722 Langmuir 2003, 19, 722-730

Applications of Multipore Model for the Mechanism

Identification during the Adsorption of Dye on Activated
Carbon and Bagasse Pith
Danny C. K. Ko, David H. K. Tsang, John F. Porter, and Gordon McKay*
Department of Chemical Engineering, Hong Kong University of Science and Technology,
Clear Water Bay, Kowloon, Hong Kong SAR, China
Received July 19, 2002. In Final Form: October 9, 2002

A solution to the intraparticle diffusion model has been developed on the basis of the pore structure of
adsorbents. This global model incorporates the mechanism of external film mass transport, intraparticle
macropore surface diffusion, and adsorption at the surface in the micropores limited by a first-order reaction
kinetic rate constant. To assess the model, it has been applied to the adsorption of acid and basic dyes
onto two adsorbents, active carbon and bagasse pith. The two adsorbents were selected because of their
differences: active carbon has a large BET surface area and microporous structure, whereas bagasse pith
has a small surface area and only a spongelike macropore structure. The branched pore model developed
and tested is able to distinguish these physical-structural characteristic differences between the two
adsorbents using two very different adsorbates.

1. Introduction ing systems, many models for liquid-phase adsorption are

Adsorption is the selective removal of a component or
impurity in a fluid by contacting the fluid with a solid
adsorbent; it is widely accepted as an effective purification
method both for drinking waters and wastewaters. The
ability of novel adsorbents to remove color from waste-
waters has been of major interest for many years; these
include silica,1,2 chitosan,3-6 Fullers earth,7 and peat.8
The materials in the present research have been
previously used for color removal by waste banana pith,9
palm branches,10 orange peel pith,11 and bagasse pith.12-14
The second adsorbent, activated carbon, is the most widely
used adsorbent for dye removal.15-17
To design adsorption treatment systems, a knowledge
of the diffusion mass transport/kinetic processes is es-
sential, and many such models have been developed. Due
to the many varied chemical systems, the nature of
different adsorbents, and the different designs of contact-
* To whom all correspondence should be addressed. E-mail:
kemckayg@ust.hk. Phone: (852)-2358-8412. Fax: (852)-2358-0054.
(1) McKay, G. Chem. Eng. Sci. 1987, 39, 129-138.
(2) McKay, G. AIChE J. 1984, 30, 692-697.
(3) McKay, G.; Blair, H. S.; Gardner, J. R. J. Appl. Polym. Sci. 1982,
27, 3043-3057.
(4) Seo, T.; Kanbara, T.; Iijima, T. J. Appl. Polym. Sci. 1988, 36,
1443-1451.
(5) Yoshida, H.; Kishimoto, N.; Kataoka, T. Ind. Eng. Chem. Res.
1994, 33, 854-859.
(6) Yoshida, H.; Kishimoto, N.; Kataoka, T. Ind. Eng. Chem. Res.
1995, 34, 347-355.
(7) McKay, G.; Otterburn, M. S.; Aga, J. A. Water, Air, Soil Pollut.
1985, 24, 307-322.
(8) McKay, G.; Allen, S. J. Can. J. Chem. Eng. 1980, 58, 521-526.
(9) Namasivayam, C.; Kanchana, N. Chemosphere 1992, 25, 1691-
1705.
(10) Nassar, M. M.; Hamoda, M. F.; Radwan, G. H. Water Sci. Technol.
1995, 32, 27-32.
(11) Namasivayam, C.; Muniasamy, N.; Gayatri, K.; Rani M.;
Ranganathan, K. Bioresour. Technol. 1996, 57, 37-43.
(12) McKay, G.; El Geundi, M.; Nassar, M. M. Water Res. 1987, 21,
1513-1520.
(13) McKay, G.; El Geundi, M.; Nassar, M. M. Process Saf. Environ.
Prot. 1997, 74, 277-288.
(14) McKay, G. Adsorption 1998, 4, 361-372.
(15) McKay, G.; Al Duri, B. Chem. Eng. Sci. 1988, 43, 1133-1142.
(16) Porter, J. F.; McKay, G.; Choy, K. H. Chem. Eng. Sci. 1999, 54,
5863-5885.
(17) Choy, K. K. H.; Porter, J. F.; McKay, G. J. Chem. Eng. Data
2000, 45, 575-584.
limited in their range of applications. External film mass
transport models exist for film diffusion,18 models exist
for film-pore diffusion,19,20 and there is an extensive range
of solutions for film-surface diffusion21-24 and certain
combined diffusion models.25-27
In the present paper a combined diffusion model has
been developed on the basis of the branched pore model.28
The adsorption of a basic dye and an acidic dye onto a
highly porous adsorbent, activated carbon, and a low
surface area, low porosity adsorbent, bagasse pith, has
been used to test the scope of the model. The effect of
initial dye concentrations has been studied for the four
systems.
2. Model Description
Previous analyses on the single- and double-resistance
mass transfer models2,29,30 show that such models agree
with experimental results but only over very limited
regions of the adsorption periodsindicating that another
internal mass transfer resistance or a variable solid
diffusivity becomes rate controlling at long time periods.
A model was presented by Peel and co-workers,28 incor-
porating three mass transfer resistances and dividing the
adsorbent particle into two regions of different diffusion
(18) Furusawa, T.; Smith, J. M. Ind. Eng. Chem. Fundam. 1973, 12,
197-203.
(19) Spahn, H.; Schlunder, E. U. Chem. Eng. Sci. 1975, 30, 529-537.
(20) Cheung, C. W.; Porter, J. F.; McKay, G. Langmuir 2002, 18,
650-656.
(21) Hand, D. W.; Crittenden, J. C.; Thacker, W. E. J. Environ. Eng.
Div. (Am. Soc. Civ. Eng.) 1983, 109, 82-92.
(22) Tien, C. Adsorption Calculations and Modeling; Butterworth-
Heinemann: Newton, MA, 1994.
(23) Do, D. D. Adsorption Analysis: Equilibrium and Kinetics;
Imperial College Press: London, 1998.
(24) McKay, G. Chem. Eng. J. 2001, 81, 213-221.
(25) Crittenden, J. C.; Luft, P.; Hand, D. W.; Oravitz, J.; Loper, S.
W.; Ari, M. Environ. Sci. Technol. 1985, 19, 1037-1043.
(26) Chen, B. N.; Hui, C. W.; McKay, G. Langmuir 2001, 17, 740-
748.
(27) Choy, K. K. H.; Porter, J. F.; McKay, G. Adsorption 2001, 7,
231-247.
(28) Peel, R. G.; Benedek, A.; Crowe, C. M. AIChE J. 1981, 27, 26-
31.
(29) Mathews, A. P.; Weber, W. J., Jr. AIChE Symp. Ser. No. 73,
1976, 91-107.
(30) Fritz, W.; Schlunder, E. U. Chem. Eng. Sci. 1974, 29, 1279-
1282.

