You are on page 1of 8

Nanocrystalline TiO2 Photocatalytic Membranes with a Hierarchical

Mesoporous Multilayer Structure: Synthesis, Characterization, and


Multifunction**

FULL PAPER

DOI: 10.1002/adfm.200500658

By Hyeok Choi, Anna C. Sofranko, and Dionysios D. Dionysiou*


A novel solgel dip-coating process to fabricate nanocrystalline TiO2 photocatalytic membranes with a robust hierarchical mesoporous multilayer and improved performance has been studied. Various titania sols containing poly(oxyethylenesorbitan
monooleate) (Tween 80) surfactant as a pore-directing agent to tailor-design the porous structure of TiO2 materials at different
molar ratios of Tween 80/isopropyl alcohol/acetic acid/titanium tetraisopropoxide = R:45:6:1 have been synthesized. The sols
are dip-coated on top of a homemade porous alumina substrate to fabricate TiO2/Al2O3 composite membranes, dried, and calcined, and this procedure is repeated with varying sols in succession. The resulting asymmetric mesoporous TiO2 membrane
with a thickness of 0.9 lm exhibits a hierarchical change in pore diameter from 26, through 38, to 511 nm from the top to
the bottom layer. Moreover, the corresponding porosity is incremented from 46.2, through 56.7, to 69.3 %. Compared to a
repeated-coating process using a single sol, the hierarchical multilayer process improves water permeability significantly without sacrificing the organic retention and photocatalytic activity of the TiO2 membranes. The prepared TiO2 photocatalytic
membrane has great potential in developing highly efficient water treatment and reuse systems, for example, decomposition of
organic pollutants, inactivation of pathogenic microorganisms, physical separation of contaminants, and self-antifouling action
because of its multifunctional capability.

1. Introduction
Along with new and increasingly stringent regulations for
the discharge of effluents, which require a more reliable and
sustainable treatment process, the development of cost-effective membrane manufacturing has accelerated the use of membrane technology in wastewater treatment and reuse.[1,2] Membrane-based separations are energy efficient and cost effective
when optimized.[3] They represent promising alternatives to
conventional water and wastewater treatment processes as well
as air-purification and gas-separation processes.[1,2,4,5] Microporous and mesoporous inorganic membranes have attracted con-

[*] Prof. D. D. Dionysiou, H. Choi


Department of Civil and Environmental Engineering
University of Cincinnati
Cincinnati, OH 45221-0071 (USA)
E-mail: dionysios.d.dionysiou@uc.edu
A. C. Sofranko
Department of Chemical Engineering, University of Virginia
Charlottesville, VA 22904-4741 (USA)
[**] The authors acknowledge the financial support from the Office of
Biological and Physical Research of the National Aeronautics and
Space Administration (NRA Grant No: NAG 9-01475) and the
National Science Foundation (Grant No. BES 0448117 through a
CAREER Award to D. D. Dionysiou and Grant No: OCE 0304171).
Ms. Sofranko was a summer research fellow supported by the
National Science Foundation REU Site Program in Membrane
Applied Science and Technology (NSF Award No: 139 438, R. W. Millard, Ph.D., Principal Investigator).

Adv. Funct. Mater. 2006, 16, 10671074

siderable attention for water treatment due to their excellent


thermal, chemical, and mechanical stability, and their reusability after burning over conventional polymeric membranes (e.g.,
cellulose derivatives, polysulfone, polyamide, poly(vinylidene
fluoride) (PVDF), and polytetrafluoroethylene (PTFE)).[6,7]
Among the materials used for the preparation of inorganic
membranes, TiO2 photocatalysts have gained tremendous popularity, since a membrane skin layer composed of TiO2 material allows only small molecules and water to penetrate the
membrane pores (i.e., separation function) and its photocatalytic activity allows decomposition of recalcitrant organic pollutants in water.[8,9] The photocatalytic activity is also expected
to affect the anti-biofouling properties of the membrane, which
is one of the main goals of both research and industry in the
field of membrane-based water and wastewater treatment.[8,10]
TiO2/UV systems can generate hydroxyl radicals efficiently. In
fact, TiO2 photocatalysis has been found to be extremely effective for complete mineralization of virtually all organic compounds and for inactivation of pathogenic microorganisms
present in contaminated water.[1114]
An inorganic membrane generally consists of a macroporous
substrate providing mechanical strength for an overlying thin
active layer.[15,16] It is known that the quality of the underlying
support layer determines, to a large extent, the quality of the
top skin mesoporous titania membrane. A support surface with
large pore irregularity would cause cracking and pinholes in
the overlying skin layer because of uneven stress development
over the coatings during the solgel process. In order to reduce
surface roughness, conventional mechanical polishing could be

