Rongjing Yan,
Li Kang,
Yuxiu Sun* and
Jingbo Zhang*
Key Laboratory of Inorganic-Organic Hybrid Functional Material Chemistry, Ministry of Education, Tianjin Key Laboratory of Structure and Performance for Functional Molecules, College of Chemistry, Tianjin Normal University, Tianjin 300387, China. E-mail: hxxyzjb@mail.tjnu.edu.cn
First published on 22nd March 2018
Cu2ZnSnS4 (CZTS) thin film solar cells become an interesting research topic due to some advantages of the CZTS thin film such as having nontoxic and abundant components, a low price and excellent optoelectronic properties. In this work, a solution-based preparation method was developed to fabricate a CZTS solar cell with a superstrate structure of FTO/TiO2/CdS/CZTS/P3HT/Cu by using mixed solvent. Nanocrystalline TiO2 porous thin film was used as the bottom layer for deposition of CZTS to increase the interfacial area of CZTS. To deposit CZTS inside the porous structure leading to a good contact of CZTS with porous TiO2 thin film, the CZTS precursor particle size is successfully regulated by changing the volume ratios of N,N-dimethylformamide and ethanol. More importantly, small size CZTS precursor particles can easily enter into the porous structure of nanocrystalline thin film leading to a good interfacial contact, which allowed the effective improvement of the light-to-electric conversion efficiency for the present superstrate CZTS solar cell. This work may provide a promising way for the design of high-efficient superstrate solar cells.
Most CZTS photovoltaic devices are fabricated with a configuration of Mo/CZTS/buffer/metal oxide nanostructures/metal electrodes,4 where the p-type semiconductor absorber layer directly contacts with the molybdenum back prepared by sputtering, spray pyrolysis, electrodeposition, thermal evaporation, etc. Previous reports have shown that this configuration has some disadvantages such as deterioration of the metal oxide due to the thermal instability of the absorber layer leading to formation of the buffer interface.5 In addition to these limitations, in CZTS solar cells there are the erosion reaction between CZTS and the Mo back contact and the surface instability of CZTS during the heat treatment.6 These problems can be ameliorated by the introduction of a superstrate structure in solar cell, where metal oxide is deposited on a smooth top substrate and then annealed at high temperature to prevent direct contact of CZTS and the conducting substrate.7 In previous work, the CZTS growth on TiO2 thin films offered the highest conversion efficiency of 0.25%.8,9 Most of following works are to change the morphology of metal oxides to improve the conversion efficiency of CZTS solar cells.10 There are also some of works focusing on using non-toxic buffer material11 or replacing the counter electrode4 to achieve the desired efficiency of solar cells. The reported conversion efficiency of the superstrate structure was gradually increased from 0.84% to 3.63%.12–17 The superstrate structure is beneficial in supplying enough driving force to facilitate electron injection,18 the CZTS absorber-TiO2 conductor interface plays a crucial role in the device's efficiency, however, there is few study to discuss this issue.
In 2013, Wang et al. found the main reason for the low conversion efficiency of CZTS solar cell is that CZTS is only distributed on the surface of TiO2 nanoparticles and does not fill in the voids of the film like dye molecules.19 In the same year, Li et al. pointed out that smaller size and more evenly distributed CZTS nanoparticles are expected to achieve a higher conversion efficiency of CZTS solar cell.20 It is well-known that the performance of thin film solar cells depends greatly on the morphology of the absorber layer. Therefore, good absorber layer morphology is the key to improve the conversion efficiency of solar cell. There are several ways to obtain different morphologies. It is important to choose a solvent. Here, we mainly explore the impact of the mixed solvent on the morphology of CZTS. On the one hand, good solvents should have high solubility for the metal source and the sulfur source, and low surface tension to make the solution be easy to completely cover without streaks or voids. On the other hand, they should have proper boiling point and vapor pressure allowing the solvent to be evaporated easily, and have minimal residue leading to microstructure stability during a rapid volatilization process. In some fields, DMF is a common solvent that meets the above requirements.21 Here, in order to improve the coverage of CZTS on TiO2 nanoparticles, we tried to add ethanol in DMF solvent for preparation of CZTS precursor solution, since solvent surface tension has different effects on CZTS growth processes. This treatment can improve the TiO2 surface wettability and make better contact interface. We hope that the addition of ethanol can make the CZTS precursor solution enter into the porous TiO2 film better and improve the photoelectric conversion efficiency of superstrate structure solar cell. Therefore, we explored the effect of the mixed solvent ratio on the morphology of CZTS precursor. The size of CZTS particles significantly decreases with the increase of ethanol with the lower surface tension. The prepared TiO2/CZTS thin film was used as a photoelectrode to fabricate the superstrate CZTS thin film solar cell. The highest light-to-electric conversion efficiency for a superstrate structure CZTS solar cell was achieved to 2.15%.
