Flexible solar cells based on graphene-ultrathin silicon Schottky junction

Abstract

We developed a flexible graphene–silicon (Gr–Si) photovoltaic device with high reliability and stability. Ultrathin Si film was fabricated via an anisotropic Si etching method, and exhibited excellent flexibility. Different from the traditional graphene transfer approach, polymethylmethacrylate (PMMA) film remained, by which the physical damage of graphene resulting from the PMMA dissolution process is avoided. Moreover, PMMA film could serve as an antireflection layer that reduces the reflectance from 40% to lower than 20%. The power conversion efficiency of a PMMA–Gr–Si film solar cell reached 5.09%, which far exceeds the efficiency of a Gr–Si solar cell with the same thickness of Si film of 10.6 μm. More importantly, the PMMA film worked as a packaging material to improve the device stability. The PMMA–Gr–Si solar cell could keep 93% of the original efficiency after bending 60 times. The simple, low-cost and flexible photovoltaic device shows promising prospects in potential applications for portable and wearable electronic products.

---

1. Introduction

Photovoltaic technology is considered a promising green approach for easing the energy crisis. Until now, silicon has captured a dominant market share,^1,2 because of natural abundance, high conversion efficiency, and mature process technology. However, the conventional p–n junction Si solar cells with the thickness of 200–300 μm were rigid and unbendable, which was difficult to meet some practical environmental demands, such as human body, textiles, hot-air balloon, and so on. Hence, the flexible photovoltaic panels have attracted increasing interests. These flexible solar cells are lighter, thinner, foldable, and portable, so could be used as excellent conformal solar generators for applications in spacecraft, new energy vehicles, and wireless electronic devices.^3,4 Especially, wearable solar cells could be integrated into clothing, knapsack, and tent as power supply for cell phone, mp4, ipad, and other portable electronics,^5,6 which is more attractive.

Different types of flexible solar cells have been extensively researched, including amorphous-Si, organic, hybrid, dye-sensitized, copper indium gallium selenide (CIGS) photovoltaic devices.^7–12 Noticeably, an efficient transparent conductive layer was used as surface electrode for the photo-generated carrier transport and collection. So the transparent electrode is one of the most significant parts in flexible solar cells. However, the ordinary transparent conductive materials, such as metal mesh, indium tin oxide (ITO), and ZnO:Al film, could not meet the comprehensive requirements of light weight, high transparency, non-toxicity, well flexibility, and foldability.^13–15

Graphene, a two-dimension crystal film with single atomic layer, was considered as an ideal candidate of next-generation transparent electrode, due to high carrier mobility, high optical transmittance, low sheet resistance, and excellent mechanical property.^16,17 After graphene film was transferred onto semiconductor surface, this Schottky junction was a simple and effective cell structure to separate photo-generated carriers for photovoltaic process. In 2010, Li et al. reported few-layers graphene–Si (Gr–Si) solar cell with efficiency up to 1.5%, and then the power conversion efficiency (PCE) of silicon-pillar-array solar cell was enhanced from 2.9% to 4.35% through HNO₃ doping by Feng et al.^18,19 Later, the PCE of 8.6% was attained on the basis of a pristine efficiency of 1.9% by employing bis-(trifluoromethanesulfonyl)amide (TFSA) doping graphene.²⁰ In 2013, Shi et al. improved the pristine efficiency of 3.78% to 14.5% by coating TiO₂ as antireflection and HNO₃ doping.²¹ However, the aforementioned Gr–Si solar cells were fabricated from bulk silicon wafers, and were still rigid and inflexible. On the other hand, the reported pristine efficiency of the Gr–Si solar cell was still low below 4%.^18–23 Although the improvement of the PCE could be achieved by chemical doping, the doping efficiencies were easy to degenerate seriously in the air condition, due to the instability of chemical doping.

Here, we developed a flexible Gr–Si solar cell with high reliability and stability. Firstly, ultrathin crystal Si film even with the thickness of about 10 μm, which shows excellent flexibility and bendability, could be fabricated by a simple alkaline solution anisotropic etching method.^24,25 Then monolayer graphene was transferred onto Si film by polymethylmethacrylate (PMMA) mediated approach, to form Gr–Si Schottky junction. In our case, the PMMA film remained. We found that PMMA film could not only increase the absorption of incident light, but also worked as packaging material to improve the device stability. The PCE of PMMA–Gr–Si solar cell reached 5.09%, which far exceeds the efficiency of Gr–Si solar cell with same thickness of Si film. Meanwhile, the PMMA–Gr–Si solar cell could keep 93% of the original efficiency after 60 times bends. Our results demonstrate a flexible, simple-structure photovoltaic device for the potential applications of portable and wearable electronics.

