Photocatalysis has been intensively investigated as it has potential for exploitation in environmental remediation (pollutant degradation) and clean energy production (H₂ production, CO₂ reduction) ^{[1, 2, 3]}. The metal oxide semiconductors, such as TiO₂, tungsten oxide (WO₃), Bi₂O₃, and ZnO, have been shown to be efficient as photocatalysts in different processes ^{[4, 5, 6]}. Among these metal oxides, WO₃ is of great interest because of its attractive properties such as small energy band gap (from 2.4 to 2.8 eV), stable physicochemical properties, non-toxicity, and resistance to photocorrosion ^[7].

However, a high charge recombination rate of photo-induced electron-hole pairs is a critical drawback in the development of WO₃ as an efficient photocatalyst ^[8]. Several methods have been developed to improve the photocatalytic activity of WO₃. Zhang et al. ^[9] synthesized monoclinic WO₃ nanoplates by a one-step template-free hydrothermal route through a topochemical conversion process of H₂WO₄ nanoplates. The as-prepared m-WO₃ nanoplate exhibited a superior photocatalytic activity for the degradation of rhodamine B (RhB) under visible light irradiation. Lin et al. ^[10] modified WO₃ with an ionic liquid [Bmim]I through a facile impregnation method. The presence of a surface-bound imidazolium ring was demonstrated to effectively suppress the recombination of photoexcited electron-hole pairs.

Additionally, it has been shown that proper junctions formed in semiconductor-based photocatalysts could lead to enhanced activity. Oh et al. ^[11] prepared WO₃/MWCNT-TiO₂ composites using a modified sol-gel method. It is found that the WO₃/MWCNT-TiO₂ photocatalyst showed efficient charge separation during UV irradiation. Our previous work ^[12] indicated that the phase junction formed between the anatase and rutile TiO₂ could greatly enhance photoexcited charge migration between the two phases that, in turn, enhanced the charge separation. Recently, Jaroniec et al. ^[13] proposed a new “surface heterojunction” concept to explain the difference in the photocatalytic activity of anatase TiO₂ with co-exposed (001) and (101) facets. The phase junction enhanced charge separation and thus improved photocatalytic efficiency, and has been demonstrated for a variety of materials. α-Ga₂O₃/β-Ga₂O₃ ^[14], α-Bi₂O₃/β-Bi₂O₃ ^[15], and WO₃/WO₃·H₂O ^[16] phase junction structures have been successfully synthesized and exhibit high photocatalytic activity for H₂ production and RhB degradation. Therefore, the formation of the phase junction becomes an efficient way to produce semiconductor metal oxides exhibiting high photocatalytic activity.

WO₃ exists mainly in four polymorphs, which are monoclinic, triclinic, orthorhombic, and tetragonal, along with a hexagonal phase ^[17]. Among these crystalline phases, the monoclinic phase has been reported as the most stable phase ^[18]. The monoclinic (m-WO₃) and hexagonal (h-WO₃) phases of WO₃ have attracted the most attention because the m-WO₃ is an attractive candidate for photocatalytic applications and h-WO₃ exhibits an open-tunnel structure and rich intercalation chemistry ^[19]. We are interested in determining if we can fabricate the phase junction formed between m-WO₃ and h-WO₃ by a simple method, as well as understanding if the phase junction in WO₃ enhances its photocatalytic activity. However, the relationship between the phase junction in WO₃ and the photocatalytic activity is not well understood.

Generally, a material with different crystalline phases possesses a different band gap and flat band. Therefore, a phase junction could be built between different crystal phases given the two phases are in close contact, which will lead to efficient electron-hole separation and higher catalytic reactivity ^[20]. According to Ref. ^[21], the valence band position and conduction band position of h-WO₃ are lower than those of m-WO₃, which implies that a phase junction could be established between m-WO₃ and h-WO₃.

