In this era of rapid economic development with an ever-increasing use of fossil fuels, environmental pollution is fast becoming one of the most challenging issues faced by mankind. One of the hazardous gaseous pollutants generated by fossil fuel combustion is NO, which not only causes acid rains but also affects the human respiratory system [1-7]. In recent times, semiconductor photocatalysis has attracted a lot of attention since it provides sustainable solutions to the problems surrounding environmental pollution [7-13] and energy crisis [14-17]. TiO₂ has been considered as the most important semiconductor photocatalyst owing to its strong oxidative power, cost effectiveness, excellent biocompatibility and long-term stability against photochemical corrosion [18, 19]. However, TiO₂ can only be excited by UV light (bandgap of 3.2 eV), which only accounts for 3%–5% of the energy in the solar spectrum [20, 21]. Therefore, exploration of a novel semiconductor photocatalyst which can harvest visible or IR light is both imperative and extremely challenging [13].

Recently, considerable attention has been given to Bi-based semiconductor photocatalysts such as BiOBr [22], BiOCl [12], Bi₂O₂CO₃ [23], Bi₄MoO₉ [11], BiPO₄ [7] and Bi₂WO₆ [24]. One of these materials, Bi₂WO₆, has a bandgap of 2.6–2.8 eV and consists of perovskite-like slabs of WO₆ and (Bi₂O₂)²⁺ layers [25-28]. This unique layered structure of Bi₂WO₆ makes it an important visible-light-responsive photocatalyst. Despite this attractive property, the visible photoreactivity of Bi₂WO₆ is not high enough for practical applications [24, 29]. According to literature, the apparent quantum efficiency of Bi₂WO₆ at 420 nm is less than 5% [30]. Up to now, many strategies have been used to improve the photocatalytic activity of Bi₂WO₆ such as controlling the morphology [26, 29], surface modification [27, 29, 31], ion doping [32], introducing defects [33] and forming heterojunctions [34]. For example, it has been reported that flower-like Bi₂WO₆ microspheres exhibit improved photocatalytic activity compared to plate-like Bi₂WO₆ [35].

Very recently, surface plasmon resonance (SPR) effect of noble metals such as gold [36] and silver [37] has been used to enhance the visible photocatalytic activity of semiconductor photocatalysts [38]. Huge local electric field enhancement was observed at the surface of gold nanoparticles (NPs) due to the interaction with the electric and magnetic field of light [36]. Similarly, the excitation of electron-hole (e^--h⁺) pairs was boosted in TiO₂ with the enhanced near-field of silver NPs [37], improving the photocatalytic activity of TiO₂.

In this study, SPR effect was used to stimulate the excitation of photo-generated e^--h⁺ pairs in Bi₂WO₆ by deposition of Bi nanospheres on the surface of flower-like Bi₂WO₆. The photocatalytic activity of Bi₂WO₆ microspheres was evaluated by the oxidation of NO in ppb level under irradiation by an LED lamp (λ > 400 nm). We systematically studied the effect of Bi loading amount on the properties and performance of Bi₂WO₆ microspheres. Although SPR effect of Bi has been used to improve the visible photoreactivity of some semiconductor photocatalysts such as BiOBr [22, 39], Bi₄MoO₉ [11], and BiPO₄ [7], the present work is the first report on the SPR effect of Bi on Bi₂WO₆ photocatalyst, to the best of our knowledge.

Microspheric Bi₂WO₆ precursor was synthesized according to the procedure detailed in Ref. [40]. Typically, 2.0 mmol of Bi(NO₃)₃·5H₂O, 2.0 mmol of Na₂WO₄·2H₂O, and 0.1 mmol of sodium dodecyl sulfate were dissolved in 30 ml of water, which was first sonicated for 10 min and then stirred for 20 min to form a uniform solution. The resulting solution was transferred to a 100-ml Teflon-lined stainless steel autoclave and heated in an oven at 180 ^oC for 24 h. After cooling down to to room temperature, the light-yellow precipitates were collected by centrifugation, washed several times with ethanol and n-hexane, and dried in an oven at 80 ^oC for 4 h.

Bi-nanospheres-modified Bi₂WO₆ microspheres (Bi/Bi₂WO₆) were prepared by reduction of Bi(NO₃)₃·5H₂O using ethylene glycol as reductant. Following this, 0.42 g of the prepared Bi₂WO₆ was added to 80 ml of ethylene glycol solution, in which 0.2 mmol of Bi(NO₃)₃·5H₂O had been dissolved. After ultrasonication for 10 min and stirring for 2 h, the suspensions were transferred to Teflon-lined stainless-steel autoclave and heated at 180 ^oC for 10 h. The obtained grey precipitates thus were separated by centrifugation and washed with distilled water. The resulting cake was dried at 80 ^oC for 4 h.

