Volatile organic compounds (VOCs) are generally harmful pollutants that can greatly damage people's health [1]. Photocatalysis is regarded as an effective way to decompose the VOCs into CO₂, H₂O, and other harmless compounds [2]. Nanostructured photocatalytic materials have attracted much interest in recent years because of their large surface area and specific morphologies [3, 4]. In particular, TiO₂ nanomaterials are promising photocatalysts because of their chemical inertness, non-toxicity, and low cost, and have been studied extensively in the past decade [5-8]. However, the wide band gap (E_g) [9, 10] of TiO₂ means that it only responds to ultraviolet (UV) light, which accounts for just 5% of the total solar energy [11-13]. In addition to TiO₂, other titanium compounds such as alkali metal titanates have also drawn some attention for use in photocatalysis [14-16]. Titanate materials are easy to synthesize and display excellent chemical and mechanical properties. Hydrothermal treatment of TiO₂ powder in a concentrated alkaline solution is a commonly used approach to prepare alkali metal titanates with nanowire and nanotube morphologies [9, 17]. Such materials display obvious photocatalytic activity under UV illumination [18]. For example, it has been reported that Na₂Ti₆O₁₃ materials can decompose 4-chlorophenol under UV illumination [19, 20]. Like TiO₂, these materials respond to UV light because of their large E_g. Other interesting technological applications for Na₂Ti₆O₁₃ such as capacitors, dielectric sensors, and biosensors have been also reported [21].

Through coupling with a noble metal, such as Au, Pt, or Ag, TiO₂ and alkali metal titanates can make use of visible light in photocatalysis by localized surface plasmonic resonance (LSPR) absorption [22-26]. For example, TiO₂ showed a notable photocurrent response in the visible light region after being coupled with Ag nanoparticles because of the LSPR effect [27-30]. Deposition of Au nanoparticles on titanate nanobelts also resulted in obvious visible-light photocatalytic activity [31]. In addition to noble metals, loading of oxide co-catalysts onto TiO₂ can also increase its photocatalytic activity via surface electronic structure coupling [32, 33]. Amorphous copper oxide (Cu_xO) and copper hydroxide (Cu(OH)_x) nanoclusters have attracted much interest as co-catalysts. The deposition of Cu²⁺ nanoclusters onto Na₂Ti₃O₇ greatly increased its photocatalytic activity in the oxidation of cyclohexa-1, 4-diene [34]. This is because using Cu²⁺ nanoclusters as surface active sites can effectively enhance the photocatalytic hydrogen generation of TiO₂. Irie et al. [35] reported that interfacial charge transfer (IFCT) between Cu²⁺ and rutile TiO₂ could induce the photocatalytic oxidation reaction, resulting in decomposition of gaseous isopropanol or acetaldehyde. However, it was also found that Cu²⁺ nanoclusters might increase charge recombination and resulted in decreased photocatalytic activity [36]. Choi and co-workers reported that photoinduced electron transfer to grafted Cu(Ⅰ) or Cu(Ⅱ) might lead to the in situ formation of metallic Cu, which can act as an effective co-catalyst [37, 38].

The above photocatalyst materials mainly exist in the form of ultrafine powders, which are usually difficult to separate and recycle after photocatalysis, inevitably leading to secondary pollution during their practical application. Photocatalytic materials that are easily separated and recycled are needed. In the present research, a novel membrane composed of ultralong sodium titanate nanofibers is prepared by a modified hydrothermal reaction and used to remove indoor VOCs. Both Au nanoparticles and Cu(Ⅰ) nanoclusters are loaded on the surface of the sodium titanate membrane to access LSPR absorption and accelerate electron-hole separation, respectively. The obtained Au-Cu(Ⅰ)@Na₂Ti₆O₁₃ hybrid membrane presents high photocatalytic activity under visible- light illumination. The membrane resists self-aggregation, and can be easily separated and recycled. This self-standing membrane is attractive for air purification under visible-light illumination.

