Enhanced microwave absorption material of ternary nanocomposites based on MnFe₂O₄@SiO₂, polyaniline and polyvinylidene fluoride

Abstract

The core–shell structural MnFe₂O₄@SiO₂ nanocomposite has been successfully fabricated on a large-scale via a simple mechanical stirring method at room temperature. Subsequently, the synthesized MnFe₂O₄@SiO₂ nanoparticles were compounded with polyaniline (PANI) and polyvinylidene fluoride (PVDF) to form ternary nanocomposites. The composites filled with 20 wt% MnFe₂O₄@SiO₂ and 20 wt% PANI in a PVDF matrix show the most excellent wave absorption performance. The minimum reflection loss value can reach −25.73 dB at 12.32 GHz, and the effective frequency bandwidth (RL < −10 dB) ranges from 10.72 to 14.40 GHz. Moreover, the possible microwave absorption mechanism has been also discussed in detail.

---

Introduction

Nowadays, with the development of GHz telecommunications, military and commercial radar systems and other communication equipment, electromagnetic (EM) radiation has been considered to be the fourth largest source of environmental pollution followed by water, air and noise pollution.^1,2 This EM radiation can interfere with the precision of electronic equipment, generate energy consumption and be harmful to human health as well.^3,4 In order to avoid these serious consequences, the development of microwave absorption (MA) materials has been extensively studied. In the past few decades, exploratory efforts have focused on ferrite absorbing materials such as Ni–Zn ferrite,⁵ CoFe₂O₄,^6,7 Fe₃O₄,^8,9 and Co₃O₄.^10,11 However, the large density, poor corrosion resistance, narrow absorption bandwidth and agglomeration effects of ferrite absorbers greatly restrict their technical applications.¹²

To improve the chemical homogeneity and enhance the EM wave absorption properties of ferrites, many scientists have shown much interest in constructing core–shell structures. For example, Du et al. prepared core–shell Fe₃O₄@C composites with different thicknesses of the carbon shell, and their unique microstructure endowed them with enhanced microwave absorption properties.¹³ In addition, Liu et al. synthesized a series of Fe₃O₄/C core–shell nanospindles with different shell thickness and then embedded them into a polyvinylidene fluoride (PVDF) matrix. The results displayed that the Fe₃O₄/C/PVDF composite with a thick carbon shell exhibited strong electromagnetic wave absorbing ability. The minimum reflection loss reached −38.8 dB with a thickness of 2.1 mm.¹⁴ Mesoporous Fe₃O₄@ZnO sphere decorated graphene (GN–pFe₃O₄@ZnO) composites with uniform size were synthesized by Sun et al. EM wave absorption properties of epoxy containing 30 wt% GN–pFe₃O₄@ZnO were investigated and the results indicated that the absorption bandwidth with reflection loss (RL) values less than −10 dB was up to 11.4 GHz, and the minimal RL was almost −40 dB.¹⁵ Furthermore, Ren et al. fabricated a three-dimensional (3D) SiO₂@Fe₃O₄ core/shell nanorod array/graphene architecture. The measured electromagnetic parameters showed that the 3D architecture exhibited excellent electromagnetic wave absorption properties. More than 99% of electromagnetic wave energy could be attenuated by the 3D architecture with an addition amount of only 20 wt% in the paraffin matrix.¹⁶

Apart from constructing core–shell structural nanomaterials, another effective approach for improving their microwave absorption properties is to blend dielectric and magnetic materials together. This is mainly because pure ferrite has high permeability and low permittivity, leading to poor impedance matching. In recent studies, as a typical dielectric loss material, polyaniline (PANI) has been extensively used in microwave absorption applications due to its high conductivity, broad response bandwidth, low density, easy preparation, and good environmental stability.¹⁷ To obtain a proper impedance matching between dielectric loss and magnetic loss, PANI is often mixed with magnetic materials to fabricate enhanced electromagnetic wave absorbers. The FeNi₃@SiO₂ nanoparticles were successfully synthesized by Ding et al. and dispersed into the rGO–PANI nanosheets through a hydrothermal process. The minimum reflection loss of FeNi₃@SiO₂@rGO–PANI achieved was −40.18 dB at 14.0 GHz with the thickness of 2.4 mm and the effective absorption bandwidth (RL < −10 dB) was 6.64 GHz.¹⁸ Wang et al. synthesized a ternary composite of Ni/polyaniline (PANI)/reduced graphene oxide (RGO) and proved that the as-prepared Ni/PANI/RGO composite showed enhanced electromagnetic absorption properties compared with Ni/RGO.¹⁹ Liu et al. fabricated core-multishell MWCNT/Fe₃O₄/polyaniline (PANI)/Au hybrid nanotubes by a facile layer-by-layer technique and investigated their microwave absorption properties. It turned out that the lowest reflection loss of MWCNT/Fe₃O₄/PANI/Au hybrid nanotubes could reach as low as −60 dB.²⁰ Moreover, the excellent EM wave absorbency of a carbon fiber (CF)/Co_0.2Fe_2.8O₄/PANI heterostructure composite and PANI/carbonyl iron powder (CIP)/Fe₃O₄ composite were studied by Fang et al.²¹ and He et al.,²² respectively.

