Bi2O3 cocatalyst improving photocatalytic hydrogen evolution performance of TiO2

Accepted Manuscript Title: Bi2 O3 cocatalyst improving photocatalytic hydrogen evolution performance of TiO2 Author: Difa Xu Yang Hai Xiangchao Zhang Shiying Zhang Rongan He PII: DOI: Reference:

S0169-4332(16)32904-X http://dx.doi.org/doi:10.1016/j.apsusc.2016.12.171 APSUSC 34718

To appear in:

APSUSC

Received date: Revised date: Accepted date:

17-11-2016 19-12-2016 20-12-2016

Please cite this article as: Difa Xu, Yang Hai, Xiangchao Zhang, Shiying Zhang, Rongan He, Bi2O3 cocatalyst improving photocatalytic hydrogen evolution performance of TiO2, Applied Surface Science http://dx.doi.org/10.1016/j.apsusc.2016.12.171 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Bi2O3 cocatalyst improving photocatalytic hydrogen evolution performance of TiO2 Difa Xu, Yang Hai, Xiangchao Zhang, Shiying Zhang, Rongan He* Hunan Key Laboratory of Applied Environmental Photocatalysis, Changsha University, Changsha, 410022, PR China

Graphical abstract

Highlights 1.

Bi-Bi2O3-anatase-rutile TiO2 multijunction photocatalyst was prepared

2.

Bi2O3 quantum dots with size of 2-3 nm were uniformly distributed

3.

Improved H2 evolution was noticed in glycerol-water mixture

4.

Optimal amount of Bi2O3 was found to be 0.89 mol%

Abstract Photocatalytic hydrogen production using water splitting is of potential importance from the viewpoint of renewable energy development. Herein, Bi2O3-TiO2 composite photocatalysts presented as Bi-Bi2O3-anatase-rutile TiO2 multijunction were first fabricated by a simple impregnation-calcination method using Bi2O3 as H2-production cocatalysts. The obtained multijunction samples exhibit an obvious enhancement in photocatalytic H2 evolution activity in the presence of glycerol. The effect of Bi2O3 amount on H2-evolution activity of TiO2 was investigated and the optimal Bi2O3 content was found to be 0.89 mol%, achieving a H2-production rate of 920 mol h–1, exceeding that of pure TiO2 by more than 73 times. The enhanced mechanism of photocatalytic H2-evolution activity is proposed. This study will provide new insight into the design and fabrication of TiO2-based hydrogen-production photocatalysts using low-cost Bi2O3 as cocatalyst. Keywords: TiO2; Bi2O3 quantum dots; Impregnation-calcination method; Hydrogen evolution; Charge carrier mechanisms

1. Introduction Hydrogen (H2) is considered as the cleanest renewable energy. However, the large-scale H2 production is mainly based on fossil fuels splitting or water electrolysis, which has led to high energy consumption and severe environmental pollution. Since the pioneering work of Fujishima and Honda on semiconductor photocatalysis [1], the direct H2 production via water splitting using solar energy has been extensively investigated as a promising way to meet the ever-growing global energy demands [2-7]. Although a variety of semiconductors are discovered for water splitting to produce H2 under light irradiation, titanium dioxide (TiO2) is regarded as one of the most stable H2-production semiconductors [8,9]. However, the H2-production efficiency of TiO2 is still limited, mainly due to the rapid recombination of conduction band (CB) electrons and valance band (VB) holes resulting from coulomb force [10,11]. It is well accepted that loading cocatalysts on the surface of TiO2 is effective for inhibiting the recombination of electron-hole pairs. Platinum (Pt) and palladium (Pd) are known as the best cocatalysts for photocatalytic H2 production, as they can efficiently trap the electrons, provide reaction sites, lower the over potential and suppress the backward reaction [12-14]. Unfortunately, Pt and Pd are noble metals that cannot be used for large-scale applications. Therefore, the possibility of utilizing earth-abundant metals as cocatalysts has attracted much attention, such as Cu, Ni, Co, Bi and Fe [15]. Among them, Bi-based cocatalysts have proven beneficial for TiO2-based photocatalytic pollutant degradation [16-24]. Nevertheless, there are few reports on photocatalytic H2 production over Bi2O3 decorated TiO2 to our knowledge [25-26]. Especially, the photocatalytic H2-production mechanism of Bi-based cocatalysts is still under debate. For example, Zhao et al. report that the enhanced photocatalytic H2 production of Bi-doped TiO2 (Bi/TiO2) nanotubes is due to doping energy level of Bi(3+x+) species below the CB of TiO2 [27]. For photocatalytic H2 production, it is crucial for semiconductors to have a more negative CB potential than H+/H2 redox potential. However, Sajjad et al. indicate that the redox potential E0 (Bi(3+x+)/Bi3+) of Bi-doped TiO2 is 0.29 eV (vs. NHE, pH = 7) [19], which is obviously more positive than that of H+/H2 (0.42 eV, vs.