10.1021/la020654g CCC: $25.00 2003 American Chemical Society

Published on Web 01/11/2003
Adsorption of Dye on Activated Carbon and Bagasse Pith Langmuir, Vol. 19, No. 3, 2003 723

rates. These two regions are loosely termed macropores distributed regions. Rb is the rate of transfer of adsorbate
and microporessrelatively rapid diffusion and adsorp- from the macropore network to the micropores or branch
tion occur in the macropores, and the remaining slow pores.
approach to equilibrium occurs in the micropores. The 3. Micropore Mass Balance.
three mass transfer resistances are (1) diffusion of solute
from solution, across the external liquid boundary layer qb
onto the adsorbent surfaces, (2) diffusion onto and (1 - f) ) Kb(qm - qb) ) Rb (4)
t
adsorbing in the macropore region, and (3) diffusion onto
and adsorbing in the micropore region. In addition, the following initial and boundary conditions
This branched-pore kinetic model (BPKM), based on apply
external mass transfer, with a fraction, f, of the total
adsorption capacity belonging to the macropores and a qm(r,0) ) 0 (5)
fraction, 1 - f, belonging to the micropores, has been used
to predict the performance of a batch adsorber. A computer qb(r,0) ) 0 (6)
program has been developed to generate theoretical
concentration-time curves based on this model, and the Ct(t)0) ) C0 (7)
mass transfer parameters have been adjusted to maximize
qm(R,t) ) qs(t) (8)
the correlation with experimentally measured concentra-
tion-time decay curves by a best fit approach. The four qm
mass transfer parameters of the model are the external (0,t) ) 0 (9)
mass transfer coefficient, Kf; the effective diffusivity, Deff; r
the lumped micropore diffusion rate parameter, Kb; and In previous models,28,31 the coupling between the liquid
the fraction of the total adsorption capacity allocated to and solid phases was achieved by equating the fluxes at
macropores, f. These each have an effect on the shape the solid-liquid interface:
of the concentration decay curves.

( )
To describe transport into the micropores, a lumped qm
parameter approach has been used. The particle is Kf A(Ct - Cs,t) ) fDeff AF (10)
partitioned into two regions: with a fraction, f, of the total r r)R

adsorption capacity attributed to macropores and a

Equation 10 ignores the contribution from the micropore
fraction, (1 - f), constituting the micropores. Transport
fraction of the adsorbent; therefore, in the present model
between the two regions, which are both distributed
eq 11 has been used:
radially, is described by a linear driving force expression

( )
between the local macro- and micropore concentrations. qm qb fDeff 2qm kf A
The present solution method modifies the original model f + (1 - f) ) 2 r ) (C - Cs)
used by Peel and co-workers28,31,32 and is based on solving t t r r r FV
the four key transport controlling equations using a (11)
Crank-Nicolson finite difference approach33 combined Equations 2-4 and 11 are first reduced to dimensionless
with banded matrix techniques.34 form using the substitutions
Mathematical Development. Equilibrium as de-
scribed by the equilibrium isotherms holds at the interface Ct Cs,t q
between the liquid and solid phases. In this case the model C) , Cs ) , Q) (12)
uses the Langmuir isotherm C0 C0 q0

KLCs r (Defft)
qs ) (1) ) and ) (13)
1 + aLCs R R2
The mass balance equations for the various sections of The change variable
the system can be written as follows:
1. Liquid-Phase Mass Balance. ) 2 (14)
dCt is introduced to give a finer spatial grid near the surface
V ) -Kf A(Ct - Cs,t) (2) of the particle, where concentration gradients are highest.
dt
The resultant set of equations is
2. Macropore Mass Balance. If the fraction f represents
that part of the total adsorptive capacity utilized in the Qm 2Qm Qm 1 - f Qb
macropore region, the balance can be written ) 4 2 + 4 - (15)
f

f
t
) 2
r r
r (
qm fDeff 2 qm
r
- Rb ) (3)
Qb

) PBiot
f
(Q - Qb)
1-f m
(16)

C FBiot
The introduction of this fraction is equivalent to dividing ) -3f (C - Cs) (17)
the adsorbent volume and mass into two homogeneously SFact