Based on the abovementioned research, our group synthesized spinel MnFe₂O₄ nanoparticles by a simple hydrothermal process and then coated them with SiO₂ to form a core–shell structural MnFe₂O₄@SiO₂ nanocomposite. Subsequently, according to the electromagnetic complementary theory, the magnetic MnFe₂O₄@SiO₂ nanocomposite and conducting polyaniline (PANI) were mixed with a polyvinylidene fluoride (PVDF) matrix simultaneously to develop the ternary MnFe₂O₄@SiO₂/PANI/PVDF composites. We also investigated their EM wave absorption properties in detail.

Experimental section

The conducting polyaniline (PANI) was purchased from Cool Chemical Science and Technology (Beijing) Co., LTD. All other chemicals were of analytical grade and used without further purification.

Preparation of core–shell structural MnFe₂O₄@SiO₂ nanocomposite

The spinel MnFe₂O₄ nanoparticles were synthesized according to our previous research.²³ To obtain the MnFe₂O₄@SiO₂ nanocomposite, 100 mg of MnFe₂O₄ nanoparticles was added to a mixture composed of 20 mL of absolute ethanol, 3 mL of deionized water and 2 mL of ammonia water. After sonicating for 30 min, the suspension was poured into a three-necked flask with a volume of 250 mL. Subsequently, 100 μL of tetraethyl orthosilicate (TEOS) was added into the suspension and then stirred mechanically for another 30 min. Finally, the resultant brown product was washed with deionized water and absolute ethanol several times and dried at 60 °C for 12 h for further characterization.

Preparation of MnFe₂O₄/PVDF, MnFe₂O₄@SiO₂/PVDF and MnFe₂O₄@SiO₂/PANI/PVDF films

MnFe₂O₄/PVDF, MnFe₂O₄@SiO₂/PVDF and MnFe₂O₄@SiO₂/PANI/PVDF films were obtained by dissolving PVDF in N,N-dimethylformamide (DMF) under magnetic stirring at room temperature. After the solution was transparent, various amounts of MnFe₂O₄ nanoparticles, MnFe₂O₄@SiO₂ nanocomposite, and PANI were added and sonicated for 30 min and then poured into glass Petri dishes. Finally, it was dried in an oven at 100 °C for 3 h.

Characterization

XRD analyses were carried out on an X-ray diffractometer (D/MAX-1200, Rigaku Denki Co. Ltd, Japan). The XRD patterns with Cu Kα radiation (λ = 1.5406 Å) at 40 kV and 40 mA were recorded in the range of 2θ = 10–70°. Scanning electron microscopy (SEM) images were achieved by a FEI Quanta 250 field-emission gun environmental scanning electron microscope at 15 kV with the samples obtained from the thick suspension dropping on the silicon slice. Field emission scanning electron microscopy (FE-SEM) was conducted on a JSM-6700F microscope. Transmission electron microscopy (TEM) and high-resolution TEM (HRTEM) investigations were carried out on a JEOL TEM-2100 microscope. The magnetic properties were carried out on a Quantum Design superconducting quantum interference device (SQUID) magnetometer (MPMS-7) at 300 K.

EM absorption measurement

The composites used for EM absorption measurements were prepared by mixing the products with PVDF in different mass percentages, respectively. The mixtures were then pressed into cylindrical-shaped samples (Φ_out = 7.00 mm and Φ_in = 3.04 mm). The complex permittivity and permeability values were measured in the 2–18 GHz range with a coaxial wire method by an Agilent N5230C PNA-L Network Analyzer.

