A simple and effective crystal growth inhibitor for high performance solid-state dye-sensitized solar cells

Abstract

Three simple imidazolium-type ionic liquids with benzene cores (abbreviated as [TMImB][Br] and [TMImB][TFSI]) have been successfully designed, synthesized and characterized. The chemical structure, thermal and electrochemical properties have also been investigated in detail. Moreover, it is revealed that adding 20 wt% [TMImB][TFSI] could effectively suppress the crystal growth of the typical ionic conductor 1-ethyl-3-methylimidazolium iodide (EMII). Meanwhile, [TMImB][TFSI] can facilitate the EMII-based solid-state electrolyte to form a much smoother surface morphology than EMII alone, which can improve interfacial electrochemical contact among EMII and porous TiO₂ films. Therefore, the resultant solid-state DSSC with [TMImB][TFSI] exhibits a higher efficiency of 5.66% than the DSSC without crystal growth inhibitors, and displays better long-term stability than the DSSC with conventional EMIBF₄. These preliminary results provide us with more opportunities to explore new crystal growth inhibitors with special chemical structures for high performance ssDSSCs.

---

1. Introduction

Due to their high efficiencies, simple assembly technology and low cost, dye-sensitized solar cells (DSSCs) have attracted considerable academic and industrial interest.¹ Recently, impressive photoelectric conversion efficiencies of 13% have been successfully obtained by combining porphyrin sensitizers with cobalt(II/III)-based redox mediators in organic liquid electrolytes.^2,3 However, the evaporation of volatile components and leakage of organic liquid electrolytes are still unavoidable, greatly restraining the long-term performance of DSSCs.⁴ To ultimately overcome these drawbacks, solid-state electrolytes including p-type semiconductors (CuI, CuSCN, etc.),^5,6 organic hole conductors^7,8 and ionic polymers,^9,10 especially small molecule ionic conductors^11–15 have been proposed to replace the liquid electrolytes, demonstrating promising potential in the future application of solid-state dye-sensitized solar cells (ssDSSCs).

Unfortunately, easy crystallization of these solid-state electrolytes (e.g. EMII^12–14) can form large voids and discontinuous films, which will lead to incomplete filling of electrolytes into porous TiO₂ films. Therefore, the major shortcoming destroys dye regeneration and increases internal resistance, reducing the photocurrent generation. However, crystal growth inhibitors, including 1-ethyl-3-methylimidazolium tetrafluoroborate (EMIBF₄),^13,16 1-propyl-3-methylimidazolium tetrafluoroborate (PMIBF₄)^11,17 and 1-propyl-3-methylimidazolium iodine (PMII),^14,15 have often been used to restrain the crystal growth of ionic conductors and improve the device performance. For example, Zhao et al. added PMIBF₄ to an organic ionic crystal MA-II to prepare solid-state electrolytes for ssDSSCs. Because of effective crystal growth inhibition, better filling of solid-state electrolytes and smoother surface without voids improved the device efficiency from 0.20% to 0.40%.¹¹ Wang and coworkers designed several solid-state ester and hydroxyethyl functional ionic conductors,^13,14 and demonstrated excellent long-term stability and high efficiencies over 6% by employing EMIBF₄ and PMII as crystal growth inhibitors. Recently, in our previous report,¹⁸ a kind of high viscous light-scattering crystal growth inhibitors have been developed to fabricate high performance ssDSSCs, and display better long-term stability than that of conventional low viscous EMIBF₄-based devices. Therefore, exploring new crystal growth inhibitors offer more opportunities to demonstrate their basic research interest and extensively promising applications for ssDSSCs.

Here, a new kind of ionic liquids with benzene core are designed, synthesized and characterized using a range of measurements. Acting as an effective crystal growth inhibitor, the ionic liquid can effectively restrain the crystal growth behaviors of classical EMII with high ionic conductivity, and form much smoother surface morphology for better electrochemical contact. The resultant DSSCs are further investigated by the photocurrent density–voltage curves and electrochemical impedance spectra for the photovoltaic photovoltaic and interfacial charge transfer information, respectively. These attempts offers us a feasible method to explore other novel crystal growth inhibitors with special chemical structure for high performance solid-state DSSCs.