(31) McKay, G.; McKee, S.; Walters, H. R. J. Chem. Eng. Sci. 1987,
42, 1145-1151.
(32) Yang, X. Y.; Al-Duri, B. Chem. Eng. J. 2001, 83, 15-23.
( )
Qm
)1
) (
FBiot
2 )
(C - Cs) (18)

(33) Crank, J.; Nicolson, P. Math. Proc. Cambridge Philos. Soc. 1947, In previous work, the authors28,31 used the coefficients 4,
43, 50-67.
(34) The NAG Fortran Library Introductory Guide Mark 16; The 6, 1 instead of 4, 4, 1 on the right-hand side of eq 15. Yang
Numerical Algorithms Group Ltd: 1993. and Al Duri32 did not state the resultant set of equations.
724 Langmuir, Vol. 19, No. 3, 2003
Figure 1. Pore size distribution of bagasse pith.
Equations 15-18 are the transformed equations for
macropore and micropore mass balance equations, the
coupling equation, and the liquid-phase mass balance
equation, respectively, where
KfC0R
FBiot ) (19)
FfDeffq0
KbR2
PBiot ) (20)
fDeff
VC0
SFact ) (21)
mq0
The initial and boundary conditions (eqs 5-9) become
Qm(,0) ) 0 (22)
Qb(,0) ) 0 (23)
C(0) ) 1 (24)
Qm(1,t) ) Qs(t) (25)
The set of equations 15-18 is solved using Crank-
Nicolson finite difference techniques. The solution pro-
cedure is similar in methodology to previous work31 using
the trapezium rule, central finite difference formula, and
forward difference solution method to obtain the solution
matrix, which can be used using standard subroutines.34
3. Experimental Techniques
Adsorbents. Pith. The Egyptian bagasse used in this inves-
tigation was provided from the Abou-Korkas sugar mill, El-Mina,
Egypt. The depithing operation was performed in the sugar mill.
The moisture content of the bagasse pith was 14.5 (0.5%, and
it was not subjected to any form of pretreatment prior to use. The
pith particles were sieved in the laboratory to obtain a discrete
size range of 500-710 m. Although the particles appeared
irregularly shaped under a microscope, they approximated more
closely spheres than cylinders or parallel pipes for the size range
under investigation. Thus, they were assumed to be spheres
having a diameter given by the arithmetic mean value between
respective mesh sizes, that is 605 m.
The Egyptian bagasse pith was subjected to chemical analysis,
and the results obtained were as follows: pentosan, 27.0%;
R-cellulose, 53.7%; lignin, 20.2%; ash, 6.6%; alcohol soluble, 7.5%.
The BET surface area and the pore volume of the bagasse pith,
which were measured using a Micrometrics Accelerated Surface
Area and Porosimetry instrument (ASAP 2000), were found to
be 1.5 m2 g-1 and 0.041 cm3 g-1, respectively. The pore size
Ko et al.
Figure 2. Pore size distribution of activated carbon.
distribution of the bagasse pith is shown in Figure 1, which has
demonstrated that most of the pores inside the pith are in the
macropore region.
Activated Carbon. The activated carbon used in this inves-
tigation was type F400 provided from Filtrasorb Corporation. It
was sieved in the laboratory to obtain a discrete size range of
500-710 m. Although the particles appeared irregularly shaped
under a microscope, they approximated more closely spheres
than cylinders or parallel pipes for the size range under
investigation. Thus, they were assumed to be spheres having a
diameter given by the arithmetic mean value between respective
mesh sizes, that is 605 m. The BET surface area and the pore
volume of the GAC, which was measured using a Coulter
Omnisorp 100 CX unit, were found to be 1150 m2 g-1 and 0.766
cm3 g-1, respectively. The pore size distribution of the activated
carbon is shown in Figure 2, which has demonstrated that most
of the pores inside the activated carbon are in the mesopore and
micropore region. The pH value of the activated carbon is
measured on the basis of ASTM Method D3838-80. The result
shows that the activated carbon has a pH of 7.15 ( 0.05.
Adsorbates. The adsorbates used in the experiments are listed
below: (1) Acid Red 114 (AR114) was supplied by Sigma-Aldrich
Chemical Co. (2) Acid Blue 80 (AB80) was supplied by Ciba
Specialty Chemicals. (3) Basic Red 22 (BR22) was supplied by
Ciba Specialty Chemicals. (4) Basic Blue 69 (BB69) was supplied
by Ciba Specialty Chemicals. The dyestuffs were used as the
commercial salts. The structures and the molecular weights of
these dyes are given in Figure 3.
Preparation of Calibration Curve. The dyes were made
up in stock solutions of concentration 1000 ppm and were
subsequently diluted to the required concentrations with deion-
ized water. Calibration curves for each dye were prepared by
recording the absorbance values for a range of known concentra-
tions of dye solutions at the wavelength for maximum absorbance
of each dye on a spectrophotometer. The value max was found
from a scan for the dyes spectrum. These values of max were
used in all subsequent investigations using these dyes. The max
values for Acid Red 114 and Acid Blue 80 are 522 and 626 nm,
respectively, while the max values for Basic Red 22 and Basic
Blue 69 are 538 and 585 nm, respectively.
Equilibrium Isotherms. Equilibrium studies were carried
out to determine the adsorption behavior for the dyes onto bagasse
pith and activated carbon. Each system was given a contact time
of 5 days for dye-pith systems and 21 days for dye-carbon
systems to ensure equilibrium had been reached. For the dye-
carbon systems, the isotherms were determined by agitating 0.05
g of activated carbon with 0.05 dm3 of the dye solution for 21
days at a constant temperature of 20 ( 2 C. For the dye-pith
systems, the isotherms were determined by agitating 0.25 g of
pith with 0.05 dm3 of the dye solution for 5 days at a constant
temperature of 20 ( 2 C.
Batch Adsorber Studies. The adsorber vessel used for the
contact time experiments was based on the standard mixing tank
configuration. A 2 dm3 plastic beaker of internal diameter 0.13
m was used and contained a volume of 1.7 dm3 of dye solution.
A six-bladed, flat, plastic impeller provided good mixing in the
vessel. The diameter of the impeller and the blade height were
0.065 and 0.013 m, respectively. A Heidolph type RZR1 variable-
Adsorption of Dye on Activated Carbon and Bagasse Pith Langmuir, Vol. 19, No. 3, 2003 725

