A novel approach to formulate high flux multifunctional ultrafiltration membranes from photocatalytic titania composite precursors on multi-channel tubular substrates

Abstract

Anatase rich titanium dioxide ultrafiltration membranes with high filtration rates have been successfully developed on multi-channel tubular alumina substrates via aqueous sol–gel method from titania–alumina composite precursors containing 30 wt% alumina. The composite membrane material exhibited anatase phase stability above 800 °C and retained a BET surface area of 64 m² g⁻¹ even after calcination at 700 °C. Supported membranes on multi-channel substrates with an active layer thickness of 4 μm gave a water flux value of 215 L m⁻² h⁻¹ coupled with 80% rejection of Bovine Serum Albumin (BSA) with molecular weight 66 kD at 2 bar pressure. This is much higher compared to a flux of 27 L m⁻² h⁻¹ obtained for a single component titania membrane layer. The composite membrane materials showed excellent photocatalytic activity under UV irradiation such that a solution containing Methylene Blue (MB) dye showed 96% dye degradation within 2 h. Porous disc shaped substrates coated with the active titania composite layer showed methylene blue degradation of 44% under identical conditions. The present results point towards an excellent pathway for the development of multifunctional ultra-filtration membranes for water purification and also for other separation applications where separation together with photocatalysis will be of great importance.

---

1. Introduction

Filtration systems are important for a wide range of applications,¹ in which water filtration is the most important, due to concerns over the continuing availability of adequate clean water from natural sources.² Membrane filtration of water has been discussed in technical literature for the last 30–50 years, yet in the absence of cost effective solutions, the topic remains as important as ever. New membrane materials as well as systems or any improvements in existing ones will have far reaching implications on the quality of life, specifically in the developing world. Although polymeric membrane based filtration systems are widely used, certain limitations such as frequent replacement of filtration modules, lack of durability with respect to liquids of extreme acidity/alkalinity, temperature stability and problems related to the safe disposal of the polymeric membrane elements after use are still to be resolved.³ Some of these issues are very critical, when dealing with industrial water recycling and ceramic membranes are considered as potential replacements for many of such systems.⁴ The advantages of ceramic membranes are unparalleled and include good thermal and chemical stability, high mechanical resistance, long life and good defouling properties.⁵ Ceramic membrane elements primarily consist of a macroporous ceramic substrate over which layers of particles with controlled particle sizes and morphology are deposited for a gradual reduction in pore size from an initial size of 1–10 μm down to the final size of <1 nm depending on the type of filtration application, for e.g. final membrane size required is ∼0.3 nm for gas separation.^6,7 Accordingly, the water/liquid/gas permeation, which is a measure of the filtered output, would normally decrease. Pore size and distribution are the two most important parameters that directly influence the efficiency of membranes. Titanium dioxide is one of the most widely used material for the fabrication of the top membrane layers,^8–10 in view of the easiness in the preparation of nano sized pores as well as its chemical and thermal stability. Alumina, zirconia and silica and their composites are also widely used for ceramic membrane fabrication. Other developments in the area of ceramic membranes have considered the application of advanced processing techniques for fabricating less tortuous pores¹¹ as well as providing functionality in terms of hybrid membrane layer structure¹² of the membrane or properties of the materials.¹³ Incorporation of membrane materials with photocatalytic activity is one of the very important approaches discussed in the literature.

Organic contaminants and oxidisable matter present in the feed water could be decomposed in an eco-friendly manner during the filtration process in order to enhance the purity of permeate water. Out of the many approaches reported, use of photoactive materials in the top layer has been found most effective.¹⁴ Photo-catalysis has also potential to solve problems related to the fouling of membranes. Titanium oxide in nano size, doped suitably to match wide spectrum activity has been reported as a candidate material for top layer in ceramic ultra-filtration membrane fabrication.¹⁵

By suitably doping titanium oxide, a well-known semiconductor photocatalyst, it is possible to shift the anatase phase stability to higher temperature as well as increase the surface pore characteristics in addition to the improved photoactivity. Wide ranges of pore sizes can be achieved by controlling the sol gel process of preparation and such membranes reportedly have higher water flux, in addition to photocatalytic activity. Further, titania is non-toxic and therefore can be used safely in filtration applications.¹⁶

