1. Introduction

In response to global energy demand expected to reach terawatt (TW) level by 2050, extensive use of renewable energies is essential without affecting the environment. Solar energy is a kind of renewable energy, which can meet the world energy demand in TW level with a large amount of solar energy to spare [1]. A potentially effective way to use solar energy is to use photovoltaic (PV) devices to convert it directly into electrical energy. From the very beginning of the invention of the solar cell, it has been improving every day. The first generation crystalline-Si (c-Si) based devices has been a leader of mainstream the solar PV market. It occupy approximately 95% of the global PV market share, while all other thin films PV technologies (i.e. CdTe, Cu(In,Ga)Se₂ (CIGS) and thin film Si) occupy only 5% [2]. However, at present, c-Si technology and c-Si solar cells have the limitations of thick and expensive Si wafer, high processing temperature, high-vacuum fabrication process and other complicated and expensive processing technologies. Over the past five decades, new cost-effective and earth-abundant thin film materials and device fabrication technologies have been explored and developed by many researchers to make more efficient and environment friendly solar cells at lower cost. During this period, many TFSCs technologies came out, of which very few such as CdTe, CIGS, and perovskite solar cells have got widespread attention because their efficiency crosses commercial demarcation. Despite remarkable advances of CdTe and CIGS technology, the use of toxic and rare earth materials in CdTe solar cells, material or resource constraints, and complicated fabrication process of CIGS technology are the delimiting factors for the industrial scalability of the aforementioned solar cells [3,4] and therefore, these materials need to replace for future clean energy technologies [5]. In the last 10 years, kesterite structure such as Cu₂ZnSn(S/Se)₄ with its non-toxic, low-cost, earth-abundant elements were investigated vigorously as PV absorbers to overcome the material bottlenecks in CIGS and CdTe technology. So far, the record experimental efficiency of CZTS solar cells is 13.0% in laboratory level [6]. As CZTS is a defect-prone system, the large V_oc deficiency (Eg/q – V_oc) in CZTS/CZTSe devices is due to high defects density and large potential fluctuations at the energy band edges. However, recently Atowar et al. [7,8] theoretically anticipated over 30% PCE of CZTS and CTSe-based solar cells by reducing non-radiative recombination losses through the formation of appropriate energy band structure at buffer/absorber interface and utilizing a suitable BSF layer. As a result, it is obvious that in order to find inexpensive, non-toxic, less defective, easily controllable stoichiometry, chemically stable and earth-abundant materials for efficient TFSCs, further exploration and development of PV absorber materials is needed.

Ternary Copper-antimony-sulfide and copper-bismuth-sulfide (commonly known as CAS and CBS, respectively) systems have been gaining attention in PV communities as an alternative active layer materials for TFSCs, due to their inexpensive, nontoxic and earth-abundant elements [9–15]. The first theoretical study on CuSbS₂ thin films was conducted by Dufton et al. in 2012 [16] and identified CuSbS₂ thin films as future absorber materials for TFSCs. In 2013, Kumar and Persson [17] affirmed the primary finding from Dufton and predicted very high optical absorption of CuSbS₂ thin films in the visible and infrared region, which is due to localized p-like Sb state near to the conduction band edge. Four main phases of copper-antimonysulfide system are CuSbS₂ (chalcostibite), Cu₃SbS₄ (famatinite), Cu₁₂Sb₄S₁₃ (tetrahedrite), and Cu₃SbS₃ (skinnerite) with an energy band gap that depends on the crystal structure. Among all these phases, Cu₁₂Sb₄S₁₃ and Cu₃SbS₄ with their low thermal conductivity are being studied as thermoelectric materials [18,19]. The less-explored I-V-VI₂ chalcogenides (CuSbS₂) thin films with basic optoelectronic properties such as near-ideal direct band gap ranging from 1.38-1.58 eV, a high optical absorption in broad spectrum (α ≈ 10⁵ cm⁻¹), and moderate acceptor concentration (10¹⁵–10¹⁸ cm⁻³) and the low melting point (551 ^°C) and chemically stable phase [20], seems to be an upcoming solar absorber material for PV applications [21–23]. Concerning the production cost, the development of inexpensive and easily scalable PV absorber layers depends largely on growth techniques. Many synthesis routes including physical and chemical processes such as chemical bath deposition (CBD) [12], spray pyrolysis [24], heating glass/Sb₂S₃/Cu layer in low vacuum condition [25], hybrid ink method [26], electrodeposited Sb-Cu alloys over sulfur [27], co-evaporation of Cu, Sb and S pure elements [23], low-temperature atomic layer deposition (ALD) [28] and low-cost spin coating method [29], to prepare CuSbS₂ thin films are reported in the literature. Therefore, as a potential active layer material for TFSCs, interest in CuSbS₂ thin films is increasing due to its relatively inexpensive synthesis process than other TFSCs technology, non-toxicity of all elements, and Sb, which is more abundant and less demanding than indium and gallium. In addition, unlike CZT(S/Se) system, the theoretical calculations of CuSbS₂ thin films shows that low-energy defects are far away from the band gap center, resulting in a low density of bulk recombination centers in CuSbS₂ thin films [30].

