A Simple and Ligand-Free Synthesis of Light and Durable Metal-TiO2 Polymer Films with Enhanced Photocatalytic Properties

optimized


Introduction
[21][22][23] Apart from the formation of the Schottky barrier, loading of plasmonic metals such as Ag, Cu, and Au on the TiO 2 can also enhance the photocatalytic activity of TiO 2 via different mechanisms because of the localized surface plasmonic resonance effect.[26] In constructing such highly active hybrid metal NP-TiO 2 nanomaterials, it is important to control the size, morphology, and distribution of the doped metal NPs and to create intimate contact between them and the TiO 2 surface. [2,20,27,28][30] The even bigger challenge for solgel methods which use pre-synthesized metal NPs is to create intimate contact between the TiO 2 and metal NPs, since the metal NPs are normally covered by organic molecules that are introduced during the synthesis and/or added to stabilize them.37] In this report, we present a simple approach to fabricate robust, flexible, and lightweight metal-TiO 2 -polymer SENSs which are convenient to use and possess superior photocatalytic efficiency and durability.The procedure is straightforward and involves interfacial self-assembly of P25 TiO 2 NPs followed by physical vapor deposition (PVD) of metal NPs.We show that interfacial self-assembly allows powdered TiO 2 NPs to be deposited into a densely packed layer on a support for the formation of TiO 2 SENSs which maximizes the proportion of the TiO 2 NPs available for catalysis.The PVD process allows the metal NPs to be deposited uniformly and with direct control over their size and loading, without the need for surface-ligands.Since the selfassembly is also achieved without the use of organic ligands this approach creates intimate contact between metal and TiO 2 NPs which allows for unhindered electron-transfer between them, maximizing the performance enhancement given by the metal NPs.The product metal-TiO 2 SENSs were not only dramatically lighter and more flexible that standard commercial TiO 2 -based photoactive glass but they also out-performed it by a maximum of ≈18× in standard dye degradation experiments.

Results and Discussion
As illustrated in Figure 1, the key step in the formation of the required TiO 2 (Evonik P25) SENSs is the assembly of TiO 2 NPs at a liquid-liquid (water-dichloromethane) interface.10] Therefore, in this work non-adsorbing "promoter" molecules were added to the mixture.][40][41][42] In practice, this assembly process simply involved shaking an aqueous dispersion of TiO 2 NPs with dichloromethane, which contained micromolar concentrations of promoters (Figure 1, step 1, see Experimental Section for details).This created particle-coated emulsion droplets which coalesced to form a two-layer liquid system with a 2D array of TiO 2 NPs at the interface (Figure 1, step 2).When polystyrene was pre-dissolved in the dichloromethane, subsequent evaporation of dichloromethane could lead to deposition of a layer of polystyrene onto the assembled particles on the side of the interface which the oil resided.This yielded a flexible polymer film carrying a densely packed monolayer of TiO 2 particles on one side (Figure 1, step 3).The as-prepared TiO 2 SENSs were very thin (≈2.5 µm thick as shown in Figure S1, Supporting Information) but physically robust, which allowed them to be handled and deployed easily in routine use, for example in the following treatment with metal sputtering and their eventual application as convenient photocatalysts.