Kimberly et al. reported fabrication of defect free asymmetric alumina ultrafiltration membranes having pore sizes in the range of 20–30 nm with a thickness of ∼2 μm and molecular weight cut-off (MWCO) in the range of 35 000–44 000 g mol⁻¹. The membrane showed high porosity and good water permeability.¹⁷ Bae et al. succeeded in preparing 3 μm thick Al₂O₃–TiO₂ top layers with crack-free microstructure even after calcination up to 900 °C.¹⁸ They have used a polymeric sol composition for the formation of the top layer on the supports. Kumar et al., Lin et al. and Zaspalis et al. reported certain properties of titania as well as alumina-titania membranes having γ-alumina and anatase titania layers.^19–21 Bjorkerk et al. also studied diphasic membranes containing γ-alumina and anatase phase in La₂O₃ doped Al₂O₃–TiO₂ ceramic films.²² La₂O₃ doping arrested the reorganization of the γ-alumina phase, thus preventing the formation of other intermediate alumina phases. Van Gestel et al. studied good thermal, mechanical and chemical stability of alumina–titania membranes and reported the properties of membranes having mesoporous Al₂O₃/ZrO₂ interlayer and micro porous ZrO₂ top layer developed by sol–gel preparation. The optimized membranes showed a molecular weight cut-off (MWCO) of 200–300.²³ Ten Elshof et al. synthesized organic–inorganic hybrid membranes for dewatering applications.²⁴ Several researchers have also attempted to combine photocatalysis with membrane separation techniques using TiO₂ powder photocatalysts for removal of organic pollutants from water. Suspended TiO₂ powder photocatalysts were separated by ultrafiltration or microfiltration after photocatalytic reaction. Falaras et al. studied application of membranes for water purification under visible light.²⁵ Romanos et al. studied advantages of photocatalytically active ultra-filtration technology involving CVD derived TiO₂ membranes and reported that 82% efficiency in the removal of methyl orange from the permeate effluent.²⁶ Tsuru et al. synthesized porous TiO₂ membranes with pore sizes in the range of 2.5–22 nm by a sol–gel method for the decomposition of MeOH and EtOH in a photocatalytic membrane reactor.²⁷ Bottino et al. investigated the influence of Ti/Al ratio, amount of an organic additive, dip-time, the number of depositions and drying and calcination conditions on the structure of resulting titania–alumina membrane and transport properties.²⁸ The issues in the development of composite coatings are the control of homogeneity of dispersion of the composite particles in the coating precursor, control of viscosity of the coating solution, as well as surface roughness of the substrate surface. A combination of all the above factors is important in realising membranes having excellent separation performance along with superior functionality in other selected aspects. However, the deposition of defect-less intermediate and functional membrane layers having controlled pore structural characteristics are the most important points with respect to the development of such high performance membranes. We expect that the proper tailoring of sol–gel processing conditions are adequate to accomplish membrane materials with controlled pore size as well as crystal structure capable of forming such membrane layers and therefore attempted the work reported in this study.

Here, we introduce membranes having two layers made of alumina–boehmite and titania–boehmite composite precursors to tailor a membrane with structure and pore-size distribution that allows higher filtration rates as well as enhanced photoactivity. Compared to membranes having titania single component layers, water flux measured were much higher for the newly developed membranes. Synthesis of the coating precursors and the coating process were optimized to result in graded pores from the substrate to the top layer. The substrate used was multi-channel porous sintered alumina tubes. The composite membrane precursor has been prepared by an aqueous sol–gel process, without employing any organic solvents. Influence of the composition of the composite sol on the properties of the final membrane is presented. In addition, the effect of alumina phase on the crystal structure stability of titania and hence the structural formation of the resulting composite membranes leading to their superior separation performance is elaborated.

2. Experimental

2.1. Materials

Multi-channel porous α – alumina tubes of length 200 mm, outer diameter 25 mm and having nineteen channels each of diameter 3 mm were used as substrates. The tubes were procured from Bharat Heavy Electricals Limited (BHEL, Bangalore, India). Porous alumina discs of diameter 25 mm and 3 mm thickness made in-house by powder compaction and sintering were used for trial coating experiments as well for the fabrication of supported membranes for photocatalytic activity measurement. The porosity of the disc shaped substrate was measured as 50% by Archimedes' method according to ASTM 378-88 standard. Photograph and SEM image of the porous tubular and disc shaped substrates are shown in Fig. 1 and 2.

Fig. 1 Photograph and SEM image of the 19 channel porous tubular substrate.

Fig. 2 Photograph and SEM image of the porous alumina disc substrate.

Boehmite powder (of average particle size 150 nm) was purchased from PURAL SBI Condea Chemicals, Germany and alumina powder (of average particle size 0.35 μm) from ACC-ALCOA Chemicals, Kolkata, India. Ammonium hydroxide, aluminium nitrate and polyvinyl alcohol (of molecular weight 1, 25 000) were procured from SD Fine Chemicals (Mumbai, India). Nitric acid was obtained from Merck India. Titanium(IV) oxysulphate (of purity 99.99%) was obtained from Sigma Aldrich Chemicals (Steinheim, Germany). Hydroxyethylcellulose (HEC) was obtained from Biochemika, Germany.