Results and discussion

Fig. 1a shows the typical XRD pattern of the synthesized MnFe₂O₄ nanoparticles, MnFe₂O₄@SiO₂ nanocomposite, MnFe₂O₄@SiO₂/PANI/PVDF nanocomposites and PVDF. All the diffraction peaks can be readily indexed to the face-centered cubic structure of jacobsite ferrite (JCPDS no. 10-0319). The narrow sharp peaks confirm that the MnFe₂O₄ nanoparticles are highly crystallized without any other impurities. In addition, the comparison of the XRD pattern for MnFe₂O₄@SiO₂/PVDF nanocomposites and PVDF indicates that the MnFe₂O₄@SiO₂ nanoparticles compound well with PVDF. From the SEM images of the MnFe₂O₄ and MnFe₂O₄@SiO₂ nanoparticles in Fig. 1b and c, it can be seen that the average diameter of the MnFe₂O₄ nanoparticle and MnFe₂O₄@SiO₂ nanocomposite in the ranges of 50–80 nm and 100–150 nm, respectively. Moreover, from the TEM image and high-resolution TEM image of the MnFe₂O₄@SiO₂ nanocomposite in Fig. 1d and Fig. S1,† it is obviously seen that the MnFe₂O₄ nanoparticles are coated with SiO₂. The point-scan EDX spectrum displayed in Fig. S2† also verifies the TEM result. Moreover, after coating with SiO₂, the dispersity of the MnFe₂O₄@SiO₂ nanocomposite in DMF is much better than that of the MnFe₂O₄ nanoparticles, as shown in Fig. S3.†

Fig. 1 (a) XRD patterns of a. MnFe₂O₄ nanoparticles, b. MnFe₂O₄@SiO₂ nanocomposites, c. MnFe₂O₄@SiO₂/PVDF membrane, d. PVDF; (b) SEM image of MnFe₂O₄ nanoparticles; (c) SEM image and (d) TEM image of MnFe₂O₄@SiO₂ nanocomposites.

In order to determine how the MnFe₂O₄@SiO₂ nanocomposite and PANI disperse in PVDF, the FESEM characterization and elemental maps of MnFe₂O₄@SiO₂/PANI/PVDF membrane are displayed in Fig. 2. The elemental mapping of N, Si, F, Mn and Fe in area-scanning mode indicates that the PANI and MnFe₂O₄@SiO₂ nanocomposite disperse well in the PVDF matrix.

Fig. 2 FESEM image of the fracture section of the MnFe₂O₄@SiO₂/PANI/PVDF membrane and corresponding elemental mapping images of N, Si, F, Mn and Fe.

The magnetic properties of the as-synthesized MnFe₂O₄ nanoparticles and MnFe₂O₄@SiO₂ nanocomposite were measured at room temperature. As observed in Fig. 3a, both samples show a typical hysteresis loop in their magnetic behavior. The magnified views of the hysteresis loops for these two samples at low applied fields (Fig. 3b and c) indicate that they are typical soft magnetic materials. The saturation magnetization (M_s), remanent magnetization (M_r), and coercivity (H_c) values of the MnFe₂O₄ nanoparticles and MnFe₂O₄@SiO₂ nanocomposite are 65.1 emu/g, 4.37 emu/g, 35.5 Oe and 47.9 emu/g, 3.98 emu/g, 52.7 Oe, respectively. The difference in M_s values of these two nanoparticles may be attributed to the cladding of nonmagnetic SiO₂.²⁴

Fig. 3 (a) Room-temperature hysteresis loops of MnFe₂O₄ nanoparticles and MnFe₂O₄@SiO₂ nanocomposites. The magnified views of the hysteresis loops at low applied fields of (b) MnFe₂O₄ nanoparticles and (c) MnFe₂O₄@SiO₂ nanocomposites.

To investigate the microwave absorption property, the reflection loss (RL) coefficients of the as-obtained samples were calculated from the measured relative complex permittivity and permeability values. According to transmission line theory, the reflection loss (RL) can be defined with the following equations:²⁵

where Z_in is the input characteristic impedance, which can be expressed as:²⁶where ε_r and μ_r are the complex permittivity and permeability of the composite absorber, respectively, ƒ is the frequency, d is the thickness of the absorber, and c is the velocity of light in free space.

