You are on page 1of 9

Drying of Supported Catalysts: A Comparison of Model Predictions and

Experimental Measurements of Metal Proles


Xue Liu,

Johannes G. Khinast,

and Benjamin J. Glasser*


,
Department of Chemical and Biochemical Engineering, Rutgers UniVersity, 98 Brett Road, Piscataway,
New Jersey 08854, and Institute for Process Engineering, Graz UniVersity of Technology, Inffeldg. 21A,
A-8010 Graz, Austria
Supported metal catalysts are used in many industrial applications. Experiments have shown that drying may
signicantly impact the metal distribution within the support. Therefore we need to have a fundamental
understanding of drying. In this work, a theoretical model is established to predict the drying process, and the
model predictions are compared with experimental measurements of a nickel/alumina system. It is found that
egg-shell proles can be enhanced by increasing the drying temperature or the initial metal concentration if
the metal loading is low. For high metal loadings, nearly uniform proles are observed after drying. We have
also investigated how breakage of the liquid lm inside the pores of the support can affect the metal distribution
during drying. It was found that lm-breakage has a signicant impact on the metal distribution, and it is
important to correctly capture lm-breakage in the model in order to get good experimental agreement.
1. Introduction
Supported catalysts are used in a variety of industrial
processes, ranging from catalytic converters and the production
of petroleum to the production of new drugs. These catalysts
consist of a porous support, one or more active catalytic
materials deposited on the support, and in some cases a
modier.
1
With respect to the distribution of the active
component in the support, four main categories of metal proles
can be distinguished, that is, uniform, egg-yolk, egg-shell, and
egg-white proles.
2,3
The choice of the desired metal prole is
determined by the required activity and selectivity, and can be
tailored for specic reactions and/or processes. Although the
development and preparation of supported catalysts have been
investigated for many years, many aspects of the various catalyst
manufacturing steps are still not fully understood, and in industry
the design of catalysts is predominated by trial and error
experiments, which are expensive and time-consuming, and do
not always offer assurances on the nal manufacturing results.
The preparation of supported catalysts usually involves three
steps: impregnation, drying, and reduction and calcination.
Experimental work has shown that the metal distribution within
the support is mainly determined by the impregnation and drying
steps.
4-9
Therefore, to achieve an optimum metal prole a
fundamental understanding of both impregnation and drying is
crucial. However, most studies on controlling the metal proles
in catalysts have focused on the impact of the impregnation
step. There are still many questions regarding the impact of
drying that remain unanswered.
The effect of drying on the metal distribution and catalyst
properties has been studied experimentally by Wu et al.
10,11
who
investigated the impact of various preparation procedures on
the mechanical strength of solid catalysts and showed that drying
has a signicant effect on the catalysts mechanical properties.
Santhanam et al.
12
examined the nature of the Pd precursors
and the adsorption of Pd complexes during and after drying
with different adsorption strengths. They showed that for strong
adsorption there is no migration of the metal through the pellets
during drying, while for weak adsorption migration does occur,
leading to a modied nal prole. Li et al.
9
studied the Ni
distribution during the preparation of Ni/alumina catalyst pellets
and compared the experiments with simulations. They showed
that the simulation tted the experimental data well, if the metal
redistribution during drying was considered. Other work has
focused on the characterizations of the physicochemical pro-
cesses that occur during the preparation of supported catalysts
using nuclear magnetic resonance,
13-17
and spatially resolved
Raman and UV-visible-NIR spectroscopy.
18-20
Computer simulations have also been used to predict the
impregnation and drying of supported catalysts. Theoretical
models describing the impregnation step have been reported in
a number of papers.
6,7,21-23
Because of the complexity of the
drying step, only a few theoretical models have been reported
for drying. However, experiments and simulations have shown
that the drying procedure can signicantly affect the metal
prole established during impregnation, if adsorption of the
metal component on the support surface is weak or mode-
rate.
12,24-27
Neimark et al.
5,28
were among the rst to theoreti-
cally study the metal redistribution during drying. They used a
dimensionless number to characterize slow drying and fast
drying regimes, and their theory is in agreement with the
experiments of Komiyama et al.,
8
who showed that a very high
drying rate can result in a uniform prole, while a relatively
low drying rate favors an egg-shell prole. More detailed drying
models were formulated by Uemura et al.,
29
Lee and Aris,
6
and
Lekhal et al.
24-26
They considered the effects of the capillary
ow and metal diffusion and simulated the metal migration
during drying. Recently, the sensitivity of the metal distribution
during impregnation and drying with respect to the physical
and processing parameters was examined by Liu et al.
27
They
also considered the effect of crystallization in their model, and
showed that metal crystallization has a signicant effect on the
generation of egg-shell proles for relatively high metal
concentrations.
It is of particular interest to compare simulation results and
experimental measurements to validate the theory and determine
the key parameters used to predict the metal distribution during
the preparation of supported catalysts. Most previous studies
* To whom correspondence should be addressed. Tel.: 732- 445-
4243. Fax: 732-445-2581. E-mail: bglasser@ rutgers.edu.

Rutgers University.

Graz University of Technology.


Ind. Eng. Chem. Res. 2010, 49, 26492657 2649
10.1021/ie9014606 2010 American Chemical Society
Published on Web 02/15/2010
only focused on a comparison of theory and experiments for
the impregnation step. A systematic comparison for the drying
step has not been reported as of yet. Thus, the objective of this
paper is to predict the metal distribution during drying and to
compare the simulation results with experimental measurements.
Furthermore, it is of interest to investigate the fundamental
mechanisms occurring during drying, and to study the impact
of the processing parameters and material properties on the nal
metal distribution.
2. Model and Experiment Setup
2.1. Model Equations. In the present work, we studied the
drying of a Ni/alumina system, which is widely used in
processes like hydrogenation, hydrodesulfurization, and steam
reforming of hydrocarbons.
30,31
During the drying process,
several phenomena are taking place simultaneously: heat transfer
from the hot gas to the wet support, solvent evaporation near
the external surface, solvent convective ow toward the external
surface, and metal diffusion and adsorption inside the support.
An accurate drying model must include all these phenomena.
In this work, we extend the model proposed by Lekhal et
al.
24
by considering a cylindrical geometry and the impact of
breakage of the liquid lm inside the pores of the support (lm-
breakage). In our model the following parameters have an
impact on the drying process: the metal diffusion coefcient,
the equilibrium constant of adsorption and desorption, the
intrinsic permeability, the initial metal concentration in the
solvent, the drying temperature, the humidity of the drying air,
and the lm-breakage parameters. It is important to note that
although the results presented are chosen for a Ni/alumina
system, the methodology is entirely general. It is not limited to
specic active components and supports.
There are two main assumptions in our model: (1) During
drying the metal concentration in the solution is below its
solubility; therefore, crystallization is not considered. (2) The
equilibrium adsorption constant can be assumed to be a constant
during the drying process. These assumptions have been made
in order to arrive at a model that can simply yet accurately
describe the important physical processes taking place during
drying. Our model can capture convective ow in the gas and
liquid phases, metal convection, diffusion, and adsorption on
the porous support as well as heat transport.
The following equations describe the drying process:
Equations 1 and 2 represent the mass balances of the drying
medium (air) and the solvent (water). Equations 3 and 4
represent the mass balances of the metal dissolved in the liquid
and deposited on the support. Equation 5 is the energy balance.