2. Experimental

Materials

1-Methylimidazole, 1,3,5-tris(bromomethyl)benzene ([TMB]-[Br]) and iodine (I₂) were purchased from Alfa Aesar and used as received. Lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) was kindly provided by Rhodia and used as received. TiCl₄, H₂PtCl₆ and organic dye Z907 were purchased from Aldrich, and all the chemical reagents were used without further treatment. 1-Ethyl-3-methylimidazolium iodide (EMII) and 1-ethyl-3-methylimidazolium tetrafluoroborate (EMIBF₄) were synthesized according to the previous report.¹⁹ Fluorine-doped tin oxide (FTO) glass electrodes (8 Ω Sq⁻¹), and slurries containing 20 nm-sized mesoporous and 200 nm-diameter light-scattering TiO₂ colloidal were purchased from Dalian Hepat Chroma Solar Tech. Co., Ltd (China).

Synthesis of 1,3,5-tris-(1-methylimidazol-3-ethyl)-benzene bromide [TMImB][Br] and 1,3,5-tris-(1-methylimidazol-3-ethyl)-benzene bis(trifluoromethanesulfonyl)imide [TMImB][TFSI]

[TMImB][Br] was synthesized by the reaction of [TMB][Br] (6.03 g, 10 mmol) with a excess of 1-methylimidazole (2.87 g, 35 mmol) in ethanol, and the solution was reacted at 60 °C under a nitrogen atmosphere for 48 h. After evaporation of the solvent, the crude product was washed with ethyl acetate and diethyl ether three times, respectively, and dried under vacuum for 24 h to provide 9 g [TMImB][Br] (yield: 78.4%, light white power). ¹H NMR (400 MHz, d₆-DMSO): 9.46 (s, 3H), 7.87 (t, 3H), 7.77 (t, 3H), 7.59 (s, 3H), 5.49 (s, 6H), 3.91 (s, 9H). Formula mass: 602.97. Microanalysis calcd: C, 41.83%; H, 4.48%; N, 13.94%. Found: C, 41.72%; H, 4.38%; N, 13.76%.

Anion-exchange of [TMImB][Br] with superfluous LiTFSI aqueous solution was utilized to produce [TMImB][TFSI].²⁰ In a typical experimental procedure, 3.02 g (5 mmol) [BMImB]Br was added to 5.17 g (18 mmol) LiTFSI solution. The mixture was stirred at room temperature for 24 h. The resulting orange liquid of [TMImB][TFSI] was washed with cold water several times and then dried at 80 °C for 24 h in a vacuum oven (yield: 85.8%). ¹H NMR (400 MHz, d₆-DMSO): 9.15 (s, 3H), 7.76 (t, 3H), 7.70 (t, 3H), 7.41 (s, 3H), 5.42 (s, 6H), 3.88 (s, 9H). Formula mass: 1131.72. Microanalysis calcd: C, 22.29%; H, 2.39%; N, 11.14%; S, 17.00%. Found: C, 22.22%; H, 2.17%; N, 11.09%; S, 16.72%.

Device fabrication, characterization and measurements

The fabrication of DSSCs¹⁵ is represented with the configuration of FTO glass/TiO₂/dye/solid-state electrolyte/Pt/FTO glass in the ESI.† The detailed characterization and photovoltaic measurements are also described in ESI.†

3. Results and discussion

Scheme 1 shows the facile synthetic procedure and chemical structures of synthesized ionic liquids. It can be found that simple two steps will fabricate the resultant product. The purity and chemical structure are confirmed by ¹H NMR (Fig. S1 and S2, ESI†). In addition, Fig. 1 also shows typical photographs of [TMImB][Br], [TMImB][TFSI], EMII and EMIBF₄, respectively. Moreover, solid-state electrolytes A–C have been fabricated by ionic conductor EMII and 20 wt% I₂, combining with synthesized [TMImB][TFSI] and referenced EMIBF₄ as crystal growth inhibitors, respectively. It could be seen that all the three electrolyte samples are solid-state at room temperature.

Scheme 1 General synthetic procedures for the preparation of ionic liquids [TMImB][Br] and [TMImB][TFSI].

Fig. 1 Typical photographs of [TMImB][Br], [TMImB][TFSI], EMII, EMIBF₄ and electrolytes A–C, respectively.

