A.
Sotto
a,
J.
Kim
b,
J. M.
Arsuaga
a,
G.
del Rosario
c,
A.
Martínez
d,
D.
Nam
b,
P.
Luis
e and
B.
Van der Bruggen
*f
aDepartment of Chemical and Energy Technology, Rey Juan Carlos University, 28933-Móstoles, Madrid, Spain
bDepartment of Environmental Engineering, Inha University, 100, Inha-ro, Incheon, Republic of Korea
cTechnological Support Center, Rey Juan Carlos University, 28933-Móstoles, Madrid, Spain
dDepartment of Basic Sciences Applied to Engineering, Technical University of Madrid, Madrid, Spain
eMaterials and Process Engineering (iMMC-IMAP), Université Catholique de Louvain, Place Sainte Barbe 2, 1348 Louvain-la-Neuve, Belgium
fDepartment of Chemical Engineering, KU Leuven, W. de Croylaan 46, B-3001 Leuven, Belgium. E-mail: Bart.VanderBruggen@cit.kuleuven.be
First published on 21st February 2014
A new ultrafiltration membrane was developed by the incorporation of binary metal oxides of the type Ti(x)Zr(1−x)O2 inside polyethersulfone. Physico-chemical characterization of the binary metal oxides demonstrated that the presence of Ti in the TiO2–ZrO2 system results in an increase of the size of the oxides, and also their dispersity. The crystalline phases of the synthesized binary metal oxides were identified as srilankite and zirconium titanium oxide. The effect of the addition of ZrO2 can be expressed in terms of the inhibition of crystal growth of nanocrystalline TiO2 during the synthesis process. For photocatalytic applications the band gap of the synthesized semiconductors was determined, confirming a gradual increase (blue shift) in the band gap as the amount of Zr loading increases. Distinct distributions of binary metal oxides were found along the permeation axis for the synthesized membranes. Particles with Ti are more uniformly dispersed throughout the membrane cross-section. The physico-chemical characterization of membranes showed a strong correlation between some key membrane properties and the spatial particle distribution in the membrane structure. The proximity of metal oxide fillers to the membrane surface determines the hydrophilicity and porosity of modified membranes. Membranes incorporating binary metal oxides were found to be promising candidates for wastewater treatment by ultrafiltration, considering the observed improvement in flux and anti-fouling properties of doped membranes. Multi-run fouling tests of doped membranes confirmed the stability of permeation through membranes embedded with binary TiO2–ZrO2 particles.
Membrane fouling is considered one of the main drawbacks in the performance of ultrafiltration membranes, often related to hydrophobicity. The hydrophobic surface of ultrafiltration membranes depends on the hydrophobic nature of the polymeric membrane surface. The fouling mechanism can be described in terms of adsorption, deposition or accumulation of foulants of organic chemical nature on the membrane surface.4,5 Adsorbed molecules increase the hydrophobicity of the membrane or may decrease the pore diameter, leading to a decline of water permeability.6 One of the methods proposed in the recent literature to avoid membrane fouling is the use of hydrophilic particles as additives to the membrane compositions. Due to the hydrophilic nature of additives immersed in the membrane structure, the undesired hydrophobic interactions that take place between organic foulants and the membrane surface can be reduced considerably. Enhanced membranes with better anti-fouling properties lead to an improved performance in terms of permeability, solute rejection, or membrane lifetime. Many types of inorganic fillers have been proposed as polymeric membrane nanofillers. In the first place, TiO27–11 is undoubtedly the preferred material for this goal, but also silica (SiO2),12,13 carbon nanotubes,14 zeolites,15,16 alumina (Al2O3),17 silver (Ag),18 zirconia (ZrO2),19 zinc oxide (ZnO)20 and zerovalent iron (FeO)21 are of potential interest in this field.