g
and
l
are the volume fractions of the gas and liquid phases,
C
g,a
(mol/m
3
) and C
l,s
(mol/m
3
) are the concentrations of the
drying air and the liquid solvent, respectively. C
l,i
(mol/m
3
) and
C
s,i
(mol/kg) are the concentrations of the metal dissolved in
the solvent and adsorbed on the support. N
g,a
(mol/(m
2
s)) and
N
g,v
(mol/(m
2
s)) are the uxes of the air and solvent vapor.
N
l,s
(mol/(m
2
s)) and N
l,i
(mol/(m
2
s)) are the uxes of the
solvent and the dissolved metal. F
s
(kg/m
3
) is the apparent
density of the porous support. R
i
(mol/kg/s) represents the rate
of metal adsorption, which is described using a Langmuir
model.
32-34
where k
ads
(m
3
/mol/s) and k
des
(s
-1
) are the adsorption and
desorption constants of the metal component. C
sat
(mol/kg)
denotes the metal saturation concentration. In this model we
assume adsorption is not the limiting step. Consequently, the
adsorbed metal is in equilibrium with its dissolved precursor,
and the equilibrium adsorption constant can be calculated as
K
eq
) k
ads
/k
des
. Although the value of the equilibrium adsorption
constant may change during drying,
26
in this work we assume
it to be a constant. h
g,i
(J/mol) represents the enthalpy of the air
or solvent vapor. h
l
(J/mol) and h
s
(J/kg) denote the enthalpy of
the liquid and solid. (J/(m s K)) is the effective thermal
conductivity.
Film-breakage is an important phenomenon during drying.
At the beginning of drying, the water phase is continuously
distributed in the support. As evaporation proceeds, isolated
domains are gradually formed in the liquid phase. Finally, the
liquid is only found in the isolated domains. To consider
the effect of lm-breakage in our model, a factor R is added to
the ux terms in eqs 2, 3, and 5. Neimark et al.
5
showed that
lm-breakage is related to the support pore structure and the
water content in the support. Therefore, for a given solid carrier,
we assume that the lm-breakage factor, R, is a function of the
water volume fraction.
where R
1
represents the water volume fraction when lm-
breakage starts, and R
2
represents the water volume fraction
when the solvent only exists in the isolated domains and the
water ux completely stops. We assume that when the value of
the water volume fraction is between R
1
and R
2
, R linearly
decreases with a decrease in the water volume fraction indicating
that the ux in the liquid phase linearly decreases with a decrease
in the water content.
The gas-phase uxes N
g,a
and N
g,v
are assumed to follow the
dusty gas model (DGM),
35
which considers the effect of
molecular diffusion, Knudsen diffusion, and viscous ow.
In eq 8 P
g
(Pa) is the total gas pressure, R (8.314 J/(mol K))
is the gas constant, T (K) is the temperature, x
g,i
represents the

t
(
g
C
g,a
) ) -
1
r

r
(rN
g,a
) (1)

t
(
l
C
l,s
) ) -
1
r

r
(RrN
l,s
+ rN
g,v
) (2)

t
(
l
C
l,i
) ) -
1
r

r
(RrN
l,i
) - F
s
R
i
(3)

t
(C
s,i
) ) R
i
, i ) metal component (4)

t
(

i)1
2

g
C
g,i
h
g,i
+
l
C
l,s
h
l
+ F
s
h
s
) )
-
1
r

r
(

i)1
2
rN
g,i
h
g,i
+ RrN
l,s
h
l
- r
T
r
)
(5)
R
i
) k
ads
C
l,i
(C
sat
- C
s,i
) - k
des
C
s,i
, i ) metal component
(6)
R ) 1 when
f
g R
1
R )
(R
1
-
f
)
(R
1
- R
2
)
when R
1
>
f
> R
2
R ) 0 when
f
e R
2
(7)
-
P
g
RT
x
g,i
-
x
g,i
RT
(
1 -
KK
g,eff
P
g

g
D
Knud
)
P
g
)

