A new type of sulfide-based, solid-state dye material that is sensitive to visible radiation was assessed as a potential replacement for commercial ruthenium complex dyes in a dye-sensitized solar cell (DSSC) assembly. The In2S3 crystals on the surface of the TiO2 bottom blocking layer were grown as a solid-state dye material. Scanning electron microscopy of In2S3 revealed a microsized, 3D-connected sheet-like shape, which was confirmed by X-ray diffraction to be a beta-structure. The efficiency of the dye-sensitized solar cells assembled with a layer grown with In2S3 increased with increasing In2S3 mole concentrations to 0.05 M (1.02%) but decreased at concentrations greater than 0.6~0.8%. This suggests that crystalline In2S3 acts as a dye sensitized to visible radiation, but the short-circuit current density is too low compared to the commercially available ruthenium dye. This suggests that In2S3 crystals did not grow densely but were bulk-grown with large pores, resulting in a smaller amount of In2S3 per unit area. Two IPCE curves were observed, which were assigned to TiO2 and In2S3, meaning that the TiO2 surfaces were covered completely with In2S3 crystals. The exposure of TiO2 eventually leads to a reaction with the electrolytes, resulting in lower quantum efficiency.

1. Introduction

The increasing demand for energy and the limited oil reserves have highlighted the need for the development of alternative forms of renewable energy. Solar energy is one option proposed to deal with the energy crisis because it is abundant, unlimited, and clean. Sunlight can be harvested and converted directly to electricity using photovoltaic devices [15]. Since O’Regan and Gratzel first reported dye-sensitized solar cells (DSSCs) in 1991 [6], there has been continuous research into DSSCs because of their numerous advantages over silicon solar cells, such as low cost, easy assembly, and relatively high efficiency [710]. DSSCs are one of the most promising methods for future large-scale energy production from renewable energy sources. DSSCs are composed of four major parts, a semiconducting cathode, Pt anode, iodine electrolyte, and photosensitive dye [11], which is the most important component. The photosensitizer, that is, the dye, in DSSCs should fulfill the following essential characteristics:(1)the excited state energy level of the photosensitizer should be higher than that of the conduction band edge of the semiconductor;(2)the absorption spectrum of the sensitizer should cover as much as of the visible light and even the near-infrared region as possible;(3)the dye molecule should bind to the surface of the semiconductor efficiently through certain anchoring groups;(4)the oxidized state level of the sensitizer should match the redox potential of the electrolyte;(5)the sensitizer should be photo- and electrochemically and thermally stable.

Over the last 20 years, ruthenium (II) complexes based on carboxylic pyridine ruthenium compounds, such as N3 [12], N719 [13], N907 [14], and black dye [15], have shown a clear lead in the solar cell performance. The best solar-to-electricity conversion efficiency of >10% was achieved using these ruthenium-complex photosensitizers. On the other hand, ruthenium-complex photosensitizers tend to deteriorate the lifespan of the cell, making the photosensitizers unstable under thermal stress and light soaking [16, 17].

Therefore, it is important to explore novel inorganic sensitizers composed of transition metals. Recently, as an alternative to dye sensitizers, narrow-band gap semiconducting quantum dots, such as CdS [18], CdSe [1], PbS [19], Sb2S3 [20], and CuInS2 [21], have also been tested as photosensitizers. These cells are called quantum dot-sensitized solar cells (QDSSCs). The use of semiconductor QDs as sensitizers has the following advantages in solar cell applications over conventional dyes: possible tuning of the visible response by changing the QD size [22], large intrinsic dipole moment resulting from quantum confinement and a high extinction coefficient [23], possible multiple electron-hole pairs [24], and a potential theoretical maximum efficiency of 31% [25]. Although the above advantages are quite promising, the efficiencies of QDSSCs (typically below 5%) are much lower than that of DSSCs (ca. 10%). One of the reasons for the poor performance of QDSSCs is the difficulty in assembling the QDs on a mesoporous TiO2 matrix to obtain a well-covered QD layer on the TiO2 surface. Therefore, new metal sulfide materials including QDs, which are easier to manufacture, absorb visible light easily, and have excellent electron-transfer ability, are needed.

