Synthesis and characterization of nano-composite alumina-titania ceramic membrane for gas separation

Article abstract of DOI:10.1111/j.1551-2916.2006.01183.x

A ceramic composite membrane was prepared using a commercial titania ceramic membrane coated by alumina oxide via a sol-gel technique where polyvinyl alcohol (PVA) was used as a binder. The characteristic of the membrane was analyzed in terms of the effect of PVA concentration and sintering temperature on viscosity, pore size, density, porosity and surface area of the membrane. Two vol% of PVA solution containing 4 g of PVA in 100 mL of water was adequate to achieve an appropriate porosity level to avoid cracks on the gel layer. Sol viscosity and pore size of the membrane essentially increased when the PVA concentration was increased. The density of the membrane increased as the sintering temperature increased. The porosity level however, decreased when the temperature was increased. The composite membrane was further characterized in terms of permeability of pure gas at low-temperature region (301 K) where an experimental platform has been developed to perform the permeability studies.

Full text of DOI:10.1111/j.1551-2916.2006.01183.x

J. Am. Ceram. Soc., 89 [10] 3187–3193 (2006)
DOI: 10.1111/j.1551-2916.2006.01183.x
r 2006 The American Ceramic Society
ournal
J
Synthesis and Characterization of Nano-Composite Alumina–Titania
Ceramic Membrane for Gas Separation
A.L. Ahmad,w M.R. Othman, and N.F. Idrus
School of Chemical Engineering, Engineering Campus, Universiti Sains Malaysia, Seri Ampangan, 14300 Nibong Tebal,
Seberang Perai Selatan, Penang, Malaysia
A ceramic composite membrane was prepared using a commer-
cial titania ceramic membrane coated by alumina oxide via a
sol–gel technique where polyvinyl alcohol (PVA) was used as a
binder. The characteristic of the membrane was analyzed in
terms of the effect of PVA concentration and sintering temper-
ature on viscosity, pore size, density, porosity and surface area
of the membrane. Two vol% of PVA solution containing 4 g of
PVA in 100 mL of water was adequate to achieve an appropri-
ate porosity level to avoid cracks on the gel layer. Sol viscosity
and pore size of the membrane essentially increased when the
PVA concentration was increased. The density of the membrane
increased as the sintering temperature increased. The porosity
level however, decreased when the temperature was increased.
The composite membrane was further characterized in terms of
permeability of pure gas at low-temperature region (301 K)
where an experimental platform has been developed to perform
the permeability studies.
organic metal salts or metal or organic compounds such as met-
al alkoxides followed by the addition of acid and binders. The
thickened sol is then deposited or coated as a layer on a porous
support by dip- or spin-coating, followed by gelation of the layer
upon drying to form gel. Finally, the membrane is calcined and
sintered to form a ceramic membrane.
Gas separation across a porous ceramic composite membrane
may take place through different mechanisms, namely Knudsen
diffusion, surface diffusion or laminar flow. These mechanisms,
however, depend on the pore size of the membrane. The con-
tribution of surface diffusion is claimed to be significant in a
micro pore membrane (diameter o2 nm) and when the process
is operated at low-temperature and high-pressure region.2,3 For
a meso porous membrane having pore diameter between 2 and
50 nm, two mechanisms, normally Knudsen and surface diffu-
sion or the combination of both, occur in the membrane system.
However, Knudsen diffusion appeared to play a major role in
gas transport through alumina membrane.4 Viscous flow or Po-
isseuille flow was equivalent to laminar flow due to the assump-
tion that the flow across the cylindrical tube wall was laminar. In
brief, Poisseuille flow 5 viscous flow 5 laminar flow. Laminar
flow and surface diffusion, on the other hand, have minimal ef-
fect on the gas permeation rate, particularly at high-temperature
region and for an extremely fine pore membrane.5 For the case
of a composite membrane with separation layer system, the per-
meability equation can be calculated based on serial average
permeability from each separation layer as follows:
I. Introduction
CERAMIC composite membrane can be developed using
many methods. One of the common methods is by using
A
a metallic gel layer coated on the ceramic membrane. The gel
layer acts as the perm-selective to enhance the separation, and
sol–gel technique is the most common method for this purpose.