j)1
j*i
2
x
g,j
N
g,i
- x
g,i
N
g,j
D
g,ij
+
N
g,i
D
Knud
(8)
2650 Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010
vapor or air mole fraction in the gas phase, D
g,ij
(m
2
/s) and D
Knud
(m
2
/s) are the effective binary and Knudsen diffusion coefcients
estimated from the kinetic gas theory,
36
and K
g,eff
is the intrinsic
permeability of the gas phase, which has the form
Jones
37
has shown that eq 9 can describe experimental data
very well. In this work, the water vapor pressure is calculated
using the Antoine equation
38
and the Hailwood-Horrobin
equation
39
with parameters tted by Simpson.
40
We assume that
the convective ow in the liquid phase follows Darcys law,
41
where K
l,eff
is the relative permeability of the liquid phase,
l
(Pa s) is the viscosity of the liquid phase, K (m
2
) represents
the intrinsic permeability, and P
l
(Pa) is the liquid phase
pressure, which is equal to the local gas pressure less the
capillary pressure P
c
(Pa).
41
In the present work P
c
is described
using the form proposed by Perre et al.,
42
where (N/m) represents the surface tension, and M
l,s
is the
molecular weight of the liquid solvent. The ux of the dissolved
metal is described by the Nernst-Planck equation,
43
which takes
into account the effect of convective ow of the solvent
(capillary ow), diffusion due to the metal concentration
gradient, and migration caused by electrical charges. It takes
the form
where D
l,i
(m
2
/s) is the effective diffusion coefcient of the
dissolved metal, Z
i
is the charge of the metal component, F
(96500 C/mol) is the Faraday constant, and (V) represents
the electrostatic potential. In this work we assume there is no
external current and the electroneutrality condition is satised
in the support. The gradient of the electrostatic potential, which
is a function of the number of charges and the concentration
gradient of the charged components, is determined by the no-
current equation:
44
where n is the total number of ionic species in the liquid phase.
The constitutive relations proposed by Jones
37
are adopted for
the relative permeability K
l,eff
.
The boundary conditions are the zero-ux conditions at the
support center and the Neumann conditions at the support
surface.
24,25
The resulting system of nonlinear partial differential
equations is spatially discretized by a nite volume method.
45
Then the resulting set of ordinary differential equations is solved
by LIMEX, which is efcient for solving highly stiff differential-
algebraic equations.
46
2.2. Experiment Setup and Parameter Measurement. The
system studied in this work is a nickel/alumina system. Nickel
nitrate powders (Sigma-Aldrich) were used as metal precursors
and cylindrical -alumina pellets provided by Saint-Gobain were
used as solid carriers. The pellets are 3 mm in diameter and
around 10 mm in length with a void volume fraction of 0.3
cm
3
/g and a surface area of 200.7 m
2
/g. The basic experimental
protocol includes the following steps: (1) The solid support is
preheated in an oven at 120 C for 12 h. (2) The dry alumina
supports are immersed in nickel nitrate solutions for impregna-
tion. The pH of the solutions is adjusted by adding nitric acid
or NH
4
OH. To investigate the effect of the metal concentration,
we changed the concentration of the nickel nitrate solutions from
0.01 to 4 M. Usually we hold the impregnation time sufciently
long such that a uniform prole can be obtained after impregna-
tion, representing an equilibrium state. (3) The catalyst samples
are dried in an oven at a constant temperature. The drying
temperature is varied between 22 and 180 C. (4) Calcination
is carried out at 500 C for 2 h. During impregnation and drying,
nickel nitrate gives the catalyst a green color. During calcination,
nickel nitrate becomes nickel oxide, and thus, the catalyst color
changes. The gray or black color of the samples after calcination
is most likely due to some deviation from ideal 1:1 stoichiometry
of the NiO.
47
The nickel concentration in the solution is measured using a
UV-visible spectrophotometer at a wavelength of 190 nm. To
obtain the standard curve, seven samples were prepared with
the Ni(NO
3
)
2
concentration equaling 0.001, 0.005, 0.01, 0.025,
0.05, 0.075, and 0.1 M. Then the absorbance value of each
sample was measured by the UV-visible apparatus. From
experiments we found that there is a linear relation between
the nickel concentration, C
Ni
and the absorbance value, A
uv
.
Using a linear regression, we can obtain the equation:
This equation can be used to calculate the nickel concentration
in the solution during impregnation. To investigate the metal
proles after drying or calcination, we cut the catalyst samples
in half in the radial direction and measured the radial nickel
prole using micro-X-ray uorescence spectroscopy (micro-
XRF).
To solve our drying model, we need to measure several
parameters.
9,32
In our work, a Langmuir equation is used to
describe the adsorption and desorption processes (see eq 6).
Under equilibrium conditions, the rate of the metal adsorption
is equal to the rate of the metal desorption (R
M
) 0). Therefore,
eq 6 can be rewritten as
From eq 16 it can be seen that a straight line is obtained
when plotting 1/C
eq
versus 1/C
s
. Then the values of C
sat
and
K
eq
can be calculated from the line intercept and the slope. Six
samples with Ni(NO
3
)
2
concentration equal to 0.01 M, 0.02 M,
0.04 M, 0.06 M, 0.08 M, and 0.1 M were prepared. Each sample
contained 100 mL of Ni(NO
3
)
2
solution and 1 g of alumina
support with the pH equal to 6.5. The value of C
s
can be
calculated based on the Ni mass balance in the system, since
the amount of the metal in the solution before impregnation
minus the amount of the metal in the solution after impregnation
equals the amount of the metal adsorbed on the support. From
Figure 1, it can be seen that the amount of metal deposited on
the supports increases rapidly at the beginning of impregnation.
K
g,eff
) 1 - 1.11
(

)
(9)
N
l,s
) -C
l,s
KK
l,eff

l
P
l
(10)
P
c
) 1.364 10
5

l
c
l,s
M
l,s
F
s
)
-0.63
(11)
N
l,i
) -C
l,i
KK
l,eff

l
P -
l
D
l,i
c
l,i
-
l
C
l,i
Z
i
D
l,i
F
RT

l
,
i ) metal component (12)

i)1
n
z
i
N
l,i
) 0 (13)
K
l,eff
)
(

)
3
(14)
C
Ni
) 0.0878A
uv
(15)
1
C
eq
)
1
C
sat
K
eq

1
C
s
+
1
C
sat
(16)
Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010 2651
Then, the adsorption rate decreases due to the increase in the
surface coverage of the active sites. After approximately 3 days
a plateau is reached indicating an equilibrium state, from which
we can obtain the equilibrium metal concentration in the solution
C
eq
and the corresponding metal load on the support C
s
. Using
eq 16, we obtained C
sat
) 0.3 mol/kg and K
eq
) 0.2 m
3
/mol
when plotting 1/C
eq
versus 1/C
s
.
At the beginning of impregnation, the effect of desorption
can be neglected. Thus, the decrease in the metal concentration
in the solution is mainly due to the accumulation of the metal
adsorbed on the support. Therefore,
where C
0
represents the initial metal concentration in the
solution.
9,32
When plotting dC
0
/dt versus C
0
, a straight line can
be obtained and the value of the kinetic adsorption constant
k
ads
can be calculated from the slope. To obtain k
ads
, we prepared
ve samples with Ni(NO
3
)
2
concentrations equal to 0.01, 0.02,
0.03, 0.04 and 0.05 M, at a pH equal to 6.5. To reduce diffusion
effects during impregnation we ground the pellet supports into
powders. The particle size was between 150 and 250 m. We
used a sieve to remove large particles, and then used water to
wash out ne particles. The powder supports were dried in the
oven at 120 C for 12 h before being used. For each sample,
the value of the Ni concentration was measured at 10 min
intervals after impregnation started. Then the value of dC
0
/dt
was calculated. By plotting dC
0
/dt versus C
0
, we obtained k
ads
) 6.5 10
-5
m
3
/(mol/s).
The diffusion coefcient of nickel nitrate in water was taken
from the work of Takahashi et al.
48
as D ) 6 10
-10
m
2
/s.
The permeability is based on the support pore size distribution.
The pore size distribution was measured by Saint-Gobain using
a mercury volume test. The porosity of the support is around
0.67 with 80% small pores (average 7 nm) and 20% large pores
(average 500 nm). Then the permeability of the support can be
calculated using the modied Ergun equation.
49
Using eq 18, we obtained a permeability of K ) 5 10
-16
m
2
.
In general, the base case conditions used in our simulations
are pH ) 6.5, C
sat
) 0.3 mol/kg, K
eq
) 0.2 m
3
/mol, k
ads
) 6.5
10
-5
m
3
/(mol/s), D ) 6 10
-10
m
2
/s, K ) 5 10
-16
m
2
,
and 30% relative humidity. The initial metal concentration in
the solution C
0
was varied from 0.04 to 4 M, and the drying
temperature T
bulk
was varied from 22 to 180 C. A uniform initial
metal distribution was utilized indicating that impregnation
reached an equilibrium state.
3. Results and Discussion
3.1. Experimental Results. Typical experimental drying
results are shown in Figure 2 for T
bulk
) 60 C. In Figure 2a
the water volume fraction can be seen to decrease with time
until a plateau is reached. In Figure 2b the drying rate is equal
to the weight of the water evaporated from the support per
kilogram dry support per minute. At the beginning of the process
the drying rate is constant. After about 40 min the drying rate
decreases and nally the water content in the support is reduced
to 1% after 75 min (see Figure 2a) indicating the end point of
drying. Similar results have been reported in previous studies.
25
During drying the metal concentration in the liquid phase
increases due to evaporation of water. This may greatly affect
the solution properties, the drying rate, and the metal distribu-
tion. Figure 3 shows experimental measurements of the evolu-
tion of the drying rate for different initial metal concentrations
at a drying temperature of 60 C. The lines here are included
as a guide for the eye. By comparing the curves for C
0
) 0 M
(water only) and C
0
) 0.1 M, we nd that for a low initial
Figure 1. The variation of the concentration of the metal deposited on the
support with the impregnation time.
dC
0
dt
) -k
ads
FC
sat
C
0
(17)
K )

