Carbon 42 (2004) 20092017

www.elsevier.com/locate/carbon

Adsorption of phenanthrene on activated carbons:

Breakthrough curve modeling
R. Murillo, T. Garc
a, E. Ayl

on, M.S. Call
en, M.V. Navarro, J.M. L

opez, A.M. Mastral

Department of Energy and Environment, Instituto de Carboqu
mica, CSIC, Maria de Luna, M Luesma Castan 4, 50015 Zaragoza, Spain
Received 26 January 2004; accepted 5 April 2004
Available online 4 May 2004

Abstract
The modeling of breakthrough curves obtained in the adsorption of phenanthrene as model compound on dierent activated
carbons is described. All the runs were performed in a xed bed reactor at atmospheric pressure, with a process temperature of 150

C and low contaminant concentrations. Therefore, experimental conditions were similar to the observed in the ue gases of energy
generation systems. This work is mainly focused on the study of how adsorbent characteristics (surface area and micropore size
distribution) inuence the kinetics of the adsorption process. First, equilibrium values are found from the breakthrough curves and
they are satisfactory tted to the Freundlich isotherm. Using the obtained equilibrium values together with the linear driving force
model as kinetic expression, the breakthrough curve modeling is achieved. It was found that this model ts all the breakthrough
curves and it is a useful tool for modeling purposes. Values for the phenanthrene surface and eective diusion coecients are
calculated and reported, and a relationship with the microporosity is found. As it was expected, it is observed that the phenanthrene
molecule nds kinetic restrictions for the diusion in those carbons with narrow microporosity, especially in those with a mean pore
diameter close to the molecular size.
 2004 Elsevier Ltd. All rights reserved.
Keywords: A. Activated carbon; B. Coke; C. Adsorption properties; D. Modeling

1. Introduction
Polycyclic aromatic hydrocarbons (PAH) are an
important family of environmental pollutants that are
mainly generated during combustion processes [13].
PAH emissions, due to their high volatility and reactivity, can be released, not only supported onto particulate matter (PM) but also in the gas phase [4]
downstream the dust collection systems. PAH supported
on the PM can be easily trapped using the proper systems such as cyclones, electrostatic precipitators, fabric
lters, etc. However, the atmospheric PAH emission
control becomes more dicult.
Dierent technologies for gas-phase emissions control have been developed during the last years [5,6]. The
two technologies that are the most promising alternatives to reduce gaseous PAH emissions are: catalytic
PAH destruction [7] and PAH adsorption/concentration
*

Corresponding author. Tel.: +34-976-733-977; fax: +34-976-733318.

E-mail address: amastral@carbon.icb.csic.es (A.M. Mastral).
0008-6223/$ - see front matter  2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2004.04.001

on carbonaceous materials [8]. Activated carbons are

extensively used to remove contaminants from exhaust
gases [911]. The gas is generally passed through a bed
of activated carbon where the contaminant molecules
are transferred to the solid phase (adsorption process).
Information about the adsorption of pure organic vapors [1216] for environmental purposes has been recently published. In these previous works, it was shown
that the adsorption of PAH by sorbents from waste hot
gas emissions is inversely proportional to their volatility.
These studies also indicated that other chemical species
in the gas stream like moisture [14] and CO2 [15] have a
negative inuence on phenanthrene (Phe) adsorption
capacity. First, the determined parameters showed that,
for a carbonaceous material, the higher the steam percentage (in volume) in the gas stream, the lower its Phe
adsorption capacity; and second, it was also observed
the existence of a negative inuence of CO2 in the
adsorption capacity of Phe. That is probably due to a
competitive eect between CO2 , H2 O and Phe molecules
for the primary adsorption sites within the carbonaceous
material and also the increase of the surface hydrolity.

2010

R. Murillo et al. / Carbon 42 (2004) 20092017

Nomenclature
C0
Ct
De
Dk
Dm
Ds
K
Kf
kf
kp
L
LT
M
nf
p
p0
q
Q
r0
Rp
SBET
tf
tr
T
u

gas inlet concentration (mol/m3 )

gas outlet concentration (mol/m3 )
eective diusivity (m2 /s)
Knudsen diusivity (m2 /s)
molecular diusivity (m2 /s)
surface diusivity (m2 /s)
equilibrium constant
Freundlich equation constant
uid mass-transfer coecient (m/s)
intrapellet mass-transfer coecient (m/s)
bed length (cm)
height of the mass-transfer zone (m)
molecular weight (g/mol)
Freundlich equation exponent
nitrogen partial pressure (Pa)
nitrogen saturation pressure (Pa)
nitrogen adsorption capacity (cm3 (STP)/g)
total inlet ow (m3 /s)
mean pore radius (m)
equivalent radius of pellet (m)
surface area calculated according to BET
isotherm (m2 /g)
bed saturation time (s)
breakthrough time (s)
temperature (K)
supercial velocity (m/s)