Table 1. Langmuir and Freundlich Isotherm Constants for Different Acid Dyes on Activated Carbon and Pith
Langmuir isotherm Freundlich isotherm
adsorbent dye KL (dm3 g-1) aL (dm3 mg-1) SSE KF (dm3 g-1) 1/n SSE
carbon BB69 202 0.23 12897 302 0.26 60960
carbon AB80 25.7 0.20 39.2 38.2 0.29 909
carbon AR114 14.9 0.14 92.9 28.1 0.29 331
pith BB69 16.0 0.10 10.3 38.2 0.24 5432
pith BR22 18.1 0.24 1.9 26.1 0.19 1447
pith AR114 1.10 0.05 0.8 3.89 0.30 85.0

kinetic model and must be accurately represented to

maintain confidence in the accuracy of the model and its
results. The equilibrium data were analyzed using the
Langmuir and Freundlich isotherms:

KLCe
Langmuir isotherm: qe ) (26)
1 + aLCe

Freundlich isotherm: qe ) KFCe1/n (27)

Although the surface of activated carbon F400 is known

Figure 3. Structure and molecular weight of four different
acid and basic dyes.
speed motor was used to drive the impeller using a 0.005 m
diameter plastic shaft. Six plastic baffles were spaced evenly
around the vessel circumference to avoid the formation of a vortex
and the consequential reduction in relative motion between liquid
and solid particles, and power losses due to air entrainment at
the impeller. The polystyrene baffles used were flat strips 0.2 m
long and 0.01 m wide. Kinetic data were collected for a series of
system parameters. Three mL samples were extracted at selected
time intervals (up to a maximum of 12 h) using a 10 mL syringe.
The concentration was determined, and the results were recorded
as concentration versus time decay curves. Prior to use in the
equilibrium or batch experiments, the activated carbon and
bagasse pith were rinsed with deionized water and dried at 110
C for 24 h to remove the fine dust and adsorbed impurities from
the adsorbent surface. These preliminary treatments served as
a purity and moisture control on the adsorbent.
4. Results and Discussion
Equilibrium Isotherms. The equilibrium surface
concentration is an integral part of the branched pore
to be a heterogeneous surface, the Langmuir equation
was found to give the best data correlation despite its
normal applicability to homogeneous equal energy site
adsorption. Consequently, the Langmuir form of equation
was used in the present work to describe the equilibrium
relationship, although it must be regarded as an empirical
form of data correlation. The comparisons of the fittings
of the Langmuir and Freundlich isotherms on the ex-
perimental equilibrium data for the different dyes on
activated carbon and pith, and the isotherm constants
are presented in Table 1 together with the sum of the
squares of the errors (SSE) values. Figures 4 and 5 show
the Langmuir equilibrium isotherms for the different dyes
on activated carbon and pith, respectively.
Adsorption Mechanism. Activated carbon has a high
adsorption capacity toward a range of organic pollutants,
which arises from the complexity of the chemical surface
groups. The surface of activated carbon materials contains
at least five markedly different types of surface groups
such as carboxylic, lactonic, phenolic, carbonyl, and etheric
types. This diversity of surface groups makes the surface
chemistry much more versatile than that of other
adsorbents.35-38 The F-400 demonstrates a high sorption
capacity for both H+ and OH- ions (amphoteric behavior),
and it is this behavior that enables the activated carbon
to manifest a reactivity for many organic and inorganic
adsorbates.35 Since most carbon materials have a higher
surface basicity than surface acidity, it is expected that
the activated carbon shows a higher adsorption affinity
for H+ ions compared to OH- ions.35,39
The carbon surface basicity can be expressed in the
form40-42
R-OH + H+ f R-OH2+ (28)
(35) Al-Degs, Y.; Khraisheh, M. A. M.; Allen, S. J.; Ahmad, M. N.
Water Res. 2000, 34, 927-935.
(36) Boehm, H. P. Carbon 2002, 40, 145-149.
(37) Strelko, V., Jr.; Malik, D. J. J. Colloid Interface Sci. 2002, 250,
213-220.
(38) Leon y Leon, C. A.; Radovic, L. R. Chemistry and Physics of
Carbon; Marcel Dekker: New York, 1994.
(39) Moreno-Castilla, C.; Carrasco-Marin, E.; Mueden, A. Carbon
1997, 35 (5), 1619-1629.
(40) Corapcioglu, M. O.; Huang, C. P. Carbon 1987, 25, 569.
(41) Noh, J. S.; Schwarz, J. A. J. Colloid Interface Sci. 1989, 130,
157-164.
(42) Abotsi, G. M. K.; Scaroni, A. W. Carbon 1990, 28, 79.
726 Langmuir, Vol. 19, No. 3, 2003
Figure 4. Langmuir adsorption isotherm for different dyes on activated carbon.
Figure 5. Langmuir adsorption isotherm for different dyes on pith.
The acid dissociation reaction for surface groups can be
written as37
R-COOH f R-COO- + H+ (29)
R-OH f R-O- + H+ (30)
Undoubtedly, the pH value of the dye solution also plays
a role in the adsorption process. Any oxide surface creates
a charge (positive or negative) on its surface. This charge
is proportional to the pH of the solution which surrounds
the oxide particles.35
The large difference between the adsorption capacities
of basic and acid dyes on both adsorbents is connected to
the fact that the surfaces of pith and activated carbon in
contact with water are more negatively charged. Acid Red
114 is an example of a dye which ionizes in aqueous
solution to form an anionic colored component D- and a
Ko et al.
cation, Na+. The approach of an acidic dye anion will suffer
Coulombic repulsion due to the presence of the strong
anionic groups on the surface. However, the net positive
surface during the adsorption process of the acid dye has
the effect of attracting the negative portion of the acid
dye.35
Basic Red 22 is an example of a dye which will ionize
to give the colored cationic dyebase, and this will undergo
attraction on approaching the anionic surface, resulting
in a much greater affinity and adsorption capacity for
basic dyes onto pith or activated carbon rather than the
acidic dyes. These results agree with the literature, which
shows that activated carbon has a higher surface basicity
than surface acidity.35,39 On the other hand, the pH of the
carbon is found to be around 7.15, which shows that the
carbon surface is not protonated. This gives less evidence
for the chemisorption or ion exchange of the basic dye
(cationic dye) with the H+ ion to occur. Similar observations
Adsorption of Dye on Activated Carbon and Bagasse Pith Langmuir, Vol. 19, No. 3, 2003 727