Although many studies have assessed a range of metal sulfide materials for solar cell applications, there are no reports of a detailed study of In2S3 for solar cell applications. Therefore, In2S3 was evaluated as a photosensitizer in this study. The In2S3 is an excellent electron donor, and three different structures are as follows: yellow -In2S3 has a defect cubic structure; red β-In2S3 has a defect spinal, tetragonal structure; and -In2S3 has a layered structure. The β-form is believed to be the most stable form at room temperature. In addition, it is diamagnetic and exhibits n-type semiconductivity with an optical band gap of 2.1 eV [26]. In2S3 crystals are grown directly onto the surface of a TiO2 seed layer electrode using a chemical bath deposition (CBD) method [27], and its photosensitive effect in DSSCs is discussed.

2. Experimental

2.1. In2S3 Crystals Grown over TiO2 Seed Blocking Layers and Their Characterization

In2S3 crystals were grown on the surface of the TiO2 seed blocking layer using a general CBD method [28]. First, a sol-gel dip-coating technique was used to fabricate a TiO2 seed blocking layer onto a fluorine-doped tin oxide (FTO) conducting glass plate (Hartford, 30 ohm cm−2 and 80% transmittance in visible region). A FTO glass substrate, which covered one side with black tape, was dipped into a 0.1 M TTIP ethanol solution fixed at pH = 2.5 with acetic acid and aged for 24 h. The prepared TiO2 blocking layer film was then annealed at 450°C for 30 minutes in air to remove the additives. In the secondary step, the In2S3 crystals were grown on the TiO2 seed blocking layer using thiourea ((NH2)2C=S, 99.9%, Junsei Chemical, Japan) and indium chloride (InCl3, 99.9%, Junsei Chemical, Japan) as the S and In precursors, respectively. Indium chloride (0.001~0.5 M) was dissolved in distilled water, and thiourea (0.0015~0.75 M) was then added slowly with stirring. The solution was fixed to pH = 2.5 using acetic acid. After stirring homogeneously for 2 h, the final solution was transferred to an autoclave, which included a TiO2 seed blocking layered glass, and treated thermally at 80°C for 24 h. The resulting glass was washed with ethanol and dried at 60°C for 24 h. Here, a TiO2 paste was not used. Instead, the TiO2 seed blocking layer eventually acted as a working electrode.

2.2. Assembly of TiO2 Seed Blocking/In2S3 Photosensitizer Layered-Solar Cells

To prepare a solar cell, a working TiO2 seed blocking/In2S3 photosensitizer layered electrode was covered with a Pt-coated FTO counterelectrode, and the edges of the cell were sealed with a sealing sheet (PECHM-1, Mitsui-Dupont Polychemical) at 110°C. The used redox electrolyte consisted of 0.5 mol KI, 0.05 mol I2, and 0.5 mol 4-tert-butylpyridine with an acetonitrile solvent. Sufficient electrolyte was injected into cells using a vacuum pump.

2.3. Characterization

The prepared TiO2 (bottom)/In2S3 (top) films were examined by X-ray diffraction (XRD; MPD, PANalytical) using nickel-filtered CuK radiation (40 kV, 30 mA). The surface morphology of the film was examined by scanning electron microscopy (SEM, S-4100, Hitachi). Cyclic voltammetry (CV, BAS 100B) of the TiO2 and In2S3 powder disc was performed at room temperature at a scan rate of 100 mV s−1 in an electrolyte composed of 0.5 mol KI, 0.05 mol I2, and 0.5 mol 4-tert-butylpyridine in acetonitrile. Platinum wire was used as the working and counterelectrodes, and Ag/AgCl was used as the reference electrode. The UV-visible (Cary 500 spectrometer) spectra of TiO2 (bottom)/In2S3 (top) films were obtained using a reflectance sphere in the range, 200~800 nm.

