Sputtered single-phase kesterite Cu₂ZnSnS₄ (CZTS) thin film for photovoltaic applications: Post annealing parameter optimization and property analysis Open Access

Photovoltaic solar power generation has grown exponentially in recent decades due to increasing energy consumption. Researchers in this field have focused on mastering low-cost and high efficient photovoltaic device manufacture. The most widely used Si-based solar cells exhibit high conversion efficiencies of up to 24.5%.¹ Thin film solar cells are becoming increasingly prominent due to their high power conversion efficiencies (PCEs) and direct and tunable band gap energies (E_g), and because less material is required to construct them compared to conventional Si-based solar cells. The PCE of an amorphous silicon (a-Si) thin film can reach 13.6%,² whereas solar cells based on copper indium selenide (CIS), copper indium gallium selenide (CIGS), and cadmium telluride (CdTe) offer higher PCEs of up to 22%.³ Another potential solar absorber material is the p-type chalcogenide semiconductor Cu₂ZnSnS₄ (referred to as CZTS), which is derived from the chalcopyrite structure of CIGS by replacing In and Ga with less expensive and more earth-abundant elements such as Zn and Sn.^4–7 This material also has the advantage that it does not contain cadmium, which is toxic. Solar cells based on CZTS show great promise, as they exhibit high absorption coefficients⁸ (>10⁴ cm⁻¹) and a tunable band gap (E_g ∼ 1.45 eV–1.6 eV) that can boost the PCE. Among the various possible structures of CZTS, kesterite shows greater stability than stannite and wurzite, as kesterite possesses lower energy than those of the other two structures.⁹ 

Different methods of growing CZTS thin films have been reported, including vacuum and nonvacuum (i.e., solution-based) deposition approaches. The use of vacuum-based technologies such as thermal evaporation,¹⁰ sputtering,^11,12 electron beam evaporation,¹³ and pulsed laser deposition (PLD)^14,15 increases manufacturing costs but results in high-quality thin film. Manufacturing costs can be lowered through the use of solution-based techniques such as sol-gel dip¹⁶ and spin¹⁷ coating, chemical bath deposition,¹⁸ screen printing,^19,20 spray pyrolysis²¹ and electrodeposition,²² and successive ionic layer adsorption and reaction (SILAR).²³ 

In 2018, Yan et al.²⁴ reported that heterojunction heat treatment of a CZTS-based solar cell permitted a PCE of >10%. Green et al.²⁵ achieved a PCE of 11% and an open circuit voltage of 730.6 mV using a CZTS-based solar cell. A record PCE of around 12.6% was attained for a Cu₂ZnSnS₄-based solar cell fabricated by Wang et al. using a pure hydrazine solution approach.²⁶ Saha and Alam reported that a maximum efficiency of 17.59% at 550 nm and an open circuit voltage of 940 mV were observed for a silver mixed CZTS (ACZTS) cell with a CdS/ACZTS/CZTS/ITO structure, according to optoelectronic simulations.²⁷ Nevertheless, the PCEs of CZTS cells are yet to reach the efficiencies of solar cells based on CIGS and CdTe. According to the Shockley–Queisser theory,²⁸ a maximum PCE of 32.2% is possible using CZTS. Further research into CZTS solar cells is, therefore, needed to improve their PCEs, as this would lead to a more prominent role for CZTS solar cells in energy generation in the future.

The application of kesterite in solar cells has attracted considerable attention despite the commercialization of direct band gap absorbers such as CdTe, CIS, and CIGS/Se. Great efforts have been directed into the design of an optimized device structure that can accommodate current CZTS absorbers, which have relatively high defect levels and short carrier life times. The usage of reduced materials necessitates the application of an ultrathin crystalline absorber film. The issue of optical losses²⁹ and implementation of light trapping^30,31 can play a vital role for higher PCE of the CZTS film.