The main focus is on the use of transition metal oxides to manufacture ultrafiltration membranes with a more hydrophilic character. As the membrane hydrophilicity increases, the adverse hydrophobic interactions that lead to the formation of a fouling layer are mitigated.22 Transition metal oxides have a wide variety of surface structures, which determine their elevated surface energy and hence their outspoken hydrophilic nature.23,24 Other significant approaches to improve the hydrophilicity of ultrafiltration membranes have been proposed, including the self-segregation of hydrophilic polymers and binding of nanoparticles to the membrane surface.25,26 In general, the incorporation of hydrophilic materials inside the membrane structure promotes the formation of membranes with a more open structure, with a higher average pore size and porosity, and with enhanced hydrophilicity. Both effects facilitate water transport through the membrane structure, thus increasing the membrane's water permeability.27 However, the use of metal oxides as nanofillers in membranes has some drawbacks. For example, high concentrations of metal oxide particles may lead to negative effects instead of an improvement, especially with respect to water permeability.28
Furthermore, a disproportionate increase of the membrane pore size provokes a non-equitable tradeoff between permeability and selectivity affecting negatively the general performance of the membrane.29 Another important concern is the environmental risk associated with the use of nanomaterials and the integrity of membrane functions.7
The influence of some properties, such as the particle size and especially its spatial particle distribution on the membrane performance, has not been studied yet. In this context, it is important to cite the studies reported by Maximous et al. These authors showed the influence of the distribution pattern of Al2O3 and ZrO2 particles on the performance of polyethersulfone modified membranes.30,31 Recently, Sotto et al. suggested that it is possible to reach a noteworthy enhancement in the permeability and membrane anti-fouling properties for an unusually low range of TiO2 concentrations (below 0.4 wt%).32 Following this approach, Arsuaga et al. used low concentrations (0.4 wt%) of TiO2, ZrO2 and Al2O3 as hydrophilic particles, and observed that the particle density of single metal oxides in the vicinity of the membrane surface affects positively the pore formation and prevents the decrease of permeate flux as a result of the deposition/adsorption of BSA molecules on the membrane's upper layer.33 The objective of this work is to prepare and characterize a new class of composite membranes applied to wastewater treatment, for which binary oxide metals have been synthesized in different stoichiometric combinations of TiO2 and ZrO2 in order to produce membranes with different permeation characteristics. In this work the influence of physico-chemical properties of the particles and their spatial distribution inside the membrane structure on the morphology and performance of new binary metal oxide doped membranes has been explored.
The crystal structure of the samples was studied by XRD measurements by using a Philips X-PERT MPD diffractometer (Cu Kα radiation) with a step size and counting time of 0.02° and 10 s, respectively. Appropriate sample preparation to obtain well-dispersed, isolated particles was accomplished. Metal oxide particles dispersed in ethanol were sonicated by ultrasound irradiation for 30 min. Structural characterization was carried out by transmission electron microscopy (TEM) on a PHILIPS TECNAI-20 electron microscope operating at 200 kV.
The TiO2 and ZrO2 molar concentrations of the samples were determined by inductively coupled plasma spectroscopy (ICP) with a VARIAN VISTA AX apparatus. The percentage of UV-vis reflectance of powder samples was measured by diffuse reflectance spectroscopy using a Varian Cary 500 Scan UV-vis-NIR spectrophotometer (integrating sphere) instrument.
Water uptake tests were conducted to evaluate the adsorption of water by the composite membranes in order to determine the overall membrane porosity (Pr%), using the following equation:
![]() | (1) |
A Varian Cary 500 Scan UV-vis-NIR spectrophotometer was used to determine the solute concentration for permeate and feed samples containing BSA in Milli-Q water. The rejection was calculated by:
![]() | (2) |
To describe flux decline, the water flux is defined as a function of transmembrane pressure (ΔP), viscosity (η) and total membrane resistance (Rt):
![]() | (3) |
Metal oxide | Expected | Measured (ICP) | ||
---|---|---|---|---|
TiO2 (%) | ZrO2 (%) | TiO2 (%) | ZrO2 (%) | |
A | 100 | 0 | 98 | 0 |
B | 67 | 33 | 66 | 34 |
C | 50 | 50 | 53 | 57 |
D | 33 | 67 | 35 | 65 |
E | 0.0 | 100 | 0 | 96 |
The X-ray powder diffraction patterns of synthesized binary metal oxides investigated in this study are shown in Fig. 1.