![]() | ||
Fig. 1 Surface SEM images of CZTS thin films prepared from precursors in DMF (a), DMF![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
By adjusting the reactivity of the precursor and the coordination ability of the solvent, the average particle size of the CZTS nanocrystals is in the range of 10 to 60 nm in different cases as seen from Fig. 1S.† The shape of nanoparticles is spherical. The film prepared with the DMF solution consists of inhomogeneous particles. The aggregation of difference size particles leads to some defects and cracks, which will degrade the performance of solar cell. The particle size distribution was estimated from SEM surface images by counting more than 100 particles and shown in Fig. 2. When the precursor solution is mixed with a small molecule ethanol, the average particle size is decreased from 51.0 nm (DMF) to 32.6 nm for the sample made from the mixed solution with the volume ratio of 9:
1 (DMF
:
ethanol). And the average particle size was further decreased to 28.6 nm in the case of their volume ratio of 1
:
1. Due to the addition of ethanol, the distribution becomes more narrow meaning the particles become more homogeneous in size.
![]() | ||
Fig. 2 Particle size distribution of CZTS particles prepared from precursors in DMF (a), DMF![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
In order to promote crystallization of CZTS on the surface of TiO2 nanoparticles, the precursor film was further subjected to a sulfurization treatment at 550 °C for 60 min. Fig. 3 shows top-view SEM images of the typical CZTS thin films. The CZTS morphologies on nanoporous TiO2 show a sharp change compared with the film deposited directly on the flat substrate.21 It is observed from the SEM image that the film is dominated by several large grains. As seen from Fig. 3a, they are relatively dense with a majority of grains bigger than 0.5 μm, and a small part of grains less than 0.2 μm. In addition, the addition of ethanol solvent makes the film appear more uniform and smooth as shown in Fig. 3b and c. However, using ethanol as the solvent, the grain growth was not complete after the sulfurization treatment, the number of holes becomes more and the surface is not flat (Fig. 3d).
![]() | ||
Fig. 3 Surface SEM images of the sulfurized CZTS prepared from precursors in DMF (a), DMF![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
The cross-sectional SEM images of the CZTS film on TiO2 nanocrystalline thin film prepared with the mixed solvent of DMF and ethanol with a volume ratio of 1:
1 before and after sulfurization were shown in Fig. 4. Surface of the film before sulfurization is smooth and homogeneous, no significant cracks or voids were observed as shown in Fig. 4a. The interface between CZTS and nanocrystalline TiO2 is not clear meaning CZTS particles enter into porous structure of nanocrystalline thin film. The film thickness is approximately 600 nm including the TiO2 layer of about 200 nm thick and the CZTS layer of about 400 nm. After the sulfurization treatment, the grains become larger and extended through the CZTS layer, the thickness is almost unchanged as shown in Fig. 4b.
![]() | ||
Fig. 4 SEM cross section images of the films prepared from DMF![]() ![]() ![]() ![]() |
![]() | ||
Fig. 5 Cross section EDS color map of the CZTS thin film prepared from precursors in DMF![]() ![]() ![]() ![]() |
The XRD pattern of the CZTS thin film prepared with the mixed solvent of DMF and ethanol with a volume ratio of 1:
1 after annealed at 550 °C for 60 min in sulfur atmosphere was shown in Fig. 6. Sample shows three major diffraction peaks at 28.6°, 47.6° and 56.3°, which can be indexed to the (112), (220) and (312) in the kesterite structure of Cu2ZnSnS4 (JCPDS no. 26-0575), respectively. No noticeable impurities can be observed. The lattice parameters, a = b = 5.41 Å, c = 10.8 Å, calculated from the pattern, are in good agreement with the standard values for tetragonal CZTS.25
![]() | ||
Fig. 6 XRD patterns of the CZTS thin film prepared from precursors in DMF![]() ![]() ![]() ![]() |
The XRD pattern is not enough to identify the phase purity of the synthesized product because some binary (cubic ZnS) and ternary (tetragonal Cu2SnS3) chalcogenides have similar lattice parameters or pattern characteristics to CZTS. Thus, Raman spectrum on the CZTS surface was measured to detect the presence of the secondary phase. Fig. 7 shows the Raman spectrum of the CZTS thin film prepared with the mixed solvent of DMF and ethanol with a volume ratio of 1:
