Low-Temperature Synthesis of Anatase/Rutile/Brookite TiO₂ Nanoparticles on a Polymer Membrane for Photocatalysis

Abstract

Removing pollutants from water by using the photocatalyst TiO₂ is a highly-promising method. A large amount of work has been done to increase the activity of TiO₂, whereas the main two findings are increasing the surface area and applying mixed phase modifications (anatase, brookite, and rutile). Here, we present a method to directly synthesize non-agglomerated TiO₂ nanoparticles with different crystal phase ratios via low temperature dissolution-precipitation (LTDRP) on a porous microfiltration membrane (polyethersulfone). The amount of hydrochloric acid and the temperature was varied between 0.1–1 M and 25–130 °C, respectively, while the concentration of titanium precursor (titanium(IV) isopropoxide) was kept unchanged. The TiO₂ nanoparticles and the membrane were thoroughly characterized using X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), measuring the water contact angle and permeation flux, and examining the degradation of methylene blue. The mixed phase anatase/brookite with a main component being anatase exhibited the highest photocatalytic activity in removing methylene blue. Higher synthesis temperature induces enhanced crystallinity and, subsequently, the degradation rate of methylene blue was improved. Additionally, the photocatalytic activity remains high and unchanged for up to nine repeated cycles, i.e., full recovery of the photocatalytic properties is sustained.

1. Introduction

Water is one of the most important resources for humans and animals, but with a growing population it becomes more and more contaminated by chemicals, such as cleaning agents, personal care products, sweetening agents, insecticides, herbicides, and pharmaceuticals [1,2,3]. These organic micropollutants reach the water system from different origins, like human urine and feces, incorrect waste removal, waste water, or from industries [2]. Although the reported concentrations found in surface waters are at low concentrations (ng L⁻¹ to mg L⁻¹ range), undesired physiological effects for humans and animals cannot be neglected [2]. The current waste water treatment plants were not designed to purify water from many emerging contaminants (ECs), so that a constant entry is inevitably. There is a high demand for the development of intelligent systems to remove ECs from waste water. A fourth treatment stage, at the end of the waste water treatment, could remove ECs from water.

Advanced oxidation processes (AOPs, for example ozone, hydrogen peroxide, and titanium dioxide) efficiently clean waste water of organic micropollutants by mineralizing the organic molecules to water and carbon dioxide. The degradation is based on the generation of hydroxyl radicals, which oxidize organic molecules effectively and non-selectively [4]. The heterogeneous photocatalyst titanium dioxide (TiO₂) holds many advantages over other AOPs, like its photo and chemical stability, nontoxicity, reusability, and its low price [4].

TiO₂ can be either applied as a suspension [5] or bound to a support material e.g., on membranes [6], on carbon nanotube arrays [7], or on graphene oxide nanosheets [8]. The advantage of a supported TiO₂ system is that TiO₂ does not need to be extensively removed from the cleaned water. A suspension shows higher surface area compared to bound TiO₂, but taking a porous material like a polymer membrane as a support will increase the overall surface area [9]. The surface loss effect of a supported TiO₂ system is decreased. Additionally, the molecules to be degraded are transported directly to the surface of TiO₂ using a membrane reactor, and the removal of pollutants is enhanced.

Next to the surface area the crystallite type also significantly impacts the resulting photocatalytic activity [10]. Amorphous TiO₂ is not photoactive [11]. TiO₂ possesses three different types of modifications, namely anatase, rutile, and brookite [4]. Anatase is considered to have a higher photocatalytic activity compared to rutile, although the band gap is larger (3.2 eV for anatase vs. 3.0 eV for rutile) [12,13]. However, anatase, as an indirect band gap semiconductor (rutile is a direct band gap semiconductor), exhibits a longer lifetime for photo-excited electrons and holes. Additionally, anatase possesses a lower average effective mass of photo-generated electrons and holes. Therefore, these electrons and holes can migrate with a fast rate from the bulk to the surface, which results in a low recombination rate [13]. Few studies observed the photocatalytic activity of brookite as the synthesis of pure brookite is technically difficult [14]. In these studies [15,16,17] brookite is described as a better photocatalyst compared to anatase and rutile because electrons are trapped at moderate depths, so that both electrons and holes are still accessible for reduction and oxidation reactions. Generally, a mixture of these modifications gives higher photocatalytic activities due to synergistic effects like interfacial electron transfer [12,18,19,20,21,22]. With a mixture of anatase and rutile the recombination of photo-generated electrons and holes is reduced, and the absorption of light is increased by rutile [12,18,19]. A mass ratio of 3:7 (rutile to anatase) gives best photocatalytic activities [18]. A mixture of anatase and brookite, as well as rutile and brookite, has also been observed to gain fast degradation rates of organic molecules via photocatalysis [20,21].

