UV and visible light photocatalytic activity of Au/TiO2 nanoforests with Anatase/Rutile phase junctions and controlled Au locations

To magnify anatase/rutile phase junction effects through appropriate Au decorations, a facile solution-based approach was developed to synthesize Au/TiO2 nanoforests with controlled Au locations. The nanoforests cons®isted of anatase nanowires surrounded by radially grown rutile branches, on which Au nanoparticles were deposited with preferred locations controlled by simply altering the order of the fabrication step. The Au-decoration increased the photocatalytic activity under the illumination of either UV or visible light, because of the beneficial effects of either electron trapping or localized surface plasmon resonance (LSPR). Gold nanoparticles located preferably at the interface of anatase/rutile led to a further enhanced photocatalytic activity. The appropriate distributions of Au nanoparticles magnify the beneficial effects arising from the anatase/rutile phase junctions when illuminated by UV light. Under the visible light illumination, the LSPR effect followed by the consecutive electron transfer explains the enhanced photocatalysis. This study provides a facile route to control locations of gold nanoparticles in one-dimensional nanostructured arrays of multiple-phases semiconductors for achieving a further increased photocatalytic activity.

When compared with nanoparticles, one-dimensional (1D) semiconducting nanostructures such as nanorods and nanowires have been found to enhance the photocatalytic activity of TiO 2 with distinct charge transport capabilities 30,31 . Imposing branches on the surface of well-aligned 1D nanostructures to form branched nanowire arrays, also termed as nanotrees or nanoforests, further increases the performance because of the enhanced active sites and light harvesting capability [31][32][33][34] . In practical photocatalysis for wastewater treatments, TiO 2 thin films avoid the nuisance powder recovering procedure. It is thus of great importance to develop TiO 2 thin films with high photocatalytic activity. Based on the literature, it can be anticipated that decorating noble metals on controlled locations in TiO 2 nanoforests could achieve enhanced photoelectrochemical performances.
In this work, an architectural design was implemented to study the possible effects of the location of Au nanoparticles decorated on TiO 2 nanoforests with anatase/rutile phase junctions. Simply altering the order of the fabrication step ( Fig. 1), Au nanoparticles were located preferably either on the interface of anatase/rutile phase junctions, or on rutile branches, which in turn affected readily the photocatalytic efficiency towards photodegradations of rhodamine B in water under the illumination of either UV or visible light. Efforts were made to clarify the possible mechanisms that cause the distinct photocatalytic activity.

Results and Discussions
Morphology and phase characterizations. Supplementary Figure S1 shows FESEM images of the TiO 2 nanoforests and those after Au-loading at various locations. It can be seen that quasi-aligned 1D branched nanowires covered homogeneously the Ti substrates. The Au-loading procedures induced no significant change in morphologies. Figure 2 illustrates FESEM, TEM, HRTEM, selected area electron diffraction (SAED), and EDS analysis results of the Branched-Au-NW (refer to Fig. 1 for the sample ID). The thickness of the TiO 2 nanoforests is ca. 1 μ m. An intermediate layer ca. 1 μ m in thickness, which consisted of compact nanoparticles, can also be seen between the top layer and the substrate (Fig. 2a-c).
