Dimethyl phthalate (DMP) is widely used as an additive in the manufacture of plastics, polyvinyl acetates and cellulosics even though it is an endocrine-disrupting chemical and has great potential to interfere with the hormonal control systems of humans. With the manufacture and use of DMP, it is inevitably discharged into the environment ^{[1, 2, 3]}. It is important to effectively remove DMP from aqueous systems to minimize its adverse effects. An attractive method to remove DMP is photocatalytic treatment using solar radiation in the presence of a photocatalyst. In this respect, a semiconductor photocatalyst that is inexpensive and stable under ambient conditions is required for use in clean technology.

In recent years, applications of semiconductors in the elimination of environmental pollutants from aquatic solutions have been attracting increasing attention because of their high physicochemical stability and photocatalytic performance ^{[4, 5]}. Semiconductors like titanium dioxide (TiO₂) ^{[6, 7, 8, 9]}, molybdenum trioxide (MoO₃) ^{[10, 11, 12, 13]}, vanadium pentoxide (V₂O₅) ^{[14, 15, 16]}, tungsten oxide (WO₃) ^{[17, 18, 19]}, cerium oxide (CeO₂) ^[20], zinc oxide (ZnO) ^[21], and cadmium sulfide (CdS) ^{[22, 23]}, are widely used as photocatalysts. However, a common drawback of these photocatalysts is the fast recombination of photogenerated electron-hole pairs in them, which lowers their photocatalytic efficiency. Several methodologies have been used to decrease the recombination rate of photogenerated electron-hole pairs and to enhance the photocatalytic efficiency of semiconductor photocatalysts, such as doping with metals (gold, vanadium, tungsten, strontium) ^{[24, 25, 26, 27]} or nonmetals (carbon, nitrogen, sulfur) ^{[28, 29, 30]}, surface modification ^{[31, 32]}, and combination with another semiconductor ^{[33, 34]}. In our previous work, we used the semiconductor compound MoO₃ as a dopant to improve the photocatalytic performance of V₂O₅ and obtained the predicted improvement in performance ^[35]. It is well known that heterostructures formed by integrating two or more semiconductors can theoretically improve photocatalytic efficiency because the photogenerated electrons can migrate from a semiconductor with a higher conduction-band (CB) minimum to another with a lower CB minimum.

Besides being coupled with other semiconductors to improve the photocatalytic performance of the resulting composites ^{[36, 37]}, MoO₃ has also been widely used because of its unique structure and chemical properties ^[10]. What will happen when MoO₃ is used as the host and V₂O₅ as the dopant? In the present contribution, we explore the influence of a molar ratio of V to Mo, n(V)/n(Mo) on the structure and photocatalytic activity of the resulting V₂O₅/MoO₃ composites. The V₂O₅/MoO₃ composites are fabricated via electrospinning, a useful technique for the preparation of composites with controllable hierarchical features ^[38]. The effect of n(V)/n(Mo) on the structure, morphology, surface properties, and optical absorption of the composites, as well as their application in the degradation of DMP are analyzed.

All reagents were of analytical grade and used as purchased from commercial suppliers without further purification. V₂O₅/MoO₃ composites with different n(V)/n(Mo) were synthesized by the following electrospinning and calcination processes. Polyvinyl pyrrolidone (PVP, 1 g) was dissolved in ethanol (9 mL) and stirred for 10 h to form solution (A). Ammonium molybdate tetrahydrate ((NH₄)₆Mo₇O₂₄·4H₂O, 0.4618 g) and ammonium metavanadate (NH₄VO₃, 0.0382 g) (n(V)/n(Mo) = 1/8 ) were dissolved in 50% ethanol solution (4 mL) and then stirred for 10 h to form solution (B). Solution (B) was added to solution (A), and then the resulting mixture was stirred for 10 h at room temperature. A viscous gel of PVP/(NH₄)₆Mo₇O₂₄/NH₄VO₃ formed during stirring. The as-obtained gel was transferred to a syringe, and a piece of copper wire connected to a high-voltage generator was inserted into the gel. A direct current voltage of 16 kV was applied for electrospinning. A piece of flat aluminum foil was placed 16 cm under the tip of the syringe to collect the composite fibers. The obtained composite fibers were calcined at a heating rate of 1 °C/min and held for 3 h at 300, 400 or 500 °C in air. Composites with n(V)/n(Mo) = 1/8, 1/6, 1/4, 1/2 and 1/1, which are denoted as VM-8, VM-6, VM-4, VM-2 and VM-1, respectively, were prepared under the same conditions. For comparison, pure MoO₃ and V₂O₅ photocatalysts were also prepared under the same conditions.