The novelty of the present work is the postannealing parameter optimization after successful growth of quaternary CZTS thin film on a transparent glass substrate using a single sputtering system at 200 °C with a delicate CZTS target. When designing a highly efficient solar cell, the appropriate absorber layer thickness for light trapping can be determined by comparing the absorption length with the carrier diffusion length in the layer.³² This layer thickness affects the overall device design.²⁹ The present study demonstrates the effectiveness of adjusting the annealing parameters (the pressure and the temperature) to obtain the CZTS film of the desired thickness for use in solar cells. Sulfurization is an indispensable step as sulfur is apt to be deficient in most cases, which also makes the system expensive. To obtain the desired stoichiometry, sulfurization is obvious for some deposition processes. In addition, the selenization method could have some advantages in obtaining smooth, uniform, and compact films. However, both selenization and sulfurization processes are skipped here to employ a low cost method. As it is a quaternary compound, CZTS contains some binary and ternary phases that need to overcome to obtain the pure phase only. Moreover, there are numerous issues to address, such as the grain size, surface smoothness, roughness, stoichiometric ratio, volatility of tin, and pinholes in the film surface when seeking to obtain a highly textured CZTS film with good coverage. In the investigation reported here, the crystal structure, surface morphology, composition, and optical properties of postannealed CZTS thin film samples were rigorously studied to better understand and optimize the growth and phase formation of CZTS thin films. This knowledge should facilitate attempts to fabricate CZTS thin films with the optimal stoichiometry and photovoltaic properties for use in solar cells.

Soda lime glass (SLG) substrates (3 × 3 cm²) were washed with detergent, brushed, and sequentially cleaned with methanol, ethanol, acetone, and deionized water in an ultrasonic bath for 10 min each. The glass substrates were then rinsed with deionized water and dried. The quaternary CZTS was deposited by a radiofrequency (RF) magnetron sputtering system (NSC-4000, NANO-MASTER, Inc., USA) operated at 80 W for 2 h. The target used in this study was manufactured by Kurt J. Lesker Co. (USA), and was $2 ″$ in diameter and $0.12 5 ″$ thick. The composition of the target was 2:1:1:4 C:Zn:Sn:S, and its purity was 99.99%. A voltage was applied between the target material (cathode) and the substrate (anode) to be coated with the target material. Electrons from the surface of the target ionized the process gas, resulting in the formation of a plasma. The chamber pressure was reduced as much as possible to stop the background gases from reacting with the film or sputter target. Careful control of the partial pressures of the reactive gases led to the growth of a thin film on the substrate via sputtering. The target to the substrate distance was set to 7 cm. When the base pressure reached 6.3 × 10⁻⁶ Torr, precursors were deposited on the SLG inside a sputter coater at a working pressure of 4.5 mTorr. An Ar flow of 5 SCCM was maintained throughout the process. The substrate holder was rotated at 20 rpm to maximize film uniformity and target utilization. The temperature of the chamber during deposition was maintained at 200 °C.

A successful heat treatment strategy for CZTS consisting of a two-stage process in which precursors were prepared at a substrate temperature of <300 °C and then annealed at high temperature (around 500 °C) has been reported.¹⁵ In the present work, postannealing was conducted at two different temperatures, 470 °C and 560 °C, under an N₂ atmosphere in a tube furnace (GSL-1100X, MTI Corporation, USA). Samples were maintained at the selected temperature at four different base pressures (150 Torr, 250 Torr, 350 Torr, or 450 Torr) for 30 min. Afterwards, the samples were left to cool within the annealing chamber.

X-ray powder diffraction (XRD) and Raman spectroscopy were performed on a bare CZTS absorber layer to characterize the microstructure of the CZTS film. XRD patterns were recorded using an EMMA X-ray diffractometer (GBC Corporation, Australia) using Cu-Kα radiation with a wavelength of 1.5406 Å. The unit was operated at 35 kV and 28 mA. Raman spectra (excitation wavelength: 785 nm) were obtained using a MacroRAM Raman spectrometer (Horiba Scientific, Japan) with a tunable output power of 7 mW–450 mW in the backscattering configuration. This spectrometer was equipped with a 685 gr/mm holographic grating for superior stray light rejection. An UV-visible (UV-Vis) spectrophotometer (UH4150, Hitachi, Japan) was used to study the optical absorbance at visible wavelengths. The thickness of the film was measured with a surface profilometer (Dektak, Bruker, USA), and the precursor morphology was examined by atomic force microscopy (AFM; Flex AFM, C3000, Nanosurf, Switzerland) and scanning electron microscopy (SEM; EVO18, Carl Zeiss, UK). The chemical composition of the film was measured by energy-dispersive X-ray spectroscopy (EDX) using a Team EDS system (EDAX, AMETEK, USA; beam voltage: 15 kV) attached to the SEM unit.

