You are on page 1of 9

Separation and Purification Technology 48 (2006) 288–296

Experimental and modelling studies on fixed bed adsorption


of As(III) ions from aqueous solution
Tony Sarvinder Singh, K.K. Pant ∗
Department of Chemical Engineering, Indian Institute of Technology, Hauz Khas, New Delhi 110016, India

Received 1 March 2005; received in revised form 11 July 2005; accepted 19 July 2005

Abstract

This paper deals with the experimental investigation on removal of arsenic [As(III)] ions from drinking water by activated alumina and iron
oxide impregnated activated alumina (IOIAA). Effect of inlet flow rate, sorbent bed height and initial As(III) concentration on the adsorption
of As(III) from aqueous solution were studied. Increase in throughput volume was observed with increase in bed height whereas inverse
relationship of flow rate and initial As(III) concentration with removal of arsenite ions was observed by these sorbents. Compared to activated
alumina, iron oxide impregnated activated alumina was found more effective in removing As(III) ions. The dynamics of adsorption process
was modelled by bed depth service time (BDST) and pore diffusion model. Adsorption rate constant (ka ) was found to increase with increase
in flow rate indicating the overall system kinetics was dominated by external mass transfer in the initial part of the adsorption in the column.
Critical bed depth (Z0 ) increased with increase in flow rate for both the adsorbent. Relatively lower critical bed height was observed for
As(III) removal onto IOIAA (0.56 cm) compared to AA (1.12 cm) at identical flow rate (0.083 cm3 /s). Time required for traveling a unit length
of adsorber bed varied from 17,280 to 43,920 s (4.8–12.2 h) and 21,240–54,360 s for AA and IOIAA depending upon the conditions. Pore
diffusion model explained the breakthrough behaviour for As(III) removal with a high degree of correlation.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Arsenic; Breakthrough studies; BDST model; Pore diffusion model; Adsorption zone

1. Introduction A variety of technologies have been used for the


treatment of As(V) in water which includes conventional
Increasing concern about the effect of toxic metals co-precipitation, lime softening, filtration, ion exchange,
released from different natural and anthropogenic sources reverse osmosis and membrane filtration [2–5] but due to
has resulted in more stringent environmental regulations in the excessive use of chemicals, bulky sludge, high cost,
recent years. Occurrence of arsenic, a commonly occurring these techniques are not feasible at small-scale or household
toxic metal in natural ecosystem, can be associated with level. Most of the available studies in published literature
natural conditions or the industrial practices of mankind. Nat- are for As(V) species and limited work has been reported
ural arsenic is generally associated with sedimentary rocks for As(III) removal in fixed bed [6,7].
of marine origin, weathered volcanic rocks, fossil fuels and It is well known from the chemistry of arsenic that As(III)
geothermal areas [1]. Arsenic readily substitutes for silicon, is more toxic and mobile than As(V). Due to presence of
ferric, iron and aluminium in crystal lattices of silicate miner- neutral ionic specie of As(III) in the pH range of most surface
als and therefore, it is possible for it to occur in all geological waters (6.5–7.5), this form of arsenic is very difficult to treat
materials. by different technologies.
Different types of sorbents have been attempted for As(III)
removal up to various degrees of success. Some of the most
Abbreviations: AA, activated alumina; IOIAA, iron oxide impregnated
activated alumina; R2 , correlational coefficient
common sorbents tried for the removal of arsenic are iron
∗ Corresponding author. Tel.: +91 11 26596172; fax: +91 11 26521120. oxide, manganese oxides and different polymeric materials
E-mail address: kkpant@chemical.iitd.ac.in (K.K. Pant). and alumina [8–13]. The use of different oxidizing agent

1383-5866/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.seppur.2005.07.035
T.S. Singh, K.K. Pant / Separation and Purification Technology 48 (2006) 288–296 289