As shown in Fig. 2, the chemical structures of [TMImB][Br] and [TMImB][TFSI] are further confirmed by FTIR spectra. The bands attributed to the imidazolium cations at 3358, 3634, 1561 and 1571 cm⁻¹ are clearly identified.²¹ The adsorption peaks at 3159, 3079 and 2964 cm⁻¹ are attributed to the C–H stretching vibration mode of the imidazole ring and the alkyl chains. Moreover, the C=N stretching and asymmetric vibration belonging to the imidazole ring is also observed at about 1628, 1620 and 1158 cm⁻¹. Compared with [TMImB][Br], new absorption peaks at 1354, 1194 and 1054 cm⁻¹ can be assigned as the TFSI anion of [TMImB][TFSI], indicating the successful anion exchange.²²

Fig. 2 FTIR spectra of (a) [TMImB][Br] and (b) [TMImB][TFSI].

The influence of the counteranions on the thermal stability of synthesized ionic liquids is also shown in Fig. 3a. It can be found that the initial decomposition temperatures of [TMImB][Br] and [TMImB][TFSI] are 245 and 409 °C, respectively. However, [TMImB][TFSI] exhibits much better thermal stability than [TMImB][Br], probably because of more stable anion's structure and weaker destructive interaction to benzene core of TFSI⁻ than Br⁻.²³ Meanwhile, the values of carbon residue of [TMImB][Br] and [TMImB][TFSI] are 21.7 wt% and 18.2 wt%, which is potentially applied in porous carbon materials.²⁴ These results demonstrate that synthesized [TMImB][TFSI] indeed offers a high thermal stability, far beyond the range of interest for solid-state electrolytes.²⁵ The DSC curves are also represented in Fig. 3b. The solid-state [TMImB][Br] has a relatively high melting point ∼138.5 °C. However, besides the melting point around −33 °C, [TMImB][TFSI] displays a liquid state without any endothermic or exothermic peak from 0 to 180 °C. These results are also in accordance with the photographs in Fig. 1. Therefore, [TMImB][TFSI] can be chosen as a new kind of crystal growth inhibitor for solid-state electrolytes. As shown in Fig. 3c, bare electrolyte A with EMII/10 wt% I₂, exhibits a solid-state form with a melting point about 74 °C, and is prone to be a crystal at room temperature. Compared with that of electrolyte A, the melting point (MP) of electrolyte B with controlled EMIBF₄ decreases down to 55 °C. Moreover, electrolyte B with 20 wt% addition of [TMImB][TFSI] exhibits a decreased MP at 49 °C because of more effective crystal growth inhibitor. It is well known that ionic crystallization of EMII based solid-state electrolytes usually makes them difficult into the porous TiO₂ films, forms voids and results in obvious reduce of device efficiencies of ssDSSCs.^11,13–18 Fortunately, crystal growth inhibitors can effectively restrain the crystallization of EMII, and facilitate them to penetrate across the porous TiO₂ films for better device performance.

Fig. 3 (a) TGA curves of synthesized [TMImB][Br] and [TMImB][TFSI]; (b, c) DSC curves of [TMImB][Br], [TMImB][TFSI] and electrolytes A–C, respectively.

As shown in Fig. 4, the surface morphologies of EMII-based electrolytes coated on TiO₂ photoanodes is further investigated the effect of synthesized [TMImB][TFSI] and referenced EMIBF₄. Because of easy crystallization, electrolyte A with EMII/20 wt% I₂ exhibits large grains and cracks (Fig. 4a). However, the rough surface can be effectively improved by adding liquid-state two kinds of crystal growth inhibitors (Fig. 4b and c). Moreover, compared with EMIBF₄ in Fig. 4b, the unique structure of [TMImB][TFSI] with benzene cores probably results in slightly smoother surface morphologies shown in Fig. 4c. In addition, as shown in Fig. S3,† the cross-section SEM images further demonstrate synthesized [TMImB][TFSI] can effectively prevent EMII from recrystallization, and elemental analysis indicates that electrolyte C is well-dispersed across the whole device (Fig. S3d†). These results can provide effective information on surface morphologies to optimize the following photo-current characteristics of solid-state DSSCs.

Fig. 4 SEM images of (a) electrolyte A with EMII/10 wt% I₂, (b) electrolyte B with EMII/EMIBF₄/10 wt% I₂, and (c) electrolyte C with EMII/[TMImB][TFSI]/10 wt% I₂, respectively.