2.2. Preparation of precursors

Alumina–boehmite slurry (coating precursor for intermediate layer). Stable suspension of boehmite was first prepared by dispersing 10 g of boehmite powder in 500 mL of distilled water and by adjusting the pH value to ∼3.5. 40 g of alumina (Al₂O₃) powder was then slowly added to the boehmite (AlOOH) suspension kept under stirring. Poly vinyl alcohol (PVA) was used as binder for preparing the coating solution, the solid content of which was ∼10 wt%. Detailed procedure for the preparation of the slurry and coating solution are available in our earlier publications.¹⁴

Boehmite sol (constituent for coating precursor for functional layer). 125 g of Al(NO₃)₃·9H₂O was first dissolved in 1 L of distilled water and was maintained at a temperature of 90 °C. 100 mL of ammonium hydroxide (25% solution) was slowly added to this mixture. The obtained precipitate was filtered and washed repeatedly with hot distilled water (60 °C). The nitrate free aluminium hydroxide precipitate thus obtained was dispersed in 1000 mL of distilled water which was peptized by adding 20% nitric acid under stirring condition until the pH was below 5. Detailed procedure for the preparation is reported in our earlier publications.^14,29

Titania sol (constituent for coating precursor for functional membrane layer). 157 mL titanium(IV) oxysulphate was hydrolysed using ammonium hydroxide (25% solution) in 1 L of distilled water. The resulting precipitate was filtered and washed with hot distilled water. The sulphate free titanium hydroxide precipitate was dispersed in 1000 mL of distilled water, followed by peptization by adding 20% nitric acid until the pH was ∼2. Detailed procedure for the preparation of titania (TiO₂) sol can be found in our earlier publications.^30–32 Coating suspension was prepared by mixing titania sol with 1 wt% of the organic binder hydroxyethyl cellulose (HEC); the sample thus made is denoted as TH.

Titania–alumina composite sol (coating precursor for functional membrane layer). Titania–alumina composite precursor (TAl) was prepared from titania and boehmite sol. For the preparation of TAl composite precursor sol, varying wt% of boehmite sol was added to the titania sol and stirred for ∼3 h at room temperature. The sol thus produced was denoted as TAl_X, where “X” corresponds to the weight percentage of boehmite in the sol, which was varied between 0 and 50 wt% in this study. As in the case of pure titania, the composite sol was modified with 1 wt% of HEC to make the dip solution for coating; samples are denoted as THAl_X. Supported and unsupported membrane with pure titania top layer was used to compare against the performance of titania–alumina composite membranes.

2.3. Coating procedure

Coating was performed using an automatic coating system fabricated by Plastomek, Kerala, India. Alumina–boehmite slurry was coated on the substrate to form the intermediate layer and titania–boehmite composite sol was coated over it to form the functional membrane layer.¹⁴ Coating procedure and a schematic structure of the resulting membrane are shown in Fig. 3.

Fig. 3 Procedure for the coating process and schematic representation of the membrane showing different layers on the porous substrate.

As shown in Fig. 3, polyvinyl alcohol (PVA, 1 wt%) and hydroxyethyl cellulose (HEC, 1 wt%) were added to the precursor sols for controlling the viscosity and also to act a binders during membrane layer formation. Coated substrates were dried at 60 °C/59% RH using a humidity controlled oven. The resulting membrane had a 3-layered structure as shown in Fig. 3 consisting of an alumina substrate, alumina intermediate layer made from slurry containing alumina & boehmite and finally a titania/alumina functional layer made from titania and boehmite sols. For the evaluation of photocatalytic activity of supported membranes, sintered, porous alumina discs coated at identical conditions were employed.

The dip solution was also dried in petri dishes at similar conditions to prepare titania and titania–alumina unsupported membranes (TH and THAl) for pore size analysis. The supported and unsupported membranes were calcined at 600, 700 or 800 °C; the heating rate was 1° min⁻¹ and soaking time was 3 h in all cases.

2.4. Characterization

The porosity and pore size distribution of the tubular support were measured by Mercury porosimetry analysis (Micrometrics, Auto Pore IV 9500) in the pressure range 0.01–200 MPa.

Particle size distribution of sol was measured using photon correlation spectroscopy (Zetasizer 3000 HSA, Malvern Instruments Ltd., UK). X-ray diffraction was measured using Cu K_α radiation (1.542 Å) using a Philips X'pert X-ray Diffractometer. The amount of anatase, rutile and brookite phases in the titania sample was estimated using the method of Zhang and Gribb's, eqn (A.1) and (A.2).

where, W_B and W_A are the mass fraction of brookite and anatase respectively, the coefficients K_A and K_B have values 0.886 and 2.721 respectively. I_B and I_A are the integrated intensity of the brookite (121) and anatase (101) peaks respectively.

Porosity, pore size and surface area of the unsupported titania–alumina membrane material was measured using N₂ adsorption method (Micrometrics, Gemini 2375). The samples were degassed under vacuum at 200 °C for 2 h before the measurement.