2. Experimental

2.1 Ultrathin Si film

The Si film was etched from polished (100) Si wafers (n-type, 400 μm, 0.05–0.2 Ω cm⁻¹) in KOH solution with concentration of 50 wt% at 90 °C. In the KOH etching process, the rough thickness of the Si film could be simply judged by surface color illuminated under white light.

2.2 Graphene preparation

Monolayer graphene was grown on copper foil by the traditional chemical vapor deposition (CVD) method.²⁶ 3 wt% poly(methyl-methacrylate) (PMMA), which was dissolved in ethyl lactate, was spin-coated onto the graphene at 6000 rpm for 40 s. Then the sample baked at 100 °C for 10 min. The PMMA–Gr–Cu was tailored into the size of about 1 cm. The Cu foil was removed in 0.5 M ferric nitrate solution for 10 h by floating the PMMA–Gr–Cu. The remaining PMMA–Gr film was rinsed in deionized water, H₂O/HCl/H₂O₂, deionized water respectively. Finally, the PMMA–Gr film was transferred onto targeting substrate.

2.3 The cell fabrication

At first, the Si film should go through RCA (Radio Corporation of America) cleaning procedure to avoid the metal ion contaminations. After the cleaning, the Si film was immersed in a diluted 2% HF solution for 30 s to remove the oxide layer. The large-size Si film was cut into a size of about 1–1.5 cm. Then liquid-state Ga–In alloys was carefully brushed onto the one side of Si film as back electrode, and the surface of a thin copper tape-polyethylene terephthalate (PET) substrate. So, the Si film could stick to the flexible Cu-PET substrate because of the adhesion of the Ga–In layer between Si film and Cu. Afterwards, the Si film pulled the graphene out of water. After dried naturally in the air, the sample was put on a hot plate for well contact between Si and PMMA–Gr. The Ag paste was served as front electrode. The square region, which was enclosed by the Ag paste, was defined as effective area (0.075 cm²) which was measured by optical microscope. Meanwhile, the area outside the Ag electrode was enclosed by black shading tape.

2.4 Characterizations

The precise thickness of the Si film was characterized by high-resolution scanning electron microscope (SEM; JSM-7800F). Surface morphology and roughness of the Si film and PMMA layer was measured by atomic force microscope (AFM). The quality of graphene was characterized by Raman spectroscopy (inVia Reflex) with 532 nm laser, and the sheet resistances were measured by four-point probe. The transmittance and reflectance of samples were obtained by UV-visible spectrophotometer (Lambda 35) with integrating sphere. The performance of solar cells was measured by using Keithley 2400 source meter and a solar simulator under AM 1.5G illumination.

3. Results and discussion

In our Si etching process, the etching rate maintain about 80 μm per hour. The thickness of Si film could be controlled by etching time. When the thickness was reduced to less than 15 μm, Si film exhibited excellent flexibility, as shown in Fig. 1a. In Fig 1b, the ultrathin Si film could be wrapped around finger with the thickness of 10 μm, which shows excellent flexibility. The thickness of silicon could be estimated by the color under the illumination of white light. The ultrathin Si film in Fig. 1c shows dull red, which corresponds to the thickness of around 10 μm, because of high transmittance of Si film on infrared spectrum. On the other hand, the SEM image of cross section in the inset of Fig. 1e determines that the thickness is 10.6 μm. For this ultrathin Si film, it could be tailored with scissors like a piece of paper.

Fig. 1 Optical image of Si film with the thickness of (a) about 15 μm and (b) 10 μm, and the Si film could be wrapped around a finger; (c) optical image of the 10 μm-thick Si film under the illumination of white light; (d) an AFM image and (e) EDS spectrum of Si film; the inset of (e) is a SEM image of cross section of the 10.6 μm Si film.

