Fully stoichiometric Cu₂BaSn(S_1−xSe_x)₄ solar cells via chemical solution deposition

Kesterite (Cu₂ZnSn(S_1−xSe_x)₄, CZTSSe) has shown a promising future in photovoltaic (PV) solar cells [1–4] and photoelectrochemical (PEC) devices [5] due to Earth-abundance [6], low toxicity and a similar structure to copper–indium–gallium–selenide (CuInGaSe₂). As CuInGaSe₂ has reached commercial levels with power conversion efficiencies (PCEs) > 22% [7], kesterite materials have attracted much attention in the past few years [8]. Since the first device was reported in 2009 [9], the PCE of CZTS solar cells has been improved from 0.66% to 12.6% within five years [1]. However, the PCE has been limited to below 13% due to the large open circuit voltage (V_oc) deficit [4, 10–13], which is caused by anti-site defects and associated band tail inside the material [14]. In previous reports, the primary cause of anti-site defects is the similar ionic radius between Cu⁺ (0.91 Å) and Zn²⁺ (0.88 Å) [15].

According to the theoretical simulation, anti-site defects can be avoided by the cation being substituted with a different ionic radius [16], such as (Cu, Ag)₂ZnSnS₄ [17], Cu₂FeSnS₄ [18], Cu₂CdSnS₄ [19], Cu₂NiSnS₄ [20] and Cu₂CoSnS₄ [21]. Considering the ionic radius difference between zinc (Zn, 0.88 Å) and barium (Ba, 1.49 Å) [22], Cu₂BaSnS₄ (CBTS) is regarded as the most promising candidate of CZTS [23, 24]. Since being first reported in 2016 [8], CBTS-based solar cells and PEC devices have obtained great breakthroughs with PCEs of up to 5% and high photocurrent of 12.08 mA cm⁻² at 0 V/RHE in only one year [6]. It is well known that a tuned bandgap and determined band position are vital to facilitate light utilization and charge transport for photoelectric devices [25]. CBTS with a tunable bandgap can be achieved by substituting S with Se [26], which makes it suitable for application in solar cells and PEC devices [3, 24]. In addition, a tunable bandgap and structure change in Cu₂BaSn(S_1−xSe_x)₄ (CBTSSe) may have huge potential in thermoelectrics [10, 27].

Methods for the synthesis of CBTSSe have been reported in the last few years [6, 8, 28–36]. The most common method is a two-step process including physical vapor deposition (PVD) and post sulfurization [6, 32, 36, 37]. CBTS films obtained by this method have shown good properties. However, the high cost of co-sputtering hinders its further application. Solution-based methods have attracted great attention due to low cost, large scale and easy composition control [28, 31, 33, 38–43]. The first solution-processed Cu₂BaSn(S_1−xSe_x)₄ film was synthesized from a thiol-amine solution by McCarthy and Brutchey in 2018 [31]. Using a molecular solution method, CBTS solar cells obtained a PCE of 1.72% [28]. Tunable bandgap CuBaSnS_4−xSe_x (x < 2.4) films (from 1.56 to 1.86) were synthesized by a thiol-amine solvent mixture [31]. However, there are difficulties in the synthesis of solution-processed Cu₂BaSnSe₄ (CBTSe) films due to the low solubility of barium salt and difficulty in substituting S by Se completely [31, 44]. Here, we introduced a thionothiolic acid metathesis method for the synthesis of CBTSSe with fully stoichiometric Se/(S + Se) ratios (from zero to one); properties with different ratios were also investigated. CBTSSe solar cells with different ratios were fabricated to achieve the best PCE of 2.01% in current solution-processed reports.

