Spontaneous growth of silver nanotrees dominated with (111) crystalline faces on monolithic activated carbon

Abstract

An environmentally friendly wet chemistry approach is used to synthesize silver nanotrees with a single step. The silver nanotrees with single crystalline (111) faces were synthesized without any external reducing or stabilizing agents in water media, involving a galvanic-cell mechanism between the silver nanostructure and biomass-derived monolithic activated carbon (MAC). By varying the concentration of [Ag(NH₃)₂]NO₃, silver nanoflowers, nanowires and nanorods can also be selectively obtained. The silver morphology can also be tuned by some anions such as PO₄³⁻, Cl⁻ and OH⁻ in MAC, which afford a more flexible morphological control in this galvanic-cell mechanism than those reported previously. Importantly, the obtained silver dendrites are found to yield excellent catalytic activity for the reduction of 4-nitroaniline by borohydride.

---

1. Introduction

Controlling the morphology of noble metal nanocrystals has been an active research area as the physical and chemical properties of the metal nanocrystals can be easily tuned by tailoring their size and shape.^1,2 Among various possible morphologies, metal nanocrystals with highly branched morphologies are an exciting new class of nanomaterials due to their attractive hyperbranched structures, physicochemical properties, great potential application in catalysis,^3,4 enhanced fluorescent immunoassays,⁵ photonics applications,⁶ SERS enhancement etc.⁷ However, since most noble metals adopt face-centered cubic (fcc) microstructure, there is no intrinsic driving force for growing into highly anisotropic structure such as hyperbranches. Obtaining multipods of noble metals is therefore extremely challenging.⁸ Until now, much effort has been employed to design hyperbranched silver nanocrystals with well-defined sizes and crystallinity.⁸ Various methods have been developed for the preparation of hyperbranched silver nanocrystals, including electrochemical reduction,⁹ γ-irradiation,¹⁰ and template approaches.¹¹ In most of the preparation processes, organic reducing agents, such as tetrathiafulvalene and ascorbic acid¹² are often used. Some recent efforts have been devoted to the synthesis of dendritic silver through a galvanic replacement reaction by using Zn or Al as templates to react with silver ion with a proper concentration.¹³ In a similar work, morphological transition from dendrites to nanobelts in the electroless deposition of silver by galvanic displacement of copper seeds was investigated; the morphology of the silver metal was largely dependent on the seed size.¹⁴ These galvanic replacement methods avoid the utilization of organic reagents and are more eco-friendly. However, another type of metallic ions is released into the solution upon the synthesis of noble metal nanocrystals.

Monolithic activated carbon (MAC) is a kind of porous carbon materials, which has been extensively used as absorbent, electrode and catalyst support. Recently, we demonstrated that MAC can be used as both support and reductant in the preparation of dendritic silver.¹⁵ A detailed investigation indicated that these silver nanoparticles were formed via the spontaneous reduction by the reductive groups on surface and inside micropores of MAC through a galvanic cell reduction mechanism, similar to the galvanic replacement mechanism mentioned above. Due to abundant functional groups on the surface and inside micropores of MAC, no additional reducing agent was needed in the preparation process. Moreover, unlike the galvanic replacement by another metal, the functional groups are covalently bound to the carbon substrate and may not be released into the solution upon oxidation. In our previous work,¹⁶ the silver nanobelts can grew abundantly on MAC, which provides a straightforward protocol for preparing belt Ag nanostructure. In that work, it was found that the Cl⁻ anions play a key role to induce the formation of the silver nanobelts. Herein, we demonstrate that PO₄³⁻ can be used as another structural inducer for the direct formation of silver metal with various morphologies on the surface of MAC in [Ag(NH₃)₂]⁺ precursor solution at room temperature. The advantages of this silver production method include the simplicity (no need for surfactants), the high yield (almost all of [Ag(NH₃)₂]⁺ precursor can be transformed to silver), the amenability to control and the diversity of silver morphology. In particular, silver nanotrees can be obtained in a high yield with proper [Ag(NH₃)₂]⁺ concentration. The nanotrees are dominated by (111) faces and possess unique local morphology, and exhibit excellent catalytic activity for the reduction of 4-nitroaniline in the presence of borohydride.