Fig. 4a shows the theoretical reflection loss (RLs) of various nanocomposites in the frequency range of 2–18 GHz with a thickness of 2.0 mm. It turns out that the ternary MnFe₂O₄@SiO₂/PANI/PVDF composites with 20 wt% MnFe₂O₄@SiO₂ and 20 wt% PANI present the most enhanced EM wave absorption property. The minimum reflection loss value can reach −25.73 dB at 12.32 GHz, and the effective frequency bandwidth (RL < −10 dB) is from 10.72 to 14.40 GHz. Moreover, as shown in Fig. 4b–f, the three-dimensional presentations of the calculated theoretical RLs for all nanocomposites with different thicknesses (2–5 mm) in the range of 2–18 GHz indicate that the microwave absorption ability can be tuned effectively by the thickness and filler contents of the absorbers.

Fig. 4 (a) Microwave RL curves of the composites with a thickness of 2.0 mm in the frequency range of 2–18 GHz. Three-dimensional representations of the RL of (b) 20 wt% MnFe₂O₄/PVDF composites; (c) 20 wt% MnFe₂O₄@SiO₂/PVDF composites; (d) 20 wt% PANI/PVDF composites; (e) 20 wt% MnFe₂O₄@SiO₂ + 20 wt% PANI/PVDF composites; and (f) 30 wt% MnFe₂O₄@SiO₂ + 20 wt% PANI/PVDF composites.

In order to study the possible microwave absorption mechanism, the complex permittivity and permeability of various samples was measured in the frequency range of 2–18 GHz, where the real parts (ε′ and μ′) symbolize the storage capability of the electric and magnetic energy, and the imaginary parts (ε′′ and μ′′) stand for the dissipation loss capability.²⁷ The frequency dependence of the relative permittivity and permeability is shown in Fig. 5. For ternary MnFe₂O₄@SiO₂/PANI/PVDF composites, the ε′ and ε′′ values increase largely compared with that of MnFe₂O₄/PVDF and MnFe₂O₄@SiO₂/PVDF nanocomposites after the interfusion of PANI (Fig. 5a and b). This is mainly because of the higher electrical conductivity of PANI. However, as shown in Fig. 5c and d, the μ′ and μ′′ values are relative poor. The increment of ε′ may be attributed to the electronic and dipolar polarization provided mainly by PANI. More polarization will result in more energy dissipation. With increasing frequency, the variation of ε′ and ε′′ changes slowly due to the movement of electrons and the dipole cannot keep up with the variation of frequency.²⁸

Fig. 5 Frequency dependence on (a) real and (b) imaginary parts of the complex permittivity; (c) real and (d) imaginary parts of the complex permeability of samples.

Except for electronic and dipolar polarization, interfacial polarization is another important polarization mechanism. In general, interface polarization arises when the dielectric constant or conductivity of neighboring phases differ from each other at the testing frequencies.²⁹ The combination of MnFe₂O₄@SiO₂ and PANI with PVDF will contribute to form more of an interface, leading to more interfacial polarization. Furthermore, the difference in complex permittivity between MnFe₂O₄@SiO₂, PANI and PVDF would generate interface scattering, which is beneficial to the wave absorption property.³⁰

In addition, we calculated the dielectric loss tangents (tan δ_ε = ε′′/ε′) and magnetic loss tangents (tan δ_μ = μ′′/μ′) of all samples. As displayed in Fig. S4,† it was found that the maximum values of tan δ_ε and tan δ_μ are 0.53 and 0.22, respectively. The mixture of MnFe₂O₄@SiO₂ and PANI show better complementarities between the dielectric loss and the magnetic loss, which results in a significantly enhanced microwave absorption performance.³¹ Nevertheless, the higher tan δ_ε values illustrate that the main influencing factor for wave absorption performance is dielectric loss. Moreover, as shown in Fig. 6, with the mixture of PANI, enhanced electrical conductivity facilitates a large attenuation constant, which can be written according to transmission line theory as:

where f is the frequency of the electromagnetic wave and c is the velocity of light. The larger attenuation constant will result in more dielectric loss, which is beneficial to the wave absorption performance.

Fig. 6 Attenuation constants of various samples.

