Ag₈SnS₆: a new IR solar absorber material with a near optimal bandgap

Abstract

We report the synthesis and photovoltaic properties of a new ternary solar absorber – Ag₈SnS₆ nanocrystals prepared by successive ionic layer adsorption reaction (SILAR) technique. The synthesized Ag₈SnS₆ nanocrystals have a bandgap E_g of 1.24–1.41 eV as revealed from UV-Vis and external quantum efficiency (EQE) measurements. Its photovoltaic properties were characterized by assembling a liquid-junction Ag₈SnS₆ sensitized solar cell for the first time. The best cell yielded a J_sc of 9.29 mA cm⁻², a V_oc of 0.23 V, an FF of 31.3% and a power conversion efficiency (PCE) of 0.64% under 100% incident light illumination using polysulfide electrolyte and Au counter electrode. The efficiency improved to 1.43% at a reduced light intensity of 10% sun. When the polysulfide was replaced by a cobalt electrolyte with a lower redox level, the V_oc increased to 0.54 V and PCE increased to 2.29% under 0.1 sun, a respectable efficiency for a new solar material. The EQE spectrum covers the spectral range of 300–1000 nm with a maximum EQE of 77% at λ = 600 nm. The near optimal E_g and the respectable photovoltaic performance suggest that Ag₈SnS₆ nanocrystals have potential to be an efficient IR solar absorber.

---

1. Introduction

Semiconductor nanocrystals are now receiving much research attention in the field of photovoltaics. This is due to the unique properties of semiconductor NC materials such as band gap tenability^1,2 and high absorption coefficient.³ Moreover, multiple exciton generation effect and hot electron injection in these solar absorbers could improve the solar cell efficiency beyond the Shockley–Queisser limit.^4–9 Semiconductor nanocrystals materials can be prepared by solution-based processing, which has the advantages of ease of fabrication and low-cost over Si based solar cell processing.

A requirement for a good solar absorber material is that the energy gap E_g should be near 1.4 eV in order to produce a maximal output power.¹⁰ To date, the most widely studied solar absorber materials have been the binary metal sulfides and selenides such as CdS, CdSe, PbS, PbSe, Sb₂S₃ and Ag₂S etc.^11–16 For a binary semiconductor, its E_g is a fixed value. Only a small number of binary semiconductor satisfies the E_g = 1.4 eV requirement. This limits the number of binary semiconductors suitable for solar absorbers. An advantage of ternary semiconductors is that E_g can be tuned by varying the ratios of the constituent elements, leading to formation of a broad range of new compounds that could be potential solar absorber materials. Solar cells based on ternary metal sulfides have been relatively less explored due to the difficulty in the material synthesis (the widely studied Cu–In–Se system being an exception).

The family of I–III/IV–VI (Cu- and Ag-based) ternary metal sulfides, such as AgInS₂,¹⁷ AgBiS₂,¹⁸ AgSbS₂,¹⁹ Ag₃SbS₃,²⁰ CuInS₂,¹⁶ Cu₂SnS₃ (ref. 21 and 22) and Cu₄SnS₄,^23,24 have been investigated for their applications in photovoltaics and photocatalytics. Among them, Ag₈SnS₆ owns several important properties such as an ideal E_g of 1.3–1.5 eV, which is near the optimal E_g of 1.4 eV for a solar cell, and high absorption coefficients of α ∼ 10⁴ cm⁻¹ in the visible range.²⁵ Moreover, the three elements contained in Ag₈SnS₆ – Ag, Sn and S – are nontoxic and environmentally friendly. Ag₈SnS₆ nanocrystals have been synthesized for various purposes^26–30 such as counter electrodes in dye-sensitized solar cells,³¹ photocatalytic dye degradation,³² photoelectrochemical salt-water splitting³³ and thermoelectrics.³⁴ Ag₈SnS₆ solar cells have recently been reported with an efficiency of 0.25%.³⁵ Here, we report the simple solution-based preparation of Ag₈SnS₆ nanocrystals and investigation of their photovoltaic properties. Ag₈SnS₆ nanocrystals were directly grown on the surfaces of mesoporous TiO₂ nanoparticles using the successive ionic layer adsorption and reaction (SILAR) technique. Liquid-junction quantum dot-sensitized solar cells based on Ag₈SnS₆ nanocrystals are demonstrated for the first time. The material crystal structure, morphology and optical property have been investigated. The dependence of photovoltaic performance on SILAR conditions, sun intensity and type of electrolyte have also been studied.