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1067

FULL PAPER

H. Choi et al./Hierarchical Mesoporous Multilayer TiO2 Photocatalytic Membranes


applied, but it is a time-consuming procedure impractical for
some membrane configurations. Therefore, it is important to
develop a simple and effective strategy for modifying the support surface to facilitate deposition of a defect-free TiO2 membrane. The use of an intermediate layer between the macroporous substrate and the mesoporous (or microporous) skin layer
has been introduced to overcome flow resistance, surface
roughness, and inherent support defects.[16,17] However, the formation of an intermediate layer itself also needs multiple coatings of an additional sol composed of a different material with
relatively large size particles. This results in increased preparation cost and time and reduced water permeability.
Alternatively, repetition of the dip-coating procedure using a
single sol until the desired thickness and homogeneity of the
coating layer are obtained has been suggested.[18,19] In this
method, the thick symmetric TiO2 coating layer consists of a
number of single coating layers, which results in an increase of
hydraulic resistance. The ideal membrane structure would be
composed of a thin asymmetric multilayer with increasing pore
size and porosity from the top to the bottom of the TiO2 skin
layer (i.e., a hierarchical multilayer). Compared to the usual
repeated coating prepared using the same mesoporous overlayers, a hierarchical multilayer can improve water permeation
flux significantly, maintaining the same organic retention and
photocatalytic activity as the typical repeated-coating process.
In this study, we report the preparation of photocatalytic
TiO2 membranes with a hierarchical mesoporous multilayer
structure using a solgel method modified with surfactant. It is
well known that the presence of templates such as surfactants
in solgel chemistry plays a crucial role in creating the porous
structure of the TiO2 inorganic network, which reduces the
hydraulic resistance of TiO2 membranes and enhances their
photocatalytic activity.[2023] In this method, the surfactant concentration largely affects the structural properties of the TiO2
material.[24,25] Controlling materials at the nanoscale makes it
possible to develop new types of catalytic membrane products
with tailor-designed properties exhibiting a hierarchical change
in the diameter of the mesopore from the top to the bottom of
the TiO2 skin layer.[26,27] The multifunctional capability of the
prepared TiO2 photocatalytic membrane, including use in decomposition of organic pollutants, inactivation of pathogenic
microorganisms, physical separation of contaminants, and selfantifouling action, are investigated.

the TiO2 and oxidation states other than TiO2. In order to determine its structural and crystallographic properties, the TiO2
material was collected by scraping the thin film surface and
analyzed. The structure and porosity of TiO2 material at the
nanoscale should not be affected by such an experimental procedure, and the nanostructure of TiO2 material in the membrane is considered identical to that in the film.[15,18]
Figure 1 shows the morphology of the nanostructured anatase TiO2 material. The control TiO2 material shown in Figure 1a demonstrated no distinct mesoporous structure. On the
other hand, TiO2 materials prepared with the surfactant,
poly(oxyethylenesorbitan monooleate) (Tween 80), were
highly porous and exhibited a distinct pore structure, as shown
in Figures 1b, e, and f. Compared to Figure 1b, Figures 1e and f
presented relatively larger pore sizes, which suggests that the
pore size increased slightly with increasing surfactant concentration. As illustrated in Figure 1c, all the TiO2 materials have
a slightly collapsed spherical (cubic) bicontinuous structure
with a highly interconnected network, inducing high porosity
values of 46 to 69 % as surfactant concentration increases.[28]

2. Results and Discussion


2.1. Surfactant Effect on the Properties of the TiO2 Material
Regardless of surfactant addition, significant organic content
and further weight loss of the TiO2 material calcined at temperatures greater than 500 C were not observed during Fourier
transform IR (FTIR) spectroscopy and thermogravimetric
analysis (TGA). A bulk energy-dispersive X-ray (EDX) elemental analysis showed that the TiO2 material was composed
of mainly Ti and O elements at a stoichiometry of around
1:1.891.95, possibly due to the presence of defect structures in

1068 www.afm-journal.de

Figure 1. Transmission electron microscopy images of TiO2 material prepared at: a) R = 0, bd) R = 1 at different magnifications, e) R = 2, and
f) R = 3, where R is the molar ratio of surfactant (Tween 80) to titania precursor (titanium tetraisopropoxide).

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2006, 16, 10671074

H. Choi et al./Hierarchical Mesoporous Multilayer TiO2 Photocatalytic Membranes

(a)

250

R=3

-1

Volume Adsorbed (cm g )

300

200
R=2

150

R=1

100
50

R=0

0
0.0

0.2

0.4

0.6

0.8

1.0

Relative Pressure (Ps/Po)

(b)

1.6

R=0
R=1
R=2
R=3

-1

Pore volume (cm g )

2.0

1.2
0.8
0.4
0.0
5

10

15

20

25

30

Pore diameter (nm)

Figure 2. a) N2 adsorptiondesorption isotherms and b) pore size distribution of the TiO2 material.

for a well-developed mesoporous material. A hysteresis loop in


the isotherms was observed with dissimilar shapes for the adsorption and desorption branches. The sharp drop on the desorption branch can be assigned to the presence of smaller
pores in the titania walls.[29] These results implied different
pore throat diameters and slightly disordered pores, in agreement with transmission electron microscopy (TEM) images.
Figure 2b demonstrates that the pore size distribution of TiO2
materials can be controlled by varying the surfactant concentration. The pore size distribution was shifted from 26,
through 38, to 511 nm when R was increased from 1 to 3.