The photovoltaic efficiency of the TiO2 seed blocking/In2S3 photosensitizer layered-solar cells was calculated from the - curves that had been determined using a Sun 2000 solar simulator (ABET technology) equipped with a xenon lamp (max. 150 W). The light intensity of the solar simulator was equivalent to one sun at AM 1.5. The instrument was calibrated using a c-Si reference cell. The spectral output of the lamp was matched in the region, 350–800 nm, using a Schott KG-5 sunlight filter to reduce the mismatch between the simulated and true solar spectrum to less than 2%. The incident light intensity and active cell area were 100 mWcm−2 and 0.40 cm2 (0.8 × 0.5 cm), respectively. The solar cell was measured at room temperature using a shading mask and the aperture area was used for the calculation. The major factors determining the efficiencies of the solar cells were the open-circuit voltage (), short-circuit current density (), and fill factor (FF).

Using the same equipment, the equilibrium conditions are important considerations in the IPCE measurements, where the spectral response is obtained under short-circuit conditions. In AC mode, the test SC was illuminated by monochromatic light with the addition of simulated AM 1.5 G solar light to represent the solar cell working conditions. The photocurrent responding to monochromatic light was measured precisely using a lock-in amplifier at a frequency that was also applied to a mechanical chopper to modulate the monochromatic light. The total photocurrent could be estimated from the IPCE spectra by integrating the incident photon flux density, (), and IPCE () over the wavelength of incident light, which is expressed as [28] where is the photocurrent, is the electron charge, and is the wavelength.

3. Results and Discussion

3.1. Physicochemical Properties of the TiO2 Seed Blocking/In2S3 Photosensitizer Layered Film

Figure 1 shows XRD patterns of the prepared TiO2 seed blocking/In2S3 photosensitizer (0.001, 0.01, 0.05, 0.1, and 0.5 M) layered films. Anatase TiO2 [29] and tetragonal tin oxide [30] induced from the FTO glass were observed in the In2S3 film sample grown using the 0.001 M-In precursor, but no peaks were assigned to β-In2S3. On the other hand, with the exception of the 0.001 M-In precursor sample, all the diffraction peaks of the samples were indexed to β-In2S3 with a Fd-3m space group, which are in good agreement with the data reported in the literature [31]. Compared to the standard card, the relative intensities of the peaks at 14.3° (111 plane), 23.4° (220), 27.4° (311), 28.5° (222), 33.4° (400), 36.2° (331), 40.9° (422), 43.7° (511), 47.8° (440), 55.8° (533), 59.3° (444), 67.0° (731), 69.6° (800), 77.1° (751), 79.1° (840), and 88.9° (844) were obvious. The (311) XRD peak is normally the strongest. The mean crystalline sizes of the samples containing 0.001, 0.01, 0.05, 0.1, and 0.5 M of the In2S3 photosensitizer, which was calculated using Scherrer’s equation, , were 48.10, 19.78, 36.21, 18.10, and 21.72 nm, respectively.

Figures 2(a) and 2(b) show a high-magnification SEM image (top view and side view) of the In2S3 crystals grown over the TiO2 seed blocking layer. The TiO2 seed blocking layer thickness was approximately 0.5 μm in all samples and the depths of the grown In2S3 crystals increased from 0.7 to 5.0 μm. The 3D-structured and connected sheet-like In2S3 particles were observed. According to the In2S3 concentration, the sheets on the top view were largely grown and the pore sizes were larger. The densities also increased with increasing In2S3 concentration to 0.05 M but decreased at concentrations >0.05 M.