1. The major peaks at 338 cm−1 and 286 cm−1 are attributed to the CZTS A1 mode. The A1 phonon mode is pure anion mode, which puts to the vibration of S atoms surrounded by motionless neighboring atoms.26 Other possible binary phases such as ZnS, Sn2S3 and CuxS for the obtained CZTS thin film cannot be excluded based on the Raman spectrum.27
X-ray photoelectron spectroscopy (XPS) was used to further confirm the presence of four components (Fig. 8). The peaks at 932.4 eV and 952.2 eV on Cu 2p spectra (Fig. 8a) correspond to 2p3/2 and 2p1/2 (generated by spin orbital splitting), respectively. Two peaks are divided into 19.8 eV. No satellites are found in the higher binding direction of the peak. Thus, it can be concluded that only Cu+ is present in the sample, indicating that Cu2+ was reduced during this process. For the Zn 2p spectrum, between 2p3/2 and 2p1/2, there is a spin orbital division of 23.1 eV (Fig. 8b). The peaks of Zn 2p appear at 1022.3 eV and 1045.4 eV, being consistent with the standard distribution of Zn2+. The two lines corresponding to Sn of 3d5/2 and 3d3/2 were observed at 487.2 eV and 495.6 eV, respectively. Sn4+ was confirmed by a peak resolution of 8.4 eV (Fig. 8c). It indicates that Sn2+ in the starting material is oxidized to Sn4+. The analysis of the sulfur core state of the annealed CZTS thin film (Fig. 8d) indicates the peaks at 161.8 eV and 162.7 eV are corresponding to the S 2p3/2 and S 2p1/2.28,29
![]() | ||
Fig. 8 XPS of the CZTS thin film prepared from precursors in DMF![]() ![]() ![]() ![]() |
![]() | ||
Fig. 9 Schematic diagram of superstrate FTO/TiO2/CdS/CZTS/P3HT/Cu solar cell and the corresponding bandgap alignment of each layer. |
![]() | ||
Fig. 10 J–V curves of CZTS solar cells based on CZTS thin films prepared from precursors in different solvents. |
Unfortunately, the performance of the superstrate CZTS solar cell is poor with a short-circuit photocurrent density (Jsc) of 3.5 mA cm−2. The low photocurrent density is usually due to presence of interfacial states as recombination sites and chemical composition of CZTS. TiCl4 hydrolysis was usually used to form a dense TiO2 layer to modify these recombination sites. However, the conversion enhancement is not obvious. Alternately, a thin layer of ZnO was electrodeposited on the porous surface of nanocrystalline TiO2 thin film. One of advantages for electrodeposition is the uniform layer can be formed on a rough electrode surface. Due to porous structure of nanocrystalline TiO2 thin film, the electrodeposition of ZnO could better modify surface defects and improve conversion efficiency of the superstrate solar cell. Firstly, the thickness of the electrodepositing ZnO layer was optimized (Fig. 5S†), the photovoltaic parameters of CZTS solar cells with the ZnO layer deposited for different times were collected in Table 1. Electrodeposition of ZnO layer for 0.5 min shows the best passivation effect towards the recombination sites and gives highest conversion efficiency.
Electrodeposition time (min) | Jsc (mA cm−2) | Voc (mV) | ff (%) | η (%) |
---|---|---|---|---|
0 | 3.5 | 430 | 34.1 | 0.51 |
0.3 | 4.5 | 450 | 30.9 | 0.63 |
0.5 | 12.6 | 400 | 42.3 | 2.15 |
0.7 | 6.1 | 460 | 34.5 | 0.97 |
Further, the effect of mixed solvent with different volume ratios on the performance of superstrate CZTS solar cell was studied and J–V curves corresponding to different solar cells were shown in Fig. 10. Jsc values of solar cells with the CZTS thin films prepared in DMF solvent, mixed solvents with volume ratios of 9:
1 and 1
:
1 were 4.9 mA cm−2, 6.4 mA cm−2 and 12.6 mA cm−2, respectively. The increase in Jsc indicates that the electron–hole recombination rate in the 1
:
1 case is lower than that of other two samples. This may be due to the fact that the formed small CZTS particles enter into porous structure forming a well interfacial contact with nano-TiO2/CdS layer, which contributes to the rapid transfer of electrons. Therefore, the solar cell based on CZTS prepared by the 1
:
1 mixed solvent has the highest energy conversion efficiency (η) of 2.15%, followed by the solar cells based on the 9
:
1 solvent (0.86%) and DMF (0.68%). However, only ethanol was as solvent to prepare CZTS, there is no expected effect, the conversion efficiency is only 0.10%.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8ra01095a |
This journal is © The Royal Society of Chemistry 2018 |