To better evaluate the influence of [TMImB][TFSI] as a crystal growth inhibitor on the device performance, electrolytes A–C are further employed to fabricate solid-state DSSCs (devices A–C). The photocurrent density–voltage (J–V) curves are shown in Fig. 5a, and the photovoltaic parameters of the ssDSSCs are also summarized in Table 1. For preliminary device A, open-circuit voltage (V_oc), short-circuit photocurrent density (J_sc), fill factor (FF) are 0.673 V, 10.64 mA cm⁻², and 0.672, respectively, yielding a satisfactory power conversion efficiency (PCE) of 4.81%, which is much higher than those efficiencies ∼4% in previous reports.^16,17 Due to poor interfacial contact of EMII's recrystallization, many formed interspaces and voids in the porous TiO₂ films will lead to an obvious decrease of efficiencies and long-term stability. However, conventional EMIBF₄ can improve the interfacial wetting properties among EMII and porous TiO₂ films, and overcome the crystal growth behavior of EMII, resulting in obvious performance enhancement. Thus, device B shows a V_oc of 0.678 V, a J_sc value of 12.66 mA cm⁻² and a FF value of 0.679, yielding a PCE of 5.83%. However, using the new crystal growth inhibitor [TMImB][TFSI], device C produces a V_oc of 0.673 V, a J_sc value of 11.69 mA cm⁻² and a FF of 0.719, corresponding to a PCE of 5.66%. The PCE is slightly lower than that of device B, probably due to lower conductivity of electrolyte C (6.24 × 10⁻⁴ S cm⁻¹) than electrolyte B (6.78 × 10⁻⁴ S cm⁻¹). Fig. 5b also represents the IPCE as a function of wavelength. The broad feature appears covering the visible spectrum range from 400 to 650 nm, corresponding to the typical absorption of dye Z907 on TiO₂ film. Additionally, the maximum IPCE values at 530 nm are 52.51%, 71.69% and 63.52% for devices A–C, respectively, indicating high light harvesting efficiency, fast electron injection, effective dye regeneration and charge collection for the high temperature solid-state DSSCs.²⁶

Fig. 5 (a) J–V characteristics of devices A–C measured under simulated AM 1.5 solar spectrum irradiation at 100 mW cm⁻². Cell area: 0.25 cm². (b) The IPCE vs. wavelength profiles for typical devices A–C, respectively.

Table 1 The performance of ssDSSCs with three solid-state electrolytes (devices A–C) under simulated AM 1.5 solar spectrum illumination at 100 mW cm⁻², and EIS results for devices A–C at −0.70 V in the dark. Cell area: 0.25 cm²

Device^a	Conductivity^b (10^−4 S cm^−1)	Jsc (mA cm^−2)	Voc (mV)	FF	PCE (%)	R1 (Ω)	R2 (Ω)	R3 (Ω)
a The three solid-state electrolytes for devices A–C containing EMII/10 wt% I2, EMII/20 wt% EMIBF4/10 wt% I2 and EMII/20 wt% [TMImB][TFSI]/10 wt% I2, respectively.b Corresponding conductivity of EMII-based solid-state electrolytes.
A	8.22	10.64	673	0.672	4.81	0.41	36.16	5.11
B	6.78	12.66	678	0.679	5.83	0.50	42.07	8.59
C	6.24	11.69	673	0.719	5.66	0.42	38.54	9.21

---

To investigate the interfacial charge transfer processes of devices A–C, as shown in Fig. 6, electrochemical impedance spectra (EIS) are further measured at −0.70 V in the dark, and the resistance values are also listed in Table 1. In addition, the equivalent circuit by Autolab software (the inset in Fig. 6) is also used to fit the experimental data of the devices. As shown in Fig. 6, from high to low frequency, three semicircles are observed in the typical Nyquist plots. R_s, R₁, R₂, and R₃ represent series resistance, charge-transfer resistance at the counter electrode, the resistance of dyed-TiO₂/electrolyte interface, and the Nernst diffusion in the electrolyte, respectively.²⁷ All the R₁ values for devices A–C lower than 0.5 Ω indicates that reduction of I₃⁻ to I⁻ is favourable at the Pt counter electrode/electrolyte interface. Generally, the semicircle in the intermediate frequency region is larger, the electron recombination at the TiO₂/electrolyte interface is weaker.^28,29 Compared with the R₂ value for device A (36.16 Ω), introducing crystal growth inhibitors can increase the resistance values of the TiO₂/electrolyte interface for device B (42.07 Ω) and device C (38.54 Ω), weaking the undesirable electron recombination. These results show good consistency with the previous J–V values, and hence devices B and C display better device performance than referenced device A. However, adding EMIBF₄ and [TMImB][TFSI] can slight increase the R₃ values in the EIS curves and reduce the conductivity of the electrolyte, which is undesirable and unavoidable.