Table 2. Constant Mass Transfer Parameters (Kf, Deff, Kb, and f) for the Adsorption of Different Dyes on Activated
Carbon
dye mass (g) C0 (mg dm-3) 104Kf (cm s-1) 1010Deff (cm2 s-1) Kb (s-1) f (-) SSE
AB80 1.7 50 4.20 4.42 2.17 10-2 0.5719 36.0
1.7 75 4.20 4.42 2.17 10-2 0.5719 38.6
1.7 100 4.20 4.42 2.17 10-2 0.5719 3.3
1.7 150 4.20 4.42 2.17 10-2 0.5719 111.3
1.7 200 4.20 4.42 2.17 10-2 0.5719 264.9
0.4 100 4.20 4.42 2.17 10-2 0.5719 6.1
0.8 100 4.20 4.42 2.17 10-2 0.5719 2.5
1.2 100 4.20 4.42 2.17 10-2 0.5719 1.4
1.7 100 4.20 4.42 2.17 10-2 0.5719 3.3
2.2 100 4.20 4.42 2.17 10-2 0.5719 3.6
AR114 1.7 50 4.69 4.73 4.09 10-3 0.5086 78.2
1.7 75 4.69 4.73 4.09 10-3 0.5086 53.4
1.7 100 4.69 4.73 4.09 10-3 0.5086 15.6
1.7 150 4.69 4.73 4.09 10-3 0.5086 40.9
1.7 200 4.69 4.73 4.09 10-3 0.5086 144.2
0.4 100 4.69 4.73 4.09 10-3 0.5086 1.3
0.8 100 4.69 4.73 4.09 10-3 0.5086 5.6
1.2 100 4.69 4.73 4.09 10-3 0.5086 1.7
1.7 100 4.69 4.73 4.09 10-3 0.5086 15.6
2.2 100 4.69 4.73 4.09 10-3 0.5086 15.5
BB69 0.637 381 6.57 2.96 2.54 10-2 0.6483 463.0
0.637 480 6.57 2.96 2.54 10-2 0.6483 109.1
0.637 596 6.57 2.96 2.54 10-2 0.6483 321.1
Total SSE ) 893.2

Table 3. Constant Mass Transfer Parameters (Kf, Deff, Kb, and f) for the Adsorption of Different Dyes on Pith
dye mass (g) C0 (mg dm-3) Kf (cm s-1) Deff (cm2 s-1) 102Kb (s-1) f (-) SSE
AR114 3.4 50 5.55 10-4 7.06 10-9 4.36 0.9999 5.8
3.4 75 5.55 10-4 7.06 10-9 4.36 0.9999 3.9
3.4 100 5.55 10-4 7.06 10-9 4.36 0.9999 7.9
3.4 125 5.55 10-4 7.06 10-9 4.36 0.9999 3.3
3.4 150 5.55 10-4 7.06 10-9 4.36 0.9999 4.5
BR22 1.7 100 9.31 10-3 1.37 10-8 1.62 0.9999 1118.1
1.7 150 9.31 10-3 1.37 10-8 1.62 0.9999 234.6
1.7 200 9.31 10-3 1.37 10-8 1.62 0.9999 21.9
1.7 250 9.31 10-3 1.37 10-8 1.62 0.9999 146.1
1.7 300 9.31 10-3 1.37 10-8 1.62 0.9999 503.8
BB69 1.7 100 4.05 10-3 1.07 10-8 1.44 0.9999 283.1
1.7 150 4.05 10-3 1.07 10-8 1.44 0.9999 280.8
1.7 200 4.05 10-3 1.07 10-8 1.44 0.9999 39.3
1.7 250 4.05 10-3 1.07 10-8 1.44 0.9999 375.7
1.7 300 4.05 10-3 1.07 10-8 1.44 0.9999 1186.4
Total SSE ) 4215.3

