Constructing Bi24O31Cl10/BiOCl heterojunction via a simple thermal annealing route for achieving enhanced photocatalytic activity and selectivity

This work reports on the construction of a Bi24O31Cl10/BiOCl heterojunction via a simple thermal annealing method. The X-ray diffraction (XRD) results indicated that the phase transformation from BiOCl to Bi24O31Cl10 could be realized during the thermal annealing process. The high-resolution transmission electron microscopy (HRTEM) images, X-ray photoelectron spectroscopy (XPS) binding energy shifts, Raman spectra and Fouier transform infrared spectroscopy (FT-IR) spectra confirmed the formation of the Bi24O31Cl10/BiOCl heterojunction. The obtained Bi24O31Cl10/BiOCl photocatalyst showed excellent conversion efficiency and selectivity toward photocatalytic conversion of benzyl alcohol to benzaldehyde under visible light irradiation. The radical scavengers and electron spin resonance (ESR) results suggested that the photogenerated holes were the dominant reactive species responsible for the photocatalytic oxidation of benzyl alcohol and superoxide radicals were not involved in the photocatalytic process. The in-situ generation of Bi24O31Cl10/BiOCl heterojunction may own superior interfacial contact than the two-step synthesized heterojunctions, which promotes the transfer of photogenerated charge carriers and is favorable for excellent photocatalytic activities.

Regarding the future environmental and energy concerns, the development of green and sustainable chemical conversions has attracted enormous interest [1][2][3] . Alcohol oxidations are one of the most frequently investigated reactions because of their industrial essentiality in the commercial synthesis of multifarious materials, such as plastics, perfumes, paints, etc [4][5][6] . Compared with conventional methods, photocatalytic technology is considered to be a green, reliable and economic method for the oxidation of alcohols into the corresponding aldehydes due to the massive solar energy and O 2 [7][8][9][10][11] . Semiconductor titanium dioxide (TiO 2 ) is universally regarded as an efficient photocatalyst toward decomposition of various organic pollutants [12][13][14][15][16][17] . Moreover, it also displays photocatalytic activity toward the oxidation of benzyl alcohol to benzaldehyde under UV-light and visible-light irradiation, which shows high conversion efficiency (> 99%) and selectivity (> 99%) 18,19 . Recently, considerable attention has been devoted to another series of semiconductors, the bismuth-based semiconductors. BiOCl is a V-VI-VII ternary semiconductor, consisting of internal structure of [Bi 2 O 2 ] 2+ layers sandwiched by two slabs of Cl atoms which induces the growth of BiOCl along a particular axis 20 . It often shows high photocatalytic performance than TiO 2 (P25, Degussa) under UV-light irradiation due to its unique layered atomic structure, which favors the transfer and separation of photogenerated charge carriers and subsequently enhances the photocatalytic activity 21,22 . However, BiOCl is a wide-band-gap (3.17 ~ 3.54 eV) semiconductor 23,24 , which leads to a poor photocatalytic performance under visible light irradiation.