Table I lists the annealing conditions and the average film thickness of the samples measured using both a surface profilometer and SEM. There was no significant variation in the thickness among the samples annealed under different conditions. Figure 1 shows a SEM image of the cross-section of a CZTS thin film.

The deposited CZTS films exhibited an amorphous nature. Polycrystalline CZTS thin films with the crystal structure of kesterite were obtained by annealing. The annealing process causes the structure of the crystal to shift from a nonequilibrium state to a state closer to thermodynamic equilibrium. The XRD profiles of CZTS thin films annealed under different conditions are shown in Fig. 2 (JCPDS-004751). The XRD profiles of all the samples exhibit a particularly intense (112) peak as well as less intense (200), (220), and (312) peaks. It is worth noting that the peak positions do not shift depending on the annealing conditions, indicating that the CZTS phase is stable.

Bragg peaks from the Cu_2−xS (x = 1) phase were observed in the XRD patterns of the thin films annealed at 470 °C (JCPDS-06-0464). The presence of a Cu_2−xS impurity phase indicates a significantly increased Cu fraction in the film, although this promotes the growth of single-phase CZTS. Sun et al. reported the appearance of this impurity in CZTS films deposited at 350 °C–400 °C.³³ Tang et al. reported a diffraction peak from a Cu_xS phase in a film that had been sulfurized at 450 °C.³⁴ This peak disappeared when the CZTS film was annealed at 500 °C, 525 °C, or 550 °C. Similarly, in the present investigation, Bragg peaks from a CuS phase were seen for the samples annealed at a higher temperature (560 °C). It is possible that at this annealing temperature, the Cu:Zn:Sn:S ratio has reached ideal stoichiometric condition, which diminishes impure phases such as CuS. Thus, the Bragg peaks disappear when the higher annealing temperature is applied.³⁵ 

Figure 3 shows that Bragg peaks³⁶ for the (112), (200), (220), and (312) planes appear in the XRD patterns of the CZTS thin film samples.^16,17 Two more characteristic peaks of kesterite that are visible in Fig. 2 at 2θ = 38.07° and 52.06°can also be seen in Fig. 3. This figure also shows that the (112) peak of S-560-450 is sharper than that of S-470-450, indicating that the peak sharpness increases with annealing temperature, in agreement with previous reports.^33,37

In addition, the full width at half maximum (FWHM) of a diffraction peak increases with decreasing crystallite size. The average crystallite size can be estimated from the FWHM of the most intense peak (112) using the Scherrer formula³⁸ (assuming that the nanocrystals are spherical) as

where β is the FWHM in radians, λ is the wavelength, and k is the Scherrer constant. The calculated average crystallite size, D, of each thin film sample is shown in Table II. It is clear that the samples annealed at the higher temperature (560 °C; samples S-560-150,S-560-250, S-560-350, and S-560-450) have lower FWHMs and larger crystallite sizes than the samples annealed at the lower temperature (470 °C). The crystallization temperature plays an important role in the morphology of the final structure. Increasing the temperature causes greater mobility and surface diffusion, which allows the deposited material to migrate and coalesce into larger islands, leading to a larger average crystallite size and a lower FWHM. The dislocation density δ is also lower for the samples annealed at a higher temperature, which indicates that these samples have fewer lattice imperfections and decreased strain.^39,40 However, the crystallite size does not appear to vary significantly with increasing base pressure.