![]() | ||
Fig. 1 XRD patterns of pure TiO2 and ZrO2 and TiO2–ZrO2 binary oxide powders. (A) TiO2, (B) TiO2(67%)–ZrO2(33%), (C) TiO2(50%)–ZrO2(50%), (D) TiO2(33%)–ZrO2(67%) and (E) ZrO2. |
The crystalline phases of the synthesized metal oxides A–E were identified as shown in Table 2. The crystal phase recognition procedure was accomplished by combining the XRD experiments (Fig. 1) with high-resolution transmission electron microscopy (HRTEM) imaging. In Fig. 2, the digital diffraction pattern (DDP) obtained from the polycrystals is shown as the inset.
Powder sample | d hkl (Å) | Relative intensity | Miller indices | DRX phase | Reference pattern* |
---|---|---|---|---|---|
A | 34![]() |
30 | 200 | Anatase (tetragonal) | 00-004-0477 |
23![]() |
35 | 220 | |||
18![]() |
35 | 222 | |||
16![]() |
15 | 422 | |||
14![]() |
15 | 204 | |||
B | 36![]() |
20 | 200 | Srilankite (orthorhombic) | 0-035-0584 |
28![]() |
100 | 211 | |||
23![]() |
15 | 220 | |||
21![]() |
20 | 222 | |||
C | 3498 | 40 | 200 | Zirconium titanium oxide (orthorhombic) | 00-034-0415 |
2886 | 100 | 211 | |||
2399 | 60 | 220 | |||
2106 | 30 | 222 | |||
1448 | 20 | 422 | |||
D | 36![]() |
10 | 200 | Zirconium titanium oxide (orthorhombic) | 01-080-1783 |
29![]() |
100 | 211 | |||
15![]() |
40 | 422 | |||
E | 38![]() |
10 | 200 | Zirconium oxide (monoclinic) | 01-078-0048 |
29![]() |
90 | 211 | |||
13![]() |
10 | 220 | |||
11![]() |
5 | 422 |
The observed diffraction patterns suggest the existence of a periodic arrangement of atoms conforming to different sets of atomic planes. Based on the information provided by the geometry of the diffraction directions, the crystal system can be recognized. In addition, the intensities of the diffracted rays are related to the nature of the atoms and the positions occupied in the crystal lattice, so that their measurement is a method for obtaining three-dimensional information about the internal crystal structure.
Considering the XRD measurements shown in Fig. 1 the most intense peaks corresponding to every powder sample can be recognized. Synthesized TiO2 (powder A) occurs as aggregates of nano-crystals in a predominant anatase phase (Fig. 2 and 5). Its distinctive peaks appear at 2θ: 25.31, 48.08, and 62.69, which correspond to the diffraction planes of (1 0 1), (2 0 0) and (2 1 3) Miller indices, respectively. For the zirconia sample (powder E), the predominant monoclinic phase of the ZrO2 micro-crystals was observed, which is in agreement with the literature.34 The monoclinic phase is obtained as a result of hydrolysis of the precursor of ZrO2. These results are in good agreement with the literature.35,36 In the binary metal oxides (powder B–D) highly crystalline particles still occur. Only several weak peaks of TiO2 are present in the XRD pattern of Ti(x)Zr(1−x)O2 samples. In agreement with the results shown in Fig. 2 the binary metal oxides mostly exist in the form of ZrO2 micro-crystals. The ZrO2 contribution inhibits the formation of TiO2 crystalline particles.