From the viewpoint of solid state physics, details of the band edge potential are primarily determined by the static potential within the unit cell of a semiconductor 32 . Any symmetry and component perturbations can have consequence on the electronic structures and physical properties. Since the potential of conduction band minimum (CBM) and valence band maximum (VBM) are mainly related to Bi 6p and Bi 6s orbitals respectively, the regulation of CBM and VBM of bismuth-based semiconductors can be achieved by adjusting the Bi content 33 10 33 have been found to show visible light driven photocatalytic activities, which are regarded as ideal candidates for the construction of heterojunctions with BiOCl. These NSSs have narrower band gap, faster transfer of charge carriers and more efficient separation of photogenerated electron-hole pairs 37 . Furthermore, they have the approximate crystalline architecture relative to their corresponding typical stoichiometric semiconductors (TSSs). As a nontypical stoichiometric bismuth-based semiconductor, Bi 24 O 31 Cl 10 is widely known as a product of the thermal decomposition of BiOCl 38 Figure 1 displays the XRD patterns of BiOCl and the calcined samples. The XRD pattern of sample B-RT is assigned to tetragonal BiOCl (JCPDS NO. 06-0249). With an increase of annealing temperature, the XRD peaks belonging to Bi 24 O 31 Cl 10 with a monoclinic structure (JCPDS NO. 75-0887) emerges. No apparent diffraction peaks belonging to BiOCl are observed when the temperature increased up to 600 °C. The enlarged XRD patterns of all samples in the range of 2θ = 20 ~ 40° are also presented to further verify the transformation process from tetragonal BiOCl to monoclinic Bi 24 O 31 Cl 10 (Fig. 1b). A weak peak located at 32° is assigned to Bi 24 O 31 Cl 10 in XRD pattern of sample B-450. The other three typical strong peaks nearby 30° are observed in sample B-500, which indicates that large amount of Bi 24 O 31 Cl 10 is produced at reaction temperature of 500 °C. Further increase of reaction temperature induces the emergence of more diffraction peaks belonging to Bi 24 O 31 Cl 10 and all the XRD peaks belonging to Bi 24 O 31 Cl 10 phase are only left at 600 °C (B-600). On the other hand, no XRD peak of Bi 24 O 31 Cl 10 phase is observed in sample B-400, which may suggest that no observable phase transformation occurs or the Bi 24 O 31 Cl 10 does not possess sufficient long-range order to be checked by XRD. DTA-TG curves ( Figure S1) of sample B-RT shows that there is an exothermic peak at about 400 °C, suggesting that the phase transformation of BiOCl occurs as the temperature achieving to 400 °C, which result is consistent with the XRD results.  Figure S2a). After calcination at 400 °C, the nanosheet edges and angles of sample B-400 are distinct and differentiable (Fig. 2b). The gradual increased temperature leads to the morphological transformation from compact sphere to loose structure as well as irregular nanosheets to square analogs (Fig. 2c,d). Furthermore, the nanosheets of BiOCl become wider and thicker with an increase of annealing temperature. Sample B-600 (pure Bi 24 O 31 Cl 10 ) presents square-like plate structure with 1 ~ 2μ m in width and ~0.1 μ m in thickness (Figs 2f and S2b). It could also be observed that the sheet-shaped structure with narrower width and thinner thickness decreases, while the plate-shaped structure increases by elevating the annealing temperature, which result is consistent with the BET results ( Figure S3) that sample B-600 has the lower specific surface area (S BET ) than sample B-RT.  Fig. 3c) clearly presents the crystalline planes of (101) and (110) of BiOCl, respectively. Sample B-450 keeps the same diameter, but the shape of the nanosheets becomes regular (Fig. 3d,e). To further confirm the chemical state and chemical composition of the as-prepared samples, X-ray photoelectron spectroscopy (XPS) analysis was applied and the results are shown in Fig. 4. The survey scans of samples B-RT, B-450 and B-600 distinctly reveal the co-existence of Bi, O and Cl elements without other impurities, excluding adventitious carbon-based contaminant. The two primary peaks at ~159.0 eV and ~164.0 eV in Bi 4f XPS spectra result from the spin orbital splitting photoelectrons of Bi 4f 7/2 and Bi 4f 5/2 , respectively 41 . There is an obvious red-shift in the Bi 4f binding energy with increasing the temperature to 600 °C. Variations in the elemental binding energies are generally related to the difference in chemical potential and polarizability of involved elements 42,43 . Thus, the binding energy shift in sample B-450 is possibly attributed to the interaction between BiOCl and Bi 24 O 31 Cl 10 , which result is similar to the SnO 2−x /g-C 3 N 4 44 and TiO 2 /ZnPcGly 45 . It is reported that the increase or decrease in electron concentration could enhance or reduce the electron screening effect, which would weaken or strengthen the binding energy 46  . However, the span between the two binding energy peaks maintains the same value of 5.3 eV, which suggests that Bi exists in the chemical state of Bi 3+ in both BiOCl and Bi 24 O 31 Cl 10 .