Therefore, this study attempts to fabricate a phase junction in a WO₃ photocatalyst, and then to investigate the effect of the phase junction in the WO₃ photocatalyst on its photocatalytic activity. The thermal decomposition method was used to fabricate a phase junction in WO₃ by tuning the calcination temperature and the calcination time, and the phase junction (m-WO₃/h-WO₃) formed between hexagonal and monoclinic WO₃ demonstrated an enhanced photocatalytic activity for RhB degradation.

Ammonium tungstate hydrate (H₄₀N₁₀O₄₁W₁₂·xH₂O, APT) was of analytical grade and used without further purification. For the thermal decomposition method, the white APT powder was calcined at 600-1000 °C for 4 h in a static atmosphere of air, and then was naturally cooled to room temperature. The obtained samples were labeled as WO₃-T, where T represents the calcination temperature (°C) used to obtain the samples. APT powders were also calcined at 800 °C for different times from 8 to 30 h and were denoted as WO₃-t, where t represents calcination time (h).

Thermogravimetry and differential thermal analysis (TG-DTA) was performed using a TG/DTA-6300 thermal analyzer at the rate of 10 °C/min under air atmosphere (60 mL/min). The structure characterization was carried out by powder X-ray diffraction (XRD) using a Rigaku MiniFlex diffractometer with a Cu K_a radiation source. The morphologies of the samples were investigated using scanning electron microscopy (SEM, Quanta 200 F) and high-resolution transmission electron microscopy (HETRM, JEOL, JEM-2100). X-ray photoelectron spectra (XPS) were obtained using a Thermo ESCALAB 250Xi X-ray photoelectron spectrometer. The BET surface area was determined by N₂ adsorption-desorption isotherms using a NOVA 4200e surface area and pore size analyzer from Quantachrome. A Hitachi F-4500 spectrophotofluorometer was used to record the photoluminescence spectra (PL) using an excitation wavelength of 350 nm.

A set of photocatalytic degradation experiments were performed with the following procedure. Photodegradation of RhB was carried out in a 100-mL Pyrex reactor filled with de-ionized water (60 mL) containing RhB (10 mg/L) and the WO₃ sample. The suspension was stirred for 30 min in the dark to obtain adsorption-desorption equilibrium of the dye before illumination. After irradiation with a 125-W Hg lamp (λ = 365 nm, Shanghai Feilipu Yaming Lighting Co.), a 3-mL aliquot was taken every 30 min and immediately centrifuged. The RhB concentration in the clear solution was analyzed by optical characteristic absorption (22PC spectrophotometer, Shanghai Lengguang Technology Co., Ltd.) at a wavelength of 553 nm for a RhB solution. The blank experiments without a mercury lamp and without catalysts in the RhB solution under the same conditions are also compared. The blank study shows that mere photolysis can be ignored for RhB.

Fig. 1 shows TG-DTA curves of APT in a dynamic atmosphere of air. The TG curve shows five main mass-loss steps with a total of 11.8% loss of the original mass of APT. Four mass-loss steps totaling 10.6% (2.2% + 2.1% + 1.2% + 5.1% = 10.6%) and four main endothermic peaks at 84, 130, 208, and 275 °C could be observed in the DTA curve, which are attributed to the loss of absorbed water, crystal water, structural water, and loss of ammonia. In addition, two exothermic peaks are observed at 355 and 440 °C. The peak at 440 °C can be attributed to crystallization of the amorphous phase of WO₃ ^[22]. In this study, calcination temperatures higher than 600 °C were chosen to prepare WO₃, which facilitated a high degree of crystallinity.

Fig. 1. TG-DTA curves of APT in a dynamic atmosphere of air.

Fig. 2 shows the XRD patterns of APT calcined at different temperatures. When the WO₃ sample was calcined at 600 °C, diffraction peaks at 2θ = 23.1°, 23.6°, 24.3°, and 34.1° were observed, which represent the indices of the (002), (020), (200), and (202) planes of m-WO₃ (PDF ^#43-1035), respectively. These results indicate that the WO₃-600 °C sample is in the monoclinic phase. Moreover, the XRD peaks are strong and sharp, indicating that the WO₃-600 °C sample exhibits high crystallinity. The XRD pattern of the WO₃-700 °C sample is almost the same as that for the WO₃-600 °C sample. On heating to 800 °C, three weak peaks appear at 2θ = 14.0°, 28.2°, and 36.6°, which arise from h-WO₃ (PDF ^#33-1387), in addition to the typical peaks of the monoclinic phase. This implies the co-existence of hexagonal and monoclinic phases in the WO₃-800 °C sample, while m-WO₃ is the main constituent in the crystalline phase.