Adv. Funct. Mater. 2006, 16, 10671074

Moreover, the corresponding structural properties of TiO2 materials were enhanced from 147, to 155, to 159 m2 g1 for the
BrunauerEmmettTeller (BET) specific surface area, and
from 46.2, to 56.7, to 69.3 % for materials porosity.
The underlying mechanism for the formation of a highly porous structure with controlled pore size is associated with the
role of the surfactant, which can act as a pore-structure-forming agent.[25,30] Amphiphilic molecules such as surfactants exist
in a wide range of ordered structures in their condensed
states.[31] Titanium alkoxide precursors, when added into the
self-organized surfactant solution, are hydrolyzed and condense around the surfactant templates. By increasing the surfactant concentration, the number of water molecules available
for association with the hydrophilic head group of the surfactant decreases, resulting in a decrease in the degree of hydration of the surfactant head group and thus a decrease in the
effective head-group area. Based on the critical-packing parameters, which determine the packing geometry of surfactant
assembly, a reduction of the head-group area increases the critical packing parameters, favoring a less-curved geometry (e.g.,
flat lamellar phase) rather than the more-curved spherical micelles.[24] During thermal treatment, the surfactants and other
organic residues are removed, leaving a pore structure that
mimics the properties of the surfactant packing geometry.
However, in this study, it was difficult to identify a change in
such pore geometry, since the calcination temperature of
500 C is believed to be high enough to destroy the initial ordered pore structure of TiO2 material.[21]
Moreover, the high surfactant concentrations used are expected to cause more pore coalescence and multimicellar interactions during heat treatment, resulting in high porosity and
large pores. Long-range ordering of the TiO2 pore network is
not definitive proof of improved materials for applications in
photocatalytic membranes. This is because the enhancement of
the crystallinity of the TiO2 material from 67 % at 350 C,
where the ordered pore structure started to collapse, to 94 % at
500 C, and the high amount of interconnection of the pore network during heat treatment at high temperatures can overwhelm disadvantages caused by the decrease in pore ordering
of TiO2 material.[21] As a result, in this solgel method, the controlled hydrolysis of the TiO2 precursor is achieved through
indirectly taking on water, and the stable incorporation of the
titania inorganic network into self-organized surfactant molecules could induce the formation of TiO2 photocatalysts with
tailor-designed structural properties. In addition, as the concentration of templating materials in a sol increases, higher porosity and relatively larger pore sizes could be created.
Figure 3 shows X-ray diffraction (XRD) spectra of the TiO2
materials. All peaks are assigned to the anatase crystal phase,
which is known to be the most-active phase. The relatively
large width of the peaks indicates a small crystallite size, which
was in good agreement with the TEM results. The crystallite
size was estimated to be approximately 12.4 nm at R = 0,
9.20 nm at R = 1, 8.85 nm at R = 2, and 9.46 nm at R = 3, using
Scherrers equation from the XRD peak broadening analysis at
(101).[32] The 810 nm crystallite size of TiO2 has been reported to be the optimum for high catalytic activity.[33]

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

FULL PAPER

This interconnected pore-network phase would be much more


attractive than a 2D hexagonal phase for applications in photocatalysis requiring diffusion of species into and out of the pore
network, and in membrane separations requiring water permeability. Figure 1d at high magnification showed obvious
46 nm pores and many randomly oriented 510 nm nanocrystallites with sets of clearly resolved lattice fringes, giving evidence that the TiO2 material was highly crystalline. The porous
structure was thermally stable, since the structure was largely
retained until heat treatment at 600 C.
The pore size controllability of the TiO2 materials was evidenced by measuring N2 adsorptiondesorption isotherms, as
shown in Figure 2. Compared to the N2 isotherms of the control TiO2 (representing a nonporous material), those of TiO2
prepared with Tween 80 exhibited type IV isotherms, typical

www.afm-journal.de 1069

FULL PAPER

H. Choi et al./Hierarchical Mesoporous Multilayer TiO2 Photocatalytic Membranes

Figure 3. XRD patterns of TiO2 material. TTIP: titanium tetraisopropoxide.

The main structural characteristics deduced from the analyses so far are summarized in Table 1. In spite of the high heattreatment temperature of 500 C for the TiO2 material prepared with Tween 80, the BET surface area of above 147 m2 g1,
pore volume of above 0.221 cm3 g1, and porosity of above
46 % were significantly high compared to other research results

the anatase crystal phase. Even though the film thickness increased with increasing surfactant concentration, the amount
of TiO2 catalyst decreased due to the high materials porosity.
The results so far show that the solgel method has good potential for application in fabricating hierarchical mesoporous
multilayer TiO2 membranes with less hydraulic resistance.
The bandgap energy Eg of the TiO2 materials was slightly
larger than the 3.23 eV value for typical bulk anatase TiO2 particles. This blue-shift was caused by small crystallites less than
10 nm, owing to the quantum size effect.[35] The crystallite size
effect on the bandgap energy was consistent with the observed
result that smaller crystallites in TiO2 prepared with Tween 80
have higher bandgap energies. The TiO2 stoichiometry in the
material surface measured using X-ray photoelectron spectroscopy (XPS) was slightly different from that in the bulk measured using EDX due to the presence of many hydroxyl groups
at the material surface. High-resolution XPS analysis for determining the Ti oxidation states at 2p3/2 level, corresponding to a
binding energy of 453460 eV, showed that TiIV is predominant, with a small fraction of TiIII. It should be noted that the
surfactant addition in this solgel method did not affect the
TiO2 stoichiometry and titanium oxidation state. However,
compared to the relatively pure control TiO2 at R = 0, TiO2 materials prepared with Tween 80 had trace amount of impurities
including carbon, fluoride, nitrogen, and phosphorous.