By varying the amount of Bi(NO₃)₃·5H₂O, a series of Bi/Bi₂WO₆ hybridized photocatalysts with different mass ratios of Bi to Bi₂WO₆ were obtained (Table 1). For simplicity, the prepared samples are named as Sx, where x% represents the calculated mass ratio of Bi to Bi₂WO₆. For example, sample S10 refers to the Bi/Bi₂WO₆ hybridized photocatalyst with a Bi to Bi₂WO₆ mass ratio of 10%. Note that sample S0 was also prepared in an ethylene glycol medium, but without the addition of any Bi(NO₃)₃·5H₂O during the treatment of Bi₂WO₆ precursor.

For comparison, pure Bi sample (denoted as SBi) was also prepared under identical conditions except for the addition of Bi₂WO₆ precursor.

The crystalline structure of the samples was investigated by X-ray diffraction (XRD, D8 Advance, Bruker, Germany). The microscopic morphology of the catalyst was observed by transmission electron microscopy (TEM, Tecnai G20) and field emission scanning electron microscopy (SEM, Hitachi, Japan), with the acceleration voltages set to 200 and 20 kV for the TEM and SEM, respectively. UV-Vis diffuse reflectance spectrum (DRS) was measured using a spectrophotometer (Shimadzu UV-2550, Japan) using BaSO₄ as the reference. Photoluminescence (PL) spectra were obtained on a fluorescence spectrophotometer (F-7000, Hitachi, Japan). X-ray photoelectron spectroscopy (XPS) was performed using a Multilab 2000 XPS system. A nitrogen-adsorption instrument (ASAP 2020, USA) was used to obtain information regarding the surface area and pore structure of the photocatalyst.

Electron spin resonance (ESR) spectroscopy (JES FA200 spectrometer) was used to record free radicals by mixing the photocatalysts in a 40 mmol/L 5, 5-Dimethyl-1-pyrroline N-oxide (DMPO) solution tank (methanol dispersion for DMPO-·OOH/O₂·^- and aqueous dispersion for DMPO-·OH), illuminated with a visible LED lamp (λ = 420 nm).

Photocurrent and electrochemical impedance spectroscopy (EIS) were performed in a three-electrode configuration on an electrochemical station (CHI760e, China), using Bi₂WO₆/ITO as the working electrode, Pt plate as the counter electrode, and Ag/AgCl as the reference electrode, in a aqueous solution (0.4 mol L^-1) of Na₂SO₄ (as the electrolyte). During the photocurrent measurements, an LED lamp which emitted mainly at 420 nm (3 W, Shenzhen LAMPLIC) was used as the light source to excite the Bi₂WO₆/ITO electrode.

The photocatalytic activity of the system was evaluated by oxidation of NO at room temperature. The reaction was carried out in a 4.5-L stainless steel rectangular reactor which was covered by quartz glass in the upper layer. A visible LED lamp (150 W) with a cut-off filter (λ > 400 nm) was used as the light source. The photocatalyst powders (0.3 g) were deposited on the surface of a watch glass (diameter of about 11.5 cm). The NO gas was supplied from a compressed gas cylinder having 50 ppm of NO (diluted with N₂) in line with the standards recommended by the U.S. National Institute of Standards and Technology. An air stream produced by a zero-air generator (Advanced Pollution Instrumentation, A Teledyne Technologies Co., Model 701) was used for diluting the initial NO concentration to approximately 600 ppb. When the gas system reached adsorption-desorption balance, the lamp was turned on. A chemiluminescence NOx analyzer (Advanced Pollution Instrumentation, Teledyne Technologies, Model T200) was used for monitoring the concentrations of NO continuously at a gas-flow rate of 1.0 L·min^-1.

XRD was used to identify the phase structure and crystallinity of the photocatalyst. It can be seen from Fig. 1 that the diffraction peaks of Bi₂WO₆ precursor are in good agreement with those of the pure orthorhombic Bi₂WO₆ (JCPDS card no. 39-0256), where the diffraction peaks appearing at 28.3^o, 32.8^o, 47.1^o and 55.8^o correspond to the (131), (002), (202) and (133) planes, respectively [26, 29]. After further solvothermal treatment of Bi₂WO₆ precursor in Bi(NO₃)₃·5H₂O/ethylene glycol solution, a sharp diffraction peak appears at 2θ = 27.2^o due to the formation of metallic Bi (JCPDS card no. 44-1246) [41, 42]. This indicates that Bi/Bi₂WO₆ hybridized photocatalyst has been successfully prepared. Note that the diffraction peaks for metallic Bi cannot be observed for samples S0, S1 and S2, possibly due to the smaller particle sizes of Bi nanoparticles.