The optical images of an as-prepared solid TiO 2 SENS and its precursor colloid are shown in Figure 2a.While the TiO 2 colloid appeared to be white and opaque, the as-prepared TiO 2 SENS was almost colorless and transparent, suggesting that the SENS carried only a very thin layer of TiO 2 NPs.In fact, inductively coupled plasma optical emission spectroscopy (ICP-OES) analysis showed that there was only ≈0.8 g of TiO 2 NPs per square meter of TiO 2 SENS.This also explains the extremely weak TiO 2 signals obtained from TiO 2 SENSs in the following X-ray diffraction (XRD) analysis (Figure 2b).Despite being extremely weak, the XRD patterns obtained from TiO 2 SENSs were identical to those obtained from TiO 2 NPs, indicating that the preparation process had an insignificant effect on the crystalline structure of the assembled TiO 2 NPs.The nanoscale morphology of the surface TiO 2 layer in the product TiO 2 SENSs was characterized using SEM.As shown in Figure 2c, the plan view of the surface of a typical TiO 2 SENS sample confirmed a densely packed and exposed layer of P25 TiO 2 NPs.The corresponding TEM image can be found in Figure S2, Supporting Information.To study the thickness of the TiO 2 layer, a crack at the edge of the SENS was probed with SEM at a tilted angle.As shown in Figure 2d, the SEM revealed a TiO 2 layer which was tens of nanometers thick.More importantly, the image clearly demonstrated that the TiO 2 NP layer sat on top of the polymer support and was therefore exposed for further functionalization or applications.The observed morphology is also consistent with our previous work which showed that for a range of different NPs the product SENSs contained particles which were partly anchored onto the polymer support with the bulk of their surfaces exposed. [7,38]This morphology constitutes a significant advantage over conventional TiO 2 film catalysts where a large proportion of TiO 2 NPs are inaccessible during catalytic reactions since they are buried within the substrate or beneath other layers of catalyst that sit closer to the surface. [6]etal NPs were deposited onto the TiO 2 SENSs using PVD treatment, as shown in Figure 3a.Au NPs were used for proofof-principle studies since Au is one of the most widely used electron-sink material for TiO 2 photocatalysts. [19,20]In practice, our method is extremely versatile since PVD can be generally applied to introduce hundreds of different nanomaterials onto the TiO 2 SENSs under mild conditions within seconds.
As shown in Figure S3, Supporting Information, the XRD and UV/vis profile for TiO 2 SENSs sputtered with Au NPs were identical to bare TiO 2 SENSs, which suggested that Au was only present in extremely small amounts.To confirm the presence of Au NPs, XPS of the sputtered and bare TiO 2 SENSs was conducted.As shown in Figure 3b, XPS characterizations of the sputtered TiO 2 SENSs showed a distinct Au (0) peak at 83.2 eV (Au 4f 7/2 peak), which was slightly down-shifted compared to the 84.0 eV value of bulk Au, indicating a lower oxidation state of the sputtered Au nanoparticles compared to bulk Au.Correspondingly, XPS characterization of Ti 2p peaks showed increased binding energies of Ti after Au sputtering, indicating a higher oxidation state of Ti, and more generally, suggesting strong interactions and electron-transfer between the Au and TiO 2 NPs (Figure 3b and Figure S4, Supporting Information). [43,44]To study this effect, density functional theory (DFT) calculations were performed to simulate the interaction between an Au cluster and anatase TiO 2 {1, 0, 1} slab at room temperature.As shown in Figure 3d, the calculated adsorption energies of Au on a clean TiO 2 {1, 0, 1}, and defected TiO 2 {1, 0, 1} slab with an oxygen vacancy were −0.79 and −2.11 eV, respectively, which confirms the strong interaction between the Au and TiO 2 photocatalyst suggested by the XPS results.
