At present, the rapid consumption of natural resources and the intensification of environmental pollution have seriously affected the society development as well as ecosystem and human health [1]. However, conventional physicochemical technologies are not so effective in removal of these environmental concerns [2, 3]. Therefore, exploring advanced strategies to ease this dilemma is imperative. As typical environmentally friendly technology, several kinds of advanced oxidation processes (AOP) technologies have been extensively used in various fields for pollutant removal. In particular, semiconductor-based photocatalysis was recognized as the most promising one because of its high efficiency, economic feasibility, and successful utilization of solar energy [4]. Whereas, many traditional semiconductors were ultraviolet-light-responsive (ULD) photocatalysts, which could only be excited by ultraviolet light (a small fraction of the solar spectrum, 2%~4%) [5]. Besides, the rapid recombination of the photo-generated charge carriers in the single-phase photocatalyst also restrict its photocatalytic (PC) activity to a certain extent. Thus, the search of suitable photocatalysts with strong visible light response, high photo-generated carrier separation efficiency and good stability has been an intensively pursued topic [6].

Due to the suitable band gap, easy preparation, good chemical tenability and low-cost, graphitic carbon nitride (g-C₃N₄) has received increasingly attention as one of the most promising metal-free photocatalysts [7]. Although the conduction band (CB) of g-C₃N₄ is sufficiently negative, the PC performance of g-C₃N₄ was far from satisfactory with less positive valance band (VB) and fast recombination of photo-generated electrons and holes. To solve this problem, various attempts have been investigated to improve the visible light photocatalytic efficiency of g-C₃N₄ such as doping with non-metals [8], coupling with metals [9], and combining with semiconductors [10, 11]. In particular, semiconductor composite photocatalyst, which construct a heterojunction interface between different components with matching energy levels, could facilitate the interfacial charge transfer and enhance the separation of photo-generated carries [3]. However, the charge carrier migration in the common p-n heterojunction could reduce both the oxidizing power of holes and the reducing power of electrons, thereby limiting the further exercise of PC activity. Recently, numerous photocatalysts based on Z-scheme heterojunction (e.g., LaCoO₃/g-C₃N₄ [12], MoO3/Bi₂O₄ [13], WO₃/MoS₂ [14], etc.) have attracted great attention in the field of photocatalysis. It has been accepted that the newly proposed Z-scheme heterojunction could avoid this shortcoming, in which the migration of electrons is from the less negative CB of one semiconductor to the less positive VB of the other semiconductor, thus simultaneously prolonging the lifetime of the more reactive charge carriers and preserving the strong redox ability of remaining electrons and holes [15, 16].

In principle, the construction of band match, effective intimate contact interfaces and improved carrier separation efficiency are crucial to efficient Z-scheme photocatalytic system [17]. As an important metal-oxide semiconductor, bismuth(Ⅲ) oxide (Bi₂O₃) was usually applied as a sensitizer to realize the visible light response [18-20]. By now, six polymorphic forms of Bi₂O₃ have been identified and reported, containing two stable phases (α- and δ-phases) and other four metastable phases (β-, γ-, ε- and ω-phases). Among those polymorphs, α-Bi₂O₃ has been most frequently studied because of its thermal stability and environmental benign characteristics [21]. Notably, benefitting from a less negative CB of Bi₂O₃, which is close to the VB of g-C₃N₄, Bi₂O₃ was considered as a promising candidate for coupling with g-C₃N₄ to construct Z-scheme heterojunction. Moreover, the VB of g-C₃N₄ is about 3.13 eV, positive enough to prompt the photo-generated holes on the VB of Bi₂O₃ to exhibit strong oxidation capacity. So far, there have been some reports on the fabrication of Bi₂O₃/g-C₃N₄ composites for photocatalytic purposes [22]. α-Bi₂O₃ was usually prepared by calcining Bi(NO₃)₃5H₂O crystal, and the morphology of the synthesized α-Bi₂O₃ were mostly granular-shaped or rodshaped. As far as we know, the related research on the control of the synthesis of α-Bi₂O₃ by introducing precursors has not been reported. Additionally, different preparation methods could obtain Bi₂O₃ samples with different polymorphic forms or different morphology. Consequently, according to previous study, BiOCOOH with Sillen structure has been testified to show satisfying degradation activity under simulated sunlight [23, 24]. Despite this, pristine BiOCOOH still suffers from the rapid electron-hole recombination rate, which undermines the photocatalytic performance and restrains the practical application. In this study, we applied BiOCOOH as precursor to obtain α-Bi₂O₃. Besides, since temperature responsible for the morphology, structure and other physicochemical properties of catalysts, and the PC performance of catalysts with different physicochemical characterization vary greatly [25]. Therefore, to explore a new idea of Bi₂O₃ preparation, we optimize the calcination temperature for the photocatalyst synthesis based on BiOCOOH.