Up to now, the quantitative studies of PAH adsorption on activated carbons are limited to apply dierent
adsorption isotherm models in a xed packed bed [17].
However, to improve the industrial design of these
adsorbents performance and to nd the best operating
conditions, the adsorption processes need to be modeled. Numerous models have been published in recent
years to describe the phenomena of xed bed adsorption
[1820]. Model equations are conservation equations (in
this case only mass conservation applies), kinetic law at
the uid/solid interface (adsorption equilibrium isotherm), kinetic law of mass transfer, boundary conditions and initial conditions.
Several publications on the adsorption modeling of
volatile organic compounds (VOC) on activated carbons
and zeolites can be found in the literature [2124]. In
these references, the transport properties of dierent
small organic molecules like acetone, toluene, methyl
ethyl ketone, heptane and octane are evaluated tting
the experimental results with the linear driving force
(LDF) model. Hence, Brosillon et al. [21] investigated
the kinetic performance of zeolites for VOC abatement
observing that a constant value of eective diusivity is
found independent of the nature and the amount of

VCO2
VN 2
VT
w
W
W
W0

Wtr

narrow micropore volume (cm3 /g)

total micropore volume (cm3 /g)
total pore volume (cm3 /g)
carbonaceous material introduced into the
reactor (kg)
equilibrium adsorption capacity (mol/kg)
concentration adsorbed on the surface of
adsorbent (mol/kg)
adsorbate concentration in the solid in equilibrium with the gas inlet concentration (mol/
kg)
adsorption capacity at the breakthrough
(mol/kg)

Dimensionless numbers
uq R
Re
Reynolds number lg p
l
Sc
Schmidt number qg D
m
Sh

Sherwood number

Rp kf
Dm

Greek symbols
e
porosity of bed
porosity of particle
ep
gas density (kg/m3 )
qg
bed density (kg/m3 )
ql
sp
pore tortuosity
surface tortuosity
ss
l
gas viscosity (Pa s)

VOC adsorbed. Malek and Farooq [24] studied the kinetic behavior of dierent small VOC molecules in their
adsorption on activated carbon and silica gel nding
that the adsorbate transport inside the adsorbent is
essentially by Knudsen and surface ow. However, none
of these authors have focused on the PAH adsorption
modeling and the relationship between kinetic and
equilibrium parameters with the structural characteristics of the adsorbent. In this work, the successful
application of the LDF model to VOC adsorption is
used to predict the breakthrough curves of PAH on
activated carbons at low concentrations. In addition, the
relationship between kinetic parameters and structural
properties is also assessed.

2. Theory
The mathematical model of the isothermal, dynamic
adsorption breakthrough process in a xed bed is based
on transient material balance, gas phase and intraparticle mass transfer, the adsorption equilibrium relationship, boundary conditions and initial conditions. In
this model, both the adsorption equilibrium and kinetic

R. Murillo et al. / Carbon 42 (2004) 20092017

aspect are taken into account. The model assumptions

are: the pressure drop throughout the reactor is very
low; the system is isothermal for low Phe concentrations; the vapor is considered as an ideal gas; and the
plug ow is assumed. According to all these assumptions
the model equations are as follows:
Mass balance on xed bed. If the ow rate is constant
and the eect of axial dispersion is neglected, the mass
balance on a portion of the bed is given by
oC
oW t
oC
q1
u
0
e
ot
ot
oz

Intraparticle mass transfer. Mass transfer can be described by the LDF model [25]. This model is frequently
used for this purpose because it is analytical, simple and
physically consistent. According to the LDF model, the
rate of adsorption of a single adsorbate into an adsorbent particle is given by
oW
kp W   W
ot

where the kp parameter is used as an adjustable

parameter and W  is the adsorbate concentration in the
adsorbent surface (mol/kg).
Equilibrium model. The adsorption capacities were
deduced from equilibrium adsorption isotherms using
the Freundlich isotherm model [19]. The validity of this
model for the representation of the Phe equilibrium data
in terms of the inlet concentration will be later explained
in detail.
W Kf  C 1=nf

Initial conditions.
W 0

for z P 0 and t 0

C C0
C0

for z 0 and t P 0
for z > 0 and t 0

Estimation of mass-transfer coecient, kp . This

parameter can be obtained by tting simulation results
and experimental data. The LDF mass-transfer coecient is related to mass-transfer resistance as follows:
R2p
1
Rp

K
K
15ep De
kp 3kf

where K is the equilibrium constant.