2. Experimental

2.1. Preparation of TiO₂ electrodes

Fluorine-doped tin oxide glass (FTO, Pilkington, sheet resistance ∼ 7 Ω □⁻¹) was cleaned with acetone, methanol, and deionized water successively in an ultrasonic bath. It was then coated with three layers of TiO₂ film with different particle sizes and film thicknesses. First, a 90 nm thick compact TiO₂ film was made by spin-coating (2000 rpm, 1 min) a 0.2 M titanium tetraisopropoxide solution (TTIP) onto an FTO substrate, followed by annealing at 190 °C for 5 min. Next, a mesoporous TiO₂ (mp-TiO₂) layer (Dyesol 30NR-D, particle size 30 nm, ∼10 μm thick) was coated on top of the compact layer by the doctor blade technique and heated at 125 °C for 5 min. Finally, a TiO₂ scattering layer (Dyesol WER2-O, particle size 300 nm, ∼5 μm thick) was coated on top of the mp-TiO₂ layer and heated at 500 °C for 15 min.

2.2. SILAR growth of Ag₈SnS₆ nanocrystals

The formation of Ag–Sn–S nanocrystals was made by a two-stage SILAR process. In the first stage, Ag–S nanoparticles were grown on the mp-TiO₂ electrode. In the second stage, Sn–S nanoparticles were grown on top of the Ag–S nanoparticles. Post-annealing transformed the Ag–S/Sn–S double-layered structure into the Ag–Sn–S phase. An Ag–S SILAR cycle was performed by dipping the TiO₂ electrode in the solution of AgNO₃ (0.1 M) for 30 s accompanied with rinsing and drying the electrode. The step resulted in the adsorption of Ag ions on the pores of mp-TiO₂. The electrode was then dipped in a Na₂S (0.1 M) solution. This process is called one SILAR cycle and creates the Ag₂S nuclei. Repetition of this process is necessary to increase the particle size of Ag₂S nuclei. In the second-stage SILAR cycle for Sn–S, the same procedure was followed except changing the AgNO₃ cation precursor solution with SnCl₂ solution to make Sn–S. To achieve high performance, the numbers of Ag₂S and Sn–S SILAR cycles need to be optimized. The best ratio for Ag₂S and Sn–S SILAR cycles were found to be 3 : 2. For example, a sample with 12 Ag₂S cycles has 8 Sn–S cycles. To simplify discussion, a sample with Ag₂S (12 cycles)/Sn–S (8 cycles) will be referred to as sample with SILAR(12) herein. Finally, the Ag₂S/Sn–S deposited mp-TiO₂ electrode was annealed at 400 °C for 10 min under flowing N₂ gas, which resulted in the formation of ternary phase Ag₈SnS₆ nanoparticles.

2.3. Fabrication of solar cells

The solar cells were fabricated by a sandwich-type assembly of the Ag₈SnS₆ nanocrystal-immobilized TiO₂ photoelectrode and counter electrode (CE) using a 190 μm-thick parafilm thermoplastic spacer and sealant. Gold (Au) was used as the CE. A polysulfide electrolyte containing 0.5 M Na₂S, 2 M S, 0.2 M KCl and 0.5 M NaOH in a methanol/DI water (volume ratio 7 : 3) solution was used as a redox mediator. The cobalt electrolyte consisted of 0.2 M Co²⁺ (Co[PyPz]₃[PF₆]₂ salt-Dyesol), 0.05 M Co³⁺ (Co[PyPz]₃[PF₆]₂-Dyesol), 0.2 M LiClO₄, and 0.2 M t-butyl pyridine with acetonitrile solution, where Py is pyridine and Pz is pyrazole. The electrolyte was injected separately through a predrilled hole on the counter electrode into the cell. The Au CE film, ∼70 nm in thickness, was deposited onto an FTO substrate by sputtering evaporation.