Therefore, in this study, α-Bi₂O₃ and g-C₃N₄ were employed to construct a Z-scheme composite photocatalyst through calcination of BiOCOOH/g-C₃N₄ precursors at different temperature. The α-Bi₂O₃/g-C₃N₄ nanocomposite photocatalyst exhibited excellent photocatalytic performance. Subsequently, the effects of various calcination temperatures on the physicochemical properties and photocatalytic efficiencies of as-obtained samples were systematically investigated (Text S1 in Supporting information). Finally, a possible photocatalytic mechanism of the degradation of amido black 10B dye over as-prepared Z-scheme α-Bi₂O₃/g-C₃N₄ composite photocatalyst was proposed.

The information of chemicals which were applied in this study were introduced in Text S2 (Supporting information).

Pure g-C₃N₄ powder was prepared by calcination with dicyandiamide as precursor. Typically, dicyandiamide (3 g) powder was put into an alumina crucible and calcined at 550 ℃ for 4 h with a rate of 10 ℃/min. The resulting yellow product was collected and ground into powders for further use. BiOCOOH/g-C₃N₄ precursor was prepared by a hydrothermal method as follow: 1 mmol Bi(NO₃)₃·5H₂O was dissolved in 5 mL DMF by stirring. Subsequently, the solution was added into 40 mL deionized water containing 1 mmol g-C₃N₄ powder and ultrasonication for 10 min to ensure that all of the reagents were dispersed homogeneously. Afterwards, the resulting solution was transferred into the Teflon-lined autoclave, which was maintained at 120 ℃ for 12 h and then cooled to room temperature naturally. The obtained precipitates were collected and washed for several times with deionized water and absolute ethanol, respectively. Finally, the products were dried at 60 ℃. At this point, the precursor BiOCOOH/g-C₃N₄ was obtained. The pure BiOCOOH was synthesized following the above procedures without the addition of g-C₃N₄. For the preparation of the α-Bi₂O₃/g-C₃N₄ composites, the BiOCOOH/g-C₃N₄ precursor was put in a covered crucible and calcined at a certain temperature (T = 300, 400, 450, 500 ℃) for 3 h. The final as-obtained samples with different calcination temperature (300, 400, 450, 500 ℃) were respectively denoted as BC-300, BC-400, BC-450, and BC-500. The B-400 sample is the one synthesized similarly but without the addition of g-C₃N₄.

In this work, amido black 10B dye was used as representative pollutants to estimate the PC activity of the prepared samples. The PC reaction was carried out at room temperature in 80 mL quartz reactor containing 20 mg as-prepared photocatalysts and 50 mL amido black 10B aqueous solution of 10 mg/L. A 35 W xenon lamp with light density of 67 mW/cm² was served as the external light source. Prior to irradiation, the suspensions were magnetically stirred in the dark for about 60 min to ensure the establishment of an adsorption-desorption equilibrium. Subsequently, the xenon lamp was switched on. At given time intervals, certain amount of the solution after taking from the reactor and filtering through a 0.45 μm membrane filter were measured at the characteristic wavelength of 613 nm using an Evolution 300 UV–vis spectrophotometer.