The surface morphology of Si film could be characterized by AFM, and be shown in Fig. 1d. The average roughness is about 6.73 nm. Furthermore, the EDS spectrum in Fig. 1e shows high peak from Si and no peaks from other metal atoms, which means that the Si etching method in high concentration aqueous KOH solution could not bring the contamination of potassium ion on Si surface.

Fig. 2a shows the Raman spectroscopy of graphene on a SiO₂/Si substrate. The 2D peak at about 2700 cm⁻¹ has a half peak width of ∼38 cm⁻¹, and the intensity ratio of 2D to G peak (I_2D/I_G) is 2.9, implying the graphene film with single layer.²⁷ The D peak is so weak that the intensity ratio of D to G peak (I_D/I_G) is less than 0.1, which demonstrates that the graphene has low defect density. High quality of graphene provides important foundation for high-performance Gr–Si solar cells. Meanwhile, the PMMA–Gr layer could keep high transmittance of in the visible range, which is important for ensuring the light to get to the Si wafer.

Fig. 2 (a) Raman spectroscopy of the graphene used in solar cells. Inset is the transmittance of the PMMA–graphene; (b) SEM image of the surface morphology of transferred graphene onto SiO₂ after removing PMMA by acetone; (c) the reflectance of the Si (black), flexible Si film (red), graphene–Si (Gr–Si) (blue) and PMMA–graphene–Si (PMMA–Gr–Si) (green); (d) AFM measurement for the thickness and roughness of PMMA film.

In general, the conventional Gr–Si solar cells are usually treated with acetone to dissolve PMMA layer. However, we found that the PMMA film could not only protect the structural integrity of graphene from physical damage, but also could avoid the instability of cell's performance due to chemical adsorption. Fig. 2b clearly illustrates that the graphene after suffering PMMA dissolution process could bring a lot of cracks, wrinkles, and PMMA residues, which would result in the quality-degradation of graphene. After removing PMMA, the sheet resistance of graphene increased from ∼400 to ∼1200 Ω sq⁻¹, the obvious loss of the quality is disadvantageous to further application. Furthermore, the reflectance was investigated in Fig. 2c. For bare Si, the reflectance is about 40% in the visible spectrum range, which means a lot of loss for light absorption. The reflectance of Si film is fundamentally the same as the base Si, which is a result of well surface roughness of ultrathin Si film showed in Fig 1d. The reflectance of the Gr–Si is similar to the bare Si. However, the reflectance of the PMMA–Gr–Si is lower than 15% in visible light range, which attributed to PMMA layer that served as antireflection coating to effectively improve the light absorption for higher efficiency.

The AFM measurement of the PMMA layer was shown in Fig 2d. The surface roughness is about 0.6 nm, which is negligible compared with the visible wavelength (400–900 nm), thus the PMMA layer could be recognized as planar antireflection. The thickness of the PMMA is about 60–70 nm with the refractive index of the PMMA (n_PMMA ≈ 1.6) is between Si (n_Si ≈ 4) and air (n_air ≈ 1), thus the PMMA layer could serve as antireflection layer to reduce the light reflectance from Si surface.

The PMMA–Gr–Si film solar cell could be described as a typical sandwich structure, which consists of a PMMA protective layer, ultrathin Si film and monolayer graphene between them, as shown in Fig. 3a. In the cell, the graphene does not only operate as an effective transparent electrode for charge transportation and collection, but also forms Schottky junction with n-Si, then photo-generated electron–hole pairs in Si film would be separated by built-in electric field.²⁸ The electrons would be collected by the Ga–In cathode, while the holes are pulled to the graphene.

Fig. 3 (a) Schematic illustration of the photovoltaic device based on flexible PMMA–Gr–Si sandwich structure; (b) J–V characteristics of the PMMA–Gr–Si film and Gr–Si film solar cells under illumination and (c) dark J–V characteristics with the inset showing the ideality factor (n); (d) the external quantum efficiency (EQE) of the cells.

The current density–voltage (J–V) characteristics of flexible PMMA–Gr–Si and Gr–Si solar cells with the same Si film thickness of 10.6 μm under AM 1.5G illumination were investigated in Fig. 3b. The values of open circuit voltage (V_OC), short circuit current density (J_SC) and fill factor (FF) of the PMMA–G–Si film solar cell are respectively 0.443 V, 19.04 mA cm⁻² and 60.4%, while the corresponding values of the Gr–Si film solar cell are 0.416 V, 12.4 mA cm⁻² and 25.2%. Noticeably, the PCE of the PMMA–Gr–Si film solar cell reaches 5.09%, nearly 4 times higher than that of G–Si film solar cell. The dramatic increase of the PCE (from 1.30% to 5.09%) can be attributed to the improvement of FF and J_SC, which are increased by about 150% and 240%, respectively.