Although there are many reports on the electrical, optical and structural properties of CuSbS₂ thin films that have emphasized its potential as a PV absorber, so far only a few reports have demonstrated its practical use in PV devices [31]. Using, chemical bath deposition process, the first CuSbS₂-based solar cells with V_oc of 345 mV and very low J_sc of 0.2 mA/cm² were reported in 2005 [32] and since then, different methods have been utilized to construct CuSbS₂-based solar cells. However, all of the CuSbS₂-based solar cells demonstrated so far have been struggling with very poor PCE largely below 2% [23,27,30,33–36]. In 2016, Banu et al. demonstrated CuSbS₂-based solar cells using a hybrid ink process (non-vacuum) and achieved the highest PCE of 3.22% with a V_oc of only 470 mV [26]. It is widely acknowledged that the PV performance of CuSbS₂-based PV devices is mostly limited by their defects in absorber layer, dominant interfacial recombination of minority carriers, and unfavourable heterojunction configurations [37]. Moreover, the effective photo-generated carrier separation in CuSbS₂ thin films is limited by the short diffusion length of carriers [33], thereby resulting in a low J_sc.

CuSbS₂-based solar cells reported so far use CdS as the pn junction partner, because the CuSbS₂-based device structure is directly inherited from its CIGS counterpart. However, a “cliff-like” conduction band offset (CBO) is formed at the CuSbS₂/CdS interface [30,38], which is not within the favourable region of 0 to -0.4 eV “spike-like” CBO [39]. This results in a low V_oc in CuSbS₂ based devices. Therefore, more study is needed to reduce the recombination of minority carriers and to align the energy band at the CuSbS₂/ETL interface to improve the efficiency of CuSbS₂-based PV devices [38]. Metal oxides are widely used in TFSCs as ETL. TiO₂ is a wide band gap (E_g= 3.26 eV) n-type semiconductor with high electron affinity has been used successfully in perovskite and dye-sensitized solar cells [40–42], thus it is a promising candidate as ETL layer for fabricating high-performance solar cells as well.

In this work, we demonstrate significant improvements of V_oc, J_sc, and thereby, PCE in CuSbS₂-based solar cells by using Mg-doped CuAlO₂ thin films as the BSF layer and TiO₂ thin films as ETL to form heterojunction on both sides of the CuSbS₂ absorber layer. In general, for the BSF layer in solar cells, direct band gap thin films materials with high carrier density and mobility, and low electron affinity are desired requirements. Thermally stable ternary delafossite copper aluminate (CuAlO₂) is a p-type semiconductor. It has relatively low electron affinity compared to CuSbS₂-absorber layer, tunable band gap in the range of 2.5 to 3.5 eV [15,43], low resistivity [43], and tunable acceptor concentration, making it suitable as the BSF layer for the CuSbS₂-based solar cell. Carrier concentration in CuAlO₂ thin films can be increased by incorporating excess oxygen and divalent species into the delafossite structure. Ruijian Liu et al. [44] reported that carrier concentration of CuAlO₂ thin films can be increased to the order of 10¹⁸ cm⁻³ with 6.0% of Mg doping. As far as we know, for the first time, we design and simulate CuSbS₂-based heterojunction solar cells with the novel device structure: Al/ITO/n-TiO₂/p-CuSbS₂/p⁺-CuAlO₂:Mg/Au(100) using SCAPS-1D. The impact of physical parameters of different layers on the PV performances of the newly designed device has been investigated.

2. Methodology, material parameters, and device structure

Computer-aided design and simulation are becoming more and more compelling in the wafer-based silicon PV industry. To design and evaluate silicon PV devices, simulation tools such as PC1D, AFORS-HET, ATLAS, and Sentaurus TCAD are commonly used [45–47]. Due to the complex structure and quantum-mechanical behavior of thin-film devices, the development of thin-film PV simulation tools is quite slow as compared with wafer-based silicon simulation tools. To date, most of the TFSC simulation packages are based on 1D, which are freely available to the PV community. Some of the popular thin-film PV simulation packages are AMPS-1D, wxAMPS, SCPAS-1D, etc. [48–50]. According to Burgelman [50], an ideal thin-film solar cell simulator should meet some requirements such as allowing multiple layers (minimum six layers), capable of handling graded materials, capable of describing the possible discontinuities in the energy band at the interface between layers etc. Based on the requirements specified in the Refs. [50], we searched popular simulation software packages to select the most suitable package for CuSbS₂-based solar cell simulation. SCAPS-1D was selected because it meets all the requirements listed in the Refs. [50] for 1D devices. It also provides the most extensive set of simulation outputs, including energy band diagrams, J-V, QE, C-V, and C-f. SCAPS-1D simulation package adopts Poisson-drift-diffusion (PDD) equation sets to compute the electrical properties of PV devices. PDD model involves three coupled equations, namely, Poisson, carrier transport, and continuity equations of free holes and electrons. Under one-dimensional and steady-state conditions, SCAPS-1D solves these three coupled differential equations with appropriate boundary conditions to compute numerically the electrical properties of PV devices.