2.4. Material characterization and photovoltaic measurements

The crystal structure and phase purity of the prepared material were studied using a high-resolution X-ray diffractometer (XRD, Bruker D8 SSS). The particle size and morphology were determined by a transmission electron microscope (TEM, Joel JEM-2010). Optical properties were studied using a Hitachi U-2800A UV-Vis spectrophotometer. The photovoltaic performance was studied by measuring photocurrent–voltage (I–V) curves using a Keithley 2400 source meter under 100 mW cm⁻² light illumination from an Oriel 150 W Xe lamp with an Oriel bandpass filter simulating the AM 1.5 solar spectrum. External quantum efficiency (EQE) was measured using the monochromatic light generated from an Acton monochromator with a 250 W tungsten halogen lamp. The active area, defined by a metal mask, was 3 mm × 3 mm.

3. Results and discussion

3.1. Morphology and crystal structure characterization

The structural property of the prepared Ag–Sn–S ternary phase was investigated by studying its X-ray diffraction (XRD) pattern after annealing at 400 °C under nitrogen atmosphere. The XRD pattern shown in Fig. 1 indicates the formation of orthorhombic Ag₈SnS₆ structure that is in agreement with the reference JCPDS no. 00-038-0434. The diffraction planes given at (311), (120), (411), (022), (122), (510), (313), (322), (123), (031), (611), (131), (603) and (424) are characteristic of the orthorhombic Ag₈SnS₆ structure. The (022) peak shows the strongest peak intensity compared with other peaks signifying the preferential orientation or major growth of the crystal along (022) plane. The other peaks due to the TiO₂ and FTO substrates were also observed. The calculated lattice constants for orthorhombic Ag₈SnS₆ are a = 15.29, b = 7.57 and c = 10.71 Å, which are in good agreement with the lattice constants of the data in the JCPDS database. For comparison, the major peaks associated with the binary Ag₂S and SnS₂ compounds are displayed in the bottom panel. It is clear the Ag₈SnS₆ XRD pattern does not contain the two starting binary phases, indicating successful formation of the Ag₈SnS₆ single phase. The deposited Ag–S/Sn–S double-layered particles are mostly amorphous before heat treatment (see Fig. S1, ESI†). Only a weak Ag₂S peak at 37° corresponding to the (200) plane can be observed. The average particle size of Ag₈SnS₆ nanocrystals, calculated by the Scherrer's formula³⁶ given in eqn (1) using the strongest (022) peak, is about 17.4 nm.where D is the average particle size, λ is the wavelength of the incident X-rays, β is the value of the full width at half maximum, K is a numerical constant and θ is the Bragg angle.

Fig. 1 XRD spectra of Ag₈SnS₆ nanocrystals along with the reference JCPDS 00-038-0434, reference patterns of SnS₂ and reference pattern of Ag₂S.

Fig. 2a displays a TEM image of bare TiO₂ nanoparticles. The rectangular shaped TiO₂ particles have round corners and an average length of 30 nm. Fig. 2b displays a TEM image of Ag₈SnS₆ nanocrystals grown on the surface of TiO₂ nanoparticles. The Ag₈SnS₆ nanocrystals, marked by red arrows, are randomly distributed all over the pores of mesoporous TiO₂ with no visible aggregation. Fig. 2c shows the size distribution of Ag₈SnS₆ nanocrystals prepared with eight SILAR cycles. The particle size of Ag₈SnS₆ nanocrystals is the range of ∼10–20 nm with a distribution peak near 15 nm. This size is in agreement with the value (17 nm) calculated using Scherrer's formula.

