arsenite to 200230 mesh of AA grains, the models
failed to describe the experimental data. The model
was then adjusted for the diusion coecient to nd
the best ts to the experimental data. As shown in
Fig. 5(a), the tted models, including both Langmuir
and Freundlich isotherm cases, reasonably describe
the experimental adsorption kinetics for the 200230
mesh case, and the best tted D
p
is 5 10
8
cm
2
/s.
This tted D
p
was also used to predict the kinetic
data for another initial arsenite concentration
(Fig. 5(b)). A similar degree of t was found for the
model predictions and experimental data. Compared
to the previous cases (80100 and 140150 mesh
cases), D
p
for the 200230 cases is only
1
4
of their
values, implying that a dierent pattern of pore
structure may be present for dierent sizes. This
dierence will be discussed in the Tortuosity section.
To extend the applicability of the model to another
arsenic species, the model was used to predict for the
adsorption of arsenate onto dierent sizes of AA
grains. It is expected that the diusion coecient for
arsenite in water is very close to that for arsenate.
Therefore, the extracted D
p
for arsenite was applied
for the model predictions for arsenate. It is clear to
see in Fig. 6 that the models well predict the
experimental data for arsenate/activated alumina
systems 80100 mesh case (Fig. 6(a)) when incorpo-
rated with both Freundlich and Langmuir isotherms.
In the 200230 mesh case, the models capture the
feature of the experimental data, although the
Langmuir isotherm case produced better prediction
than the Freundlich isotherm case. The discrepancy
between two model predictions may be attributed to
that the Langmuir isotherm has better description of
the partition behavior than Freundlich isotherm for
this specic condition. These ndings reinforce that
the model appropriately describes the transport and
adsorption processes of arsenic species within AA
grains.
Fig. 4. Adsorption kinetic data and model predictions of
arsenite onto (a) 80100 mesh-activated alumina (initial
concentration=1.1 mg/L, pH=8.8); (b) 140150 mesh-
activated alumina (initial concentration=3.1 mg/L,
pH=8.8).
Fig. 5. Adsorption kinetics of arsenite onto 200230 mesh-
activated alumina: (a) tted models and experimental data
(initial concentration=7.3 mg/L, pH=8.8); (b) model
predictions and experimental data (initial concentra-
tion=5.0 mg/L, pH=6.9).
Tsair-Fuh Lin and Jun-Kun Wu 2054
Tortuosity
The tortuosity factor, t, represents the ease of
solute diusion within the pore, and is dened as
t D
b
=D
p
7
A tortuous path and pore constrictions may reduce
the diusion ux and cause t to be greater than unity,
while pore inter-connections may increase the ux
(Wang and Smith, 1983). As a result of decreasing
likelihood for inter-connections between pores, tor-
tuosity was observed to increase rapidly as the
porosity decrease (Kim and Smith, 1974). Generally,
t values of between 2 and 6 are found for zeolites
(Ruthven, 1984), while values between as high as 10
100 have been found for low-porosity soils
(e
p
=0.0140.15) (Lin et al., 1994) and nickel oxide
grains (e
p
=0.03) (Kim and Smith, 1974). When the
solute molecular size is signicant with respect to the
pore size, restricted diusion needs to be considered,
and equation (7) can be modied as (Sattereld et al.,
1973)
t=K
r
D
b
=D
p
8
where K
r
is a constrictivity factor (1).
An empirical correlation presented by Chantong
and Massoth (1983) to describe the restrictive eect is
in agreement with experimental results from a few
studies and is shown below:
K
r
1:03 exp4:5l 9
where l is the ratio of critical molecular diameter
(Sattereld et al., 1973) to pore diameter.
Although the correlation was developed based on
the measured diusion of polyaromatic compounds
in alumina with l between 0.04 and 0.4 (Chantong
and Massoth, 1983), the correlation should be able to
provide a useful insight for the diusion of arsenic in
this study. Since pore diameter of activated alumina
is not measured in this study, a simple estimation
proposed by Sattereld (1981) is employed to
quantify the mean pore diameter, d
p
(m).
d
p

4e
p
A
s
r
s
10
where A
s
is the specic surface area of AA grains
(m
2
/g).
According to equation (10), the mean-pore dia-
meters for the AA samples were around 3.4 nm for
the 80100 mesh grains, and 1.6 nm for the 200230
mesh grains. These values are close to that measured
by Rosenblum and Cliord (1983) for 2048 mesh
Alcoa F-1 activated alumina, which is 1.6 nm. Since
the diameter of H
2
AsO
4

ion is around 0.8 nm (Bodek


et al., 1988), based on equation (10) the constrictivity
factors are estimated to be 0.36 and 0.11 for 80100
mesh and 200230 mesh grains, respectively.