Fig. 2 compares the SEM and TEM images of Bi₂WO₆ precursor (Figs. 2(a)–(c)) and S0 sample (Figs. 2(d)–(f)). It can be seen that both samples have a similar flower-like morphology, and the diameter of the flower-like Bi₂WO₆ assembly from nanosheets is about 2 µm. This indicates that ethylene glycol has little effect on the Bi₂WO₆ precursor. Similar to sample S0, sample S10 maintains the flower-like morphology. However, several smooth nanospheres with diameters 200–500 nm were also observed on the surface of sample S10. These are most likely to be Bi nanospheres (see discussion below) [42]. Therefore, we can conclude that Bi nanospheres are produced as a result of the reduction of Bi(NO₃)₃·5H₂O, and not of the Bi₂WO₆ precursor itself.

Sample S10 was further characterized by elemental mapping using energy dispersive X-ray spectroscopy (EDX). From Fig. 3, it can be clearly seen that the density of Bi is higher at the place indicated by the arrow, while the densities of elements W and O in this area are lower than average. This indicates that the nanospheres are composed of metallic Bi. The molar ratio of Bi to W on the surface of flower-like Bi₂WO₆ (area A) is 1.61, while the ratio increases to 6.27 (by a factor of 3.90) on the surface of nanosphere (area B) as shown in Fig. 4. This sharp increase in the concentration of Bi from area A to B in sample S10 further confirms the composition of the nanospheres.

The photocatalytic activity of the photocatalyst depends on the efficiency of the following sequential processes [43]: (1) separation of the photo-generated carriers after light absorption, (2) migration of photo-generated electrons and holes from the bulk to the surface of the photocatalyst, and (3) capture of electrons and holes to initiate redox reaction with adsorbed substrates. Therefore, light-harvesting ability is of great importance to the photoreactivity of the photocatalyst [44]. Here, UV-visible diffuse reflectance spectrum was used to evaluate the light-harvesting ability of the photocatalyst. From Fig. 5, it can be seen that the optical absorption edge of the yellowish Bi₂WO₆ (sample S0) is about 436 nm corresponding to a bandgap of 2.84 eV, consistent with earlier reports [26, 30].

After the deposition of Bi nanospheres, we can clearly see that the light-harvesting ability of Bi₂WO₆ microspheres are conspicuously enhanced, which can be attributed to a charge-transfer transition between the Bi spheres and Bi₂WO₆ microspheres. A distinct SPR peak centered at 533 nm can be observed for pure Bi nanospheres. When compared to the pristine Bi₂WO₆ microspheres, the Bi/Bi₂WO₆ hybridized photocatalyst exhibits an additional broad absorption peak around 530 nm, which is the characteristic SPR peak of Bi nanospheres. The SPR effect of Bi nanospheres can boost the excitation of e^--h⁺ pairs of Bi₂WO₆ microspheres under visible-light irradiation, improving the photoreactivity of Bi₂WO₆.

Fig. 6 compares the nitrogen adsorption-desorption isotherms of pristine Bi₂WO₆ microspheres (sample S0) and Bi-nanospheres modified Bi₂WO₆ microspheres (sample S10). It can be seen that the isotherms of both photocatalysts are type Ⅳ with one hysteresis loop at a relative pressure range of 0.5–1.0. This indicates the presence of mesopores (type H3), corresponding to slit-shaped pores that are generally associated with plate-like particles, consistent with their sheet-like morphology [45, 46]. The isotherms of samples S0 and S10 almost coincide, indicating their similar structures. Table 1 summarizes the BET surface areas and pore structures of the photocatalysts. It can be seen that all samples have comparable BET surface area (40–50 m²/g), pore volume (0.09–0.12 cm³/g) and average pore size diameter (8–10 nm). Therefore, BET surface area and pore structure are not critical factors affecting the photoreactivity of Bi₂WO₆ microspheres.

XPS was used to characterize the elemental composition and the chemical state of Bi₂WO₆ precursor, pristine Bi₂WO₆ (S0) and Bi/Bi₂WO₆ (S10). From the XPS survey spectra, it can be seen that all samples contain Bi, W, O and small amounts of C. The C contained in the samples arises from the residual organic solvent or from the hydrocarbons present in the XPS instrument itself [47].

