Introduction

The rapid increase in the global population and the widespread use of technology have caused a significant rise in energy demand1. To meet this demand, large amounts of fossil fuels are being used, which leads to serious environmental problems. The dependence on conventional energy sources is depleting the Earth’s limited resources and causing ecological damage. Renewable energy, especially solar energy, offers a cleaner and more sustainable solution. Therefore, it is important to develop efficient methods for capturing and storing solar energy. Recent progress in science and technology has made it possible to use non-toxic materials with high light absorption in semiconductor applications. Crystalline silicon has long been the main material in the photovoltaic (PV) industry but there is growing interest in materials that can maintain a high efficiency while reducing the material usage. Copper zinc tin sulfide (CZTS) is a promising material for making thin-film solar cells (TFSCs) on a large scale. It has a direct bandgap of 1.4–1.6 eV and a high absorption coefficient greater than 10⁴ cm−12. According to previous studies, CZTS-based solar cells have achieved power conversion efficiencies (PCEs) between 6.7% and 8.4%3. Adding a Zn1−xCdₓS buffer layer increased the PCE to 9.2%4, and improved heterojunction designs have reached nearly 11%5.

However, to date, CZTS devices face intrinsic material-related issues that severely affect the charge dynamics. High defect levels arise from Cu/Zn anti-site and stoichiometric deviations during growth, leading to deep-level recombination centers which reduce the minority carrier lifetimes below 10 ns6,7,8,9. Also, the formation of a secondary phase (mainly Cu2SnS3 and ZnS) creates carrier trapping centers10,11 and shunt pathways12, while non-optimal band alignment at the back contact causes severe recombination losses with velocities exceeding 105 cm/s13,14. All these factors contribute to the degradation of the device performance, usually leading to a limited open-circuit voltage (< 730 mV) and a low fill factor (< 70%) in the standard device configurations5,16,17.

Recent advances in interface engineering have identified back-surface field (BSF) layers as a transformative approach to mitigate these limitations18,20. An appropriate BSF architecture could introduce a sufficient conduction band offset to reflect minority carriers maintaining a near-zero valence band offset for unimpeded hole transport and providing chemical passivation of interfacial defects20,21,22. As far as we know, while several prior studies have explored individual BSF materials or specific heterojunction structures, the literature lacks a unified and comparative analysis of multiple BSF candidates under consistent modeling conditions.

This study aims to enhance the PCE of CZTS-based solar cells by incorporating a proper BSF, which improves the carrier collection efficiency and reduces the recombination phenomena at the rear-contact surface. The electrical properties of the proposed solar cell are examined both with and without a BSF. Different sets of simulations, focused on Al: ZnO/ZnO/CdS/CZTS/Mo structures assumed to be grown on Mo substrates, aid in identifying the most effective interfacial layer acting as BSF material at the interface among well-promising candidates drawn from diverse material classes (oxides, sulfides, halides, and organic semiconductors), such as Poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA), Zinc Phosphide (Zn3P2), Tin(II) Sulfide (SnS), Molybdenum Oxide (MoOₓ), Copper(I) Iodide (CuI), Copper Zinc Tin Sulfide (CNTS), Vanadium Pentoxide (V2O5), and Copper(I) Oxide (Cu2O). These materials were selected by considering their appropriate band line-up with CZTS, high carrier mobility, and chemical compatibility23,24,25,26,27. The study provides a comprehensive and systematic numerical comparison of eight different BSF materials integrated into the same cell structure under uniform and realistic simulation conditions, accounting for critical physical parameters like interface defect densities, series and shunt resistances, and performance under elevated temperatures. Beyond evaluating conventional PV parameters, namely the open-circuit voltage (VOC), short-circuit current density (JSC), fill-factor (FF), and power conversion efficiency (PCE), the study examines key physical mechanisms such as energy band alignment (CBO/VBO), carrier generation and recombination dynamics, defect tolerance thresholds, and contact work-function sensitivity.

The manuscript is structured as follows. Section 1 (Introduction) describes the main physical properties of CZTS needed in the design of optimized devices and shows some previous results on this topic. Section 2 (Methodology) describes the simulation process used in the SCAPS-1D simulator. Section 3 presents the simulation results and their analysis, highlighting the role of the different design parameters and the inclusion of a BSF layer in determining the device performance. The last section (Conclusion) summarises the results, evaluates their relevance, and identifies potential areas for future research.