Tetrabutyl titanate (TBOT), ethanol, acetic acid, and polyvinylpyrrolidone K30 (PVP K30) were used to prepare the precursor materials for hydrothermal reactions. The precursor materials were synthesized by the electrospray method, as reported elsewhere [39]. Briefly, PVP K30 (0.3 g) was dissolved in ethanol (2.5 mL) and acetic acid (3.8 mL) by stirring for 10 min. TBOT (7.0 mL) was added to the mixture, which was then vigorously stirred for 1 h to obtain a yellow PVP@TBOT sol–gel. The PVP@TBOT sol was aged in the dark for 10 h. The PVP@TBOT sol was placed into a syringe pump and then automatically pumped through a metal needle at a fixed speed of 2.5 mL/h. A voltage of 15 kV was applied to the metal needle, causing the PVP@TBOT sol to atomize into very tiny droplets, which were sprayed on an aluminum target. After electrospraying, the aluminum target was dried in air for one day, and then the electrosprayed materials were collected from the aluminum target for further hydrothermal treatment.

Na₂Ti₆O₁₃ nanowires were synthesized in fresh NaOH solution by the hydrothermal method. Precursor (0.2 g) was added to a sealed Teflon reactor (volume: 50 mL) containing aqueous NaOH (30 mL, 10 mol/L). The stirred mixture was heated at 180 ℃ in an oil bath for 12 h. If the solution was not stirred during the hydrothermal reaction, no nanofibers were obtained. After cooling to room temperature, the hydrothermal products were washed with deionized water several times until the pH of the filtrate reached ~7. The sample was transferred into ethanol and vigorously stirred for 10 min. The mixture was heated at 80 ℃ to remove ethanol and obtain a membrane. The membrane was annealed in air at 450 ℃ for 2 h to increase its crystallinity.

Au nanoparticles and then Cu(x) nanoclusters were loaded on the membrane by thermal deposition using HAuCl₄ and CuCl₂·2H₂O, respectively. The membrane was immersed in 5 × 10^-3 mol/L HAuCl₄ aqueous solution at 70 ℃. The pH was adjusted to 6 using 0.2 mol/L NaOH solution. The immersion time was regulated to adjust the amount of Au deposited on the membrane. After immersion for a certain time, the sample was washed with deionized water several times, dried in air at 70 ℃, and then annealed at 300 ℃ for 2 h to generate Au nanoparticles. The resulting samples are denoted as Au(M)@Na₂Ti₆O₁₃, where M is the time immersed in HAuCl₄ solution (in minutes).

Na₂Ti₆O₁₃ membranes were also immersed in 1.60 × 10^‒3 mol/L CuCl₂ aqueous solution at 90 ℃. The immersion time was used to adjust the amount of Cu(x) loaded on the membrane. Each sample was washed with deionized water several times and then dried in air at 70 ℃. Finally, each sample was heated at 150 ℃ for 2 h to generate Cu(x) on the membrane. These samples are denoted as Cu(x)(N)@Na₂Ti₆O₁₃, where N is the time immersed in CuCl₂ solution (in minutes), and x is the valence state of Cu. In the same manner, each Au(M)@Na₂Ti₆O₁₃ membrane was immersed in 1.58 × 10^‒3 mol/L CuCl₂ solution to produce hybrid Au(M)-Cu(x)(N)@  Na₂Ti₆O₁₃ membranes.

Surface images of the pure Na₂Ti₆O₁₃, Au(5)@Na₂Ti₆O₁₃, Cu(x)(40)@Na₂Ti₆O₁₃, and Au(5)-Cu(x)(40)@Na₂Ti₆O₁₃ membranes were observed by field-emission scanning electron microscopy (FE-SEM; S-4800, Hitachi, Tokyo, Japan) and transmission electron microscopy (TEM; JEM2100F, JEOL, Tokyo, Japan). The Au and Cu atomic contents in the samples were determined by inductively coupled plasma (ICP) optical emission spectrometry (Prodigy 7, Leeman Labs, USA). X-ray diffraction (XRD) patterns were collected with a grazing- incidence X-ray diffractometer (Empyrean, PANalytical, Almelo, the Netherlands) using Cu K_α radiation as the X-ray source. The surface chemical compositions of the Au(5)@Na₂Ti₆O₁₃, Cu(x)(40)@Na₂Ti₆O₁₃, and Au(5)-Cu(x)(40)@Na₂Ti₆O₁₃ membranes were analyzed with an X-ray photoelectron spectrometer (XPS; VG Multilab 2000, Thermo Scientific, USA) with Al K_α radiation as the X-ray source. The binding peak of C 1s electrons at 284.6 eV was used as a reference. Ultraviolet- visible (UV-vis) diffuse reflectance spectra of samples were measured using a UV-Vis spectrophotometer (UV-3600, Shimadzu, Japan) in wavelength range from 200 to 800 nm.