Fig. 1 displayed the XRD spectra of g-C₃N₄, BiOCOOH, BiOCOOH/g-C₃N₄ and the calcination products of BiOCOOH/g-C₃N₄. Obviously, pure g-C₃N₄ sample (Fig. 1a) showed two typical diffraction peaks at 13.09° and 27.47, corresponding to the (100) and (002) crystal planes of g-C₃N₄ (JCPDS No. 87-1526), respectively [26]. All the diffraction peaks of the as-prepared BiOCOOH sample could be readily indexed to the tetragonal BiOCOOH (JCPDS No. 35-0939), which was in good agreement with previous reported literatures [27]. Only the characteristic peaks assigned to BiOCOOH were observed in the BiOCOOH/g-C₃N₄ composite, which might be ascribed to the low crystallinity of g-C₃N₄. Similar phenomenon was also observed in g-C₃N₄/Bi₄O₇ [28] and BiOIO₃/g-C₃N₄ heterostructured composite [3]. Fig. 1b exhibited the XRD patterns of the samples after the 300-500 ℃ calcination of the precursor BiOCOOH/g-C₃N₄ composite. Obviously, the sample obtained after 300 ℃ calcination was Bi₂O₂CO₃/g-C₃N₄ composite because the main detectable diffraction peaks of BC-300 could be matched with the tetragonal phase of Bi₂O₂CO₃ according to the PDF card (JCPDS NO. 41-1488) [29] and the small peak at 27.4° represented the (002) planes of g-C₃N₄. The generation of Bi₂O₂CO₃ was due to the following reaction between BiOCOOH and O₂ at 300 ℃ (Eq. 1):

(1)

However, after calcination at 400 ℃, the diffraction peaks of Bi₂O₂CO₃ was disappeared and the major characteristic peaks at around 24.58°, 25.75°, 26.90°, 27.40°, 28.01°, 33.25°, 35.03°, 37.63°, 39.94°, 42.11°, 44.92°, 46.37°, 48.60°, 52.20°, 54.78°, 55.43°, 57.72°, 63.58° and 71.40° should be attributed to the (-102), (002), (-112), (-121), (012), (-202), (-212), (-113), (-222), (-123), (023), (041), (-104), (-322), (-241), (-224), (024), (-115) and (-161) lattice planes of monoclinic α-Bi₂O₃ (JCPDS No. 71-2274) [30]. This illustrated that the α-Bi₂O₃/g-C₃N₄ composite was obtained successfully. This phenomenon could be explained by the fact that when the calcination temperature reached to 400 ℃ or higher, BiOCOOH would be constantly oxidized by fully reaction with the O₂ in the air atmosphere and eventually transformed to Bi₂O₃. The characteristic peak of g-C₃N₄ was not detected in the α-Bi₂O₃/g-C₃N₄ composite sample because of the overlapping between the sharp peaks of (120) crystal plane of α-Bi₂O₃ at 27.41° and (002) crystal plane of g-C₃N₄ at 27.47° [31]. Besides, as for B-400, there was the detection of monoclinic α-Bi₂O₃ without other impurity peaks, proving that α-Bi₂O₃ could be controllably synthesized through the calcination of BiOCOOH. Notably, the XRD peaks intensity of α-Bi₂O₃ decreased and the diffraction peaks appeared to be blunt as the calcination temperature rose from 400 ℃ to 500 ℃, which indicated that the increase of temperature would affect the crystallinity of the α-Bi₂O₃/g-C₃N₄ composite.

Fig. S1 (Supporting information) showed the FT-IR spectra of g-C₃N₄, B-400, BC-300, BC-400, BC-450 and BC-500 composites, in which the surface functional groups and bonding methods of the as-prepared samples were further explored. For g-C₃N₄ sample, the sharp peak in 808 cm^-1 could be assigned to the typical breathing vibration of the triazine unites [32]. The strong band from 1200 cm^-1 to 1600 cm^-1 were corresponding to the typical stretching modes of C-N heterocycles. Typically, the bands at 1241 cm^-1 and 13, 234 cm^-1 were ascribable to stretching vibration of connected units of C-N(-C)-C (full condensation) or C-NH-C (partial condensation). While those at 1406, 1458, 1571, and 1641 cm^-1 were due to typical stretching vibration modes of heptazinederived repeating units [33]. In addition, the broad band located at 3000~3500 cm^-1 were caused by the stretching vibration of the N-H, indicating the g-C₃N₄ obtained was of incomplete polycondensation [34]. For α-Bi₂O₃, it was known that absorption bands at 423 cm^-1 and 844 cm^-1 were due to the symmetrical stretching vibration of Bi-O bond of BiO₃ species [35]. Whereas, the peak at 504 cm^-1 was due to the stretching vibration of Bi-O bonds of BiO₆ octahedral [36]. Furthermore, the peak at 1390 cm^-1 was attributed to Bi-O stretching vibration [33]. It was clear that the FT-IR spectrum of B-400 shows the characteristic bands of α-Bi₂O₃, suggesting the generation of α-Bi₂O₃, in consistent with XRD. As for the BC-400, BC-450, and BC-500 samples, the typical bands of α-Bi₂O₃ and g-C₃N₄ could be observed, which could be fully demonstrated that α-Bi₂O₃/g-C₃N₄ can be obtained by calcination of BiOCOOH/g-C₃N₄ precursors. It was worth mentioning that with the increase of temperature, the spectra display clear vibration modes of α-Bi₂O₃. However, the band vibration of g-C₃N₄ could not be obviously observed. This may be due to the fact that high temperature might break the vibration mode of g-C₃N₄ as well as its response of FT-IR.