The t/K
r
(eective tortuosity) values inferred from
this study, as summarized in Table 3, are either 50 or
100 for both arsenate and arsenite within 80100 and
140150 mesh AA grains, and are 110500 for the
200230 mesh grains. The more porous samples (80
100 and 140150 mesh) do possess smaller eective
tortuosity compared to those for the less porous
samples (200230 mesh), indicating that the general
trend for the extracted eective tortuosity factors was
correct. In fact, a factor of 4 is generally observed for
the dierence between the porous and less porous
granular samples, which is very close to the dierence
between the corresponding interpreted constrictivity
factors (0.36/0.11), suggesting that the model is
reasonable.
Theoretically, t/K
r
values for dierent compounds
with same constrictivity factor in the same adsorbent
should be the same because they have identical
diusional pathways. The t/K
r
values for both
arsenate and arsenite solutes within the same mesh
sizes of AA grains are almost the same, even though
that their experimental concentrations and isotherms
were quite dierent. The fact of identical t/K
r
values
for two dierent species in the same activated
alumina further substantiates that the application
of the diusion model for the transport of arsenate
and arsenite solutes within AA grains is appropriate.
Fig. 6. Adsorption kinetic data and model predictions of
arsenate onto (a) 80100 mesh-activated alumina (initial
concentration=4.4 mg/L, pH=6.9); (b) 200230 mesh-
activated alumina (initial concentration=6.1 mg/L,
pH=6.9).
Adsorption of arsenite on activated alumina 2055
CONCLUSIONS
The adsorption of arsenite and arsenate onto
activated alumina is governed by both the surface
charge of AA and the form of arsenic species in the
water. As a result of that, pH is a strong factor in the
uptake of both arsenite and arsenate by activated
alumina. The uptake of arsenite is much less than
that of arsenate for AA in most pH conditions,
because under most pH conditions for natural water,
arsenate is present in negatively ionic form and
arsenite is in non-ionic form. The Freundlich and
Langmuir isotherm equations were able to describe
the partitioning behavior for the system at dierent
pH. The pore diusion model, coupled with the
observed Freundlich or Langmuir isotherm equa-
tions, captured the observed experimental adsorption
kinetic curves for arsenite and arsenate at various
conditions, including dierent initial arsenic concen-
tration, grain size of activated alumina, and system
pH. The models were able to predict the experimental
data fairly well, and pore diusion coecients can be
extracted for each case. The interpreted tortuosity for
dierent cases was found to be very close, even
though that the tested conditions substantially
diered from one another. The agreement among
the tortuosity factors indicated that the adsorption
and diusion of arsenate and arsenite can be
simulated by the proposed model.
Acknowledgements}This work was supported by Environ-
mental Protection Administration, Taiwan, ROC,
under contract EPA-88-J1-02-03-403. Discussions about
arsenic diusion in water with Professors H. P. Wang and S.
G. Su at National Cheng Kung University are also
acknowledged.
REFERENCES
Anderson M. A., Ferguson J. F. and Gavis J. (1976)
Arsenate adsorption on amorphous alumina hydroxide.
J. Colloid Interface Sci. 54, 391399.
APHA, AWWA and WPCF (1996) Standard Methods, 19th
ed. American Public Health Association, Washington,
DC, USA.
Ball W. P., Buchler C. H., Harmon T. C., McKay D. M.
and Roberts P. V. (1990) Characterization of a sandy
aquifer material at the grain scale. J. Contam. Hydrol. 5,
253295.
Black G. R. (1965) Particle density. In Methods of Soil
Analysis part 1, eds. C. A. Black, D. D. Evans, J. L.
White, L. E. Ensminger and F. E. Clark, pp. 371373.
Am. Society of Agronomy, Madison, WI.
Bodek I., Lyman W. J., Reehl W. F. and Rosenblatt D. H.
(1988) Environmental Inorganic Chemistry. Pergamon
Press, NY, USA.
Bowers A. R. and Huang C. P. (1985) Adsorption
characteristics polyacetic amino acids onto Hydrous g-
Al
2
O
3
. J. Colloid Interface Sci. 105, 197215.
Brunauer S. P., Emmett P. H. and Teller E. (1938)
Adsorption of gases in multimolecular layers. J. Am.
Chem. Soc. 60, 309319.
Chantong C. and Massoth F. E. (1983) Restrictive diusion
in alumina. A. I. Ch. E. J. 29, 725731.
Chen S. L., Dzeng S. R., Yang M. H., Chiu K. H., Shieh G.
M. and Wai C. M. (1994) Arsenic species in groundwaters
of the blackfoot disease area, Taiwan. Environ. Sci.
Technol. 28, 877881.
Chen H. W., Frey M. M., Cliord D., McNeill L. S. and
Edward M. (1999) Arsenic treatment considerations.
J. Am. Water Works Assoc. A 91, 7485.
Cliord D. A. (1990) Ion exchange and inorganic adsorp-
tion. In Water Quality and Treatment, 4th edn., ed F. W.
Pontius. American Water Works Association, McGraw-
Hill, Inc., NY, USA.
Cliord D. A. and Lin C. C. (1991) Arsenic(III) and
Arsenic(V) removal from drinking water in San Ysidro,
New Mexico. USEPA Project Summary, EPA-600-S2-9-
011, Cincinati OH, USA.