The high resolution XPS spectra in Bi 4f, W 4f and O 1s regions are shown in Fig. 7(B), Fig. 7(C) and Fig. 7(D), respectively. It can be seen that Bi₂WO₆ precursor and sample S0 have similar peak positions in Bi 4f, W 4f and O 1s regions which further confirm that ethylene glycol has little effect on Bi₂WO₆, consistent with the XRD (Fig. 1) and SEM/TEM (Fig. 2) characterization results. The two peaks with binding energies of 159.4 and 164.7 eV correspond to Bi 4f_7/2 and Bi 4f_5/2, respectively (Fig 7(B)) [26, 48, 49]. The peaks with binding energies of 35.7 and 37.8 eV originate from W 4f_7/2 and W 4f_5/2, respectively (Fig. 7(C)) [26, 50]. The peak with binding energy of 530.5 eV can be attributed to O 1s (Fig. 7(D)) [26, 48]. When compared with those of Bi₂WO₆ microspheres (sample S0), the binding energies for all the Bi 4f, W 4f and O 1s peaks of sample S10 reduced by 0.3–0.5 eV. The reduction in binding energy implies a corresponding increase of electron cloud density. This indicates that electrons have a tendency to be transferred from Bi nanospheres to Bi₂WO₆ due to the built-in Mott-Schottky effects at the interface between Bi metal and Bi₂WO₆ [42].

According to literature, the XPS characteristic peaks for the Bi-Bi bonds of elemental Bi are centered at 156.8 and 162.1 eV [3]. However, in the present study, the XPS peaks for Bi-Bi bond were not found in the Bi/Bi₂WO₆ hybrid (sample S10), which is possibly due to the oxidization of the surface of Bi nanospheres.

Fig. 8(A) compares the photocatalytic oxidation curves of NO in presence of the different photocatalysts under visible irradiation. It can be seen that pure Bi nanospheres, Bi₂WO₆ precursor and pristine Bi₂WO₆ (sample S0) exhibit poor photocatalytic activity with NO removal rate of 7.7%, 8.6% and 12.3%, respectively. This is due to the quick recombination of photo-generated carriers [28]. Upon modification with Bi metal, the photocatalytic activity of Bi₂WO₆ microspheres improves steadily, with the highest NO removal rate (53.1%) achieved for sample S10. However, with further increase in the loading amount of Bi nanospheres to 20 wt%, the photoreactivity of Bi/Bi₂WO₆ begins to decrease. The NO removal rate for S20 reduces to 44.5%.

To further determine the surface interaction between Bi nanospheres and Bi₂WO₆ microspheres, we measured the photoreactivity of the physically mixed Bi/Bi₂WO₆ sample (Smix) which contains 10 wt% Bi nanospheres. It was found that the NO removal rate of Smix is only 16.9%, much smaller than that of sample S10 (53.1%). This indicates the strong interaction between Bi nanospheres and Bi₂WO₆ microspheres exists only in the Bi/Bi₂WO₆ hybridized photocatalyst.

Not only the photocatalytic activity, but also the stability of the photocatalyst is of great importance from the viewpoint of practical applications [51]. Therefore, the catalytic activity of the S10 photocatalyst was tested during repeated use, and the experimental results are shown in Fig. 8(B). It can be seen that the NO removal rate only slightly decreases after cycling the catalyst 5 times, indicating excellent stability of the Bi/Bi₂WO₆ hybridized photocatalyst.

To account for the enhanced photocatalytic activity of Bi/Bi₂WO₆ hybridized photocatalyst, we performed PL, photocurrent measurements, and EIS of the photocatalyst.

PL spectrum is commonly used to analyze the recombination rate of photo-generated carriers of the photocatalyst [49]. Fig. 9(A) shows the PL spectra of pristine Bi₂WO₆ (sample S0) and Bi₂WO₆ modified with different amounts of Bi metal (samples S10 and S20). It can be seen that pure Bi₂WO₆ sample shows the highest PL peak intensity, indicating that the recombination rate of photo-generated carriers in pure Bi₂WO₆ is the highest, and the deposition of Bi nanospheres can retard the recombination process. Therefore, it is not hard to understand the positive effect of Bi nanospheres on the photoreactivity of Bi₂WO₆ (Fig. 8).

Photocurrent is a measure of the ability of the photocatalyst to generate and transfer of photo-generated charge carriers after excitation [52]. From Fig. 9(B), it can be seen that the photocurrent in Bi/Bi₂WO₆ hybridized photocatalyst (samples S10 and S20) is much higher than that of pristine Bi₂WO₆ photocatalyst (sample S0).