had also been reported.35,43 The adsorption mechanism of

the basic dye on activated carbon was attributed to the
presence of negatively charged carboxylate anionic surface
functional groups on the carbon.44
These mechanistic steps are likely to be rapid at the
surface sites. However, the rate of dye removal from
solution is most likely to be controlled by diffusion rate
processes, as discussed later.
Application of the Branched Pore Model. To
determine the model parameters for each dye-sorbent
system, the model was initially used to determine a set
of the four mass transport parameters for each experi-
mental concentration-time decay curve. An average was
then taken for each parameter for each of the different Figure 6. Effect of the initial concentration on the adsorption
of Basic Blue 69 on activated carbon (m ) 0.637 g) with constant
dye-sorbent systems. The model was then executed with Kf, Deff, Kb, and f.
these average parameter values, and the SSE values were
determined for each set of experimental data. The values trends in that the errors are greatest at lowest and highest
of the four parameters for the adsorption of different dyes C0 for both dye systems, suggesting that Deff may vary
on activated carbon and pith are shown in Tables 2 and with the initial dye concentration.
3, respectively. Since it has previously been established that surface
The results are shown in Figures 6 and 7 for the effect diffusivity can vary with liquid concentration and surface
of C0 of BB69 on activated carbon and pith, respectively. loadings,23,45-49 in the second application of the model only
The results show that both sets of data exhibit similar
(45) Gilliland, E.; Baddour, R. F.; Perkinson, G. P.; Sladek, K. J. Ind.
Eng. Chem. Fundam. 1974, 13, 95-100.
(43) Huang, C. P.; Wu, M. H. Water Res. 1977, 11, 673-679. (46) Muraki, M.; Iwashima, Y.; Hayakawa, T. J. Chem. Eng. Jpn.
(44) Mattson, J. S.; Mark, H. B. Activated Carbon Surface Chemistry 1982, 15, 34-39.
and Adsorption from Solution; Marcel Dekker: New York, 1971. (47) Suzuki, M.; Fujii, T. AIChE J. 1982, 28, 380-385.
728 Langmuir, Vol. 19, No. 3, 2003
Figure 7. Effect of initial concentration on the adsorption of
Basic Blue 69 on pith (m ) 1.7 g) with constant Kf, Deff, Kb, and
f.
Figure 8. Effect of initial concentration on the adsorption of
Basic Blue 69 on activated carbon (m ) 0.637 g) with variable
Deff and constant Kf, Kb, and f
Figure 9. Effect of initial concentration on the adsorption of
Basic Blue 69 on pith (m ) 1.7 g) with variable Deff and constant
Kf, Kb, and f.
three of the mass transport parameters were fixed (Kb, Kf,
and f) while Deff was retained as a variable. Figures 8 and
9 show the influence of varying Deff on the concentration-
time decay curves for adsorption of BB69 on carbon and
pith, respectively. The best fit parameters and SSE values
under these conditions are presented in Tables 4 and 5
for the adsorption of different dyes on activated carbon
and pith, respectively.
Comparing the SSE values in Table 5 with those in
Table 3, all the SSE values in Table 5 are very low. This
quality is reflected in the visual agreement between the
experimental and theoretical concentration decay curves
(48) Miyahara, M.; Okazaki, M. J. Chem. Eng. Jpn. 1992, 25, 408-
414.
(49) Min, Y. H.; Lee, K. S.; Yoon, C. S.; Do, L. M. J. Mater. Chem.
1998, 8, 1225-1232.
Ko et al.
shown in Figures 8 and 9. Good agreement is observed for
all ranges of carbon masses and initial dye concentration
examined. Taking the dye on pith systems as examples,
the use of a variable Deff results in a significant improve-
ment of the fit, with the SSE value decreasing from 4215
to 583. The largest improvements were observed for decay
curves measured at high initial dye concentrations.
Previous attempts to account for variable surface
diffusivities have examined the dependence on fractional
surface coverage, , which is the same form as the Darken
equation for the Langmuir isotherm:23,50
Deff,
Deff ) (31)
1-
This form did not correlate the Deff value in the present
solution, again suggesting the heterogeneous nature of
the carbon surface even though the equilibrium data could
be correlated by a Langmuir type equation.
Suzuki and Fujii47 demonstrated that the strong
dependence of the surface diffusion coefficient on the
amount adsorbed q could be interpreted in terms of the
change of the heat of adsorption with surface coverage:
Ds ) Ds0(q)n (32)
where n is the exponent in the Freundlich isotherm
However, using the Freundlich exponent based equation
still did not enable Ds to be correlated with surface
coverage. This indicates that Deff may be a complex
function dependent on both the isotherm and the adsorbed
species concentration/mobility expression.
By considering a general form of eq 32, the effective
diffusivities as surface diffusivities may be correlated by
the following expression:
Deff ) Deff,(q
j /q0) (33)
where qj is the averaged solid-phase concentration for each
set of experimental data and q0 is the saturated solid-
phase concentration in equilibrium with C0.
Figures 10 and 11 show plots of Deff against q j /q0 for a
dye-carbon and a dye-pith system, and the expressions
for the optimized diffusion coefficients as a function of
surface coverage for different dyes are given in Table 6.
The exponents based on the best fit Freundlich equation
for BB69 on carbon and pith are 3.85 and 4.17, respectively;
the best fit exponents, namely, 2.46 and 2.50, reflect the
poorness of fit of the Freundlich equation for these systems.
The deviation from the proposed equations in Figures 10
and 11 may be due to the fact that the points are not
individual Deff values for individual experimental data
points but rather averaged Deff values for final q j /q0 values
for experimental systems, not individual points.
Mass Transfer Processes. The results in Table 4
suggest that the adsorption of both acid and basic dyes
onto activated carbon is a predominantly surface diffusion
mechanism with a concentration dependent variable
surface diffusivity. The diffusion process occurs throughout
a complex structure of macropores and micropores with
a fractional adsorptive capacity from 0.50 to 0.65 for the
different dye carbon systems. The exact value will depend
on a number of factors such as dye molecular size, dye-
carbon affinity, and surface loadings.
Table 5 shows the results for the three dye-pith
systems. In the case of the acid dye AR114 the diffusivity
(50) Darken, L. S. Trans. Am. Inst. Min., Metall. Pet. Eng. 1948, 175,
184-201.
Adsorption of Dye on Activated Carbon and Bagasse Pith Langmuir, Vol. 19, No. 3, 2003 729