Fig. 2 TEM images of (a) 30 nm bare TiO₂ nanoparticles, (b) Ag₈SnS₆ nanocrystals deposited on TiO₂ nanoparticles and (c) size distribution of Ag₈SnS₆ nanocrystals.

3.2. Optical property

Fig. 3a shows the UV-Visible transmission spectra T(λ) of Ag₈SnS₆ nanocrystals with different numbers of SILAR cycles n. The corresponding optical absorbance A(λ), calculated from the optical transmission value using the relation , is displayed in Fig. 3b. The transmission T(λ) decreases with increasing n, indicating increasing light absorption by the increased amount of material with increasing SILAR cycles. The absorbance A(λ) increases with increasing n, which is, again, the result of increasing material. A notable feature of the transmission spectra is that the small transmission (T(λ) ≤ 5%, n = 12 sample) for wavelengths < 700 nm. This implies the Ag₈SnS₆ nanocrystals can absorb nearly all the photon energy with wavelength lower than 700 nm (the visible and near IR regions of the solar spectrum). The large light absorption ability is an advantageous property for a solar material.

Fig. 3 Optical spectra (a) transmission, (b) absorbance and (c) Tauc plots (αhν)² versus hν.

Fig. 3c shows the Tauc plot, (αhν)² versus hν, of the Ag₈SnS₆ nanocrystals where h is the Planck constant and ν is the frequency. The optical band gap E_g can be estimated by finding the x-intercept of an extrapolated Tauc plot. E_g decreases with increasing SILAR cycles n as: 1.76, 1.54, 1.41 eV for SILAR cycles of 6, 9, 12, respectively. The decrease in E_g with increasing n is attributed partly to the quantum size effect – a large SILAR cycle produced larger particles and, hence, a lower E_g. The E_g observed here is in agreement with the literature values for Ag₈SnS₆ nanocrystals prepared by other synthesis methods (1.31–1.45 eV).^31,33,35 Moreover, The E_g of 1.41 eV (n = 12 sample) is equal to the optimal E_g (1.4 eV) for a solar absorber, which is favorable property for a solar material.

3.3. Photovoltaic performance

The photovoltaic performance of a SILAR-prepared QDSSC is strongly dependent on the number of SILAR cycles n. An insufficient or excess amount of semiconductor absorber material leads to low efficiencies. Fig. 4 displays the J–V curves for Ag₈SnS₆ QDSSCs with various numbers of SILAR cycles n. The CE was Au and the electrolyte was polysulfide. Table 1 lists the photovoltaic parameters – short-circuit current density (J_sc), fill factor (FF), open-circuit voltage (V_oc) and power conversion efficiency (PCE). The efficiency increased with SILAR cycles n, reaching a maximal PCE of 0.68% at n = 10 (sample no. 3), then decreased again for a larger SILAR cycle of n = 11 (sample no. 4). The optimal sample (no. 3) has the photovoltaic parameters of J_sc = 9.01 mA cm⁻², open-circuit voltage V_oc = 0.26 V and FF = 28.9%. These are typical results for QDSSCs prepared by SILAR. Initially, the amount of material deposited on the mp-TiO₂ electrode was low, leading to insufficient solar light harvesting and low efficiencies. The deposited material increased with the SILAR cycle, leading to improved efficiencies. This result had been observed in CdS and other semiconductor solar cells as reported previously.^37,38 After reaching the optimal SILAR cycle of 10, a further increase in SILAR cycle led to over-saturated material among the porous spaces within the mp-TiO₂ matrix, which hindered the flow of liquid electrolyte and led to a reduced efficiency.

Fig. 4 J–V curves of Ag₈SnS₆ QDSSCs with various numbers of SILAR cycles n.