Fig. 9(C) compares the EIS spectra of the photocatalysts. It can be seen that: (1) all samples have only one arc/semicircle on the EIS plane display, which indicates that only the surface-charge-transfer step is involved in the photocatalytic reaction, and (2) sample S10 possesses the smallest arc radius on the EIS Nyquist plot. The smaller the radius of the arc, the more effective is the separation of the photogenerated electron-hole pair and faster the interfacial charge transfer to the electron donor/electron acceptor [53]. Therefore, the loading of Bi metal can stimulate the generation and transfer of photo-generated electrons and holes, enhancing the photocatalytic activity.

Reactive oxygen species (ROSs) such as hydroxyl radicals (·OH) and super oxygen radicals (·O_2^-) are important radicals responsible for the oxidation of NO [3, 54]. Therefore, radical trapping experiment was also performed in the present study, and the experimental results are shown in Fig. 10. It can be seen that hardly any ·OH and ·O_2^- radicals can be detected from pristine Bi₂WO₆ (sample S0) or Bi/Bi₂WO₆ hybridized photocatalyst (sample S10) in the dark. However, after irradiation for 10 min, signals originating from ·OH and ·O₂^- radicals were observed for pristine Bi₂WO₆ and Bi/Bi₂WO₆ hybridized photocatalyst, and the intensities of the signals for ·OH and ·O₂^- radicals in Bi/Bi₂WO₆ suspensions were much stronger than in pristine Bi₂WO₆. These results suggest that more electrons and holes present in the Bi/Bi₂WO₆ hybridized photocatalyst produce ROSs (instead of undergoing quick recombination) when compared to pristine Bi₂WO₆, consistent with the observed trend in photocatalytic activity (Fig. 8).

Since all the photocatalysts have similar BET surface areas and pore structures (Table 1), the enhanced photocatalytic activity of Bi₂WO₆ microspheres can be attributed to the introduction of Bi nanospheres. The conduction band (CB) position of Bi₂WO₆ flowers was reported to be -0.32 eV (vs. NHE) [29], which is higher than the flatband position of Bi nanospheres (-0.17 eV) [3, 55]. Therefore, the photo-generated electrons prefer to migrate from the CB of Bi₂WO₆ microspheres to Bi nanospheres, which are then captured by adsorbed oxygen to form superoxide radicals (E_O2/·O2-^o = -0.046 V vs. NHE) [56]. As the bandgap of pristine Bi₂WO₆ sample is 2.84 eV (Fig. 5), the valence band (VB) potential of Bi₂WO₆ microspheres can be calculated as +2.52 eV. This value of the VB potential is high enough to oxidize H₂O to produce hydroxyl radicals (E_H2O/·OH^o= 2.3 V vs. NHE) [57]. Both ·O_2^- and ·OH radicals are important ROSs that are responsible for the oxidation of NO. However, the concentrations of ·O₂^- and ·OH radicals over pristine Bi₂WO₆ microspheres are very low (Fig. 10), mainly due to the quick recombination of carriers. After modification of Bi₂WO₆ microspheres with Bi nanospheres, the photo-generated carriers can transfer from the CB of Bi₂WO₆ microspheres to Bi nanospheres, retarding the recombination. In addition, the near-field enhancement produced on Bi nanospheres by the SPR effect can significantly enhance energy of electrons, which then consequently boosts the separation and migration of photo-generated carriers [25]. Then, highly concentrated ROSs such as ·O_2^- and ·OH radicals are produced in Bi/Bi₂WO₆ to oxidize NO (Figs. 10 and 11).

However, the photoreactivity of Bi/Bi₂WO₆ begins to decrease when the concentration of Bi is high, which may be due to the light-filtering effect of Bi nanospheres, as excesssive Bi can block the visible-light irradiation on Bi₂WO₆ microspheres (Fig. 5) [3].

Bi/Bi₂WO₆ hybridized photocatalyst was successfully fabricated by solvothermal treatment of Bi₂WO₆ in Bi(NO₃)₃ solution using ethylene glycol as reductant. The photocatalytic NO removal rate of Bi₂WO₆ increased from 12.3% to 53.1% after loading 10 wt% of Bi nanospheres under visible LED lamp irradiation. This sharp increase in visible photocatalytic activity of Bi/Bi₂WO₆ hybridized photocatalyst was attributed to the SPR effect of Bi nanospheres. This SPR effect not only facilitates the transfer of photo-generated electrons, but also boosts the separation and migration of photo-generated carriers, enhancing the production of ROSs such as ·O_2^- and ·OH radicals that are responsible for the oxidation of NO. Bi/Bi₂WO₆ hybridized photocatalyst also exhibits excellent photocatalytic stability, making it a potential candidate to be used in air purification systems.