2. Experimental procedure

The MAC samples used to grow silver nanotrees were derived from corn straw, and were provided by Qufu Tembton Technology Co. Ltd. Shandong, China. Their structural and compositional information are presented in ESI Tables S1 and S2.† These MAC samples were activated by H₃PO₄ and may contain some PO₄³⁻ residue, which is the structural inducer for the growth of silver nanotree. To change the structural inducer to Cl⁻ or OH⁻, the PO₄³⁻-containing MACs, (hereafter denoted as PO₄³⁻-MAC) were ultrasonicated in HCl (1 mM, 30 min) or NaOH (1 mM, 30 min) aqueous solutions. Elemental analysis with the energy-dispersive X-ray spectrum of Cl⁻-MAC and OH⁻-MAC are presented in ESI Table S3.† To grow silver nanocrystals, MACs were firstly cut into small pieces and ultrasonicated in DI water to remove contaminants. The silver nanocrystals were growing on MACs by immersing MACs into freshly-prepared [Ag(NH₃)₂]NO₃ solution at different concentrations in the dark to avoid silver photoreduction. All experiments were carried out at room temperature (ca. 25 °C). Then, the silver nanocrystal products were detached from MAC with tweezers.

3. Results and discussion

The formation mechanism of silver nanotrees is illustrated schematically in Fig. 1. When PO₄³⁻-MAC contacts with aqueous solution of silver precursor, PO₄³⁻ ions are released and combined with Ag⁺ to form Ag₃PO₄ and deposited on the surface of MAC (Fig. 1, step 1). Similar to the mechanism suggested previously,¹⁶ insoluble silver salt are reduced consequently to metal nanocrystal by the reductive groups on the surface of MAC (Fig. 1, step 2). Due to the good electrical conductivity and abundant functional groups of MAC, a galvanic cell is formed between crystals on the outer surface of MAC (as cathode) and the reductive functional groups on the inner pores (as anode), which are connected by the conductive carbon skeleton (Fig. 1, step 3). The continuous growth of the silver nanotree on the MAC surface is driven actually by the reactions of a galvanic cell.

Fig. 1 The schematic of the proposed mechanism for the in situ growth of silver nanotree on MAC.

Fig. 2 shows the digital photograph, scanning electron microscope (SEM), transmission electron microscope (TEM) images and selected area electron diffraction (SAED) pattern of the silver nanotree, which was prepared by immersing PO₄³⁻-MAC into 10 mM [Ag(NH₃)₂]NO₃ for 24 h. In Fig. 2a, it can be seen abundant silver crystals grow on a small piece of MAC. It is worth noting that silver with multi-level branched nanostructures is similar to trees (Fig. 2b) with stems and leaves (Fig. 2c). The surface of the silver nanotree is rough and protuberant, which are the aggregation of nanoplate-like entities. Notably, silver leaves and branches could be separated by ultrasonic treatment for a few minutes. The silver branches will precipitate off gradually, yielding a colloidal solution of silver leaves. TEM image of silver leaves (Fig. 2d) indicates that the leaves have irregular plate structure. HRTEM images of silver leaves (Fig. 2e) exhibit fringes with a spacing of 2.5 Å, which can be ascribed to the formally forbidden 1/3(422) reflections. The SAED pattern of the silver leaves (Fig. 2f) shows a hexagonal symmetry with two sets of spots. The weak spots (circled) with a spacing of 2.50 Å and a weaker intensity corresponded to the formally forbidden (1/3)(422) reflection. The strong spots (squared) with a spacing of 1.44 Å could be indexed to the allowed (220) reflection (interruption from background made some spots of this strong set look dark). The appearance of weak spots corresponding to the forbidden (1/3)(422) reflection suggests that the facets parallel to the TEM grid are flat. The SAED patterns taken from various regions along an individual structure are similar, indicating that each silver leaf, though irregular in shape, has a dominant crystal orientation.

Fig. 2 Typical images of silver nanotree prepared on MAC. (a) Optical photograph. (b) SEM overview. (c) The SEM of the “leaves” of silver nanotree, as marked by a square in the left part of (b). (d) TEM image of the leaves (e) HRTEM of the square-marked region in (d). (f) Selected-area-electron-diffraction (SAED) pattern of the squared region in (d).