The typical indoor air pollutant acetaldehyde was chosen as the target organic to evaluate the photocatalytic activity of the membranes. The photocatalysis experiments were performed in a transparent cylindrical glass reactor with a volume of 500 mL and quartz window. The light source was a high-power Hg lamp with a 420-nm cut-off long-pass filter to provide visible light. The membrane sample (200 mg) was placed in the sealed glass photocatalytic reactor, and the temperature was kept at 25 ℃. Before starting the photocatalytic experiments, fresh air with a relative humidity of 50% was flowed through the reactor for 10 min. Then, the reactor was irradiated with UV light overnight to remove organic pollutants on the sample surface. The reactor was then filled with fresh air. Acetaldehyde (120 ppm) was injected into the reactor and the reactor was placed in the dark until the concentration of acetaldehyde became stable. A gas chromatograph (GC-2014C, Shimadzu, Japan) was used to analyze the concentration of acetaldehyde during the photocatalytic reaction.

Fig. 1(a) shows the Ti 2p_1/2 and Ti 2p_3/2 spectra of pure Na₂Ti₆O₁₃, Au(5)@Na₂Ti₆O₁₃, Cu(x)(40)@Na₂Ti₆O₁₃, and Au(5)- Cu(x)(40)@Na₂Ti₆O₁₃ membranes. All of the Ti 2p spectra are symmetrical, indicating the +4 valence of Ti atoms. Because the work function of Na₂Ti₆O₁₃ is higher than that of Au, electrons will move from Na₂Ti₆O₁₃ to Au when they are connected. The decrease of electron density near Ti atoms leads to a positive shift of the Ti 2p spectra in the presence of Au, as shown in Fig. 1(a). In contrast, no obvious shift was observed when Na₂Ti₆O₁₃ was loaded with Cu(x), indicating that there was no charge exchange between these species. Fig. 1(b) displays the Au 4f_7/2 and Au 4f_5/2 spectra of the Au(5)@Na₂Ti₆O₁₃ and Au(5)-Cu(x)(40)@Na₂Ti₆O₁₃ membranes, which indicate that the Au nanoparticles are metallic. In the Cu 2p spectra (Fig. 1(c)), Cu 2p_1/2 and Cu 2p_3/2 peaks were observed at 952.5 and 932.5 eV, respectively. The 20 eV splitting of these peaks is consistent with the presence of Cu(Ⅰ). Considering that the characteristic shake-up satellite lines of Cu(Ⅱ) are absent, the presence of Cu(Ⅱ) can be excluded [40]. These results are consistent with the membrane fabrication process; lower temperature calcination aids the formation of Cu(Ⅰ) species, whereas higher temperature treatment leads to the formation of Cu(Ⅱ) species [41].