Fig. 2. XRD patterns of APT calcined at different temperatures.

Martínez-de la Cruz et al. ^[23] prepared m-WO₃ particles by the precipitation method using APT as the starting material and PEG as the template and steric stabilizer. It was also found that a long calcination time promoted the formation of the hexagonal structure of WO₃ in addition to m-WO₃. The XRD peaks of h-WO₃ hardly changed for the WO₃ sample calcined at temperatures up to 900 °C. However, the intensities of the typical peaks of h-WO₃ were decreased for the WO₃-1000 °C sample (Fig. 2(b)).

The morphologies and particle sizes of the WO₃-T samples were investigated by SEM (Fig. 3). As shown in Fig. 3(a) and (b), the WO₃-600 °C sample consists of dispersed large and fine particles with irregular polyhedral shapes. Bulky components with dimensions up to 10 mm were observed, and the sizes of the fine particles were approximately 200-500 nm for the WO₃-600 °C sample. It is obvious that the rod-like components (circled by a dashed line in Fig. 3), assigned to the hexagonal phase, appear in the WO₃-800 °C sample (Fig. 3(c) and (d)). This shows the co-existence of the monoclinic and hexagonal phases in the WO₃-800 °C sample, which is in agreement with the results from XRD (Fig. 2). Moreover, it can be seen that m-WO₃ and h-WO₃ particles are in close contact, and m-WO₃ particles are located on the surface of h-WO₃ particles with a rod-like shape.

Fig. 3. SEM images of WO3-600 °C (a, b), WO3-800 °C (c, d), and WO3-1000 °C (e, f).

Crucial information on the spatial distribution of the h-WO₃ and m-WO₃ particles was further provided by HRTEM (Fig. 4). The HRTEM image of WO₃-800 °C shows an example of the intimate contact between m-WO₃ and h-WO₃ phases. As shown in Fig. 4, the small particles and rod-like particles exhibit fringes with a lattice spacing of 3.779 and 6.361 Å, respectively, which can be indexed into the (020) atomic plane of m-WO₃ and (100) atomic plane of h-WO₃, respectively. HRTEM and SEM results (Fig. 3) reveal that the phase junction can be built between the m-WO₃ particles and h-WO₃ nanorods.

Fig. 4. HRTEM image of WO3-800 °C.

From Fig. 3, no apparent changes in the morphology of the m-WO₃ particles were observed in the WO₃-1000 °C sample, as compared with the WO₃-800 °C sample. However, the samples obtained after calcination at 1000 °C showed a minor tendency to form large agglomerates, as shown in Fig. 3(e). Moreover, it should be noted that it is not easy to observe rod-like particles in the WO₃-1000 °C sample, and the number of rod-like particles decreased for WO₃-1000 °C as compared with the WO₃-800 °C sample, from the SEM image. This result is in agreement with that from XRD (Fig. 2).

It is reported that triangular and hexagonal channels are formed along the structure in h-WO₃ bronzes owing to the special arrangement of corner-sharing WO₆ octahedra ^[24]. Alkaline (Na⁺, K⁺, Cs⁺ and so forth) or NH₄⁺ ions are thought to be located in hexagonal channels ^[25]. Szilágyi et al. ^[24] studied the formation of h-WO₃ during the annealing of hexagonal ammonium tungsten bronze (HATB), which is obtained by heating APT ((NH₄)₁₀[H₂W₁₂O₄₂]·4H₂O) in H₂. They proposed that, in general, some ions or molecules were needed in the hexagonal channels to maintain a hexagonal tungsten oxide structure. Moreover, residual NH₄⁺ and NH₃ (from APT) in the hexagonal channels seemed to be vital for stabilizing h-WO₃: when NH₄⁺ and NH₃ were completely released, the hexagonal framework collapsed in an exothermic reaction into m-WO₃. Thus, combined with the results from previous studies, h-WO₃, which forms at 800 °C, may transform into m-WO₃ by increasing the calcination temperature to 1000 °C because of the release of NH₄⁺ ions and NH₃ molecules (from APT starting material) in the hexagonal channels at high temperature. Thus, the amount of rod-like h-WO₃ decreased for WO₃-1000 °C from Fig. 2. Further work is in progress to study the mechanism of the phase transformation between h-WO₃ and m-WO₃ in our lab.