Most methods deposit TiO₂ nanoparticles in a separate process on the membrane [23,24,25,26,27,28] or embed it directly in the membrane casting solution [29,30,31,32,33,34,35,36]. Disadvantages of the mentioned methods are aggregation of nanoparticles [9,26,28,37] and high sealing of nanoparticles in the bulk of the membrane polymer [9,23], respectively. The direct synthesis of TiO₂ nanoparticles on a membrane via hydrolysis of titanium alkoxides leads to finely- and evenly-distributed nanoparticles [9,37]. However, a separate step has to be included as the synthesized TiO₂ is amorphous and, therefore, not photocatalytically active. Crystallization is often performed by calcination at temperatures above 400 °C [12,18,19,38], but would degrade the supporting membrane material. Low-temperature crystallization with water [18,39,40,41] is a low-cost alternative compared to calcination. Water acts as a catalyst linking TiO₆²⁻ octahedra together to share edges leading to anatase [18,39,40,41]. Additionally, calcination of anatase to rutile at high temperatures (>500 °C) leads to agglomeration of nanoparticles, i.e., lower surface area [42,43,44]. However, only the crystal form of anatase was generated and, as stated above, a mixed phase is aspired to increase photocatalytic performance.

Here, we enhanced the synthesis to gain a mixture of TiO₂ phases (namely anatase, rutile, and brookite) of nanoparticles on a polyethersulfone (PES) membrane. The photocatalytic activity, i.e., the reaction rate constant, increased by more than three times compared to the membrane with pure anatase TiO₂ nanoparticles [9]. Additionally, the TiO₂ nanoparticle membrane completely degraded methylene blue for nine repeated cycles. TiO₂ nanoparticles were synthesized via hydrolysis (low temperature dissolution-precipitation LTDRP) [42,45] of titanium alkoxide using hydrochloric acid (HCl) in different concentrations and at various temperatures ranging from 0.1 M to 1 M and 25 °C to 130 °C, respectively.

The membranes modified with TiO₂ were thoroughly characterized. The morphology was examined via scanning electron microscopy (SEM), the TiO₂ content with X-ray photoelectron spectroscopy (XPS), the hydrophilic properties by water contact angle measurement, the crystal composition via X-ray diffraction (XRD), the water permeation with dead-end filtration setup, and the photocatalytic activity by degrading methylene blue (MB).

2. Results and Discussion

2.1. Crystal Phase

The formed crystal composition depends on the acid concentration and temperature used during synthesis. Changing the amount of titanium alkoxide [46] and time of reaction [47] also has an impact on the crystal composition, but was not investigated here. The three TiO₂ modifications develop due to a different linking of the TiO₆²⁻ octahedra [22,42,46,47,48] either by a solid-state rearrangement (anatase) or by a dissolution precipitation reaction (rutile) [49]. Anatase consists of edge-sharing TiO₆²⁻ octahedra, rutile shares an edge along the c axis to form chains and corner-shared bonding among these chains, and brookite is more complicated as it shares edges and corners [46]. The mechanism of crystal forming by combining TiO₆²⁻ octahedra has been discussed elsewhere and includes different theories using a thermodynamic and kinetic approach [47,50] or looking at the amount of OH⁻ present at different conditions [42,48].