Another significant benefit of the PVD process is that it allows the size and loading of metal NPs to be easily and reproducibly controlled with sputtering duration.The effect of Au sputtering was characterized by TEM, as shown in Figure 3d-f and Figure S5, Supporting Information.Since the polymer layer of the polymer-supported Au-TiO 2 SENSs was sufficiently thick to interfere with direct TEM analysis, the TEM samples were fabricated by scooping the interfacial TiO 2 arrays directly from the liquidliquid interface with TEM grids.TiO 2 layers deposited on TEM grids were similar to those obtained with the polystyrene support but had lower uniformity and particle coverage (Figure S2, Supporting Information).Nonetheless they were sufficiently uniform to represent the actual TiO 2 surface on a TiO 2 SENS and allowed the size and distribution of Au NPs created by PVD deposition to be determined.For convenience, Au-sputtered TiO 2 samples are denoted Au (x) -TiO 2 , where x represents the length of Au sputtering in seconds.Based on a count of 100 individual Au NPs, the size distribution of Au NPs for each sputtering time was obtained, which is shown in insets in Figure 3e-g and Figure S5, Supporting Information.The sizes of the Au particles increased from 2.8 ± 0.7 to 3.1 ± 0.9 and 6.5 ± 1.7 nm as the sputtering time was increased from 5 to 15 and then 30 s.The mass of Au deposited was determined using ICP-OES analysis and was found to rise from 1.67 to 6.85 and finally 11.27 µg cm −2 , which translates to 0.021, 0.086, and 0.141 g/g TiO 2 .It is notable that at the longest sputtering time the deposited Au NPs became slightly irregular and polydisperse due to the growth and/or merging of Au NPs.However, this was not a significant issue since this did not occur at the 15 s, which gave the optimal catalytic efficiency (see below), and the overall picture is that sputtering offers a near ideal route to depositing small (<10 nm) and uniformly distributed Au NPs on the surface of TiO 2 .This contrasts drastically with colloidal systems, where the presence of organic ligands and Ostwald ripening cause issues, and with systems with in situ generated particles, such as photo-deposition of Au NPs on the surface of TiO 2 , where the morphology and uniformity of the deposited particles   they can be folded to fit in practically any reaction containers, which in this case was a small sample vial containing 10 mL of MO solution (Figure S8, Supporting Information).Figure 4b shows that Au sputtering for any time between 5 and 30 s increased the rate of MO decomposition compared to the control TiO 2 SENS.Apart from improvement in electron-hole separation efficiency, it has been shown that the enhancement in photocatalytic activity of metal-TiO 2 systems might also arise from an improvement in adsorption properties. [45,46]To study this the adsorption of MO on TiO 2 and Au (15) -TiO 2 SENSs were studied using UV-vis spectroscopy.As shown in Figure S9, Supporting Information, the adsorption kinetics and capacity of the TiO 2 and Au (15) -TiO 2 SENSs were found to be identical, which suggests that the enhancement in photocatalytic activity of the Au (15) -TiO 2 SENSs arose mainly from improved electronhole separation efficiency due to the Au NPs acting as electron sinks.
Figure 4c summarizes the photocatalytic activity of Au (x) -TiO 2 SENSs with different Au sputtering times.These results were based on tests conducted with three different batches of metal-TiO 2 SENSs.In general, the photocatalytic activity of the SENS increased as the length of Au sputtering was raised from 5 to 15 s.The maximum reaction rate constant for the degradation of MO was 1.95 × 10 −3 min −1 cm −2 at 15 s of sputtering time, which was ≈4× that of the control TiO 2 SENSs (0.48 × 10 −3 min −1 cm −2 ).A further increase in the duration of Au sputtering, however, led to a decrease in photocatalytic activity.This may be due to the increase in recombination centers as the loading of Au NPs increased. [27]In addition, the increase in Au NP loading can lead to reduction in exposed TiO 2 surface.Since both Au and TiO 2 are needed for catalyzing the degradation of MO, this means that the optimum amount of Au would be determined by the need to balance the number of electron sinks provided by the Au NPs with the amount of exposed area of TiO 2 surface. [5,27]Apart from this, increasing the surface coverage of Au NPs will also block more of the incoming light from reaching the TiO 2 catalyst surface. [5,27]In order to confirm that the degradation of MO arises from photocatalytic effect of the SENSs rather than from its direct absorption of UV photons, photocatalytic degradation of 4-chlorophenol which has low absorbance at 365 nm was studied under the same conditions.The results, shown in Figure S10, Supporting Information, were found to be similar to those obtained for MO.