i
3
i
d
ci
2
200
(18)
Figure 2. Variation of (a) the water volume fraction and (b) drying rate
with the drying time at T
bulk
) 60 C and C
0
) 100 mol/m
3
.
Figure 3. Effect of the initial metal concentration on the drying rate at
T
bulk
) 60 C for (a) catalyst samples and (b) solution samples. The lines
here are included as a guide for the eye.
2652 Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010
metal concentration (C
0
< 0.1 M), the effect of the metal
concentration on the drying rate is not signicant (see Figure
3a). For C
0
> 2 M, however, the drying rate is signicantly
reduced. In Figure 3a, the drying time for C
0
) 0.1 M is around
50 min. In contrast, the drying time required for C
0
) 4 M is
more than 75 min. We believe that this is due to the decrease
in the vapor pressure and the increase in the solvent viscosity
with an increase in the metal concentration.
50
Therefore, drying
is much slower for high metal concentrations, and the drying
time required for high metal concentrations is much longer than
for low metal concentrations. To eliminate the effect of the
support pore size distribution and pore network on the drying
rate and only focus on the contribution of the initial metal
concentration, 1 mL of solution (no support) with a certain
amount of Ni(NO
3
)
2
was dried in the oven at 60 C. In Figure
3b, we show results for ve samples with Ni(NO
3
)
2
concentra-
tion equal to 0 M (only water), 0.1, 0.5, 2, and 4 M. It is clear
that for a low initial metal concentration (C
0
< 0.1 M), after an
initial increase the drying rate reaches a plateau and then reduces
rapidly at the end of drying. For a high initial metal concentra-
tion (C
0
> 2 M), however, the drying rate is much lower and
the drying rate evolution becomes quite different. The plateau
region observed in the low initial metal concentration conditions
disappears and the drying rate gradually reduces with time. This
is because for high metal concentration conditions, the amount
of the metal precursor is comparable to the amount of water so
the increase in the molar ratio of the metal precursor in the
liquid phase during drying becomes signicant leading to a
gradual decrease in the water vapor pressure.
50
In contrast, for
low metal concentration conditions the amount of water is much
higher than the amount of the metal precursor. Therefore,
although the molar ratio of the metal precursor keeps increasing
during drying its effect on the change of the water vapor pressure
is negligible. If we compare the drying rate evolution curves
shown in Figure 3 panels a and b, we nd that the curve shapes
and the extent of the decrease in the drying rate with the initial
metal concentration look quite similar for the two cases. This
indicates that the effect of the initial metal concentration on
the drying procedure during preparation of supported catalysts
is important. For moderate or high metal loading, an accurate
drying model must be capable of capturing the change of the
solvent properties due to the increase in the metal precursor
concentration during drying.
After impregnation, the metal inside the support has two
forms: metal dissolved in the solvent or metal adsorbed on the
support. From past studies we know that drying can change the
distribution of the metal dissolved in the solvent, while its effect
on the metal already adsorbed on the support is much smaller.
12
After impregnation, the ratio of the amount of the metal
dissolved in the solvent to that adsorbed on the support is
determined by the adsorption strength and the initial metal
concentration in the solvent. The effect of adsorption strength
on the metal proles during drying has been reported in previous
work.
25,27
It was found that drying can modify the metal proles
only for weak adsorption, while its effect is not signicant for
strong adsorption.
The impact of the initial metal concentration on the nal metal
distribution after drying is shown in Figure 4a-c. Clearly, the
total metal left in the support after drying increases with an
increase in the initial metal concentration. From Figure 4a,b,
we can see that for a uniform initial condition, an egg-shell
prole is obtained if the metal load in the system is low or
moderate. This is due to the effect of convection which drives
the metal to move toward the support surface. If the initial metal
concentration is sufciently high (C
0
> 3 M), nearly uniform
proles can be observed after drying (see Figure 4c). This may
be related to three mechanisms. (1) For C
0
> 3 M, the drying
rate is greatly reduced (see Figure 3a), which favors a nal
uniform distribution. (2) If the metal concentration is sufciently
high, during drying the support pores can be blocked by the
accumulation of metal crystals due to adsorption and crystal-
lization. This pore-blockage mechanism can greatly reduce the
water transport and the metal redistribution during drying.
Similar results have been observed in previous work. Sietsma
et al.
47
investigated the preparation of Ni/SiO
2
catalysts via the
impregnation and drying method. They found that with 4.2 M
initial metal concentration, the average crystal size after drying
was 9 nm which was around the same size as the mesopore
Figure 4. Effect of the metal concentration on the metal proles after drying at T
bulk
) 60 C: (a) low metal concentrations; (b) moderate metal concentrations;
(c) high metal concentrations. Effect of the metal concentration on the metal proles after calcination: (d) low metal concentrations; (e) moderate metal
concentrations; (f) high metal concentrations.
Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010 2653
diameter of the SBA-15 support they used. (3) Since the melting
point of Ni(NO
3
)
2
is 56 C, part of the nickel nitrate could be
melted when the samples were dried at 60 C. For high metal
loading conditions (C
0
> 3 M), the liquid Ni(NO
3
)
2
may lead to
the liquid phase remaining continuous during drying, and thus
lm-breakage would not occur. This will favor a nal uniform
distribution. The effect of lm-breakage on the metal distribution
during drying will be further discussed in the following section.
For practical use of the catalyst it is of interest to study the