Table 1 Photovoltaic performance of Ag₈SnS₆ quantum dot-sensitized solar cells with various numbers of SILAR cycles. Electrolyte: polysulfide. Light intensity: 1 sun. Counter electrode: Au

Sample no.	SILAR cycle	Jsc (mA cm^−2)	V (V)	FF (%)	PCE (%)
1	6	7.29	0.19	34.07	0.47
2	9	7.77	0.20	32.45	0.50
3	10	9.01	0.26	28.90	0.68
4	11	8.08	0.23	30.25	0.56
5	12	7.62	0.24	29.91	0.55

---

The photovoltaic performance of a QDSSC could be improved by measuring J–V curves under reduced light intensities. Fig. 5 displays the J–V curves of the best Ag₈SnS₆ cell under various light intensities. The cell had an Au CE and a polysulfide electrolyte. Table 2 lists the photovoltaic parameters. As the incident light intensity was reduced from 100% sun to 10% sun, the PCE increased from 0.64% to 1.43%, a significant increase of 123%. The improved PCE is due to (a) an increase in FF from 31.3% to 39.7%, an improvement of 27%; and (b) an improvement in J_sc due to the reduction in carrier recombination (explained below). For an ideal solar cell, the current density J_sc should be linearly proportional to the number of incident photon n_ph. As the light intensity was reduced from 1 to 0.1 sun, the number of incident photons n_ph was also reduced to 0.1n_ph. Hence, the linear response model predicted that J_sc at 0.1 sun should equal to 9.29 mA cm⁻² (1 sun) × 0.1 = 0.929 mA cm⁻². However, the experimental data in Table 2 shows a much higher J_sc of 2.25 mA cm⁻² under 0.1 sun. Semiconductor nanocrystals prepared by SILAR inherently contain a large number of surface defects, which act as recombination sites for photocarriers. The mechanism of recombination can be revealed by analyzing the dependence of J_sc on light intensity I₀. Analysis of the data in Table 2 yielded a sublinear relation: J_sc ∝ I₀^α where α = 0.67. Nelson has explained the sublinear J_sc–I₀ relation in terms of the multiple trapping model for carrier recombination.³⁹ A solar cell with multiple trapping would experience reduced carrier recombination at low light intensities and yield improved performance. The most important effect of low light intensities is the increase in electron lifetime τ_n under low light. Miyashita et al. and A. C. Fisher et al. had observed in dye-sensitized solar cells that τ_n increased by ∼3–10 times as the light intensity was decreased from 100 to 10% sun.^40,41 The significantly improved performance under low light intensities indicates that carrier recombination is the limiting process in the liquid-junction Ag₈SnS₆ sensitized solar cell at high incident light intensities. The competition between recombination and extraction of free charges determines FF, J_sc and inturn the PCE of the cell. At low light intensities, the charge carrier recombination process is reduced, which improves the charge extraction process and the performance of the solar cell.⁴²

Fig. 5 J–V curves of Ag₈SnS₆ QDSSCs under various light intensities. Electrolyte: polysulfide.

Table 2 Photovoltaic performance of Ag₈SnS₆ quantum dot-sensitized solar cells under various sun intensities

Sun intensity	Jsc (mA cm^−2)	Voc (V)	FF (%)	η (%)
100%	9.29	0.23	31.3	0.64
50%	5.85	0.22	35.4	0.92
25%	3.42	0.19	36.6	0.96
10%	2.25	0.16	39.7	1.43

---

Reducing light intensities also has an effect on V_oc. As seen in Table 2, V_oc decreased monotonically with light intensity. As the light intensity decreased from 100% to 10% sun, V_oc decreased from 0.23 to 0.16 V. This can be explained as follows. The theoretical maximal V_oc of a QDSSC is the difference between the quasi Fermi level E_F of TiO₂ and the redox potential E_redox of the electrolyte, i.e.,

V_oc = E_F − E_redox. (2)

The Fermi level E_F depends on the electron density n_CB in the conduction band (CB) of TiO₂ according to:⁴³

E_F = k_BT ln(n_CB)

A reduced light intensity generates a reduced number of photoelectrons n_CB, leading to a lower E_F and V_oc as shown in Table 2.