It is attractive for the application on a porous or composite/
asymmetric membrane. For that purpose, a sol solution is first
prepared using an organo-metallic oxide followed by the addi-
tion of binder. The viscous sol is then deposited or coated on the
support membrane by dip- or spin-coating, after which it is dried
in a controlled environment to form the perm-selective layer on
the support membrane. In the final stage, the membrane was
calcined and sintered to form an appropriate ceramic composite
membrane. One of the most prominent advantages of this tech-
nique is that it can produce an extremely fine pore distribution.
However, this process requires a specific binder with the right
concentration as the density and viscosity are very much de-
pendent on those parameters. Furthermore, appropriate calci-
nations and sintering temperature, and also the right choice of
membrane support, are very important in order to avoid cracks
on the perm-selective gel layer and undesirable pore size distri-
bution.
!
ꢀ1
n
X
1
Kaverage
¼
(1)
Ki
i¼1
where, Ki is permeability of pure gas, i across the membrane.3
Titania membrane is a commonly used ceramic membrane
for separation of gas mixture for both low and high temperature
processes such as manufacture of high purity hydrogen and wa-
ter gas shift reaction in the coal gasification process to produce
hydrogen from coal. It has received significant attention in re-
cent years because of its unique characteristics, such as high
water flux, semi-conductivity, catalysis and chemical resistance,
compared with other membrane materials.
In the current study, alumina membrane was selected to be
coated on the titania ceramic membrane to form the separation
layer. Alumina was used as the separation layer because of its
stability at high temperature and mechanical strength. Alumina
membrane is widely used for gas separation because of its high
gas separation factor and very fine pores that increase gas per-
meability.
The use of sol–gel technique to produce a ceramic membrane
for the separation and filtration process has been extensively
studied. It is known to be the most suitable process for making
thin porous film. The technique is simple and accurate in order
to synthesize fine pores separation layer on the membrane.1 In
the preparation of the sol, the starting materials are usually in-
The objective of this research is to synthesize and characterize
a novel ceramic alumina–titania nano-composite membrane for
gas separation at low-temperature region (301 K). The ceramic
alumina–titania composite membrane was developed using
sol–gel technique. The effect of binder on gel viscosity, mem-
brane surface morphology and pore size were analyzed. The ce-
ramic alumina–titania membrane was then synthesized and
characterized. The effect of sintering temperature on membrane
R. Koc—contributing editor
Manuscript No. 21288. Received December 26, 2005; approved May 16, 2006.
This work was financially supported by the research grant provided by Universiti Sains
Malaysia, Penang, Malaysia, that has enabled the production of this article.
wAuthor to whom correspondence should be addressed. e-mail: chlatif@eng.usm.my
3187

3188
Journal of the American Ceramic Society—Ahmad et al.
Vol. 89, No. 10
morphology and the pore size were examined. The ceramic
composite membrane was further characterized for separation
and permeability of pure gases (hydrogen, nitrogen, carbon
monoxide and carbon dioxide). The separation of binary gas
mixture at high-temperature region (1073K) has been studied
elsewhere.6 An experimental rig was designed and fabricated to
perform gas analysis and permeability. The commercial ceramic
titania membrane and ceramic alumina–titania composite mem-
brane were compared to study the effect of alumina-coated layer
on the permeability of the gases.
4001 to 10001C. The gel was sintered using a programmable
muffle furnace.
For composite titania membrane, commercial titania (sup-
plied by Fine Continents (USA) Corporation, Newark, NJ) was
used as a support. The external part of the membrane was ad-
hered with cellophane tape. The internal part was dipped into
boehmite sol for 27 s and dried under controlled conditions be-
fore the cellophane tape was discarded. The thickness of mem-
brane layer and dipping time depends on the viscosity of the sol.
The coated membrane was sintered at 4001C to obtain appro-
priate coating thickness.
(2) Membrane Characterization
II. Experimental Procedure
Quantachrome Autosorb (Quantachrome Corporation, Boyn-
ton Beach, FL) was used to determine the pore size distribution,
surface area and N2 adsorption–desorption isotherms of the
membrane. Surface morphology, structure of membrane, in-
cluding membrane thickness and separation layer, were analy-
zed using microscope and scanning electron microscope (SEM)
model Leica Cambridge S-360 (Leica Manufacturer, Cam-
bridge, U.K.). Two types of membrane samples were analyzed
by SEM; unsupported ceramic alumina membrane and support-
ed alumina membrane (ceramic alumina–titania composite
membrane). The surface and cross section of the alumina mem-
brane were analyzed to determine the homogeneity, texture and
the thickness of separation layer at the support.