The crystal structure of samples was determined by X-ray diffraction (XRD) patterns recorded on a Rigaku D/Max-IIB diffractometer with Cu K_α (λ = 0.15405 nm) radiation at a scan rate of 4°/min in the range of 2θ = 10°-80°. The morphology of samples was observed by field-emission scanning electron microscopy (FE-SEM, FEI-Philips XL-30) and transmission electron microscopy (TEM, Philips T20ST). The specific surface areas (SSA) of samples were measured at liquid N₂ temperature using the Brunauer-Emmett-Teller method (BET, JW-K). X-ray photoelectron spectra (XPS) were recorded on an ESCALAB-MKII photoelectron spectrometer with Al K_α (1468.6 eV) radiation as the excitation source. Ultraviolet-Visible diffuse reflectance spectra (UV-Vis DRS) were obtained by a Lambda 900 UV-Vis-NIR spectrophotometer (Perkin-Elmer). The identification of degradation intermediates of DMP was performed by high-performance liquid chromatography-tandem mass spectrometry (OA_SPE Waters Xevo TQ_S).

Photocatalysis experiments using the composites were performed in a self-assembled photoreactor at room temperature. The photoreactor used a 500-W high-pressure xenon lamp as a visible-light source (λ > 420 nm), which was surrounded by a water-cooling quartz jacket and with a UV cutoff filter. For the photocatalytic degradation of methylene blue (MB) solution, MB aqueous solution (10 mg/L, 100 mL) and catalyst (50 mg) were stirred in the dark for ca. 30 min to let the catalyst disperse completely and establish an adsorption-desorption equilibrium between MB and catalyst. At given time intervals, 10-mL aliquots of the reaction mixture were withdrawn and centrifuged; the absorption of the resulting clear solutions at λ_max = 664 nm was examined using a UV-Vis 756B spectrophotometer. Photocatalytic degradation of DMP solution by the catalysts was tested under similar conditions. DMP solution (40 mg/L, 60 mL) and catalyst (100 mg) were stirred in the dark for ca. 30 min. Every hour, a 10-mL aliquot of the reaction mixture was withdrawn and centrifuged; the absorption of the clear solution at λ_max = 230 nm was detect by a UV-Vis 756B spectrophotometer. As control experiments, the photocatalytic degradation of both substrates was also tested in the absence of catalyst.

The crystalline phases of pure MoO₃, V₂O₅, and the V₂O₅/MoO₃ composites were analyzed by XRD. As shown in Fig. 1(a), all of peaks observed in the XRD patterns of the composites can be indexed to orthorhombic MoO₃ (JCPDS 05-0508) when n(V)/n(Mo) < 1/6; no characteristic peaks corresponding to vanadium oxides are observed. When n(V)/n(Mo) > 1/6, both the diffraction peaks of orthorhombic MoO₃ (JCPDS 05-0508) and orthorhombic V₂O₅ (JCPDS 41-1426) are observed. The intensity of the peaks corresponding to orthorhombic V₂O₅ increased with n(V)/n(Mo) (Fig. 1(a)). Fig. 1(b) depicts the amplified MoO₃ (021) reflection of the samples. The MoO₃ (021) peak shifted to larger diffraction angle as n(V)/n(Mo) increased until VM-1, where the shift ceases. Fig. 1 indicates that when n(V)/n(Mo) is low, V dopes into the MoO₃ crystal lattice to form a V-doped MoO₃ crystal. As the amount of V ions increases, V₂O₅ and V-doped MoO₃ crystals coexist, forming V₂O₅/V-doped MoO₃ heterojunctions. In this structure, V₂O₅ crystals should be dispersed on the surface of V-doped MoO₃ crystals. Based on these results, it can be deduced that V₂O₅/V-doped MoO₃ heterojunctions form when n(V)/n(Mo) > 1/6.

Fig. 1. (a) XRD patterns of (1) MoO3; (2) VM-8; (3) VM-6; (4) VM-4; (5) VM-2; (6) VM-1; (7) V2O5. (b) Enlarged view of the MoO3 (021) peaks of (1) MoO3; (2) VM-8; (3) VM-6; (4) VM-4; (5) VM-2; (6) VM-1.

For VM-2, the theoretical MoO₃ content (f_M) is 75.99%; however, the experimental value is 80.55%, which was obtained from the integrated intensities of the MoO₃ diffraction line (I_M), and V₂O₅ diffraction line (I_V) using the following phase analysis equation ^[39]:

The experimental content of MoO₃ in the V₂O₅/MoO₃ composites is higher than the theoretical one, which supports that some V ions dope into the MoO₃ lattice while others form V₂O₅ crystals or exist in other forms.