The low-magnification TEM image shows clearly a typical branched nanowire that is decorated with Au nanoparticles (Fig. 2e). The average diameter of the nanorod branch is ca. 10 nm and the length is ca. 45 nm; while the Au particles have an approximately spherical shape that is 8-9 nm in diameter. The HRTEM image (Fig. 2d) exhibited that the backbone is formed by many tiny crystal grains with a lattice spacing of ca. 0.35 nm, which can be attributed to the (101) facet of anatase TiO 2 . The fringe with inter-plane spaces of ca. 0.32 nm on the branch can be discerned, which is attributed to the (110) crystal plane of rutile TiO 2 . Locating between the anatase backbone and the rutile branch, fringes with inter-plane spaces of ca. 0.235 nm can be seen clearly, which can be assigned to the (111) facet of Au. Hence, the HRTEM observations suggest that, for the Branched-Au-NW specimen, the well-crystalized single-crystalline rutile nanorods grew along the polycrystalline anatase TiO 2 nanowires with an arbitrary angular orientation; whilst most Au nanoparticles located at the interface between the backbone and the branch, which contacted closely with both anatase and rutile. Figure 2f demonstrates the SAED pattern of the branched nanowire shown in Fig. 2e. Multi-rings characteristic of a mixture of anatase and rutile can be discerned, which is in good accordance with the HRTEM observations (Fig. 2d). The EDS mapping (Fig. 2g) suggests the homogenous distributions of Ti, O, N, and S throughout the Branched-Au-NW. A spot corresponding to an Au nanoparticle located between the anatase trunk and rutile branch can be discerned.
The crystal structure of the TiO 2 nanoforests can find support from both XRD patterns and Raman spectra. XRD peaks corresponding to both anatase and rutile can be discerned, besides those arising from the metallic Ti substrates (Fig. 3a). It is not surprising that three specimens exhibited similar XRD patterns, indicating that the phase structure of TiO 2 does not change after the deposition of gold nanoparticles. Trace srilankite TiO 2 can also be discerned. The corresponding Raman spectrum suggests more clearly the coexistence of anatase, rutile, and srilankite (Fig. 3b). Both XRD (Fig. 3a) and SAED patterns (Fig. 2f) exhibit no signals corresponding to Au, which can be contributed to the minor amounts of Au nanoparticles. It is noted that only one peak appeared in both XRD patterns and Raman spectra, which is a weak proof of the existence of srilankite TiO 2 . However, a previous study 35 revealed that, the srilankite TiO 2 was detected on a low temperature derived TiO 2 thin film. In that case, Raman peaks were identified at 168, 314, 354 and 425 cm −1 , which have been contributed to srilankite TiO 2 . Also, the XRD peak located at ca. 31.5° was identified. Therefore, we believe it is safe to ascribe the present XRD peak located at ca. 31.5° (Fig. 3a) and the Raman peak located at ca. 314 cm −1 (Fig. 3b) to srilankite TiO 2 .
Au contents. The atomic ratio of Au/Ti was measured by ICP-MS to be 0.36% and 0.51%, for the specimens of Branched-Au-NW and Branched-NW-Au, respectively. By using the defined Au-loading parameters (identical HAuCl 4 solution and photo-reduction time), the Branched-Au-NW contained less Au nanoparticles when compared with the Branched-NW-Au. This may be contributed to the different phase compositions of the TiO 2 films before Au-decorations. For the Branched-Au-NW, Au-loading was performed on poorly crystallized anatase TiO 2 nanowires; whilst for the Branched-NW-Au, Au-loading was conducted on TiO 2 nanoforests consisted of rutile TiO 2 branches that grow radially around the poorly crystallized anatase TiO 2 trunk 34 . It is well established that the anatase/rutile phase junctions facilitate charge separations [24][25][26] and hence the photocatalytic efficiency for the Au-loading, which explained the higher content of gold nanoparticles decorated on Branched-NW-Au. It is also argued that Au nanoparticles form more easily on the rutile surface because a number of oxygen vacancies act as the crystal nucleation sites 13 .
The surface compositions of the Branched-Au-NW were analyzed by XPS. Figure 4a shows that the surface layer of the Branched-Au-NW is composed of Ti, O, Au, N, and S, which is consistent to the EDS mapping images (Fig. 2g). The XPS spectrum of Ti 2p exhibits two dominant peaks, which correspond to Ti 2p 1/2 at 464.6 eV and Ti 2p 3/2 at 458.8 eV, indicating that Ti exist in the form of Ti 4+ . The separation between the two peaks is 5.8 eV (Fig. 4b), which is in agreement with the XPS data in the literature 7 . As shown in Fig. 4c, the O 1 s spectrum can be fitted by two components: a higher binding energy (BE) peak near the 531.7 eV, originating from the hydroxyl group (-OH) 36 , and a lower BE, in the vicinity of 530.1 eV, originating from the crystal lattice oxygen (Ti-O-Ti).