To further change the phase composition in the WO₃ sample, the APT powder was calcined at 800 °C for different times. Fig. 5 displays the XRD patterns of the WO₃-t samples. It can be seen that the mixed phases of m-WO₃ and h-WO₃ were obtained for all the WO₃-t samples. WO₃-8 h sample is mainly composed of m-WO₃ with a small amount of h-WO₃. More intense peaks of h-WO₃ appeared when the annealing time was increased from 8 to 12 h (Fig. 5(b)), suggesting an increase of h-WO₃ content in the mixed phases. However, the diffraction peaks of h-WO₃ gradually decreased in intensity after calcination for 12 h. With a further increase of the calcination time up to 24-30 h, WO₃ samples were again mainly composed of m-WO₃ with a small amount of h-WO₃.

Fig. 5. XRD patterns of APT calcined at 800 °C for different times.

The XPS technique was used to determine the chemical states of WO₃ samples. The overview XPS spectrum of the WO₃-12 h sample in Fig. 6(a) shows the corresponding peaks of O 1s, C 1s, and W 4f, confirming the existence of W and O elements in the WO₃-12 h sample. The fine spectrum in Fig. 6(b) displays two peaks at 35.4 and 37.5 eV, which suggests that the W exists in the W⁶⁺ form ^[26]. These two peaks are attributed to the spin-orbit splitting of the W 4f components (W 4f_7/2 and W 4f_5/2). For O 1s XPS (Fig. 6(c)), a peak at 530.6 eV was found, which can be assigned to lattice oxygen O^2- of WO₃ ^[27]. According to XPS studies, the W/O atomic ratio was estimated to be 1:3, which is consistent with tungsten(VI) trioxide. Therefore, the XPS results agree well with the results from the XRD pattern of the WO₃-12 h sample.

Fig. 6. XPS spectra of the WO3-12 h sample. (a) Survey spectrum; (b) W 4f; (c) O 1s.

The morphologies of the WO₃-t samples were determined by the SEM analysis and are shown in Fig. 7. The m-WO₃ particles exhibit a bulky shape with heavy aggregates, while h-WO₃ particles exhibit a rod-like shape. Therefore, substances containing m-WO₃ and h-WO₃ can be identified simply by the particle morphology. The WO₃-8 h sample (Fig. 7(a)) is composed of m-WO₃ particles and some h-WO₃ particles. As compared with the WO₃-8 h sample, the number of rod-like particles obviously increases for the WO₃-12 sample, which is in accordance with the XRD results (Fig. 5). It is shown that h-WO₃ particles can be easily observed in the WO₃-16 h sample although the number of h-WO₃ particles decreases with an increase of the calcination time to 16 h. With a further increase in the calcination time to 30 h, the WO₃ products are composed of a large number of m-WO₃ particles coupled with a small number of h-WO₃ particles, as shown in Fig. 7(g).

Fig. 7. SEM images of WO3-8 h (a, b), WO3-12 h (c, d), WO3-16 h (e, f), and WO3-30 h (g, h).

The results from XRD, SEM, and HRTEM demonstrate that well-defined m-WO₃/h-WO₃ phase junctions of WO₃ can be fabricated by the phase transformation between m-WO₃ and h-WO₃. It is observed that the ratio of m-WO₃ and h-WO₃ in the WO₃ products is mainly controlled by the calcination time. Moreover, from SEM and HRTEM images, it can be seen that m-WO₃ crystals are sporadically patched on the surface of the h-WO₃ rod-like particles, resulting in the exposure of both m-WO₃ and h-WO₃ on the surface.