NHE, pH = 7). Meanwhile, it is reported that the enhanced photocatalytic activity of Bi2O3/TiO2 composite is mainly due to the heterojunction effect [20-22]. The CB and VB potentials of Bi2O3 are calculated to be 0.31 and 2.59 eV, respectively [16], at the point of zero charge (close to the isoelectric point of 3.5 [28]), thus the correctional CB and VB potentials of Bi2O3 are 0.10 and 2.38 eV (vs. NHE, pH = 7), respectively. This means that the photogenerated electrons will spontaneously transfer from the CB of TiO2 (0.6 eV, vs. NHE, pH = 7) to that of Bi2O3. However, the CB potential of Bi2O3 is still less negative than the H+/H2 potential. Therefore, more efforts are necessary for the elucidating the mechanisms behind Bi-enhanced photocatalytic H2 production over TiO2. Herein, employing commercial TiO2 (Degussa P25, Degussa) as support, we synthesized Bi2O3 quantum dots (QDs) decorated TiO2 (Bi2O3-QDs-TiO2) nanocomposite photocatalyst by the impregnation method followed by calcination at 450 °C. The photocatalytic H2-production activity of the samples in glycerol/water solution was measured under ultraviolet (UV) light. Surprisingly, H2 production is greatly enhanced after decorating a small number of Bi2O3-QDs on the surface of TiO2, suggesting that Bi2O3-QDs are excellent cocatalysts for H2 production in TiO2 photocatalytic system. Considering that Bi is a non-noble metal, this motivates us to investigate the origins of the enhanced photocatalytic H2-production activity of TiO2 with the aid of Bi2O3-QDs. Meanwhile, this work also provides a new insight into the photocatalytic mechanism of Bi-based cocatalysts. 2. Experimental 2.1. Sample preparation All the reagents were of analytical grade and were used without further purification. Distilled water was used in all experiments. Commercially available Degussa P25 TiO2 powder (P25) was used as the source of TiO2. Bi2O3 quantum dots decorated TiO2 (Bi2O3-QDs-TiO2) photocatalysts were prepared by the conventional impregnation and calcination method. In a typical synthesis, P25 (1.0 g) was dispersed in 40 ml of HNO3 (0.05 mol L1) aqueous solution, and then a certain volume of Bi(NO3)3 (0.05 mol L1) and HNO3 (0.05 mol L1) mixed aqueous solution was added. The atomic ratios of Bi to Bi and Ti [Bi/(Bi+Ti)], which

hereafter were designated as RBi, were 0, 0.2, 0.5, 1.0, 2.0, 12 and 100 nominal atomic % (at. %) (see Table 1), and the obtained samples were labeled as B0, B0.2, B0.5, B1.0, B2.0, B12 and Bi2O3, respectively. The suspension were stirred for 12 h at room temperature, and then at 100 °C for 12 h to remove HNO3 and water. Finally, the obtained solid samples were calcined at 450 °C for 3 h in air. 2.2. Characterization The phase structures of the obtained samples were determined by Xray diffraction (XRD) on an HZG41BPC Xray diffractometer. The average crystallite size was calculated using the Scherrer formula (d = 0.9λ/Bcosθ, where d represents crystallite size, λ is Cu Kα wavelength (0.15418 nm), B refers to the full width at half maximum intensity (FWHM) in radians and θ is Bragg’s diffraction angle) after correcting the instrumental broadening. The chemical compositions of the samples were measured by inductively coupled plasma atomic emission spectrometry (ICP-AES) using an optima 4300 DV spectrometer (PerkinElmer). Transmission electron microscopy (TEM) analyses were conducted by a F20 S-TWIN electron microscope (Tecnai G2, FEI Company), using a 200 kV accelerating voltage. The X-ray photoelectron spectroscopy (XPS) measurement was made in an ultrahigh vacuum VG ESCALAB 210 electron spectrometer equipped with a multichannel detector. UV-vis diffused reflectance spectra

were

obtained

from

a

Shimadzu

UV2550

UV-vis

spectrotometer.

BrunauerEmmettTeller (BET) specific surface area (SBET) of the powders was analyzed by nitrogen adsorption in a Micromeritics ASAP 2020 nitrogen adsorption apparatus (USA). Photoluminescence (PL) spectra were measured at room temperature on a Hitachi F-7000 Fluorescence Spectrophotometer. 2.3 Photocatalytic hydrogen production The photocatalytic H2-production tests of the obtained samples were carried out at ambient temperature and atmospheric pressure in a 100 mL Pyrex glass flask with the openings sealed by silicone rubber septum. Four ultraviolet light-emitting diodes (UV-LEDs, 3W, 365 nm) were employed as light sources, which were positioned 1 cm away from the reactor in four

different directions. The focused intensity and areas on the flask for each UV-LED was ca. 80.0 mW/cm2 and 1 cm2, respectively. In a typical measurement, 50 mg of Bi2O3-QDs-TiO2 photocatalyst was suspended in 80 mL of 0.1 M glycerol/water solution and then dispersed by ultrasonic treatment for 3 min. To remove the dissolved oxygen, nitrogen was continuously bubbled into the mixed suspension for 40 min and the openings were immediately sealed, ensuring an anaerobic condition of photocatalytic H2-production system. To prevent the sedimentation of particles, the photocatalytic H2-production reaction was carried out under continuous magnetic stirring. At certain time points after light irradiation started, 0.4 mL of headspace gas was sampled by a micro-injector for analysis of H2 on a gas chromatograph (Shimadzu GC-14C, TCD) with nitrogen carrier gas and 5Å molecular sieve column. The quantum efficiency (QE) was calculated according to the following equation.