biological changes occurring in the adsorbent. Column


Nomenclature adsorption experiments were performed to study the adsorp-
tion dynamics. Main objective of the fixed bed adsorption
ap particle radius (cm)
is to reduce concentration in the effluent so that it does
b Langmuir constant (cm3 /g)
not exceed permissible limit (breakthrough concentration
C arsenic concentration in solution (g/(cm3 L))
Cb ). The most important criterion in the design of fixed
C0 initial sorbate concentration (g/cm3 )
bed adsorption systems is the prediction of column break-
Cb breakthrough concentration = 0.01 g/cm3
through or the shape of the adsorption wave front, which
Ce equilibrium Arsenic concentration in solution
determine the operation life span of the bed and regeneration
(g/cm3 )
times.
Ct arsenic concentration in solution at time t
Mathematical models facilitate the design and analysis of
(g/cm3 )
full-scale systems by reducing the number of pilot-scale test
DL axial dispersion coefficient (cm2 /s)
required to evaluate various operating conditions and design
DP Fick pore diffusivity (cm2 /s)
parameters for sorption. The design of the adsorption pro-
ka rate constant in BDST model (cm3 /(g s))
cess is based on the accurate production of breakthrough
Kf external mass transfer coefficient (cm/s)
curves. Various mathematical models have been developed
L column length (cm)
to describe contaminant sorption over different adsorbents in
m mass of the adsorbent (g)
many of the diverse applications for which it is used [17–19].
MV mass of the adsorbent per unit volume of
The main obstacle in using mathematical models for
particle free adsorbate solution (g)
design and analysis of sorption system is that they typi-
N0 adsorption capacity (g/cm3 )
cally require many input parameters, some of which may be
q adsorbed phase concentration (g/kg)
site-specific and can be obtained only experimentally. The
qe amount of solute uptake per unit mass of
parameters for the column study were based on the results
adsorbent at equilibrium (g/kg)
obtained in batch experiments [13,20].
qm Langmuir isotherm constant (g/kg)
Present study was carried out to evaluate the performance
r radial distance (cm)
of activated alumina (AA) and iron oxide impregnated
Re Reynolds number (dimensionless)
activated alumina (IOIAA) for As(III) removal in fixed bed
Sc Schmidt number (dimensionless)
mode. The effect of process parameters, such as inlet flow
t time (s)
rate, adsorbent bed height and initial adsorbate concentration
T temperature (K)
on the shape of breakthrough curves was investigated. The
v linear flow rate (cm/s)
dynamics of adsorption process was modelled by bed depth
V interstitial velocity (cm/s)
service time (BDST) and pore diffusion model. The study
V0 velocity at inlet of the fixed bed (cm/s)
forms a part of overall experimental investigations to develop
x amount of arsenic adsorbed in solid phase (g)
an adsorption process to remove arsenic from water.
z axial distance (cm)
Z0 critical bed depth (cm)

Greek letters 2. Materials and method


ε bed porosity
εp particle porosity All the chemicals used in the study were of high purity
ρ density (kg/m3 ) analytical grade and double distilled/de-ionised water was
ρs density of adsorbent (kg/m3 ) used in the preparation of all the solutions. All the experi-
ments were carried out at room temperature (298 ± 2 K) and
Subscript atmospheric pressure. Method for the preparation of IOIAA
L bulk fluid has been discussed elsewhere and the final iron oxide content
s solid phase as determined by UV–vis spectrophotometer was 10% (wt%)
[20]. Arsenite [As(III)] stock solution (1000 g/cm3 ) was pre-
pared by dissolving dehydrated sodium arsenite (NaAsO2 )
impregnated materials for As(III) removal has been well doc- (Merck, Germany) in the de-ionized water. Dissolution of
umented in the literature [14–16]. NaAsO2 also includes addition of dilute HCl. pH of the
Most of the reported studies for arsenic removal in the solution was adjusted using either 0.1 M HCl or 0.01N
literature have been conducted in batch operation. Batch NaOH.
experimental data is often difficult to apply directly to fixed
bed adsorber because isotherm cannot give accurate data 2.1. Determination of arsenic concentration
for scale up since a flow column is not at equilibrium. The
adsorbent rarely becomes totally exhausted in commercial Quantitative determination of arsenic was done by
processes, and the isotherms cannot predict chemical or Graphite Tube-Atomic Absorption Spectrophotometer
290 T.S. Singh, K.K. Pant / Separation and Purification Technology 48 (2006) 288–296

(Varian Spectra AA-860) by using hollow cathode lamp at Table 1


a wavelength of 193.7 nm. The instrument was calibrated Physico-chemical properties of the activated alumina and iron oxide impreg-
nated activated alumina
by using As(III) standard solutions of 0.025, 0.050 and
0.10 ␮g/cm3 , which were prepared in diluted HNO3 . All Properties Activated Iron oxide impregnated
alumina (AA) activated alumina (IOIAA)
the arsenic concentration determinations were carried out
in triplicates and the average value of the same has been Particle form Spheres Spheres
Color White Reddish brown
reported in the manuscript. All the readings having deviation Particle size (mm) 0.2 ± 0.02 0.2 ± 0.02
more than ±7% were discarded and repeated. Surface area (m2 /g) 365 ± 5 200 ± 10
Pore volume (cm3 /g) 0.42 0.27
2.2. Fixed bed adsorption studies Bulk density (g/cm3 ) 0.800 0.850

