Platinum-nanoparticle-loaded bismuth oxide: an efficient plasmonic photocatalyst active under visible light

Abstract

Employing a commercial semiconductor with a positive conduction band level, we investigated a new plasmonic photocatalyst, Pt/Bi₂O₃, with high photocatalytic activity for decomposition of environmental organic pollutants under visible light.

---

Introduction

Heterogeneous photocatalysis has always been an urgent issue with regard to the important technological and societal benefits resulting from the removal of toxic organic pollutants in wastewater and in the purification of drinking water.¹ Not being able to use visible light, however, hinders the practical applications of most photocatalysts.² Modified photocatalysts, such as N-, C-, and S-doped TiO₂,³ have been much studied in order to extend the absorption edge into the visible light region, but such photocatalysts often suffer serious problems, for instance, their low quantum yields. In recent research, nano-sized Ag was employed, allowing the development of highly efficient and stable plasmonic photocatalysts due to its plasmon resonance (PR) in the visible region.⁴ Importantly, Huang et al. pointed out that the dipolar character of the surface plasmon state of Ag nanoparticles (NPs) was critical for enhancing the separation of plasmon-induced electron–hole pairs.⁴ In addition to Ag NPs, Au NPs can also be photoexcited under visible light illumination due to their PR. Tian et al. indicated that the photoexcitation of electrons from the Au NPs to the TiO₂ conduction band (CB), and the simultaneous transfer of compensatory electrons from the solution-phase electron mediator (Fe^2+/3+) to the Au NPs, resulted in the achievement of spatial charge separation in the Au/TiO₂ system.⁵ With regard to this, it should be noted that the potential of the Fe^2+/3+ mediator is more positive (Fe³⁺ + e⁻→ Fe²⁺, +0.77 V vs. NHE) than that of the TiO₂ CB (−0.3 V vs. NHE) and that of the O₂ reduction potentials (single-electron reduction process: O₂ + e⁻→ O₂⁻; −0.56 V vs. NHE; or multi-electron reduction process: O₂ + 2H₂O + 4e⁻→ 4OH⁻; +0.40 V vs. NHE), so that Fe^2+/3+ pairs (rather than the CB level of TiO₂) give out electrons to the Au NPs.

Unlike Au or Ag, platinum is an important catalytic metal but usually shows no plasmonic activity. However, recent literature has demonstrated that Pt NPs can also be plasmon-excited by choosing an appropriate excitation wavelength.⁶ If this is true, after plasmon excitation, the holes generated on the surface of Pt NPs can be used for decomposing organic molecules so long as the electrons can be transferred successfully. Accordingly, with the aim of synthesizing a plasmonic photocatalyst active under visible light, we wanted to use a metal oxide with a positive CB level, where the transfer of plasmon-induced electrons can be achieved. Bismuth oxide (Bi₂O₃) is of this type.⁷ In the present study, density functional theory (DFT) calculations, together with the literature,^7,8 show that Bi₂O₃ can serve as an electron carrier owing to its positive CB level (+0.4 V vs. NHE), which lies exactly in the potential range of the multi-electron reduction of O₂ (O₂ + 2H₂O + 4e⁻→ 4OH⁻; +0.40 V vs. NHE).

The above discussions so far led us to prepare a plasmonic photocatalyst by dispersing metallic Pt NPs onto Bi₂O₃ substrates. In addition, considering that home-made supports synthesized using various methods could lead to different particle sizes, crystal phases, morphologies, and thus different catalytic performance, the use of commercially available Bi₂O₃ is strongly desired for initial screening; catalyst optimization will be the subject of further research.