effect of calcination on the distribution of the metal in the
support. Figure 4 panels d-f show the metal distribution after
calcination with the variation of the metal concentration from
0.05 to 4 M. It is clear that for all cases studied in this work
the metal distribution after drying and after calcination is similar,
indicating that the effect of calcination on the metal redistribu-
tion is not signicant. Therefore, it is reasonable to assume that
for our specic systems the metal prole obtained after drying
can be used to predict the nal metal distribution of the catalysts.
3.2. Comparison of Experiments and Simulations. In this
section we focus on low metal loads where the initial metal
concentration is less than or equal to 0.1 M. For these cases,
the effect of the metal ions on the solvent properties during
drying is small, and pore-blockage and crystallization are
negligible. Therefore, the nal metal distribution is determined
by the initial metal concentration (C
0
), adsorption strength (K
eq
,
k
ads
, C
sat
), drying conditions (T
bulk
), transport properties (D
l,i
, K)
and lm-breakage conditions (R
1
, R
2
). Given a specic
metal-support system, the parameters for adsorption, transport,
and lm-breakage are xed and cannot be adjusted in a
straightforward manner. Thus, the nal metal prole can be
controlled mainly by changing the initial metal concentration
and the drying temperature.
The variation of the water volume fraction in the support
during drying for different drying temperatures is shown in
Figure 5, where the symbols represent the experimental data
and the lines represent the simulation results. To investigate
the effect of lm-breakage, two sets of simulation results are
presentedsone including and one excluding the effects of lm
breakage. In the simulations with lm-breakage, we assume R
1
) 0.53 representing the situation where lm-breakage starts as
the water evaporation transits from the large pores to the small
pores (R
1
)voidage volume fraction x percentage of small pores
in the void ) 0.67 0.8), and R
2
) 0.013 below which the
liquid phase is completely discontinuous (solvent ux ) 0). The
value of R
2
was chosen on the basis of the regression of
experimental data for C
0
) 0.04 M, and thereafter we held this
value a constant for other cases. R
2
) 0.013 corresponds to the
mass ratio of water in the support equal to 2%. In general the
value of R
2
is related to the hydrophilic or hydrophobic
properties of the solvent on the support, the size of the small
pores, and the pore network in the support.
51
The structure of
the pore network has a signicant effect on the transport of the
solvent during drying. Neimark et al. proposed that the point
where the liquid phase becomes completely discontinuous (i.e.,
R
2
) can be calculated on the basis of a coordination number for
the support if the porous space can be represented as a system
of intersecting channels and the coordination number is the
average number of channels meeting at a lattice site.
28
In Figure 5, it is clear that drying is much faster at higher
drying temperatures. When drying is carried out at room
temperature (T
bulk
) 22 C), drying is very slow and an
unacceptable amount of water remains in the support at the end.
When the drying temperature is above 60 C, the mass fraction
of the water in the support can be reduced to 1% within a
reasonable amount of time. From Figure 5 it can be seen that
for low to moderate drying rates (T
bulk
< 60 C), the effect of
lm-breakage on the change in the water content inside the
support during drying is small. When the drying temperature is
high, the drying rate decreases slightly when lm-breakage is
taken into account. This effect is observed since for fast drying
rates lm-breakage occurs at the support surface fairly early in
the drying process. For all three cases shown in Figure 5, the
comparison between the experiments and simulations is gener-
ally good, and the effect of lm-breakage is rather small.
Corresponding to the data in Figure 5, Figure 6 shows
experimental data and simulation results for the drying rate. The
simulation results in the gure include the effect of lm-
breakage. Simulations without lm-breakage have also been
carried out (not shown) and they are very similar to the
simulation results in Figure 6. As before, the drying temperature
is 22, 60, and 115 C, respectively. The variation of the
Figure 5. Effect of the drying temperature on the water volume fraction
for C
0
) 100 mol/m
3
, R
1
) 0.53, and R
2
) 0.013.
Figure 6. Effect of the drying temperature on the drying rate for C
0
) 100
mol/m
3
, R
1
) 0.53 and R
2
) 0.013: (a) T
bulk
) 22 C; (b) T
bulk
) 60 C; (c)
T
bulk
) 115 C.
2654 Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010
temperature in the support during drying was also examined in
this work (not shown). For a low drying temperature, initially
the temperature in the particle decreases due to evaporation,
reaches a plateau, and then increases (T
bulk
) 22 C). Corre-
sponding to the change of the particle temperature, the drying
rate shows three stages (see Figure 6a), i.e., a decrease in the
rate, a plateau stage, and a second decrease in the rate. The
sharp drop in the drying rate in the rst stage is due to the sharp
drop in the particle temperature at the beginning of drying. The
third stage occurs when the drying front starts to move from
the support surface to the center due to the loss of water in the
support. For a moderate drying temperature (T
bulk
) 60 C),
the particle temperature only changes slightly before reaching
a plateau. Thus, only two stages can be observed in Figure 6b.
For a high drying temperature (T
bulk
) 115 C), initially the
temperature in the particle increases, reaches a plateau, and then
increases again. Therefore, we observe a preheating period,
followed by a constant-rate period, and a falling-rate period (see
Figure 6c). Similar results have been reported in previous
work.
25
In general, the simulation results match the experimental
measurements fairly well.
The nal metal proles for different drying temperatures are
shown in Figure 7 when the initial metal concentration is 0.04
M. Clearly, more metal is accumulated near the surface with