The morphology and microstructure of as-prepared samples were investigated by SEM and TEM measurement. Fig. 2 showed the SEM images of the pure g-C₃N₄, BiOCOOH/g-C₃N₄ precursors and α-Bi₂O₃/g-C₃N₄ composites obtained at different calcination temperature. The pure g-C₃N₄ (Fig. 2a) displayed a typically aggregated morphology with a number of micrometer size blockbased flakiness and particles. Without calcination, the BiOCOOH/g-C₃N₄ precursor (Fig. 2b) exhibited a flower-like microsphere. The close-packed microflowers consisted of two-dimensional thin flakes with thickness of 30~50 nm and edge lengths of 1~2 μm. Besides, many agglomerates of g-C₃N₄ with a diameter of 0.5~1 μm irregularly attached on the surface of thin flakes. After 300 ℃ calcination (Figs. 2c and d), the surface of the precursor layer became matte, which might be attributed to the fact that a large amount of small molecular gas, such as CO₂ and H₂O, were released from the precursor during calcination. Consequently, the numerous small pores appeared on the surface of the flower-like microspheres. At this point, the thickness of petals layer were in range of 46~82 nm. Seen from the Figs. 2e–j, when the calcination temperatures were between 400~500 ℃, the flower-like microsphere morphology of precursors were basically maintained in the asobtained samples. However, the thickness of the lamella that made up the microflowers increased obviously, while the spherical particles gradually formed and adhered to its surface. This phenomenon could be explained as the increase of calcination temperature leads to the collapse of laminar structure. Notably, as the calcination temperature increases, the diameter of the spherical particles became larger and the surface became smoother. Interestingly, the rod-like structure with a diameter of 400 nm and length of 600 nm even appeared in the samples after calcination at 500 ℃.

Fig. S2a (Supporting information) exhibited the TEM images of as-obtained BC-400 composite which showed a stackmorphology. The large black area represented the α-Bi₂O₃, while a few bits of grey agglomerates with irregular morphology were belonged to g-C₃N₄. The corresponding HR-TEM image (Fig. S2b in Supporting information) further displayed the existence of the heterojunction between g-C₃N₄ and α-Bi₂O₃, and the lattice fringes of 0.324 nm could be well agreed with the spacing of (120) planes of α-Bi₂O₃ [37]. It is obvious that there was the formation of heterojunction between g-C₃N₄ and α-Bi₂O₃ in BC-400 composite.