(1) Preparation of Alumina Membrane
The complete flow diagram of membrane preparation is shown
in Fig. 1. The optimum molar ratio of aluminum sec-butoxide,
HNO3 and water used in the experiment to prepare the sol was
1:0.07:100 using the sol–gel method as developed by Yoldas.7
The boehmite sol was prepared by dissolving 12.8 mL aluminum
sec-butoxide in 90.3 mL de-ionized water while simultaneously
heated to 901C and vigorously stirred. 3.1 mL nitric acid
(HNO3) and 2.1 mL Polyvinyl alcohol (PVA) binder were
then added to the mixture and stirred under continuous condi-
tions at 901C under reflux for 24 h.
After reflux, the resulting sol mixture was poured into a Petri
dish. Next, the sol was dried at room temperature for 24 h to
obtain dried gel before sintering at temperatures ranging from
(3) Experimental Platform
An experimental rig was designed and fabricated to perform
mixed gas analysis and pure gas permeability experiments. Pure
gases used during the analysis were nitrogen, hydrogen, carbon
dioxide, and carbon monoxide supplied by Sitt Tatt Industrial
Gases Sdn. Bhd (Penang, Malaysia). The purity of the respective
gases was over 99.5%. The pressure of the feed gases from the
cylinders was regulated using factory-calibrated regulators sup-
plied by Malaysian Oxygen Berhad (Petaling Jaya), whereas the
flow of the gases was controlled individually via a mass flow
controller (AALBORGs) supplied by Bioclear Sdn. Bhd (Pu-
chong, Selangor). Four mass flow controller’s transducers were
each connected to a controllers command module. All of the
transducers are capable of operating at pressure of up to 500 psi
(3400 kPa). The mass flow controller’s command module is ca-
pable of displaying four channel readouts in addition to con-
trolling the flow of feed gas at flow rate up to 100 mL/min per
channel. Both flow meter and controller command modules re-
port accuracy within 70.1% of full scale.
Al(OC H )
4
9 3
+H O
2
Vigorous stirring
Add HNO – to peptize
3
the sol
Add polyvinyl alcohol
(PVA) – binder
A complete flow system is depicted in Fig. 2. The rig allows
flexibility of conducting both co-current and cross flow exper-
iments. However, only the cross flow experiment was selected.
According to Tsai et al.,8 cross flow pattern eliminates the prob-
lem of back diffusion, and the use of permeate side sweeping
gases, such as in co-current flow patterns commonly used in
laboratory practice, are impractical in industrial application be-
cause they increase energy consumption.8
Reflux – to stabilize the
sol
Dry to form thin
membrane gel layer
(4) Gas Permeability
For a gas system, pure gases from mass flow controllers were
allowed to converge at a union cross and flow into different size
tubing in order to create turbulence that would produce a ho-
mogenized gas. The gas was then diverted to the membrane
module. For compositional analysis of a single gas in the per-
meate stream, a gas chromatography was utilized (Shimadzu
model C-R8A Chromatopac, Shimadzu, Japan). It was con-
nected online utilizing a molecular sieve column (2 m length, 4
mm OD, 3 mm ID, 60/80 mesh range), Porapak Q column (2 m
length, 4 mm OD, 3 mm ID, 50/80 mesh range) and TCD de-
tector. The analysis began with identification of peaks for pure
gases and standard mixed gases, followed by determination of
the desired gas composition. For accurate and reproducible rep-
Calcine to decompose
boehmite crystal into:
γ-Al O + H O
2
3
2
Sinter at 400C, 600C, 800C and
1000C to form thin ceramic
alumina (Al O 0 layer
2
3
Fig. 1. Alumina membrane preparation using sol–gel technique.