Figure 2 shows FE-SEM images of all samples, along with a TEM image and EDS data for VM-2. Flakes of crystalline MoO₃ have a smooth surface (Fig. 2(a)); crystalline V₂O₅ displays a block-like granular shape (Fig. 2(b)) ^[35]. As seen in Fig. 2(c)-(g), the morphology of V₂O₅/MoO₃ composites is different from that of pure MoO₃ and tends toward the structure of pure V₂O₅ as n(V)/n(Mo) increases with more and more V₂O₅ particles deposited on the MoO₃ flakes. The TEM image of VM-2 clearly reveals that V₂O₅ particles have deeply rooted in the MoO₃ flakes (Fig. 2(h)), indicating that heterojunctions have formed at the interface between MoO₃ and V₂O₅. None of the samples show fiber morphology, which is ascribed to their calcination at high temperature ^[35]. The EDS data of VM-2 in Fig. 2(i) is consistent with the presence of Mo, V, and O elements and n(V)/n(Mo) of around 1/2, which is consistent with the nominal value. The Au signal can be attributed to the Au coating; no other impurities are observed.

Fig. 2. FE-SEM images of (a) MoO3; (b) V2O5; (c) VM-8; (d) VM-6; (e) VM-4; (f) VM-2; (g) VM-1; (h) TEM images of VM-2; (i) EDS data of VM-2.

SSA is an important factor that affects the photocatalytic activity of photocatalysts. This is because a large SSA aids absorption of light, organics and OH groups. Table 1 lists the SSAs of MoO₃, V₂O₅ and the V₂O₅/MoO₃ composites. The SSA of pure V₂O₅ is much higher than that of pure MoO₃. The SSA of VM-8 is lower than that of pure MoO₃, which is attributed to the V₂O₅ particles blocking the secondary pores between MoO₃ flakes. The SSA of the V₂O₅/MoO₃ composites increases gradually with n(V)/n(Mo) from 1/8 to 1/1. The increase of SSA of the composites may be caused by the formation and growth in the number or size of V₂O₅ particles.

Table 1 SSA, band gap (BG), photodegradation efficiency (PE) of MB, generation rate (GR) of phthalic acid over different samples. Sample SSA (m2/g) BG (eV) PE (%)* GR (%)** MoO3 0.7680 2.82 47.62 64.58 VM-8 0.6932 2.34 65.56 — VM-6 — 2.30 73.60 75.57 VM-4 1.5374 2.30 81.68 — VM-2 — 2.25 89.23 82.20 VM-1 2.6664 2.21 78.16 — V2O5 8.5757 2.13 49.95 65.20 * After irradiation for 140min; ** After irradiation for 5 h.

Table 1 SSA, band gap (BG), photodegradation efficiency (PE) of MB, generation rate (GR) of phthalic acid over different samples.

To explore the surface composition and chemical states of the V₂O₅/MoO₃ composites in more detail, the samples were characterized by XPS (Fig. 3). The peaks in the spectra are assigned to O, V, Mo, and C; no other impurities were found. The C impurity originates from a carbon-based contaminant (Fig. 3(a)). Fig. 3(b) presents the Mo 3d spectra of pure MoO₃, VM-6 and VM-2. The main peaks centered at 232 and 235 eV indicate that Mo exists as Mo⁶⁺ ions in the samples ^[40]. Fig. 3(c) shows the V 2p spectra of VM-6, VM-2 and pure V₂O₅. For pure V₂O₅, the peaks at 515.58 and 516.94 eV correspond to V³⁺ and V⁵⁺ ions, respectively, indicating that there are V³⁺ and V⁵⁺ ions in the sample ^[41]. Meanwhile, the peaks of VM-6 at 516.54, 517.43, and 518.09 eV are consistent with V³⁺, V⁴⁺, and V⁵⁺ ions, respectively. The peaks of VM-2 at 516.87 and 517.85 eV correspond to V⁴⁺ and V⁵⁺ ions ^[25], respectively. The presence of V³⁺ and V⁴⁺ ions as well as V⁵⁺ ones is normal in semiconductors containing V₂O₅ ^[42]. Moreover, the fitted Gaussian Lorentzian peaks of V 2p spectra for VM-6 and VM-2 in Fig. 3(c) show that the binding energy of different valence states of V increases slightly compared with those of pure V₂O₅. Compared with that of pure MoO₃, the binding energy of Mo in VM-6 and VM-2 decreases slightly (Fig. 3(b)). Both of these binding energy variations indicate that the chemical environments of V and Mo have changed in the composites compared with in the pure materials because of doping or heterostructure formation. Notably, the presence of V ions of various valence can lead to formation of more oxygen vacancies, which enhances the surface adsorption ability of V₂O₅/MoO₃ composites for water and organic species ^[22].