The crystalline phase type was observed with XRD. Direct measurement of TiO₂ nanoparticles on the membrane is not possible as the amount of TiO₂ in comparison to the support PES membrane is too low and the membrane surface too rough to use a small incident angle. Therefore, the remaining non-attached TiO₂ nanoparticles were analyzed. The generated TiO₂ nanoparticles contained the phase type anatase, brookite, and rutile according to the adjusted acid concentration and temperature (Figure 1). Acid concentration as well as the temperature has a major impact on the formed crystal composition because the kinetics of reactions taking place was changed and the amount of OH⁻ was varied by temperature and acid [42,46,50].

At 90 °C (Figure 1a,c) the anatase and brookite content decreased from 60% to 4% and 24% to 0%, respectively, and the rutile portion increased from 15% to 96% by raising the acid concentration from 0.25 to 1 M. A similar behavior was observed at 120 °C and 130 °C. Thus, it appears that at higher acid concentrations the formation of rutile is favored. Brookite was formed at intermediate concentrations of acid and vanished before anatase to form rutile. A similar behavior was found in the literature [47]. At lower acidity more OH⁻ ligands are formed, so that edge-sharing of TiO₆²⁻ octahedra is much more likely to occur during dehydration reactions and anatase is formed [46]. With higher acidity the amount of OH⁻ ligands is lower. Edge-shared bonding is minimized because it needs two dehydration reactions taking place at the same time. Corner-sharing takes place more easily to gain rutile phase [46]. Another approach explaining the different effects of acid considers the different mechanism of crystallization, i.e., comparing the solid-state mechanism leading to anatase and the dissolution precipitation mechanism leading to rutile [49]. With increasing acid concentration, the surface of titania becomes protonated and titania solubilized. A dissolution precipitation process is possible so that rutile is formed [49].

At low temperatures (room temperature and 60 °C) the impact of acid concentration was less pronounced. At high acid concentration (0.5 and 1 M) only rutile was formed. With 0.25 M HCl the sample was either amorphous (room temperature: RT) or yielded 84% of anatase and 26% of rutile (no brookite). The amount of OH⁻ ions in water is changing with temperature, according to the equilibrium law (Le Chatelier's principle). The dissociation of water is an endothermic process. At room temperature the concentration of OH⁻ is at [OH⁻] = 2 × 10⁻¹⁴ M with an acid concentration of [H⁺] = 0.5 M. Increasing the temperature to 130 °C will increase the concentration of OH⁻ by about 119 times ([OH⁻] = 2.4 × 10⁻¹² M). The increase is not linear (ionic product K_w at 25 °C is 1 × 10⁻¹⁴ M², at 60 °C about 10 × 10⁻¹⁴ M² at 90 °C about 30 × 10⁻¹⁴ M², at 120 °C about 93 × 10⁻¹⁴ M², and at 130 °C about 119 × 10⁻¹⁴ M²) [51]. Thus, at lower temperatures (here, at room temperature and at 60 °C) the amount of OH⁻ is low and, therefore, edge-sharing is preferred to form rutile. With an acid concentration of 0.25 M the OH⁻ concentration is still high enough to form anatase next to rutile. At room temperature the crystallization was too slow to form crystallites [50]. Longer reaction time would generate also crystals of TiO₂. With higher acid concentration or temperature the crystallization rate is increased and crystal structures occurred.

Changing the temperature affects also the crystal composition (see Figure 1b,d). Again, a critical change occurred between 60 and 90 °C. At low acid concentration (0.25 M) the composition changed from amorphous (RT), to anatase-rutile (60 °C, 84% and 16%, respectively). At higher temperatures (90, 120, and 130 °C) also brookite was formed and the composition did not change in between 90, 120 °C, and 130 °C, significantly (anatase-brookite-rutile 90 °C: 61-24-15%; 120 °C: 70-26-4%; 130 °C: 61-27-12%). At a concentration of 0.5 M HCl only rutile was formed at room temperature and 60 °C, while at higher temperatures (90, 120, and 130 °C), next to rutile, anatase and brookite were also generated. With a high acid concentration (1 M) again only rutile was developed at temperatures below 60 °C. At temperatures above 90 °C a small amount of anatase was also generated, in addition to rutile.