The schematic of (Al/ITO/n-TiO₂/p-CuSbS₂/p⁺-Mg:Cu AlO₂/Au (100)) n-p-p⁺ heterojunction device considered in this simulation is shown in Fig. 1(a). In the solar cell layout, n-type TiO₂ was selected as the heterojunction partner with the active p-type CuSbS₂ absorber layer forming the n-p heterojunction interface that collects the photogenerated carriers (electrons) efficiently. A window layer on the top of TiO₂ layer is used consisting of highly transparent conductive and low-cost ITO. The p⁺-CuAlO₂ layer is used as the BSF layer, which improves the photo-generated carrier separation and collection. In this simulation, the shunt and series resistance have been considered as 10³ Ω.cm² and 2.5 Ω.cm², respectively. Simulation of the considered device was performed under global AM 1.5 spectrum at 1 sun illumination (100 mW.cm⁻²) from the window layer (ITO) side at 300 K temperature. The physical properties of different layers and baseline parameters, as it is important and ultimately determine the relative accuracy of simulated results, used for computing the PV response of CuSbS₂-based solar cells are selected from experimental data, various reasonable estimates, or literature which are given in Table 1. The optical absorption spectra for the different layers, namely, TiO₂, CuSbS₂, and CuAlO₂:Mg of the device under consideration are taken from the previously reported experimental values [58–61]. As the bulk recombination loss is concerned, a single-acceptor/donor-like bulk defects with Gaussian energetic distribution is introduced into the bulk region of each semiconductor layer. Losses due to minority carrier recombination at each interface are considered by incorporating reasonable neutral interfacial defects, shown in Table-1. The values of electrical parameters for the front and back-contact are listed in Table 2.

 

Fig. 1. (a) The schematic structure of the proposed (Al/ITO/n-TiO₂/p-CuSbS₂/p⁺-CuAlO₂:Mg/Au) n-p-p⁺-heterojunction solar cell, and (b) J-V characteristics under illumination for CuSbS₂-based single and heterojunction solar cells with different device configurations.

Download Full Size | PDF

Table 1. Baseline material parameters, used for simulating CuSbS₂-based heterojunction ultrathin film solar cells [51–54].^a

View Table | View all tables in this article

Table 2. Electrical properties of front and back contact materials used in simulation

View Table | View all tables in this article

3. Results and discussion

3.1 Enhancement of the PV performance of the CuSbS₂-based solar cells

The basic model of the CuSbS₂-based solar cell (Cell-1) reported so far has the traditional Mo/CuSbS₂/CdS/n-ZnO/Al device structure, in which CdS is used as the ETL layer and Mo is used as the back-contact material, which has been developed, simulated and obtained PCE of 12.21%. This is the close approximation to the reported result in Ref. [62]. The J-V relationship of the CuSbS₂/CdS device (Cell-1) is shown in Fig. 1(b). As mentioned earlier, the low PV performance of the Cell-1 is due to the dominant interface recombination, high series resistance, and unfavourable heterojunction configurations. This CuSbS₂ based solar cell uses CdS as ETL, which is toxic and more convenient to form a “cliff-like” CBO at the CuSbS₂/CdS interface, thereby acting as the barrier to the photo-induced electrons transport and hindering the PV performance of the device. The J-V characteristics of Cell-2 is illustrated in Fig. 1(b). The CdS layer of the Cell-1 is replaced by another suitable TiO₂ ETL in Cell-2. The solar cell having TiO as ETL (Cell-2) exhibits slightly enhanced PV performance: V_oc ≈ 0.708 V, J_sc ≈ 31.59 mA/cm², FF ≈ 57.88%, and PCE ≈ 12.93%. This improvement in the PV performance of Cell-2 relative to Cell-1 is associated with the replacement of the CdS layer with the TiO₂ layer.

It is worth mentioning that the application of low-work function metal back contact on photovoltaic devices will result in the formation of Schottky barriers that impede photo-generated carriers (holes), thereby reducing V_oc and pernicious to the device’s performance. In fact, a feature called “rollover” is observed in the J-V curves of Cell-1 and Cell-2 (see Fig. 1(b)), indicating that a barrier for photogenerated carriers is developed at the back contact. Moreover, in case of Mo back-contact, the formation of a thin MoS₂ layer is highly probable during CuSbS₂ film growth on the Mo substrate because of the high reactivity of S with the Mo. A thin MoS₂ layer between Mo back-contact and absorber layer can also adversely affect the cell performance. Therefore, this result suggests that Mo back-contact is not suitable for the CuSbS₂-based solar cell, and it needs to be replaced by another suitable back-contact metal with a higher work function than Mo. Metal with higher work function may weaken the barrier of the photogenerated carriers at the back-contact.