an increase in the drying temperature. This is due to the
competition between convection and diffusion. For low drying
temperatures, diffusion dominates the drying procedure, which
leads to a uniform proles (see Figure 7a). For relatively high
drying temperatures, convection dominates the drying process
at the early stages of drying, which transports the water and
metal ions toward the external surface, leading to pronounced
egg-shell proles. Although at the late stages of drying, diffusion
may control the drying process, causing the metal to move
toward the support center and the metal distribution to atten,
the nal metal distribution still remains egg-shell (see the
experimental data in Figure 7b,c). One can see from Figure 7
that lm-breakage is very crucial to capture the egg-shell proles
observed in the experiments. If the effect of lm-breakage is
not considered, egg-white proles are predicted for relatively
high drying temperatures (see Figure 7b,c). When the effect of
lm-breakage is taken into account, egg-shell proles can be
obtained for relatively high drying temperatures (see Figure
7b,c). In all the cases studied in Figure 7, the egg-shell proles
are greatly enhanced and the simulations show a good agreement
with the experiments if the effect of lm-breakage is considered.
This is due to the effect of lm-breakage on convection and
diffusion. If lm breakage occurs at the early stage of drying,
it will reduce the liquid ux and thus reduce the effect of
convection. At the late stages of drying, diffusion may dominate
the drying process, and its effect is also reduced by lm-
breakage. Therefore, we believe lm-breakage suppresses the
egg-shell distribution at the early stages of drying and favors
the egg-shell distribution at the late stages of drying. In Figure
7, our simulations show that the egg-shell distribution is
enhanced by lm-breakage, indicating that under the drying
conditions used in Figure 7 lm-breakage does not start at the
very early stage of drying so its main contribution is to reduce
the effect of diffusion. On the other hand, if the drying rate is
sufciently high and lm-breakage occurs at the surface nearly
immediately, the effect of convection should be reduced and
the extent of the egg-shell distribution should be reduced.
To validate our hypothesis, we experimentally measured the
metal proles for different drying rates (see Figure 8). The lines
in Figure 8 are only meant as a guide for the eye. For a low
drying temperature (T
bulk
) 25 C) the effect of convection and
lm-breakage is not signicant so a nearly uniform prole is
observed. For a relatively high drying temperature (T
bulk
) 120
C) convection accumulates the metal near the surface at the
early stage of drying and lm-breakage reduces the effect of
back-diffusion at the late stage of drying so a more pronounced
egg-shell prole can be obtained. If the drying rate is sufciently
high (T
bulk
) 500 C) isolated domains occur at the surface
nearly immediately during drying and the metal has no time to
migrate from the center toward the surface so the egg-shell
prole is less pronounced. Similar results have been observed
in previous work.
8
In Figure 9, we compare the nal metal proles for experi-
ments and simulations for a higher initial metal loading of C
0
) 0.1 M. We found that for all the cases studied, lm-breakage
enhances the egg-shell proles, and simulations show good
agreement with the experiments only if the effect of lm-
breakage is considered. This is in agreement with our previous
results at lower initial metal concentrations (see Figure 7).
From Figure 7 and 9, it can be seen that for the cases we have
examined, lm-breakage must be considered if one is to capture
the metal proles observed in the experiments. Therefore, it is of
Figure 7. Effect of the drying temperature on the nal metal prole for C
0
) 40 mol/m
3
, R
1
) 0.53, and R
2
) 0.013: (a) T
bulk
) 22 C; (b) T
bulk
) 115
C; (c) T
bulk
) 180 C.
Figure 8. Experimental measurement of the nal metal distribution for C
0
) 100 mol/m
3
. The lines here are drawn to guide the eye.
Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010 2655
interest to investigate the sensitivity of the metal distribution after
drying to the lm-breakage parameters, R
1
and R
2
. Generally, the
egg-shell prole is enhanced with an increase in R
1
(not shown).
However, when R
1
is sufciently high (R
1
> 0.53) we nd that the
metal distribution changes only slightly when further increasing
this number. This is because two mechanisms occur with the
variation of R
1
. For a high R
1
value, lm-breakage occurs at the
beginning of drying, which reduces the water ux toward
the surface, and thus suppresses the accumulation of the metal
at the surface. With continued drying, metal starts to move back
to the support center due to the gradient of the metal concentration
in the solvent. Film-breakage can reduce this back diffusion, and
this reduction effect increases with an increase in R
1
. Therefore,
for a high R
1
value lm-breakage suppresses the egg-shell prole
at the earlier stages of drying and favors the egg-shell prole at
the later stages of drying. Consequently, the effect of R
1
on the
nal metal proles is due to the compensation of these two
contributions. To enhance the egg-shell prole, an optimum R
1
is
required. The egg-shell prole can be greatly enhanced with
increasing the value of R
2
(not shown). This is because the variation
of the value of R
2
has only a slight effect on the early stage of
drying, while its effect on the nal stage of drying is signicant.
Therefore, for a high value of R
2
(R
2
)0.13), the pronounced egg-
shell prole formed in the early stage of drying may be still
observed at the end of drying.
The sensitivity analysis was also carried out for other
parameters based on our nickel/alumina system (not shown).
In general, the egg-shell proles can be enhanced by increasing
the permeability and uniform proles can be obtained by
increasing the diffusion coefcient. This is in agreement with
our previous work.
25,27
In our specic case the adsorption
process is much faster than the transport process; we found that
the metal redistribution is not sensitive to the variation of the
kinetic adsorption constant.