K

W0 q
C0

The uid-phase mass-transfer coecient, kf , was

calculated from the following correlation proposed by
Wakao and Funazkri [26]:
Sh 2 1:1  Sc1=3  Re0:6

where Sh, Sc and Re are respectively the Sherwood,

Schmidt and Reynolds numbers. kf parameter can be
directly obtained from the Sh number because the rest of

2011

the parameters are physical constants or known process

values.
The eective diusivity (De ) was determined from Eq.
(4) using the value of kp and it is related to molecular,
Knudsen, and surface diusivities by means of [21]
1
sp
1

D
1ep
k
De Dm

sp

ep

Ds
ss

where the Knudsen diusivity value is [19]

r
T
Dk 9700r0
M

Fitting methodology. kp was determined by tting the

experimental data with the mathematical model by using
a computer program specically developed for this
purpose. The numerical method used was nite dierence method with a forward nite dierence explicit
scheme. Eqs. (1)(3) were simultaneously solved taking
into account the initial conditions. Dierent kp values
were supposed and the computer program was run
obtaining a theoretical breakthrough curve. Then, the
experimental and the theoretical data were compared
and the correlation coecient was calculated. Correlation coecients close to 0.99 were obtained for all the
breakthrough curves. Once the optimum kp value was
found, the only unknown variable in Eq. (4) was De
because the rest of the variables are input data, like Rp
or ep , or calculated, like K through Eq. (5) or kf through
Eq. (6). In the same way, once De has been calculated
Eq. (7) is used to deduce the value of Ds , the only unknown variable. In Eq. (7), in addition to the input data,
like ep and Dm , and the calculated variable, Dk , it is
necessary to suppose the values of the pore and surface
tortuosities. A value of 4 was adopted for both parameters because it is a common value used in the
adsorption of hydrocarbons [21,24,27].

3. Experimental
Three dierent activated carbons were used to obtain
their Phe adsorption isotherms. All materials were provided to their direct application in PAH adsorption.
ICB5 was a coke from German Rhenish lignite supplied
by RWE Rheinbraun. The supplier did not inform
about the activation procedure. ICB8 was an activated
carbon from apricot stones obtained in a two step process. In the rst step, apricot stones were carbonized in
N2 at 800
C for 2 h. In the second step, the carbonization product was activated with steam at 800
C for 2
h and a steam ow rate of 10 g/min. The heating rate
was 20
C/min for both steps. Finally, ICB12 was an
activated carbon from Spanish lignite. Activated carbon
preparation was carried out by carbonization of the
original raw material, followed by activation in CO2 .

2012

R. Murillo et al. / Carbon 42 (2004) 20092017

Table 1
Activated carbons textural parameters
Sample

SBET (m2 /g)

VN2 (cm3 /g)

VCO2 (cm3 /g)

VT (cm3 /g)

VCO2 =VN2

ICB5
ICB8
ICB12

273
636
574

0.12
0.33
0.29

0.11
0.28
0.25

0.20
0.40
0.33

0.92
0.85
0.86

250

q [cm3 (STP)/g]

200

150

100
ICB5
ICB8

50

ICB12
0
0.00

0.20

0.40

0.60

0.80

p/p0

Fig. 1. Activated carbons nitrogen adsorption isotherms.

1.00

0.05

Pore Volume (cm3/g/nm)

Both, carbonization and activation reactions, were carried out in a horizontal furnace. The carbonization was
conducted in N2 ow (80 ml/min) at 850
C with a
heating rate of 5
C/min for 2 h. The activation process
was carried out at the same temperature and heating
rate and the activating agent used was CO2 with a ow
rate of 80 ml/min for 1 h.
The activated carbons used in this work were characterized through their morphological properties and
the obtained data are compiled in Table 1. The three
porous materials were characterized by N2 and CO2
adsorption at 77 and 273 K (see Fig. 1), respectively,
using a Quantachrome Autosorb 1 analyzer. Their total,
VN2 , and narrow, VCO2 , micropore volumes were calculated by applying the DubininRadushkevich (DR)
equation to the N2 and CO2 isotherm data, respectively.
The total pore volume (VT ) was obtained from the N2
volume adsorbed at a relative pressure of 0.95. Micropore size distributions were calculated by applying the
density functional theory (DFT) to the N2 isotherm data
(see Fig. 2).
Phe adsorption experiments were conducted using the
experimental rig which was described in detail in previous publications [17]. The experimental system was used
to generate gas streams containing Phe at the required
concentration. The Phe gas generation was obtained by
sublimating reagent grade solid Phe contained in a
quartz cylindrical saturator (6.8 mm ID) into a stream
of pure helium (10 ml/min). The saturator was kept at a
xed temperature by a cylindrical ceramic furnace driven by a PID temperature controller (1
C accuracy),