The mechanism for photocatalytic degradation of MO is illustrated in Figure 4e.Upon UV illumination of Au (x) -TiO 2 SENSs, electrons are excited from the valence band of TiO 2 to the conduction band while generating holes in the valence band (Equation ( 1)).The excited electrons are then transferred rapidly to the Au NPs attached on the surface of TiO 2 .Presumably the transfer of electrons from TiO 2 to Au does not significantly lower the reduction potential of photogenerated electrons, the reduction potential of photogenerated electrons equals to the energy level at the bottom of conduction band of TiO 2 , which is ≈−0.5 V. [47][48][49] Therefore, photogenerated electrons are capable of reducing MO directly or alternatively reacting with electron acceptors, such as O 2 , which is present on the catalyst surface, thereby generating superoxide radical anions (E 0 (O 2 / O 2 • − ) = −0.28V, Equation ( 2)).49] Since it is possible that OH• radicals and O 2 • − radicals were both involved in the photocatalytic degradation of MO, corresponding radical scavengers (benzoquinone for O 2 • − radicals and tert-butyl alcohol for OH• radicals) were added to MO solution to provide further insight into the photocatalytic mechanism.It was found that the addition of both scavengers led to decrease in MO degradation efficiency.The degradation efficiency dropped by over 80% after addition of tert-butyl alcohol, indicating that OH• played the main role in degrading MO in the current photocatalytic system. [50,51]O T iO (3) To showcase the versatility of our approach Pt (x) -TiO 2 SENSs were produced via the same procedure.The size and distribution of Pt NPs were determined by analyzing the corresponding TEM images (Figure S11, Supporting Information), which followed the same pattern as for the Au NPs discussed above.As a result, the general trends for the effect of Au and Pt sputtering were also found to be quite similar, in that the photocatalytic activity of three different batches of Pt (x) -TiO 2 SENSs also showed a rise and then a fall with increased Pt sputtering time (Figure 4d).The increase in photocatalytic activity with Pt was found to be even higher than that observed with Au.This is most likely due to a larger difference between the work functions of Pt and TiO 2 compared to Au and TiO 2 , which enables electron-hole pairs to be more efficiently separated in Pt (x) -TiO 2 SENSs, thereby leading to significantly more enhanced photocatalytic activity. [18,52,53]The maximum reaction rate constant for MO degradation with Pt (30) -TiO 2 SENSs was almost 6× (2.89 × 10 −3 min −1 cm −2 ) than that of TiO 2 SENSs.Moreover, since we have previously shown that even the pure TiO 2 SENSs were already ≈3× more photocatalytically active than their commercial counterparts, Pilkington Activ, this means the optimal Pt (30) -TiO 2 SENSs are ≈18× more photocatalytically active than Pilkington Activ.
Apart from being highly photocatalytically active, the metal-TiO 2 SENSs were also found to be stable under the UV irradiation conditions used for photocatalysis.Although it has been reported that polystyrene can be photo-oxidized by TiO 2 , [54] our previous research suggested that the polystyrene side of these TiO 2 SENSs remained unchanged after week-long intense UV irradiation. [7]Moreover, as shown in Figure 5a, due to the strong interaction between metal and TiO 2 , the Au (15) -TiO 2 SENSs can sustain at least four complete cycles of reaction without any noticeable changes to their photocatalytic activity.This is supported by the XPS results in Figure 5b, which show that the oxidation state of the Au NPs the in Au (15) -TiO 2 SENSs remained unchanged even after 12 h of continuous reaction.Similarly, ICP-OES analysis revealed that the there was a negligible change in the amount of Au on the sputtered film, which shows the excellent durability of the product SENSs (Table S1, Supporting Information).

Conclusions
In summary, we have shown a simple and rapid method to prepare highly photocatalytically active, light, and durable metal-TiO 2 SENSs under mild conditions.The hybrid SENSs contained a near monolayer of TiO 2 NPs anchored onto the surface of a thin, flexible polystyrene film support, with small (<10 nm) Au or Pt NPs distributed evenly on the TiO 2 surface.Importantly, our approach is completely ligand-free, which allows intimate contact between the metal NP electron-sinks and TiO 2 catalyst to be achieved without the need for any further processing steps.These properties combine to make the product metal-TiO 2 SENSs highly photocatalytically active, as measured by the rate of degradation of MO and 4-chlorophenol.As a result, the Au-TiO 2 and Pt-TiO 2 SENSs were measured to be 12× and 18× more photocatalytically active than their TiO 2 -based commercial counterparts.Moreover, these SENSs can be easily deployed and recovered in practical photocatalytic applications thanks to the flexible but robust polymer support.Since both the self-assembly method for film production and PVD of electron-sinks are generally applicable to a variety of materials, this means that the current method can be readily exploited to prepare highly active, durable, and easy-to-use photocatalytic films from various combinations of photocatalysts and electron-sinks, which will be pave the way for the production of functional materials for direct applications or for the construction of more sophisticated devices.Preparation of TiO 2 SENSs: Polymer-supported TiO 2 SENSs were prepared using the self-assembly method previously reported by our group with slight modifications. [7]Typically, 5 mL of TiO 2 colloids which were diluted by a factor of 2000 were mixed with 0.14 mL of 1 mm of TBA + NO 3 − and 3 mL of polystyrene/dichloromethane (0.06 g mL −1 ) solution.The mixture was vigorously shaken for ≈1 min to facilitate the migration of TiO 2 NPs to the surface of emulsion droplets.After shaking, the emulsion droplets covered with TiO 2 NPs were poured immediately into a petri dish covered with a lid to prevent solvent evaporation.After all the emulsion droplets had coalesced to form an interfacial 2D array of TiO 2 NP-clusters, the cover of the petri dish was removed to allow evaporation of dichloromethane at room temperature, which led to the deposition of a thin polystyrene film on the TiO 2 array at the interface.After complete evaporation of the dichloromethane phase, solid TiO 2 SENSs were mounted onto Sellotape for further characterizations and applications.