Fox K. R. (1989) Field experience with point-of-use
treatment systems for arsenic removal. J. Am. Water
Works Assoc. 81, 94101.
Fox K. R. and Sorg T. J. (1987) Controlling arsenic,
uoride, and uranium by point-of-use treatment. J. Am.
Water Works Assoc. 79, 8184.
Frey M. M., Chwirk J., Kommineni S., Chowdhury Z., and
Narasimha R. (2000) Cost implications of a lower arsenic
MCL. AWWA Research Foundation.
Fuller C. C., Davis J. A. and Waychunas G. A. (1993)
Surface chemistry of ferrihydrite: part 2. Kinetics of
arsenate adsorption and coprecipitation. Geochim. Acta
57, 22712282.
Hathaway S. W. and Rubel F. J. (1987) Removing arsenic
from drinking water. J. Am. Water Works Assoc. 79,
6165.
Kim K. K. and Smith J. M. (1974) Diusion in nickel oxide
pellets: eects of sintering and reduction. A. I. Ch. E. J.
20, 670678.
Lin T. F. (1998) Best available technologies and cost analysis
for source water treatment. Report Number EPA-87-J1-
02-03-08, National Cheng Kung University, Taiwan
(in Chinese).
Table 3. Estimated tortuosity factors for the transport of arsenate and arsenite within activated-alumina grains
Grain size (mesh) Adsorbate Initial conc. (mg/L) pH t/K
r
(eective tortuosity)
a
Freundlich
b
Langmuir
c
80100 Arsenite 5.58 8.8 50 100
Arsenite 1.07 8.8 50 50
Arsenate 4.37 7.0 50 110
140150 Arsenite 3.08 8.8 50 100
200230 Arsenite 7.33 8.8 200 200
Arsenite 5.01 6.9 250 500
Arsenate 6.08 6.9 110 200
a
Best tted t/K
r
values are calculated assuming that D
b
=1 10
5
cm
2
/s.
b
Isotherm described by Freundlich equation.
c
Isotherm described by Langmuir equation.
Tsair-Fuh Lin and Jun-Kun Wu 2056
Lin T. F. (1999) Technology development of arsenic removal
from ground water. Report Number EPA-88-J1-02-03-
403, National Cheng Kung University, Taiwan
(in Chinese).
Lin T. F., Little J. C. and Nazaro W. W. (1994) Transport
and sorption of volatile organic compounds and water
vapor within dry soil grains. Environ. Sci. Technol. 28,
322330.
Lin T. F., Little J. C. and Nazaro W. W. (1996) Transport
and sorption of organic gases in activated carbon. J.
Environ. Engng., ASCE 122, 169175.
Raven K. P., Jain A. and Loeppert R. H. (1998) Arsenite
and arsenate adsorption on ferrihydrite: kinetics, equili-
brium, and adsorption envelopes. Environ. Sci. Technol.
32, 344349.
ROCEPA (1998) National Drinking Water Quality Stan-
dards. Environmental Protection Administration, Tai-
wan, ROC.
Rosenblum E. and Cliord D. (1983) The equilibrium
arsenic capacity of activated alumina. US EPA Report
Number: EPA-600/2-83-107.
Ruthven D. M. (1984) Principles of Adsorption and
Adsorption Processes. Wiley, NY, USA.
Sattereld C. N. (1981) Mass Transfer in Heterogeneous
Catalysis. MIT Press, Cambridge, MA, USA.
Sattereld C. N., Colton C. K. and Pitcher W. H. (1973)
Restricted diusion in liquids with ne pores. A I Ch.E. J.
19, 628635.
Stumm W. and Morgan J. (1981) Aquatic Chemistry. 2nd
ed. Wiley New York, USA.
Tien C. (1994) Adsorption Calculations and Modeling.
ButterworthHeinemanm, Boston, MA, USA.
Wakao N. and Funazkri T. (1978) Eect of uid dispersion
coecients on particle-to-uid mass transfer coecients
in packed beds. Chem. Engng. Sci. 33, 13751384.
Wang C.-T. and Smith J. M. (1983) Tortuosity factors for
diusion in catalyst pellets. A. I. Ch. E. J. 29, 132136.
World Health Organization (WHO) (1996) Guidelines for
Drinking Water Quality, Vol. 2 Health Criteria and Other
Supporting Information, 2nd ed. World Health Organiza-
tion, Geneva.
Xu H., Allard B. and Grimvall A. (1991) Eect of
acdication and natural organic materials on the mobility
of arsenic in the environment. Water Air Soil Pollut.
5758, 269278.
Yuan J. R., Ghosh M. M. and Teoh R. S. (1987)
Adsorption of arsenic on hydrous oxides. In Management
of Hazardous and Toxic Wastes in the Process Industries,
eds S. T. Kolaczkowski, B. D. Crittenden, pp. 363371.
Elsevier Applied Science, London, UK.
Adsorption of arsenite on activated alumina 2057

You might also like