number of reacted electrons  100 number of incident photons number of evolved H 2 molecules  2   100 number of incident photons

QE[%] 

3. Results and discussion 3.1. Phase structures and morphology XRD patterns of the samples B0, B0.2, B0.5, B1.0, B2.0 and B12 are shown in Fig. 1. Pure TiO2 (sample B0) shows the expected anatase (JCPDS 21-1272) and rutile (JCPDS 21-1276) phases, which are also observed for all other photocatalysts. However, no significant XRD peaks belonging to metallic Bi species can be found up to sample B12 with 12% of Bi/(Bi+Ti) atomic ratio. For sample B12, there are four new diffraction peaks at 2θ = 28.0, 30.3, 32.8 and 66.5° corresponding to the (221), (311), (400) and (004) planes of β-Bi2O3 (JCPDS 74-1374), respectively, indicating the existence of crystalline Bi2O3. No characteristic diffraction peaks of Bi2O3 are observed for samples B0.2, B0.5, B1.0 and B2.0 due to the relatively low contents of Bi2O3, also implying the small size and high dispersion of Bi species [29]. It is reported that Bi-doped TiO2 (Bi-TiO2) might be formed by treating Bi(NO3)3 and TiO2 at high temperature. In that case, because the radius of Bi3+ ions (0.103 nm) is larger than that of Ti4+

ions (0.061 nm), Bi doping will increase the lattice spacing of TiO2, as indicted by a shift of the (101) peak of TiO2 at 25.0o towards a lower angle [30-32]. However, no obvious peak shift is observed for all Bi2O3-TiO2 samples, implying that no Bi-doped TiO2 is formed.

Fig. 1. XRD patterns of samples B0, B0.2, B0.5, B1.0, B2.0 and B12 (the XRD intensity of sample B12 is enlarged 3 times that of raw data). The average crystallite sizes of the obtained samples are calculated by the Scherrer equation and presented in Table 1. It is found that the average crystallite size of TiO2 is almost the same (ca. 23 nm for anatase and 44 nm for rutile) for all samples, suggesting that loading Bi2O3 do not obviously change the crystallite size of TiO2. This is because the calcination temperature of 450 oC is not high enough to stimulate the further growth of TiO2. Accordingly, no Bi-doped TiO2 is formed due to the insufficient energy to incorporate Bi species into the lattice of TiO2.

Table 1. Effect of RBi on physicochemical properties and quantum efficiency (QE) of Bi2O3-QDs-TiO2 samples. RBi

Samples

Bi (mol.%)

Crystalline size

SBET

Average

Pore volume

Porosity

QE

(ICP-AES)

(nm) a

(m2/g)

pore size

(cm3/g)

(%)

(%)

0

B0

0

23.0 (A), 42.4 (R)

46

13.7

0.14

34.1

0.05

0.2

B0.2

0.17

23.1 (A), 43.5 (R)

44

31.1

0.39

59.0

1.8

0.5

B0.5

0.36

23.6 (A), 43.5 (R)

42

27.6

0.23

46.0

3.0

1.0

B1.0

0.89

23.7 (A), 43.8 (R)

42

22.2

0.21

43.8

3.7

2.0

B2.0

1.74

24.0 (A), 44.1 (R)

42

19.4

0.20

42.6

2.3

12

B12

12.4

26.6 (A), 45.4 (R)

40

18.6

0.19

43.2

0.12

100

Bi2O3

100

29.2 (Bi2O3)

0.12

-

-

-

0

a

A and R denote anatase and rutile, respectively.

The morphology and microstructures of sample B1.0 are investigated by TEM (Fig. 2). The low-magnification image (Fig. 2a) exhibits that sample B1.0 consists of crystalline NPs with size of about 20–65 nm. This agrees well with the crystallite sizes calculated by the Scherrer equation based on XRD results. According to the data shown in Table 1, the NPs with larger size can be ascribed to rutile TiO2 and those with smaller size belong to anatase TiO2. It can be observed from the high-magnification TEM image (Fig. 2b) that there are many small NPs with size of 2–3 nm uniformly dispersed on the surface of TiO2 NPs. The chemical composition of the B1.0 sample is determined by energy-dispersive X-ray spectroscopy (EDS) (data not shown), which shows that only Bi, Ti and O elements can be observed in sample B1.0. Combined with XRD and EDS results, those small NPs can be identified as Bi2O3-QDs and labeled in Fig. 2b.