Table 1. Physicochemical properties of TiO2 materials. BJH: BarrettJoynerHalenda. [a]


Parameter

Control
R=0

Surfactant concentration
R=1

R=2

R=3

BET specific surface area [m2 g1]


Pore volume [cm3 g1]
Porosity [%]
BJHadsorption pore diameter [nm]
BJHdesorption pore diameter [nm)
Crystal phase
Crystallite size [nm]
Film thickness [nm]
TiO2 mass [lg cm2 ]
Band gap energy [eV]
Ti/O stoichiometry in bulk
Ti/O stoichiometry in surface
TiO2 (IV)/Ti2O3 (III)
Traces [at %]

22.7
0.037
12.6
5.65
5.38
Anatase
12.4
87
29.6
3.26
1:1.94
1:2.06
1:0.15
C: 0.8
F: < 0.2
N: < 0.2
P: <0.2

147
0.221
46.2
4.04
3.72
Anatase
9.20
103
21.6
3.34
1:1.92
1:2.18
1:0.13
C: 1.3
F: 0.7
N: 0.7
P: 0.3

155
0.337
56.7
5.37
4.98
Anatase
8.85
110
18.5
3.36
1:1.89
1:2.19
1:0.13
C: 1.5
F: 0.8
N: 0.7
P: < 0.2

159
0.579
69.3
6.89
6.29
Anatase
9.46
113
13.5
3.29
1:1.95
1:2.17
1:0.12
C: 1.8
F: 0.9
N: 0.9
P: 0.3

[a] Scraped from TiO2 thin films on glass substrate.

reported and to the control TiO2 material, with a surface area


of 22.7 m2 g1, pore volume of 0.037 cm3 g1, and porosity of
12.6 %.[20,34] The BarrettJoynerHalenda (BJH) pore diameter based on the adsorption branch was similar to that based on
the desorption branch, implying good homogeneity of pores.
The film thickness of around 100 nm per coating was measured
using environmental scanning electron microscopy (ESEM),
and it increased slightly due to the increased viscosity of the
sols with increasing surfactant concentration. The amount of
TiO2 mass was calculated from the pore volume and density of
1070 www.afm-journal.de

2.2. TiO2 Multicoating


The integrity of the TiO2 skin layer and its incorporation
with the support layer are crucial factors in the fabrication of
defect-free TiO2 membranes. In preliminary experiments using
ESEM to examine the morphology of TiO2/Al2O3 composite
membranes, we observed that in this solgel method at least
three coating layers (i.e., T = 3) are necessary to fabricate a
TiO2 skin layer with good integrity and without significant
cracks and pinholes. A multicoating with three layers was fabricated with sols at different concentrations in the order R = 3,
R = 2, and R = 1 in succession, or a repeated-coating with three
layers was fabricated with a single sol at R = 1 (otherwise specified, R = 2 or R = 3). The top layer of the TiO2 membrane in
both cases was prepared with the sol at R = 1 to achieve the
same extent of organic retention. Figure 4 shows an ESEM image of the multicoating TiO2/Al2O3 composite membrane. The
thickness of the TiO2 skin layer was uniform at approximately
0.9 lm, which is three times thicker than the TiO2 coating on a
smooth glass substrate due to the surface irregularity and porous structure of the alumina substrate. The TiO2 membranes
were relatively well incorporated into the alumina substrate.
In order to visually investigate how the multicoating layers
are developed and connected to each other at the nanoscale,
the sols were coated onto glass substrates in the same manner
as the TiO2/Al2O3 membranes, since the TiO2 multicoating
layer on the alumina substrate does not satisfy sample conditions required for TEM analysis. The multicoating layer was
carefully scratched from the film and sonicated for 1 h to thoroughly disturb the multilayer, in order to observe a contrast be-

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2006, 16, 10671074

H. Choi et al./Hierarchical Mesoporous Multilayer TiO2 Photocatalytic Membranes

FULL PAPER

2 m

Figure 4. ESEM image of TiO2 membrane.

tween the layers. In general, it is very difficult to find the multilayer since a side-view section of the film hardly appears under
TEM analysis conditions. Such a multilayer structure is illustrated in Figure 5. Each layer, with a thickness of around
100 nm, was well developed, and the hierarchical mesoporous
structure previously explained in Figures 1 and 2 is also shown.
Taking into account the long sonication time of 1 h, the multilayer structure is considered to be robust.