Fig. 3. (a) XPS spectra, and (b) Mo 3d, (c) V 2p and (d) O 1s spectra of the samples. (1) MoO3; (2) VM-6; (3) VM-2; (4) V2O5.

The O 1s spectra of the samples are provided in Fig. 3(d). For pure MoO₃, the peak at 530.87 eV is attributed to crystal lattice oxygen (O_Mo-O). The peaks of pure V₂O₅ at 529.75, 532.07, and 533.16 eV correspond to crystal lattice oxygen (O_V-O), surface hydroxyl groups (O_O-H) ^[43], and adsorbed water on the surface of V₂O₅, respectively. The peaks of VM-6 and VM-2 at 530 and 531 eV are ascribed to crystal lattice oxygen, O_V-O and O_Mo-O, respectively, while that at 533 eV originates from adsorbed water on the surface of the composites ^[22].

To examine the optical properties of the samples, their UV-Vis DRS were recorded over the wavelength range of 200-800 nm at room temperature. As shown in Fig. 4, pure MoO₃ and V₂O₅ display absorption edges at about 440 and 580 nm, respectively. The absorption edges of VM-8, VM-6, VM-4, VM-2 and VM-1 show a marked red shift compared with that of pure MoO₃, appearing at 530, 540, 540, 550, and 560 nm, respectively. VM-6 and VM-4 show almost the same absorption edge, which may be because VM-6 and VM-4 are at the transition point between V-doped composite and heterojunction formation. The band gaps of the samples were calculated using Equation (2) and are listed in Table 1.

Fig. 4. UV-Vis diffuse reflectance spectra of (1) MoO3; (2) VM-8; (3) VM-6; (4) VM-4; (5) VM-2; (6) VM-1; (7) V2O5.

As shown in Table 1, the band gaps of the samples decreased from 2.82 to 2.21 eV when n(V)/n(Mo) in V₂O₅/MoO₃ composites increased from 0/1 to 1/1. It is obvious that either V-doping or heterojunctions can effectively extend the photoresponse range and decrease the band gap of the composites, which should improve their photocatalytic performance, particularly under visible-light irradiation.

The photocatalytic degradation of MB was chosen as a model reaction to evaluate the photocatalytic performance of the V₂O₅/MoO₃ composites and determine the optimal n(V)/n(Mo). Fig. 5 shows the photodegradation of MB without catalyst and over different photocatalysts under visible-light irradiation. The self-degradation of MB was 7.52%, and the photodegradation over all samples (after subtraction of the photodegradation of MB without photocatalyst) are listed in Table 1.

Fig. 5. Photodegradation of MB over (1) MoO3; (2) VM-8; (3) VM-6; (4) VM-4; (5) VM-2; (6) VM-1; (7) V2O5; (8) No photocatalyst.

All the V₂O₅/MoO₃ composites show higher photocatalytic activity than pure MoO₃ and V₂O_5. V₂O₅/V-doped MoO₃ samples (Fig. 5(4, 5, 6)) exhibit better photocatalytic performance than V-doped MoO₃ samples (Fig. 5(2, 3)). Moreover, VM-2 displays the highest photocatalytic activity of the V₂O₅/MoO₃ composites; therefore, the optimal n(V)/n(Mo) should be 1/2.

The photocatalytic activity and efficiency of a semiconductor are related to many factors, including band gap and SSA ^[20]. Under the same reaction conditions, the narrower band gap, the higher the photocatalytic efficiency, and the larger the SSA, the higher the photocatalytic activity. Therefore, the photocatalytic activity and efficiency of VM-6 are higher than that of VM-8 because of the narrower band gap and larger SSA of VM-6. Because VM-4 has a larger SSA than VM-6, the photocatalytic activity of VM-4 is slightly higher than that of VM-6 even though the band gaps of VM-4 and VM-6 are almost the same. When n(V)/n(Mo) is too high, the excess V₂O₅ on the MoO₃ surface may cover active sites ^[41], which is why VM-1 shows lower photocatalytic activity than VM-2. Overall, the suitable n(V)/n(Mo) of VM-2 resulting in a narrow band gap and large SSA are responsible for its high photocatalytic activity.