2.1. Chemicals and materials

2.2. Preparation of the CBTSe precursor solution

2.3. Deposition of the CBTSSe absorber layer

2.4. Device fabrication

2.5. Characterization

The materials and precursor solution are shown in figure S1. In order to investigate the reaction in this solution, the CBTS precursor is studied by thermogravimetric analysis (TGA) and attenuated total internal reflectance Fourier transform infrared spectrum (ATR-FTIR). The TGA data in figure 1(a) indicates that there are two decomposition procedures from room temperature to 200 °C and almost no reaction happens when the temperature is higher than 200 °C. This result also can be demonstrated by the ATR-FTIR spectrum. The ATR-FTIR spectra of the precursor annealed at different temperatures are shown in figure S2. Compared with the precursor annealed at 100 °C, the ν_C-OH bend (1030–1130 nm), ν_CH2 twisting (920–1050 nm) and ν_CNH stretch (1540 nm) disappear while temperature reaches 200 °C [31]. In addition, there is no infrared peak at 2570 cm⁻¹ for ν_S-H stretch at any temperature, which suggests that the sulfhydryl has reacted or decomposed completely. In order to study the structure of annealed films, figure 1(b) shows the X-ray diffraction (XRD) pattern of as-grown films. The XRD pattern well matches the COD reference (COD 4000567) [6, 8] with lattice constants a = 11.1215 Å, b = 11.2373 Å and c = 6.7531 Å, suggesting that a well-crystallized CBTSe film with Ama2 structure was synthesized. As is well known, there is a similar crystal structure for CBTSe and chalcogenides impurity (BaSe (PDF 18-0191), Cu₂SnSe₄ (PDF 16-670)). Furthermore, we utilize Raman spectroscopy to identify the purity of the film. The Raman spectrum in figure 1(c) has two dominant peaks at 191 cm⁻¹ and 240 cm⁻¹ [34]. There is no obvious impure phases like SnSe (107 cm⁻¹, 181 cm⁻¹) and Cu₂SnSe₃ (179 cm⁻¹) in the final films [30]. Figure 1(d) presents the transmission electron microscope (TEM) images of CBTSe. The interplanar spacing of 0.40 nm and 0.33 nm can be indexed to the planes of [211] and [031], respectively. The corresponding dihedral angle of 40° is in accord with the calculated angle between the [211] and [031] planes in the hexagonal phase. The optical properties of the CBTSe films were characterized by an ultraviolet-visible (UV–vis) absorption curve and photoluminescence (PL) spectrum excited by a 532 nm laser in figure 1(e). The absorption wavelength and PL emission peaks are approximately 685 nm and 695 nm, respectively. The 10 nm shift between UV–vis and PL peak indicates less band tail inside the CBTSe film [30]. The full-width-half-maximum of the PL spectrum is around 66 nm, which is much smaller than CZTS and CuInGaSe₂ [15]. It shows as-grown CBTSe having a high quality. The plot of (αhv)² versus hv in figure 1(f) shows that the bandgap is estimated to be 1.70 eV corresponding with the emission peak from the PL spectrum.

Zoom In Zoom Out Reset image size

To identify the multi-elemental valence state of the CBTSe film, high-resolution X-ray photoelectron spectroscopy (XPS) of CBTSe is measured in figure 2. The survey of CBTSe shown in figure S3 indicates the constituent of Cu, Ba, Sn and Se. The Cu 2p region shown in figure 2(a) is fitted by a double set of peaks at 931.5 eV and 951.6 eV with a splitting of 20.1 eV, which is well matched with the Cu¹⁺ [40]. The Ba 3d spectrum in figure 2(b) can be fitted by a double set of peaks at 779.3 eV and 794.6 eV with a splitting value of 15.3 eV, which agrees with Ba²⁺ [31]. The Sn 3d data in figure 2(c) can be fitted by a double set of peaks with 485.7 eV and 494.1 eV, which is assigned to Sn⁴⁺ [31]. Additionally, the lack of satellite peaks in the Sn 3d region indicates that Sn²⁺ has been oxidized to Sn⁴⁺ during the annealing process. The Se 3d data in figure 2(d) could be divided into two fitting peaks at 54.1 eV and 53.3 eV with a splitting of 0.8 eV, indicating an oxidation state of −2 for Se in CBTSe [40]. Besides, there is a sub-peak at higher binding energy (55.3 eV). A similar spectrum was reported in solution-processed CBTSe, which might come from the selenium-oxide by oxygen absorbed on the film surface [29].

Zoom In Zoom Out Reset image size

The morphology of CBTSe film was conducted by a scanning electron microscope (SEM). Figure 3(a) and (b) shows the top-view and cross-section SEM images of CBTSe. The images present a dense, large grain (with average size of 3.3 μm) and pinhole-free film grown on Si/SiO₂ substrate. It is worth noting that the grain size of CBTSe films can be grown as large as ∼7.0 μm. Large grain is of great benefit to reduce recombination in the grain boundary, which promotes a device's photoelectric performance. Figure 3(b) displays a thickness around 1.5 μm of the compact CBTSe films. It is reported that a kesterite absorber with a thickness of ∼1.5 μm shows the best properties [45]. Figure 3(c) reveals the atomic force microscope (AFM) images to gauge the roughness of film. The rough surface may result from the non-uniformity of grain size in the CBTSe film. It is worth noting that the surface roughness can be reduced by tuning the Se/(S + Se) ratio in the next part. Figure 3(d) presents the composition analysis of as-grown film by an energy dispersive spectrum (EDS). As seen from figure 3(e)–(g), the homogeneous images indicate the continuous and uniform distribution of Cu, Ba and Sn inside the film. The atomic ratios of Cu, Ba and Sn are coincidental to the stoichiometry as well as the ratio of raw materials. The distribution of Se (figure 3(h)) is basically observed well along with the SEM image in figure 3(d). This may be attributed to a little bit being Se-rich on the surface (see table S1 (Se/Cu + Ba + Sn = 1.1)).