A disadvantage of the Ag₈SnS₆ QDSSC is that the V_oc had a low value of 0.24 V, which lowered the attainable efficiency. One way to improve V_oc is to use a different electrolyte. As eqn (2) indicates, a lower redox potential E_redox will produce a higher V_oc. Here the polysulfide electrolyte in the Ag₈SnS₆ QDSSC was replaced by the cobalt electrolyte. Cobalt electrolytes have a redox level of 0.86 V relative to NHE (normal hydrogen electrode) whereas the polysulfide electrolyte has a redox level of −0.77 V.⁴⁴ The difference is 1.57 V. The cobalt electrolyte produced a much larger V_oc of 0.54 V (Table 3) under 1 sun, as compared to 0.24 V of the polysulfide electrolyte. The PCE of the cobalt-electrolyte Ag₈SnS₆ QDSSC is 0.53% under 1 sun. However, at the reduced light intensity of 0.1 sun, the PCE greatly increased to 2.29% (Fig. 6), which is 60% higher than that (1.43%) of the polysulfide cell.

Table 3 Photovoltaic performance of Ag₈SnS₆ quantum dot-sensitized solar cells using a cobalt electrolyte

Light intensity	Jsc (mA cm^−2)	Voc (V)	FF (%)	PCE (%)
1 sun	3.20	0.54	30.61	0.528
0.1 sun	1.18	0.48	40.60	2.29

---

Fig. 6 J–V curves of an Ag₈SnS₆ QDSSC employing cobalt electrolyte under 1 and 0.1 sun.

A known disadvantage of cobalt electrolytes is its low diffusion coefficient and slow kinetics for hole transfer in electrolytes, resulting a low J_sc. So the cobalt electrolyte works better under low light intensities when the small number of photoelectrons is small. A cobalt electrolyte will produce a low J_sc and a high V_oc. The large gains in V_oc and FF compensate loss in J_sc and increase the PCE to 2.29% under 0.1 sun.

Fig. 7 displays the EQE spectrum for the best Ag₈SnS₆ QDSSCs in the 300–1000 nm wavelength range, where a maximum EQE value of 77% at λ = 600 nm was achieved. The cell started to produce a current at the onset wavelength of 1000 nm. The E_g of Ag₈SnS₆ nanocrystals, calculated from the onset wavelength of the EQE spectrum using the relation: , is 1.24 eV, which is close to the optical E_g value of 1.41 eV shown in Fig. 4. Therefore, we conclude that the Ag₈SnS₆ nanocrystal has an E_g value of 1.24–1.41 eV. This E_g equals to the optimal E_g (1.4 eV) for a solar absorber, showing its potential as a solar absorber material. The area under the EQE curve represents the total integrated current density J_ph generated from the solar cell. J_ph can be calculated from the EQE curve using the following equation:

where Φ(λ) is the incident photon flux and e is the elementary charge. The integrated EQE resulted in a maximum J_ph of 25.5 mA cm⁻² (shown in the right axis of Fig. 7). The EQE measurement is basically equivalent to a low-light I–V measurement because the single-wavelength light dispersed from a monochromator has a relatively low power intensity. Compared to Table 2, the integrated J_ph of 25.5 mA cm⁻² is consistent with the J_sc of 2.25 mA cm⁻² under 0.1 sun (i.e. 25.5 × 0.1 sun = 2.55 mA cm⁻²).

Fig. 7 EQE spectrum.