To investigate the photocatalytic performance of the composites in greater detail, we examined their ability to photodegrade DMP in water. It has been reported that one of the main intermediates formed in the photodegradation of DMP is phthalic acid ^[2]. Fig. 6 shows the generation rate of phthalic acid over different samples under visible-light irradiation. The self-degradation of DMP does not occur under the same conditions. VM-2 shows higher photocatalytic activity than pure MoO₃, V₂O₅, and VM-6. This further confirms that the heterostructures have excellent photocatalytic activity, and can effectively photodegrade organic species in water. Among the samples, VM-2 shows the highest photocatalytic activity, with its degradation efficiency of MB reaching up to 89.23%. The composites also show excellent catalytic performance in the photodegradation of DMP in water. As discussed above, the V₂O₅/MoO₃ composites exhibit better photocatalytic performance than pure MoO₃ and V₂O₅, which is attributed to a synergistic effect between doping and heterostructures: the dopant effectively narrows the band gap of MoO₃ and heterostructures improve the separation rate of photogenerated electron-hole pairs.

Fig. 6. Generation rates of phthalic acid over (1) MoO3; (2) V2O5; (3) VM-6; (4) VM-2.

The XRD results indicated that V was doped into the crystal lattice of $Mo{{O}_{3}}\left( {{V}_{2}}{{O}_{5}}\underrightarrow{2Mo{{O}_{3}}}2V_{Mo}^{,}+{{V}_{{\ddot{O}}}}+5{{O}_{\text{o}}} \right)$ to different degrees in the samples; XPS results revealed that V ions exist in multiple valence states. The existence of a V 3d impurity level narrows the band gap of pure MoO₃, shortening the transmission distance of charged particles ^{[25, 26]}. Both the electron transition from the valence band (O2p) to the t_2g level of the V3d orbitals and the d-d transition of V may be driven by absorption of visible light ^[44]. This is the reason why all of the V₂O₅/MoO₃ composites show good photocatalytic performance under visible-light irradiation.

Because V₂O₅ particles are in close contact with the V-doped MoO₃ flakes, staggered energy levels may form between the heterojunctions. According to the known mechanism of electronic transmission and a previous report ^[33], we propose a possible photocatalytic mechanism for V₂O₅/V-doped MoO₃ particles, which is outlined in Fig. 7. The potentials of the CB and valence band (VB) of the semiconductors in Fig. 7 can be evaluated by Equation (3) ^[45].

Fig. 7. Photocatalytic mechanism of V2O5/V-doped MoO3 composites.

where X is the absolute electronegativity of the semiconductor, E^c is the energy of free electrons on the hydrogen scale (~4.5 eV), and E_g is the band gap energy of the semiconductor.

When V₂O₅/V-doped MoO₃ particles are irradiated with visible light, the electrons on the surface of V₂O₅ and V-doped MoO₃ are excited simultaneously. The photogenerated electrons of photoexcited V₂O₅ might inject into the CB of MoO₃ of lower impurity energy level than that of V₂O₅, which results in a large number of electrons accumulating in MoO₃. The oxygen species adsorbed on the catalyst surface then accept electrons to form superoxide radical anions (^•O₂^-), hydrogen peroxide (H₂O₂), and hydroxyl radicals (^•OH) ^[4]. Meanwhile, the holes generated in the VB of MoO₃ can be easily transferred into the VB of V₂O₅ because the VB potential of V₂O₅ is higher than that of MoO₃. Overall, this promotes the separation of electron-hole pairs and lowers the probability of electron-hole recombination. Furthermore, the accumulated holes could facilitate formation of ^•OH ^[22]. ^•O₂^- and ^•OH are known to be mainly responsible for the photocatalytic degradation of organic species, and they reacted with MB or DMP adsorbed on the catalyst surface to produce CO₂, H₂O and other substances ^{[25, 37]}. The specific catalytic processes are as follows:

The MB adsorbed on the catalyst surface can also act as a photosensitizer, absorbing visible light and transferring electrons to the CB ^[46]. This may be the reason why the photocatalytic degradation efficiency of MB is higher than that of DMP over the catalysts.

V₂O₅/MoO₃ composite photocatalysts have been prepared by a simple process. The resulting composites include both doping and heterostructures, which enhance their photocatalytic performance. Results indicate that the photocatalytic activity of the V₂O₅/MoO₃ composites is higher than that of pure MoO₃ for the degradation of MB and DMP under visible-light irradiation. The optimal n(V)/n(Mo) is 1/2, and the degradation efficiency of MB over the catalyst with this ratio reached 89.23%. The catalysts also showed excellent catalytic performance for the photodegradation of DMP.