With the difference of dielectric loss, the magnetic loss of classified ferrite originates from eddy current loss, hysteresis loss, residual loss, ferromagnetic resonance loss and intragranular domains wall loss on the basis of Van Der Zaag's research.³² However, the hysteresis loss is negligible in the weak field, and the domain wall resonance loss commonly occurs at MHz frequencies.³³ Thus, the eddy current loss and natural resonance may be responsible for the EM wave attenuation over the 2–18 GHz frequency range. The eddy current loss can be evaluated by the following equation:

μ′′ ≈ 2πμ₀(μ′)²σd²f/3 (3)

where σ (S cm⁻¹) is the electrical conductivity and μ₀ (H m⁻¹) is the permeability in vacuum. If the reflection loss results from the eddy current loss effect, the values of C₀ (C₀ = μ′′(μ′)⁻²f ⁻¹) are constant when the frequency varies. From Fig. 7, it can be observed that the value of C₀ is almost constant within the frequency range from 10 to 18 GHz, indicating that the ferromagnetic MnFe₂O₄ nanoparticles possess an obvious eddy current effect for the microwave energy dissipation.

Fig. 7 The C₀-f curve of various samples.

Another mechanism for magnetic loss is natural resonance, which can be expressed by the following equation,

H_a = 4|K₁|/3μ₀M_s (4)

where H_a is the anisotropy energy, |K₁| is the anisotropic coefficient, and M_s is the saturation magnetization. As shown in Fig. 3a, the M_s value of MnFe₂O₄@SiO₂ nanocomposites is lower than that of MnFe₂O₄ nanoparticles. Thus, the anisotropic energy is higher for the MnFe₂O₄@SiO₂ composites, which is helpful for the improvement of the EM absorption performance especially at high frequency.^34,35

Conclusion

In summary, the ternary MnFe₂O₄@SiO₂/PANI/PVDF composites have been fabricated by a simple method. Through the investigation of their microwave absorption properties, the results indicated that the composites filled with 20 wt% MnFe₂O₄@SiO₂ and 20 wt% PANI in PVDF matrix show the most excellent wave absorption performance in a frequency range of 2–18 GHz. The minimum reflection loss value can reach −25.73 dB at 12.32 GHz, and the effective frequency bandwidth (RL < 10 dB) falls in a range from 10.72 to 14.40 GHz. The main loss mechanism includes electronic and dipolar polarization, interfacial polarization, eddy current loss and natural resonance loss.

Acknowledgements

This study was supported by the National Natural Science Foundation of China (51472012) and the Fundamental Research Funds for the Central Universities.