Fixed bed adsorption studies were conducted to evaluate


were estimated from the batch experiments and discussed
column performance for As(III) removal on AA and IOIAA.
elsewhere [13,20]. Fixed bed adsorption studies were
Experiments were conducted in a 2.5 cm i.d., 30.0 cm length
conducted at initial pH 7.6 and 12.0 for activated alumina
vertical down flow Perspex column packed with different
and iron oxide impregnated activated alumina, respectively.
adsorbent (AA and IOIAA) at optimum pH. It was observed
At these respective pH, maximum sorption capacity for
in earlier studies [13,20] that maximum As(III) removal was
As(III) removal was observed [13,20].
obtained at pH values of 7.6 and 12.0 for AA and IOIAA,
respectively. It may be noted that final pH of the treated
water by IOIAA was 7.2. Preliminary experiments carried 3.1. Effect of bed height on the performance of
out with two different column diameters (2.5 and 5.0 cm breakthrough
i.d.) revealed that column diameter had negligible effect on
breakthrough performance when D/d ratio is more than 12. Sorbent bed height strongly affects the volume of solution
An overhead feed water reservoir (20 × 103 cm3 capacity) treated or throughput volume. Effect of bed height on adsorp-
with constant head and fitted with a flow control valve was tion of As(III) by AA and IOIAA is shown in Figs. 1 and 2,
placed over the column through which As(III) solution was respectively. As it is evident from these figures, an increase in
allowed to flow down. Influent feed flow rate was supplied column depth increased the treated volume due to high con-
and maintained throughout the experiment by the use of vari- tact time. At relatively lower contact time, the curve becomes
able flow peristaltic pump (Neolab India). At the exit of the steeper showing the faster exhaustion of the fixed bed. The
column, flow rate was also controlled so as to get steady state steepness of the breakthrough curves is a strong function of
conditions in the column. Sampling of effluent was done bed height. For As(III) removal by AA, the treated volume
at pre-determined time intervals in order to investigate the increased from 14 × 103 cm3 (14 L) to 46 × 103 cm3 (46 L)
breakthrough point (BTP). The effluent samples were filtered as the bed height was increased from 6 to 18 cm (Fig. 1).
through 0.45-␮m filter paper (Whatman No. 1), diluted and At a bed height of 6 cm, the sorbent gets saturated early as
acidified in HNO3 (AR grade, Merck, Germany) on 5% (v/v) compared to higher bed heights. The curves followed char-
basis and analysed for As(III) concentration. Effects of inlet acteristic S-shape profile, which is associated with adsorbate
feed flow rate (0.016–0.083 cm3 /s) and adsorbent bed height of smaller molecular diameter and more simple structure.
(6–18 cm) and initial adsorbate concentration (0.1 × 10−6 to
0.3 × 10−6 g/cm3 ) was investigated on the performance of
breakthrough for the adsorption of As(III) by AA and IOIAA.
The corresponding empty bed contact time (EBCT) in the bed
was in the range 222–1128 s.

3. Results and discussion

Various physico-chemical properties of the AA and


IOIAA used for the removal of arsenite ions from water have
been given in Table 1. Comparison of the various properties
showed a decrease in BET surface area of AA upon impreg-
nation with iron oxide, which could be due to the diffusion of
iron oxide particles into the pores of the adsorbent. Reduction
of porosity further confirms this observation.
Results of the batch experiments showed that incorpora-
tion of iron oxide enhance the metal uptake capacity [20]. Fig. 1. Effect of bed height on adsorption of As(III) by activated alumina
The adsorption parameters for the fixed bed adsorption (AA) [flow rate = 0.033 cm3 /s, C0 = 0.5 × 10−6 g/cm3 , pH 7.6].
T.S. Singh, K.K. Pant / Separation and Purification Technology 48 (2006) 288–296 291