Zoom In Zoom Out Reset image size

A tunable bandgap is of great importance for the wide application of kesterite materials [8, 31, 34, 40]. In this work, different Se/(S + Se) ratios of CBTSSe thin films were synthesized to tune the bandgap. By changing the concentration of selenium vapor during the annealing process, we can obtain four kinds of CBTSSe films with different ratios. As shown in table S2 (atomic ratio) and figure S4 (EDS), sample 1 was sulfureted in a pure sulfur atmosphere, and then the Se/(S + Se) ratio was calculated to be zero expectedly. Samples 2 and 3 were annealed at different concentrations of selenium vapor atmosphere, and the ratios were calculated to be 0.36 and 0.62, respectively. Sample 4 was annealed in a pure selenium atmosphere. The result indicates that the Se ratio increases to one, which can be considered to be CBTSe. Therefore, CBTSSe with a tunable Se/(S + Se) ratio can be synthesized by regulating the concentration of selenium vapor during the annealing process. The characterization of CBTSSe films with different Se/(S + Se) ratios were measured to uncover the phase and composition changes. Figure 4(a) shows the XRD patterns of CBTSSe with different ratios. Sample 1 matches well with trigonal CBTS (P3₁, JCPDS 30-0124) [8]. As the ratio varies from samples 1 to 3, a low degree shift is detected in full range of the XRD pattern. Figure 4(b) shows the magnified XRD pattern from 20° to 30°. (103) and (104) peaks shift towards the lower degree obviously. It can be recognized to be a structure change for the anion substitution of S by Se [8, 41]. It is interesting that the XRD pattern of sample 4 is totally different from the others, as discussed above, which indicates that the crystal structure converts from P3₁ to Ama2 (COD 3000567) [6, 14, 26, 28, 30]. The unit cell parameters of CBTSSe films with different ratios are calculated and shown in table S3, which indicates a change in lattice parameters and is in accordance with a previous report [34]. For further investigation of the structural change of films, the Raman spectrum in figure 4(c) shows that the main characteristic peak of sample 1 appears at 339 cm⁻¹ [24, 32]. As the ratio decreases, the characteristic peak at 339 cm⁻¹ (Sn and S vibration) gradually decreases until it disappears. Meanwhile, the peak at 250 cm⁻¹ (S and Se vibrations) increases at a lower wavelength [41]. As the ratio varies from samples 1 to 4, the characteristic peaks change from 215 cm⁻¹ and 339 cm⁻¹ to 191 cm⁻¹ and 241 cm⁻¹. These changes indicate the substitution of S by Se. Figure 4(d) represents the bandgap estimated from αhv versus hv. As seen, the bandgap decreases from 2 to 1.68 eV as the ratio varies from zero to 0.62. Nevertheless, the bandgap increases from 1.68 eV to 1.70 eV when the ratio increases from 0.62 to 1. It is found that the smallest bandgap is 1.68 eV at 0.62 of Se/(S + Se), which corresponds to the crystal structure converting from P3₁ to Ama2 [6, 8]. PL spectra of CBTSSe films with different ratios are presented in figure 4(e). There is an obvious red shift in emission wavelength with the ratio of Se/(S + Se) increasing from 0 to 0.62. As the ratio reaches one, the wavelength presents a blue shift from 745 nm to 700 nm, corresponding to the bandgap increase. The varying trends of the PL emission peak and bandgap versus the Se/(S + Se) ratio are summarized in figure 4(f). Both measurements show the same results that the lowest bandgap appears at the ratio of 0.62. To one's interest, there is anti-variation in optical property with wider bandgap while crystal structures change from P3₁ to Ama2.