Results and discussion

Bi₂O₃ loaded with Pt NPs (denoted as Pt/Bi₂O₃) was prepared via a simple photoreduction method.⁹ The morphology of Pt/Bi₂O₃ was examined by TEM (JEM-2010, Fig. S1†) and FE-SEM (MIRA II-LMH, Fig. S1), its crystal structure by XRD (Rigaku D/Max-2550pc, Fig. S2), its surface area by BET (Micromeritics TriStar II) and its chemical composition by XPS (KRATOS AXIS ULTRA-DLD, Fig. S3). The photoactivity of Pt/Bi₂O₃ was initially evaluated by oxidative decomposition of aqueous acetaldehyde (AcH) under visible light illumination from a mercury lamp (λ > 400 nm).† In this case, Bi₂O₃ alone exhibits limited photoactivity. Intriguingly, its photoactivity increased significantly with increasing Pt content to a maximum at ca. 3.0 wt%, with the activity dropping gradually at higher Pt content. Fig. 1 illustrates that the rate of CO₂ evolution over Pt/Bi₂O₃ (10.37 μmol h⁻¹) was ca. 10-fold and 4-fold higher than that of bare Bi₂O₃ (1.01 μmol h⁻¹) and N-TiO₂ NPs (2.58 μmol h⁻¹) at room temperature (ca. 20 °C), respectively. To clarify that the decomposition of AcH over Pt/Bi₂O₃ was not caused by regular photolysis or catalysis, we also carried out the experiments under visible light irradiation without catalysts (photolysis) and in the dark with the catalysts (catalysis). In the former experiment, no CO₂ evolution was detected, while the latter showed that a tiny quantity of CO₂ over Pt/Bi₂O₃ was generated within 20 min, with no further change during the subsequent period in the dark. These results fully confirmed that Pt/Bi₂O₃ acts as an active photocatalyst for the decomposition of AcH under visible light illumination.

Fig. 1 Decomposition curves of acetaldehyde solution over Pt/Bi₂O₃, N-TiO₂ and Bi₂O₃ under visible light irradiation (λ > 400 nm) in the air.

The most notable application of the present photocatalyst is the photodecomposition of formaldehyde (FAD), another typical indoor air pollutant. As shown in Fig. 2, it is worth noting that the decomposition of FAD over Pt/Bi₂O₃ under visible light illumination proceeded with a rate (151.8 μmol h⁻¹) comparable to that of TiO₂ NPs (Degussa P25) driven by UV light (189.4 μmol h⁻¹), and was 34.4-fold higher than that of visible-light-irradiated N-TiO₂ (4.41 μmol h⁻¹). The oxidative ability of Pt/Bi₂O₃ is extremely high when we consider this is a visible-light-driven photocatalytic system.

Fig. 2 Decomposition curves of formaldehyde solution over Pt/Bi₂O₃, P25 and N-TiO₂ under light irradiation in the air.

In addition to Pt/Bi₂O₃, Pt-loaded TiO₂ (Pt/TiO₂) and Pt-loaded CdS (Pt/CdS) were also prepared by photoreduction for comparison. As indicated in Fig. 3, when methanol was selected as the target organic compound, the visible photoactivity of Pt/Bi₂O₃ was much higher than that of Pt/TiO₂ and Pt/CdS. It should be mentioned that although TiO₂ with a band gap of 3.2 eV can only be photoexcited by UV light, Pt/TiO₂ exhibits appreciable visible photoactivity (even higher than that of Pt-loaded CdS, whose band gap is only 2.4 eV). In order to determine the reusability and stability of Pt/Bi₂O₃, methanol was also chosen as the target compound, and the life-cycle tests are shown in Fig. S4†.

Fig. 3 Decomposition curves of methanol solution over Pt/Bi₂O₃, Pt/TiO₂ and Pt/CdS under visible light irradiation (λ > 400 nm) in the air.

The band structures of Bi₂O₃ were evaluated through plane-wave-based DFT calculations. Fig. 4 shows the band structure and density of states (DOS) for Bi₂O₃, from which it is noteworthy that the Bi6p atomic orbitals are separated into two parts – an occupied part and an unoccupied part. The former hybridizes with the O2p band and partially contributes to the valence band (VB), whereas the latter is the main component for the CB. The presence of Bi6p electrons in the upper valence band provides p–p repulsion for the VB maximum, which results in a narrowing of the band gap. The electron density contour maps of the HOMO and LUMO levels related to photoexcitation are shown in Fig. S5†. Fig. 5 compares the UV/Vis absorption spectrum (Lambda 900, Perkin Elmer) of Pt/Bi₂O₃, Bi₂O₃ and N-TiO₂. N-TiO₂ exhibits a visible absorption shoulder due to the nitride at ca. 450 nm.¹⁰ In contrast to bare Bi₂O₃, a distinct feature of the UV/Vis absorption spectrum of Pt/Bi₂O₃ is its strong absorption in the visible region (up to 700 nm), comparable to that of the UV region. The spectrum is similar to that of the adsorption band of Ag NPs, as reported by Huang et al.⁴ Further, Ikeda et al. stated that nano-sized Pt can also absorb visible light owing to its PR.⁶ Thus the existence of the wide absorption band seems to be attributable to the surface plasmon state of Pt NPs.