4. Conclusion

We demonstrated liquid-junction Ag₈SnS₆ QDSSCs prepared by SILAR on a mp-TiO₂ electrode. The orthorhombic Ag₈SnS₆ nanocrystals have an E_g of 1.24–1.41 eV and an average size of 15 nm. The cell employing polysulfide electrolyte exhibited a V_oc of 0.23 V, and a PCE of 1.43% at 0.1 sun. In contrast, the cell employing cobalt electrolyte yielded a higher PCE of 2.29% at 0. 1 sun. The near optimal E_g and the broad absorption band suggest that Ag₈SnS₆ nanocrystal could be a promising candidate material for solar cells.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors are grateful to the financial support from the Ministry of Science and Technology of the Republic of China under grant No. MOST 106-2112-M-005-006.

References

1. P. V. Kamat, J. Phys. Chem. C, 2008, 112, 18737–18753.
2. A. Kongkanand, K. Tvrdy, K. Takechi, M. Kuno and P. V. Kamat, J. Am. Chem. Soc., 2008, 130, 4007–4015.
3. I. Moreels, K. Lambert, D. D. Muynck, F. Vanhaecke, D. Poelman, J. C. Martins, G. Allan and Z. Hens, Chem. Mater., 2007, 19, 3023–3030.
4. R. D. Schaller and V. I. Klimov, Phys. Rev. Lett., 2004, 92, 186601–186604.
5. R. J. Ellingson, M. C. Beard, J. C. Johnson, P. Yu, O. I. Micic, A. J. Nozik, A. Shabaev and A. L. Efros, Nano Lett., 2005, 5, 865–871.
6. V. I. Klimov, J. Phys. Chem. B, 2006, 110, 16827–16845.
7. J. B. Sambur, T. Novet and B. A. Parkinson, Science, 2010, 330, 63–66.
8. O. E. Semonin, J. M. Luther, S. Choi, H.-Y. Chen, J. Gao, A. J. Nozik and M. C. Beard, Science, 2011, 334, 1530–1533.
9. W. A. Tisdale, K. J. Williams, B. A. Timp, D. J. Norris, E. S. Aydil and X.-Y. Zhu, Science, 2010, 328, 1543–1547.
10. T. Zdanowicz, T. Rodziewicz and M. M. Zabkowska-Waclawek, Sol. Energy Mater. Sol. Cells, 2005, 87, 757–769.
11. G. Larramona, C. Chone, A. Jacob, D. Sakakura, B. Delatouche, D. Pere, X. Cieren, M. Nagino and R. Bayon, Chem. Mater., 2006, 18, 1688–1696.
12. N. Guijarro, T. L. Villarreal, I. Mora-Seró, J. Bisquert and R. J Gómez, J. Phys. Chem. C, 2009, 113, 4208–4214.
13. V. G. Pedro, C. Sima, G. Marzari, P. P. Boix, S. Giménez, Q. Shen, T. Dittrich and I. M. Seró, Phys. Chem. Chem. Phys., 2013, 15, 13835–13843.
14. Y. Itzhaik, O. Niitsoo, M. Page and G. Hodes, J. Phys. Chem. C, 2009, 113, 4254–4256.
15. A. Tubtimtae, K. L. Wu, H. Y. Tung, M. W. Lee and G. J. Wang, Electrochem. Commun., 2010, 12, 1158–1160.
16. S. L. Castro, S. G. Bailey, R. P. Raffaelle, K. K. Banger and A. F. Hepp, J. Phys. Chem. B, 2004, 108, 12429–12435.
17. T. Torimoto, T. Adachi, K. I. Okazaki, M. Sakuraoka, T. Shibayama, B. Ohtani, A. Kudo and S. Kuwabata, J. Am. Chem. Soc., 2007, 129, 12388–12389.
18. P. C. Huang, W. C. Yang and M. W. Lee, J. Phys. Chem. C, 2013, 117, 18308–18314.