Fig. 6 Nyquist plots of EIS spectra measured for devices A–C at −0.70 V in the dark. The inset is the equivalent circuit used to fit the impedance spectra.

The long-term stability of devices A–C is further investigated via an accelerating aging test at 50 °C with successive one sun visible-light soaking in Fig. 7. It can be found that, without crystal growth inhibitors, device A exhibits an obvious decrease in the efficiency (63% after 50 days) in this period. Notably, devices B–C containing crystal growth inhibitors still remain at 79% and 88% of their initial efficiency after an accelerating aging test of 50 days, respectively. However, device B with EMIBF₄ suffers from more obvious efficiency attenuation than device C with [TMImB][TFSI] during the testing period, probably because of the lower viscosity value (39 cP at 25 °C) leading to the slight leakage of EMIBF₄ at 50 °C than [TMImB][TFSI] (486 cP at 25 °C). The results demonstrate [TMImB][TFSI] acting as a new crystal growth inhibitor, could effectively suppress the crystallization of ionic conductors at high operating temperature, and therefore result in excellent long-term stability of ssDSSCs.^30–32

Fig. 7 Time-course variation of normalized PCE for devices A–C with successive one sun light soaking during the accelerated aging test at 50 °C, respectively.

4. Conclusions

In summary, a kind of simple and effective ionic liquid [TMImB][TFSI] with benzene core is designed, synthesized and characterized with high purity and yield. The thermal and electrochemical properties of [TMImB][TFSI] have been also investigated. Moreover, [TMImB][TFSI] can effectively restrain the crystal growth of ionic conductor EMII. Addition of 20 wt% [TMImB][TFSI] to EMII still remain solid-state property with a melting point ∼49 °C. Meanwhile, [TMImB][TFSI] can facilitate EMII-based solid-state electrolyte to form much smoother surface morphology than EMII alone, which can improve interfacial electrochemical contact among EMII and porous TiO₂ films. Furthermore, the resultant solid-state DSSC with [TMImB][TFSI] exhibits a higher efficiency of 5.66% than the DSSC without crystal growth inhibitors, and displays better long-term stability than the DSSC with conventional EMIBF₄. These preliminary results provide a facile method to synthesize new ionic liquids and electrolytes for high performance solid-state DSSCs for future practical applications.

Acknowledgements

This work was supported by “973 Program – the National Basic Research Program of China” Special Funds for the Chief Young Scientist (2015CB358600), the National Natural Science Foundation of China (21504061, 21422103), the Natural Science Foundation of Jiangsu Province (No. BK20140311), University Science Research Project of Jiangsu Province (No. 13KJB150033), Jiangsu Fund for Distinguished Young Scientist (BK20140010), the Open Foundation of Jiangsu Key Laboratory of Thin Films and the Priority Academic Program Development (PAPD) of Jiangsu Higher Education Institutions.