2.3. Water Permeability and Organic Retention


Figure 6 shows water permeability and poly(ethylene glycol)
(PEG) retention of some TiO2/Al2O3 composite membranes
prepared using multicoating and repeated-coating processes,
and the properties are summarized in Table 2. The permeate
water flux of the membranes with one or two coating layers
was not significantly different from that of the porous substrate
itself. After three coatings, for both the multicoating membrane and the repeated-coating membrane, the TiO2 skin layer
was almost homogeneous, had a thickness of 0.9 lm, and exhibited a significant decrease in the permeate water flux. The
Figure 5. TEM images of TiO2 films with a multicoating layer: a) three
permeability coefficient decreased from 11.0 L m2 h1 bar1
layers prepared at R = 3, 2, and 1 in succession, b) the boundary between
2 1
1
(1 bar = 100 kPa) for the alumina support to 7.69 L m h bar
R = 3 and 2, and c) the boundary between R = 2 and 1.
for the multicoating membrane, and 6.71 L m2 h1 bar1 for the
repeated-coating membrane prepared with R = 1. The correwas 13 000 Da, which was similar to 12 000 Da for the
sponding hydraulic-filtration resistance increased from
repeated-coating layer. This result implies that most of the
3.23 1013 m1 for the alumina support to 4.59 1013 m1 for
organic retention occurred at the very surface of membranes,
the multicoating membrane, and 5.26 1013 m1 for the rethe top layer prepared at R = 1. The MWCO of the membranes,
peated-coating membrane. The water permeability of the mulwhich is equivalent to a molecule diameter of approximately
ticoating layer was significantly increased to 15 %, compared
to the repeated-coating membrane.
The water permeability of the TiO2
Table 2. Properties of TiO2 membranes.
membranes was in good agreement with
their PEG retention. As the number of
Parameter
Substrate
Multicoating
Repeated-coating
coating layers increased, lower-molecularT=1
T=2
T=3
T=3
(T=3)
(T=3)
weight PEG was rejected by the memR=3
R=3-2 R=3-2-1
R=1
(R=2)
(R=3)
branes. At least three coating layers were
3.23
3.39
3.73
4.59
5.26
4.71
3.93
Filtration resistance, Rm [61013 m1]
needed for both the multicoating mem11.0
10.4
9.45
7.69
6.71
7.48
8.98
Permeability coeff., Lp [L m2 h1 bar1]
brane and the repeated-coating mem1.00
0.95
0.86
0.70
0.61
0.68
0.82
Lp/Lp,substrate
Thickness [lm]
22 [mm]
na
0.6
0.9
0.9
0.9
0.9
brane to ensure the removal of small molMolecular weight cut-off [Da]
0.1 [lm] >>20k
>>20k
13k
12k
17k
>20k
ecules. The molecular weight cut-off
na
na
>96.19
161.1
188.4
147.4
107.5
Mass of TiO2 [lg cm2]
(MWCO) of the multicoating membrane

Adv. Funct. Mater. 2006, 16, 10671074

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.afm-journal.de 1071

(a)
Substrate
T=0

60
-1

T=1, R=3
T=2, R=3-2

-2

T=3, R=3-2-1
T=3, R=1-1-1

40

20

Normalized concentration, C/Co

80

Flux (L m h )

FULL PAPER

H. Choi et al./Hierarchical Mesoporous Multilayer TiO2 Photocatalytic Membranes

(a)

UV without TiO2

1.0

TiO2 without UV

0.8
0.6

T=3, R=1-1-1

0.4
T=3, R=3-2-1

0.2
0.0
0.0

0
0

0.5

1.0

1.5

2.0

2.5

3.0

Reaction time (h)

-2

Retention (%)

100

(b)
T=3, R=1-1-1
T=3, R=3-2-1

80
T=2, R=3-2

60
40

T=1, R=3

Log inactivation of E. coli (N/N0)

Pressure (kg cm )

(b)
0

TiO2 without UV

-1
-2

UV without TiO2

-3
TiO2 / UV

-4

20
Substrate
T=0

0.5

1.0

1.5

2.0

2.5

3.0

Reaction time (h)

104

67 nm, is relatively consistent with the pore size distribution


of the TiO2 material prepared at R = 1 (Fig. 2). As a result,
compared to a repeated-coating process using a single sol, the
hierarchical multilayer process prepared with the various sols
improved water permeability significantly, without sacrificing
organic retention of the TiO2 membranes.

2.4. Photocatalytic Activity and Antifouling Properties


The photocatalytic activity of TiO2 membranes was examined in terms of methylene blue (MB) dye decomposition and
Escherichia coli inactivation, as shown in Figure 7. Compared
to no direct photolysis of MB in the absence of TiO2 photocatalyst, the TiO2/UV photocatalytic system effectively decolorized the initial blue color of MB within 3 h. The photocatalytic
activity of the multicoating membrane was similar or even
slightly higher than that of the repeated-coating membrane. In
order to achieve 4-log inactivation of E. coli, the TiO2/UV photocatalytic process needed less than 1.5 h, while the UV disinfection required more than 3 h. Considering that 0.050
0.200 % m/v (TiO2 mass/reaction volume) of TiO2 photocatalyst is usually used in suspension for the decomposition of contaminants in water, the high photocatalytic activity per TiO2
mass in these membranes (0.0200.023 % m/v) shows great
promise.[36] The high photocatalytic efficiency of the TiO2
membrane is attributed to their porous structure with high sur-

face area, which is able to facilitate adsorption of water contaminants and effective utilization of UV light.
Inherent anti-biofouling in a membrane is highly important
in membrane research and industry. As shown in Figure 8, interestingly, TiO2 membranes irradiated by UV exhibited less
flux decline and no significant fouling formation over time,
compared to a control experiment without UV. This difference
is attributed to the photocatalytic and photolytic activity of the
TiO2/UV system. While organic contaminants and microorganisms attach at the TiO2 membrane surface and interact to
form an adsorption fouling layer in static condition, they are

23.5
-1

Figure 6. a) Water permeability and b) PEG retention of TiO2 membranes.