19. W. C. Yang and M. W. Lee, J. Electrochem. Soc., 2014, 161, H92–H96.
20. C. L. Chou, N. Suriyawonga, B. Aragawa, J. B Shib and M. W. Lee, J. Electrochem. Soc., 2016, 163, H445–H449.
21. Z. Jia, Q. Chen, J. Chen, T. Wang, Z. Li and X. Dou, RSC Adv., 2015, 5, 28885–28891.
22. H. D. Shelke, A. C. Lokhande, J. H. Kim and C. D. Lokhande, J. Colloid Interface Sci., 2017, 506, 144–153.
23. Q. Chen, X. Dou, Z. Li, Y. Ni, J. Chen, F. Zhou, Y. Yamaguchi and S. Zhuang, Optik, 2014, 125, 3217–3220.
24. Y. Chen, L. Ma, Y. Yin, X. Qian, G. Zhou, X. Gu, W. Liu, X. Wu and F. Zhang, J. Alloys Compd., 2016, 672, 204–211.
25. C. L. Lu, Z. Lin, Y. W. Zhang, S. Y. Liu and M. Yang, Chin. Phys. B, 2015, 24, 017501.
26. C. An, K. Tang, G. Shen, C. Wang, L. Huang and Y. Qian, Mater. Res. Bull., 2003, 38(5), 823–830.
27. B. Li, Y. Xie, J. Huang, H. Su and Y. Qian, J. Solid State Chem., 2000, 149(2), 338–340.
28. Q. Li, Y. Ding, X. Liu and Y. Qian, Mater. Res. Bull., 2001, 36, 2649–2656.
29. B. H. Shambharkar and A. P. Chowdhury, RSC Adv., 2016, 6, 10513–10519.
30. M. Deng, S. Shen, Y. Zhang, H. Xu and Q. Wang, New J. Chem., 2014, 38, 77–83.
31. Q. He, S. Huang, C. Wang, Q. Qiao, N. Liang, M. Xu, W. Chen, J. Zai and X. Qian, ChemSusChem, 2015, 8, 817–820.
32. W. Q. Hu, Y. F. Shi and L. M. Wu, Cryst. Growth Des., 2012, 12, 3458–3464.
33. K. W. Cheng, W. T. Tsai and Y. H. Wu, J. Power Sources, 2016, 317, 81–88.
34. T. Ghrib, A. L. A. Otaibi, M. A. Almessiere, I. B. Assaker and R. Chtourou, Chin. Phys. Lett., 2015, 32, 127402.
35. L. Zhu, Y. Xu, H. Zheng, G. Liu, X. Xu, X. Pan and S. Dai, Sci. China Mater., 2018, 61, 1549–1556.
36. A. L. Patterson, Phys. Rev., 1939, 56, 978–982.
37. M. Ganjian, M. Kolahdouz, A. Aletayeb, M. Norouzi, P. Ebrahimi, S. Pourjafari and M. S. S. Mousavi, Vacuum, 2017, 146, 548–553.
38. M. Marandi, E. Rahmani and F. A. Farahani, J. Electron. Mater., 2017, 46, 6769–6783.
39. J. Nelson, S. A. Haque, D. R. Klug and J. R. Durrant, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 205321–205329.
40. M. Miyashita, K. Sunahara, T. Nishikawa, Y. Uemura, N. Koumura, K. Hara, A. Mori, T. Abe, E. Suzuki and S. Mori, J. Am. Chem. Soc., 2008, 130, 17874–17881.
41. A. C. Fisher, L. M. Peter, E. A. Ponomarev, A. B. Walker and K. G. U. Wijayantha, J. Phys. Chem. B, 2000, 104, 949–958.
42. D. Bartesaghi, I. C. Pérez, J. Kniepert, S. Roland, M. Turbiez, D. Neher and L. J. A. Koster, Nat. Commun., 2015, 6, 7083.
43. J. Nelson, The Physics of Solar Cells, Imperial College Press, London, 2003, p. 59,  DOI:10.1142/p276.
44. F. Hao, P. Dong, Q. Luo, J. Li, J. Lou and H. Lin, Energy Environ. Sci., 2013, 6, 2003–2019.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8ra08734b

---