Figures 3(a) and 3(b) present the UV-visible absorption spectra and Tauc’s plot of the TiO2 seed blocking/In2S3 photosensitizer (0.001, 0.01, 0.05, 0.1, and 0.5 M) layered films, respectively. This is a practical method for estimating the energy gap from an extrapolation of the absorption edge intercept on the -axis or of that at the excitonic (shoulder) peak, called . The absorption spectra had a steeper absorption edge, which is similar to the results obtained from the Tauc equation [32]. The absorption band of the TiO2 film normally appears at approximately 350 nm. According to the concentrations of the grown In2S3, the absorption bands shifted to a higher wavelength around the maximum, 450~650 nm. The band gap of semiconductor materials is closely related to the wavelength absorbed, where the band gap decreases with increasing absorption wavelength. The band gap can be calculated from the following Tauc equation: , where is the photon energy, is the absorption coefficient, is a constant relative to the material, and the exponent is a value that depends on the nature of the transition (2 for a direct allowed transition, 2/3 for a direct forbidden transition, and 1/2 for an indirect allowed transition). The band gap obtained by extrapolating the linear portion of the graph in the 0.05 M In2S3 photosensitizer layered on TiO2 seed blocking (700 nm) was approximately 1.72 eV. In contrast, the TiO2 seed blocking layered film had a band gap of 2.81 eV. Generally, the band gap of anatase TiO2 is approximately 3.02 eV. On the other hand, the absorption in the visible region was attributed to the transition from the ground state to a few defects. This excitation character of the absorption spectra indicated the excellent crystal quality of the semiconductor. A redshift was observed in the optical absorption spectra of the TiO2 hexagonal columns, which indicated that the TiO2 particles showed quantum confinement related effects. This absorption in the visible region was attributed more to the transition from the ground state to a few defect-related deep states [3335]: the value of for bulk CdS was 2.5 eV, but a wide range of values have been reported for CdS thin films depending mainly on the deposition technique. Evaporated CdS films with band gaps as low as 2.2 eV and chemically deposited CdS films with band gaps as high as 2.6 eV have been reported. These high values for CdS have been attributed to quantum confinement effects due to the small grain size of the polycrystalline films. Consequently, the band gap can be varied according to the particle size or crystal defects. In addition, the band gap indicates that its electronic properties changed due to the small particle sizes of TiO2, indicating that the energy of the lowest excited state of the sample and heat treatment depend on their size: the highest energy is reached at a smaller particle size. The band gap energies could be explained by the volume conservation law, which states that an increase in at least one lattice constant should be compensated for by a decrease in another and vice versa, as is the case in the low-energy regions. Therefore, anatase (2.81 eV) shows an almost linear decrease in the band gap with increasing volume or decreasing lattice constant. A smaller band gap is achieved by compressing the lattice constant [36, 37]. Because the semiconductor electrode plays an important role in accepting electrons and donating them to the solar cells, its light absorption is important. This suggests that In2S3 effectively absorbs a large range of visible light similar to commercial ruthenium dyes.

3.2. Photovoltaic Efficiency and Electronic Properties of the TiO2 Seed Blocking/In2S3 Photosensitizer Layered-Solar Cells

Figure 4 shows the - curve (up) and photovoltaic efficiencies (down) of the TiO2 seed blocking/In2S3 photosensitizer layered-solar cells. The TiO2 nanoparticles typically used in DSSCs have a diameter of approximately 10~30 nm, through which it is easy for the long wavelengths of light to permeate. Unfortunately, long-wavelength light is not used effectively in solar cells. In recent years, DSSCs assembled with a second layer film with a 4~5 μm thickness comprised of large nanoparticles or nanopores coated on the top of a dense TiO2 electrode exhibited improved efficiency. The added second layer consisting of TiO2, 400 nm in size, acts as a scattering material for increasing the incident light path, resulting in an array of photons that in turn improve the current density. This highlights the importance of the scattering materials in a DSSC assembly. On the other hand, a TiO2 thick film as a second layer or scattering material was not used, and only a 0.5 μm TiO2 seed blocking layer and 0.7~5 μm In2S3 photosensitizer layer were used. The 0.5 μm TiO2 seed blocking layered-solar cell showed a power conversion efficiency of only 0.20% with  V and  mA/cm2. This corresponds to the cell without a dye. On the other hand, the cases of the TiO2 seed blocking/In2S3 photosensitizer layered-solar cells showed a higher . In particular, the current density increased until a In2S3 photosensitizer content of 0.05 M. The conversion efficiency was maximized to 1.20% in the TiO2 seed blocking/0.05 M In2S3 photosensitizer layered-solar cell. This suggests that the In2S3 crystalline acts as a dye sensitized to visible radiation, even though the short-circuit current density is too small compared to the use of the commercially available ruthenium dye. This suggests that In2S3 crystals are not grown densely and are bulk-grown with rather large pores, resulting in the existence of a smaller amount of In2S3 per unit area. Moreover, the exposure of TiO2 eventually leads to a reaction with the electrolytes and a smaller quantum efficiency. Chemical bath deposition provides a high surface coverage of crystals, but a lack of capping agents leads to a broad size distribution as well as a higher density of surface defects of crystals [38]. Therefore, a photoanode with a high crystal loading, fewer surface traps, and strong crystals/TiO2 electronic coupling is essential for constructing highly efficient solar cells.