Aiming at this point, another solar cell (Cell-3) was modeled and simulated, in which the back-contact metal (Mo) of Cell-2 was replaced by nickel (Ni) (its work function is higher than Mo), and Fig. 1(b) demonstrates the J-V characteristic curve for the Cell-3. The Cell-3 provides PCE of 18.89%, which is still below the Shockley-Queisser (SQ) limit for a single-junction solar cell with the absorber layer bandgap of ∼ 1.5 eV [63]. This is due to the short photogenerated carrier lifetime in CuSbS₂ absorber as compared to CIGS and CdTe absorber. However, when the CuAlO₂:Mg is added as BSF layer to the heterojunction of Cell-3, the efficiency of the device with double-heterojunction increases to ≈ 22.23%. The final cell with CuAlO₂:Mg as BSF layer is identified as Cell-4 and the J-V characteristic curve of the Cell-4 is also presented in Fig. 1(b), from which the main PV parameters are derived. The PV performance parameters which are derived from the J-V curves of all cells are presented in Table 3. Therefore, data in Table 3 and the simulated J-V characteristics curves suggest a superior PV response for Cell-4 with respect to other cells studied herein.

Table 3. PV performance parameters for CuSbS₂-based solar cells with different cell configurations

View Table | View all tables in this article

3.2 Energy band diagram of the CuSbS₂-based heterojunction solar cell

The energy band matching at the interface between two semiconductors is very important for the proper functioning of their bulk properties. The energy band diagram under illumination of (Al/ITO/n-TiO₂/p-CuSbS₂/p⁺-CuAlO₂:Mg/Au) heterojunction solar cell (Cell-4) is shown in Fig. 2. It has been specified in the introduction section that the low V_oc in conventional CuSbS₂-based solar cell (Cell-1) is due to the poor energy band alignment with the CdS ETL. The too higher values of the spike and cliff CBO at the heterojunction interface (ETL/absorber interface) will reduce V_oc and J_sc, respectively. Therefore, for high-efficiency solar cells, the energy mismatch at the heterojunction interface should be minimal. It is observed from Fig. 2 that a small energy mismatch (spike-like CBO) between the conduction band of CuSbS₂-absorber and TiO₂ ETL is developed, according to literature which is beneficial, ensuring an easy transport of photoelectrons at the TiO₂/CuSbS₂ interface, whereas blocking photogenerated holes at the TiO₂/CuSbS₂ interface, as the valance band edge of TiO₂ ETL layer is much lower in energy than the CuSbS₂ absorber. On the other hand, the holes from the CuSbS₂ absorber layer are easily extracted by the CuAlO₂:Mg BSF layer and transported to the back electrode, since the VBO between CuSbS₂-absorber and CuAlO₂:Mg BSF layer is very small. In contrast, the large CBO between the CuSbS₂ absorber and CuAlO₂:Mg BSF layers will prevent photoelectrons from moving to the CuAlO₂:Mg BSF layer.

 

Fig. 2. Schematic energy band diagram of the proposed (Al/ITO/n-TiO₂/p-CuSbS₂/p⁺-CuAlO₂:Mg/Au) heterojunction solar cell.

Download Full Size | PDF

3.3 Built-in-potential of the CuSbS₂-based solar cell

The built-in potential at junction yields a significant electric field to split the photogenerated electron-hole pairs in the space charge region (SCR). Therefore, the carrier recombination in the SCR area decreases as the built-in potential increases. There are two heterojunctions in our proposed device structure, namely, n-p heterojunction formed by the p-CuSbS₂ absorber and n-TiO₂ ETL layer and p⁺-p heterojunction formed by the p⁺-CuAlO₂:Mg and p-CuSbS₂ absorber layer with the carrier densities: N_A in p-CuSbS₂, N_D in n-TiO₂, and N_a in p⁺-CuAlO₂:Mg layers, respectively. The junction potential, which is defined as the splitting of the quasi-Fermi levels in both n-type and p-type semiconductor, of the n-p heterojunction and p⁺-p heterojunction can be expressed as the following expressions [65]:

Consequently, the calculated potentials of the n-p junction, p⁺-p interface, and n-p-p⁺ heterojunctions are ≈ 0.95 V, ≈ 0.83 V, and ≈ 1.78 V, respectively. This high potential is due to the double-heterojunction structure of the solar cell we designed. When the built-in potential is high, carrier recombination and trapping are lower under irradiation than when the built-in potential is low. The reduced photo-generated carrier trapping at the defect sites in the interface region further increases the effective bulk field under illumination, thereby reinforced the performance of the device as it is observed in Cell-4.

3.4 Recombination analysis of the CuSbS₂-based solar cells

Among various recombination processes in solar cells, Shockley-Read-Hall (SRH) recombination is known to be a dominant recombination process which deteriorates the cell performance mostly. Qualitative and quantitative information about the rate of carrier recombination at different regions of the device such as at the ETL/absorber interface (R^i,f), at BSF layer/absorber interface (R^i,b), in the SCR zone (R^d), and in the bulk region (R^b) can be obtained using SRH recombination statistics and analytical approximation [67]. The recombination rates depend on various factors such as thickness, carrier density, carrier lifetime, bias voltage, working temperature of the devices, etc. However, the temperature (T) and the bias voltage (V)-dependent recombination rates can be expressed as follows [67]:

Using Eq. (9) and Eq. (10), the recombination coefficient in the depletion region is