ICB8

0.04

ICB12
ICB5

0.03
0.02
0.01
0.00
0.0

1.0

2.0

3.0

4.0

Pore Width (nm)

Fig. 2. Micropore size distribution obtained by DFT procedure.

which allows controlling the inlet Phe concentration in

each run.
Before starting each experiment, it was necessary to
reach a constant and known concentration of Phe in the
adsorbent tubes inlet stream, by passing the saturator
outlet gas stream to the ame ionization detector (FID)
until these concentrations were obtained. When these
were reached, the gas stream was driven through the
adsorbent tubes starting the reaction time, which lasted
until saturation was reached (C C0 ). The FID detector
was previously calibrated and the linear range determined. All the concentrations used in the adsorption test
were inside this linear range.
00
The adsorbent tubes used were 1/4 teon tubes that
contained a bed composed of 25 mg of adsorbent (100
200 lm average particle size diameter) and 2.0 g of sand.
The mixture was held in place by two plugs of silanized glass wool, to provide enough bed length (9 cm)
assessing in this way a uniform ow throughout the
reactor. Blank tests were carried out to check the inert
material adsorption capacity. The temperature of the
adsorption reactor was 150
C and xed with an accuracy of 1
C inside a chromatographic furnace. The
carrier gas ow rate was controlled with a mass ow
controller and had a value of 10 ml/min in all the runs.
Adsorption isotherms were constructed from the
obtained breakthrough curves. Equilibrium adsorption
capacities, W (mol Phe/kg carbonaceous material), were
calculated by integrating the corresponding breakthrough curves and by applying Eq. (9).


Z tf
Q
C0 tf 
Ct dt
9
W
w
0

R. Murillo et al. / Carbon 42 (2004) 20092017

4. Results and discussion

The morphological characteristics of the activated

carbons used were calculated from their N2 and CO2
adsorption isotherms. Fig. 1 contains the N2 adsorption
isotherms obtained for the studied materials indicating
the dierences in the porosity between them. It is observed that these adsorption isotherms are close to type I
of the BDDT classication [19] typical of microporous
carbon materials. The N2 adsorption isotherms of these
activated carbons also show a H4 type hysteresis loop
[28], characteristic of the slit-shaped pores where the
adsorption and desorption branches are parallel. The
textural parameters of these activated carbons calculated from their N2 and CO2 adsorption isotherms are
collected in Table 1. It can be observed that ICB8 has
the highest porosity whereas ICB5 shows the lowest.
The same trend is shown for the total micropore volumes, VN2 , the narrow micropore volumes, VCO2 , the
total pore volumes, VT , the apparent surface area, SBET ,
and also observed in their N2 adsorption isotherms.
More details about the dierences in micropore sizes can
be deduced from the analysis of the micropore size
distributions obtained according to the DFT procedure
(see Fig. 2). It is observed that ICB5 has mainly a unimodal distribution with the majority of pores between
0.7 and 1.4 nm, where the PAH retention could be favored by the overlapping of the adsorption potentials.
ICB8 and ICB12 have a wide unimodal distribution with
a signicant volume in pores between 1.4 and 2 nm. The
Phe molecular size has been calculated by using Materials Studio software, and a value of 0.6 nm has been
obtained. Therefore micropores of this size or even twice
or three times the size of this molecule, where the
adsorption potentials overlapping occurs, will be
appropriate for the adsorption process. It is observed in
DFTs (Fig. 2) that most of the micropores of the three
carbon materials would be within this range (0.61.8
nm) and therefore, they would be appropriate materials
for Phe adsorption.
4.2. Modeling of the Phe adsorption isotherms
The Phe adsorption isotherms obtained at 150
C of
the three activated carbons above described are shown
in Fig. 3. These Phe isotherms are close to type I of the
BDDT classication as it was observed for N2 adsorption isotherms. Thus, the initial part of the adsorption
isotherm represents micropore lling and the slope of
the plateau at higher concentrations is due to multilayer
adsorption on the non-microporous surface. It is also
observed that adsorbents with a high SBET and high
microporosity (ICB8 and ICB12) improve the Phe
retention because higher adsorption capacities are found

3.0
2.5

W (mol Phe/kg)

4.1. Activated carbons characterization

2013

2.0
1.5
ICB5
ICB8
ICB12

1.0
0.5
0.0
0

0.02

0.04

0.06

0.08

0.1

Co (mol/m3)
Fig. 3. Phe adsorption isotherms for the three activated carbons.