October 2006
Synthesis and Characterization of Nano-Composite Alumina–Titania Ceramic Membrane
3189
was increased. From Table I and Table II, the abrupt change in
coating time as viscosity increased when PVA in sol was in-
creased from 2% to 3% was observed. However, more contin-
uous change in coating time was observed from Table III and
Table IV. This is because of significant change of viscosity at the
region as viscosity is related to the coating time and shown as:
H (Channel 1)
Feed gas
N (Channel 4)
CO (Channel 3)
t
¼ 30:91 ꢀ 0:698 m
(2)
ðsecÞ
ðcpÞ
Mass flow controller
(4-channels)
where t is coating time and m is viscosity of sol.
Thermocouple
display
The higher PVA content and concentration in the sol formed
a membrane layer with fewer to no cracks as observed from
Tables I, II, III and IV. The membrane was free from defects
such as cracks and pinholes when it was observed under a mi-
croscope, although a slightly deformed shape was observed for
the membrane developed using at least 2 vol% PVA containing
4 g PVA dissolved in 100 mL de-ionized distilled water. The
deformation that existed was actually due to the fluctuation of
humidity level during the drying process. However, as PVA
content was increased, the deformation was not observed. This
might be because the increase of PVA content helped in achiev-
ing steady drying.
Heating zone
Switch valve
no. 1
Pressure
gauge no.1
Pressure
gauge no.2
Open/close
valve no. 2
Open/close
valve no. 1
The mean pore size of the membrane was determined directly
from Autosorb analysis. The pore size of the membrane in-
creased with the increase in PVA addition as indicated in Tables
I, II, III and IV. Das and Maiti9 reported that higher PVB con-
tent promotes agglomeration of the ceramic particles causing
wider pore size and forming closed pores during firing. A min-
imum pore size was only achieved when a minimum amount of
PVA was used. Nevertheless, the existence of cracks on a mem-
brane was normal. Therefore, higher PVA concentration was
required to form a crack-free membrane, but with implication of
forming larger pores.
Condensor
Back
pressure
regulator
Mass flow meter no. 1
Switch valve no. 2
GC
Vent
Mass flow
meter no. 2
Vent
Switch valve no. 3
Vent
The amount of PVA used in the sol was optimized to form a
membrane with small pore size and without crack or other
structural defects. It was found that, 2 vol % of 4 g PVA dis-
solved in 100 mL de-ionized distilled water was sufficient to
successfully prepare the desired membrane without cracks or
pinholes. The alumina membrane prepared using the recom-
mended PVA concentration exhibits very narrow pore distribu-
tion with mean pore diameter of 1.6 nm.
Fig. 2. Schematic diagram of the experimental rig.
resentation of results, each reading was only recorded when the
permeate flow was stable.
Two units of CO detectors (model Txgard-IS/CO) with
detection ranging from 0 to 500 ppm and Gasman II/CO
(dangerous level of CO detection) were used to notify operators
of dangerous CO level in the rig vicinity in case of incidence such
as CO leaking from the tank or pipe connection.
(2) Effect of Sintering Temperature
Figure 3 shows the SEM picture of the cross section of the syn-
thesized alumina–titania composite membrane sintered at
4001C. The thin alumina separation layer was coated on the
commercial titania membrane. The texture of the membrane
was rough and nonhomogeneous at the support. The thickness
of the coating layer was about 5 mm.
Table V shows the effect of sintering temperature on the
membrane characteristic—density, porosity, pore size and sur-
face area. The density of the membrane increased as the sinte-
ring temperature increased. However, the porosity of the
membrane decreased from 26.2% to 8.2% as the sintering tem-
perature was raised from 4001 to 10001C. The decrease in po-
rosity was due to the decrease of micro pore volume with
increase in temperature because most of the micro pores evolved
to form meso pores and macro pores at high temperature.
III. Result and Discussion
(1) Effect of PVA Addition on the Gel Viscosity, Membrane
Morphology and Pore Size
The effect of PVA addition in boehmite sol during peptization
on the viscosity, coating time, pore size and Brunauer–Emmet–
Teller (BET) surface area is presented in Tables I, II, III and IV.