RhB was used as a representative organic substance to investigate and compare the photocatalytic activity of the WO₃ samples (Fig. 8). The degradation of RhB could be described by the first-order kinetics of ln(C_t/C₀) versus reaction time (t): -ln(C_t/C₀) = k_apt, where k_ap is the apparent reaction rate constant, and C₀ and C_t are the initial concentration and the concentration at reaction time t of RhB, respectively.

Fig. 8. Photocatalytic activity of the degradation of RhB on the WO3 calcined at different temperatures.

After 180 min of irradiation, 41% and 30% of RhB were degraded by the WO₃-600 °C and WO₃-700 °C catalysts, respectively. The lower degradation ability may be ascribed to the decrease of the surface area from 1.216 to 0.867 m²/g with increasing calcination temperature from 600 to 700 °C. From Table 1, WO₃-600 °C, WO₃-700 °C, WO₃-800 °C, WO₃-900 °C, and WO₃-1000 °C samples exhibit a photocatalytic performance with k_ap determined as 0.00298, 0.00198, 0.00383, 0.00440, and 0.00232 min^-1. It is observed that the photocatalytic activity increases with increasing calcination temperature to 800 °C. The WO₃-900 °C sample exhibits the highest photocatalytic activity although the surface area of the WO₃-900 °C sample (0.779 m²/g) is lower than those of WO₃-600 °C and WO₃-700 °C. Therefore, we concluded that the enhancement in the photocatalytic activity of the WO₃-800 °C and WO₃-900 °C samples arose from the phase junction between m-WO₃ and h-WO₃, the presence of which is confirmed by SEM and HRTEM results. The formation of the phase junction may improve the charge separation efficiency of photoexcited electrons and holes in the WO₃-800 °C and WO₃-900 °C samples. Thus, WO₃-800 °C and WO₃-900 °C samples show a higher photocatalytic activity than WO₃-600 °C and WO₃-700 °C containing the monoclinic phase.

Table 1 First-order rate constants (kap) and relative coefficients (R2) for the degradation of RhB under UV light irradiation for different calcination temperatures and different calcination times. Catalyst kap (min-1) R2 WO3-600 °C 0.00298 0.96308 WO3-700 °C 0.00198 0.99605 WO3-800 °C 0.00383 0.97547 WO3-900 °C 0.00440 0.99094 WO3-1000 °C 0.00232 0.97461 WO3-8 h 0.00818 0.99267 WO3-12 h 0.01025 0.99569 WO3-16 h 0.00853 0.99833 WO3-24 h 0.00680 0.99614 WO3-30 h 0.00672 0.99858

Table 1 First-order rate constants (kap) and relative coefficients (R2) for the degradation of RhB under UV light irradiation for different calcination temperatures and different calcination times.

By further increasing the calcination temperature to 1000 °C, the photocatalytic activity of WO₃-1000 °C (surface area is 0.617 m²/g) decreased. However, the WO₃-1000 °C sample shows a comparable photocatalytic activity to WO₃-600 °C although the photocatalytic activity of WO₃-1000 °C decreases compared with that of WO₃-800 °C and WO₃-900 °C. From the above discussion, the decrease of the photocatalytic activity for WO₃-1000 °C arises from the decrease in the amount of h-WO₃ when the calcination temperature is raised to 1000 °C, which decreases the amount of the exposed m-WO₃/h-WO₃ phase junction.

To further confirm the phase junction effect of WO₃, the photocatalytic activities of WO₃-t samples that were calcined at 800 °C for different times were investigated (Fig. 9) because the m-WO₃ and h-WO₃ phase composition can be carefully controlled by calcination at 800 °C for different times. When the APT powder was calcined at 800 °C for 8-30 h, all samples are in the mixed phases of m-WO₃ and h-WO₃ from XRD results (Fig. 5), and all the above samples show higher photocatalytic activities compared with WO₃-600 °C containing m-WO₃ (Fig. 9). Moreover, the photocatalytic activities were improved by gradually increasing the calcination time range from 8 to 12 h. The increase in the photocatalytic activity may mainly arise from the extent of the phase junction when the calcination time is increased from 8 to 12 h because the XRD results indicate that the amount of h-WO₃ increases gradually with increasing calcination time. From Table 1, the WO₃-12 h sample exhibits the highest photocatalytic activity with k_ap of 0.01025 min^-1. With further increases of the calcination time to 24 and 30 h, the photocatalytic activity decreases.