The XRD patterns of the pure Na₂Ti₆O₁₃, Au(5)@Na₂Ti₆O₁₃, Cu(Ⅰ)(40)@Na₂Ti₆O₁₃, and Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membranes are shown in Fig. 2. The pure Na₂Ti₆O₁₃ membrane presents a pattern consistent with the main phase of Na₂Ti₆O₁₃. The diffraction peak at 25.1° could be assigned to the (110) lattice plane of Na₂Ti₆O₁₃ or (100) lattice plane of anatase TiO₂. The diffraction peak at 54.3° could originate from the (514) lattice plane of Na₂Ti₆O₁₃ or (105) lattice plane of anatase TiO₂. However, according to the TEM analysis (Fig. 4), only single-phase Na₂Ti₆O₁₃ was observed. Therefore, the above two peaks are assigned to the diffractions of the Na₂Ti₆O₁₃ phase because there is no anatase TiO₂ in the Na₂Ti₆O₁₃ membrane. No diffraction peaks from Cu metal and its oxides were observed in the XRD patterns of the Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ and Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membranes because of the low amount of loaded Cu(Ⅰ). The XRD patterns of the Au(5)@Na₂Ti₆O₁₃ and Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membranes display diffraction peaks at 38.2°, 66.4°, and 77.5°, which are assigned to the (111), (220), and (311) lattice planes of metallic gold, respectively. The XRD patterns indicated that loading Cu(Ⅰ) on the Na₂Ti₆O₁₃ membrane surface greatly decreased the amount of Au nanoparticles loaded previously.

Fig. 3(a) shows that the electrosprayed precursor materials possess dense spherical morphology with rough surfaces. A digital photograph of the as-prepared pure Na₂Ti₆O₁₃ membrane with a diameter of ca. 10 cm is illustrated in Fig. 3(b). Fig. 3(c) reveals that the membrane consists of ultralong Na₂Ti₆O₁₃ nanofibers. The large pores between the nanofibers allow air to flow through the membrane, so it can be used as a support for photocatalytic air purification. Fig. 3(d) indicates that the nanofiber surfaces of the Au(5)@Na₂Ti₆O₁₃ membrane are covered with gold nanoparticles with an average size of ~10 nm. Meanwhile, Fig. 3(e) shows that the nanofibers do not show change obviously after being loaded with Cu(Ⅰ) to form the Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane, which may be because of the small size of the Cu(Ⅰ) nanoclusters. Fig. 3(f) reveals that the Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane contained less Au nanoparticles on the nanofiber surface than that of the Au(5)@Na₂Ti₆O₁₃ membrane. This shows that some of the Au nanoparticles were removed during the loading of the Cu(Ⅰ) nanoclusters, which is in accordance with the results of XRD analysis. The loading of Cu(Ⅰ) and Au was also observed by high-resolution TEM, as shown in Fig. 4(a) and (b). Many spherical Au nanoparticles were present on the Na₂Ti₆O₁₃ nanofiber surface. Fig. 4(b) reveals that the Cu(Ⅰ) nanoclusters had a size of ~2.5 nm. Considering the low-temperature deposition process used to load the Cu(Ⅰ) nanoclusters, the above results indicate that the Cu(Ⅰ) nanoclusters are composed of amorphous Cu₂O and Cu(OH). The TEM results revealed the small size of the Cu(Ⅰ) nanoclusters, supporting the conclusion that Cu(Ⅰ) is more stable than Cu(Ⅱ) when the particle size is very small [36]. The actual loadings of Au and Cu(Ⅰ), Au nanoparticle size, and Cu(Ⅰ) nanocluster size of Au(5)@Na₂Ti₆O₁₃, Cu(Ⅰ)(40)@Na₂Ti₆O₁₃, and Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ are summarized in Table 1.

Fig. 5 shows the UV-vis diffuse reflectance spectra of pure Na₂Ti₆O₁₃, Cu(Ⅰ)(40)@Na₂Ti₆O₁₃, Au(5)@Na₂Ti₆O₁₃, and Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membranes. Au and Cu(Ⅰ) loading had clear effects on the absorption spectra of the membranes. The strong absorptions from 300 to 400 nm are caused by the band-to-band transition of Na₂Ti₆O₁₃. The small amount of anatase phase present in the materials may also contribute to this absorption. For the Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane, the grafting of Cu(Ⅰ) increased the absorption intensity in the regions from 400 to 500 nm and 700 to 800 nm. The absorption at 700–800 nm was assigned to the Cu(Ⅰ) d-d transition. The absorption at 400–500 nm may arise from the direct IFCT from the valence band (VB) of Na₂Ti₆O₁₃ to Cu(Ⅰ) [42]. The plasmonic absorption peaks of the Au(5)@Na₂Ti₆O₁₃ and Au(5)-Cu(Ⅰ)(40)@ Na₂Ti₆O₁₃ membranes appeared at ~560 nm, resulting in the violet color of the Au-modified samples.