The XPS spectrum of Au can be separated to two peaks, with a lower BE at 83.4 eV and a high BE at 87.1 eV, corresponding to Au 4f 7/2 and Au 4f 5/2 , respectively (Fig. 4d). The typical Au 4f 7/2 peak locates at 84.0 eV 37 for bulk metallic gold; the slight shift in BE to a lower value can be ascribed to the redistribution of the electrons at the Au-TiO 2 contact interfaces, because of the difference in the work function between Au (5.27 eV) and TiO 2 (4.1 eV). This is an indication that Au nanoparticles interact with the adjacent TiO 2 14,37,38 . The electron transferring from TiO 2 to Au nanoparticles is therefore facilitated, which increases the valence charge density of Au atoms and reduces the binding energy of Au in the TiO 2 film. The binding energy of N 1 s locates at ca. 399.8 eV, which can be assigned to nitrogen species bonding to various surface oxygen sites N-O, or N-N, and N-C bonds ( Fig. 4e) 39 . The incorporation of N into the Branched-Au-NW film is believed to result from the decomposition of melamine during the fabrication of the titanate nanowires. After the H 2 SO 4 treatment, sulfate ions also incorporate into the TiO 2 film, which gave the XPS peak at 168.9 eV (Fig. 4f) 36 . Table 1 lists the surface compositions of the two Au-decorated TiO 2 nanoforests derived by the XPS analysis. The atomic Au/Ti ratio determined by the XPS analysis is ca. 0.27% for Branched-Au-NW, which is lower than that of Branched-NW-Au. The atomic Au/Ti ratios evaluated by XPS roughly agreed with that derived by the ICP-MS measurement.
Growth of the TiO 2 Nanoforests with Controlled Au-location. Figure 5a,b shows that, Au nanoparticles were intimately decorated on the surface of the poorly crystallized anatase TiO 2 nanowires before the final H 2 SO 4 treatment. That is to say, in the current tactic to synthesize Branched-Au-NW, Au nanoparticles were firstly deposited on the surface of the nanowires which were just subjected to the intermediate calcination. During the final H 2 SO 4 treatment, the poorly crystallized anatase TiO 2 nanowires were partly attacked by H 2 SO 4 , which released hydrated Ti(IV) ions into the acid solution 34 . Once the Ti(IV) ions accumulated to a critical concentration, nucleation and subsequent growth of rutile TiO 2 branches around the anatase TiO 2 trunk occurred. It seems that the edges formed between the nanowires and the decorated Au nanoparticles provide heterogeneous nucleation sites for the TiO 2 branches, which resulted in the preferred locations of the Au nanoparticles around the junctions of the anatase trunk and rutile branch.
When the Au-decoration procedure was moved to be the final step, the Au nanoparticles did not locate preferably on the anatase/rutile phase junctions any more. Because of certain "tip" effects, most of Au nanoparticles located on the surface of the rutile branch (Fig. 5c,d), rather than on the interface of rutile branch and anatase backbone. In addition, for the Branched-NW film, because the anatase trunk is surrounded by rutile branches, Au nanoparticles had a higher possibility to sit on rutile. Therefore, simply altering the Au-loading order fulfilled the control in the location of Au nanoparticles, which affects readily the resultant photocatalytic performance of the Au-decorated TiO 2 nanoforests, as will be discussed later.  PL and UV-Vis DRS Characterizations. Figure 6 shows the PL spectra of the three specimens, all of which possessed a feature that consists of two emission peaks in the UV-visible range. The UV emission centered at 400 nm is related to the electron transition from the valence band and the conduction band 38 , and the emission centered at around 608 nm may arise from the recombination of photo-generated holes with the electrons in singly occupied oxygen vacancies 40 . It can be seen that, the intensity of UV emission decreased in the order of Branched-NW, Branched-NW-Au, and Branched-Au-NW. Therefore, the Au-decoration suppresses the charge recombination, which is closely related to the location of Au nanoparticles. Figure 7a illustrates the UV-Vis diffuse reflectance spectra collected from the three specimens. The Branched-Au-NW film exhibited the lowest reflectance over the visible light region, which suggests the highest visible light harvesting capability when compared with the Branched-NW and the Branched-NW-Au. The increasing absorption can be ascribed to the LSPR effect arising from Au nanoparticles. A strong oscillation of the metal's surface free electrons with the varying electric field of the incident light absorbs the energy of photon  and conveys to electrons to form surface plasmons, which decay to hot electron-hole pairs; as a result, the light response of the Au-decorated specimens in the visible light region is enhanced 41,42 .