Notes and references

1. M. Grätzel, Nature, 2001, 414, 338–344.
2. A. Yella, H. W. Lee, H. N. Tsao, C. Y. Yi and A. K. Chandiran, Science, 2011, 334, 629–634.
3. S. Mathew, A. Yella, P. Gao, R. Humphry-Baker, B. F. E. Curchod, N. Ashari-Astani, I. Tavernelli, U. Rothlisberger, M. K. Nazeeruddin and M. Grätzel, Nat. Chem., 2014, 6, 242–247.
4. J. H. Wu, Z. Lan, J. M. Lin, M. L. Huang, Y. F. Huang, L. Q. Fan and G. G. Guo, Chem. Rev., 2015, 115, 2136–2173.
5. G. R. A. Kumara, A. Konno, K. Shiratshchi, J. Tsukahara and K. Tennakone, Chem. Mater., 2002, 14, 954–955.
6. V. P. S. Perera, P. K. D. D. P. Pitigala, P. V. V. Jayaweera, K. M. P. Bandaranayake and K. Tennakone, J. Phys. Chem. B, 2003, 107, 13758–13761.
7. U. Bach, D. Lupo, P. Comte, J. E. Moser, F. Weissörtel, J. Salbeck, H. Spreizer and M. Grätzel, Nature, 1998, 395, 583–585.
8. N. J. Jeon, J. Lee, J. H. Noh, M. K. Nazeeruddin, M. Grätzel and S. I. Seok, J. Am. Chem. Soc., 2013, 135, 19087–19090.
9. J. N. De Freitas, A. F. Nogueira and M. A. De Paoli, J. Mater. Chem., 2009, 19, 5279–5294.
10. Y. Zou, W. Xiang, S. Chen, S. Fang, X. Zhou, J. Zhang and Y. Lin, Chem. Commun., 2009, 4, 3895–3897.
11. Y. Zhao, J. Zhai, J. L. He, X. Chen, L. Chen, L. B. Zhang, Y. X. Tian, L. Jiang and D. B. Zhu, Chem. Mater., 2008, 20, 6022–6028.
12. D. M. Li, D. Qin, M. H. Meng, Y. H. Luo and Q. B. Meng, Energy Environ. Sci., 2009, 2, 283–291.
13. H. Wang, X. Zhang, F. Gong, G. Zhou and Z. S. Wang, Adv. Mater., 2012, 24, 121–124.
14. B. Yu, Y. Cao, J. Zhang, M. Wang, R. Li, P. Wang, M. Nazeeruddin, T. Sekiguchi and M. Grätzel, Nat. Mater., 2008, 7, 626–630.
15. H. Cao-Cen, J. Zhao, L. Qiu, D. Xu, Q. Li, X. Chen and F. Yan, J. Mater. Chem., 2012, 22, 12842–12850.
16. C. Li, M. Ye, R. Vittal and K. Ho, J. Mater. Chem., 2011, 21, 15471–15478.
17. C. Li, L. Lin, P. Chen, R. Vittal and K. Ho, J. Mater. Chem., 2010, 20, 3619–3625.
18. S. Cong, Q. Yi, Y. Wang, J. Zhao, Y. Sun and G. Zou, J. Power Sources, 2015, 280, 90–96.
19. P. Bonhôte, A. P. Dias, N. Papageorgiou, K. Kalyanasundaram and M. Grätzel, Inorg. Chem., 1996, 35, 1168–1178.
20. J. Zhao, F. Yan, L. Qiu, Y. Zhang, X. Chen and B. Sun, Chem. Commun., 2011, 47, 11516–11518.
21. J. Zhao, F. Yan, Z. Chen, H. Diao, F. Chu, S. Yu and J. Lu, J. Polym. Sci., Part A: Polym. Chem., 2009, 47, 746–753.
22. J. Zhao, X. Shen, F. Yan, L. Qiu, S. Lee and B. Sun, J. Mater. Chem., 2011, 21, 7326–7330.
23. H. Shirota, T. Mandai, H. Fukazawa and T. Kato, J. Chem. Eng. Data, 2011, 56, 2453–2459.
24. J. S. Li, X. Q. Wang, H. M. Luo, G. A. Baker and S. Dai, J. Am. Chem. Soc., 2009, 131, 4596–4597.
25. M. Grätzel, J. Photochem. Photobiol., A, 2004, 164, 3–14.
26. S. Cong, Y. Wang, Q. Yi, J. Zhao, Y. Sun, M. Shen and G. Zou, J. Mater. Chem. A, 2014, 2, 20147–20153.
27. J. Zhao, B. Sun, L. Qiu, H. Caocen, Q. Li, X. Chen and F. Yan, J. Mater. Chem., 2012, 22, 18380–18386.
28. Y. Geng, Y. Shi, L. Wang, B. Ma, R. Cao, Y. Zhu, H. Dong and Y. Qiu, Phys. Chem. Chem. Phys., 2011, 13, 2417–2421.
29. Z. Huo, D. Dai, K. Wang, F. Kong, C. Zhang, X. Pan and X. Fang, Sol. Energy Mater. Sol. Cells, 2007, 91, 1959–1965.
30. Q. Li, J. Zhao, B. Sun, B. Lin, L. Qiu, Y. Zhang, X. Chen, J. Lu and F. Yan, Adv. Mater., 2012, 24, 945–950.
31. C. Shi, L. Qiu, X. Chen, H. Zhang, L. Wang and F. Yan, ACS Appl. Mater. Interfaces, 2013, 5, 1453–1459.
32. S. Li, L. Qiu, C. Shi, X. Chen and F. Yan, Adv. Mater., 2014, 26, 1266–1271.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra09921a

‡ Authors with equal contributions.

---