Fig. 2. TEM images of sample B1.0 at low magnification (a) and high magnification (b). 3.2. XPS analysis To further investigate the surface chemical composition of the obtained samples, XPS characterization is performed. According to the XPS survey spectra shown in Fig. 3a, both samples B0 and B1.0 exhibit sharp photoelectron peaks at the binding energies of 458 (Ti 2p), 531 (O 1s) and 285 eV (C 1s), suggesting the existence of Ti, O and C element. Among them, C element is mainly from the residual carbon and/or adventitious hydrocarbon. However, some new peaks at the binding energy of approximately 158 eV are observed for sample Bi1.0, which can be indexed as Bi element. Fig. 3b shows the high-resolution XPS spectrum of Bi 4f region. The XPS fitting plots of Bi species are divided into two groups, where the binding energies at 159.1 (Bi 4f7/2) and 164.5 (Bi 4f5/2) eV belongs to Bi3+ in Bi2O3, and those at 156.9 (Bi 4f7/2) and 162.5 (Bi 4f5/2) eV are ascribed to Bi0 (metallic Bi) . This suggests that both Bi2O3 and metallic Bi are present in sample B1.0. The easy reduction of beta Bi oxides (β-Bi2O3) is supported by previous work, in which zerovalent Bi is a common issue for β-Bi2O3 with small particle size obtained from 450 oC calcination [33]. It is also reported that metallic Bi might play very important roles in enhancing the photocatalytic activity of semiconductors [30,33,34]. However, no metallic Bi is detected by XRD due to the relatively low XRD intensity and Bi content. Furthermore, the binding energies of Ti species are unchanged before and after the deposition of Bi2O3-QDs onto the surface of TiO2. This confirms that no Bi-doped TiO2 is formed, which is in agreement with the above XRD results.

Fig. 3. XPS survey spectra of samples B0 and B1.0 (a), and high-resolution XPS spectrum of Bi 4f of sample B1.0 (b). 3.3. UV-Vis diffuse reflectance spectrum Fig. 4 shows the UV-vis diffuse reflectance spectra of samples B0, B1.0, B12, and Bi2O3. The absorption edge of pure TiO2 (sample B0) is about 400 nm, corresponding to the intrinsic absorption of rutile TiO2 with a band gap of 3.1 eV. On the contrary, bulk Bi2O3 shows higher absorbance in the visible light region (400–550 nm) than pure TiO2. Therefore, it is not surprising that the visible light absorption of Bi2O3-QDs-TiO2 (samples B1 and B12) increases with increasing RBi value, suggesting the photosensitization effect of Bi2O3-QDs to TiO2 [35]. Obviously, the increased light absorption is in favor of the photocatalytic activity enhancement of semiconductors. However, the absorption edge of samples B1.0 and B12 is not greatly changed compared to pure TiO2, further indicating that Bi species are not

incorporated into the lattice of TiO2, but are only deposited on the surface of TiO2.

Fig. 4. UV-vis diffuse reflectance spectra of samples B0, B1.0, B12 and Bi2O3. 3.4. BET surface areas and pore size distributions Fig. 5 shows the N2 adsorption/desorption isotherms and the corresponding pore-size distribution curves (inset) of samples B0 and B1.0. The isotherms of samples B0 and B1.0 are type IV with type H3 hysteresis loops. The adsorption capacity of the samples is very high at the relative pressure (P/P0) near to 1.0, implying the presence of large mesopores and macropores. Meanwhile, the hysteresis loops of the samples exhibit an unlimited absorption at high P/P0 range from 0.8 to 1.0, indicating the existence of slit-like pores formed from the aggregation of plate-like TiO2 particles. The pore size distributions of samples range from 2 to 70 mm with a centralization at about 27 and 35 nm for samples B0 and B1.0, respectively, further suggesting the existence of mesopores and macropores. Notably, such porous structures can provide efficient transportation channel for reactant and product molecules, giving rise to an enhanced photocatalytic reaction.

Fig. 5. Nitrogen adsorption-desorption isotherms and the corresponding pore-size distribution curves (inset) of samples B0 and B1.0. Table 1 shows quantitative details on BET specific surface area (SBET), pore volume, porosity and average pore size of samples BT0, BT0.2, BT0.5, BT1.0, BT2.0, BT12 and Bi2O3. The presence of Bi2O3 QDs in the nanocomposite photocatalysts exerts an obvious influence on the pore structure and SBET of the obtained samples (Table 1). With RBi increasing from 0.2 to 12, the average pore size, pore volume, porosity and SBET slightly decrease due to the presence of Bi2O3 and its relatively large density.