Experimental Section
Preparation of Au/Pt Sputtered TiO 2 SENSs: Au/Pt NPs were sputtered onto TiO 2 SENSs using a plasma-assisted reactive direct current sputtering deposition (PAR-DC-MS) system.The sputtering process was carried out at room temperature and in argon environment with a constant pressure of 7 × 10  Instrumentations: Scanning electron microscopy (SEM) was performed using a Quanta FEG 250 at an acceleration voltage of 20 kV under high chamber vacuum with standard SEM copper tape or carbon tape mounting.Transmission electron microscopy (TEM) was performed using a Joel JEM-1400 Plus TEM.To mimic metal-TiO 2 SENSs, TEM samples were prepared by lifting TiO 2 2D arrays from the water/ dichloromethane interface using TEM grids, which were then sputtered with Au/Pt under the same conditions used for sputtering TiO 2 SENSs.XRD patterns of TiO 2 SENSs before and after Au sputtering were obtained on a PANalytical X'Pert Pro X-ray diffractometer equipped with a copper X-ray source (40 kV, 40 mA).All XRD analysis was performed ex situ on a spinning stage with 2θ ranging from 5° to 90° using a step size of 0.017° and a step time of 15 s.XPS was performed with a Kratos AXIS Ultra DLD apparatus equipped with a monochromated Al Kα radiation X-ray source (10 kV, 15 mA), a charge neutralizer and a hemispherical electron energy analyzer.Survey scans were acquired at an analyzer pass energy of 60 eV and high-resolution narrow scans were performed at a constant pass energy of 20 eV in 0.1 eV steps with a dwell time of 500 ms.The pressure in the analysis chamber was maintained below 5 × 10 −9 torr for data acquisition.The spectra were then analyzed using CasaXPS and corrected for charging using the C 1s feature at 284.8 eV.UV-vis spectra were recorded on a Perkin Elmer Lambda 800 UV-vis spectrometer with a scan rate of 400 nm min −1 in the range 200-800 nm.Inductively coupled plasma-optical emission spectrometry (ICP-OES) measurements were performed with an Agilent 5100 Synchronous Vertical Dual View ICP-OES equipped with a SeaSpray nebulizer, a single-pass cyclonic spray chamber, a 1.8 mm DV i.d.injector torch and a Vista Chip II CCD detector.The sample for ICP-OES measurement was prepared by digesting metal-TiO 2 SENS in nitric acid in a microwave digester to extract the metal elements.