In order to explore the effect of [Ag(NH₃)₂]NO₃ concentration on the morphology evolution of silver nanotree, different concentrations of silver precursor are implemented. Fig. 3 shows SEM images of the products obtained by immersing MAC in [Ag(NH₃)₂]NO₃ solutions with different concentrations. At an extremely high [Ag(NH₃)₂]NO₃ concentration (100 mM), flower-like silver are produced (Fig. 3a). At a moderate [Ag(NH₃)₂]NO₃ concentration (10 mM), the Ag nanotree are formed (Fig. 3c, similar to Fig. 2c). As the concentrations of [Ag(NH₃)₂]NO₃ is reduced to 0.5 mM, silver nanowires (or nanobelts) with good crystallization grows on the MAC surface (Fig. 3b). As the concentration of [Ag(NH₃)₂]NO₃ was further reduced to 0.001 mM, the rodlike silver formed on the MAC (Fig. 3d). Above results seem to indicate the morphology of the silver nanocrystals is strongly dependent on the concentration of the [Ag(NH₃)₂]NO₃, and the silver nanotree can only be abundantly obtained with a proper concentration of [Ag(NH₃)₂]NO₃.

Fig. 3 SEM images of the products obtained by immersing MAC in [Ag(NH₃)₂]NO₃ solutions at concentration of (a) 100 mM, (b) 0.5 mM, (c) 10 mM, (d) 0.001 mM for 2 hours and (e) XRD of silver products corresponding to SEM images (a)–(c) in this figure.

The crystal structures of the silver nanoflower, silver nanowire to silver nanotree was carefully investigated by XRD analysis, and the results are shown in Fig. 3e. The five diffraction peaks can be indexed to diffraction from the (111), (200), (220), (311), and (222) of face-centered cubic (fcc) silver (JCPDS card file, 04-0783). The anisotropy and crystallinity of these silver products were taken from the XRD spectra by comparing the ratios of (111) and (200) peak intensities. The I₍₁₁₁₎/I₍₂₀₀₎ ratio of nanotree is 22.2, which is far higher than the 2.50 of the standard value for fcc silver crystals.¹⁷ This ratio gives a measurement for the relative degree of anisotropy in the crystal growth process, where higher ratios are correlated to the preferential stacking of (111) planes of nanotree. The results indicate that the nanotrees are the mostly dominated by (111) crystal planes.

To understand the exact mechanism responsible for the morphological evolution and the I₍₁₁₁₎/I₍₂₀₀₎ ratio, the time required for full growth of the silver nanocrystals (i.e., the amount and morphology of silver nanocrystals formed on MAC will no longer change after that time) was carefully measured. It was found that full growth of the silver nanoflower, nanowire and nanotree required 1, 3 and 24 hours respectively. Thus, the formation of the silver nanoflower, prepared with high [Ag(NH₃)₂]⁺ concentration, needs shortest time. It seems that with a higher [Ag(NH₃)₂]⁺ concentration, the growth process is mainly controlled by a kinetic factor which is determined by the rate of incorporation of new Ag⁰ adatoms, the (100), (110) and (111) planes grow simultaneously, thereby, I₍₁₁₁₎/I₍₂₀₀₎ will not deviate significantly from standard JCPDS card.¹⁸ On the other hand, silver nanotree is formed in a solution with moderate [Ag(NH₃)₂]⁺ concentration. Under this circumstance, the thermodynamic factor could be more important factor for the growth of silver crystal. As a result, the growth of relatively stable (111) planes is favorable.¹⁹ The morphological change from branch silver to 1-D growth with reduced [Ag(NH₃)₂]⁺ concentration is in agreement with that on kinetically controlled overgrowth.²⁰ The higher concentration of precursor leads to a non-equilibrium growth condition and lower concentration of precursor results in a quasi-equilibrium growth condition. Therefore, a proper silver precursor concentration is important for the formation of silver nanotree crystals.

To further illustrate the growth process of silver nanotree, the morphologies of the silver nanocrystals at different time were evaluated by SEM images. The silver nanocrystals were obtained with a 10 mM [Ag(NH₃)₂]NO₃ solution, in which silver nanotree can be eventually formed, for 1 min, 3 min, 15 min, 30 min, 12 h and 48 h of reaction time. During the first 1 min, many small cubes and some tetrahedrons appeared on the surface of MAC (Fig. 4a). Elemental analysis with the energy-dispersive X-ray spectrum (EDX) (Fig. 4g) illustrate the presence of the elements, such as O, P and Ag, which suggest that the cubes are the compounds of the O, P and Ag elements. Only a small amount of carbon impurity is observed, which possibly originated from MAC. As the reaction proceeded to about 3 min (Fig. 4b), the cubes almost disappeared, particles with some degree of anisotropy, including some multipods, start to appear. After 15 min (Fig. 4c), silver plates with irregular edges dominated the surface of MAC. After 30 min (Fig. 4d), nearly all of these plates became leaf-like silver with indented edges. These indented silver leaves further grow to hyperbranched-structure till the formation of silver nanotrees at about 12 h (Fig. 4e). The structure would not change significantly after that (Fig. 4f). According to the SEM analysis in Fig. 4, we may conclude that insoluble silver phosphate induce the formation of silver nanotree. Moreover, there is a structure transition from the cubic silver phosphate to indented silver plates and finally to dendritic silver nanotree with increasing time.