Results and Discussion
The chemical compositions of Bi, Cl and O in various samples as well as the variation of Bi/Cl molar ratio as a function of annealing temperature are displayed in Fig. 4c and Table S1. As shown in Fig. 4c, there exists a monotonic increase of Bi/Cl molar ratio with an increase of annealing temperature. When the temperature increases to 600 °C, the Bi/Cl molar ratio reaches 2.295, which is very close to the theoretical value 2.4 of Bi 24 O 31 Cl 10 . This observation indicates the phase transformation from pure BiOCl to Bi 24 O 31 Cl 10 . It could be possibly accepted that if the Bi/Cl molar ratio is larger than the theoretical value of BiOCl, the phase transformation occurs. Thus, 450 °C could be recognized as the initial phase transformation temperature in our experiment, which is consistent with the XRD result.  (Fig. 5a), there are two distinguishable Raman active bands at 140 cm −1 and 198 cm −1 which are assigned to the A 1g and E g internal Bi-Cl stretching modes 47,48 , respectively. However, the band related to the motion of oxygen atoms at about 400 cm −1 36,49 is very weak and nearly unnoticeable. With an increase of annealing temperature, the Raman peak assigned to A 1g shifts to higher wavenumbers. This phenomenon could be ascribed to the formation of heterojunction between BiOCl and Bi 24 O 31 Cl 10 , because the interfacial contact might produce intrinsic stresses on the crystal structure and alter the periodicity of the lattice 37,50 . However, for sample B-600, there exists a new band located at 115 cm −1 , which is close to that of pure Bi 24 O 31 Cl 10 ( Figure S4) 33 Figure S5). Furthermore, it can be identified in Fig. 5b that the band at 523 cm −1 exhibits a blue shift and the peak located at 442 cm −1 is gradually distinguishable, verifying the interfacial interactions caused by the construction of the heterojunction between BiOCl and Bi 24  . Based on the results from HRTEM, XPS, Raman and FT-IR spectra, it could be concluded that the Bi 24 O 31 Cl 10 /BiOCl heterojunction is successfully constructed, which is probably helpful for the transfer and separation of photogenerated charge carriers as well as the improvement of photocatalytic activity.
The photocatalytic performance of catalysts is related to the light absorption, thus the UV-vis diffuse reflectance spectroscopy (DRS) was adopted to determine the visible light harvesting ability of BiOCl and calcined samples (Fig. 6a)  It is accepted that the band gap energy of a semiconductor can be evaluated by the following equation: where α, v, E g , and A are the absorption coefficient, light frequency, band gap energy, and a constant, respectively. The parameter n is determined by the characteristics of the transition in a semiconductor (i.e., n = 1 for direct transition or n = 4 for indirect transition). In order to specify the n values of BiOCl and Bi 24 O 31 Cl 10 , the density functional theory (DFT) calculations are carried out (Fig. 6c,d). The calculated Fermi level is set at an energy of zero eV in the band gap, indicating typical intrinsic semiconducting characteristics in the electronic structure. Fig. 6c (left) shows that the conduction band minimum (CBM) and the valence band maximum (VBM) are located at Z and R point, respectively. It indicates that BiOCl is an indirect band gap semiconductor with a band gap of 2.63 eV, which is close to the previous DFT calculations 55,56 . The calculated band structure and density of states (DOS) (Fig. 6c right) imply that the CB of BiOCl mainly consists of Bi6p orbitals, whereas the VB is contributed by hybridized Cl2p and O2p orbitals. It could be inferred from Fig. 6d that Bi 24 O 31 Cl 10 is also an indirect band gap semiconductor with a band gap of 2.11 eV, which is consistent with the previous DFT calculations 39 . The CB of Bi 24 O 31 Cl 10 mainly consists of Bi6s and Bi5p orbitals, whereas the VB has major contribution from the hybridized Bi6s, Cl3p and O2p orbitals.