4. Conclusions
We established a theoretical model to predict the metal
distribution during drying and compared the simulation results
with experimental measurements for a nickel/alumina system.
The adsorption and transport parameters used in the simulations
are obtained from separate experiments/calculations.
From the experiments, several interesting phenomena were
observed. (1) We found that egg-shell proles can be enhanced
by increasing the drying temperature and the initial metal
concentrations, if the metal load in the system is low or
moderate. For high metal loadings, nearly uniform metal proles
are observed from the experiments. (2) We compared the metal
proles after drying and after calcination and showed that for
our specic situation the effect of calcination on the metal
distribution is not signicant. Thus, the metal proles obtained
after drying can be used to predict the nal metal distribution
of the catalysts. (3) By plotting the variation of the water content
and the drying rate with the drying time for different initial
metal concentrations, we found that if the initial metal concen-
tration is high the solvent properties may change dramatically
during drying because of water evaporation and high metal
concentration in the liquid phase.
We also compared the simulations with experiments to
validate our theory. Since the effect of crystallization and pore-
blockage is not considered in our model, our comparison only
focused on low metal load conditions. To investigate the effect
of lm-breakage on the metal redistribution during drying, we
assume that once lm-breakage occurs the solvent ux linearly
decreases with the decrease in the water volume faction until
the water volume fraction reaches a certain point, at which the
liquid ux completely stops and the metal is enclosed in isolated
liquid domains. We found that lm-breakage is crucial to capture
the metal proles observed in the experiments and the simula-
tions show an excellent agreement with experiments if the effect
of lm-breakage is considered.
In summary, the goal of this study is to better understand
the fundamental mechanisms during drying, and to determine
the key parameters used to generate a desired metal prole, using
theoretical simulations and experiments. We have compared
experiments and simulations for low metal concentration
conditions (C
0
< 0.1 M). For moderate and high metal
concentrations, crystallization may become important and the
change of the solvent properties during drying due to the
increase in the metal concentration in the solvent may greatly
affect the drying process. Pore-blockage may also become
important at high metal concentrations. It remains to be seen
what is the relative importance of these additional phenomena
that occur at moderate and high metal concentrations, and future
work should investigate how these phenomena interact to impact
drying.
Acknowledgment
We wish to acknowledge partial nancial support for this
work from the National Science Foundation and the Rutgers
Catalyst Manufacturing Science and Engineering Consortium.
Literature Cited
(1) Ertl, G.; Knozinger, H.; Weitkamp, J. Preparation of Solid Catalysts;
Wiley-VCH: Weinheim, Germany, 1999.
(2) Shyr, Y. S.; Ernst, W. Preparation of nonuniformly active catalysts.
J. Catal. 1980, 63, 425432.
(3) Gavrilidis, A.; Varma, A.; Morbidelli, M. Optimal distribution of
catalyst in pellets. Catal. ReV.-Sci. Eng. 1993, 35, 399456.
Figure 9. Effect of the drying temperature on the nal metal prole for C
0
) 100 mol/m
3
, R
1
) 0.53, and R
2
) 0.013: (a) T
bulk
) 22 C; (b) T
bulk
)
115 C; (c) T
bulk
) 180 C.
2656 Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010
(4) Maatman, R. W.; Prater, C. D. Adsorption and exclusion in
impregnation of porous catalytic supports. Ind. Eng. Chem. Fundam. 1957,
49, 253257.
(5) Neimark, A. V.; Kheifez, L. I.; Fenelonov, V. B. Theory of
preparation of supported catalysts. Ind. Eng. Chem. Prod. Res. DeV. 1981,
20, 439450.
(6) Lee, S.-Y.; Aris, R. The distribution of active ingredients in supported
catalysts prepared by impregnation. Catal. ReV.-Sci. Eng. 1985, 27, 207
340.
(7) Kotter, M.; Riekert, L. Impregnation type catalysts with nonuniform
distribution of the active component. Part II: Preparation and properties of
catalysts with different distribution of the active component on inert carriers.
Chem. Eng. Fundam. 1983, 2, 3138.
(8) Komiyama, M.; Merrill, R. P.; Harnsberger, H. F. Concentration
proles in impregnation of porous catalysts: nickel of alumina. J. Catal.
1980, 63, 3552.
(9) Li, W. D.; Li, Y. W.; Qin, Z. F.; Chen, S. Y. Theoretical prediction
and experimental validation of the egg-shell distribution of Ni for supported
NidAl
2
O
3
catalysts. Chem. Eng. Sci. 1994, 49, 48894895.
(10) Wu, D. F.; Zhou, J. C.; Li, Y. D. Mechanical strength of solid
catalysts: Recent developments and future prospects. AIChE J. 2007, 53,
26182629.
(11) Wu, D. F.; Li, Y. D. Effects of the impregnating and drying process
factors on the mechanical properties of a PCoMo/Al
2
O
3
hydrotreating
catalyst. Stud. Surf. Sci. Catal. 2002, 143, 101109.
(12) Santhanam, N.; Conforti, T. A.; Spieker, W. A.; Regalbuto, J. R.
Nature of metal catalyst precursors adsorbed onto oxide supports. Catal.
Today 1994, 21, 141156.
(13) Bergweff, J. A.; Lysova, A. A.; Alonso, L. E.; Koptyug, I. V.;
Weckhuysen, B. M. Probing the transport of paramagnetic complexes inside
catalyst bodies in a quantitative manner by magnetic resonance imaging.
Angew. Chem., Int. Ed. 2007, 46, 72247227.
(14) Bergweff, J. A.; Lysova, A. A.; Alonso, L. E.; Koptyug, I. V.;
Weckhuysen, B. M. Monitoring transport phenomena of paramagnetic metal-
ion complexes inside catalyst bodies with magnetic resonance imaging.
Chem.sEur. J. 2008, 14, 23632374.
(15) Lysova, A. A.; Koptyug, I. V.; Sagdeev, R. Z.; Parmon, V. N.;
Bergwerff, J. A.; Weckhuysen, B. M. Noninvasive in situ visualization of
supported catalyst preparations using multinuclear magnetic resonance
imaging. J. Am. Chem. Soc. 2005, 127, 1191611917.
(16) Koptyug, I. V.; Fenelonov, V. B.; Khitrina, L.; Sagdeev, R. Z.;
Parmon, V. N. In situ NMR imaging studies of the drying kinetics of porous
catalyst support pellets. J. Phys. Chem. B 1998, 102, 30903098.