Fig. 4 SEM images showing the growth process of silver nanotree on MAC. The samples were obtained after immersing MAC in 10 mM, [Ag(NH₃)₂]NO₃ solution for (a) 1 min, (b) 3 min, (c) 15 min, (d) 30 min, (e) 12 h, (f) 24 h and (g) EDX spectrum of the cubic crystal-seed silver in (a).

Dendritic fractals are phenomena generally observed in nonequilibrium growth; the diffusion-limited aggregation mode and the cluster–cluster aggregation mode have been widely used to interpret and analyze these fractal phenomena.²¹ In our reaction system, the lowering of cell potential via using [Ag(NH₃)₂]⁺ precursor, rather than Ag⁺, together with the mild reducing capacity of biomass-derived monolithic activated carbon, create a stable environment growth of silver nanotree. At the early stage, the reaction process is dominated by a nonequilibrium condition due to higher concentration, and a dendritic morphology is formed. If the reaction time is long enough, the precursor concentration should drop to such a level that the reaction process is dominated by quasi-equilibrium or equilibrium conditions. Therefore, plenty of leaves are growing on the silver branches. We believe that the observations here imply an evolution from nonequilibrium to equilibrium. It is consistent with the predictions from Nernst equation.^16,20 The decrease of the precursor concentration will result in the change of cell potential, which, in turn, has led to the preparation of silver crystals with different morphology.²⁰

In addition to concentration of the silver precursor solution, it was found that the type of anions of inorganic salts, such as Cl⁻, OH⁻ or PO₄³⁻ in MAC also has great influence on the morphology of silver nanocrystal. Fig. 5 shows SEM images of the silver nanocrystals obtained on Cl⁻-MAC, PO₄³⁻-MAC, and OH⁻-MAC respectively, (refer to Experimental procedure and ESI Tables S1–S3† for the preparation and structural/elemental properties of these materials) by immersing the MACs in 10 mM [Ag(NH₃)₂]NO₃ solution. In Fig. 5a, silver belt was obtained on Cl⁻-MAC, which is consistent with our previous work.¹⁵ The spherical silver appeared on the surface of OH⁻-MAC (Fig. 5c). At the same [Ag(NH₃)₂]NO₃ concentration (10 mM), the Ag nanotree grow on the surface of PO₄³⁻-MAC with PO₄³⁻ (Fig. 5b, this work). These results demonstrate inorganic salt in MAC can induce the formation of different silver morphology. From the above experiments, it was found that the silver nanotree only grew abundantly on PO₄³⁻-MAC.

Fig. 5 SEM images of MAC immersed in 10 mM [Ag(NH₃)₂]NO₃, (a) silver nanobelt on MAC with Cl⁻, (b) silver nanotree on MAC with PO₄³⁻ and (c) spherical silver on MAC with OH⁻.

In previous work, the formation of insoluble AgCl on MAC in the initial reaction step is also found to be indispensable for the growth of Ag nanobelts with single crystalline.¹⁵ However, hollow silver cubes were morphosynthesized (through AgCl cubes) on MAC with enhanced Cl⁻ pre-absorption,²² instead of a continuous heterogeneous seeded growth. Different kinds inorganic salt crystalline seed initiation in MAC change the initial cell potential through galvanic cell. Compare the standard reduction of AgCl|Ag (0.22 V), Ag₂O|Ag (0.34 V), [Ag(NH₃)₂]NO₃|Ag (0.38 V), Ag₃PO₄|Ag (0.69 V) and AgNO₃|Ag (0.80 V), it could be observed that there is significant increase in the reduction potential. Moreover, the high solubility of [Ag(NH₃)₂]NO₃ and the complexation structure may provide a good buffer solution of Ag⁺ near the Ag deposition region. Because of kinetically controlled overgrowth, the higher initial potential lead to faster growth rate, aggregation of crystal and the gradual growth of dendritic silver crystals. The exact mechanism for the morphological control of Ag crystals by inorganic salt crystalline seed initiation is worthy of further investigation.