Fig. 2. Effect of bed height on shape of breakthrough curve for adsorption


of As(III) by IOIAA [flow rate = 0.033 cm3 /s, C0 = 0.5 × 10−6 g/cm3 , pH
12.0]. Fig. 3. Effect of flow rate on breakthrough curves for adsorption of As(III)
by AA [C0 = 0.5 × 10−6 g/cm3 , bed height = 6 cm, pH 7.6, flow rate (cm3 /s):
() 0.016; () 0.033; () 0.083].
Higher throughput volume was observed in case of IOIAA
as compared to AA.
The rate at which adsorption zone travels through the flow rate. This decrease in breakthrough time can be due to
bed decreased with the bed depth, suggesting that beds of relatively low contact time between solute and sorbent which
an increased height may be required for As(III) adsorption. further results in low diffusion of solute into the pores of the
The typical sorption capacity for AA and IOIAA were 0.310 sorbent.
and 0.380 g/kg, respectively, at bed height of 18 cm. Rela- As As(III) ions moves from bulk solution to the surface of
tively higher sorption capacity was obtained for adsorption of film surrounding the sorbent particle, a concentration gradi-
As(III) by IOIAA (Fig. 2). The increase in the metal uptake ent develops at the interface, which allows the solute particle
capacity with the increase of bed height in the column is to penetrate through the film, and reaches the surface of par-
probably due to improvement in surface properties of the ticle. At higher flow velocities, film surrounding the particle
adsorbent, which provide more binding sites for the adsorp- breaks thereby reducing the adhesion of sorbate to the sorbent
tion. The breakthrough time also increased with increase particle.
in bed height. For the adsorption of As(III) over IOIAA. The breakthrough capacity of the sorbent was found to
57 × 103 cm3 water could be treated which was approxi- decrease with flow rate and was in agreement with the find-
mately 63.5% (vol%) when the exit As(III) concentration ing of published literature. The lower quantity of As(III)
was below the permissible limit (0.01 × 10−6 g/cm3 ). ions is in contact with the sorbent thereby thus reducing the
efficiency of metal uptake by AA and IOIAA. As the veloc-
3.2. Effect of flow rate on the performance of ity through the bed decrease, the depth of adsorption zone
breakthrough curves decreases because there is more time for adsorption on each
layer.
Investigations were made by varying flow rate from
0.016 to 0.083 cm3 /s for initial As(III) concentration of
0.5 × 10−6 g/cm3 (0.5 mg/L) at pH 7.6 and 12.0 for acti-
vated alumina and iron oxide impregnated activated alu-
mina, respectively. Breakthrough plots between Ct /C0 versus
throughput volume at different flow rates for AA and IOIAA
are given in Figs. 3 and 4, respectively. Inverse relationship
between feed flow rate and metal uptake was observed for
adsorption. It is evident from these figures that at the low flow
rate, relatively higher throughput volume was observed while
much sharper breakthrough obtained at higher flow rates.
Increasing flow rates from 0.016 to 0.083 cm3 /s reduced the
volume of water treated from 19.2 × 103 to 13.8 × 103 cm3
for As(III) treatment over activated alumina. Corresponding
decrease in sorption capacity from 0.384 to 0.276 g/kg and
0.460 to 0.324 g/kg was observed when flow rate was raised Fig. 4. Effect of flow rate on breakthrough curves for adsorption of As(III)
from 0.016 to 0.083 cm3 /s. Also time required to reach break- onto IOIAA [C0 = 0.5 × 10−6 g/cm3 , bed height = 6 cm, pH 7.6, flow rate
through (Ct = 0.01 × 10−6 g/cm3 ) decreased with increase in (cm3 /s): () 0.016; () 0.033; () 0.083].
292 T.S. Singh, K.K. Pant / Separation and Purification Technology 48 (2006) 288–296

It was observed that the sorbent gets saturated faster at


higher concentrations. It can be seen from Figs. 5 and 6 that at
lower inlet As(III) concentrations breakthrough curves were
dispersed. Breakthrough occurred very late and the surface of
the adsorbent was saturated with As(III) at a relatively longer
time. This fact is probably associated to the availability of
sorption sites around or inside the adsorbent particles, able to
capture As(III) at sufficient retention time. In a second stage,
with the gradual occupancy of these sites, the uptake becomes
less effective. Even after the breakthrough occurred, the col-
umn was still capable of accumulating As(III), although at a
progressively lower efficiency.

3.4. Modelling of breakthrough curves


Fig. 5. Effect of initial As(III) concentration on % As(III) removal by acti-
vated alumina (AA) [flow rate = 0.083 cm3 /s, pH 7.6, bed height = 18 cm, The shape of the breakthrough curves was found to be
initial As(III) concentration (10−6 g/cm3 ): () 0.1; () 0.2; () 0.3]. affected by the fluid velocity, concentration of solute in feed
and length of adsorber bed (bed height). The break point may
3.3. Effect of initial arsenic concentration on be sharply defined in some cases and poorly in some other.
breakthrough curve Steep slopes of breakthrough curves were obtained for sys-
tems that exhibit high film transfer coefficients, high internal
The adsorption performance of AA and IOIAA was diffusion coefficients or flat adsorption isotherm [22].
tested at various inlet As(III) concentrations varying from By material balance it can be shown that the area between
0.1 × 10−6 to 0.3 × 10−6 g/cm3 (0.1–0.3 mg/L). The sorption curve and a line at Ct /C0 = 1.0 is proportional to the total
breakthrough curve for AA and IOIAA obtained at differ- solute adsorbed if the entire volume comes to equilibrium
ent initial sorbate concentrations are shown in Figs. 5 and 6, with the feed. The dynamic behaviour of the column was
respectively. Inverse relationship between initial As(III) con- predicted with the Adam Bohart and pore diffusion model in
centration and throughput volume was observed. the present investigation.
There is relatively a slower transport due to decreased dif- Bohart and Adams [23] proposed a relationship between
fusion coefficient and decreased mass transfer coefficient at bed depth and time taken for breakthrough to occur. This
low metal concentration [21]. Binding sites are quickly filled approach was focused on the estimation of characteristic
at higher initial concentration resulting a decrease in break- parameters such as maximum adsorption capacity (N0 ) and
through time. Volume throughput obtained for As(III) adsorp- kinetic constant (ka ) from Adam Bohart model. The funda-
tion was more than 150 × 103 cm3 (150 L), 75 × 103 cm3 mental equations describing the relationship between Ct /C0
(75 L) and 56 × 103 cm3 (56 L) at initial arsenic concentra- and t in a continuous system were established for the adsorp-
tion of 0.1 × 10−6 , 0.2 × 10−6 and 0.3 × 10−6 g/cm3 , respec- tion of sorbate on a adsorbent, is known as bed depth service
tively (Fig. 6). time (BDST) model. Bed depth service time model has been
extensively used for predicting breakthrough curves for phe-
nol and designing of columns [21–24].
This model assumes that the adsorption rate is proportional
to residual capacity of the sorbent and the concentration of
the sorbing species. The service time was related to process
conditions and operating parameters as:
 