Unfortunately, detailed phase diagrams for the calcinated samples of TiO2–ZrO2 at relatively low temperatures are not available in the literature. Troitzsch et al.37 reported that as a result of slow kinetics of synthesis experiments performed at temperatures below 1200 °C the composition of these samples results in metastable phases in the TiO2–ZrO2 system. The sol–gel method, used in this study (calcination temperature 550 °C), is prone to such effects, so that the equilibrium state of those samples is uncertain.38,39
Considering the possible presence of a combination of phases described as Ti(x)Zr(1−x)O2, the analysis and identification of the major phases from the diffraction patterns (Fig. 2) of these powders were accomplished. For this purpose, the spacing between the individual planes dhkl was measured using an image digital processor (Digital-Micrograph 3.6) from every digital diffraction pattern (DDP). Subsequently, Miller indices were calculated.38 Once the Miller indices are known, the identified phases by XRD are corroborated with the reference patterns summarized in the database.43,44 Srilankite, a mineral with the chemical formula Zr2.20Ti1.80O8.00, was found to be the most probable crystalline phase for the powder B, synthesized at lower contents of TiO2, whereas zirconium titanium oxide chemically described as ZrTiO4 appears to be the predominant phase as a result of the increasing Zr content. Both chemical species yield adequate information about the crystal structure of TiO2–ZrO2 samples synthesized in this work.
Surface physico-chemical properties are relevant characteristics of the metal oxide particles. For this reason, hydrophilicity and zeta potential of metal oxides were evaluated. In all studied cases the observed contact angles were lower than 10°, suggesting a great hydrophilic character for the synthesized materials that could enhance the water permeation properties of doped membranes.
The zeta potential of the synthesized particles was measured for an aqueous solution containing 0.5 wt% of metal oxide at different pHs. Fig. 3 shows a typical amphoteric behavior of the electrophoretic mobility of metal oxide particles as a function of solution pH, indicating that particles have a lower tendency to agglomerate when they are electrically charged. Solution pH affects the hydrodynamic diameter of particles by changing the particle surface charge and zeta potential. TiO2 particles show a higher dimension with average sizes around 3 μm. Agglomeration of metal oxide particles can dominate when the solution pH (ca. 6) is close to the isoelectric point because the repulsive force between particles caused by electrostatic interaction can be mitigated.32
![]() | ||
Fig. 3 Zeta potential (triangles) and average size (circles) of metal oxide particles as a function of aqueous solution pH. |
Assuming that these metal oxides are used as semiconductors in different applications (e.g., as catalysts for water treatment, due to the high oxidation power42), the band gap energy of these materials was determined as a measure of their electric conductivity. As shown in Fig. 4, a gradual increase (blue shift) in the band gap is seen with an increase in the amount of Zr loadings. The band gap of the binary oxide samples was quite similar, indicating that their photoabsorbance properties are poorly related to their different compositions. The presence of Ti, even at low contents, determined the conductivity of synthesized binary metal oxides.40
![]() | ||
Fig. 4 Band gap energy of the synthesized metal oxides derived from the diffuse reflectance measurements. |
In order to evaluate the morphology (size and shape) of individual particles, TEM images of synthesized powders were taken. Fig. 5 shows the synthesized binary oxides as microparticles formed by aggregation of semi-spherical nanosized crystals with polyhedral shape and with different particle size distributions as a result of their composition. Table 3 summarizes some relevant size characteristics of individual nanoparticles. It can be seen that the incorporation of TiO2 into the general composition of synthesized metal oxides increases the individual particle size and at the same time the dispersion rate also increases, considering the standard deviation values.
Metal oxide type | Minimum (nm) | Maximum (nm) | Average size ± S.D. (nm) |
---|---|---|---|
TiO2 | 20.4 | 75.1 | 41.5 ± 16.5 |
TiO2(50%)–ZrO2(50%) | 24.3 | 187.8 | 85.3 ± 45.7 |
TiO2(33%)–ZrO2(67%) | 23.7 | 175.0 | 109.0 ± 57.1 |
ZrO2 | 4.4 | 11.6 | 7.9 ± 2.3 |
In general, all analyzed metal oxides show a tendency to form agglomerates held by weaker van der Waals forces.34 The microparticle size distributions of selected binary metal oxides were determined from SEM and are shown in Fig. 6. The EDS analysis of the element composition for each metal oxide is exhibited together with the corresponding particle size distribution.