The catalytic activity of the resulting silver nanostructures was examined for reduction of 4-nitroaniline to p-phenylenediamine, which is an important chemical for the synthesis of textile and rubber antioxidants.²³ Traditional catalytic reduction of 4-nitroaniline was mostly conducted by using PbBi₂Nb₂O₉ and metal nanoparticles.²⁴ In an early study, crystal facet-dependent catalytic activity of different lattice planes of gold toward 4-nitroaniline reduction indicates that the catalytic activity for the reduction reaction follows the order of (110) > (100) > (111).²⁵ In this work, silver nanotrees and nanowires, which have different ratio of crystal facets, were compared for the catalytic reduction of 4-nitroaniline. For a typical catalysis reaction, 1 ml of 15 mM sodium borohydride aqueous solution was added to 1.5 ml 4-nitroaniline (0.1 mM) aqueous solution. The reaction mixture was maintained at 25 °C using a water bath. The absorption spectra of the mixture were recorded every 10 second in the range of 300–650 nm immediately after adding silver nanotree or silver nanowire (ca. 10 mg) to the reaction mixture. Fig. 6 shows the UV-vis absorption spectra of the mixture as a function of catalytic reduction time. Before initialization of the catalytic reaction, a strong characteristic absorption peak ascribed to 4-nitroaniline was observed at 380 nm. This band diminished quickly after introduction of the silver nanotree or nanowire to the aqueous solution of 4-nitroaniline/NaBH₄, indicating the occurrence of 4-nitroaniline reduction.²⁶ The reduction finished in 120 s and 190 s for the silver nanotrees and nanowires, respectively (Fig. 6a and b). The plot of ln A against reaction time (t) in Fig. 6c exhibits a linear correlation, suggesting the pseudo-first order kinetic model of the reduction process. The rate constant k of the silver products for the reduction are 9.708 × 10⁻¹ min⁻¹ (silver nanotrees), 4.704 × 10⁻¹ min⁻¹ (silver nanowires), respectively, which were calculated by the slopes of the fitted lines. Because the ratio of (111) : (200) for silver nanotree is much larger than that for silver nanowire, it seems that the (111) crystal facet for silver nanostructures is more favourable to the catalytic reduction of 4-nitroaniline than the other crystal facets. The crystal facet-dependent catalytic behaviour might be associated with the surface adsorption of 4-nitroaniline on silver metal as was demonstrated previously.²⁵ Moreover, the rate constant k obtained in the present study has increased 1–2 orders as compared with the values reported in the literature under similar reaction conditions.²³ The results clearly indicate that the high-yield and low-cost silver nanotree can be employed as an excellent catalyst for the reduction of 4-nitroaniline to p-phenylenediamine.

Fig. 6 Time-dependent UV-vis absorption spectra for the borohydride reduction of 4-NA using (a) silver nanotree, (b) silver nanowire catalysts. (c) ln A versus time plots using silver nanotree and silver nanowire, as catalysts at 25 °C.

4. Conclusion

In summary, we have successfully synthesized Ag nanostructures with multiple morphologies within a facile and organic-free reaction system by simply adjusting the silver ion concentration and the kinds of inorganic salts loaded on monolithic activated carbon. Among different morphologies, silver nanotrees were obtained in a high yield, and the growth process and mechanism are analysed in this paper. Our experiments provide very useful information to understand the growth mechanism of morphology and/or structure-controlled nanocrystals. The present experimental method has also been extended to other systems to investigate the growth behavior of Pd and Au (ESI S4†) branches. The obtained silver nanotree are found to yield excellent catalytic activity for the reduction of 4-nitroaniline by borohydride. This economical and environmental sustainable synthetic concept could ultimately enable the fine-tuning of material responses to magnetic, electrical, optical, and mechanical stimuli.

Acknowledgements

Project supported by the National Natural Science Foundation of China (No. 21601059), the National Natural Science Foundation of China (No. 21473067), the Science and Technology Department of Jiangsu Province (No. BY2014085), the Education Department of Jiangsu Provincial (No. 14 KJB430006), the State Oceanic Administration (No. 201305007) and the Environmental Protection Department of Jiangsu Province (No. 2012010).