3.5. PL spectra The transfer and separation efficiencies of photogenerated charge carriers can be identified by PL emission spectra, in which lower PL emission intensity indicates decreased recombination of the charge carriers. As shown in Fig. 6, there are three main emission peak groups for samples B0 and B1.0. The strongest PL peak at 398 nm is ascribed to the transition emission under light irradiation with the energy equivalent to the band gap energy of TiO2 (3.1 eV). The PL peaks at 451 and 468 nm are attributed to free exciton, and the other two peaks at 482 and 493 nm are due to bound exciton resulting from the band edge. It is found that the PL intensity of sample B1.0 is significantly decreased relative to pure TiO2 (sample B0), indicating a lower recombination rate for the photogenerated electrons and holes of sample B1.0 (Bi2O3-QDs-TiO2). Importantly, the decreased charge carrier recombination is in favor

of the enhanced photocatalytic activity of the samples.

Fig. 6. Comparison of PL spectra of samples B0 and B1.0. 3.6. Photocatalytic H2-production activity Photocatalytic H2-production activity of the obtained samples is evaluated using glycerol as scavenger under UV-LED illumination. In control experiments without light irradiation or photocatalyst addition, no H2 production can be detected, suggesting that H2 is only produced from photocatalytic reaction in the presence of light and photocatalyst. The photocatalytic H2 production performance of the samples is shown in Fig. 7. Pure Bi2O3 is inactive for photocatalytic H2 production because the CB level of Bi2O3 (0.10 eV) is less negative than H+/H2 potential (–0.42 eV), as a result, no H2 production is detected. Meanwhile, the photocatalytic H2-production activity of pure TiO2 (B0) is very low (ca. 12.6 mol h–1 g–1) due to the rapid charge recombination, the large overpotential and the fast backward reaction of TiO2. Interestingly, the photocatalytic H2-production activity of TiO2 is remarkably improved when a small number of Bi2O3-QDs is decorated onto the surface of TiO2, suggesting that Bi2O3-QDs played an important role in improving the photocatalytic H2-production activity of TiO2. The photocatalytic H2-production activity of Bi2O3-QDs-TiO2 increases with increasing the content of Bi2O3-QDs (RBi) from 0 to 1.0, and sample B1.0 shows the highest photocatalytic H2-production activity, which is 920 mol h–1 g–1 with an apparent quantum efficiency (QE) of 3.7%. This is 73 times higher than that of pure TiO2 under UV-LED irradiation. Moreover, The B0 sample exhibits the sufficient photocatalytic stability in the

recycling hydrogen evolution measurement. However, the photocatalytic H2-production activity of Bi2O3-QDs-TiO2 decreases as the value of RBi further increased from 1.0 to 12, and sample B12 shows a drastically decreased H2-production rate, indicating an important influence of Bi2O3-QDs content on the photocatalytic H2-production activity of TiO2.

Fig. 7. Photocatalytic activity of samples B0, B0.2, B0.5, B1.0, B2.0, B12 and Bi2O3 on the photocatalytic H2 production from an 0.1 M glycerol aqueous solution under UV-LED irradiation. 3.7 Photocatalytic mechanism Although Bi2S3 with negative CB potential can split water to produce hydrogen [36,37], it is difficult for Bi2O3 to improve the photocatalytic H2-production activity of TiO2 by Bi-doping or heterojunction construction. Another possibility for the enhanced photocatalytic H2-production activity is that the energy band of quantum-sized Bi2O3 is tuned by the quantum confinement effect. In this case, CB level of Bi2O3-QDs will shift to a more negative position due to the energy level splitting, which gives rise to a widened band gap compared to bulk Bi2O3. For example, Hou et al. calculat that the CB potential and band gap of Bi2O3-QDs is 0.11 and 2.97 eV at the point of zero charge, respectively [21], corresponding to CB potential of –0.10 eV (vs. NHE, pH=7), based on the fact that band potential decreases by 59 mV for each increasing unit of pH [38,39]. However, the CB potential of Bi2O3-QDs is still far less negative than H+/H2 potential. The band structure analysis suggests that Bi2O3 is an electron acceptor when decorated onto the surface of TiO2. Meanwhile, the standard redox potential (E0) of Bi2O3/Bi is 0.37 eV,