Assuming an indirect transition between band gaps, the band gaps of TiO 2 can be estimated by extrapolating the tangent line in the plot of α 1/2 against hυ 43 , where α is the absorption coefficient and the hυ is the photon energy. Figure 7b demonstrates that the Branched-Au-NW possessed an indirect band gap of 2.61 eV, which is lower than the value of 2.64 eV and 2.80 eV determined for the Branched-NW-Au and Branched-NW, respectively. The relatively lower band gap of 2.80 eV for Branched-NW, when compared with bulk TiO 2 (3.2 eV for anatase and 3.0 eV for rutile), can be contributed to the N-doping (Fig. 4e) and the significant oxygen deficiency (Fig. 4c) arising from the low-temperature synthesis route. A DFT calculation by Jia et al. revealed that, the band gap of a N, S-codoped TiO 2 can be narrowed to be. 2.77 eV 44 . The further red-shift for the Au-decorated films was attributed mainly to the interaction of Au and TiO 2 , which might introduce an intra-gap level inside the band gap of TiO 2 37 . Considering the relatively lower Au content for the Branched-Au-NW (0.36%) when compared with the Branched-NW-Au (0.51%), as determined by ICP-MS, both the higher light harvesting capability and the slightly lower band gap once again convinced the importance of the Au-location. A rough explanation for the enhanced LSPR arising from Au nanoparticles in Branched-Au-NW could be that the contacting surface area between Au and TiO 2 is higher than that in Branched-NW-Au.
Photocatalytic Activity. Photocatalytic activities of the three TiO 2 nanoforests were evaluated by decomposing rhodamine B in water under UV and visible light illuminations, respectively. In absence of any photocatalysts, about 95% and 97% dye molecules remained after 60 min of UV and visible light illuminations. The dark absorption capacity is enhanced after the Au-decoration, which can be ascribed to the interaction of Au and dye molecules. For comparison purpose, thin films of Degussa P25 TiO 2 nanoparticles (ca. 3.0 μ m in thickness, refer to the literature 45 for the fabrication route), which are generally adopted as a benchmark, were also subjected to the photocatalytic activity evaluations under the identical conditions. The HAuCl 4 concentrations adopted for the Au-decoration were firstly optimized (Supplementary Figure S2). It illustrates that, for both Au-decorated films, there is an optimum HAuCl 4 concentration (0.040 mM). It can thus be inferred that, certain amounts of Au nanoparticles introduced to the TiO 2 nanoforests enhanced the photocatalytic activity. However, excess Au nanoparticles aggregate to serve as recombination centers for photogenerated charges, leading to an inferior performance 46 . Figure 8a,b indicates the photodegradation curves, which can be fitted well assuming a pseudo-first-order kinetics 47 where c is the dye concentration after illumination for a duration t, c 0 is the dye concentration after the dark adsorption, and k is the pseudo-first-order reaction rate constant, which can be obtained by the slope of the straight lines through zero. Figure 8c,d shows the corresponding fitting results and Table 2 lists the reaction rate constants, which were derived using the average data obtained from three repetitive tests. Under the UV light illumination, the reaction rate constant increased from 0.86 to 2.5 × 10 −2 min −1 after the Au-decoration of the Branched-NW. Simply controlling the location of the Au nanoparticles to distribute mainly along the anatase/ rutile phase junctions, the reaction rate constant further increased to 4.7 × 10 −2 min −1 , which is nearly 5 times higher when compared with the Branched-NW specimen. The Au location affects also the photocatalytic activity under the visible light illumination. Photocatalytic degradations of p-nitrophenol and phenol under the UV light illumination, in the presence of the various Au/TiO 2 films were also evaluated. Figure 9 and Table 2 show that, the same trend can be discerned. Therefore, it can be concluded that, the photocatalytic activity of the branched TiO 2 film was enhanced after the Au decoration and the position of Au nanoparticles really has a great impact on the photocatalytic activity. Due to the insufficient efficiency of the present Au/TiO 2 films under the visible light illumination, the photocatalytic degradations of p-nitrophenol and phenol under visible light is not presented in the current investigation. Photosensitive materials are argued to be not suitable as probe chemicals for photocatalytic activity tests, especially those for evaluation of activity under visible light 48 . Further study is thus demanded to disclose the effects of Au-locations on photodegradations of organics besides rhodamine B molecules.