To further examine the surface chemical state and composition of α-Bi₂O₃/g-C₃N₄ composite, XPS analysis was employed. Fig. S3a (Supporting information) displayed the full-scan XPS spectra of BC-400 sample, wherein C, N, O, Bi species were detected. As illustrated in Fig. S3b (Supporting information), two peaks at 284.8 and 288.1 eV could be observed in the highresolution spectrum of C 1s. The characteristic peaks could be assigned to carbon atoms (C-C bonding) in a carbon environment; i.e., amorphous carbon adsorbed on the surface and carbon atoms bond in N-containing aromatic rings (N-C=N) in the pure g-C₃N₄ and BC-400 composite [37]. The high-resolution spectrum of N 1s (Fig. S3c in Supporting information) could be described as the superposition of three peaks located at 398.5, 399.9 and 401.1 eV. The binding energies at 398.5 and 401.1 eV were corresponded to sp²-bonded N in the triazine rings (C-N=C) and amino groups (C-N-H), respectively. While the peak at 399.9 eV belongs to the tertiary N in the form of N-(C)₃ group. Apart from these, there was a weakest peak at 404.3 eV (π-excitations) could be detected in pure g-C₃N₄, but the binding energies shifted to 404.8 eV in the BC-400 composite. It implied that α-Bi₂O₃ affected the π-excitations and coupled with g-C₃N₄ via an interaction using π-electrons of CN heterocycles, rather than combined with sole C or N atoms of gC₃N₄ [38]. As for the Bi 4f XPS spectrum (Fig. S3d in Supporting information), two binding energies at 158.7 and 164.1 eV of BiOCOOH/g-C₃N₄ were attributed to Bi 4f_7/2 and Bi 4f_5/2, respectively. It demonstrated that Bi species existed in the forms of Bi³⁺ in composite [27]. Notably, the binding energies of Bi in BC-400 exhibited a negative shift relative to those of BiOCOOH/gC₃N₄, which could be attributed to the generation of α-Bi₂O₃ as well as the interaction of g-C₃N₄ with α-Bi₂O₃, resulting in inner shift of Bi 4f orbits. And the peaks at 158.5 and 163.8 eV were respectively corresponded to the Bi 4f_7/2 and Bi 4f_5/2 of α-Bi₂O₃. Similarly, the peaks of O 1s (Fig. S3e in Supporting information) were also changed, and the binding energy at 529.5 eV could be indexed to the lattice oxygen of Bi-O bonds in [Bi₂O₂]²⁺ slabs in BiOCOOH, while peak at 529.2 eV was characteristic of Bi-O binding energy in Bi₂O₃. The slight in lower binding energy could be attributed to the increased electron density in α-Bi₂O₃/g-C₃N₄ [39]. Besides, another two peaks with the binding energy of 531.2 and 532.9 eV were attributed to the hydroxyl groups or carboxyl groups [40]. In summary, the XPS results indicated that both gC₃N₄ and α-Bi₂O₃ existed in the BC-400 composite. And all the observed shifts for binging energy were ascribed to the intense interaction between α-Bi₂O₃ and g-C₃N₄, further confirming the establishment of the heterojunctions.

In order to explore the porous properties of as-prepared samples, the BET surface area and pore size distribution were measured via N₂ adsorption-desorption isotherms method. According to the IUPAC classification, the N₂ adsorptiondesorption isotherms curve of all samples could be classified as type IV isotherms as shown in Fig. S4 (Supporting information), indicating that the sample have mesoporous structure [22]. Besides, the adsorption isotherms were below the desorption isotherm corresponding to the type H₃ hysteresis loop, and this adsorption hysteresis was usually attributed to the size and shape of pores [41]. The corresponding pore size distributions of samples were shown in the inset of Fig. S4, which further indicated the presence of mesoporous, and pore size of mesoporous were not uniform ranging from 5 nm to 60 nm. The BET surface area, pore volume and pore diameter of as-prepared samples were displayed in the Table S1 (Supporting information). It can be seen that the BET specific surface area of α-Bi₂O₃/g-C₃N₄ calcined by 400, 450 and 500 ℃ were 8.149, 7.214, and 4.632 mg₂/g, respectively, which were significantly smaller than that of BiOCOOH/g-C₃N₄. And the specific surface area of resulting α-Bi₂O₃/g-C₃N₄ composite gradually decreased with the increase of temperature.

The optical properties of as-prepared catalysts, which were important indicator for evaluating PC performance [42], were studied and illustrated via UV–vis DRS. As illustrate in Fig. S5a (Supporting information), it was observed that the steep absorption edge of B-400 was about 447 nm, and the pure g-C₃N₄ exhibited absorption wavelengths from the UV to the visible range up to 464 nm. However, compared with pure g-C₃N₄, the absorption wavelength regions of the BC-X composites showed slight blue shift, except for BC-300 sample. It might be attributed to the interaction between the g-C₃N₄ and α-Bi₂O₃ and smaller specific surface area of α-Bi₂O₃/g-C₃N₄ composite.