Fig. 3c shows the J–V characteristics of PMMA–Gr–Si and Gr–Si film solar cells in the dark. It's noticeable that the PMMA–Gr–Si film solar cell has a larger forward current, demonstrating more efficient carrier transport via the graphene. According to the thermal emission theory, the non-ideal forward J–V characteristics of the solar cells can be expressed as:

where, V is the voltage across the diode, k_B is the Boltzmann constant, J_S is the saturation current density.²⁹ The diode ideality factor (n) is extracted from the dark J–V curves.¹⁸ In our case, n was determined by the interfacial condition. For Gr–Si film solar cell, the value of n is 2.9, while the value has been optimized to 1.7 for PMMA–Gr–Si film solar cell. The lower n value means less interface charge recombination and better spatial homogeneity of the Schottky barrier.^30,31 However, for the Gr–Si film solar cell, the damage from PMMA removing process would severely reduce the quality of Gr–Si Schottky junction. Hence, lower n value could account for the increase of FF and J_SC for the PMMA–Gr–Si film solar cell. Noticeably, the antireflection function of the PMMA layer is another important factor for the increase of J_SC. PMMA–Gr–Si has lower reflectance of below 15% in the visible region of 500–900 nm, which would ensure the higher light absorption for Si film. Fig. 3d is the external quantum efficiency (EQE) spectra, reflecting the capacity of utilizing the incident photons. The EQE of the PMMA–Gr–Si film solar cell is almost 60% in the range from 400 to 700 nm, which implying higher light absorption capability for sufficient electron–hole pair generation, while the value for Gr–Si film solar cell is only almost 40% in the same range.

In order to investigate the reliability of the flexible hetero-structural solar cells, the bending test was made to check the degeneration of device performance. Fig. 4a is the image of Si film with transferring PMMA–Gr layer, which exhibits good flexibility. As shown in Fig. 4b, a new PMMA–Gr–Si solar cell with rectangle active area of 0.096 cm² was assembled on flexible polyethylene terephthalate (PET) substrate for bending test. The variation of the J–V curves between 60 times bends is tiny as shown in Fig. 4c. Fig. 4d and e show the dependence of cell's performance on the bending times. The original V_OC, J_SC and FF are 0.44 V, 18.53 mA cm⁻² and 55.6% respectively, while the corresponding parameters gently decrease to 0.431 V, 18.25 mA cm⁻² and 53.9% after 60 times bends. The PCE of the cell slightly decreases from 4.53% to 4.24%, reduced by about 7%. The results clearly indicate the well physical reliability of the cell in practical application.

Fig. 4 Photographs of (a) the flexible Si film coated by PMMA–Gr layer and (b) the flexible PMMA–Gr–Si film solar cell assembled on polyethylene terephthalate (PET) substrate; (c) light J–V curves under various bend times; (d) and (e) plots of V_OC, J_SC, FF and PCE as functions of bend times.

Furthermore, we analyze the stability of the PMMA–Gr–Si film solar cells by storing the cell in air for 20 days. Fig. 5a exhibits that the J_SC, V_OC and FF slightly dropped from 19.04 to 18.51 mA cm⁻², from 0.443 to 0.439 V, and from 60.4% to 57.8%, resulting in the mild degeneration of PCE from 5.09% to 4.70%. Noticeably, the J_SC was nearly invariable, indicating that the PMMA–Gr layer could keep durably stable in the air condition. The degeneration of the cell might attribute to ohmic losses from the increase of the series resistance (R_S), which could be extracted from the slope of the d(V)/d(ln J) versus J plot from the dark J–V curves.³² Fig. 5b shows that the R_S increased from 4.0 Ω cm² to 4.8 Ω cm² after 20 days, which partly resulted from the change of the sheet resistance (R_SH) of graphene film. Fig. 5c shows that the averaged R_SH of graphene film increased from 408 to 426 ohm sq⁻¹ after 30 days in the air. The R_SH variation (ΔR_SH/R_SH) maintained lower than 5%, signifying insignificant influence to the stability of the cell's performance. Another reason for the increase of R_S might be the slow oxidation of the Si interface. The oxide layer might enhance interfacial resistance for carries transport.^33,34 Fig. 5d compares the stability of the PMMA–Gr–Si film solar cell with previously reported chemically doped Gr–Si solar cell.³⁵ The PCE of the Gr–Si solar cell degraded by nearly 50% just after 8 days storage in the air, while the PCE of PMMA–Gr–Si film solar cell only slightly decreased from 5.09% to 4.70% after 20 days, and PMMA-coated solar cell could retain the 92% of original PCE. Hence, PMMA film could work as transparent packaged materials, which improve the stability of the flexible graphene-based solar cell.