PVA was prepared by dissolving 2, 4, 6 and 8 g PVA powder
into 100 mL de-ionized distilled water. PVA with different con-
centration in the sol give a grade of characteristic viscosity
range. It was noted that the viscosity of the sol increased as
the PVA addition was increased. On the contrary, the coating
time and BET surface area decreased as PVA content in the sol
Table I. Effect of PVA Concentration on the Sol and Membrane Characteristics (2 g PVA Dissolved in 100 mL De-Ionized Distilled
Water)
Percent PVA
in Sol
Viscosity (cP)
Coating time (s)
Pore size (nm)
BET area (m2/g)
Membrane appearance
1
2
3
4
2.38
4.80
13.1
15.2
29
28
22
20
1.350
1.525
1.700
1.725
391
373
226
219
Translucent, many cracks with curl from edge
Translucent, slight crack with curl from edge
Translucent, slight crack with curl from edge
Translucent, almost without crack
PVA, Polyvinyl alcohol; BET, Brunauer–Emmet–Teller.

3190
Journal of the American Ceramic Society—Ahmad et al.
Vol. 89, No. 10
Table II. Effect of PVA Concentration on the Sol and Membrane Characteristics (4 g PVA Dissolved in 100 mL
De-Ionized Distilled Water)
Percent PVA
in sol
Viscosity (cP)
Coating time (s)
Pore size (nm)
BET area (m2/g)
Membrane appearance
1
2
3
4
3.02
5.28
14.7
15.0
29
27
21
20
1.450
1.600
1.725
1.750
355
322
219
201
Translucent, slight crack and curl from edge
Translucent, without crack
Translucent, without crack
Translucent, without crack
PVA, Polyvinyl alcohol; BET, Brunauer–Emmet–Teller.
Table III. Effect of PVA Concentration on the Sol and Membrane Characteristics (6 g PVA Dissolved in 100mL
De-Ionized Distilled Water)
Percent PVA
in sol
Viscosity (cP)
Coating time (s)
Pore size (nm)
BET area (m2/g)
Membrane appearance
1
2
3
4
8.78
27.2
32.6
39.1
25
12
8
1.675
1.725
1.780
1.870
288
205
150
182
Translucent, without crack
Translucent, without crack
Translucent, without crack
Translucent, without crack
4
PVA, Polyvinyl alcohol; BET, Brunauer–Emmet–Teller.
Table IV. Effect of PVA Concentration on the Sol and Membrane Characteristics (8 g PVA Dissolved in 100 mL
De-Ionised Distilled Water)
Percent PVA
in sol
Viscosity (cP)
Coating time (s)
Pore size (nm)
BET area (m2/g)
Membrane appearance
1
2
3
4
8.25
28.6
30.0
39.1
25
11
10
4
1.700
1.750
1.850
1.870
273
256
131
135
Translucent, without crack
Translucent, without crack
Translucent, without crack
Translucent, without crack
PVA, Polyvinyl alcohol; BET, Brunauer–Emmet–Teller.
The pore size of the membrane increased as the sintering
temperature was raised. The trend is similar with other pub-
lished data.10 The pore size increased from 1.6 to 2.0 nm as the
sintering temperature was increased from 4001 to 10001C. The
surface area of the membrane decreased with the increase of
sintering temperature. However, the decrease is in a small range
—241 m2/g compared with the work of Leenars et al.10, which
was 286 m2/g.
across alumina separation layer (Kseparator) can be calculated
from Eq. (1) using permeability of gas across composite
membrane as (Kaverage) and commercial titania membrane as
(Ktitania).
The permeability of pure gases across composite titania mem-
brane at 301 K was not proportional to the average pressure as
observed from Fig. 4. From the graph, the permeability of hy-
drogen was the highest followed by nitrogen, carbon monoxide
and carbon dioxide. The order of the permeability value follows
the molecular weight of the gases. The lower the molecular
weight of gas, the higher the permeability.
(3) Permeability of Gas Through Composite Membrane
The permeation of hydrogen, nitrogen, carbon monoxide and
carbon dioxide at low temperature, 301 K, through different
types of membranes was studied. The types of membranes used
were composite titania membrane, commercial titanial mem-
brane and alumina separation layer. The permeation of gas
The H2 permeation using composite titania membrane, com-
mercial titania membrane and alumina separation layer at 301 K
is shown in Fig. 5. Permeability of H2 through composite titania
membrane was low compared with commercial titania mem-
brane because of the small pore size in the separation layer,
as stated by Kusakabe et al.,11 and due to coarse pores of
the commercial titania membrane. The permeation of H2 was
high when permeated through the alumina separation layer.