Computational Method: DFT calculations were performed using the Vienna ab initio simulation package program at the level of Perdew-Burke-Ernzerhof functional. [55,56]Spin-polarization was also considered. [57,58]The project-augmented wave method was used to represent the core-valence electron interaction while the valence electronic states were expanded in a plane wave basis set with the energy cutoff at 450 eV.The ionic degrees of freedom were relaxed using the quasi-Newton Broyden minimization scheme until the Hellman-Feynman forces on each ion were less than 0.05 eV Å −1 .A (1 × 1 × 1) k-point mesh was used for all ab initio molecular dynamics (AIMD) simulations in order to accelerate the process, while a (2 × 2 × 1) mesh was utilized for converging the energetics.Considering the intrinsic shortcomings of DFT for calculating electronic properties of semiconductor materials, the DFT+U method was selected, where the on-site Coulomb correction was set on the Ti 3d orbitals with an effective U value of 4.2 eV, as suggested in literature. [59,60]In order to accommodate the rather large Au/Pt cluster, the model for the TiO 2 system was established using a relatively large supercell of the anatase (101) surface, which contains a three-layer p(2 × 3) periodic slab consisting of 108 atoms (36 Ti atoms and 72 O atoms) and an 11.5 Å vacuum layer. [61]AIMD simulations were used to determine the optimal structure of Au on pristine and reduced TiO 2 surfaces.The simulations were carried out in the canonical (NVT) ensemble employing Nose-Hoover thermostats.The temperature was set at 450 K that was taken from the temperature of hydrothermal treatment commonly used in experiments, and the time step was 1 fs. [62,63]More than 20 ps AIMD simulations were performed until the system became equilibrated.From the equilibrated trajectory, structural configurations at 1 ps intervals were selected and fully optimized.The adsorption energy of Au/Pt cluster (E ads ) is defined as the energy difference before and after the adsorption with respect to the gas-phase cluster molecule as shown below: where E (surface) , E (cluster) , and E (slab) are the energies for the clean surface, the Au/Pt cluster in the gas phase, and the adsorbed system, respectively.The more negative the E ads , the stronger is the binding to the surface.

Figure 1 .
Figure 1.a) Schematic illustration of the preparation of TiO 2 SENSs for Au sputtering.Step 1: Emulsification of TiO 2 colloid and dichloromethane for the formation of TiO 2 covered emulsion droplets.Step 2: Formation of TiO 2 2D arrays by coalescence of emulsion droplets.Step 3: Stabilization of TiO 2 2D arrays by depositing them on a polystyrene film.

Figure 2 .
Figure 2. a) Optical images of TiO 2 colloid (top left) and a TiO 2 SENS (top right and bottom).b) XRD patterns of TiO 2 NPs compared to a TiO 2 SENS.c) A plan view of the surface of a TiO 2 SENS imaged using SEM showing that TiO 2 NPs are densely packed and exposed to the environment.d) A tilted view of a rare crack on the surface of a TiO 2 SENS showing that there is only a thin layer of TiO 2 NPs attached to the polymer support.Scale bars in (c) and (d) correspond to 250 nm.

Figure 4 .
Figure 4. a) The change in the UV-vis spectrum of MO during photodegradation.b) Kinetic plots showing the photodegradation of MO catalyzed by TiO 2 SENSs sputtered with Au for the indicated times from 0 to 30 s. Histograms showing the rate constants for photodegradation of MO by TiO 2 SENSs sputtered with c) Au and d) Pt for the times indicated.e) Photocatalytic mechanism of Au (x) -TiO 2 under UV light irradiation.
−2 kPa and a constant current at 30 mA. Preparation of Au/Pt TiO 2 SENSs via Photo-Deposition: Au NPs were synthesized by illuminating the surface of a 5 cm 2 TiO 2 SENS immersed in 2 mL of 10 −3 m HAuCl 4 .3H 2 O aqueous solution using a UV lamp with a peak wavelength at 365 nm and an overall output of 1.05 W. The distance from the lamp to the surface of the TiO 2 SENS was kept constant at 5 cm.Degradation of MO/4-chlorophenol under UV Light: In a typical experiment, an Au/Pt deposited TiO 2 SENS was mounted on Sellotape and immersed in 10 mL of 1 × 10 −5 m MO (or 1 × 10 −4 m 4-chlorophenol) solution in the dark for1 h to reach adsorption equilibrium.The sample solution containing the catalytic SENS was then irradiated continuously using a UV-LED (forward voltage of 3.5 V DC; forward current of 0.3 A; overall power of 1.05 W) under constant stirring.The wavelength of the LED was centered at 365 nm.2.5 mL of solution sample was taken every 15 (60) min to be analyzed using a Perkin Elmer Lambda 800 UV-vis spectrometer to monitor the degradation of MO (4-chlorophenol).

Figure 5 .
Figure 5. a) Kinetic plots showing the photocatalytic degradation of MO (100 mw, 365 nm UV lamp) catalyzed by Au (15) -TiO 2 SENS over 4 reaction cycles.b) XPS spectra of Au (15)-TiO 2 film that has been irradiated for repeated reaction cycles over 2 h, showing that there is no change in Au oxidation state after repeated use.