The dislocation density,⁴¹ the number of dislocations in a unit volume of a crystalline material, is calculated as

Crystal dislocation increases the strain in nanostructured materials. In the Williamson–Hall isotropic strain model (W-HISM),⁴² the crystallite size and lattice strain are independent of each other. According to the Williamson–Hall anisotropic uniform deformation energy density model (W-HUDEDM), there are more volume defects at grain boundaries when the crystallites are very small. The internal pressure exerted by the surface tension due to volume defects creates a stress field. This additional stress at grain boundaries causes lattice strain. For simplicity, the strain, ε, of the thin film is measured as⁴¹ 

It is difficult to distinguish experimentally between tetragonal CTS, cubic CTS, sphalerite (ZnS), and CZTS as they present very similar lattice parameters.^36,43 To detect traditional issues of phase segregation (secondary phases), the Raman spectrum in the wave number region 140 cm⁻¹–495 cm⁻¹ was obtained for a thin film sample at room temperature; this spectrum is shown in Fig. 4. It presents well-defined peaks at 288 cm⁻¹, 334 cm⁻¹, and 361 cm⁻¹, corresponding to the reported characteristic peaks for polycrystalline kesterite films.⁴⁴ Moreover, it is obvious that there are no extra peaks due to the presence of other compounds such as SnS₂ (314 cm⁻¹), Cu₂SnS₃ (352 cm⁻¹ and 374 cm⁻¹), cubic ZnS (352 cm⁻¹ and 275 cm⁻¹), and orthorhombic Cu₂SnS₃ (318 cm⁻¹), which confirms that the sample consists of single-phase CZTS.^45,46

The morphology of the thin film is one of the most important properties to understand the surface phenomena. Figure 5 shows AFM micrographs of a 25 µm² region of the surface of each annealed CZTS thin film sample. These images reveal that the sputtered films have similar surface topographies and thicknesses and that the Volmer–Weber mode was the dominant growth mode. The films show far better coverage than a film deposited by a sol–gel technique.¹⁷ The variation in the root mean square (rms) surface roughness of the thin film derived from the AFM image with annealing pressure and temperature is shown in Fig. 6. This figure indicates that the surface roughness increases significantly with increasing annealing temperature. The surface roughness barely changes with increasing base pressure when the annealing temperature is 470 °C, but it changes drastically upon increasing the base pressure from 350 Torr to 450 Torr, when the annealing temperature is 560 °C. Changing the annealing temperature shifts the phase transition of the CZTS thin film.³⁷ The average grain size increases with increasing annealing temperature.⁴⁷ At high temperatures, grain coalescence and reorganization occur due to increased surface mobility. Figure 5(i) shows the topography of the CZTS thin film sample S-470-150, based on AFM data. Table II shows that when the temperature is higher than 500 °C, the grains in the film become larger.

The analysis of the AFM images shown in Fig. 6 confirmed that the grain size and surface roughness of the CZTS thin film increased with increasing annealing temperature. However, some voids were observed on the surface of the film. This was due to the significant loss of Sn at the higher annealing temperature.⁴⁸ Nevertheless, annealing appears to be useful for enhancing the crystallinity and phase formation of the film and is, therefore, beneficial for device quality. Our results indicate that CZTS thin films should be annealed at a temperature of at least 500 °C because an increased grain size leads to a lower recombination rate, which enhances the PCE.

The atomic percentages (at. %) for the CZTS thin film samples were determined using EDX and are shown in Table III. The theoretical stoichiometry of CZTS is Cu:Zn:Sn:S = 2:1:1:4. It is evident from Table III that the films are Cu rich and Zn poor compared to the expected stoichiometry. Chen et al. reported that Cu-rich and Zn-poor conditions are suitable for the growth of single-phase CZTS films.^49,50 Cu is a lighter element and, therefore, has a higher flow speed than Zn, which leads to the enrichment of Cu in the film.³³ Zn and S are more volatile and have lower melting points than the other two elements in CZTS, which results in Zn- and S-deficiency in CZTS thin films.⁵¹ However, even without sulfurization, the sulfur-to-metal ratios of the present samples are compatible with those stated in the previous reports.^45,46 The films annealed at 560 °C (S-560-150, S-560-250, S-560-350, S-560-450) are clearly Sn deficient compared to the other samples. This undesirable phenomenon is likely to be due to the increasing loss of volatile elements or compounds at higher crystallization temperatures—an effect that has a significant impact on the final composition. An additional disadvantage of applying a relatively high annealing temperature is the possibility of diffusion through substrate interfaces, which may result in the formation of solid solutions. According to Weber et al.,⁴⁸ the evaporation rates from different phases increase according to the sequence Cu₂ZnSnS₄ < Cu₄SnS₄ < Cu₂SnS₃ < SnS. SnS is lost from the CZTS thin film at a significant rate at temperatures of 550 °C and above. However, Todorov et al. reported the opposite composition i.e., Zn-rich and Cu-poor CZTS thin film yielded the highest conversion efficiency.⁵² 