at the same Phe inlet concentration. It is also observed

that Phe adsorption takes place mainly in the micropores, whose diameter is close to the Phe molecular size
because the adsorption potential is higher due to the
proximity of pore walls [12].
Three dierent models were applied to try to t the
experimental results: Langmuir, Freundlich and DR.
According to previous published results [17], all the
obtained isotherms were properly tted to the Freundlich adsorption model. The corresponding isotherm
parameters (see Eq. (3)) were determined by regression
analysis obtaining a linear expression for the Freundlich
equation. Generally, the parameter nf has a value
greater than unity, showing a favorable adsorption onto
the microporous adsorbent. The magnitude of Kf is related to the adsorbent capacity. This model is able to t
the experimental data at the studied conditions due to
the fact that it allows for the average of dierent multilayer adsorption energies and dierent surface
adsorption energies, which are both, aected by the size
and distribution of pores in an adsorbent. Table 2 shows
the parameters nf and Kf and correlation coecients
obtained by applying this model to the Phe isotherm
data. It is observed that the nf values for the three
activated carbons are greater than unity demonstrating
that the Phe adsorption at the studied experimental
condition is favorable. With respect to the Kf parameter
(see Table 2), it is observed that the minimum value is
the one corresponding to the ICB5, which at the same
time, it is the carbon material with a lower surface area
and microporosity. The other two carbon materials
(ICB8 and ICB12) have a similar value for Kf according
to their similar morphological properties (see Table 1).

Table 2
Freundlich isotherm adsorption parameters
Sample

Kf (m3 /kg)

nf

r2

Range (mol/m3 )

ICB5
ICB8
ICB12

1.9
3.5
3.8

6.2
8.3
6.2

0.97
0.96
0.99

1.71 104 1.35 102

1.48 103 7.40 102
1.84 103 8.14 102

2014

R. Murillo et al. / Carbon 42 (2004) 20092017

4.3. Breakthrough curves modeling

Once the performance of the activated carbons in Phe
retention has been demonstrated, it is needed to model
mathematically the adsorption processes to predict the
adsorbent behavior at dierent experimental conditions.
Several theoretical complex models have been proposed
in the literature [20,25] to simulate breakthrough curves.
In this paper, a simple model, the linear driving force
model has been chosen to t the experimental data because it has physical consistence and, in addition, due to
its simplicity and accuracy in reproducing the adsorption of organic molecules on dierent materials, both
organic [26] and inorganic [21]. Fig. 4 shows the experimental breakthrough curves obtained for each adsorbent at a similar Phe inlet concentration (C0 2 ppm).
The theoretical curves calculated according to the LDF
model are also shown (continuous line). It can be observed that both curves, theoretical and experimental,
are very close. Thus, the model provides a good prediction of the experimental breakthrough curves observed during the Phe adsorption in all the range of
concentrations studied. Fig. 4 also shows that the higher
the total pore volume, the higher the breakthrough time
and the better the performance of the activated carbon.
Therefore, not only the presence of micropores is fundamental for the adsorption process but also the macro
and mesoporosity increases the adsorption capacity
probably due to cooperative eects between molecules
and multilayer formation. Table 3 shows the column
characteristics obtained from the Phe breakthrough
1.2

ICB8

1.0

ICB12
C/Co

0.8

ICB5

0.6

0.4

0.2

0.0
0

10000

20000

30000

40000

50000

60000

70000

80000

t (s)
Fig. 4. Phe breakthrough curves for the three activated carbons
(C0 0:2 ppm).

curves plotted in Fig. 4 in terms of the carbon material

used. In this Table the time needed for breakthrough
(time when the outlet concentration is 2% of the inlet
concentration, tr ), the nal time (time when the outlet
and inlet concentrations are identical, tf ), the height of
the mass-transfer zone (LT ), the adsorption capacity at
the breakthrough (Wtr ), the equilibrium adsorption
capacity (W ) and nally the degree of column utilization
(Wtr =W ) are shown. It can be observed that the higher
the activated carbon surface area and porosity, the
lower the LT . A signicant correlation (r2 0:99) between the LT and the ratio VCO2 =VN2 has been found.
Therefore, the higher the VCO2 =VN2 ratio, the narrower
the microporosity and the higher the kinetic restrictions.
As it was expected, Phe molecules nd more diculty in
diusing in those activated carbons with a narrow microporisity causing a decrease in the adsorption kinetics
and the LT enlargement. Regarding to the degree of
column utilization it is observed that it is very high for
all the adsorbents and no relationship was found between this parameter and the textural properties of the
carbon materials.
Results on mass-transfer coecients and diusivities
can be obtained by simulation and they are compiled in
Table 4. According to Eq. (4) and taking into account
those values in Table 4, it is possible to calculate the
magnitudes of the external and intraparticle masstransfer resistances. It was observed that at the operating conditions used in this research and for the three
carbon materials tested, the uid mass-transfer resistance is quite small and may be neglected for all practical purpose [26]. Moreover, it has been demonstrated
that in the adsorption of hydrocarbons on activated
carbon, the intraparticle diusion in the pores takes
place predominantly through a combination of Knudsen
diusion and surface diusion [26,27]. Therefore, the
intrinsic characteristics of the carbon materials will be
the determinant factor for Phe adsorption, more than
bed properties like gas physical properties and ow, bed
porosity, bed geometry, etc. These intrinsic characteristics of the carbon materials will mainly inuence to the
eective diusion coecient because it comprises the
Knudsen diusion coecient (see Eq. (7)), which depends on the average pore size (see Eq. (8)), and the
surface diusion coecient, which is typical for each
adsorbateadsorbent couple.
The eective diusivity values were calculated with
Eqs. (4), (6) and (7), using the kp value as an adjustable