C0
ln − 1 = ln |eka N0 Z/v − 1| − ka C0 t (1)
C
A linear relationship between bed depth and service time may
be given by Eq. (2) [19]:
 
N0 Z 1 C0
t= − ln −1 (2)
C0 v ka C 0 Ct
where Ct is the effluent concentration of adsorbate in the liq-
Fig. 6. Effect of initial As(III) concentration on As(III) removal by iron
oxide impregnated activated alumina (IOIAA) [flow rate = 0.083 cm3 /s, pH
uid phase (g/cm3 ), C0 the initial concentration of sorbate in
12.0, bed height = 18 cm, initial As(III) concentration (10−6 g/cm3 ): () 0.1; the liquid phase (g/cm3 ), v the linear flow rate (cm/s), N0 the
() 0.2; () 0.3]. adsorption capacity (g/cm3 ), ka the rate constant in BDST
T.S. Singh, K.K. Pant / Separation and Purification Technology 48 (2006) 288–296 293

model (cm3 /(g s)), t the time (s) and Z is bed depth of column within limit of ±5%. The values of BDST model parameters
(cm). Eq. (2) enables the service time, t, of an adsorption as obtained from the slope and intercept of BDST plot are
bed to be determined for a specified bed depth, Z, of adsor- given in Table 2. As can be seen from Table 2, the values of
bent. In the investigation, the service time and bed depth were kinetic constants were influenced by flow rate and increased
correlated with the process parameter, such as initial arsenic with increase in flow rate indicating that overall system
concentration, flow rate and adsorption capacity. Eq. (2) can kinetics was dominated by external mass transfer in the initial
be rewritten in the form of a straight line: part of the adsorption in the column. For the two sorbents
studied for arsenic removal, the critical bed depth (Z0 ) was
t = m x Z − Cx (3) higher for As(III) removal by AA than IOIAA and increased
where mx is the slope of the BDST line and the intercept of with increase in flow rates. Calculated value of adsorption
this equation represents: capacity (N0 ) showed a marginal decrease with increasing
  flow rates. These results correlate well with the observed
1 C0 performance in the breakthrough curves and explained
Ct = ln −1 (4)
ka C 0 C experimental results for poor performance at higher flow
rates.
The slope of the BDST curves, mx , represented the time Although the BDST model provides a simple and com-
required for the adsorption zone to travel a unit length through prehensive approach for evaluating sorption column test, its
the adsorbent under the given experimental conditions. This validity is limited in the range of conditions used [22–24]. The
can be used to predict the performance of the bed, if there is a present investigation revealed that the model fits well in the
change in the initial solute concentration, C0 , to a new value initial region of the breakthrough curve indicating the mass
of solute concentration. transfer limitations. Since the accuracy of a model is gener-
The critical bed depth (Z0 ) is obtained for t = 0 and for a ally a function of number of adjustable parameters used in
fixed outlet concentration Ct = Cb , where Cb is the concen- the model. It was concluded that sound correlations of the
tration at the breakthrough defined as a limit concentration model parameters with operating variables are needed before
or a fixed percent of initial concentration: the models can be used for design and scale-up studies.
 
v C0
Z0 = ln −1 (5)
ka N 0 Cb 3.5. Pore diffusion model
The critical bed depth (Z0 ) represents the theoretical depth
The pore diffusion model used to describe the break-
of adsorbent, necessary to prevent the sorbate concentration
through curves was non-equilibrium and isothermal and
to exceed the limit concentration Cb . Applying Eq. (5) to
included non-linear adsorption isotherm [25,26]. This model
the experimental data at the breakthrough point, a linear rela-
was successfully used for the breakthrough analysis of Fluo-
tionship was found while plotting service time (t) against bed
ride ions onto AA [27]. The development of model equation
height, Z.
was based on assumptions such as the dominant mass trans-
The slope of the BDST plot decreased at low linear
fer resistance is intraprticle diffusion; the adsorbent particles
flow rates. Comparison between experimental and the-
were spherical in shape and are in a fixed position in the adsor-
oretical breakthrough curves as obtained from BDST
ber bed. The pressure gradient across the bed was assumed to
model shows good agreement between the experimental
be constant and radial dispersion and concentration gradient
and predicted values up to an initial part of breakthrough
in radial direction were neglected within the bed.
(breakpoint) was found, suggesting that the BDST model
is valid for As(III) adsorption till breakthrough point, i.e.
Ct = 0.01 × 10−6 g/cm3 (0.01 mg/L). However, there were 3.6. Model equations
some discrepancies found between the experimental and pre-
dicted curves above break points for the arsenic sorption on In fixed bed operation, a small element of fixed bed
AA and IOIAA column. The experimental sorption capacities bounded by an upper and lower surface with a distance of
for As(III) was well matched with the model predicted values z is taken as a control volume. Since the flow is assumed