The UV-Vis absorption spectra of the CZTS thin film samples depicted in Fig. 7 indicate that the samples show high optical absorbance and a large absorption coefficient, α, in the visible region (ignoring reflection and transmission losses). The nature of the transition associated with absorption can be determined using the classical equation²¹ for the absorption coefficient

Here, n = 1/2 for an allowed direct transition and n = 2 for an allowed indirect transition. The absorption coefficient, α, can be calculated from the absorbance A using⁵³ 

where I is the transmitted intensity, I_o is the incident intensity, A is the absorbance, and d is the thickness of the film. The absorption coefficients of the samples were found to be larger than 10⁴ cm⁻¹, which is consistent with previous reports.^17,21,23 Studying the optical absorption of the samples allows us to explore properties such as the band structure, refractive index, and high-frequency dielectric constant,⁴⁶ all of which are important influences on solar cell performance.

The optical band gap determines the portion of the solar spectrum that a photovoltaic cell will absorb. The band gap energy E_g for direct transitions can be calculated by extrapolating the linear portion of the (αhν)² vs hν plot, and the band gap energies of the samples calculated in this manner are shown in Table IV. These energies are observed to range between 1.47 eV and 1.51 eV. This variation in the band gap energy is largely explained by differences in the homogeneity and crystallinity of the thin films, which in turn leads to the differences in the crystallite size between the samples.⁵⁴ Note that the calculated band gap values are close to the optimum band gap for solar cells.⁵⁵ 

The refractive index (n) of the thin film is a major determinant of the possibility of total internal reflection inside the solar cell.³¹ The refractive index of each of the CZTS thin films examined in this work was calculated using the Moss relation⁵⁰ 

where k is a constant with a value of 108 eV.

The dielectric properties of a material relate to its capacity to impede the electron movement when it is polarized under the influence of an external electric field. Materials with suitable dielectric constants are needed to develop efficient solar cells.

The high-frequency dielectric constant ε_∞ of each sample was determined using the relation^17,23

The static dielectric constant ε_o of each thin film^17,23 was calculated using the relation

The band gap, refractive index, and dielectric constant data shown in Table IV for the CZTS thin films annealed under different conditions are all in good agreement with previously reported results.²³ 

Here, we report the production of thin (∼2000 nm) Se-free kesterite CZTS films without notable phase segregation or voids via sputtering. Sample properties were analyzed after postannealing treatment to find out the optimum annealing conditions for the CZTS thin film. The Raman spectra of the samples verified that they were indeed kesterite based on the presence of a large peak at 334 cm⁻¹ and additional small peaks at 288 cm⁻¹ and 361 cm⁻¹. The samples annealed at 560 °C also had larger crystallites than those annealed at 470 °C, as demonstrated by the sharper peaks in the XRD spectra of thin films annealed at 560 °C. The grains in the samples were compact, and agglomeration increased with annealing temperature, which increased the absorption coefficient. AFM analysis revealed that the rms roughness of the thin film annealed at 560 °C changed drastically upon increasing the base pressure from 350 Torr to 450 Torr. EDX analysis indicated that the CZTS samples produced without sulfurization were Cu and S rich but Zn and Sn poor. Significant SnS losses were found to occur at temperatures above 500 °C, resulting in the generation of voids on the film surface and affecting the stoichiometry of the film. UV-Vis absorption spectra indicated that the direct band gap energies (E_g) of the thin films ranged from 1.47 eV to 1.51 eV. Nevertheless, the notably improved CZTS film was realized in this work through postannealing parameter optimization, which could facilitate enhanced manufacturing throughput. We can conclude that samples annealed at a higher temperature (560 °C) show better crystallinity, as predicted, and the base pressure plays a vital role in the case of surface morphology. Future work in this area will focus on improving the morphology, chemical composition, and electric properties of the thin films, thereby enhancing their optical absorbance and stoichiometry to achieve further improvements in CZTS solar absorber quality.

The logistic support of the Solar Energy Technology Research Laboratory of IFRD, BCSIR, is gratefully acknowledged.