Table 3
Column characteristics from breakthrough curves of Phe in terms of carbon material (L 9 cm, C0
2 ppm. Total gas ow 10 ml/min)
Adsorbent

tr (s)

tf (s)

LT (cm)

Wtr (mol/kg)

W (mol/kg)

Wtr =W

ICB5
ICB8
ICB12

25 818
67 682
47 590

28 229
72 666
51 376

0.77
0.62
0.66

0.600
1.650
1.350

0.613
1.669
1.37

0.98
0.99
0.98

R. Murillo et al. / Carbon 42 (2004) 20092017

2015

Table 4
Experimental and calculated mass-transfer parameters for Phe adsorption in terms of the carbon material
kp (s1 )

C0 (ppm)

Adsorbent
ICB5
ICB5
ICB5
ICB5
ICB5
ICB8
ICB8
ICB8
ICB8
ICB12
ICB12
ICB12
ICB12

0.00040
0.00097
0.0025
0.014
0.060
0.0015
0.0070
0.028
0.074
0.0018
0.0064
0.037
0.085

Dk (m2 /s)
7

7.25 10
7.25 107
7.25 107
7.25 107
7.25 107
7.25 107
7.25 107
7.25 107
7.25 107
7.25 107
7.25 107
7.25 107
7.25 107

0.001
0.0015
0.002
0.0025
0.004
0.001
0.002
0.003
0.005
0.001
0.002
0.0035
0.0045

parameter. Dierent kp values were tested until the

correlation coecient r2 was higher than 0.99. The
variation of eective diusivity versus amount adsorbed
for each carbon material is shown in Fig. 5. It can be
observed that those carbon materials with the widest
micropore size distributions, ICB8 and ICB12 (see
DFT representations in Fig. 2), show similar and higher
eective diusivity values. This fact suggests a clear
relationship between carbonaceous material micropore
size distribution and diusion coecients, in such a way
that the wider the micropore size distribution, the lower
the kinetic restrictions for Phe adsorption and therefore,
the higher the eective diusion coecient.
Fig. 5 also shows the inuence of the protocol used
for the production of the dierent carbon materials on
the eective diusion coecient. It is observed that
ICB5 is the adsorbent showing the lowest eective diffusivity values. This carbon material, which has not
been activated, has a narrower micropore size distribution (see Fig. 2) and, therefore, the Phe molecules nd
more diculty in diusing into the particles increasing

0.5

1.5

2.5

-12.0
-12.5
-13.0
Ln(De)

-13.5
-14.0
-14.5
-15.0
-15.5

Activated with steam (ICB8)

Activated with CO2 (ICB12)
Not activated (ICB5)

-16.0
W (mol Phe/kg)
Fig. 5. Variation of eective diusivity versus amount adsorbed for
three carbon materials.

De (m2 /s)
6

3.99 10
2.94 106
1.87 106
1.05 106
4.16 107
2.60 106
1.32 106
5.72 107
4.09 107
1.80 106
1.27 106
5.07 107
2.51 107

Ds (m2 /s)
4.56 1012
5.65 1012
6.25 1012
6.48 1012
6.65 1012
6.48 1012
1.04 1011
1.15 1011
1.54 1011
3.83 1012
6.91 1012
8.44 1012
1.02 1011