Fig. 5 (a) Photovoltaic and (b) dark J–V characteristics of the PMMA–Gr–Si film solar cell in origin and after 20 days; the inset of (b) shows the comparison of the series resistance; (c) the change of the R_SH as the increase of the storage days; the inset of (c) shows the variation rate, compared with the original R_SH; (d) comparison of the PMMA–Gr–Si film solar cells with Gr–Si solar cells previously reported in ref. 35.

4. Conclusion

In summary, we reported a flexible solar cell based on graphene-ultrathin silicon film heterojunction with high reliability and stability. Compared with the conventional Gr–Si solar cells, the photovoltaic device based on Si film was thinner, lighter, and bendable. Moreover, we found that PMMA film could not only effectively reduce the reflectance of sunlight, but also worked as transparent packaging material for high stability of performance. The V_OC, J_SC, FF of PMMA–Gr–Si solar cell were 0.443 V, 19.04 mA cm⁻², and 60.4% respectively, with PCE of 5.09%, which far exceeds the efficiency of Gr–Si solar cell with same thickness of Si film. Our solar cell exhibits excellent flexible and reliable, and could provide clean energy for next-generation wearable and portable electronic products.

Author contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grant No. NSFC 11404329, NSFC 11374359) and Natural Science Foundation of Chongqing (No. CSTC2014jcyjjq50004, CSTC2012jjjq90001, CSTC 2012ggC50003, CSTC 2012ggC50001), and the Project-sponsored by SRF for ROCS, SEM.