Note that T is the number of dip-coatings.

Figure 7. a) Photocatalytic degradation of methylene blue dye and b) photocatalytic inactivation of pathogenic microorganism, E. coli, by TiO2
membranes.

-2

Molecular weight of PEG (g mol-1)

1072 www.afm-journal.de

0.0

Permeate water flux (L m h )

0
103

23.0
22.5
with UV

22.0
without UV

21.5
21.0
0

10

20

30

40

50

60

Contact time (min)

Figure 8. Antifouling properties of TiO2 membranes.

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2006, 16, 10671074

H. Choi et al./Hierarchical Mesoporous Multilayer TiO2 Photocatalytic Membranes

3. Conclusions
A novel dip-coating process to synthesize hierarchical multilayer mesoporous TiO2 membranes with improved performance was studied, employing a surfactant templating strategy
in an acetic acid based solgel route. By multicoating a porous
alumina support layer with various sols at different surfactant
concentrations in succession, the resulting asymmetric, threelayer TiO2 membrane exhibited hierarchical changes in pore
diameter and porosity from 26 nm and 46.2 %, to 38 nm and
56.7 %, to 511 nm and 69.3 % from the top to the bottom of a
TiO2 skin layer with a total thickness of 0.9 lm. Compared to a
repeated-coating process using a single sol, the hierarchicalmultilayer-coating process improved water permeability of
TiO2 membrane significantly, without sacrificing its organic retention and photocatalytic activity. This solgel method is useful for the preparation of nanostructured TiO2 membranes with
high photocatalytic activity and desirable pore structures. The
prepared photocatalytic TiO2 membranes have great implications for environmental applications due to their simultaneous
photocatalytic, disinfection, separation, and antifouling functions.

4. Experimental
Sol Preparation: Common solgel methods employ direct water addition and lead to the precipitation of amorphous particles with uncontrolled structure that are then peptized in acidic conditions to make a
homogeneous sol [38]. An acetic acid-based solgel method, where
water molecules released from the esterification reaction of acetic acid
with isopropyl alcohol during synthesis, was utilized instead of directly
adding water [39].
Various template-based solgel methods using a variety of surfactants and block copolymers have been applied in the past to synthesize
tailor-designed TiO2 catalytic materials [30,34]. In this study, the acetic
acid based solgel method was modified with surfactants. Polyoxyethylenesorbitan monooleate (Tween 80, Aldrich) surfactant was selected as
the pore-directing agent. A suitable amount of Tween 80 was homogeneously dissolved in isopropyl alcohol (iPrOH, Fisher). Acetic acid
(Fisher) was added into the solution for the esterification reaction.
When titanium tetraisopropoxide (TTIP, Aldrich) was added to the solution, hydrolysis and condensation reactions occurred through the intake of water molecules released from the esterification reaction. The
molar ratio of the ingredients was optimized at Tween 80/iPrOH/acetic
acid/TTIP = R:45:6:1, where the surfactant concentration R was varied
from 0 for the control to 13 for target structural properties. At surfactant concentrations above R = 3, film homogeneity suddenly deceased.
Dip-Coating and Heat Treatment: Fabrication of a porous substrate
made of fine alumina powder was performed by modification of wellestablished procedures [40]. A16-SG alumina powder (Alcoa Chemicals) was mixed with a suitable amount of water, placed into a mold,
and pressed at 5000 lb (1 lb = 0.453 kg) for 30 s followed by 7500 lb for
3 min using a hydraulic unit (Model 3925, Carver). The pressed disks
were calcined and sintered up to 1220 C for 7 days in a furnace (Vulcan 3-400 HTA, Neytech) with programmed multistage temperatures
and holding times. The diameter and thickness of the alumina substrate
were 21 mm and 2.2 mm, respectively. The pore size and pure-water-