Notes and references

1. Y. Zhang, Y. Huang, T. F. Zhang, H. C. Chang, P. S. Xiao, H. H. Chen, Z. Y. Huang and Y. S. Chen, Adv. Mater., 2015, 27, 2049–2053.
2. T. Xia, C. Zhang, N. A. Oyler and X. B. Chen, Adv. Mater., 2013, 25, 6905–6910.
3. R. B. Wu, K. Zhou, Z. H. Yang, X. K. Qian, J. Wei, L. Liu, Y. Z. Huang, L. B. Kong and L. Y. Wang, CrystEngComm, 2013, 15, 570–576.
4. Y. X. Gong, L. Zhen, J. T. Jiang, C. Y. Xu, W. S. Wang and W. Z. Shao, CrystEngComm, 2011, 13, 6839–6844.
5. K. K. Gupta, S. M. Abba, T. H. Goswami and A. C. Abhyankar, J. Magn. Magn. Mater., 2014, 362, 216–225.
6. P. B. Liu, Y. Huang and X. Zhang, Powder Technol., 2015, 276, 112–117.
7. P. B. Liu, Y. Huang and X. Zhang, Synth. Met., 2015, 201, 76–81.
8. H. Zhang, M. Hong, P. Chen, A. J. Xie and Y. H. Shen, J. Alloys Compd., 2016, 665, 381–387.
9. J. Zheng, H. L. Lv, X. H. Lin, G. B. Ji, X. G. Li and Y. W. Du, J. Alloys Compd., 2014, 589, 174–181.
10. X. B. Li, S. W. Yang, J. Sun, P. He, X. P. Pu and G. Q. Ding, Synth. Met., 2014, 194, 52–58.
11. G. S. Wang, Y. Wu, Y. Z. Wei, X. J. Zhang, Y. Li, L. D. Li, B. Wen, P. G. Yin, L. Guo and M. S. Cao, ChemPlusChem, 2014, 79, 375–381.
12. X. D. Cai, J. J. Wang, B. C. Li, A. H. Wu, B. C. Xu, B. Wang, H. T. Gao, L. Yu and Z. Li, J. Alloys Compd., 2016, 657, 608–615.
13. Y. C. Du, W. W. Liu, R. Qiang, Y. Wang, X. J. Han, J. Ma and P. Xu, ACS Appl. Mater. Interfaces, 2014, 6, 12997–13006.
14. X. F. Liu, X. R. Cui, Y. X. Chen, X. J. Zhang, R. H. Yu, G. S. Wang and H. Ma, Carbon, 2015, 95, 870–878.
15. D. P. Sun, Q. Zou, Y. P. Wang, Y. J. Wang, W. Jiang and F. S. Li, Nanoscale, 2014, 6, 6557–6562.
16. Y. L. Ren, C. L. Zhu, S. Zhang, C. Y. Li, Y. J. Chen, P. Gao, P. P. Yang and Q. Y. Ouyang, Nanoscale, 2013, 5, 12296–12303.
17. P. Zhang, X. J. Han, L. L. Kang, R. Qiang, W. W. Liu and Y. C. Du, RSC Adv., 2013, 3, 12694–12701.
18. X. Ding, Y. Huang, J. G. Wang, H. W. Wu and P. B. Liu, Appl. Surf. Sci., 2015, 357, 908–914.
19. Y. Wang, X. M. Wu, W. Z. Zhang and S. Huang, Synth. Met., 2015, 210, 165–170.
20. C. Y. Liu, Y. J. Xu, L. N. Wu, Z. H. Jiang, B. Z. Shen and Z. J. Wang, J. Mater. Chem. A, 2015, 3, 10566–10572.
21. J. Y. Fang, Z. Chen, W. Wei, Y. Li, T. Liu, Z. Liu, X. G. Yue and Z. H. Jiang, RSC Adv., 2015, 5, 50024–50032.
22. Z. F. He, Y. Fang, X. J. Wang and H. Pang, Synth. Met., 2011, 161, 420–425.
23. X. J. Zhang, G. S. Wang, W. Q. Cao, Y. Z. Wei, J. F. Liang, L. Guo and M. S. Cao, ACS Appl. Mater. Interfaces, 2014, 6, 7471–7478.
24. X. Lin, X. Lv, L. Wang, F. Zhang and L. Duan, Mater. Res. Bull., 2013, 48, 2511–2516.
25. G. S. Wang, L. Z. Nie and S. H. Yu, RSC Adv., 2012, 2, 6216–6221.
26. M. S. Cao, X. L. Shi, X. Y. Fang, H. B. Jin, Z. L. Hou, W. Zhou and Y. J. Chen, Appl. Phys. Lett., 2007, 91, 203110–203113.
27. X. M. Zhang, G. B. Ji, W. Liu, B. Quan, X. H. Liang, C. M. Shang, Y. Cheng and Y. W. Du, Nanoscale, 2015, 7, 12932–12942.
28. Y. J. Chen, P. Gao, R. X. Wang, C. L. Zhu, L. J. Wang, M. S. Cao and H. B. Jin, J. Phys. Chem. Lett., 2009, 113, 10061–10064.
29. T. Liu, P. H. Zhou, J. L. Xie and L. J. Deng, J. Appl. Phys., 2011, 110, 033918.
30. M. Fu, Q. Jiao and Y. Zhao, J. Mater. Chem. A, 2013, 1, 5577–5586.
31. J. W. Liu, R. C. Che, H. J. Chen, F. Zhang, F. Xia, Q. S. Wu and M. Wang, Small, 2012, 8, 1214–1221.
32. P. J. van der Zaag, J. Magn. Magn. Mater., 1999, 196–197, 315–319.
33. M. Z. Wu, Y. D. Zhang, S. Hui, T. D. Xiao, S. H. Ge, W. A. Hines, J. I. Budnick and G. W. Taylor, Appl. Phys. Lett., 2002, 80, 4404–4406.
34. J. Guo, X. Wang, P. Miao, X. Liao, W. Zhang and B. Shi, J. Mater. Chem., 2012, 22, 11933–11942.
35. J. Zhu, S. Wei, N. Haldolaarachchige, D. P. Young and Z. Guo, J. Phys. Chem. C, 2011, 115, 15304–15310.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra17076e

‡ These authors contributed equally to this work.

---