The amount of Au loaded on the membranes was changed by varying the loading time. The ICP analysis showed that the Au contents of the Au(2)@Na₂Ti₆O₁₃, Au(5)@Na₂Ti₆O₁₃, Au(10)@Na₂Ti₆O₁₃, and Au(30)@Na₂Ti₆O₁₃ membranes were 0.93, 2.85, 8.09, and 17.64 wt%, respectively. Fig. 6 shows the effect of Au content on the photocatalytic activities of the Au@Na₂Ti₆O₁₃ membrane samples. The adsorption–desorption equilibrium of acetaldehyde on the sample surface in the dark was almost achieved after 30 min. The pure Na₂Ti₆O₁₃ membrane displayed the lowest photocatalytic activity of the samples; the concentration of acetaldehyde changed only slightly over 330 min. The low visible-light activity of the pure Na₂Ti₆O₁₃ nanofiber membrane should originate from the presence of intrinsic defects, such as oxygen vacancies, because they can also induce some gap states to absorb some visible light [43, 44]. Compared with that of the pure Na₂Ti₆O₁₃ membrane, the photocatalytic activity of the membranes was greatly increased after Au deposition, which should be ascribed to function of LSPR absorption (Fig. 6). The LSPR absorption excites the electrons in Au nanoparticles to higher energy levels, which allows them to flow to the conduction band (CB) of Na₂Ti₆O₁₃ nanofibers, resulting in their separation from the holes in Au nanoparticles. As a consequence, the holes left in the Au nanoparticles have enough time to oxidize organic molecules rather than recombining with electrons, leading to the plasmonic photocatalytic effect. In addition, O₂ can be reduced by the electrons in the CB of Na₂Ti₆O₁₃. Fig. 6 reveals that the photocatalytic activity of the Au@Na₂Ti₆O₁₃ membranes first increases and then decreases with the increased loading of Au. The Au(5)@Na₂Ti₆O₁₃ membrane displayed the highest photocatalytic activity of the samples. The initial increase of photocatalytic activity upon Au deposition is mainly caused by the increase of LSPR absorption. In addition to the LSPR effect, surface plasmon resonance of Au nanoparticles can generate enhanced localized electric fields, which may also partially contribute to the increased photocatalytic effect [45-47]. Loading with a high content of Au nanoparticles resulted in decreased photocatalytic activity, indicating increased carrier recombination, which is possibly caused by the accelerated back interfacial transfer of electrons from Na₂Ti₆O₁₃ to Au nanoparticles.

Fig. 7(a) shows the time-dependent change of acetaldehyde concentration during photocatalysis by the pure Na₂Ti₆O₁₃, Cu(Ⅰ)(40)@Na₂Ti₆O₁₃, Au(5)@Na₂Ti₆O₁₃, and Au(5)- Cu(Ⅰ)(40)  @ Na₂Ti₆O₁₃ membranes under visible-light illumination. Over the Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane, 78% of acetaldehyde was decomposed after 300 min, clearly showing promising visible-light photocatalytic activity. In the Cu(Ⅰ)(40)@  Na₂Ti₆O₁₃ membrane system, it is suggested that the visible-light illumination can initiate the IFCT [48] of electrons to Cu(Ⅰ) nanoclusters; i.e., the electrons in the VB of Na₂Ti₆O₁₃ are transferred directly to Cu(Ⅰ), and part of the Cu(Ⅰ) is then reduced to Cu(0). The holes produced in the VB of Na₂Ti₆O₁₃ have strong oxidative power to induce the decomposition of organic compounds such as aldehydes. Meanwhile, the excited electrons in the Cu(Ⅰ) nanoclusters are capable of reducing the adsorbed O₂ via multi-or single-electron interfacial transfer pathways [49]. After photocatalytic reaction, the Cu(0) is easily oxidized to Cu(Ⅰ), ensuring the good stability and sustainability of the Cu(Ⅰ)-loaded membrane photocatalysts [50-52]. Consequently, the Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane exhibited photocatalytic activity under visible-light illumination [53]. For the Au(5)@Na₂Ti₆O₁₃ membrane, 93% of the acetaldehyde present was decomposed after 60 min. The photocatalytic activity was further increased by loading with Cu(Ⅰ) nanoclusters, with 98% acetaldehyde removal after 45 min of visible-light illumination.