which means that Bi2O3 can be easily reduced by photogenerated electrons. Therefore, a new photocatalytic mechanism is provided in this work (Fig. 8). Under UV-LED light irradiation, both Bi2O3-QDs and TiO2 are excited and generate electron-hole pairs. The photogenerated electrons can spontaneously transfer from the CB of TiO2 to that of Bi2O3-QDs, leading to partial reduction of Bi2O3-QDs into metallic Bi. The Fermi level (EF) of TiO2 is higher (–0.23 eV, vs. NHE) than that of metallic Bi (0.025 eV, vs. NHE) [40], which induces the formation of an built-in electric field between the interface of TiO2 and Bi2O3-QDs and drives the electrons flowing from TiO2 to metallic Bi. As a result, the Fermi level is up-shift in metallic Bi while down-shift in TiO2 until reaching the Fermi level equilibrium. At equilibrium, the apparent Fermi level (EF*) is closer to the CB of TiO2 [41,42], endowing Bi/TiO2 system with the sufficiently negative H2-production level and eliminating the overpotential of TiO2. Meanwhile, the bands of TiO2 bend upward toward surface to form a Schottky barrier [43], which not only promotes the separation but also inhibits the recombination of charge carriers. The additional electrons are accumulated in metallic Bi, which are employed as H2-production active sites and suppress the backward reaction. The photocatalytic action of metal Bi is similar to Pt, Au, Cu, Ni and Co metals that are considered as excellent cocatalysts for H2 production [44-46]. It is noteworthy that the catalytic properties of metal NPs are size-dependent, and generally metal NPs with smaller size are more active than those particles with larger size [41,47,48]. Undoubtedly, the metallic Bi with smaller size is expected to be obtained from the quantum-sized Bi2O3, which has the larger surface area, closer contact with TiO2 and is easier to be reduced compared to bulk Bi2O3. This agrees well with the reports that only Bi species with small size improves the photocatalytic H2-production activity of semiconductors [27,30]. Therefore, it is not surprising that when RBi is < 1.0, more metallic Bi is formed and deposited on TiO2 surface with increasing RBi value, resulting in an increased photocatalytic activity. On the contrary, when RBi is > 1.0, the photocatalytic activity decreases with further increase in RBi, due to the bigger size of metallic Bi. Moreover, the excessive Bi2O3 might hinder the light absorption and block the surface active sites of TiO2, which are also disadvantageous to

H2 production. Therefore, the optimal Bi2O3-QDs loading content is 0.89 mol% and sample B1.0 shows the highest photocatalytic H2-production activity in this work.

Fig. 8. Schematic illustration for the photocatalytic H2 production in Bi2O3-QDs-TiO2 photocatalysts (the magnified illustration is the transfer and separation mechanism of charge carriers between TiO2 and metallic Bi). 4. Conclusions In summary, Bi2O3-QDs decorated TiO2 (Bi2O3-QDs-TiO2) photocatalyst was successfully fabricated by a simple impregnation and calcination method for efficient photocatalytic H2 production. The obtained samples were Bi-Bi2O3-anatase-rutile TiO2 multijunction. The optimal Bi loading content was determined to be about 0.89 mol% and the corresponding H2-production rate was 920 mol h–1 g–1 with QE of 3.7% in 0.1 M glycerol solution, which was 73 times higher than pure TiO2. Under light irradiation, Bi2O3-QDs can easily be reduced to form metallic Bi, due to the sufficiently positive standard redox potential of Bi2O3/Bi (E0 = 0.37 eV). Therefore, the enhanced H2-production activity of Bi2O3-QDs-TiO2 was mainly due to the existence of metallic Bi, which was employed as H2-production cocatalyst to help the transfer and separation of photogenerated charge carriers of TiO2. This

work for the first time demonstrates that Bi2O3 can be used as a cocatalyst for photocatalytic H2-production.

Acknowledgements We are grateful for the financial support from the National Natural Science Foundation of China (51272032 and 51402025) and a project supported by Scientific Research Fund of Hunan Provincial Education Department (16B027).

References [1] A. Fujishima, K. Honda, Electrochemical photolysis of water at a semiconductor electrode, Nature 238 (1972) 37-38. [2] X. Zou, Y. Zhang, Noble metal-free hydrogen evolution catalysts for water splitting, Chem. Soc. Rev. 44 (2015) 5148-5180. [3] S. J. A. Moniz, S. A. Shevlin, D. J. Martin, Z. X. Guo, J. Tang, Visible-light driven heterojunction photocatalysts for water splitting-a critical review, Energy Environ. Sci. 8 (2015) 731-759. [4] T. Hisatomi, J. Kubota, K. Domen, Recent advances in semiconductors for photocatalytic and photoelectrochemical water splitting, Chem. Soc. Rev. 43 (2014) 7520-7535. [5] S. Cao, J. Yu, g-C3N4-based photocatalysts for hydrogen generation, J. Phys. Chem. Lett. 5 (2014) 2101-2107. [6] S. Cao, J. Yu, Carbon-based H2-production photocatalytic materials, J. Photochem. Photobiol. C 27 (2016) 72-99. [7] Q. Xiang, B. Cheng, J. Yu, Graphene-based photocatalysts for solar-fuel generation, Angew. Chem. Int. Ed. 54 (2015) 11350-11366. [8] C. P. Sajan, S. Wageh, A. A. Al-Ghamdi, J. Yu, S. Cao, TiO2 nanosheets with exposed {001} facets for photocatalytic applications, Nano Res. 9 (2016) 3-27. [9] A. Meng, J. Zhang, D. Xu, B. Cheng, J. Yu, Enhanced photocatalytic H2-production activity of anatase TiO2 nanosheet by selectively depositing dual-cocatalysts on {101}