![]() | ||
Fig. 6 Particle size distribution of synthesized metal oxides: powder A – TiO2, powder C – TiO2(50%)–ZrO2(50%), powder D – TiO2(33%)–ZrO2(67%) and powder E – ZrO2. |
The metal oxides TiO2 and ZrO2 form aggregates in micron sizes with an average particle size of around 3 μm. In particular, the prepared ZrO2 oxides are monodisperse aggregates, exhibiting a narrow particle size distribution, with a peak at 3 μm; the average particle size is smaller than the one observed for TiO2 particles. The formation of binary metal oxides increases the particle size, as well as enhancing the dispersivity of formed particle agglomerates. The observed polydispersity for binary metal oxides with a chemical structure type of Ti(x)Zr(1−x)O2 can be explained in terms of the increase of the surface area due to the dissimilar nuclei and co-ordination geometry.35 Considering the individual distributions of TiO2 and ZrO2 particles, the presence of titanium in their corresponding structures significantly alters the particle morphology. In other words, the addition of ZrO2 can effectively inhibit the excessive crystal growth of TiO2 micro-crystals during thermal processing.41
In order to explore the possible changes in the morphology of metal oxides as a result of dispersion in the solvent during the polymeric solution preparation, solid size measurements in the NMP dispersion were accomplished. The obtained particle size distributions (Fig. 7) were comparable to the metal oxide aggregates analyzed previously (Fig. 6). The particle size of metal oxides dispersed in the organic solvent ranged in the micrometric scale.
![]() | ||
Fig. 8 SEM cross-section images of synthesized membranes at two different magnifications: 800 and 5000×. An additional magnified image of membrane B is shown. |
Due to the high solubility of the solvent (NMP) in water and the low polymer content used for membrane synthesis by phase inversion, the resulting PES membranes exhibit a finger-like structure in their cross-section. The control PES membrane exhibits a typical asymmetric cross-sectional structure, consisting of a skin layer as a selective barrier, a thick finger-like sub-layer and a sponge-like bottom layer. It is well-known that the addition of low amounts (0.5 wt%) of inorganic fillers to the polymeric solution alters the morphology of the membrane by the formation of macrovoids and reduction of the thickness of the skin layer.33 Both effects are observed as a result of TiO2 addition. Moreover, the use of binary oxides Ti(x)Zr(1−x)O2 as inorganic filler particles provokes further changes in the membrane morphology. Symmetrical narrow macrovoids are formed, well fixed with the thinner membrane skin layer. The top and bottom sides of the membrane are connected through parallel channels, which can favour water permeation across the membrane. In general, the connectivity of surface pores with the sub-layers is improved after the addition of metal oxide particles.
A magnified image of membrane B is also included in Fig. 8 to allow visualization of metal oxide microparticles inside the membrane structure both in macrovoid walls and in the vicinity of the membrane surface. It is well-known that the hydrophilic character of the nanofiller increases the water transport through the membrane enhancing the membrane water permeation.
The pore size distribution of the membranes is shown in Fig. 9. The average diameters for control and doped membranes were 2.5 and 7.5 nm, respectively. As a result of the addition of inorganic particles, the average pore size of the membranes increased to around 5 nm.
The membrane porosity was investigated by two different procedures: from FEG-SEM images (surface pore density) and from uptake water experiments (overall porosity). Calculated values are summarized in Table 4. Surface porosity in terms of the ratio of pores area to the total membrane surface area yielded small values (lower than 0.5%) and are not included in Table 4. The presence of inorganic particles with specific hydrophilic nature enhances the diffusion of the non-solvent (water) into the nascent membrane during the phase inversion process. The favoured water diffusion promotes the formation of lean polymer phase points, which act as pore formation agents.30 Both effects, the increase of surface pore density and the growth of well connected and narrow macrovoids, explain the increasing free-volume in the membrane structure, which favours water sorption inside the membrane. This cooperative effect is evident for Memb. B and Memb. C, both being Ti(x)Zr(1−x)O2 modified PES membranes.