Adv. Funct. Mater. 2006, 16, 10671074

permeability coefficient were 0.1 lm and 11.0 L m2 h1 bar1, respectively. One side of the alumina substrate was dipped into the sol for
20 s and taken out with extreme care using a homemade dip-coating
device. After coating, the membranes were dried at room temperature
for 1 h, calcined in a multisegment programmable furnace (Paragon
HT-22D, Thermcraft) at a ramp rate of 3 C min1 up to 500 C, maintained at this temperature for 15 min, and cooled down naturally. This
dip-coating procedure was repeated with different sols at various surfactant concentrations, R, in succession. The number of dip-coatings
(denoted as T) was usually one or three. Due to the difficulty in directly
characterizing the properties of TiO2 materials on an alumina support
layer, easy-to-remove TiO2 films were prepared by dip-coating borosilicate glass (Micro slide, Gold Seal) with the same sols used for the preparation of the TiO2 membrane, at a withdrawal rate of 12.8 cm min1.
After the coating procedure, the other drying and calcination procedures were the same as for the TiO2 membranes. TiO2 material was collected by carefully scraping the thin film surface.
Materials Characterization: In order to determine the crystal structure and crystallinity of the TiO2 material, XRD analysis using a Kristalloflex D500 diffractometer (Siemens) with Cu Ka (k = 1.5406 )
radiation was employed. A porosimetry analyzer (Tristar 3000, Micromeritics) was used to determine structural characteristics of TiO2 materials including BET specific surface area, porosity, and pore size distribution in the mesoporous range, using nitrogen adsorption and
desorption isotherms. In order to determine the bandgap energy of
TiO2, its UV-vis light absorbance was measured using a UV-vis spectrophotometer (Hewlett Packard 8452 A). The structure of TiO2 materials
at the nanoscale was visualized using a JEM-2010F (JEOL) high-resolution transmission electron microscope with a field-emission gun. An
environmental scanning electron microscope (Philips XL 30 ESEMFEG) was used to measure the thickness of coatings and ensure the
homogeneity of TiO2 membranes. An elemental composition analysis
of the TiO2 was performed using an energy dispersive X-ray spectroscope (Oxford Isis) connected to the transmission and scanning electron microscopes. An X-ray photoelectron spectroscope (Perkin-Elmer
model 5300) with Mg Ka X-rays was used at a take-off angle of 45 for
the accurate measurement of trace elements in a few nanometers of
the material surface and oxidation states of titanium. An organic component in the TiO2 during heat treatment was monitored using Fourier
transform IR spectroscopy (Nicolet Magna-IR 760) and thermogravimetric analysis (TA instruments 2050).
Evaluation of TiO2 Membranes: The water permeability and organic
retention of the prepared TiO2/Al2O3 composite membranes were
evaluated with a home-fabricated membrane chamber. The pure water
was pressurized with a nitrogen gas at transmembrane pressures
(TMPs) from 0 to 6 kg cm2, and the permeate water flux was measured. Instead of pure water, 200 mg L1 of PEG (Fluka) solutions with
different molecular weights were pressurized at a TMP of 1 kg cm2 to
determine the molecular weight cut-off (MWCO) of the TiO2 membranes and investigate their integrity. Organic carbon concentrations in
the raw PEG solution and permeate water were measured using a total
organic carbon (TOC) analyzer (Shimadzu, TOC-V CSH). Experiments to examine the photocatalytic activity of TiO2 membranes were
conducted by immersing TiO2 membranes with 10 cm2 surface area
into a borosilicate glass reactor containing 8 mL of 30 lM methylene
blue dye (MB, Riedel-de Han), which is a commonly used dye in photocatalytic tests or 8 mL of 106107 cfu mL1 E. coli (ATCC 11 229;
cfu = colony forming units), which is a model pathogenic micro-organism for microbial inactivation tests. Two 15 W low-pressure mercury
UV tubes (Spectronics) emitting near-UV radiation with a peak at
365 nm were used at a light intensity of 3.48 mW cm2. The concentration of MB solution was determined by measuring the visible light absorbance at 664 nm using a UV-vis spectrophotometer (Hewlett Packard 8452A). The number of viable E. coli was counted using a colonyforming test where a membrane-filtration technique and ColiBlue24
broth (enrichment medium, HACH) were employed. The antifouling
properties of TiO2 membranes were investigated with a high-strength
organic solution obtained by filtering an activated sludge treating municipal wastewater using a 1.2 lm glass fiber filter. The organic solution
contained around 800 mg L1 dissolved solids (170 mg L1 organics and

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

FULL PAPER

also attacked by the photocatalytic and photolytic action and


thus the foulants are decomposed or their attachment strength
is weakened.[37]

www.afm-journal.de 1073

FULL PAPER

H. Choi et al./Hierarchical Mesoporous Multilayer TiO2 Photocatalytic Membranes


630 mg L1 inorganic) and around 120 mg L1 dissolved organic carbon.
Pure water flux was measured at a TMP of 3 kg cm2 after contacting
the membrane with the organic solution in static condition (i.e., TMP
of 0 kg cm2) in the presence or absence of UV radiation.
Received: September 26, 2005
Final version: November 4, 2005
Published online: April 10, 2006