Table 2
BDST parameters for adsorption of As(III) over AA and IOIAA at 10% of the breakthrough concentration (C0 = 0.5 g/cm3 )
Flow rate (cm3 /s) tu (s/cm) KAB (10−4 cm3 /(g s)) N0 (106 g/cm3 ) R2 Z0 (cm)
Activated alumina (AA)
As(III) 0.033 43920 0.029 0.344 0.999 0.638
0.083 17100 0.04 0.335 0.999 1.12
Iron oxide impregnated activated alumina (IOIAA)
As(III) 0.033 54360 0.032 0.559 0.998 0.471
0.083 21384 0.038 0.550 0.999 0.561
294 T.S. Singh, K.K. Pant / Separation and Purification Technology 48 (2006) 288–296

to be one-dimensional along the direction of the flow (down-


wards) based on a unit cross-sectional area. The details of the
model equations are discussed elsewhere [27] and is briefly
discussed here.

3.6.1. Liquid phase mass balance


A material balance on the adsorption bed results in a pair
of partial differential equation for solute entering the liquid
phase of the bed and adsorbing on the solid phase at z → 0:
∂Cb ∂ 2 Cb ∂Cb ∂V (1 − ε) ∂qp
= DL 2 − V − Cb − ρs (6)
∂t ∂z ∂z ∂z ε ∂t
The initial and boundary conditions were:
∂Cb
DL = −V0 (Cbin − Cb ); z = 0 (inlet), t > 0
∂z
Fig. 7. Comparison of experimental and theoretical breakthrough curves for
Cb = 0; 0 < z < L, t = 0 (7) As(III) removal at different flow rate by IOIAA–pore diffusion model.
∂Cb
= 0; z = L (outlet), t > 0
∂z
The above equations were solved numerically using the
3.6.2. Solid phase mass balance
implicit backward difference scheme [27]. With the respec-
The transport of the adsorble species from the bulk of the
tive coefficients of the corresponding unknown variable a
fluid to the external surface of adsorbent pellets constitutes
tri-diagonal matrix was formed which was solved to obtain
an important step in the overall uptake process. For adsorp-
the bulk concentration and the adsorbed phase concentration
tion in spherical pellets, the inter phase mass transfer may be
was calculated using the isotherm equations and the bulk con-
expressed as [18]:
centration.
∂qp 3Kf With this adsorbed phase concentration, the concentra-
ρs = (Cb − Cs ) (8)
∂t ap tion in pore fluid at various radial and axial positions was
calculated. With the new concentration value at pellet sur-
where qp is the average adsorbed phase concentration, Cb face whole process was repeated till the uptake of adsorbate,
and Cs the concentrations in bulk of the fluid and that at fluid obtained from successive iterations converged. The root mean
pellet interface, ap the pellet radius and Kf is the inter phase square error was calculated as:
mass transfer coefficient. Its magnitude of course depends
upon the flow conditions around the pellet [26].
1 0.5
RMSE = [(Exp − Model)2 ]
n
3.6.3. Pore diffusion equation
Structurally, alumina pellets are porous and adsorption
where n is the number of data points.
takes place exclusively in the void surface of the pellet.
Adsorbate requires to diffuse into the interior of the pellet.
Intra pellet mass transfer needs to be considered as part of
the uptake process. In this model, the surface diffusion is
neglected and only pore diffusion is taken into account.
For the diffusion of arsenic molecules through the pore
fluid, macroscopic equation is given by:
 
∂C ∂q 1 ∂ ∂C
εp + (1 − εp ) = 2 εp DP r 2 (9)
∂t ∂t r ∂r ∂r
where the term on right-hand side represents the diffusion of
solute into the pores.
Initial and boundary conditions:
C = 0; ap = 0, 0 < r < ap , t = 0
∂C
= 0; r = 0 (centre of pellet), t > 0 (10)
∂r
∂C
−Kf (C − Cs ) = DP ; r = ap , t > 0 Fig. 8. Comparison of experimental and theoretical breakthrough curves for
∂r As(III) removal at different flow rate by AA–pore diffusion model.
T.S. Singh, K.K. Pant / Separation and Purification Technology 48 (2006) 288–296 295

Fig. 10. Comparison of theoretical and predicted breakthrough curve for


Fig. 9. Comparison of experimental and pore diffusion model for arsenic adsorption of As(III) onto activated alumina–pore diffusion model.
removal onto IOIAA at different initial concentration–pore diffusion model.