Overall, only rutile occurs alone (high acid concentration and low temperature). Brookite appears ternary in combination with both anatase and rutile or just in combination with anatase (high temperatures and low acid concentration). Brookite is more easily transformed to rutile than anatase [47]. Therefore, brookite disappears first before anatase is transformed to rutile. Anatase was found to be formed together with rutile (high acid concentration and high temperature, as well as at 60 °C and 0.25 M HCl).

The crystal size of anatase, brookite, and rutile was calculated by the Scherrer formula. Overall the crystal size was growing with increased temperature and acid concentration (Supplementary Materials, Figure S1). As the anatase crystals grow larger, they become unstable. Anatase crystals migrate and align so that an in situ rearrangement of the lattice transforms them to rutile crystallites [52]. High acid concentrations, as well as large temperatures, accelerate crystal growth.

2.2. Morphology and the Amount of TiO₂

TiO₂ nanoparticles of different shape and quantity have been directly synthesized on a PES membrane. The temperature and acid concentration during synthesis had an impact on the above mentioned values.

The amount of TiO₂ on the membrane clearly increased with elevated temperatures (Figure 2 and Figure 3). The reprecipitation of TiO₂ was increased with higher temperatures and led to a higher content of TiO₂ on the membrane surface [43]. A trend of TiO₂ content with changing acid concentration is not clearly pronounced (Figure 3). At high temperatures (120 and 130 °C) the amount of TiO₂ was increased, while the acid concentration was raised from 0.1 M to 0.25 M HCl. Further increasing of the acid concentration led to a decrease of TiO₂ at high temperatures (120 and 130 °C). At temperatures of 60 and 90 °C the amount of TiO₂ was elevated with increasing acid concentration. The used concentration of HCl during synthesis changed the shape of the generated TiO₂ nanoparticles (Figure 2). At a HCl concentration of 0.25 M primarily spherical particles were formed. Only at 60 °C were rod-like particles also detected. Increasing the HCl concentration to 0.5 M produced spherical and rod-like particles, which can also be observed at HCl concentrations of 1 M at all temperatures. The rod-like particles formed flower-like structures (see Figure 2, e.g., 130 °C, 1 M HCl). The formation of rod-like particles is typically described in the literature for rutile [17,46,53], the generation of flower-like structures is known [46]. XRD (see Section 2.1) proves the assumption of the formation of rutile at higher concentrations of HCl.

2.3. Hydrophilicity and Water Permeation Flux

The hydrophilicity of the membranes with TiO₂ nanoparticles was evaluated by measuring the water contact angle on the membrane surface (Figure 4a). TiO₂ is a well-known additive to increase the hydrophilicity [36]. The amount of hydroxyl groups on the surface affects the water contact angle [54]. The unmodified membrane can be classified as hydrophilic with a value below 90° [55]. With increasing treatment temperature the water contact angle decreased, i.e., the hydrophilicity was enhanced. Values below 20° were observed for samples treated at temperatures above 90 °C. The structural morphology change visualized by SEM imaging (Figure 2) could be a reason for the change in contact angle. Structural changes have a high impact on the contact angle, which is widely described in the literature [54,56,57].

The generation of TiO₂ nanoparticles on the membrane did only reduce the water permeation flux by a small value (Figure 4b). An addition of TiO₂ nanoparticles on the membrane surface (also inner surface) should reduce the pore diameter, which would decrease the permeation flux. This effect was not observed here and can be explained by an increase in hydrophilicity (Figure 4a). Two competitive factors (hydrophilicity and pore narrowing) influence the flux, so that in sum the permeation flux did not change significantly.

2.4. Photocatalytic Activity

The photocatalytic activity was determined by degrading methylene blue over time. Methylene blue is widely used as a model compound for photocatalytic tests.