and {001} facets, Appl. Catal. B 198 (2016) 286-294. [10] P. Zhou, J. Yu, M. Jaroniec, All-solid-state Z-scheme photocatalytic systems, Adv. Mater. 26 (2014) 4920-4935. [11] J. Low, S. Cao, J. Yu, S. Wageh, Two-dimensional layered composite photocatalysts, Chem. Commun. 50 (2014) 10768-10777. [12] X. Li, J. Yu, J. Low, Y. Fang, J. Xiao, X. Chen, Engineering heterogeneous semiconductors for solar water splitting, J. Mater. Chem. A 3 (2015) 2485-2534. [13] J. Wang, P. Yang, B. Cao, J. Zhao, Z. Zhu, Photocatalytic carbon-carbon bond formation with concurrent hydrogen evolution on the Pt/TiO2 nanotube, Appl. Surf. Sci. 325 (2015) 86-90. [14] M. Hattori, K. Noda, All electrochemical fabrication of a bilayer membrane composed of nanotubular photocatalyst and palladium toward high-purity hydrogen production, Appl. Surf. Sci. 357 (2015) 214-220. [15] J. Ran, J. Zhang, J. Yu, M. Jaroniec, S.Z. Qiao, Earth-abundant cocatalysts for semiconductor-based photocatalytic water splitting, Chem. Soc. Rev. 43 (2014) 7787-7812. [16] S. Han, J. Li, K. Yang, J. Lin, Fabrication of a β-Bi2O3/BiOI heterojunction and its efficient photocatalysis for organic dye removal, Chin. J. Catal. 36 (2015) 2119-2126. [17] R. He, S. Cao, P. Zhou, J. Yu, Recent advances in visible light Bi-based photocatalysts, Chin. J. Catal. 35 (2014) 989-1007. [18] X. Meng, Z. Zhang, Bismuth-based photocatalytic semiconductors: introduction, challenges and possible approaches, J. Mol. Catal. A: Chem. 423 (2016) 533-549. [19] S. Sajjad, S. A. K. Leghari, F. Chen, J. Zhang, Bismuth-doped ordered mesoporous TiO2: Visible-light catalyst for simultaneous degradation of phenol and chromium, Chem. Eur. J. 16 (2010) 13795-13804. [20] M. Ge, C. Cao, S. Li, S. Zhang, S. Deng, J. Huang, Q. Li, K. Zhang, S. S. Al-Deyab, Y. Lai, Enhanced photocatalytic performances of n-TiO2 nanotubes by uniform creation of p-n heterojunctions with p-Bi2O3 quantum dots, Nanoscale 7 (2015) 11552-11560.

[21] J. Hou, C. Yang, Z. Wang, S. Jiao, H. Zhu, Bi2O3 quantum dots decorated anatase TiO2 nanocrystals with exposed {001} facets on graphene sheets for enhanced visible-light photocatalytic performance, Appl. Catal. B 129 (2013) 333-341. [22] X. Zhao, H. Liu, J. Qu, Photoelectrocatalytic degradation of organic contaminants at Bi2O3/TiO2 nanotube array electrode, Appl. Surf. Sci. 257 (2011) 4621-4624. [23] J. Hou, Z. Wang, S. Jiao, H. Zhu, Bi2O3 quantum-dot decorated nitrogen-doped Bi3NbO7 nanosheets: in situ synthesis and enhanced visible-light photocatalytic activity, Crystengcomm 14 (2012) 5923-5928. [24] J. Zhu, S. Wang, J. Wang, D. Zhang, H. Li, Highly active and durable Bi2O3/TiO2 visible photocatalyst in flower-like spheres with surface-enriched Bi2O3 quantum dots, Appl. Catal. B 102 (2011) 120-125. [25] N. L. Reddy, G. K. Reddy, K. M. Basha, P. K. Mounika, M. V. Shankar, Highly efficient hydrogen production using Bi2O3/TiO2 nanostructured photocatalysts under led light irradiation, Mater. Today: Proc. 3 (2016) 1351-1358. [26] B. Naik, S. Martha, K.M. Parida, Facile fabrication of Bi2O3/TiO2-xNx nanocomposites for excellent visible light driven photocatalytic hydrogen evolution, Int. J. Hydrogen Energy 36 (2011) 2794-2802. [27] W. Zhao, X. Wang, H. Sang, K. Wang, Synthesis of Bi-doped TiO2 nanotubes and enhanced photocatalytic activity for hydrogen evolution from glycerol solution, Chin. J. Chem. 31 (2013) 415-420. [28] M. Kosmulski, Compilation of PZC and IEP of sparingly soluble metal oxides and hydroxides from literature, Adv. Colloid Interface Sci. 152 (2009) 14-25. [29] J. Zhang, Y. Hu, X. Jiang, S. Chen, S. Meng, X. Fu, Design of a direct Z-scheme photocatalyst: preparation and characterization of Bi2O3/g-C3N4 with high visible light activity, J. Hazard. Mater. 280 (2014) 713-722. [30] Y. Wu, G. Lu, S. Li, The doping effect of Bi on TiO2 for photocatalytic hydrogen generation and photodecolorization of Rhodamine B, J. Phys. Chem. C 113 (2009) 9950-9955.