The photocatalytic activity of the samples was studied by monitoring the degradation of Methylene Blue (MB) dye in an aqueous suspension under UV-A exposure and continuous stirring using a magnetic stirrer. The intensity of the UV light was 0.4 mW cm⁻². 0.03 g of the TiO₂, TH and THAl samples were dispersed in 75 mL of 3.0 × 10⁻⁵ M molar aqueous MB (AR Grade, Qualigens Fine Chemicals, India) solution. The suspension was stirred in the dark for half an hour before irradiating with UV light to compensate for the effect due to the adsorption of the MB dye. The concentration of the dye was measured at different time intervals using UV-Visible spectrometer (Shimadzu, Japan, UV-2401PC). A blank dye solution was also irradiated, for about 2 h to confirm that the dye was not photobleached by UV exposure. The dye concentration remained unchanged even after irradiation for 2 h. The maximum intensity absorbance peak at 663.2 nm of MB solution was taken for measuring the degradation. The absorbance of MB solution after keeping in the dark for half an hour under stirring was taken as initial absorbance; and the absorbance after UV-A exposure in the presence of nano-composites was measured in time intervals of 20 min. The photocatalytic activity of the coated disc shaped substrate was also studied using MB degradation under identical conditions. The degradation of MB was calculated using the relation in eqn (2)

C/C₀ = A_time=t/A_time=0 (2)

where, A_time=0 is the initial absorbance of the MB solution, i.e., A₀ and A_time=t are the absorbance values after time intervals.

The diffused reflectance spectra of the calcined powders were obtained using a (UV-2401PC, Shimadzu, Japan) in the range of 200–800 nm.

HR-TEM data of calcined powder sample was taken on carbon coated copper grids (FEI Tecnai 30 G²S-Twin HR-TEM equipped with a Gatan CCD camera and operated at 300 kV). The membrane thickness, homogeneity of the coating and surface quality were examined using a scanning electron microscope (CRYO-VP, ZEISS EVO′18, Japan) operated at 15 kV. A cross flow filtration unit was used for the filtration studies. The flux of the membranes at specific pressures were measured from the weight of permeate, collected at different time intervals. Membrane permeability was estimated from the measured flux, pressure and time values.

Molecular weight cut-off of the membrane was determined using Bovine Serum Albumin (BSA) (66 000 D) (99.99%, Sigma Aldrich Chemicals, Steinheim, Germany) solution (1.98 g L⁻¹). Flux and molecular weight cut off (MWCO) are the primary tools for measuring the efficiency of the membranes, where MWCO was evaluated in terms of rejection (%). Flux is the amount of water/liquid medium permeated or transported through the unit area of the membrane at unit time intervals (L m⁻² h⁻¹) and rejection is the percentage of solid concentration which is removed from the feed water by the membrane through the filtration process.

The filtration was carried out at different pressures ranging from 1 to 6 bar. 10 mL of feed solution before and after filtration (permeate) was taken out every half hour to measure the absorbance. % rejection of BSA was calculated using the formula given in eqn (3).

R (%) = (1 − A_t/A₀) × 100 (3)

where, R is the % rejection, A₀ is the absorbance of the feed solution and A_t is the absorbance of the permeate.

3. Results and discussion

The pore size distribution of the porous alumina tubular substrates measured using mercury porosimetry is provided in Fig. 4. The plot obtained from mercury porosimetry analysis, shows the pore size distribution with respect to differential cumulative pore volume and cumulative pore area. The average pore size distribution was in the range of 0.8–2 μm. The two peaks observed in the pore size distribution curve should be due to the presence of non-cylindrical shaped pores.³³

Fig. 4 Mercury porosimetry analysis of porous alumina tubular substrates.

The average pore diameter obtained was about 1 μm and the porosity was about 46.9%. The average pore size and porosity mainly depend on the sintering temperature.³⁴ The pore size distribution observed was very narrow and absence of large size pores indicates that the porous substrate would be useful as supports for membrane fabrication. Particle size distribution is an important factor for governing the quality of the precursor sol, used for the coating purpose. The average particle size distribution of pure titania, alumina and composite sols are illustrated in Fig. 5. It is shown that the average particle size of titania sol increased with the alumina addition.

Fig. 5 Particle size measurement of titania–alumina composite sol (a) TAl₁₀, (b) TAl₂₀, (c) TAl₃₀ and (d) TAl₅₀. Particle size distribution of pure titania (TiO₂) and alumina sol (Al₂O₃) are shown in inset.

Pure titania and alumina sols exhibited average particle sizes in the range of 10–20 and 100–200 nm respectively. The average particle size in the composite sol was found to be closer to that of alumina sol, with the average particle size increasing from 160 to 280 nm along with increase in alumina concentration from 10 to 50 wt%. The OH groups present in the boehmite sol are expected to support Ti–O–Al linkages in the composite sol thereby forming particle size distribution of larger particles.^35,36 In practice, the particle size distribution of the composite sol will be determined by the cluster formation of the majority titania particles around the alumina particles when only small concentration of alumina is present in the composite sol. However, as the amount of alumina increases the cluster between alumina particles having smaller titania particles in between them could also take place. From the shape of particle size distribution curves, we expect that a change in cluster forming mechanism probably occurred when alumina concentration changed from 20% to 30% (curve b to c in Fig. 5). It should be noted that the particle size as measured with dynamic light scattering represents such secondary particles (aggregates) formed in the sol. The primary particles of titania or alumina are not expected to change during mixing of the sols. As will be discussed later, the presence of alumina in fact helps to reduce the size of titania primary particles while heat-treating.