Table 4. Variable Deff Optimization Results for the Adsorption of Different Dyes on Activated Carbon
dye mass (g) C0 (mg dm-3) 104Kf (cm s-1) 1010Deff (cm2 s-1) Kb (s-1) f (-) SSE
AB80 1.7 50 4.20 2.61 2.17 10-2 0.5719 1.5
1.7 75 4.20 3.02 2.17 10-2 0.5719 1.2
1.7 100 4.20 4.07 2.17 10-2 0.5719 0.8
1.7 150 4.20 6.94 2.17 10-2 0.5719 3.6
1.7 200 4.20 8.49 2.17 10-2 0.5719 2.1
0.4 100 4.20 2.55 2.17 10-2 0.5719 0.5
0.8 100 4.20 3.77 2.17 10-2 0.5719 0.5
1.2 100 4.20 4.15 2.17 10-2 0.5719 0.6
1.7 100 4.20 4.07 2.17 10-2 0.5719 0.8
2.2 100 4.20 4.17 2.17 10-2 0.5719 1.5
AR114 1.7 50 4.69 1.85 4.09 10-3 0.5086 3.2
1.7 75 4.69 2.79 4.09 10-3 0.5086 1.1
1.7 100 4.69 3.66 4.09 10-3 0.5086 1.7
1.7 150 4.69 6.75 4.09 10-3 0.5086 3.2
1.7 200 4.69 8.89 4.09 10-3 0.5086 2.5
0.4 100 4.69 4.82 4.09 10-3 0.5086 1.3
0.8 100 4.69 5.21 4.09 10-3 0.5086 5.0
1.2 100 4.69 4.69 4.09 10-3 0.5086 1.7
1.7 100 4.69 3.66 4.09 10-3 0.5086 1.7
2.2 100 4.69 3.88 4.09 10-3 0.5086 1.9
BB69 0.637 381 6.57 2.34 2.54 10-2 0.6483 78.9
0.637 480 6.57 3.03 2.54 10-2 0.6483 106.0
0.637 596 6.57 3.51 2.54 10-2 0.6483 63.3
Total SSE ) 281.9

Table 5. Variable Deff Optimization Results for the Adsorption of Different Dyes on Pith
dye mass (g) C0 (mg dm-3) Kf (cm s-1) Deff (cm2 s-1) 102Kb (s-1) f (-) SSE
AR114 3.4 50 5.55 10-4 8.11 10-9 4.36 0.9999 4.9
3.4 75 5.55 10-4 6.69 10-9 4.36 0.9999 3.5
3.4 100 5.55 10-4 6.52 10-9 4.36 0.9999 6.3
3.4 125 5.55 10-4 6.66 10-9 4.36 0.9999 2.2
3.4 150 5.55 10-4 7.31 10-9 4.36 0.9999 4.0
BR22 1.7 100 9.31 10-3 5.78 10-9 1.62 0.9999 1.3
1.7 150 9.31 10-3 9.62 10-9 1.62 0.9999 2.8
1.7 200 9.31 10-3 1.45 10-8 1.62 0.9999 16.0
1.7 250 9.31 10-3 1.73 10-8 1.62 0.9999 28.8
1.7 300 9.31 10-3 2.13 10-8 1.62 0.9999 79.0
BB69 1.7 100 4.05 10-3 6.63 10-9 1.44 0.9999 66.0
1.7 150 4.05 10-3 7.99 10-9 1.44 0.9999 16.4
1.7 200 4.05 10-3 1.00 10-8 1.44 0.9999 20.7
1.7 250 4.05 10-3 1.31 10-8 1.44 0.9999 149.6
1.7 300 4.05 10-3 1.59 10-8 1.44 0.9999 181.2
Total SSE ) 582.6

Figure 10. Relationship between the optimized diffusion Figure 11. Relationship between the optimized diffusion
coefficients and surface coverage for BB69 on activated carbon. coefficients and surface coverage for BR22 on pith.