Notes and references

1. M. Rycenga, C. M. Cobley, J. Zeng, W. Y. Li, C. H. Moran, Q. Zhang, D. Qin and Y. N. Xia, Chem. Rev., 2011, 111, 3669–3712.
2. B. Wiley, Y. G. Sun, B. Mayers and Y. N. Xia, Shape-Controlled Synthesis of Metal Nanostructures, Chem. –Eur. J., 2005, 11, 454–463.
3. V. Polshettiwar, B. Baruwati and R. S. Varma, Synthesis and application in catalysis, ACS Nano, 2009, 3, 728–736.
4. B. Lim, M. Jiang, P. H. C. Camargo, E. C. Cho, J. Tao, X. M. Lu, Y. M. Zhu and Y. N. Xia, Pd–Pt bimetallic nanodendrites with high activity for oxygen reduction, Science, 2009, 324, 1302–1305.
5. T. Shtoyko, E. G. Matveeva, I. F. Chang, Z. Gryczynski, E. Goldys and I. Gryczynski, Enhanced fluorescent immunoassays on silver fractal-like structures, Anal. Chem., 2008, 80, 1962–1966.
6. K. A. Dick, K. Deppert, M. W. Larsson, T. Mårtensson, W. Seifert, L. R. Wallenberg and L. Samuelson, Nat. Mater., 2004, 3, 380–384.
7. Y. Wang, P. H. Camargo, S. E. Skrabalak, H. C. Gu and Y. N. Xia, Langmuir, 2008, 24, 12042–12046.
8. B. Lim and Y. N. Xia, Angew. Chem., Int. Ed., 2011, 50, 76–85.
9. X. Qin, H. Wang, X. Wang, Z. Miao, Y. Fang, Q. Chen and X. Shao, Electrochim. Acta, 2011, 56, 3170–3174.
10. S. Wang and H. Xin, J. Phys. Chem. B, 2000, 104, 5681–5685.
11. J. P. Xiao, Y. Xie, R. Tang, M. E. N. G. Chen and X. B. Tian, Adv. Mater., 2001, 13, 1887–1891.
12. X. Wang, K. Naka, H. Itoh, S. Park and Y. Chujo, Chem. Commun., 2002, 1300–1301.
13. X. Bai, Y. Gao and L. Q. Zheng, CrystEngComm, 2011, 13, 3562–3568.
14. A. V. Avizienis, C. Martin-Olmos, H. O. Sillin, M. Aono, J. K. Gimzewski and A. Z. Stieg, Cryst. Growth Des., 2013, 13, 465–469.
15. F. Wang, Y. J. Lai, B. Y. Zhao, X. B. Hu, D. Zhang and K. A. Hu, Chem. Commun., 2010, 46, 3782–3784.
16. H. Zhao, F. Wang, Y. S. Ning, B. Y. Zhao, F. J. Yin, Y. J. Lai, J. W. Zheng, X. B. Hu, T. X. Fan, D. Zhang and K. A. Hu, Sci. Rep., 2013, 3, 1511.
17. L. Fan and R. Guo, Cryst. Growth Des., 2008, 8, 2150–2156.
18. Z. L. Wang, J. Phys. Chem. B, 2000, 104, 1153–1175.
19. X. Qin, Z. Miao, Y. Fang, D. Zhang, J. Ma, L. Zhang, Q. Chen and X. Shao, Langmuir, 2012, 28, 5218–5226.
20. R. Liu and A. Sen, Chem. Mater., 2011, 24, 48–54.
21. J. Fang, H. You, P. Kong, Y. Yi, X. Song and B. Ding, Cryst. Growth Des., 2007, 7, 864–867.
22. F. Wang, H. Zhao, Y. Lai, S. Liu, B. Zhao, Y. Ning and X. Hu, Phys. Chem. Chem. Phys., 2013, 15, 18367–18370.
23. Y. Chu, Z. Wang and Q. M. Pan, ACS Appl. Mater. Interfaces, 2014, 6, 8378–8386.
24. W. Wu, G. Liu, S. Liang, Y. Chen, L. Shen, H. Zheng, H. R. Yuan, Y. Hou and L. Wu, J. Catal., 2012, 290, 13–17.
25. C. Y. Chiu, P. J. Chung, K. U. Lao, C. W. Liao and M. H. Huang, J. Phys. Chem. C, 2012, 116, 23757–23763.
26. Q. Zhou, G. Qian, Y. Li, G. Zhao, Y. Chao and J. Zheng, Thin Solid Films, 2008, 516, 953–956.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra19399d

‡ Y. Chen and Y. Ning contributed equally to this work.

---