Now focusing on the extraction of the recombination coefficients at absorber/BSF interface ($R_0^{i,b}$). For the analysis of V_oc with monochromatic short-wavelength (SW) excitation, since the absorber band gap is much lower than the energy of the SW photon, electron and hole pairs are generated near the surface region. Therefore, the SRH recombination is localized in the vicinity of the front interface (TiO₂/CuSbS₂ interface). As a result, bulk SRH recombination has the weak impact, and rear (back) interface SRH recombination has negligible (i.e. $R_0^{i,b}$_≈ 0) impact on total SRH recombination. Therefore, under monochromatic short-wavelength (SW) illumination, the coefficient K₁and K₂ can be expressed as:

On the other hand, with long-wavelength (LW) excitation (photon energy is nearly equal to the band gap of the absorber), one can expect a nearly uniform generation profile across the absorber layer and therefore, the coefficient K₁and K₂ can be expressed:

Using the above two eq.(12) and eq.(14) and after simplification, the recombination coefficient at the back-contact interface can be expressed as:

Therefore, using the value of $R_0^d$ obtained previously from Eq. (11) and simplified Eq. (16), we can get the recombination coefficients at the back-contact interface ($R_o^{i,b}$). Finally, the recombination rate in the SCR zone and at the back-contact interface can be obtained using Eq. (5) and Eq. (7), and from known values of $R_0^d$ and $R_o^{i,b}$. In this study, aiming to extract qualitative and quantitative information about the recombination at the back-interface, two different monochromatic wavelengths (450 nm and 800 nm) are used to illuminate the designed solar cell. The intensity (G_a) dependent V_oc curves at excitation wavelength λ_exc= 450 nm, λ_exc = 800 nm, and white light (WL) for Cell-1 and Cell-4 are shown in Fig. 3.

 

Fig. 3. Intensity-dependent V_oc with different excitation wavelength of white light (WL), λ_exc = 450 nm, and λ_exc = 800 nm at constant temperature (300 K) for (a) Cell-1 and, (b) Cell-4.

Download Full Size | PDF

We have done the same V_oc analysis for all devices but intensity-dependent V_oc curves for Cell-2 and Cell-3 are not shown here. The value of the coefficients K₁and K₂ are obtained from the fitting of the intensity-dependent V_oc curves of Fig. 3 by using Eq. (9) and Eq. (10), and other recombination coefficients extracted from K₁and K₂ are provided in Table 4. From the data in Table 4, it can be evidenced that the poor PV performance of the Cell-1 is due to high carrier recombination at the SCR and the back-contact region. Replacing the CdS ETL layer of Cell-1 with TiO₂ ETL layer in Cell-2 slightly reduces carrier recombination rate at the SCR (R^d) and the back-contact region (R_i,b) from 2.55×10¹⁶ cm^-2s^-1 and 9.26×10⁸ cm^-2s^-1 to 2.43×10¹⁶ cm^-2s^-1 and 8.69×10⁸ cm^-2s^-1, respectively, results in a small enhancement of PV performance of Cell-2. However, a significant improvement of PV performance of the Cell-3 is observed, as the carrier recombination at the SCR and back-contact region is reduced (see Table 4) by diminishing barrier height for photogenerated holes at back-contact (replacing Mo with Ni). The effect of incorporating the CuAlO₂:Mg BSF layer into CuSbS₂-based solar cells (Cell-4) is obvious because the carrier recombination rate at the SCR and back-contact region is significantly reduced to 2.51×10¹⁴ cm^-2s^-1 and 2.37×10⁴ cm^-2s^-1, respectively, thereby further enhancement of PV performance of the device. The presence of the electric field at the CuSbS₂/CuAlO₂:Mg junction would contribute to increase the hole transportation from CuSbS₂ absorber to CuAlO₂:Mg layer, and can minimize the flow of minority carriers (electrons) toward the the CuSbS₂/CuAlO₂:Mg interface and thereby reducing the carrier recombination rate at the CuSbS₂/CuAlO₂:Mg interface [68]. The electric field at that junction can also improve the electric potential along the absorber, which in turn improve the separation of photo-induced carriers and enhances the collection efficiency at the respective electrodes.

Table 4. Extracted values of K₁ and K₂, and recombination coefficients for CuSbS₂-based solar cells of different configurations. ${\boldsymbol R}_0^{\boldsymbol d}$and${\boldsymbol \; R}_{\boldsymbol o}^{{\boldsymbol i},{\boldsymbol b}}$are calculated at bias voltage of 0.9 V.