is more or less constant at 7.06 ( 1.05 cm2 s-1. However, capacities of the carbon for three dyes by utilizing the
with the basic dyes the diffusivities show a distinct trend micropores. The adsorption capacities, based on qmax )
with dye capacity, suggesting a more complex surface
diffusion mechanism, probably involving interactions with Table 6. Relationships between the Optimized Diffusion
Coefficients and Surface Coverage for Different
the surface groups of the bagasse pith. The fractions, f,
Dye-Adsorbent Systems
of macropores in the pith systems are all effectively unity.
This indicates a total absence of micropores in the pith diffusion coefficient-surface coverage
structure. This is not surprising with such a cellulose based adsorbent dye expression
material, which swells in water and behaves like a sponge. carbon BB69 Deff ) (6.71 10-9)(q
j /q0)2.46
However, it does show an additional use of this multipore carbon AB80 Deff ) (1.98 10-8)(q
j /q0)2.07
carbon AR114 Deff ) (2.90 10-8)(q
j /q0)2.16
model for its ability to distinguish between the different pith BB69 Deff ) (5.03 10-8)(q
j /q0)2.50
pore structures of different adsorbents. This observation pith BR22 Deff ) (5.98 10-8)(q
j /q0)3.41
is further supported by the much higher adsorption pith AR114 Deff ) (1.76 10-8)(q
j /q0)1.28
730 Langmuir, Vol. 19, No. 3, 2003
KL/aL mg of dye/g of adsorbent, on carbon are 878 and 106
mg of dye/g of carbon for Basic Blue 69 and Acid Red 114,
respectively, whereas on pith the capacities are 160 and
22 mg of dye/g of pith for Basic Blue 69 and Acid Red 114,
respectively. The differences in the capacities reflect the
large microporous surface area of activated carbon.
5. Conclusions
From the present work the following points can be made.
1. The branched-pore model has been shown to be
applicable over a wide range of experimental conditions
for the adsorption of the different dyes on activated carbon
and pith.
2. Single values for the diffusion parameters Kf, Deff,
Kb, and f can be used to predict the concentration versus
time decay curves over much of the range of masses and
concentrations of each dye system, with reasonable
accuracy.
3. A significantly better fit of the model predictions is
obtained when Kf, Kb, and f are maintained constant but
Deff is varied. This indicates that the surface diffusivity
may vary as a function of either the liquid- or solid-phase
concentration.
4. The effective diffusion coefficients obtained by
minimization of the SSE value can be correlated with a
reasonable degree of accuracy by a function of surface
coverage.
The model can be used to identify the structure of the
adsorbent, in terms of the macropore and micropore
adsorptive capacities, and the extent of each capacity
fraction.
Acknowledgment. One of the authors, D.C.K.Ko, is
grateful to DAG and RGC, Hong Kong SAR, for the
provision of financial support during this research pro-
gram.
Nomenclature
A ) total surface area of particles (cm2)
aL ) Langmuir isotherm constants (dm3 mg-1)
Ce ) equilibrium fluid-phase concentration (mg dm-3)
C0 ) initial liquid-phase solute concentration (mg dm-3)
Cs,t ) liquid-phase concentration at particle surface at
time t (mg dm-3)
Ct ) liquid-phase concentration at time t (mg dm-3)
dp ) diameter of the particle (cm)
Deff ) effective diffusivity (cm2 s-1)
Ds ) surface diffusion coefficient (cm2 s-1)
f ) fraction of total adsorptive capacity in macropores
(dimensionless)
k ) intraparticle mass transfer coefficient (cm s-1)
Kb ) branched pore rate coefficient (s-1)
Kf ) external liquid film mass transfer coefficient (cm
s-1)
Ko et al.
KF ) Freundlich isotherm constants (dm3 g-1)
KL ) Langmuir isotherm constants (dm3 g-1)
m ) mass of adsorbent (g)
n ) Freundlich isotherm constants (dimensionless)
qb ) solid-phase concentration in micropore region (mg
g-1)
qe ) equilibrium solid-phase concentration (mg g-1)
qm ) solid-phase concentration in macropore region (mg
g-1)
q0 ) solid-phase concentration in equilibrium with C0
(mg g-1)
qs ) solid-phase concentration at particle surface (mg
g-1)
r ) radial distance from center of particle (cm)
R ) radius of particle (cm)
Rb ) rate of transfer of solute from macropore to micropore
region (mg g-1 s-1)
SSE ) sum of squares errors ) i)1 n
(Cexp,i - Ctheor,i)2
t ) time (s)
vol ) volume of batch solution (dm3)
V ) volume of particles (cm3)
C ) dimensionless liquid-phase concentration ()Ct/C0)
Cn ) value of C at nth grid point of mesh
Cs ) dimensionless liquid-phase concentration at particle
surface ()Cs,t/C0)
Qm ) dimensionless solid-phase concentration in macro-
pore region ()qm/q0)
Qb ) dimensionless solid-phase concentration in mi-
cropore region ()qb/q0)
Qs ) dimensionless solid-phase concentration at particle
surface
PBiot ) modified particle Biot number ()KbR2/fDeff)
FBiot ) modified film Biot number ()KfC0R2/FfDeffq0)
SFact ) separation factor ()VC0/mq0)
M ) last grid point of mesh corresponding to particle
surface
Qnp ) value of Qm at pth mesh point of grid and nth mesh
point of grid
Qh np ) value of Qb at pth mesh point of grid and nth mesh
point of grid
S1 ) /
S2 ) (PBiot)f/[2(1 - f)]
S3 ) 3f(FBiot)/2(SFact)
S4 ) (FBiot)/2
Greek Letters
) constant defined in eq 29 (g mg-1)
) surface coverage ()q/q0)
) reduced radial variable ()r/R)
F ) adsorbent particle density (g cm-3)
) dimensionless time variable ()Defft/R2)
) spacing of finite difference mesh in variable
) transformed radial variable ()2)
) spacing of finite difference mesh in variable
LA020654G