3.7. Analysis of breakthrough curves by pore diffusion with experimental exhaustion time has been given in
model Table 3.
There were three parameters in the model equations
The breakthrough curves obtained theoretically from external mass transfer coefficient (Kf ), liquid dispersion
pore diffusion model were compared with experimental coefficient (DL ) and pore diffusivity (DP ). Initial values
breakthrough plots and are shown in Figs. 7–10. The of these parameters were obtained using available correla-
curve has a long tail because the final molecules adsorbed tions [25,26], however, final values were estimated using
have to diffuse almost to the centre of the particle. When regression analysis for experimental curve by minimizing
both internal and external resistances are significant the the error between experimental and simulated breakthrough
breakthrough curve is S-shaped [28]. As can be seen curves.
from Figs. 7–10, the predicted model fits the experimental The value of parameters obtained at different flow rates
data very well. Comparison of simulated exhaustion time and bed height are given in Table 4. The sensitivity of the
model was also checked for these parameters and results indi-
cated that model is very sensitive to model parameters Kf , DL
Table 3 and DP .
Comparisons of experimental and theoretical exhaustion time for adsorption The predicted breakthrough curve and estimated param-
of arsenic by AA and IOIAA eters satisfactorily explain the behaviour of experimental
0.033 cm3 /s 0.083 cm3 /s results at different flow rates. There was a significant change
Experimental Simulated Experimental Simulation in the shape of breakthrough curve with the change in flow
time time time time rate. Higher film transfer coefficient and pore diffusion coef-
(10−2 s) (10−2 s) (10−2 s) (10−2 s) ficient were observed at lower initial concentration. The vari-
AA ation in the value of axial dispersion coefficient DL did not
As(III) 4500 4356 1656 1620 have any significant effect on the breakthrough curve. Thus,
As(V) 5112 4860 1944 1692 the correlations can be used without much possibility of error
IOIAA for determining the value of DP and DL . Pore diffusion model
As(III) 6660 6516 2448 2376 parameters obtained at different initial arsenic concentrations
As(V) 7200 7020 2808 2736
are given in Table 5.

Table 4
Estimation of various column parameters from pore diffusion model for adsorption of arsenic over AA and IOIAA
Flow rate (cm3 /s) Velocity (cm/s) Pore diffusion coefficient, DP Axial dispersion coefficient, Mass transfer coefficient, Kf
(cm2 /s) DL (cm2 /s) (cm/s)
Iron oxide impregnated activated alumina (IOIAA)
0.033 1.331E−02 0.5E−11 1.5E−05 5.9
0.083 3.328E−02 0.1E−10 3.8E−05 6.5
Activated alumina (AA)
0.033 1.331E−02 2.4E−12 5.8E−04 3.2
0.083 3.328E−02 2.8E−12 6.8E−04 3.95
296 T.S. Singh, K.K. Pant / Separation and Purification Technology 48 (2006) 288–296