First the adsorption of methylene blue was studied, as the adsorption already influences the photocatalytic activity (i.e., high adsorption → high surface area and high photocatalytic activity) [58]. Methylene blue is a cationic dye, which will preferably adsorb on negatively charged surfaces. TiO₂ is negatively charged in the methylene blue solution (pH 8.5) with an isoelectric point (IEP) below 6.2 [59]. Starting at a synthesis temperature of 60 °C, the adsorption of methylene blue increased with the used temperature and acid concentration during synthesis of the TiO₂ nanoparticles (Figure 5a). The TiO₂ nanoparticles synthesized with the method from Fischer et al. [9] adsorb 21% of methylene blue, which can be increased to 39% with the synthesis method described here (1 M HCl and 130 °C). The variation in methylene blue adsorption can originate from a difference in surface area, IEP, and value of zeta potential. The raising content of TiO₂ (Figure 3) and the morphology change with increasing temperature and acid concentration (Figure 2) enhances the surface area, hence, more methylene blue was adsorbed. Additionally, the IEP and zeta potential (negative) decrease with the increasing particle size [59]. Due to the increasing acid concentration larger particles (rod-like particles, Figure 2 and supplementary Materials, Figure S1) were formed, and the zeta potential and IEP [59] decreased. Therefore, the adsorption of the cationic methylene blue dye increased. Additionally, the shape of the crystals has an influence on the adsorption, as the planar facets of the rutile crystals allows the alignment of an increased amount of adsorbed molecules on the surface [17].

The degradation of methylene blue was dependent on the temperature and acid concentration at which the TiO₂ nanoparticles have been synthesized on the membrane (Figure 5b,c). The reaction rate constant of the degradation of methylene blue was calculated by plotting log(c/c₀) versus time and estimating the slope of the regression line. A high rate constant implies fast degradation of methylene blue. With increasing temperature the rate constant was enhanced. The rate constant was raised from a minimal value of 0.004 min⁻¹ (room temperature and 0.25 M HCl) to 0.0607 min⁻¹ (130 °C and 0.1 M HCl). After 60 min methylene blue was completely degraded with a TiO₂ nanoparticle membrane synthesized at 120 and 130 °C while the membrane with TiO₂ nanoparticle produced at room temperature removed only 24%. With higher synthesis temperatures the reflections of the XRD pattern was narrower (Figure 1b) indicating higher crystallinity. Increasing the crystallinity will reduce the formation of electron traps, so that fewer recombination centers of photogenerated electrons and holes are created [60]. Hence, the photocatalytic activity will be improved. Increasing the temperature further on could increase the crystallinity and photocatalytic activity. However, the chosen support (PES membrane) would be damaged if temperatures above 130 °C are applied.

The acid concentration influences the photocatalytic activity at higher temperatures (120 and 130 °C). The rate constant was decreased with raising acid concentration. At 120 °C the rate constant decreased from 0.0504 min⁻¹ (0.1 M HCl) to 0.0309 min⁻¹ (1 M HCl). The trend of degradation was contrary to the adsorption of methylene blue (Figure 5a,b). Higher adsorption would usually lead to higher photocatalytic activity as the affinity of methylene blue to TiO₂ and/or the surface area of TiO₂ is enhanced. However, the crystal phase composition also has an impact on the photocatalytic activity. A combination of a high content in anatase (91%) and low content of brookite (9%) enhances the photocatalytic activity (130 °C and 0.1 M HCl). A combination of brookite and anatase have been reported in the literature to have higher photocatalytic activity compared to pure anatase [20] and triphasic composites additionally containing rutile [22]. Pure brookite and anatase have been described to have a similar degradation rate and a much higher photocatalytic activity than rutile [17]. A combination of these two crystals (anatase and brookite) has a synergistic effect as the electron transfer between these two faces is enhanced [22].