Having these results in mind, the n values for both BiOCl and Bi 24 O 31 Cl 10 are 4. Thus, the band gap energies of pure BiOCl and Bi 24 O 31 Cl 10 could be estimated from a plot of (αhv) 1/2 versus the photon energy (hv). The intercept of the tangent to the x-axis will give a good approximation of the band gap energies for various samples. As shown in Fig. 6b, the optical band gaps of sample B-RT and B-600 are calculated to be 3.19 eV and 2.40 eV, respectively, which are close to the previously reported values 33,40,57 . It is accepted that the selective photocatalytic oxidation of benzyl alcohol to benzaldehyde using O 2 as the oxidizing agent is considered as a model reaction to evaluate the photocatalytic performance of semiconductors 58 . Figure 7a displays the benzyl alcohol conversion efficiency over various samples. Notably, all samples exhibit photocatalytic activities toward benzyl alcohol oxidation under visible light irradiation. It's noted that pure BiOCl (B-RT) with a band gap of 3.19 eV also shows a benzyl alcohol conversion efficiency of 15.4%. TiO 2 , as a wide band-gap semiconductor, also displays excellent conversion efficiency (> 99%) and selectivity (> 99%) toward benzyl alcohol oxidation under visible light irradiation. This phenomenon is ascribed to the corresponding absorption edge shifts and absorption intensity enhancement in the visible-light region, which is related to the formation of a visible-light responsive charge-transfer complex between TiO 2 and benzyl alcohol 18,19 . To specify the reason that BiOCl exhibits visible light photocatalytic activity toward benzyl alcohol oxidation, UV-vis absorption spectra of benzyl alcohol (BA)-adsorbed samples are investigated ( Figure S6). As illustrated in Figure  S6, there is nearly no obvious changes in absorption edges and intensities in visible-light region for both BA-adsorbed samples and bare samples. Thus, it is expected that the benzyl alcohol conversion efficiency over the present samples may be not related to the charge-transfer complex formed between photocatalysts and benzyl alcohol. The photocatalytic activity of BiOCl under visible light irradiation may be related to the special nanosheet structure and ′″ ′″ V V V Bi O Bi vacancy associates in BiOCl 23,59 . The conversion efficiency reaches a maximum of approximately 40.3% with increasing the annealing temperature to 450 °C, however, further increase of the annealing temperature leads to an obvious decrease in the conversion efficiency. Furthermore, all samples display > 99% selectivity toward benzaldehyde. Although the photocatalytic performance of the as-prepared Bi 24 O 31 Cl 10 / BiOCl heterojunction is lower than that of TiO 2 and Na x TaO y ·nH 2 O 1,19 , it is close to even higher than several oxyhalides, such as Bi 3 O 4 Br, BiOBr and Bi 12 O 17 Cl 2 7 (Table S2), suggesting comprehensive work needs to be further conducted for oxyhalide semiconductors in the future.
It is well known that the photocatalytic process involves the photogenerated electrons and holes, which could react with the molecular O 2 and H 2 O/HO − to yield superoxide radical ( · O 2 − ) and · OH, respectively. The new produced active species are essentially important in the catalytic reactions. To reveal the origin of the highly photocatalytic performance and selectivity for the Bi 24 O 31 Cl 10 /BiOCl heterojunction, a series of active species trapping experiments were further conducted and the results are displayed in Fig. 7b. When acetic acid (HAC) as holes scavenger is added, the conversion efficiency of benzyl alcohol decreases significantly. The addition of tetrachloromethane (CCl 4 ) and benzoquinone (BQ) used as an electron and superoxide radical scavenger respectively, makes a slight influence in the conversion efficiency. These observations suggest that photogenerated holes act as the dominant role in the photocatalytic conversion of benzyl alcohol. Moreover, if molecular nitrogen is used instead of molecular O 2 in the presence of CCl 4 during the photocatalytic process, the conversion efficiency surprisingly decreases, which suggests that molecular O 2 is specially vital in the photocatalytic reaction. That is to say, the generation of superoxide radicals, which consumes lots of the photogenerated electrons, could greatly inhibit the recombination of photogenerated charge carriers, favoring the selective oxidation of benzyl alcohol to benzaldehyde originated by photogenerated holes.