The trapping experiment was employed to disclose the possible photocatalysis mechanism (Supplementary Figure S3). It is inferred that OH• , h + , and O 2 • − all contribute to the photocatalytic reaction. The superoxide radicals O 2 • − are the major active species responsible for this photocatalytic oxidation reaction, followed by the photogenerated holes h + , and the hydroxyl radical OH• . Figure 10 provides a possible explanation for the beneficial effects arising from Au nanoparticles located between the anatase/rutile phase junctions in the present 1D branched TiO 2 nanowires. Because Supplementary Figure S3 indicates that superoxide radicals O 2 • − contribute the most to the photocatalytic reaction; only the degradation route via O 2 • − is presented in Fig. 10. It is well established that anatase/rutile phase junction effectively suppresses the recombination of photogenerated electron-hole pairs 30,[49][50][51] ; however, the exact charge transfer direction still remains a controversy 25,52 . Herein, the energetic alignment of the band edges of the anatase and rutile polymorphs of TiO 2 suggested recently by Scanlon et al. is adopted 52 . Fig. 10a, under UV light irradiation, photogenerated electrons transfer from rutile to anatase, which results in an enhanced charge separation rate and hence improved photocatalytic activity. In addition, it has been proved that O 2 reduction by the photo-induced electrons on the rutile surface is inefficient because of the low affinity between the surface and O 2 . In contrast, anatase is more active for O 2 reduction 13 . As a result, the electron transfer from rutile to anatase accelerates the photodegradation procedure, especially for which superoxide radicals O 2 • − play the key role (Supplementary Figure S3).

UV Light Illumination. As illustrated in
In case that the TiO 2 nanoforests were subjected to a final Au-decoration (Branched-NW-Au), Au nanoparticles prefer to locate on rutile surfaces (Fig. 5c,d). Because of the Shottky barrier between Au and TiO 2 semiconductor, the photoexcited electrons from rutile will also transfer to Au nanoparticles 13,14 , except for the migration to anatase. Thus, Au nanoparticles act as the electrons accepting species at the Au/rutile interface, suppressing further the recombination of photogenerated charges (Fig. 10c). The reaction rate constant thus increased from 0.86 to 2.5 × 10 −2 min −1 (UV + RhB, Table 2). When Au nanoparticles are loaded on the interface between anatase and rutile in the TiO 2 nanoforests, as for the Branched-Au-NW, they provide a path for rapid and effective electron migrations from rutile to anatase 53 , which significantly enhances the positive effects arising from the anatase/rutile phase junction (Fig. 10e). Such a function is more effective for the charge separation, contributing to an even higher reaction rate constant of 4.7 × 10 −2 min −1 (UV + RhB, Table 2). The enhanced charge separation is supported by the PL measurement (Fig. 6).    which corresponds to a wavelength of ca. 440 nm (Fig. 7); therefore, under the illumination of visible light (> 420 nm), it is possible for the rutile branch to absorb the photons with appropriate energy to initiate the photocatalytic reaction (Fig. 10b). The additional electrons arising from the LSPR effect of Au nanoparticles, which may also transfer to anatase to initiate the photodegradation reaction there (Fig. 10d), explains the increased reaction rate constant from 0.57 × 10 −2 min −1 for Branched-NW to 1.9 × 10 −2 min −1 for Branched-NW-Au (Vis + RhB, Table 2). The further enhanced reaction rate constant for Branched-Au-NW (2.3 × 10 −2 min −1 ) is in good accordance with