Fig. 4 Band dispersion and density of states for Bi₂O₃.

Fig. 5 UV/Vis absorption spectrum of (a) N-TiO₂, (b) Bi₂O₃ and (c) Pt/Bi₂O₃.

In view of the low surface areas of micron-sized powders, Bi₂O₃ scarcely has any BET specific surface area (S_BET < 0.01 m² g⁻¹) as compared to P25 (S_BET = 48.7 m² g⁻¹) and N-doped TiO₂ (S_BET = 113.2 m² g⁻¹), which results in a nearly complete lack of surface active sites where the decomposition of organics could take place. Moreover, although Bi₂O₃ has a direct band gap of 2.8 eV, which can be excited by visible light,⁷ the light-generated charge carriers in micron-sized semiconductor cannot efficiently transfer to the surface and tend to recombine in the bulk. As a result, in this work, Pt may play an important role in decomposition of organic molecules.

Generally, Pt has been recognized as a perfect electron scavenger that can greatly enhance the photoactivity of semiconductors by separating holes and electrons. However, it is hard to explain the visible photoactivity of Pt/Bi₂O₃ when one only considers the formation of Pt Schottky barriers on the surface of Bi₂O₃, because dyes with complicated conjugated structures cannot be decomposed by Pt/Bi₂O₃. For example, Pt/Bi₂O₃ showed very limited photoactivity for decomposition of methylene blue into CO₂ under visible light. If free holes can be generated in the deep VB level of Bi₂O₃ (3.13 V vs. NHE), their oxidative power should be strong enough to decompose both aldehydes and dyes into CO₂. However, if holes are generated by Pt NPs, their oxidative ability (1.2 V vs. NHE) should efficiently decompose aldehydes but not dyes, because CH and CO are able to react with surface O atoms to form CO₂, with reaction barriers below 0.7 V (vs. NHE).⁶

Furthermore, considering that Pt/Bi₂O₃ shows much higher visible photoactivity than that of Pt/TiO₂ and Pt/CdS (neither of which have acceptable photoactivities), Pt must not be a co-catalyst that serves as a normal electron scavenger. The fact is, Pt NPs can be plasmon-photoexcited, and therefore the photoactivity of Pt/Bi₂O₃ and Pt/TiO₂ is most likely due to the direct participation of the plasmon-induced holes from Pt NPs. Since the CB levels of TiO₂ (−0.3 V vs. NHE) and CdS (−0.52 V vs. NHE) are much more negative than that of Bi₂O₃, once transferred to the negative CB, the plasmon-induced electrons cannot reduce O₂ by single-electron process (O₂ + e⁻→ O₂⁻; −0.56 V vs. NHE), and would reduce O₂ by a multi-electron process (O₂ + 2H₂O + 4e⁻→ 4OH⁻; +0.40 V vs. NHE) only with difficulty. Therefore, the plasmon-induced electrons tend to be trapped by Pt NPs again; this may be the reason that Pt/TiO₂ and Pt/CdS exhibit poor photocatalytic performance.

All of the results mentioned above seem to support the full picture of the plasmon-induced photocatalytic process – namely, absorbed photons on Pt NPs are efficiently separated into electrons and holes; the electrons transfer to the CB level of Bi₂O₃,^4,5 and the holes diffuse to the surface of Pt NPs, where the decomposition of organic compounds occurs.¹¹ Since the multi-electron reduction process occurs at a potential of 0.4 V (O₂ + 2H₂O + 4e⁻→ 4OH⁻) which is the same as that of the CB level of Bi₂O₃, the electrons in the Bi₂O₃ CB level can effectively transferred to the adsorbed O₂ rather than recombining with the holes. Consequently, the charge separation can be achieved by coupling Pt NPs with Bi₂O₃.

Conclusion

In summary, we have investigated a plasmonic photocatalyst, Pt/Bi₂O₃, which has the advantage of high visible-light photoactivity for decomposing organic pollutants in mild conditions. The positive CB level of Bi₂O₃, which is able to transfer plasmon-induced electrons from the Pt NPs, plays an important role in the unique catalytic properties of Pt/Bi₂O₃. Future work will focus on preparing Au/Bi₂O₃ and Ag/Bi₂O₃ plasmonic photocatalysts. We consider that these photocatalysts could have a wide range of practical purposes.