The above result could also be proved by ESR technique. DMPO spin-trapping ESR spectra of sample B-450 to reveal the generation of active species · O 2 − and · OH are displayed in Fig. 7c and d. As shown in Fig. 7c, no characteristic ESR signal is detected either in the dark or in the visible light irradiation from 10 min to 30 min, indicating that · OH is not involved in the photocatalytic process. In Fig. 7d, there is no characteristic ESR signal observed in dark. However, the characteristic peaks of DMPO-· O 2 − adduct are detected after 10 min of visible light irradiation. Furthermore, the intensity of the DMPO-· O 2 − signals increases with prolonging the irradiation time. Combining the results of scavengers experiment and ESR spectra, it could be concluded that the photogenerated holes are the major active species in the photocatalytic conversion of benzyl alcohol under visible light irradiation, the active species · O 2 − are indeed formed during the photocatalytic process but not involved in the photocatalytic reaction. For potential applications, the stability of the heterojunction photocatalyst should be taken into consideration. Figure S7 presents the XRD patterns of sample B-450 before and after photocatalytic process. There is no structural variation between the samples before and after catalytic reaction, indicating the strong structural stability of Bi 24 O 31 Cl 10 /BiOCl heterojunction.
To investigate the photocatalytic process in detail, the relative conduction band (CB) and valence band (CB) potentials of the semiconductors should be determined. The Mott-Schottky plots of B-RT (BiOCl) and B-600 (Bi 24 O 31 Cl 10 ) are shown in Figure S8. It is found that the flat-band potential (V fb ) of BiOCl and Bi 24  − radicals greatly inhibits the recombination of photogenerated charge carriers, which is favorable for the photocatalytic performance.
To confirm the efficient separation of photogenerated charge carriers, photocurrent transient response measurements of sample B-RT, B-450 and B-600 are performed (Fig. 8c). As shown in Fig. 8c, all samples are prompt in producing photocurrent with a reproducible response to on/off cycle under visible light irradiation, suggesting that absorption of light could produce the photo-induced charge carriers and the charge carriers could transfer effectively. In comparison with B-RT and B-600, the sample B-450 displays the strongest peak intensity, implying more excellent photocatalytic activity of the Bi 24 O 31 Cl 10 /BiOCl heterojunction than the sole semiconductor counterparts.

Conclusions
A Bi 24 O 31 Cl 10 /BiOCl heterojunction has been successfully constructed through a simple thermal annealing route. Various characterization techniques confirm the construction of the Bi 24 O 31 Cl 10 /BiOCl heterojunction during the annealing process. The obtained Bi 24 O 31 Cl 10 /BiOCl photocatalyst displays excellent photocatalytic efficiency and selectivity toward the conversion of benzyl alcohol to benzaldehyde under visible light irradiation, which could reach 40.3% and > 99%, respectively. The photogenerated holes play an important role in the photocatalytic oxidation of benzyl alcohol and superoxide radicals are not involved in the photocatalytic process. The in-situ generation of heterojunction photocatalysts may provide superior interfacial contact, which is advantageous for enhancing the photocatalytic performance.

Methods
Bi 24 O 31 Cl 10 /BiOCl heterojunction synthesis. All chemical solvents and reagents were analytical grade and were used without further purification. In a typical procedure, 0.776 g Bi(NO 3 ) 3 ·5H 2 O was dissolved in 76 mL of glycerol, denoted as solution A. Then, 0.12 g KCl was dissolved in 4 mL of deionized water (solution B), which was subsequently poured into solution A. After stirring for 15 min, the mixture was transferred into a 100 mL Teflon-lined stainless steel autoclave, heated to 110 °C and kept at this temperature for 8 h. The resulting precipitate was collected by centrifugation, then washed with ethanol and deionized water for several times, and dried at 80 °C in vacuum to obtain the pure BiOCl powder (denoted as B-RT).
The thermal annealing step was performed in an air-atmosphere programmable tube furnace in the temperature range of 400 ~ 600 °C with an interval of 50 °C. The final products were denoted as B-400 ~B-600, respectively.