View Table | View all tables in this article

3.5 Performance optimization of the CuSbS₂ solar cell

3.5.1 Impact of CuSbS₂ layer thickness and carrier concentration

In this work, the PV performance of the final device with the structure of (Al/ITO/n-TiO₂/p-CuSbS₂/p⁺-CuAlO₂:Mg/Au) is optimized by tuning the thickness and carrier density of the CuSbS₂, ETL, and BSF layers. In this section, the PV response of the device is evaluated by varying carrier concentration and thickness of the CuSbS₂ layer from 10¹⁵ to 10¹⁸ cm⁻³, and 200 nm to 900 nm, respectively, keeping all the other materials parameters for different layers at the baseline values presented in Table 1 and Table 2. The computational results, i.e., the change in PV parameters of the device in response to the variation of the absorber layer’s thickness and concentration, are shown in Fig. 4. It is observed from Fig. 4(a) that except FF, all PV parameter values increase steeply up to 400 nm, stabilized between 400 nm and 500 nm, and then fall off slowly with the CuSbS₂ layer thickness. In particular, the maximal values of V_oc, J_sc, and PCE were obtained (995.4 mV, 32.266 mA/cm², and 22.55%, respectively) with a thickness of only 425 nm, which is mainly due to the strong absorption light in the visible region by the CuSbS₂ layer. However, due to enhanced carrier recombination in the thicker absorber layer, V_oc decreases slowly after 500 nm of the absorber layer thickness. According to Beer-Lambert law, the initial increase in J_scwith absorber thickness is due to more photons being absorbed within the absorber layer of the device. In addition, the initial increase in PCE is due to the fact that when the diffusion length of photogenerated carriers is greater than the absorber layer thickness, most of the carriers arrive at the electrodes before recombination, and therefore, PCE increases with the thickness of the absorber layer. However, for the thick absorber layer, bulk recombination occurs, and hence, PCE declines with the further increase in thickness. FF of the device showing a different trend of decreasing value with the absorber thickness, indicating that the series resistance increases with the increase in thickness. The optimum thickness for the absorber layer can also be estimated in the future by carried out simulation based on first principles (ab initio) study.

 

Fig. 4. PV responses of CuSbS₂-based solar cells as a function of absorber layer (a) thickness and (b) acceptor concentration.

Download Full Size | PDF

In addition to the absorber layer thickness, the PV performance of the device is significantly affected by the acceptor concentration (N_A) in the absorber layer. The p-type CuSbS₂ system has a high density of acceptor defects with low formation energy, which mostly occurs during the synthesis of this materials. To understand the effect of N_A on the PV responses of the device, N_A in the CuSbS₂-absorber layer is varied from 10¹⁵ to 10¹⁸ cm⁻³, with optimized absorber thicknesses (425 nm). The other materials parameters of different layers are kept unchanged. Figure 4(b) provides the changes of PV responses of the devices relative to acceptor density variation of the absorber layer. We observe that V_oc initially increases linearly with N_A up to 5×10¹⁷ cm⁻³, and then falls off sharply when N_A is further increased. As N_A increases, the quasi-Fermi energy level of holes moves downward and hence V_oc increased. Another aspect is that increased built-in potential, which is due to an increase of N_A, enhances photogenerated carrier separation in the SCR of the device, thereby increasing the V_oc. Any further increase in N_A resulted in sharply reduced V_oc, which is due to enhanced carrier recombination at the absorber layer. The J_sc of the device also showed a reduction with an increase in N_A beyond a certain concentration (2×10¹⁶ cm⁻³). The enhanced carrier recombination after critical concentration diminishes the minority carrier lifetime and resulting in the reduced J_sc. This is due to the fact that the Coulomb interactions of the photogenerated electrons with the overpopulated acceptor doping, may lead to more hole’s traps and recombination in the absorber layer [69]. The cumulative effect of V_oc and J_sc causes the PCE of the device to increase with increasing N_A up to 3.16×10¹⁶ cm⁻³. The overall PCE of the device decreases after the N_A concentration of 3.16×10¹⁶ cm⁻³. Therefore, it becomes important to control the acceptor concentration in the absorber layer to operate the device in the maximum PCE regime. The maximum PCE of about 22.78% for CuSbS₂-based solar cell is achieved at the absorber thickness of only 425 nm and acceptor concentration of 3.16×10¹⁶ cm⁻³ (V_oc ≈ 966 mV, J_sc ≈ 34.20 mA/cm², and FF ≈ 68.98%). Consequently, this investigation proposes that CuSbS₂-absorber might be appropriate for ultrathin solar PV devices.

3.5.2 Impact of TiO₂ ETL and CuAlO₂:Mg BSF layer thickness and carrier concentration

In this section, to comprehend the effect of ETL and BSF layer carrier density on PV response of the device, the carrier density of ETL and BSF layer is varied in the range of 10¹⁷ to 10²⁰ cm⁻³ and 10¹⁷ to 10¹⁹ cm⁻³, respectively. The simulation is performed with the optimized absorber layer thickness and carrier density obtained previously, whereas, other material parameters are kept at the baseline values. Figure 5 illustrates the change in PV responses of the device with respect to the ETL and BSF layer carrier density variation. It is worth mentioning that we additionally simulated the impact of ETL and BSF layer thickness, results are not shown here, and found that the device performance hardly changes with the thickness of ETL and BSF layer. In particular, all the PV parameters of the device slightly increase with the thickness of the BSF layer. On the contrary, the increase in ETL layer thickness slowly reduces the device performance. So, the thickness of 75 nm and 200 nm are chosen for ETL and BSF layer, respectively, to carry on further study. It is seen from Fig. 5(a) that as the ETL layer carrier concentration increases to 5×10¹⁷ cm⁻³, all PV parameters of the device increase sharply at first, then slowly increase and stabilize at a concentration level exceeding 10¹⁹ cm⁻³. However, the PV responses of the device enhance gradually as the BSF layer carrier concentration increases and finally saturates at the concentration level beyond 10¹⁹ cm⁻³. In this simulation, the maximum PV responses of V_oc: 969 mV, J_sc: 34.62 mA/cm², FF: 68.71%, and PCE: 23.05% of the proposed device are achieved at the ETL and BSF layer carrier densities of 5×10¹⁸ cm⁻³ and 3.75×10¹⁸ cm⁻³, respectively.