Fig. 9. Photocatalytic activity of degradation of RhB by the WO3-t photocatalysts and WO3-600 °C.

A mechanism for electron-hole separation and transport of an m-WO₃/h-WO₃ phase junction is proposed and is shown in Fig. 10. The valence and conduction band potentials of h-WO₃ were found to be 3.18 and 0.41 eV ^[21], whereas these values for m-WO₃ were 3.37 and 0.80 eV, respectively ^[28]. The conduction band (CB) of h-WO₃ is lower than that of m-WO₃, thus the photo-induced electrons on the conduction band of h-WO₃ transfer easily to m-WO₃ through the well-developed interface, and the photogenerated holes move in the opposite direction from the electrons. Therefore, the phase junction of m-WO₃/h-WO₃ could act as an active center for hindering the rapid recombination of photo-induced electron-hole pairs. The electrons accumulated on the CB of m-WO₃ can be scavenged by O₂ to produce a superoxide radical (^•O₂^-) and hydrogen peroxide (H₂O₂), which can interact to produce a powerful oxidant (hydroxyl radical, ^•OH) to decompose RhB. The holes accumulated in the VB of h-WO₃ take part in the oxidation process to make OH⁻ or H₂O species to produce reactive hydroxyl radicals ^{[29, 30, 31]}.

Fig. 10. A proposed mechanism for the charge transfer and separation in the m-WO3/h-WO3 phase junction.

To further investigate the effect of the phase junction, PL analysis was applied to reveal the diffusion and recombination process of photogenerated electron-hole pairs. Fig. 11 shows the PL spectra of WO₃-600 °C, WO₃-800 °C, and WO₃-12 h samples. A main peak at approximately 469 nm and several peaks in the range of 400-600 nm were observed in the PL spectrum of WO₃-600 °C, which is in agreement with the results from Ref. ^[32]. In the case of WO₃-800 °C and WO₃-12 h with an m-WO₃/ h-WO₃ junction, the shape of the curves is similar to that of WO₃-600 °C, whereas the peak intensities significantly decrease.

Fig. 11. PL emission of different photocatalysts under UV radiation.

Generally, a higher PL intensity indicates a higher recombination rate of photoexcited electron-hole, and a lower PL intensity expresses a lower recombination rate of photoexcited electrons and holes ^{[33, 34]}. This indicates that the interaction between h-WO₃ and m-WO₃ in the WO₃-800 °C and WO₃-12 h samples is favorable for interfacial charge transfer, and thus reduces the recombination rate of electron-hole pairs under light irradiation. Moreover, it can be seen that the PL intensity of the peak at 469 nm of WO₃-12 h is weaker than that of the WO₃-800 °C sample. This result indicates that WO₃-12 h exhibits a lower electron-hole recombination rate than that of WO₃-800 °C. Thus, WO₃-12 h exhibits higher photocatalytic activity than WO₃-800 °C.

WO₃ photocatalysts with different crystalline phases were prepared by thermal decomposition of APT at 600-1000 °C for 4 h and 800 °C for 8-30 h. Characterization of crystal structure, morphology, chemical composition, surface area, and optical properties of WO₃, together with their relationships with the photocatalytic degradation of RhB, reveal that a phase junction between m-WO₃ and h-WO₃ of WO₃ was fabricated by a thermal decomposition strategy to inhibit the recombination of carriers. The promotion of the photocatalytic performance is mainly ascribed to the enhanced charge separation originating from the phase junction between m-WO₃ and h-WO₃. This work provides a simple method to design and fabricate the effective WO₃-based photocatalyst.