Tsukamoto et al. 13 . Here, Au nanoparticles locate preferably at the interface of anatase/rutile, upon the visible light illumination, the LSPR induced electrons transfer from Au nanoparticles to the tightly bound rutile and then through the phase junction to well-conjugated anatase, where the electrons react with surface adsorbed O 2 to form O 2 • − to assist photodegradations of rhodamine B in water (Path 2, Fig. 10f) 13,14 . Because Au nanoparticles contact with both anatase and rutile (Fig. 2d), we believe that, photogenerated electrons arising from the LSPR effect may also transfer directly to the adjacent anatase, which additionally contribute to the enhanced photocatalytic activity (Path 1, Fig. 10f).
Cycling Performance and mineralization capability. Long-term stability is a major concern of photocatalysts. Figure 8e shows the cycling performance of the Branched-Au-NW film. For up to 10 cycles, no remarkable decay can be discerned, which evidenced the excellent stability of the present Au-decorated TiO 2 nanoforests. The present Branched-Au-NW film is also capable of inducing deep mineralization of organics in water after UV illumination for certain durations. A total organic carbon (TOC) reduction of ca. 64.2% was achieved for the rhodamine B solution after the UV light illumination for 1 h when assisted by the Branched-Au-NW film. The liquid chromatography (LC) spectra (Supplementary Figure S4), which were utilized to determine the phenol's concentration during the photodegradation procedure under the UV illumination, shows that, although the P25 film has a higher efficiency on photodegradations of phenol, less by-products can be discerned for the Au/TiO 2 photocatalyst (Branched-Au-NW). This further supports the capability of the present Au/TiO 2 photocatalyst to achieve a deep mineralization of organics in water.

Conclusion
Titania nanoforests, which consisted of anatase nanowires surrounded by radially grown rutile branches, were synthesized on metallic Ti substrates through multi-steps of H 2 O 2 oxidation, intermediate calcination, and sulfuric acid treatment. A photo-reduction technique was then applied to decorate Au nanoparticles. Altering the order in the fabrication step is effective in controlling the preferred location of Au nanoparticles. Under the UV light illumination, Au nanoparticles, which located preferably at the interface of anatase/rutile, magnify the beneficial effects arising from the anatase/rutile phase junctions; whilst under the visible light illumination, the LSPR effect followed by the consecutive electron transfer results in improved photocatalysis. This finding evidenced the feasibility to further improve photocatalytic activity of multiple-phases semiconductor arrays with one-dimensional nanostructures through carefully controlling the location of noble metals.

Methods
Synthesis of TiO 2 Nanoforests. Figure 1 demonstrates schematically the three steps in sequence to synthe-

Decoration of Au Nanoparticles at Controlled Locations. Au nanoparticles were decorated on
Branched-NW via a photo-reduction method 54 . 0.44 mg of HAuCl 4 (corresponding to an initial concentration of ca. 0.040 mM), methanol (0.2 mL), and deionized water (34 mL) were mixed in a beaker. Branched-NW was immersed into the solution, which was bubbled with pure nitrogen gas, and then subject to the UV light (5.0 mW/cm 2 ) irradiation for 5 min. The specimens were dried at 80 °C in an oven overnight, which were designated as Branched-NW-Au hereafter. To vary the location of the Au nanoparticles, in a parallel experiment, the order in Au-deposition and H 2 SO 4 treatment was exchanged. The specimens were noted as Branched-Au-NW.