 

Fig. 5. PV responses of CuSbS₂-based solar cells as a function of (a) TiO₂ ETL, and (b) CuAlO₂:Mg BSF layer carrier concentration.

Download Full Size | PDF

3.6 Effect of defects at TiO₂/CusbS₂ and CuAlO₂:Mg/CuSbS₂ interface on the device performance

The performance of PV devices is also strongly affected by interfacial defects in the heterojunction structure. In this study, the effect of the defect density at TiO₂/CusbS₂ and CuAlO₂:Mg/CuSbS₂ interface on the performance of the proposed CuSbS₂ solar cell is investigated. Figure 6 indicates simulation results for the solar cell PV parameters as a function defect density at the TiO₂/CuSbS₂ and CuAlO₂:Mg/CuSbS₂ interface in the range from 10¹² to 10²⁰ cm^-2. It can be seen that all the performance parameters of the device are declined with increasing the defect concentration at both interfaces. In particular, it is evident from the simulated results, all the PV performance parameters of the device are almost stable when the defect density reaches 10¹⁷ cm^-2 and then gradually decreases as it increases further. This performance drop is due to the increase in the carrier recombination at the interface with increasing the defect density [70,71]. With the increase of interfacial defects, the series resistance of the device also increases significantly [70]. It is worth noting that for the increase of defect density at the TiO₂/CuSbS₂ interface from 10¹² to 10²⁰ cm^-2, the conversion efficiency drops from 23.05% to 21.9%, while for the same increase in defect density at the CuAlO₂:Mg/CuSbS₂ interface, the conversion efficiency drops from 23.05% to 22.46%. Therefore, it is revealed that the performance of the proposed CuSbS₂ solar cell with heterojunction structure is more affected by the defect at TiO₂/CuSbS₂ interface than at the CuAlO₂:Mg/CuSbS₂ interface [72].

 

Fig. 6. PV responses of CuSbS₂-based solar cells as a function of defect density (a) at CuAlO₂:Mg/CuSbS₂, and (b) TiO₂/CuSbS₂ interfaces.

Download Full Size | PDF

3.7 Effect of back-contact metal work function and temperature on the PV response of the optimized CuSbS₂ device

The back-contact potential barrier mainly hampers hole transport at back-contact and it deviates the shape of the illuminated JV characteristics curve near to the V_oc of thin-film solar cells from ideality. It actually limits the current density of the device under illumination and is known as rollover effect. The presence of rollover effect in the J-V characteristics of the conventional CuSbS₂-based solar cell (in Cell-1, Mo is used as back-contact) and Cell-2 (see Fig. 1(b)) imply that a Schottky barrier is developed at the back contact, thus reducing the V_oc of the device. Therefore, we can infer that Mo back contact commonly used in CuSbS₂-based solar cells is not suitable for high-efficiency CuSbS₂-based PV devices and a stable, and low resistance back-contact that is not substantially rectifying is required for high-performance CuSbS₂-based solar cells. To study the effect of changing the work function of the back-contact metal i.e. the ohmic or rectifying behavior at the back-contact/BSF layer interface, the simulation is carried out by varying the work function of the anode material from 4.9 eV to 5.7 eV. Throughout the simulation, the thickness and carrier concentration of different layer has been fixed at the optimized value obtained previously. In Fig. 7(a), the normalized values of PV parameters in response to the variation of metal work function at the back-contact are depicted. It is observed that except J_sc, the values of all PV parameters of the device linearly increases with increasing the back- contact work function to 5.4 eV and then become steady on further increasing work function of anode materials. In contrast, the J_scis relatively less sensitive and slightly increases as the barrier height decreases on increasing back-contact metal work function. For anode materials with a work function of less than 5.4 eV, a rectifying Schottky barrier contact is developed at the anode/CuAlO₂:Mg interface, resulting in poor hole transport to the anode, reducing FF and PCE, as shown in Fig. 7(a). It is known that the energy barrier (Ψ_B) at the anode/CuAlO₂:Mg interface is given by

Here, E_g is the band gap of CuAlO₂:Mg (3.46 eV), χ is the electron affinity of CuAlO₂:Mg (2.50 eV) and φ_m is the anode’s work function. For the anode materials with work function higher than 5.4 eV, according to Eq. (17), the energy barrier is less than 0.5 eV, which may not affect the V_oc of the device [73], thereby, achieving the maximum performance of the device. This finding indicates that anode materials with relatively high work functions (> 5.4 eV) are required to achieve the maximum efficiency of CuSbS₂-based PV devices.