Figure 6 shows the comparison of N2 permeability on com-
posite titania membrane, commercial titania membrane and
Alumina
perm-
selective
skin
Table V. Effect of Sintering Temperature on the Membrane
Characteristics (2 vol% of 4 g PVA Dissolved in 100 mL
De-Ionized Distilled Water)
Titania/
support
Sintering
BET surface area Mean pore size Density (g/ Porosity
(m2/g)
Temperature (1C)
(nm)
cm3)
(%)
400
600
800
1000
322
273
91
1.6
1.8
1.9
2.0
2.0
2.3
2.7
3.0
26.2
25.0
9.4
81
8.2
Fig. 3. Scanning electron microscope micrographs of the cross section
of the sintered alumina membrane with ꢁ 500 magnification.
PVA, Polyvinyl alcohol; BET, Brunauer–Emmet–Teller.

October 2006
Synthesis and Characterization of Nano-Composite Alumina–Titania Ceramic Membrane
3191
2.6E-13
2.1E-13
1.6E-13
1.1E-13
5.5E-14
5.0E-15
1.4E+06 2.2E+06 3.1E+06 3.9E+06 4.8E+06 5.7E+06 6.5E+06 7.4E+06 8.3E+06
Average pressure, x10−1 [kg.m−1.s−2
]
H
N
CO
CO
2
2
2
Fig. 4. Permeability of pure gas across titania composite membrane. (T5 301 K, rp 5 0.68 nm (separation layer), rp5 2 nm (support)).
6.5E-13
5.5E-13
4.5E-13
3.5E-13
2.5E-13
1.5E-13
5.0E-14
1.4E+06 2.2E+06 3.1E+06 3.9E+06 4.8E+06 5.7E+06 6.5E+06 7.4E+06 8.3E+06
Average pressure, x10−1 [kg.m−1.s−2
]
Composite titania membrane
Alumina separation layer
Commercial titania membrane
Fig. 5. Permeability of H2 across composite titania membrane, commercial titania membrane and alumina separation layer at 301 K.
2.1E-13
1.9E-13
1.7E-13
1.5E-13
1.3E-13
1.1E-13
8.5E-14
6.5E-14
4.5E-14
2.5E-14
5.0E-15
1.4E+06 2.2E+06 3.1E+06 3.9E+06 4.8E+06 5.7E+06 6.5E+06 7.4E+06 8.3E+06
−1
Average pressure, x10 [kg.m−1.s−2
]
Composite titania membrane
Alumina separation layer
Commercial titania membrane
Fig. 6. Permeability of N2 across composite titania membrane, commercial titania membrane and alumina separation layer at 301 K.

3192
Journal of the American Ceramic Society—Ahmad et al.
Vol. 89, No. 10
1.7E-13
1.5E-13
1.3E-13
1.1E-13
8.5E-14
6.5E-14
4.5E-14
2.5E-14
5.0E-15
1.4E+06 2.2E+06 3.1E+06 3.9E+06 4.8E+06 5.7E+06 6.5E+06 7.4E+06 8.3E+06
Average pressure, x10−1 [kg.m−1.s−2
]
Composite titania membrane
Alumina separation layer
Commercial titania membrane
Fig. 7. Permeability of CO across composite titania membrane, commercial titania membrane and alumina separation layer at 301 K.
1.3E-13
1.1E-13
8.5E-14
6.5E-14
4.5E-14
2.5E-14
5.0E-15
1.4E+06 2.2E+06 3.1E+06 3.9E+06 4.8E+06 5.7E+06 6.5E+06 7.4E+06 8.3E+06
Average pressure, x10−1 [kg.m−1.s−2
]
Composite titania membrane
Alumina separation layer
Commercial titania membrane
Fig. 8. Permeability of CO2 across composite titania membrane, commercial titania membrane and alumina separation layer at 301 K.
alumina separation layer at 301 K. The N2 permeability on alu-
mina sol–gel membrane was lower than commercial titania
membrane.
decrease of membrane thickness. The permeability of the other
gases are lower compared with H2 due to smaller molecule size.