X-ray diffraction patterns of unsupported titania membranes having varying wt% of alumina content (0–50 wt%) and that of titania membrane with 30 wt% alumina calcined at different temperatures are shown in Fig. 6A and B respectively. Fig. 6A indicates that the alumina content in the composite sol enhances the stability of the anatase phase of titania. Brookite phase was dominant in pure titania sample as well samples having low amount of alumina (up to 20 wt%). For titania samples with 30 and 50% of alumina anatase phase has been observed for samples heat treated at 700 °C as in Fig. 6A. The bonding of higher amount of OH groups to Ti⁴⁺ centers favoring the formation of edge shared bonding is expected to lead to the stabilisation of anatase phase.³⁷ To further understand the influence of alumina addition on the structure of the membrane, crystallite sizes were calculated from the XRD plots using Scherer equation. The unsupported titania membranes at 700 °C having alumina content of 30 wt% (THAl₃₀) showed crystallite size of about 8.6 nm. The corresponding values for compositions TH, THAl₁₀, THAl₂₀ and THAl₅₀ were 16.8, 10.4, 9.0 and 11.8 nm respectively. Thus the crystallite size of titania decreased with the increase in the addition of alumina. The alumina remained amorphous in all the samples at the calcination temperature of 700 °C. The dispersion of amorphous alumina among the anatase particles should have helped to stabilize the titania crystals by grain boundary pinning, thereby hindering the grain growth.³⁸ The increased apparent activation energy for rutile nucleation at the titania–alumina interfaces³⁹ should also have hindered the anatase to rutile conversion of the titania phase.

Fig. 6 X-ray diffraction analysis of unsupported titania membrane (A) with varying wt% of alumina calcined at 700 °C (a) TH, (b) THAl₁₀, (c) THAl₂₀, (d) THAl₃₀ and (e) THAl₅₀ (B) THAl₃₀ calcined at different temperatures, (a) 600, (b) 700 and (c) 800 °C (♦: anatase & ●: brookite).

The XRD analysis clearly indicated the stability of anatase phase even at higher temperatures in the composite materials. 100% anatase phase was observed even after heat-treatment at 800 °C for the sample containing 30 wt% alumina (Fig. 6B). However, in the sample heat-treated at 800 °C, the gamma alumina phase is also clear in the XRD plots. The presence of Al₂O₃ in the titania matrix contributed towards surface area enhancement in addition to the stabilization of anatase phase by delaying the phase transformation to rutile. The anatase crystallite size increased sharply from 7 nm to 25 nm with increase in calcination temperature from 600 to 800 °C. It should be noted that the crystallite size of the sample heat treated at 700 °C was only 8.6 nm. The drastic increase in crystallite size of anatase has hence occurred between 700 °C and 800 °C, where also the crystallization of boehmite to gamma alumina phase occurred. The crystallization of alumina should have reduced the grain pinning of titania due to the loss in surface area of the alumina phase, leading to growth of titania grains as well.

N₂ adsorption/desorption isotherms of unsupported titania and titania–alumina composite membranes calcined at 700 °C are presented in Fig. 7. The shape of the isotherms of all the samples showed type-II adsorption behaviour. In addition, the hysteresis loops observed were H₃ type according to the IUPAC classifications.^40,41 The H₃ hysteresis loop is related to non-rigid slit shaped pores.⁴² As the hysteresis loop increased with the increase in alumina content, we expect that the increase in the amount of boehmite should be responsible for the typical pore morphology obtained in the composite samples. In other words, addition of alumina in titania not only enhanced the phase stability of titania but also helped to acquire and retain unique pore structure even after heat-treatment at 700 °C. Surface area, pore volume and pore diameter of alumina added titania samples are tabulated in Table 1.

Fig. 7 N₂-adsorption desorption isotherm of titania–alumina unsupported composite membranes with varying wt% of alumina heat treated at 700 °C. (a) Pure titania (TiO₂), (b) TH, (c) THAl₁₀, (d) THAl₂₀, (e) THAl₃₀ and (f) THAl₅₀.

Table 1 BET specific surface area, total pore volume and average pore size distribution of the pure and composite unsupported membranes heat treated at 700 °C

Samples	BET surface area (m^2 g^−1)	Total pore volume (cm^3 g^−1)	BJH pore diameter (nm)
TiO2	08.4	0.0012	5.3
TH	10.6	0.0017	5.7
THAl10	29.6	0.0015	5.9
THAl20	44.9	0.0017	6.3
THAl30	64.4	0.0023	6.8
THAl50	96.4	0.0025	7.3

---

BET specific surface area has increased with increase in alumina in titania matrix. However, part of the contribution towards this large increase in surface area should be from amorphous alumina present. The increase in pore size with increase in alumina content was only marginal but apparent.⁴³

Photocatalytic activity measurements of TiO₂ and unsupported titania membrane with and without alumina are illustrated in Fig. 8. The degradation measured with pure titania was only 37% while a slightly higher value of 44% could be measured with the titania unsupported membrane material without alumina (TH) sample. The addition of alumina however enhanced the photocatalytic property of titania unsupported membranes. All the titania–alumina compositions exhibited enhanced MB dye degradation. THAl₁₀, THAl₂₀,THAl₃₀ and THAl₅₀ samples showed excellent MB degradation efficiency values of 92%, 92%, 96% and 94% respectively under same experimental conditions (2 h of UV-A exposure). Although the difference in efficiency values between the composite materials were marginal we have opted to study THAl₃₀ sample in detail because it gave the highest value of 96% in the photo catalytic activity measurement.