The incident photon-to-current conversion efficiencies (IPCE) in Figure 5 indicate the number of incident photons inside the cell as well as their contribution to the conversion efficiency. The commercial ruthenium dyes used in general DSSCs that respond primarily to the wavelength of visible light were measured in the region, 300~750 nm, and reacted at approximately 530 nm, showing the highest quantum number (≈60~80%) [39]. Two IPCE curves were observed, which were assigned to TiO2 and In2S3 at a maximum of 320 and 520 nm, respectively, and the curves increased with increasing In2S3 concentration. On the other hand, the quantum efficiencies were smaller than those of the commercial ruthenium dyes because the TiO2 surfaces were not covered perfectly with In2S3 crystals. Eventually, the quantum efficiency for visible radiation decreased. The solar cell assembled with a 0.05 M In2S3 dye layer showed a maximum quantum efficiency of 22%, resulting in a lower solar conversion to electron efficiency ratio.

Figure 6 shows the cyclic voltammetry measurements for the TiO2 and In2S3 pellets. In the present study, the oxidation potentials were measured by cyclic voltammetry in distilled water solutions. The pelletized samples were used as the working electrodes, Ag/AgCl was used as the reference electrode, and 0.1 M KCl was used as the supporting electrolyte. A reversible wave is needed for the absolute potentials between the reduction () and oxidation peaks (). When such reversible peaks are observed, thermodynamic information in the form of a half cell potential can also be determined [40]. Recently, some studies devised a useful equation based on cyclic voltammetry for determining the HOMO and LUMO energy levels. The ferrocene ( versus Ag/Ag+ = +0.42 eV) potential, which is used as a standard, should be measured in the electrolyte solution using the same reference electrode, with a fixed −4.8 eV energy level in the vacuum set. The HOMO or LUMO energy levels can be calculated using the following formula: HOMO (or LUMO) or . Here, is the starting point of the redox potential, which was used more often than the peak potential value. In synthesized TiO2, the Ti(IV) → Ti(0) redox reaction appears reversible, and the absolute reduction potential () was observed at −189.3 mV. Using the Ag/AgCl reference electrode, the onset potential for reduction was determined for In2S3 powder and was −410.5 mV. Therefore, the corresponding LUMO energy levels for TiO2 and In2S3 were calculated to be −4.191 and −3.970 eV, respectively. Consequently, these values could change to −0.31 and −0.53 eV depending on the upper equation. In DSSCs, the potential location of the conduction band is more important than the band gap of the semiconductor electrode because it should always be lower than the LUMO level of the dye (normally by 0.2 eV).

A model of the potential energy diagram was suggested from the CV curves and UV-visible spectra, as shown in Scheme 1. The conduction band of the TiO2 blocking layer was located at slightly (0.22 eV) lower energy levels than the LUMO level of the In2S3 dye. This suggests that electrons donated from the LUMO of the In2S3 dye are transferred easily to the conduction band of the TiO2 film with less electron loss. The transferred electrons can move easily to the external circuit through the TCO electrode.


4. Conclusions

In this study, β-type In2S3 crystals with various concentrations were grown on the surface of the TiO2 bottom blocking layer by CBD. The grown In2S3 crystals were applied to dye-sensitized solar cells as a solid dye. Micron-sized, 3D-connected sheet-like shapes were observed by SEM. The efficiency of the dye-sensitized solar cells prepared from the 0.05 M-In2S3 crystals was the highest at 1.02%. Two IPCE curves were observed, which were assigned to TiO2 and In2S3, respectively. This means that the TiO2 surfaces are not perfectly covered with In2S3 crystals. The exposure of TiO2 can lead to a reaction with the electrolyte, resulting in an increase in resistance, an interruption of current flow, and a decrease in photoelectric efficiency. These results confirmed that the performance in DSSCs relies on a complete coating method. Therefore, more reliable coating methods will be needed to improve the efficiency.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.


This work was supported by 2013 the Innopolis Daegu Project, Research and Development Special Zones, Republic of Korea, for which the authors are very grateful.