Characterizations. Surface morphology of thin films was observed using a field emission scanning electron microscopy (FESEM, Hitachi S-4800) and a transmission electron microscopy (TEM, FEI, Tecnai G2 F20 S-TWIN, with EDS capabilities). To prepare the samples for TEM characterizations, the TiO 2 nanoforests were detached off from the Ti plate and then placed on a carbon pre-coated copper grid. The X-ray diffraction (XRD) tests were conducted on a Rigaku D/max-3B diffractmeter with Cu Kα radiation (λ = 0.154056 nm), operated at 40 kV, 36 mA. Raman spectra were collected using an Almega dispersive Raman system (Nicolet) and a Nd:YAG intracavity doubled laser operating at 532 nm with an incident power of 10 mW. The atomic ratio of Au and Ti was measured with inductively coupled plasma mass spectrometry (ICP-MS, XSeries II, Thermo Fisher Scientific, USA). For the ICP-MS measurement, the Au-decorated TiO 2 nanoforests were scratched off and dissolved in aqua regia. The X-ray photoelectron spectroscopy (XPS) spectra were collected using an ESCA spectrometer (S-Probe ESCA SSX-100S, Fisons Instrument) and monochromatized Al Kα X-ray (1486.8 eV) irradiation. The binding energy was normalized to the C 1 s energy (284.6 eV) for adventitious hydrocarbons as the indirect standard. The UV-Vis diffuse reflectance spectra were measured using a UV-Vis near-infrared spectrometer (UV-3150, Shimadzu). The ambient photoluminescence (PL) emission spectra characterizations were carried out on a fluorescence spectrophotometer (HITACHI F-4500) with an excited wavelength of 360 nm.
Photocatalytic Activity Evaluations. Titania thin films (2.5 × 2.5 cm 2 in size) were used to assist photocatalytic degradations of rhodamine B, p-nitrophenol and phenol in water (25 mL) in a Pyrex reactor with a water jacket, which is illustrated schematically in Supplementary Figure S5. Three repetitive tests were conducted to give the average value and error bar (standard deviation) as illustrated in the corresponding degradation curves. The initial concentration of rhodamine B is 0.005 mmol/L, while the initial concentrations of p-nitrophenol and phenol are 5 ppm and 10 ppm, respectively. The UV irradiation was provided by an 18 W UV lamp and the visible light by a 500 W Xe-lamp with a 420 nm UV-cut filter. The UV and visible light intensity reaching the sample was measured to be ca. 5.0 mW/cm 2 and 200.0 mW/cm 2 , respectively, using irradiance meters (model: UV-A and FZ-A, Beijing Normal university, China, measured for the wavelength of 365 nm for UV light and 400-1000 nm for the visible light). The solution was stirred and exposed to air during the photocatalytic reaction. The change in the rhodamine B and p-nitrophenol concentrations was monitored with a UV-Vis spectrophotometer (UV-1800PC, Shanghai Mapada, Shanghai, China) at a wavelength of 553 nm and 317 nm, respectively. The phenol concentration was analyzed by liquid chromatography on a Wufeng LC100 (WondaCract C18-ODS column, and 50% CH 3 OH aqueous solution as an eluent). The total organic carbon (TOC) of rhodamine B was measured with a Total Organic Carbon Analyzer (Multi N/C 3100).
To investigate the effects of the active species generated during the photocatalytic reaction under the UV illumination and in the presence of the Branched-Au-NW, the free radicals capture experiments were conducted 55 . The three major oxidants involved in photodegradations of organics in water, that is, hydroxyl radical (OH• ), hole (h + ), and superoxide radical (O 2 • − ), were trapped by adding 0.5 mL of tert-butanol (t-BuOH), 0.1 mM of ammonium oxalate (AO), and 0.5 mM of 1, 4-benzoquinone (BQ), respectively, into the rhodamine B solutions.