the resistance to adsorption. However, it is observed

that ICB8 and ICB12, which have been activated with
steam and CO2 respectively, have clearly higher eective
diusion coecients than ICB5. Therefore, activation
process not only increases the Phe adsorption capacity
because of the total porosity improvement, but also
improves the kinetics of Phe adsorption on the carbon
materials. It is also observed that the kinetics of Phe
adsorption on ICB5 are more negatively aected by the
already adsorbed Phe than for ICB8 and ICB12 since
their slopes are lower. Regarding to the activating agent
inuence, it has not been found any signicant dierence
on the posterior Phe adsorption kinetics when the activating agent is steam or CO2 .
With respect to surface diusion, it should be remarked that as a result of the concentration gradient in
the pores, a surface gradient also exists on the pore wall
along the gas-phase concentration gradient, assuming
that the adsorption equilibrium is rapidly established.
The surface ux associated with the surface concentration gradient is referred to as surface diusion. Thus, the
gas-phase diusion in pore and the surface diusion on
the pore wall proceed in parallel [29]. Table 4 shows the
surface diusion coecients obtained for the three carbon materials at every initial Phe concentration. It is
observed that, as it was expected [19], the higher the Phe
concentration (and therefore the higher the adsorption
capacity according to the obtained isotherm) the higher
the surface diusion coecient. This behavior is due to
the higher Phe concentration in the adsorbed phase than
that found in the gas phase [18,29] that improves and
increases the surface ux. Although there are some differences in the surface diusion coecient depending on
the adsorption capacity, all the numerical values are
close enough for calculating an average surface diusion
value for every carbon material, which are shown in
Table 5. From these average values, and by using Eqs.
(7) and (4), it is possible to calculate the corresponding eective diusion and global mass transfer (kp )

2016

R. Murillo et al. / Carbon 42 (2004) 20092017

Table 5
Dierent transport tted and calculated parameters in terms of carbon material
Adsorbent
ICB5
ICB8
ICB12

Fitted kp (s1 )
0.002
0.001
0.001

Average Ds (m2 /s)

12

6.77 10
7.35 1012
1.09 1011

coecients obtaining similar values to the previously

deduced in the experimental data tting. In fact, the
ttings shown in Fig. 3 were obtained by using these
calculated values observing a good concordance with the
experimental results. Finally, with respect to the relationship between surface diusion and carbon material
morphological properties, it is observed that a higher
value for this parameter has been found for ICB8 and
ICB12 that, at the same time, are the materials with a
lower VCO2 =VN2 ratio (0.85 and 0.86, respectively).
However, in the carbonaceous material with a narrower
microporosity, ICB5, which shows a higher VCO2 =VN2
ratio (0.92), the average surface diusion coecient is
lower. As it was previously commented, these results are
consistent with the trend shown by the height of the
mass-transfer zone that increases when the VCO2 =VN2
ratio closes to unity. Therefore, there is a clear relationship between the carbon material structure and their
kinetic behavior in adsorption processes, in such a way
that the narrower the microporosity, the slower the
adsorption kinetics and, therefore, the higher the height
of the mass-transfer zone.
In order to improve the abatement of Phe, and PAH
in general, in energy generation processes, activated
carbons with a developed microporosity and wide
micropore size distributions should be designed. Working with these materials, high adsorption capacities and
fast kinetics will be achieved. Therefore, longer operation times until activated carbon saturation will be
possible, not only because of the higher adsorption
capacity but also because the decreasing of the height of
the mass-transfer zone. In addition, fast kinetics imply
that high space velocities can be used reducing the size
of the equipments and therefore lower investment costs
are needed to achieve the hot gas cleaning.

Acknowledgements
Authors would like to thank to the General Council
of Arag

on (D.G.A., Spain) through the T. Garc
a and
J.M. L

opez Pre-Doc Grants and to the Spanish Ministry
of Environment (Project reference AMB2000-168) for
their partial nancial support. R. Murillo and M.S.
Call
en would like to thank the Spanish Science and
Technology Ministry for the Ram

on y Cajal Program
contract.

Calculated De (m2 /s)