The reduction of methylene blue was repeated for up to nine cycles (see Figure 5d) with two differently modified membranes (60 and 120 °C, 0.25 M HCl). The TiO₂ nanoparticles on the membrane synthesized at 120 °C do not show any activity loss over nine cycles. This is in contrast to the TiO₂ nanoparticles produced at 60 °C on the membrane as the reduction of methylene blue gradually deteriorates with repeating degradation runs. This membrane does not regain its white color after the first cycle and the membrane is colored deep blue after nine cycles (see Figure 6). Methylene blue is still adsorbed at the membrane surface while the membrane with TiO₂ nanoparticles synthesized at 120 °C is white after the first cycle and light blue after the ninth cycle (Figure 6). Although the membrane with TiO₂ nanoparticles produced at 120 °C turns light blue after nine cycles the photocatalytic activity is not declined and full recovery of the photocatalytic properties is given. An indirect proof for the stability of the membrane against an attack from OH^• radicals was obtained for the membrane synthesized with TiO₂ nanoparticles at 120 °C as the photocatalytic degradation rate was not changed after an irradiation for 900 min. An attack of OH^• radicals would have resulted in a change in photocatalytic activity. The TiO₂ nanoparticle film (synthesized at 120 °C) on the membrane acts as a protective layer as it totally covers the surface of the membrane [27]. It is well known that semiconducting photocatalysts can degrade carbon materials [61,62]. Therefore, a possible damage of the membrane material will be the focus of further studies.

3. Materials and Methods

3.1. Materials

The PES membrane (0.22 µm, Millipore Express^® PLUS membrane GPWP) was purchased from Merck Millipore, Billerica, MA, USA. Hydrochloric acid (37%) and titanium(IV) isopropoxide (TTIP, 97%) were acquired from Sigma-Aldrich, Steinheim, Germany and methylene blue (96+%) from Acros Organics (Fisher Scientific, Waltham, MA, USA). Milli-Q water was obtained from Milli-Q integral system (EMD Millipore, Billerica, MA, USA).

3.2. Nanoparticle Synthesis on the Membrane

The membranes were cut into pieces of 3 × 9 cm and two of them were put into a Teflon beaker. Thereby, the top side of the membrane (the side with smaller pores) was turned inward-looking. The beaker was filled with Milli-Q water (80 mL in total), TTIP (8.1 mL), and 37% hydrochloric acid. The amount of HCl depends on the used total concentration in the beaker varying from 0.1 to 1 M.

The beaker was closed and shaken for 4 h at room temperature. Afterwards, the membranes were tempered for 20 h at different temperatures in the beaker. After cooling, the membranes were washed with Milli-Q water by shaking 3 × 30 min. Subsequently, the membranes were dried in air.

The excessive TiO₂ suspensions were, depending on the particle size, either centrifuged or allowed to stand overnight to achieve sedimentation of the particles. The supernatant clear liquid was decanted and the sludge was blended with Milli-Q water and shaken for washing. After re-sedimentation and decantation, the sludge was dried, pestle, and used for XRD measurements.

3.3. Characterization

For SEM imaging, the samples were sputtered with a 30 nm chromium layer with a Leybold Z400 sputter system (Cologne, Germany). SEM images were taken with a Carl Zeiss Ltd. Ultra 55 SEM (Oberkochen, Germany). XPS spectra were measured at the Kratos Analytical Ltd. Axis Ultra (Manchester, UK) with a monochromatic Al Κα cathode. The X-ray source power was 150 W and the pass energy was 40 eV. To gain information about the crystal composition of the TiO₂ nanoparticles XRD measurements have been performed. It is not possible to directly perform XRD measurements of the TiO₂ layers on the membranes as the signals are superimposed by noise caused by the polymer structure of the membrane. Therefore, the excessive TiO₂ powder in suspension was refined after synthesis and used for XRD. The XRD patterns have been measured with a Rigaku Ultima IV X-ray diffraction spectrometer (Tokyo, Japan) with Cu Κα radiation (40 kV, 40 mA, scanning speed: 2 ° min⁻¹, step size: 0.05°). From the peak integrals A, the weight fractions W of the phases were calculated with Equations (1)–(3) using the response factors k (k_A = 0.886, k_R = 1 and k_B = 2.721) [22]:

$$W_A= \frac{k_A·A_A}{k_A·A_A+k_R·A_R+k_B·A_B}$$

(1)

$$W_R= \frac{k_R·A_R}{k_A·A_A+k_R·A_R+k_B·A_B}$$

(2)

$$W_B= \frac{k_B·A_B}{k_A·A_A+k_R·A_R+k_B·A_B}$$

(3)