The CO permeation across composite titania membrane,
commercial titania membrane and alumina separation
layer is shown in Fig. 7. The permeation of CO using alumina
sol–gel membrane was higher compared with composite
titania membrane. Figure 8 shows the permeation of CO2 at
301 K. The permeation of CO2 across alumina separation layer
decreased as average pressure was increased. However, the per-
meability through alumina separation layer was still higher
compared with the composite membrane. The negative slope
of CO2 permeability on alumina layer is negative due to the
occurrence of a slight back pressure when the pressure was
increased.
From Fig. 5 to Fig. 8, the permeation of H2, CO and CO2,
except N2, across alumina separation layer was higher than
composite titania membrane and commercial membrane. The
high permeability at the alumina separation layer was due to the
extremely thin layer (5 mm). Permeability was increased with the
IV. Conclusion
The crack formation in alumina membrane was avoided when
the boehmite sol with low viscosity was used. A 2 vol% of PVA
solution containing 4 g of PVA in 100 mL of water was found to
be adequate in achieving an appropriate porosity level to avoid
cracks in the gel layer. The effect of sintering temperature on
porosity, density, surface area and pore size distribution was
studied. The density of the membrane increased as the sintering
temperature was increased. However, the porosity level de-
creased as the sintering temperature was raised. As the poros-
ity decreased, the BET surface area also decreased while the pore
size of the membrane increased as the temperature was
increased. The permeability of hydrogen through composite
titania membrane was higher than nitrogen, carbon monoxide
and carbon dioxide.

October 2006
Synthesis and Characterization of Nano-Composite Alumina–Titania Ceramic Membrane
3193
6A. L. Ahmad, M. R. Othman, and H. Mukhtar, ‘‘H2 Separation From Binary
Gas Mixture Using Coated Alumina–Titania Membrane by Sol Gel Technique at
High Temperature Region,’’ Int. J. Hydrogen Energy, 29, 817–28 (2004).
7B. E. Yoldas, ‘‘Alumina Sol Preparation from Alkoxides,’’ Ceramic Bulletin,
54, 289–91 (1975).
References
1J. C. S. Wu and L. C. Cheng, ‘‘An Improved Synthesis of Ultrafiltration
Zirconia Membranes Via the Sol Gel Route Using Alkoxide Precursor,’’
J. Membr. Sci., 167, 253–61 (2000).
2J. Zaman and A. Chakma, ‘‘Inorganic Membrane Reactors,’’ J. Mater. Sci.,
92, 1–28 (1994).
8C. Y. Tsai, S. Y. Tam, Y. F. Lu, and C. J. Brinker, ‘‘Dual-Layer Asymmetric
Microporous Silica Membranes,’’ J. Membr. Sci., 169, 255–68 (2000).
9N. Das and H. S. Maiti, ‘‘Formation of Pore Structure in Tape-Cast Alumina
Membrane: Effects of Binder Content and Firing Temperature,’’ J. Membr. Sci.,
140, 205–12 (1998).
3R. J. R. Uhlhorn, M. H. B. J Huis Int’l Veld, K. Keizer, and A.J Burggraaf,
‘‘High Permselectivities of Microporous Silica Modified g-Alumina Membrane,’’
J. Mater. Sci., 8, 1135–8 (1989).
4Y. K. Cho, K. Han, and K. H. Lee, ‘‘Separation of CO2 by Modified
g-Alumina Membranes at High Temperature,’’ J. Membr. Sci., 104, 219–30
(1995).
10A. F. M. Leenars, K. Keizer, and A. J. Burggraaf, ‘‘The Preparation and
Characterization of Alumina Membranes With Ultra-Fine Pores,’’ J. Mater. Sci.,
19, 1077–88 (1984).
5H. M. Van Veen, J. P. B. M. Tol, C. W. R. Engelen, and H. J. Veringa, ‘‘High
Temperature Gas Separation With Alumina Membranes,’’ Key Eng. Mater., 62,
593–8 (1991).
11K. Kusakabe, T. Yamaki, H. Maeda, and S. Morooka, ‘‘Porous g-Alumina
Membrane Modified with Ultrafine Ziconia Particle Prepared by Reversed
Micelles Method,’’ ACS-DFC, 38 [1], 1–18 (1993).
&