Table 5 [3] C.P. Huang, L.M. Vane, Enhancing As(V) removal by a Fe(II)-treated
Pore diffusion model parameters at different initial arsenic concentrations activated carbon, J. Water Pollut. Control Fed. 61 (1989) 1596–1603.
Initial As(III) Pore diffusion Axial Mass transfer [4] J.G. Hering, Arsenic removal by ferric chloride, J. Am. Water Works
concentration coefficient, DP dispersion coefficient, Kf Assoc. 88 (1996) 155–167.
(106 g/cm3 ) (cm2 /s) coefficient, DL (cm/s) [5] S. McNeill, M. Edwards, Predicting arsenic removal during metal
(m2 /s) hydroxide precipitation, J. Am. Water Works Assoc. 89 (1997)
75–82.
Iron oxide impregnated activated alumina (IOIAA) [6] R.C. Vaishya, S.K. Gupta, Modeling arsenic(V) removal from water
0.2 0.1E−14 7.5E−03 0.135 by sulfate-modified iron oxide coated sand (SMIOCS), Sep. Sci.
0.3 0.3E−14 7.5E−03 9.95E−02 Technol. 39 (2004) 641–666.
Activated alumina (AA) [7] R.C. Vaishya, S.K. Gupta, I.C. Agarwal, Fixed-bed modelling of
0.2 1.5E−14 8.0E−03 0.095 arsenic(III) adsorption from water by sulfate modified iron oxide
0.3 1.5E−14 8.0E−03 6.55E−02 coated sand (SMIOCS), in: Proceedings of 12th International Con-
ference on Heavy Metals in the Environment, J. Phys. IV 107 (2003)
1325–1328.
4. Conclusions [8] J. Patanayak, K. Mondal, S. Mathew, S.B. Lalvani, A parametric
evaluation of the removal of As(V) and As(III) by carbon-based
adsorbents, Carbon 38 (2000) 589–596.
Adsorption of As(III) on AA and IOIAA has stronger [9] G.N. Manju, C. Raji, T.S. Anirudhan, Evaluation of coconut husk
dependence on feed flow rate, bed height and initial con- carbon for the removal of arsenic from water, Water Res. 32 (1998)
centration. Increase in bed height and initial concentration 3062–3070.
resulted in decrease in throughput volume whereas increase in [10] B.E. Reed, R. Vaughan, L.Q. Jiang, As(III), As(V), Hg, and Pb
removal by Fe-oxide impregnated activated carbon, J. Environ. Eng.
bed height leads to higher contact time thus higher throughput 126 (2000) 869–873.
volume. Relatively higher sorption capacity and throughput [11] X. Meng, S. Bang, G.P. Korfiatis, Effects of silicate, sulphate, and
volume were obtained with IOIAA. Bed depth service time carbonate on arsenic removal by ferric chloride, Water Res. 34 (2000)
model fitted well in the initial zone of breakthrough curves 1255–1261.
showing mass transfer limitations but fails in the later part [12] F. Rubel, S.W. Hathaway, Arsenic removal by coagulation, J. Am.
Water Works Assoc. 87 (79) (1987) 61.
of breakthrough curves. For the two sorbents studied for [13] T.S. Singh, K.K. Pant, Equilibrium, kinetics and thermodynamic
As(III) removal, higher critical bed depth was observed for studies for adsorption of As(III) on activated alumina, Sep. Purif.
AA compared to IOIAA and also increased with the increase Technol. 36 (2004) 139–147.
in flow rate. Sorption of As(III) from aqueous solution onto [14] L.V. Rajakovic, Sorption of arsenic onto activated carbon impreg-
AA and IOIAA exhibits breakthrough curves of asymmet- nated with metallic silver and copper, Sep. Sci. Technol. 27 (1992)
1423–1433.
ric nature, which can be simulated effectively by the pore [15] B.E. Reed, R. Vaughan, L.Q. Jiang, J. Environ. Eng. 126 (2000)
diffusion model. Such close agreement between the simu- 869–873.
lated and experimental breakthrough curves suggests that the [16] R.C. Vaishya, S.K. Gupta, Adsorption of As(III) on Ganga sand, J.
salient features of the dynamics of sorption columns can be Environ. Eng. 129 (2003) 89–92.
estimated by the mathematical form of the model equations. [17] W.J. Weber, E.H. Smith, Simulation and design models for adsorp-
tion processes, Environ. Sci. Technol. 21 (1987) 1040–1050.
The breakthrough curves for adsorption of arsenic from dilute [18] C. Tien, Adsorption Calculation and Modeling, Butterworth Heine-
solutions using AA and IOIAA showed the mutual effects of mann, Boston, USA, 1994.
the adsorption capacities and adsorption rate and explained [19] B. Chen, C.W. Hui, G. McKay, Pore surface diffusion modeling for
the dependence of the shape of the breakthrough curves on dyes from effluent on pith, Langmuir 17 (2001) 740–748.
experimental parameters. [20] S. Kuriakose, T.S. Singh, K.K. Pant, Adsorption of As(III) from
aqueous solution onto Iron oxide coated activated alumina, Water
Since the accuracy of any model is generally a function of Qual. Res. J. Can. 39 (2004) 260–268.
the number of adjustable parameters used in the model, it can [21] Z. Aksu, G.F. Ferda, Biosorption of phenol by immobilized acti-
be concluded that sound correlations of the model parame- vated sludge in a continuous packed bed: prediction of breakthrough
ters with operating variables are needed before the models curves, Process Biochem. 39 (2004) 599–613.
can be used for design and scale-up studies. The informa- [22] S.D. Faust, O.M. Aly, Adsorption Processes for Water Treatment,
Butterworth Publishers, USA, 1987.
tion provided in this study can be applied for the successful [23] G.S. Bohart, E.Q. Adams, Behavior of charcoal towards chlorine, J.
design of a module for the removal of As(III) from ground- Chem. Soc. 42 (1920) 523–529.
water knowing the physico-chemical properties of the source [24] V.J.P. Poots, G. McKay, J.J. Healy, The removal of acid dye from
water. effluent using natural adsorbents—I peat, Water Res. 10 (1976)
1061–1066.
[25] R.H. Perry, D. Green, Chemical Engineers Handbook, sixth ed.,
McGraw Hill Book Co., New York, 1981.
References [26] T. Kataka, H. Yoshida, K. Ueyama, Mass transfer for fixed bed
adsorbers, J. Chem. Eng. Jpn. 5 (1972) 132.
[1] N.E. Korte, Q. Fernando, A review of arsenic(III) in groundwater, [27] S. Ghoarai, K.K. Pant, Investigations on the column performance of
Crit. Rev. Environ. Contrib. 21 (1991) 1–39. fluoride adsorption by activated alumina in a fixed-bed, Chem. Eng.
[2] T.J. Sorg, G.S. Logsdon, Treatment technology to meet the interim J. 98 (2004) 165–173.
primary drinking water regulations for inorganic. Part 2, J. Am. [28] W.L. Maccabe, J.C. Smith, P. Harriot, Unit Operations in Chemical
Water Works Assoc. 70 (1978) 379–393. Engineering, McGraw-Hill Book Co., Singapore, 1993.

You might also like