Characterization
Detailed crystallographic information of the synthesized samples was obtained on an X-ray diffractometer (Empypean Panalytical) with Cu Ka radiation (λ = 0.15406 nm). The thermogravinetric analysis (TG) and differential thermal analysis (DTA) were carried out on a thermal analyzer (NETZSCH STA 449F3) where the sample was heated from 30 to 950 °C with a raising ramp rate of 10 °C/min under nitrogen atmosphere. The detailed morphology, structure and heterojunction feature of the samples were recorded by transmission electron microscopy (TEM) and high resolution TEM (HRTEM) on a JEM-2010 apparatus with an acceleration voltage of 200 kV. The surface state and chemical composition of the samples were analyzed by X-ray photoelectron spectroscopy Scientific RepoRts | 6:28689 | DOI: 10.1038/srep28689 (XPS), which was carried out on a Thermo Escalab 250Xi with a monochromatic Al Ka (hv = 1486.6 eV). Raman spectra were recorded on the Horiba Jobin Yvon LabRAM HR800 instrument with the laser excitation of 532 nm. Fouier transform infrared spectroscopy (FT-IR) was performed using a Bruker Tensor 27 spectrophotometer using KBr powder-pressed pellets. The UV-vis absorption spectra were measured using a UV-vis spectrophotometer (Lambda 750s) in the range of 200 ~ 800 nm. The specific surface area (S BET ) of the samples was obtained from N 2 adsorption-desorption isotherms at 77 K (ASAP 2020). Prior to the sorption experiment, the materials were dehydrated by evacuation under specific conditions (200°C, 10 h).
The photocurrent transient response measurement was carried out based on a lock-in amplifier. The measurement system is constructed by a sample chamber, a lock-in amplifier (SR 830, Stanford Research Systems, Inc.) with a light chopper (SR540, Stanford Research Systems, Inc.) and a source of monochromatic light which is provided by a 500 W xenon lamp (CHF-XM 500, Trusttech) and a monochromator (Omni-λ300, Zolix). The monochromator and the lock-in amplifier were equipped with a computer. The analyzed product is assembled as a sandwich-like structure of ITO-product-ITO, which ITO means an indium tin oxide electrode. All the measurements were performed in air atmosphere and at room temperature.
Electron spin resonance (ESR) spectra were obtained on a Brüker ER200-SRC apparatus. A frequency of about 9.06 GHz was used for a dual-purpose cavity operation. The magnetic field of 0.2 mT was modulated at 100 kHz. A microwave power of about 1 mW was employed. Other parameters for the apparatus were set at: sweep width of 250 mT, center field of 250 mT, sweep time of 2.0 min, and accumulated 5times. All measurements were performed at room temperature in air without vacuum-pumping. ESR spectra for hydroxyl radicals and superoxide radicals were conducted in methylbenzene solution (2.0 mL) and methylbenzene solution containing methyl alcohol (2 mL, the volume ratio of methyl alcohol being 20%), respectively. The experiments were processed in dark and under visible light irradiation with adding 4 mg sample and 0.05 M DMPO.
All calculations were performed with density functional theory (DFT), using the CASTEP program package. The kinetic energy cutoff is 420 eV, using the generalized gradient approximation (GGA) with the Perdew-Burke-Ernzerhof (PBE) to treat the models. Geometry optimization is carried out until the residual forces were smaller than 0.01 eV Å −1 , and the convergence threshold for self-consistent iteration was set at 5 × 10 −7 eV. Photocatalytic activity Test. Selective Oxidation of benzyl alcohol has been widely studied as a model reaction to estimate the photocatalytic performance of catalysts. The photocatalytic activity experiments were carried out in a photochemical reactor fitted with a 500 W xenon lamp and a visible-light optical filter (λ > 420 nm). 10 mL methylbenzene solution involving alcohol (1 mM) mixed with 0.05 g sample was magnetically stirred at 25 °C in water bath. Anaerobic and aerobic reactions were performed by bubbling with pure N 2 and O 2 , respectively, for at least 1 hour before visible-light irradiation. After illuminating 10 hours, the suspension was centrifuged to remove the powder and measured the concentration of the alcohol and product by GC-FID (Shimadzu GC-2014C).