The photocatalytic reactions were fitted with first-order kinetics, as shown in Fig. 7(b). Rate constants (k) were estimated by the linear fittings of time dependence of ln(C_t/C₀). The k values of the pure Na₂Ti₆O₁₃, Cu(Ⅰ)(40)@Na₂Ti₆O₁₃, Au(5)@  Na₂Ti₆O₁₃, and Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membranes are 1.0×10^-3, 5.0×10^-3, 0.05, and 0.11 min^-1, respectively. These results show that the Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane presents the highest activity of the samples, which is 110 times higher than that of the pure Na₂Ti₆O₁₃ membrane. Even though the actual Au content of the Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane was decreased during the loading of Cu(Ⅰ) nanoclusters (Table 1), the Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane exhibited higher photocatalytic activity than those of the Au@Na₂Ti₆O₁₃ and Cu(Ⅰ)@Na₂Ti₆O₁₃ membranes. This is because the Cu(Ⅰ) nanoclusters enhanced charge separation, which increased the visible-light activity of the Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane.

For the Au(5)-Cu(Ⅰ)(40)@Na₂Ti₆O₁₃ membrane system, as shown in Fig. 8, the photoexcited electrons in Au nanoparticles can transfer to the CB of Na₂Ti₆O₁₃ nanofibers. The one- dimensional single-crystalline Na₂Ti₆O₁₃ nanofibers display relatively high electron transport ability compared to that of nanocrystalline materials, which should be beneficial for the efficient separation of electrons and holes. For the samples loaded with both Cu(Ⅰ) nanoclusters and Au nanoparticles, most of the visible light was absorbed because of the LSPR of Au nanoparticles. The photoinduced electrons in Au nanoparticles can undergo interfacial electron transfer to the CB of Na₂Ti₆O₁₃ nanofibers. As a result of the lowering of the Fermi energy, acetaldehyde is then oxidized on the Au nanoparticle surface. At the same time, the electrons in the CB of Na₂Ti₆O₁₃ can be trapped by the amorphous Cu(Ⅰ) nanoclusters. It has been indicated that Cu(Ⅰ) nanoclusters could act as catalytic sites for the interfacial transfer of electrons to O₂, which would further facilitate the electron–hole separation and electron interfacial transfer [54]. As a result, the lifetimes of both the excited electrons and holes are prolonged, which would increase their probability for interfacial transfer and subsequent involvement in contaminant decomposition.

A novel self-standing visible-light-active Au-Cu(Ⅰ)@  Na₂Ti₆O₁₃ nanostructured hybrid membrane was prepared. Compared with those of pure Na₂Ti₆O₁₃, Au@Na₂Ti₆O₁₃, and Cu(Ⅰ)@Na₂Ti₆O₁₃ membranes, the hybrid membrane presented higher photocatalytic activity for acetaldehyde decomposition. In photocatalysis by the Au@Na₂Ti₆O₁₃ membranes, visible light absorption through the LSPR of Au nanoparticles induced electron transfer from the Au nanoparticles to the CB of Na₂Ti₆O₁₃. Because the loaded Cu(Ⅰ) behaved as an electron trap, the recombination of electrons in the CB of Na₂Ti₆O₁₃ with the holes remaining in Au nanoparticles was suppressed. Therefore, the lifetimes of both excited electrons and holes were prolonged, promoting their involvement in photocatalysis. This membrane material avoids aggregation during practical application, and can be easily separated and recycled after photocatalytic reaction, making it attractive for air purification.