[31] M. K. Kumar, K. Bhavani, B. Srinivas, S. N. Kumar, M. Sudhakar, G. Naresh, A. Venugopal, Nano structured bismuth and nitrogen co-doped TiO2 as an efficient light harvesting photocatalyst under natural sunlight for the production of H2 by H2O splitting, Appl. Catal. A 515 (2016) 91-100. [32] S. Bagwasi, B. Tian, J. Zhang, M. Nasir, Synthesis, characterization and application of bismuth and boron co-doped TiO2: a visible light active photocatalyst, Chem. Eng. J. 217 (2013) 108-118. [33] M. N. Gómez-Cerezo, M. J. Muñoz-Batista, D. Tudela, M. Fernández-García, A. Kubacka, Composite Bi2O3-TiO2 catalysts for toluene photo-degradation: ultraviolet and visible light performances, Appl. Catal. B 156-157 (2014) 307-313. [34] M. Nischk, P. Mazierski, Z. Wei, K. Siuzdak, N. A. Kouame, E. Kowalska, H. Remita, A. Zaleska-Medynska, Enhanced photocatalytic, electrochemical and photoelectrochemical properties of TiO2 nanotubes arrays modified with Cu, AgCu and Bi nanoparticles obtained via radiolytic reduction, Appl. Surf. Sci. 387 (2016) 89-102. [35] Y. Bessekhouad, D. Robert, J.-V. Weber, Photocatalytic activity of Cu2O/TiO2, Bi2O3/TiO2 and ZnMn2O4/TiO2 heterojunctions, Catal. Today 101 (2005) 315-321. [36] R. Brahimi, Y. Bessekhouad, A. Bouguelia, M. Trari, Visible light induced hydrogen evolution over the heterosystem Bi2S3/TiO2, Catal. Today 122 (2007) 62-65. [37] J. Kim, M. Kang, High photocatalytic hydrogen production over the band gap-tuned urchin-like Bi2S3-loaded TiO2 composites system, Int. J. Hydrogen Energy 37 (2012) 8249-8256. [38] D. Xu, B. Cheng, J. Zhang, W. Wang, J. Yu, W. Ho, Photocatalytic activity of Ag2MO4 (M = Cr, Mo, W) photocatalysts, J. Mater. Chem. A 3 (2015) 20153-20166. [39] D. Xu, B. Cheng, S. Cao, J. Yu, Enhanced photocatalytic activity and stability of Z-scheme Ag2CrO4-GO composite photocatalysts for organic pollutant degradation, Appl. Catal. B 164 (2015) 380-388. [40] C.R. Ast, H. Höchst, Fermi surface of Bi (111) measured by photoemission spectroscopy, Phys. Rev. Lett. 87 (2001) 177602.

[41] V. Subramanian, E. E. Wolf, P. V. Kamat, Catalysis with TiO2/gold nanocomposites. Effect of metal particle size on the fermi level equilibration, J. Am. Chem. Soc. 126 (2004) 4943-4950. [42] A. Wood, M. Giersig, P. Mulvaney, Fermi level equilibration in quantum dot-metal nanojunctions, J. Phys. Chem. B 105 (2001) 8810-8815. [43] A. L. Linsebigler, G. Lu, J. T. Yates Jr, Photocatalysis on TiO2 surfaces: principles, mechanisms, and selected results, Chem. Rev. 95 (1995) 735-758. [44] Q. Hu, J. Huang, G. Li, Y. Jiang, H. Lan, W. Guo, Y. Cao, Origin of the improved photocatalytic activity of Cu incorporated TiO2 for hydrogen generation from water, Appl. Surf. Sci. 382 (2016) 170-177. [45] Y. Xu, R. Xu, Nickel-based cocatalysts for photocatalytic hydrogen production, Appl. Surf. Sci. 351 (2015) 779-793. [46] P. Wang, Y. Lu, X. Wang, H. Yu, Co-modification of amorphous-Ti(IV) hole cocatalyst and Ni(OH)2 electron cocatalyst for enhanced photocatalytic H2-production performance of TiO2, Appl. Surf. Sci. 391 (2017) 259-266. [47] S. G. Kumar, K. K. Rao, Comparison of modification strategies towards enhanced charge carrier separation and photocatalytic degradation activity of metal oxide semiconductors (TiO2, WO3 and ZnO), Appl. Surf. Sci. 391 (2017) 124-148. [48] A. Lucid, A. Iwaszuk, M. Nolan, A first principles investigation of Bi2O3-modified TiO2 for visible light activated photocatalysis: the role of TiO2 crystal form and the Bi3+ stereochemical lone pair, Mater. Sci. Semicond. Process. 25 (2014) 59-67.