Zoom In Zoom Out Reset image size

In order to investigate the morphology change of CBTSSe films with different Se/(S + Se) ratios, SEM and AFM measurements are used in figure 5. The SEM images from figure 5(a) to (d) show the different ratios of Se/(S + Se) in CBTSSe films. All the films have dense and continuous structures. The grain of the CBTS (ratio: 0) is small (1 μm) and heterogeneous. As the ratio of Se/(S + Se) increases from 0 to 0.62, the grain size becomes as large as 10 μm. There is even no obvious boundary in the surface at a ratio of 0.62 (as shown in figure S5). This is very useful to the following device fabrication of solar cells. Nevertheless, the grain size becomes smaller as the ratio increases to one. Furthermore, the AFM images shown in figure 5(e) to (h) and (i) to (l) (3D images) reveal the surface roughness reduced from 48.6 nm to 36.8 nm with ratio variance from 0 to 0.62, and then the surface becomes rougher from 36.8 nm to 99.8 nm with the ratio continually increasing. That is, CBTSSe films are grown with the largest grain size and the smoothest surface at 0.62 of Se/S + Se. The EDS and Raman intensity mapping in CBTSSe films shown in figure S6 indicate the homogeneous distribution.

Zoom In Zoom Out Reset image size

CBTSSe solar cells with different ratios were studied to unveil the effect on device performance. Solar cells were fabricated by following the structure of SLG (2 mm)/Mo (800 nm)/CBTSSe (1.5 μm)/CdS (50 nm)/SnO₂ (50 nm)/IZO (100 nm)/Ag (100 nm), as shown in figure S7 (cross-sectional image of solar cells). Figure 6(a) and table 1 present the photocurrent density voltage (J–V) curves and device characteristics with difference ratios. As the ratio changes from 0.36 to 0.62, the PCE increases from 1.84% to 2.01% with higher V_OC from 0.40 V to 0.42 V and larger J_SC from 10.1 mA cm⁻² to 10.8 mA cm⁻². As the ratio further increases to 1, the PCE decreases to 1.48%. This decline of PCE may be caused by the wider bandgap, smaller grain and larger roughness in CBTSe. The device with a ratio of 0.62 shows the best performance with highest V_OC and J_SC. To detect collection efficiency for photo-induced carriers, EQE spectra of solar cells with different ratios were also measured in figure 6(b). Due to the absorption of CdS and ZnO, the EQE data shows little intensity before 400 nm. The quantum efficiency reaches a maximum value at 460 nm and ends at around 725 nm. As the ratio increases from 0.36 to 0.62, the EQE increases significantly with the highest EQE (about 62%). Then, the intensity decreases as the ratio increases to 1. All the results are consistent with the result from the J–V curve. The J_SC integrated from EQE is shown in figure S8(a) as well as agreeing with the result of J–V. In addition, the bandgap of the CBTSSe absorber is further estimated by the plot of [E*ln (1-EQE)]² versus hv on the basis of EQE data. As shown in figure S8(b), the bandgaps are estimated to be 1.77 eV, 1.74 eV and 1.78 eV, in agreement with the UV–vis curves. The sharp decline between 650 nm to 750 nm corresponds with the PL spectra. Moreover, the sharp decrease also demonstrates the absence of a band tail in the CBTSSe absorber. The J–V curves and EQE data elucidate that device performance increases as the ratio increased from 0.36 to 0.62, and obtained the best performance at a ratio of 0.62. After that, the performance reduces as the ratio increased to 1, which is identical to the results analyzed for morphology, structure and bandgap.

Zoom In Zoom Out Reset image size

In this work, we introduced a thionothiolic acid metathesis process to synthesize full stoichiometric CBTSSe films for the first time. Different stoichiometric Se/(S + Se) of CBTSSe from zero to one can tune the bandgap from 2 to 1.68 eV. The experiment shows that there is a phase change (the crystal structure) from P3₁ to Ama2 at the ratio 0.62 of Se/(S + Se) in CBTSSe films. As-grown Cu₂BaSn(S_1−xSe_x)₄ (x = 0.62) has the lowest bandgap of 1.68 eV as well as the largest grain (as large as 10 μm) and the smoothest surface. An optimized bandgap and high-quality growth ensures the best power conversion efficiency of 2.01% for CBTSSe solar cells. Our work indicates that full stoichiometric CBTSSe could be obtained by a low-cost solution process method and suggests that CBTSSe with different ratios will have promising applications in the future.

We gratefully acknowledge the support from the National Natural Science Foundation of China (21971172, 21671141, U1810119, 51774161, 51504121), the Priority Academic Program Development (PAPD) of Jiangsu Higher Education Institutions for Optical Engineering, and Jiangsu Collaborative Innovation Center of Photovoltaic Science and Engineering.