(17) Koptyug, I. V.; Kabanikhin, S. I.; Iskakov, K. T.; Fenelonov, V. B.;
Khitrina, L.; Sagdeev, R. Z.; Parmon, V. N. A quantitative NMR imaging
study of mass transport in porous solids during drying. Chem. Eng. Sci.
2000, 55, 15591571.
(18) Bergweff, J. A.; van de Water, L. G. A.; Visser, T.; de Peinder, P.;
Leliveld, B. R. G.; de; Jong, K. P.; Weckhuysen, B. M. Spatially resolved
Raman and UV-visible-NIR spectroscopy on the preparation of supported
catalyst bodies: controlling the formation of H
2
PMo
11
CoO
40
5-
inside Al
2
O
3
pellets during impregnation. Chem.sEur. J. 2005, 11, 45914601.
(19) van de Water, L. G. A.; Bezemer, G. L.; Bergweff, J. A.; Helder,
M. V.; Weckhuysen, B. M.; de Jong, K. P. Spatially resolved UV-vis
microspectroscopy on the preparation of alumina-supported Co Fischer-
Trospch catalysts: Linking activity to Co distribution and speciation. J. Catal.
2006, 242, 287298.
(20) van de Water, L. G. A.; Bergweff, J. A.; Leliveld, B. R. G.;
Weckhuysen, B. M.; de Jong, K. P. Insights into the preparation of supported
catalysts: A spatially resolved Raman and UV-vis spectroscopic study iinto
the drying process of CoMo/-Al
2
O
3
catalyst bodies. J. Phys. Chem. B 2005,
109, 1451314522.
(21) Assaf, E. M.; Jesus, L. C.; Assaf, J. M. The active phase distribution
in Ni/Al
2
O
3
catalysts and mathematical modeling of the impregnation
process. Chem. Eng. J. 2003, 94, 9398.
(22) Galarraga, C.; Peluso, E.; Lasa, H. Eggshell catalysts for Fischer-
Tropsch synthesis modeling catalyst impregnation. Chem. Eng. J. 2001,
82, 1320.
(23) Komiyama, M. Design and preparation of impregnated catalysts.
Catal. ReV.-Sci. Eng. 1985, 27, 341372.
(24) Lekhal, A.; Khinast, J. G.; Glasser, B. J. Predicting the effect of
drying on supported co-impregnation catalysts. Ind. Eng. Chem. Res. 2001,
40, 39893999.
(25) Lekhal, A.; Glasser, B. J.; Khinast, J. G. Impact of drying on the
catalyst prole in supported impregnation catalysts. Chem. Eng. Sci. 2001,
56, 44734487.
(26) Lekhal, A.; Glasser, B. J.; Khinast, J. G. Inuence of pH and ionic
strength on the metal prole of impregnation catalysts. Chem. Eng. Sci.
2004, 59, 10631077.
(27) Liu, X.; Khinast, J. G.; Glasser, B. J. A parametric investigation
of impregnation and drying of supported catalysts. Chem. Eng. Sci. 2008,
63, 45174530.
(28) Neimark, A. V.; Fenelonov, V. B.; Kheifets, L. I. Analysis of the
drying stage in the technology of supported catalysts. React. Kinet. Catal.
Lett. 1976, 5, 6772.
(29) Uemura, Y.; Hatate, Y.; Ikari, A. Formation of nickel concentration
prole in nickel/alumina catalyst during post-impregnation. J. Chem. Eng.
Jpn. 1973, 6, 117123.
(30) Reinhoudt, H. R.; Troost, R.; Langeveld, A. D. v.; van, J. A. R.;
Veen, S. T. S.; Moulijn, J. A. The Nature of the Active Phase in Sulded
NiW/-Al
2
O
3
in Relation to Its Catalytic Performance in Hydrodesulfur-
ization Reactions. J. Catal. 2001, 203, 509515.
(31) Santos, R. M.; Lisboa, J. S.; Passos, F. B.; Noronha, F. B.
Characterization of Steam-Reforming Catalysts. Braz. J. Chem. Eng. 2004,
21, 203209.
(32) Papageorgiou, P.; Price, D. M.; Gavriilidis, A.; Varma, A. Prepara-
tion of Pt/-Al
2
O
3
pellets with internal step-distribution of catalyst:
experiments and theory. J. Catal. 1996, 158, 439451.
(33) Melo, F.; Cervello, J.; Hermana, E. Impregnation of porous sup-
ports-I Theoretical study of the impregnation of one or two active species.
Chem. Eng. Sci. 1980, 35, 21652174.
(34) Sceiza, O. A.; Castro, A. A.; Ardlles, D. R.; Parera, J. M. Modeling
of the impregnation step to prepare supported Pt/Al
2
O
3
catalysts. Ind. Eng.
Chem. Res. 1986, 25, 8488.
(35) Krishna, R. A simplied procedure for the solution of the dusty
gas model equations for steady state transport in non-reacting systems.
Chem. Eng. J. 1987, 35, 7581.
(36) Wesselingh, J. A.; Krishna, R. Mass Transfer, 2nd ed.; Chichester,
U.K.: Ellis Horwood: Chichester, England, 1998.
(37) Jones, P. J. Petroleum Production; Reinhold: New York, 1946.
(38) Gomis, V.; Font, A.; Saquete, M. D. Vapour-liquid-liquid and
vapour-liquid equilibrium of the system water + ethanol + heptane at
101.3 kPa. Fluid Phase Equilib. 2006, 248, 206210.
(39) Hailwood, A. J.; Horrobin, S. Absorption of water by polymers:
Analysis in terms of a simple model. Trans. Faraday Soc. 1946, 42B, 84
92.
(40) Simpson, W. T. Predicting equilibrium moisture content of wood
by mathematical models. Wood Fiber 1973, 5, 4149.
(41) Dullien, F. A. L. Porous Media: Fluid Transport and Pore
Structure, 2nd ed.; Academic Press: San Diego, CA, 1992.
(42) Perre, P.; Moser, M.; Martin, M. Advances in transport phenomena
during convective drying with superheated steam and moist air. Int. J. Heat
Mass Transfer 1993, 36, 27252746.
(43) Newman, J. Electrochemical Systems, 2nd ed.; Prentice-Hall:
Englewood Cliffs, NJ, 1991.
(44) Krishna, R. A simplifed procedure for the solution of the dusty
gas model equations for steady state transport in nonreacting systems. Chem.
Eng. J. 1987, 35, 7581.
(45) Patankar, S. V. Numerical Heat Transfer and Fluid Flow; McGraw-
Hill: New York, 1980.
(46) Deuhard, P.; Hairer, E.; Zugck, J. One-step and extrapolation
methods for differential-algebraic systems. Numer. Math. 1987, 51, 501
516.
(47) Sietsma, J. R. A.; Meeldijk, J. D.; Versljijs-Helder, M.; Broersma,
A.; van Dillen, A. J.; de Jongh, P. E.; de Jong, K. P. Ordered mesoporous
silica to study the preparation of Ni/SiO
2
ex nitrate catalysts: Impregnation,
drying, and thermal treatments. Chem. Mater. 2008, 20, 29212931.
(48) Takahashi, R.; Sato, S.; Sodesawa, T.; Kamomae, Y. Measurement
of the diffusion coefcient of nickel nitrate in wet silica gel using UV-vis
spectroscope equipped with a ow cell. Phys. Chem. Chem. Phys. 2000, 2,
11991204.
(49) Innocentini, M. D. M.; Salvini, V. R.; Macedo, A.; Pandolfelli,
V. C. Prediction of ceramic foams permeability using Erguns equation.
Mater. Res. 1999, 2, 283289.
(50) Smith, E. B. Basic Chemical Thermodynamics; Clarendon Press:
Oxford, UK, 1993.
(51) Regalbuto, J. R. Catalyst Preparation; CRC Press, Boca Raton,
FL, 2007.
ReceiVed for reView September 16, 2009
ReVised manuscript receiVed February 1, 2010
Accepted February 2, 2010
IE9014606
Ind. Eng. Chem. Res., Vol. 49, No. 6, 2010 2657

You might also like