Membrane | Doping metal oxide | Surface pore density (pores number μm−2) | Porosity (%) |
---|---|---|---|
Memb. 0 | 4.35 | 71.26 | |
Memb. A | A | 5.19 | 86.12 |
Memb. B | B | 21.9 | 90.22 |
Memb. C | C | 17.2 | 89.08 |
Memb. D | D | 4.92 | 86.46 |
Memb. E | E | 4.80 | 83.23 |
The spatial distribution of particles in the membrane cross-section was explored through EDS by collecting the detected signal of chemical elements dispersed inside the membrane structure by X-ray maps (EDAX, USA). The option used for this analysis was EDX fast mapping that provides the capability to acquire X-ray maps at high speed in frame averaging mode. The acquisition parameters to select were: magnification 1000×, numbers of point select for the scan in x and y axis directions for mapping (matrix 256 × 200), resolutions in both directions 0.513 × 0.513 μm per pixel and the microscope operating at 20 keV.
The metal oxide pattern distributions inside the membrane structure obtained from EDS are exhibited in Fig. 10. As shown, pure TiO2 is evenly distributed through the membrane cross-section, in contrast to the pure ZrO2 pattern distribution that reveals that particle density is poor near the skin membrane layer. In addition, Fig. 10 demonstrates that microparticles are present in the membrane top layer whenever the metal oxide contains Ti.
Water permeation of synthesized membranes was tested for pure water (Milli-Q), as shown in Fig. 11B. All the doped membranes had a higher pure water flux than the control PES membrane, whereas the maximum fluxes were obtained for membranes doped with metal binary oxides. This is coherent with the observed increase of the hydrophilic character of the doped membranes. The higher flux obtained for these membranes can be further related to the increased surface pore density (Table 4) and improved connectivity of inner layers in the morphology (Fig. 8) of metal binary oxides. As the inter-connectivity between the sub-layers of the membrane increases with an increased number of surface pores of the skin layer, the membrane hydraulic resistance decreases, resulting in a higher water flux. So, there are three reasons that explain why the flux increases: increased hydrophilicity, surface pore density and pore connectivity.
![]() | ||
Fig. 12 Temporal evolution of protein solution flux and comparison of BSA solution flux with protein rejection of tested membranes. |
The incorporation of metal oxide particles into the polymer matrix caused a significant increase of the membrane flux. The PES-binary oxide doped membrane, Memb. B, had a 48% higher flux compared to the control PES membrane. Fig. 12 also compares the steady state fluxes.
In order to explore the effect of fouling on the BSA rejection, the observed rejections for each membrane are also shown in Fig. 12B. It can be seen that the rejection of the doped membranes is hardly affected by the incorporation of metal oxide fillers.
The impact of membrane modification on the membrane fouling performance was quantitatively studied using the resistance-in-series model32 as shown in Fig. 13.
![]() | ||
Fig. 13 Filtration resistances of control and doped membranes (a) intrinsic and total resistances (b) reversible and irreversible resistances. |
The effect of metal oxides addition can be studied from the observed differences between intrinsic membrane resistances obtained for control and doped membranes. In addition, fouling resistance caused by solute adsorption of foulants on the membrane pore wall or surface decreases for the modified membranes, especially for Memb. 2 and Memb. 3. In general, the fouling performance is dominated by the hydrophilic nature of the membrane. The control PES membrane has the highest total and irreversible resistance due to its lower hydrophilicity (Fig. 11A). However, the addition of metal oxides does not have a larger impact on the reversible fouling of membranes. This similar performance in reversible fouling can be caused by the low number of particles distributed at the membrane and pore surfaces, which is a disadvantage for the improvement of the membrane performance. The main difference in fouling resistance of the membranes was observed to be irreversible fouling. However, the irreversible resistance tracks closely with intrinsic resistance, suggesting that the main observed improvement is rather associated with the flux increment.
This journal is © The Royal Society of Chemistry 2014 |