Fig. 8 Photocatalytic activity measurement of TiO₂, unsupported titania membrane with varying wt% of alumina content (0 to 50 wt%), calcined at 700 °C.

Diffused reflectance spectra of unsupported titania and titania–alumina composite membrane materials (TH and THAl₃₀ calcined at 700 °C) are shown in Fig. 9 and corresponding absorbance data are shown in inset. Both the samples showed the absorbance edge in the UV region and differences are due to the brookite to anatase phase stabilization.⁴⁵ The band gap energy of brookite phase is higher than that of anatase phase.⁴⁶

Fig. 9 Diffused reflectance spectra of (a) TH and (b) THAl₃₀ calcined at 700 °C (corresponding absorbances are shown as inset).

Unsupported titania membrane exhibited a band gap of 3.38 eV, which is similar to the band gap energy of brookite phase, while in the case of composite membrane the measured band gap was only 3.2 eV, corresponding to the anatase titania. The XRD analysis results are consistent with these data.

TEM images of the unsupported titania–alumina composite membranes (THAl₃₀) are shown in Fig. 10. The particles in THAl₃₀ samples showed more or less uniformity in size and shape with an average particle size of ∼10 nm. The HRTEM images show that titania–alumina composite membrane exhibited 100% anatase phase. The lattice space corresponds to pure anatase phase, d₁₀₁ = 0.35 nm. SEM micrographs of multi-layer coatings on the substrate are shown in Fig. 11. The layered coatings consisting of alumina–boehmite (intermediate layer, I) and titania –alumina (THAl₃₀) top functional layer could be clearly identified from Fig. 11.

Fig. 10 TEM (A) and HR-TEM (B) image of unsupported titania–alumina composite membranes (THAl₃₀).

Fig. 11 SEM analysis of tubular membranes calcined at 700 °C. (A) SEM showing the cross section, (B) surface image of the functional layer and (C) cross section of the coated disc shaped substrates used for photo catalytic activity study. S – Substrate, I – intermediate layer and THAl₃₀ titania–alumina top functional layer.

The intermediate layer of alumina–boehmite showed an average thickness of ∼20 μm. The functional layer of titania–alumina showed an average thickness of ∼4 μm. The membrane surface had no cracks and the pore structure looked uniform all through the surface. Some slightly larger pores were visible in the surface micrographs although a quantitative estimation on the different types of pores in the surface is difficult from the present analysis. Disc shaped substrates showed thinner intermediate alumina composite layer but titania–alumina top membrane layer had nearly identical thickness as that of tubular substrates (Fig. 11C).

Photocatalytic efficiencies of supported membranes were measured using membrane coated porous alumina discs (Fig. 11C) under measurement conditions identical to that of unsupported membranes.⁴⁴ Discs coated with only alumina intermediate layer had no measurable photocatalytic activity. However, the presence of titania containing membrane layers improved the photocatalytic activity of the coated substrates. MB dye degradation values measured for supported membranes in presence of UV light are shown in Table 2. It should be noted that the weight of membrane material exposed to UV is considerably low here compared to the results shown in Fig. 8 and should be the reason for the differences in degradation values between the supported and unsupported membrane. The membrane area exposed was similar in all the three cases in Table 2 and the result clearly stands testimony to the superior performance of the newly developed THAl₃₀ membranes and membrane materials.

Table 2 MB dye degradation performance of layered coatings made on a porous alumina disc shaped substrate

Membrane layer & heat treatment temperature (°C)	Decomposition after 2 h of UV irradiation (%)	Decomposition after 6 h UV irradiation (%)
Alumina base/intermediate layer (I)-1100	0	0
Functional layer (TH1)-700	6	18
Functional layer (THAl30)-700	28	44

---

Membrane water permeability was determined according to the Hagen–Poisseuille equation from flux measurements. The flux data for membranes prepared from titania (TH) and titania–alumina composite (THAl₃₀) as a function of trans-membrane pressure are shown in Fig. 12.

Fig. 12 Water flux measurement of (a) uncoated substrate, (b) with alumina intermediate layer over substrate, (c) with titania–alumina layer (THAl₃₀) over intermediate layer and (d) with titania layer (TH) over intermediate layer. Both top layers were calcined at 700 °C.

Water flux measurements showed some issues in our set-up and measurements. Cake formation at the membrane surface due to impurities in water was not detected under our experimental conditions (pressure independent linear flux of titania layer (TH) confirms this), but partial pore blockage at high-pressure measurement conditions could not be ruled out in the case of titania–alumina layer (THAl₃₀).