As is well-known that for a crystalline semiconductor, the band gap energy (E_g) could be calculated by the following Kubelka-Munk transformation (Eq. 2) [22]:

(2)

where α, h, v, A, and E_g are absorption coefficient, Planck constant, light frequency, proportionality and band gap energy, respectively. n is corresponding to the properties of the transition in a semiconductor (n = 1 for direct transition and n = 4 for indirect transition). For g-C₃N₄ and Bi₂O₃, the values of n are 4 and 1, respectively. By applying this equation, the band gap of g-C₃N₄ and B-400 is 2.71 and 3.48 eV, respectively, which agrees with the previous reports [37]. The result was illustrated in Fig. S5b (Supporting information). Furthermore, the band position of photocatalyst could be also calculated by the following empirical formulas (Eqs. 3 and 4) [43]:

(3)

(4)

where E_VB is the valence band potential; E_CB is the conduction band potential; X is the absolute electronegativity of the semiconductor; E^e is the energy of free electrons on the hydrogen scale (E^e = 4.5 eV). The X values for g-C₃N₄ and Bi₂O₃ are 4.73 and 6.23 eV, respectively [22, 44]. By calculation, the E_VB and E_CB of g-C₃N₄ were about 1.58 and -1.13 eV, respectively, while the E_VB and E_CB of g-C₃N₄ were about 3.13 and 0.33 eV, respectively.

Photoelectrochemical measurements were employed to further qualitatively confirm the enhanced migration and separation of photo-generated charge carries in this study. Fig. 3a performed the transient photocurrent-curves of g-C₃N₄, B-400 and BC composites obtained by calcination at different temperature with typical onoff cycles under 35 W xenon lamp irradiation. Generally, the higher photocurrent intensity means higher separation efficiency of photo-generated charge carriers, which might cause superior PC activity. Obviously, the photocurrents of all composite samples were significantly higher than that of pure samples, which indicated that the heterojunction structure formed by combination of α-Bi₂O₃ and g-C₃N₄ was conducive to the effective separation of electron-hole pairs. It is noteworthy that the BC-400 sample was the most prominent one among them. It performed an enhanced transient photocurrent response with a photocurrent density of 1.0 μA/m², which was nearly 6.67 folds higher than that of pure α-Bi₂O₃ and g-C₃N₄ (0.15 μA/m²) under same condition. This revealed that a smaller recombination and a more efficient separation of photo-generated charge carries occurred at the interface between α-Bi₂O₃ and g-C₃N₄ in BC-400 nanocomposite. Moreover, all photocatalysts showed good repeatability and stability under several on-off cycles of intermittent illumination, indicating that the electrodes were stable and the responses were reversible. Besides, EIS measurement conducted at the open circuit potential with 35 W xenon lamp illumination was carried out to clarity the electrical conductivity of all samples. As shown in Fig. 3b, BC-400 sample exhibited the smallest arc radius on the EIS plots, indicating that the minimum charge transfer resistance existed in the BC-400 composite, which was conducive to the separation and migration of photo-generated charge carriers [45]. This phenomenon was well identical to the results of transient photocurrent measurement. Moreover, Mott-Schottky (M-S) plots of the g-C₃N₄ and BC-400 composite were carried out in the dark with a frequency of 1000 Hz. As demonstrated in Fig. 3c, the positive slopes of the plots revealed that g-C₃N₄ and BC-400 composite were n-type semiconductor. And the plots were extrapolated to 1/C² = 0 to estimate the values of flat band potential (E_FB), giving -0.32 and -0.57 V for g-C₃N₄ and BC-400 composite, respectively. The negative shift of E_FB in BC-400 composites demonstrated an increase in bending of the band edge after combining α-Bi₂O₃ with g-C₃N₄, which was benefit for the separation and transportation of photo-generated charge carriers. Based on the above discussion, the enhanced PC performance of α-Bi₂O₃/g-C₃N₄, especially BC-400 composite, could be anticipated.