6

1.98 10
6.49 106
3.96 106

Calculated kp (s1 )
0.0019
0.0012
0.0014

References
[1] Mastral AM, Call
en MS, Murillo R, Garc
a T. Polycyclic
aromatic hydrocarbons and organic matter associated to particulate matter emitted from atmospheric uidized bed coal combustion. Environ Sci Technol 1999;33:317784.
[2] Bayram, A. Generation of emission factors for PAH due to
Turkish coals burned in dierent types of combustion units. PhD,
Izmir, March 1995.
[3] Liu K, Han W, Pan WP, Riley JT. Polycyclic aromatic hydrocarbon (PAH) emissions from a coal-red pilot FBC system.
J Hazard Mater 2001;84(23):17588.
[4] Mastral AM, Call
en MS. A review on polycyclic aromatic
hydrocarbon (PAH) emissions from energy generation. Environ
Sci Technol 2000;34(15):30517.
[5] Alvim Ferraz MCM, M
oser S, Tonhaauser M. Control of
atmospheric emissions of volatile organic compounds using
impregnated active carbons. Fuel 1999;78:156773.
[6] Alonso M, Lorences MJ, Pina MP, Patience GS. Butane partial
oxidation in an externally uidized bed-membrane reactor. Catal
Today 2001;67:1517.
[7] Weber R, Sakurai T, Hagenmaier H. Low temperature decomposition of PCDD/PCDF, chlorobenzenes, PAHs by TiO2 -based
V2 O5 WO3 catalysts. Appl Catal BEnviron 1999;20:24956.
[8] Mastral AM, Garcia T, Murillo R, Call
en MS, Lopez JM,
Navarro MV. Sorbent characteristics inuence on the adsorption
of PAC. PAH adsorption with the same number of rings. Fuel
Proc Technol 2002;7778:3739.
[9] Cousins IT, Beck AJ, Jones KC. Review of the processes involved
in the exchange of semi-volatile organic compounds (SVOC)
across the airsoil interface. Sci Total Environ 1999;228:524.
[10] Rubio B, Izquierdo MT, Mastral AM. Inuence of low rank coal
char properties on their SO2 removal capacity from ue gases. 2.
Activate chars. Carbon 1998;36:2638.
[11] Adib F, Bagreev A, Bandosz TJ. Analysis of the relationship
between H2 S removal capacity and surface properties of
unimpregnated activated carbons. Environ Sci Technol
2000;34:68692.
[12] Mastral AM, Garc
a T, Call
en MS, Navarro MV, Galb
an J.
Assessment of phenanthrene removal from hot gas by porous
carbons. Energy Fuels 2001;15(1):17.
[13] Mastral AM, Garc
a T, Call
en MS, Navarro MV, Galb
an J.
Removal of naphthalene, phenanthrene, and pyrene by sorbents
from hot gas. Environ Sci Technol 2001;35:2395400.
[14] Mastral AM, Garc
a T, Murillo R, Call
en MS, L

opez JM,
Navarro MV. Moisture eects on the phenanthrene adsorption
capacity by carbonaceous materials. Energy Fuels 2002;16(1):205
10.
[15] Mastral AM, Garc
a T, Murillo R, Call
en MS, L

opez JM,
Navarro MV. Eects of CO2 on the phenanthrene adsorption
capacity of carbonaceous materials. Energy Fuels 2002;16(2):
5106.
[16] Mastral AM, Garc
a T, Call
en MS, L

opez JM, Navarro MV,
Murillo R, et al. Three-ring PAH removal from waste hot gas by
sorbents: inuence of the sorbent characteristics. Environ Sci
Technol 2002;36(8):18216.

R. Murillo et al. / Carbon 42 (2004) 20092017

[17] Mastral AM, Garc
a T, Murillo R, Call
en MS, L

opez JM,
Navarro MV. Measurements of polycyclic aromatic hydrocarbon
adsorption on activated carbons at very low concentrations. Ind
Eng Chem Res 2003;42:15561.
[18] Ruthven D. Principles of adsorption and adsorption processes.
New York: John Wiley and Sons; 1984. p. 20673.
[19] Suzuki M. Adsorption engineering. Amsterdam: Kodansha/Elsevier; 1990. p. 15186.
[20] Tien C. Adsorption calculations and modelling. Washington:
Butterworth-Heinemann; 1994. p. 12166.
[21] Brosillon S, Manero MH, Foussard JN. Mass transfer in VOC
adsorption on zeolite: experimental and theoretical breakthrough
curves. Environ Sci Technol 2001;35:35715.
[22] Delage F, Pr
e P, Le Cloirec P. Mass transfer and warming during
adsorption of high concentrations of VOCs on an activated
carbon bed: experimental and theoretical analysis. Environ Sci
Technol 2000;34:481621.

2017

[23] Pr
e P, Delage F, Le Cloirec P. A model to predict the adsorber

thermal behavior during treatment of VOC onto wet activated
carbon. Environ Sci Technol 2002;36:46818.
[24] Malek A, Farooq S. Kinetics of hydrocarbon adsorption on
activated carbon and silica gel. AIChE J 1997;43(3):76176.
[25] Sircar S, Hufton JR. Why does the linear driving force model for
adsorption kinetics work? Adsorption 2000;6:13747.
[26] Wakao N, Funazkri T. Eect of uid dispersion coecients on
particle-to-uid mass transfer coecients in packed beds: correlation of Sherwood numbers. Chem Eng Sci 1978;33(10):1375
84.
[27] Huang CC, Fair JR. Study of the adsorption and desorption of
multiple adsorbates in a xed bed. AIChE J 1988;34(11):186177.
[28] Patrick JW, editor. Porosity in carbons. UK: Edward Arnold;
1995. p. 552.
[29] Yang RT. Gas separation by adsorption processes. Stoneham:
Butterworths Publishers; 1986. p. 14193.