The rejection characteristics of the membrane were investigated using molecular weight cut off measurements.^47,48 When the water containing BSA having a particular size (66 kD) is subjected to filtration through this membrane, size above the average pore size of the membrane layer is rejected and only those particles below the cut off size are passed through. This means that if the rejection is high, the average pore size of the top layer of the membrane is small. Hence for the uncoated substrate rejection will be very low, while for the coated one, the reverse. The extent of rejection of BSA having a particular size is expressed as % rejection. The percentage rejection (MWCO) of BSA obtained for the titania–alumina membrane is provided in Fig. 13.

Fig. 13 Percentage rejection of BSA obtained for (a) Uncoated substrate, (b) with alumina intermediate layer over substrate, (c) with titania–alumina layer (THAl₃₀) over intermediate layer and (d) with titania layer (TH) over intermediate layer. Both top layers were calcined at 700 °C.

Membranes with titania/alumina functional layer showed about 80% rejection at 2 bar pressure, however membranes with pure titania functional layer showed higher rejection value >90%. This could be due to the larger pore size distribution and surface area of THAl₃₀ composite membranes compared to that of TH membranes. The larger flux and lower rejection observed for the composite membranes are in line with the pore structure characteristics measured by N₂ adsorption. Based on the MWCO value of BSA the active pore size of the membrane could be considered as ∼11 nm for practical separation purposes under similar conditions. It should be noted that the pore size value calculated from N₂ adsorption isotherm of the unsupported membrane material of THAl₃₀ was slightly smaller (6.8 nm as in Table 1). In summary, it is shown that the sol gel based tailoring of membrane materials provides a pathway to produce high performance membranes having multifunctional properties together with excellent water permeation.

4. Conclusions

A highly efficient photocatalytically active titania–alumina composite membrane was successfully developed on a multi-channel alumina tubular substrate. The multi-layered coating process could effectively reduce the pore size from 1 μm to about 11 nm. Alumina addition to titania enhanced the phase stability of anatase phase and helped to retain a high surface area value of 64.4 m² g⁻¹ compared to the surface area value of 10.6 m² g⁻¹ measured for the pure titania membrane material heat treated at 700 °C. The composite membrane material was also found to possess better photoactivity compared to pure titania showing as much as 96% MB dye degradation within 2 h. Membrane coatings on porous sintered disc shaped substrates showed methylene blue degradation of ∼44%. The novel method of solution processing and membrane fabrication reported in this work also allowed the fabrication of crack free composite membranes which gave water flux values as high as 215 L m⁻² h⁻¹ compared to 27 L m⁻² h⁻¹ measured with the pure titania membrane. The rejection rate measured with BSA was around 80% and based on the MWCO value the active pore size of the membrane could be considered as ∼11 nm for practical separation purposes under similar conditions. It is expected that the higher water permeation and rejection characteristics of the membrane together with superior photocatalytic performance will offer a pathway for the development of membrane systems useful for water purification and many other applications.

Acknowledgements

The authors Manjumol K. A., Sankar S. and Warrier K. G. K. thanks CSIR, India, for providing financial assistance through their respective fellowships. The authors acknowledge the members of Materials Science and Technology Division for providing general assistance. Mr M. R. Chandran and Mr Kiran Mohan are also acknowledged for SEM and HR-TEM analysis respectively. Mr H. Watanabe of Noritake Co. Limited is gratefully acknowledged for help and discussions related to characterization of composite membranes with Hg-porosimetry measurement.