In this work, amido black 10B dye was served as representative organic pollutants to evaluate the PC activities of a series of BC composites obtained at different calcination temperature. Meanwhile, the PC performances of pure g-C₃N₄ were also measured as references. As depicted in Fig. 4a, the degradation of amido black 10B could be almost neglected in the absence of photocatalysts after 120 min of reaction through blank test, which indicated that amido black 10B was a stable dye molecule with weak selfphotolysis. The PC degradation of pure g-C₃N₄ for amido black 10B was 62.89%, which was lower than that of composite photocatalysts. However, the degradation rate was significantly enhanced when α-Bi₂O₃ was used to combine with g-C₃N₄ to prepared nanocomposite photocatalyst. And the PC efficiency of α-Bi₂O₃/g-C₃N₄ composts with different calcination temperature (400~500 ℃) were 97.09%, 95.97%, and 85.62%, respectively. Apparently, α-Bi₂O₃/g-C₃N₄ photocatalyst that was calcined at 400 ℃ exhibited the highest PC efficiency, which mainly ascribed to the intense visible light absorbance and high separation efficiency of electron and hole pairs. However, the PC activity decreased as further increase of calcination temperature. Combined with the analysis of SEM, this phenomenon could be interpreted that the calcination temperature has a dramatically impact on the morphology of catalysts. In detail, the higher calcination temperature was, the more obvious the agglomeration would be formed. More importantly, excessive agglomeration of nanocomposite would be unfavorable to the charge transfer, thereby weakening the PC efficiency [46]. This was accordance with the above results of the PECH characterization.

Furthermore, kinetics of PC degradation of amido black 10B dye under 35 W xenon lamp light irradiation over all as-fabricated catalysts were investigated by applying the Langmuir-Hinshelwood (L-H) model with 120 min. After fitting, the plots -ln(C/C₀) versus irradiation time t were found to be linear as displayed in Fig. 4b, implying that the photodecomposition reactions followed the pseudo-first order kinetics. The apparent reaction rate constants (k) of pure g-C₃N₄, BC-300, BC-400, BC-450, and BC-500 samples were determined to be 0.00751, 0.01958, 0.02661, 0.02501, and 0.01578 min^-1, respectively. Obviously, compared with other four samples, BC-400 present the best photocatalytic performance and held the highest rate constant.

Distinguishing the construction of each primary reactive species to the PC activity was critical to the analytical of PC mechanism. Thus, reactive species trapping experiments was carried out during the PC degradation processes, in which ammonium oxalate (AO), isopropanol (IPA) and benzoquinone (BQ) were selected as the quenchers of h⁺, ^·OH and O₂^·-, respectively [46]. Fig. S6 (Supporting information) showed the effects of different scavengers on the photocatalytic degradation of amido black 10B over BC-400 photocatalyst after 120 min illumination. It could be obviously found that the addition of IPA has a little effect on the PC performance of BC-400, which indicated that OH played a minor role in the photocatalytic reaction. On the contrary, the PC degradation efficiency was significantly decreased when the AO and BQ were added, demonstrating both h⁺ and O₂ were the main active species in the degradation of amido black 10B under visible-light irradiation.