Notes and references

1. B. Parida, S. Iniyan and R. Goic, Renewable Sustainable Energy Rev., 2011, 15, 1625–1636.
2. V. V. Tyagi, N. A. A. Rahim, N. A. Rahim and J. A. L. Selvaraj, Renewable Sustainable Energy Rev., 2013, 20, 443–461.
3. D. Zou, D. Wang, Z. Chu, Z. Lv and X. Fan, Coord. Chem. Rev., 2010, 254, 1169–1178.
4. M. He, F. Qiu and Z. Q. Lin, J. Mater. Chem., 2011, 21, 17039–17048.
5. M. B. Schubert and J. H. Werner, Mater. Today, 2006, 9, 42–50.
6. M. J. Yun, S. I. Cha, S. H. Seo and D. Y. Lee, Sci. Rep., 2014, 4, 6.
7. Y. Yao, E. Brueckner, L. F. Li and R. Nuzzo, Energy Environ. Sci., 2013, 6, 3071–3079.
8. D. Shir, J. Yoon, D. Chanda, J. H. Ryu and J. A. Rogers, Nano Lett., 2010, 10, 3041–3046.
9. Z. Z. Wang, C. F. Zhang, D. Z. Chen, S. Tang, J. C. Zhang, Y. Wang, G. Q. Han, S. R. Xu and Y. Hao, IEEE Photonics J., 2015, 7, 109–109.
10. H. C. Weerasinghe, F. Z. Huang and Y. B. Cheng, Nano Energy, 2013, 2, 174–189.
11. P. Reinhard, A. Chirila, P. Blosch, F. Pianezzi, S. Nishiwaki, S. Buecheler and A. N. Tiwari, IEEE Photovoltaic Spec. Conf., 34th, 2013, 3, 572–580.
12. L. Mao, Q. Chen, Y. W. Li, Y. Li, J. H. Cai, W. M. Su, S. Bai, Y. Z. Jin, C. Q. Ma, Z. Cui and L. W. Chen, Nano Energy, 2014, 10, 259–267.
13. M. Sibinski, K. Znajdek, S. Walczak, M. Sloma, M. Gorski and A. Cenian, Mater. Sci. Eng., B, 2012, 177, 1292–1298.
14. H. Bi, F. Q. Huang, J. Liang, X. M. Xie and M. H. Jiang, Adv. Mater., 2011, 23, 3202–3206.
15. Y. Ye and L. Dai, J. Mater. Chem., 2012, 22, 24224–24229.
16. A. K. Geim and K. S. Novoselov, Nat. Mater., 2007, 6, 183–191.
17. A. K. Geim, Science, 2009, 324, 1530–1534.
18. X. M. Li, H. W. Zhu, K. L. Wang, A. Y. Cao, J. Q. Wei, C. Y. Li, Y. Jia, Z. Li, X. Li and D. H. Wu, Adv. Mater., 2010, 22, 2743–2748.
19. T. Feng, D. Xie, Y. Lin, Y. Zang, T. Ren, R. Song, H. Zhao, H. Tian, X. Li, H. Zhu and L. Liu, Appl. Phys. Lett., 2011, 99, 233505.
20. X. Miao, S. Tongay, M. K. Petterson, K. Berke, A. G. Rinzler, B. R. Appleton and A. F. Hebard, Nano Lett., 2012, 12, 2745–2750.
21. E. Shi, H. Li, L. Yang, L. Zhang, Z. Li, P. Li, Y. Shang, S. Wu, X. Li, J. Wei, K. Wang, H. Zhu, D. Wu, Y. Fang and A. Cao, Nano Lett., 2013, 13, 1776–1781.
22. X. Zhang, C. Xie, J. Jie, X. Zhang, Y. Wu and W. Zhang, J. Mater. Chem. A, 2013, 1, 6593–6601.
23. L. Yang, X. Yu, M. Xu, H. Chen and D. Yang, J. Mater. Chem. A, 2014, 2, 16877–16883.
24. S. Wang, B. D. Weil, Y. B. Li, K. X. Z. Wang, E. Garnett, S. H. Fan and Y. Cui, Nano Lett., 2013, 13, 4393–4398.
25. H. Seidel, L. Csepregi, A. Heuberger and H. Baumgartel, J. Electrochem. Soc., 1990, 137, 3612–3626.
26. X. S. Li, W. W. Cai, J. H. An, S. Kim, J. Nah, D. X. Yang, R. Piner, A. Velamakanni, I. Jung, E. Tutuc, S. K. Banerjee, L. Colombo and R. S. Ruoff, Science, 2009, 324, 1312–1314.
27. Z. Yan, J. Lin, Z. W. Peng, Z. Z. Sun, Y. Zhu, L. Li, C. S. Xiang, E. L. Samuel, C. Kittrell and J. M. Tour, ACS Nano, 2012, 6, 9110–9117.
28. X. M. Li, D. Xie, H. Park, T. H. Zeng, K. L. Wang, J. Q. Wei, M. L. Zhong, D. H. Wu, J. Kong and H. W. Zhu, Adv. Energy Mater., 2013, 3, 1029–1034.
29. S. M. Sze and K. K. Ng, Physics of semiconductor devices, John Wiley & Sons, 2006.
30. S. G. Cheng, J. Phys.: Condens. Matter, 2010, 22, 8.
31. X. C. Miao, S. Tongay, M. K. Petterson, K. Berke, A. G. Rinzler, B. R. Appleton and A. F. Hebard, Nano Lett., 2012, 12, 2745–2750.
32. S. K. Cheung and N. W. Cheung, Appl. Phys. Lett., 1986, 49, 85–87.
33. N. Tugluoglu, S. Karadeniz and S. Altindal, Appl. Surf. Sci., 2005, 239, 481–489.
34. K. H. Cui, A. S. Anisimov, T. Chiba, S. Fujii, H. Kataura, A. G. Nasibulin, S. Chiashi, E. I. Kauppinen and S. Maruyama, J. Mater. Chem. A, 2014, 2, 11311–11318.
35. T. X. Cui, R. T. Lv, Z. H. Huang, S. X. Chen, Z. X. Zhang, X. Gan, Y. Jia, X. M. Li, K. L. Wang, D. H. Wu and F. Y. Kang, J. Mater. Chem. A, 2013, 1, 5736–5740.

---