 

Fig. 7. PV responses of CuSbS₂-based solar cells as a function of (a) work function of the back-contact materials, and (b) working temperature.

Download Full Size | PDF

Material properties and PV performance of solar cells are greatly affected by the operating temperature. Generally, solar cells are installed outdoors and exposed to a temperature range of 288 K to 323 K [74], and even higher temperatures in some other applications. The reverse saturation current density (J₀) is known to be the key parameter, which is mostly affected by the temperature and band gap of the materials, affecting the PCE of the PV devices [75]. The effect of temperature on the CuSbS₂-based solar cell performances is presented in Fig. 7(b). In Fig. 7(b), normalized PV responses of the device are plotted in line with the change in the operating temperature. As expected, the nearly linear degradation in the V_oc and FF accompanied by a linear degradation in the PCE is observed on increasing the temperature. The fall in V_oc allows electrons to gain enough energy at higher temperatures and recombine with holes before entering into the depletion region and collected at electrode [76]. A slight increase in Jsc is observed, which is due to the decrease of the band gap of semiconductor with increasing temperature. The temperature coefficient C_T ((%K⁻¹) of PCE, is an indicator of PV devices, showing how higher operating temperature affects the device’s PEC, and can be expressed under standard test conditions (STC: 298 K) as below [8]:

3.8 Effect of series resistance on the PV response of the optimized device

Both series (R_s) and shunt (R_sh) resistance have a huge impact on the PV performance of heterojunction solar cells, because the appearance and slope of the J-V characteristics curve are jointly controlled by them. R_s is originated from the electrical resistance related to the contact and bulk resistance of the semiconductors (ETL, BSF, and absorber layer), while the shunt resistance mainly comes from several alternative carrier recombination paths i.e. from defects in bulk semiconductors and interfaces, and also depend on the design of the device (cell edges effect) [78]. Consequently, a larger R_s value will reduce the short-circuit current (J_sc), and a smaller R_sh value will cause the (V_oc) to decrease, and as a combined effect, the fill factor drops sharply, resulting in a low PCE. In this section, the role of the R_s and R_sh on the optimized device has been investigated. In this simulation, we have varied R_s from 1 to 10 Ω·cm², while the R_sh was kept constant at 10³ Ω·cm² and varied shunt resistance from 10² to 10⁵ Ω·cm², while keeping R_s constant at 2.5 Ω·cm². The corresponding variations of the PV parameters are shown in Fig. 8. Figure 8(a) shows that as R_s increases, V_oc initially increases slightly and then tends to saturate, while in the case of other PV parameters, it continues to decrease as observed. These findings are completely consistent with the reported literature [79].

 

Fig. 8. PV responses of CuSbS₂-based solar cells as a function of (a) series resistance, and (b) shunt resistance.

Download Full Size | PDF

It has been observed that with every 0.01 Ω increase in R_s, FF degrades by nearly 2.6%, which is consistent with traditional Si (approximately 2.5% for every 0.01 Ω increase in R_s) solar cells. However, it is interesting that although the FF degrades to 2.6%, the efficiency degrades is much lower—only 0.92%, with R_s increases by 0.01 Ω. On the contrary, reducing R_sh below 1000 Ω·cm² will cause all PV parameters to decrease rapidly (Fig. 8(b)). However, PCE and other PV parameters are less affected by R_sh > 1000 Ω·cm².

4. Conclusions

This article unveiled a CuSbS₂-based heterojunction solar cell with non-toxic buffer and BSF layers and analyzed it by SCAPS 1D simulation software. A (n-p-p⁺) heterojunction solar cell structure (Al/ITO/n-TiO₂/p-CuSbS₂/p⁺-CuAlO₂:Mg/Au) has been designed and analyzed numerically. In this study, poor PV performance of CuSbS₂/CdS heterojunction solar cells has been enhanced significantly in the presence of CuAlO₂:Mg BSF layer and Au back-contact. This improvement of PV performance of the considered device is attributed to the reduction in back-contact recombination and enhancement of built-in potential. The CuSbS₂-device structure has been optimized by tuning the thickness and carrier density of different layers to achieve the maximum PCE of the solar cell. The optimized device shows a maximum PCE of 23.05% when the CuSbS₂ absorber layer thickness is only 425 nm, indicating that it is suitable for ultra-thin solar PV devices. The study also suggests that the efficiency of novel CuSbS₂ solar cells can increase by replacing Au with Mo as back contact material. The study again portrays that although the FF of the proposed device decreases by about 2.6% with an increase of 0.01Ω in R_s, its impact on PCE is really small (only 0.92%, with R_s increases by 0.01 Ω). Based on optimization, the highest efficiency of the solar cell reached to 23.05% (V_oc= 969 mV, J_sc= 34.61 mA/cm^2, FF = 68.71%), which is very encouraging compared with traditional CuSbS₂/CdS heterojunction solar cells. Nevertheless, this work successfully demonstrates the low-cost, non-toxic, and earth-abundant CuSbS₂-based ultrathin solar cell as a potential candidate in the future PV industry.