As suggested above, a possible photocatalytic degradation mechanism of amido black 10B by the α-Bi₂O₃/g-C₃N₄ composite photocatalyst under visible light irradiation was illustrated in Fig. S7 (Supporting information). When α-Bi₂O₃ and g-C₃N₄ were combined with each other, interfacial heterojunction structure would be formed because of their firm contacting. And owing to the narrow band gap and good light absorption abilities of α-Bi₂O₃ and g-C₃N₄, both of them could be easily excited to generate photoinduced electron-hole pairs under visible light illumination. Since the E_CB of g-C₃N₄ (-1.12 eV vs. NHE) was more negative than the standard redox potential of O₂/O₂^·- (-0.33 eV vs. NHE) [31], the electros in the CB of g-C₃N₄ would be available to reduce the surface chemisorbed O₂ to produce O₂^·-, which could decompose the amido black 10B into other produces and some small molecules. However, the holes in the VB of g-C₃N₄ could not bond with water to oxide OH to give ^·OH due to the lower VB potential of g-C₃N₄ (1.58 eV vs. NHE) than the standard redox potential of ^·OH/OH (2.4 eV vs. NHE). This explicated why OH was not the main reactive species in the photocatalytic process. Similarly, the electrons in the CB of α-Bi₂O₃ also could not reduce O₂ to give O₂^·- because of its more positive CB potential than that of E₀ (O₂/O₂^·-). Therefore, combined with the results of reactive species trapping experiments, the transfer and recombination of photo-generated electros and holes in α-Bi₂O₃/g-C₃N₄ followed the Z-scheme modes, rather than the most common composite photocatalysts [47]. Consequently, when the α-Bi₂O₃/g-C₃N₄ composite was illuminated, photo-generated electrons in the CB of α-Bi₂O₃ would rapidly migrate and recombine with holes in the VB of g-C₃N₄ to form a direct Z-scheme charge separation path. Subsequently, the holes accumulated on the VB of α-Bi₂O₃ would be consumed to directly oxidize amido black 10B. In such a way, the photo-generated charge carries between the α-Bi₂O₃ and g-C₃N₄ could be effectively separated and involved in photocatalytic reactions, thereby resulting in an enhanced visible light PC performance of α-Bi₂O₃/g-C₃N₄ composite. Furthermore, all the processes of photodegradation of amido black 10B over α-Bi₂O₃/gC₃N₄ composite could be described as follows (Eqs. 5–8):

(5)

(6)

(7)

(8)

The stability and reusability were important parameter for photocatalysts in the application of wastewater treatment. Thus, in order to ascertain the stability and practicability of as-prepared α-Bi₂O₃/g-C₃N₄, BC-400 sample which exhibited the best photocatalytic activity was chosen for stability test. As shown in Fig. S8a (Supporting information), the BC-400 composite still remained high photocatalytic efficiency for amido black 10B degradation even after consecutive utilization for four recycles, suggesting that it has prominent photochemical stability. Besides, the XRD pattern of used BC-400 composite (Fig. S8b in Supporting information) also demonstrating a very stable crystal structure of it because of no extra diffraction peaks were observed after four times reuse. At this point, we can reasonably conclude that α-Bi₂O₃/g-C₃N₄ composite has superior stability, which could be used as the promising candidate in wastewater treatment.

In summary, we combined α-Bi₂O₃ with g-C₃N₄ and synthesized direct Z-scheme α-Bi₂O₃/g-C₃N₄ composite photocatalyst through facile calcination of BiOCOOH/g-C₃N₄ at different thermal treatment temperatures. The results of various characterization indicated that synthesized temperature has effects not only on the morphologies and structure of α-Bi₂O₃/g-C₃N₄ composite, but also on its optical and photoelectrochemical properties. With comparison of g-C₃N₄, the prepared α-Bi₂O₃/g-C₃N₄ composite photocatalyst exhibited much higher photocatalytic activity. Particularly, BC-400 composite displayed the highest transient photocurrent response with a photocurrent density of 1.0 mA/m² as well as the best PC efficiency towards the degradation of amido black 10B, which was approximately 1.54 folds than that of pure gC₃N₄, respectively. Besides, the reactive radical species trapping experiment confirmed that both h⁺ and O₂^·- were the main active species in the photocatalytic process, while ^.OH played a relatively small role. The improved PC activity of α-Bi₂O₃/g-C₃N₄ composite could be mainly attributed to the construction of Z-scheme heterojunction between α-Bi₂O₃ and g-C₃N₄, giving rise to efficient separation of the photog-generated charge carriers and strong redox ability for photocatalytic degradation of organic dye solution. In addition, the good stability and reusability of asprepared α-Bi₂O₃/g-C₃N₄ composite photocatalyst further indicated that it could be used as the promising candidate for application in wastewater treatment.

Declaration of competing interest

We have no conflicts of interest to declare.

Acknowledgments

This work was kindly supported by the National Natural Science Foundation of China (Nos. 51508254, 51978319), Fundamental Research Funds for the Central Universities (No. lzujbky-2017-it98), College Students' Innovative Practice Training Program of Chinese Academy of Sciences (No. Y710171040) and Key Laboratory of Comprehensive and Highly Efficient Utilization of Salt Lake Resources, Qinghai Institute of Salt Lakes, Chinese Academy of Sciences.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.cclet.2020.04.026.