[1] H. Choi, H.-S. Kim, I.-T. Yeam, D. D. Dionysiou, Desalination 2005,


172, 281.
[2] H. Choi, K. Zhang, D. D. Dionysiou, D. B. Oerther, G. A. Sorial, Sep.
Purif. Technol. 2005, 45, 68.
[3] G. Owen, M. Bandi, J. A. Howell, S. J. Churchouse, J. Membr. Sci.
1995, 102, 77.
[4] X. Tan, K. Li, W. K. Teo, AIChE J. 2005, 51, 1367.
[5] R. M. de Vos, W. F. Maier, H. Verweij, J. Membr. Sci. 1999, 158, 277.
[6] L. G. A. van de Water, T. Maschmeyer, Top. Catal. 2004, 29, 67.
[7] Y. S. Lin, Sep. Purif. Technol. 2001, 25, 39.
[8] S.-Y. Kwak, S. H. Kim, S. S. Kim, Environ. Sci. Technol. 2001, 35,
2388.
[9] D.-S. Bae, K.-S. Han, S.-H. Choi, Solid State Ionics 1998, 100, 239.
[10] H. Choi, K. Zhang, D. D. Dionysiou, D. B. Oerther, G. A. Sorial,
J. Membr. Sci. 2005, 248, 189.
[11] K. Yoo, H. Choi, D. D. Dionysiou, Catal. Commun. 2005, 6, 259.
[12] J.-M. Herrmann, Top. Catal. 2005, 34, 49.
[13] C. Chen, P. Lei, H. Ji, W. Ma, J. Zhao, H. Hidako, N. Serpone, Environ. Sci. Technol. 2004, 38, 329.
[14] D. D. Dionysiou, A. A. Burbano, M. T. Suidan, I. Baudin, J. M.
Lan, Environ. Sci. Technol. 2002, 36, 3834.
[15] T. V. Gestel, C. Vandecasteele, A. Buekenhoudt, C. Dortemont,
J. Luyten, R. Leysen, B. V. D. Bruggen, G. Maes, J. Membr. Sci. 2002,
207, 73.
[16] C.-Y. Tsai, S.-Y. Tam, Y. Lu, J. Brinker, J. Membr. Sci. 2000, 169, 255.
[17] A. J. Burggraaf, L. Cot, in Fundamentals of Inorganic Membrane
Science and Technology (Eds: A. J. Burggraaf, L. Cot), Elsevier, Amsterdam, The Netherlands 1996, p. 1.
[18] L.-Q. Wu, P. Huang, N. Xu, J. Shi, J. Membr. Sci. 2000, 173, 263.

[19] R. M. De Vos, H. Verweij, J. Membr. Sci. 1998, 143, 37.


[20] E. Stathatos, P. Lianos, C. Tsakiroglou, Microporous Mesoporous Mater. 2004, 75, 255.
[21] F. Bosc, A. Ayral, P.-A. Albouy, L. Datas, C. Guizard, Chem. Mater.
2004, 16, 2208.
[22] A. S. Deshpande, D. G. Shchukin, E. Ustinovich, M. Antonietti,
R. A. Caruso, Adv. Funct. Mater. 2005, 15, 239.
[23] H. Choi, K. Yoo, D. D. Dionysiou, in Proc. Eighth Int. Conf. Inorganic
Membranes (Eds: F. T. Akin, Y. S. Lin), Adams, Chicago, IL 2004,
p. 365.
[24] J. Z. Zhang, Z.-L. Wang, J. Liu, S. Chen, G.-Y. Liu, Self-Assembled
Nanostructures, Kluwer Academic/Plenum Publishers, New York
2003.
[25] J. Y. Ying, C. P. Mehnert, M. S. Wong, Angew. Chem. Int. Ed. 1999,
38, 56.
[26] Y. Chen, J. C. Crittenden, S. Hackney, L. Sutter, D. W. Hand, Environ. Sci. Technol. 2005, 39, 1201.
[27] K. Yoo, H. Choi, D. D. Dionysiou, Chem. Commun. 2004, 2000.
[28] N. K. Raman, M. T. Anderson, C. J. Brinker, Chem. Mater. 1996, 8,
1682.
[29] P. I. Ravikovitch, A. V. Neimark, Langmuir 2002, 18, 1550.
[30] P. Yang, D. Zhao, D. I. Margolese, B. F. Chmelka, G. D. Stucky,
Chem. Mater. 1999, 11, 2813.
[31] . Dag, I. Soten, . elik, S. Polarz, N. Coombs, G. A. Ozin, Adv.
Funct. Mater. 2003, 13, 30.
[32] S. Nakade, M. Matsuda, S. Kambe, Y. Saito, T. Kitamura, T. Sakata,
Y. Wada, H. Mori, S. Yanagida, J. Phys. Chem. B 2002, 106, 10 004.
[33] Z. Zhang, C.-C. Wang, R. Zakaria, J. Y. Ying, J. Phys. Chem. B 1998,
102, 10 871.
[34] F. Bosc, A. Ayral, P.-A. Albouy, C. Guizard, Chem. Mater. 2003, 15,
2463.
[35] A. Henglein, Ber. Bunsen-Ges. Phys. Chem. 1982, 86, 241.
[36] D. D. Dionysiou, M. T. Suidan, I. Baudin, J.-M. La, T. L. Huang,
Water Sci. Technol. Water Supply 2001, 1, 139.
[37] E. J. Wolfrum, J. Huang, D. M. Blake, P.-C. Maness, Z. Huang, J. Fiest,
W. A. Jacoby, Environ. Sci. Technol. 2002, 36, 3412.
[38] D. G. Shchukin, R. A. Caruso, Adv. Funct. Mater. 2003, 13, 789.
[39] C. Wang, Z. X. Deng, Y. Li, Inorg. Chem. 2001, 40, 5210.
[40] J. Kim, Y. S. Lin, J. Am. Ceram. Soc. 1999, 82, 2641.

______________________

1074 www.afm-journal.de

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Funct. Mater. 2006, 16, 10671074

You might also like