Notes and references

1. S. A. Yencho, PCT Int. Apple WO 2009042406, 2009.
2. F. H. Hussein, Asian J. Chem., 2012, 24, 5427.
3. A. Basile, S. Chakraborty and A. Iulianelli, Membrane and membrane reactors for sustainable growth, 2013, p. 121.
4. K. R. Colyar, J. Pellegrino and K. Kadam, Sep. Sci. Technol., 2008, 43, 447.
5. P. Hristov, A. Yoleva, S. Djambazov, R. Vladov and A. Gigova, J. Univ. Chem. Technol. Metall., 2012, 47, 471.
6. B. N. Nair, K. Keizer, T. Okubo and S. Nakao, Adv. Mater., 1998, 10, 249.
7. B. N. Nair, T. Okubo and S.-I. Nakao, Membrane, 2000, 25, 73.
8. J. Luyten, T. Van Gestel, J. Cooymans and C. Smolders, Ceram. Trans., 2000, 115, 551.
9. A. Larbot, S. Alami-Younssi, M. Persin, J. Sarrazin and L. Cot, J. Membr. Sci., 1994, 97, 167.
10. R. Vacassy, C. Guizard, V. Thoraval and L. Cot, J. Membr. Sci., 1997, 132, 109.
11. B. N. Nair, T. Suzuki, Y. Yoshino, S. Gopalakrishnan, T. Sugawara, S. I. Nakao and H. Taguchi, Adv. Mater., 2005, 17, 1136.
12. S. Gopalakrishnan, Y. Yoshino, M. Nomura, B. N. Nair and S. I. Nakao, J. Membr. Sci., 2007, 297, 5.
13. X. Li, M. Basko, F. Du Prez and I. F. J. Vankelecom, J. Phys. Chem. B, 2008, 112, 16539.
14. K. A. Manjumol, P. Shajesh, K. V. Baiju and K. G. K. Warrier, J. Membr. Sci., 2011, 375, 134.
15. V. V. Tarabara, Nanotechnology Applications for Clean Water, 2009, p. 59.
16. K. Keizer and A. J. Burggraaf, Sci. Ceram., 1988, 14, 83.
17. K. A. DeFriend, M. R. Wiesner and A. R. Barron, J. Membr. Sci., 2003, 224, 11.
18. D. S. Bae, D. S. Cheong, K. S. Han and S. H. Choi, Ceram. Int., 1998, 24, 25.
19. K. N. P. Kumar, Appl. Catal., A, 1994, 119, 163.
20. C. H. Chang, R. Gopalan and Y. S. Lin, J. Membr. Sci., 1994, 91, 27.
21. V. T. Zaspalis, W. Van Praag, K. Keizer, J. R. H. Ross and A. J. Burggraaf, J. Mater. Sci., 1992, 27, 1023.
22. U. S. Bjorkert, R. Mayappan, D. Holland and M. H. Lewis, J. Eur. Ceram. Soc., 1999, 19, 1847.
23. T. Van Gestel, H. Kruidhof, D. H. A. Blank and H. J. M. Bouwmeester, J. Membr. Sci., 2006, 284, 128.
24. H. L. Castricum, R. Kreiter, H. M. van Veen, D. H. A. Blank, J. F. Vente and J. E. ten Elshof, J. Membr. Sci., 2008, 324, 111.
25. V. Likodimos, C. Han, M. Pelaez, A. G. Kontos, G. Liu, D. Zhu, S. Liao, A. A. de la Cruz, K. O'Shea, P. S. M. Dunlop, J. A. Byrne, D. D. Dionysiou and P. Falaras, Ind. Eng. Chem. Res., 2013, 52, 13957.
26. G. E. Romanos, C. P. Athanasekou, V. Likodimos, P. Aloupogiannis and P. Falaras, Ind. Eng. Chem. Res., 2013, 52, 13938.
27. T. Tsuru, T. Kan-no, T. Yoshioka and M. Asaeda, J. Membr. Sci., 2006, 280, 156.
28. A. Bottino, A. Comite and G. Capannelli, Asia-Pac. J. Chem. Eng., 2009, 5, 242.
29. T. V. Mani, H. K. Varma, K. G. Warrier and A. D. Damodaran, J. Am. Ceram. Soc., 1991, 74, 1807.
30. K. A. Manjumol, V. S. Smitha, P. Shajesh, K. V. Baiju and K. G. K. Warrier, J. Sol-Gel Sci. Technol., 2010, 53, 353.
31. S. Sivakumar, P. K. Pillai, P. Mukundan and K. G. K. Warrier, Mater. Lett., 2002, 57, 330.
32. K. A. Manjumol, M. Jayasankar and K. Vidya, J. Sol-Gel Sci. Technol., 2015, 73, 161.
33. M. J. Moura, P. J. Ferreira and M. M. Figueiredo, Powder Technol., 2005, 160, 61.
34. P. Monash and G. Pugazhenthi, Int. J. Appl. Ceram. Technol., 2011, 8, 227.
35. A. M. Buckley and M. Greenblatt, J. Chem. Educ., 1994, 71, 599.
36. V. S. Smitha, P. Francois, U. N. Saraswathy Hareesh and K. G. Warrier, J. Mater. Chem. A, 2013, 1, 12178.
37. Q. Tay, X. Liu, Y. Tang, Z. Jiang, T. C. Sum and Z. Chen, J. Phys. Chem. C, 2013, 117, 14973.
38. S. Matsumura, T. Horiuchi, K. Yasuda, K. Kaneko, M. Watanabe and T. Masumoto, Scr. Mater., 2005, 53, 441.
39. R. Rodriguez-Talavera, S. Vargas, R. Arroyo-Murillo, R. Montiel-Campos and E. Haro-Poniatowski, J. Mater. Res., 1997, 12, 439.
40. S. Ruthstein, V. Frydman, S. Kababya, M. Landau and D. Goldfarb, J. Phys. Chem. B, 2003, 107, 1739.
41. C. Goltner-Spickermann, J. Colloid Interface Sci., 2002, 7, 173.
42. L. Vradman, L. Titelman and M. Herskowitz, Microporous Mesoporous Mater., 2006, 93, 313.
43. X. Chen and S. S. Mao, Chem. Rev., 2007, 107, 2891.
44. A. Alem, H. Sarpoolaky and M. Keshmiri, Ceram. Int., 2009, 35, 1837.
45. M. Zhou, J. Yu, S. Liu, P. Zhai and L. Jiang, J. Hazard. Mater., 2008, 154, 1141.
46. J. Y. Park, C. Lee, K. W. Jung and D. Jung, Bull. Korean Chem. Soc., 2009, 30, 402.
47. R. L. Callender and A. R. Barron, J. Mater. Res., 2000, 15, 2228.
48. R. L. Callender and A. R. Barron, J. Mater. Sci., 2001, 36, 4977.

---