The crystallite sizes of TiO₂ were determined by using Scherrer’s equation:

$$d=0.9\lambda /B\cos \theta$$

(4)

where d, λ, B, and θ are the crystallite size, Cu Kα wavelength (0.15418 nm), full width at half–maximum intensity (fwhm) of the reflection (anatase (101), brookite (121), and rutile (110)) in radians, and Bragg’s diffraction angle, respectively [63]. For contact angle measurement a piece of modified membrane was glued to an object slide. With a Krüss DSA 30E (Hamburg, Germany), a 2 µL water drop was placed on the surface. The process of drop deposition and permeation was tracked by a camera and the contact angle was determined from the video after the drop was completely placed on the membrane. This procedure was repeated on five areas of the membrane. For water permeation flux measurements, membrane circles with a diameter of d = 2.5 cm were cut and placed in a Sartorius 16249 pressure filter holder (Göttingen, Germany). The time for 100 mL of Milli-Q water passing through the membrane at an applied nitrogen pressure of p = 1 bar was measured. For each sample three membrane pieces were tested. From the measured time, the water permeation flux J could be calculated using Equation (5):

$$J=\frac{V}{t \times A \times p}$$

(5)

where J is the permeability (mL min⁻¹ cm⁻² bar), V is the volume (mL), t is time (min), A is the surface area of the membrane (cm²), and p is the pressure (bar).

The procedure of the photocatalytic degradation test has been described elsewhere [9,64]. Briefly, a circle with the diameter d = 2.5 cm was cut from the membrane. An additional hole was cut in the middle of these circle (d = 1 cm). Three of these circles were glued into vicinal pits of six-well-plates. All six pits were filled with 4 mL of a 20 µM methylene blue solution in water. The zero value of the absorption by methylene blue was measured immediately with a Tecan Infinite M200 multimode microplate reader (Männedorf, Switzerland) at a wavelength of λ = 660 nm. The samples were shaken on a radial shaker with 200 rpm without irradiation to measure the dark adsorption for 30 min. The absorption of the solution was measured in 10 min intervals. Subsequently, the samples were irradiated with a sunlamp (Heraus Original Hanau Suncare tanning tube 21/25 slim, Hanau, Germany) to perform photocatalysis for 2 h. The radiant flux density of the lamp was 11.3 ± 1.3 mW cm⁻². For the first 40 min, the absorbance of the remaining methylene blue was measured every 10 min and, afterwards, every 20 min. The non-catalytic photolysis of methylene blue has been subtracted from the photocatalytic measurement. By plotting log(c/c₀) versus the time, the observed reaction rate constants k could be derived from the slope of the regression line.

4. Conclusions

Non-agglomerated TiO₂ nanoparticles have been directly synthesized via low-temperature dissolution-precipitation on a microfiltration PES membrane with different crystallinities and varying crystal composition, including anatase, brookite, and rutile. For that purpose the amount of hydrochloric acid and reaction temperature was varied between 0.1–1 M and 25–130 °C, respectively, while the concentration of titanium precursor (titanium(IV) isopropoxide) was kept unchanged. The crystallinity and modification composition has an influence on the resulting photocatalytic activity. High crystallinity and a mixed phase of mainly anatase (79%) and a small amount of brookite (21%) give the best degradation rate of methylene blue. The photocatalytic activity was not declined and full recovery of the photocatalytic properties of at least nine repeated cycles was retained with a mixed phase sample containing anatase, brookite, and rutile (70%, 26%, and 4%, respectively) and having a high crystallinity. The degradation rate of pure rutile TiO₂ nanoparticles on a membrane with lower crystallinity subsequently declined over nine cycles. The permeation flux of the membrane did not change significantly, so that the membrane can be used for membrane reactors without any loss in flux. Additionally, the water contact angle of the membrane was improved to a value below 20° at synthesis temperatures above 60 °C. Thus, the membrane is highly suitable for water applications and low fouling can be expected. The implantation of this catalytic membrane in a reactor system for degradation of organic pollutants is highly promising.