Experimental

Pt/Bi₂O₃ was prepared by a simple photoreduction method. An aqueous suspension containing commercial Bi₂O₃ powder (Analytical Grade, Hangzhou Huadong Medicine Group Co., Ltd) was sonicated for 2 h, followed by several rinses with Milli-Q water and ethyl alcohol, and subsequent drying at 80 °C for 12 h. The resulting powder (200 mg) was added to hexachloroplatinic acid (H₂PtCl₆·6H₂O) solution (0.1 g mL⁻¹, Sigma-Aldrich) and was then exposed to UV light (λ > 300 nm) from a high pressure mercury lamp (input power 500 W, light intensity 598 μW cm⁻², Nanjing Xujiang, Ltd). After 1 h of irradiation, 10 mL methanol solution (10 vol%) was added as sacrificial reagent and the suspension was exposed to further irradiation for 1.5 h after bubbling with pure N₂ for 15 min. The resulting sample was centrifuged, washed with Milli-Q water and ethyl alcohol to remove the impurities, and collected as a powder after drying at 60 °C for 12 h. Pt/TiO₂ and Pt/CdS were prepared in the same manner. N-TiO₂ (TS-S4230) for comparative analysis was obtained from Sumitomo Chemical (Japan). P25 TiO₂ powders were supplied by Degussa Co. Ltd. Other commercial chemicals were of the highest available grade and were used without further purification.

Decomposition of organics was carried out with 40 mg of the powdered catalyst suspended in 10 mL of aqueous organic solution in a Pyrex test tube (55 mL). Photocatalytic studies of all samples were carried out in a photochemical reactor (Nanjing Xujiang, Ltd), which was fitted with a cold quartz tube and an inner UV lamp. An optical filter (Shanghai Seagull Colored Optical Glass Co., Ltd) was adopted to provide visible light with λ > 400 nm. For photocatalysis experiments, the rate of agitation was set at 700 ± 10 rpm. Gas chromatography (Agilent 6890 with Agilent column 115-3133) was utilized for quantitative measurement of gaseous CO₂ generated as the organic substrate was oxidized. Gas volumes of 200 μL were extracted from the test tubes using a microliter syringe at regular intervals.

Acknowledgements

We are grateful to Daisuke Toyoshima for the computer calculations, and Dr. Wangyang Lu and Dr. Yuyuan Yao for helpful discussions. This work was supported by the Program for Changjiang Scholars and Innovative Research Team in University (IRT 0654).

Notes and references

1. M. Fujihira, Y. Satoh and T. Osa, Nature, 1981, 293, 206–208.
2. Z. H. Yuan and L. D. Zhang, J. Mater. Chem., 2001, 11, 1265–1268.
3. X. B. Chen and C. Burda, J. Am. Chem. Soc., 2008, 130, 5018.
4. P. Wang, B. B. Huang, X. Y. Qin, X. Y. Zhang, Y. Dai, J. Y. Wei and M. H. Whangbo, Angew. Chem., Int. Ed., 2008, 47, 7931–7933.
5. Y. Tian and T. Tatsuma, J. Am. Chem. Soc., 2005, 127, 7632–7637.
6. K. Ikeda, J. Sato, N. Fujimoto, N. Hayazawa, S. Kawata and K. Uosaki, J. Phys. Chem. C, 2009, 113, 11816–11821.
7. S. Y. Chai, Y. J. Kim, M. H. Jung, A. K. Chakraborty, D. Jung and W. I. Lee, J. Catal., 2009, 262, 144–149.
8. C. H. Wang, C. L. Shao, Y. C. Liu and L. Zhang, Scr. Mater., 2008, 59, 332–335.
9. M. Harada and H. Einaga, Langmuir, 2006, 22, 2371–2377.
10. H. G. Kim, P. H. Borse, W. Y. Choi and J. S. Lee, Angew. Chem., Int. Ed., 2005, 44, 4585–4589.
11. R. F. Howe and M. Gratzel, J. Phys. Chem., 1987, 91, 3906–3909.

---

Footnote

† Electronic supplementary information (ESI) available: FE-SEM and TEM images; XRD data; XPS data; electron density contour map for Bi₂O₃. See DOI: 10.1039/b917233e

---