Modeling framework

Device structure

In this study, we adopted a reference solar cell structure inspired by the experimental research conducted by Shin et al.3, implementing a series of modifications in order to enhance the device conversion efficiency. In particular, while the reference cell involves a thin MoS2 layer at the CZTS/Mo interface as a result of the post-deposition annealing, in the proposed device we have introduced a specific BSF layer positioned between the rear contact and the CZTS absorber region as shown in Fig. 1a. In this context, the choice of the BSF material between PTAA, Zn3P2, SnS, MoOx, CNTS, CuI, V2O5, and Cu2O, and the analysis of the role played by different physical parameters of the device structure, represent the focal points of our investigation which is addressed to improve the cell performance. The energy level of materials in Fig. 1b illustrates the structure of Al: ZnO/ZnO/CdS/CZTS with various BSFs. It is evident that a sufficiently small energy barrier at the BSF/CZTS interface facilitates the easy transfer of holes from the CZTS absorber to the back electrode. This configuration may minimize recombination, as the inherent electric field at the BSF/CZTS junction effectively hinders electron flow towards the back metal contact.

Fig. 1
figure 1

(a) Schematic view of the proposed CZTS solar cell structure and (b) Energy level of materials.

Full size image

Simulation setup and parameters

The development of a reliable solar cell with an improved conversion efficiency through a deep understanding of its operational principles requires a systematic numerical strategy that reduces manufacturing costs and shortens experimental timelines. In this work, we used the 1D Solar Cell Capacitance Simulator (SCAPS) software by Gent University, which accounts for the different layer thicknesses, carrier concentrations, defect densities, and metal work functions28,29. As well known, the device behavior is investigated by solving the fundamental Poisson’s equation and the continuity equations for electrons (n) and holes (p) currents30:

$$\:\frac{\partial\:}{\partial\:x}\left({\epsilon\:}_{0}{\epsilon\:}_{R}\frac{\delta\:\psi\:}{\delta\:x}\right)=-q\left(p-n+{N}_{d}^{+}-{N}_{A}^{-}+\frac{{\rho\:}_{def}}{q}\right)$$
(1)
$$\:-\frac{\partial\:{J}_{n}}{\partial\:x}-{R}_{n}+G=\frac{\partial\:n}{\partial\:t}$$
(2)
$$\:-\frac{\partial\:{J}_{p}}{\partial\:x}-{R}_{p}+G=\frac{\partial\:p}{\partial\:t}$$
(3)

here ѱ is the electrostatic potential, εR is the relative permittivity of the semiconductor, ε0 is the vacuum permittivity, ρdef represents the defect charge density, NA- and ND+ are the ionized acceptors and donors concentrations, Jp and Jn are the current densities for holes and electrons, and G and R stand for the photo-generation rate and the net recombination rate, respectively. In addition, the transport of charge carriers is elucidated through the interplay of diffusion and drift processes modeled by the following equations:

$$\:{J}_{n}=-\frac{{\mu\:}_{n}n}{q}\:\frac{\partial\:{E}_{{F}_{n}}}{\partial\:x}$$
(4)
$$\:{J}_{p}=-\frac{{\mu\:}_{p}p}{q}\:\frac{\partial\:{E}_{{F}_{p}}}{\partial\:x}$$
(5)

where µn, p are the carrier mobilities, and EFn, p are the quasi-Fermi levels for electrons and holes, respectively.

Starting from the current-voltage and power-voltage characteristics of the device, we are able to calculate the main solar cell PV parameters, namely the fill factor (FF) and the power conversion efficiency (PCE):

$$\:FF=\frac{{P}_{max}}{{P}_{t}}=\frac{{V}_{max}{I}_{max}}{{V}_{oc}{J}_{sc}}$$
(6)
$$\:PCE=\frac{{V}_{oc\:}{J}_{sc}\:FF}{{P}_{in}}$$
(7)

As we can see, the FF is an expression of the ratio between the maximum power output (Pmax) and the theoretical power value (Pt) calculated under the open-circuit voltage and short-circuit current conditions.

Referring to Fig. 1, the different simulation parameters used in this study are summarized in Tables 1, 2, 3 and 4. They were fixed as a combination of well-documented data3,17,23,24,25,26,27,31,32,33,34,35,36,37,38, theoretical considerations, and judicious estimations. During the simulations, an operational temperature of 300 K and a standard AM1.5G worldwide sun irradiation (1000 W/m2) were assumed by default.

Table 1 Material parameters used in the simulations.
Full size table
Table 2 Back and front contact reference parameters.
Full size table
Table 3 Electrical properties for different BSF layers in the proposed solar cell.
Full size table
Table 4 Defect parameters at both czts/cds and BSF/CZTS interfaces.
Full size table

Prediction capabilities of the simulation setup

To compare and validate our calculations, we simulated the experimental results of our reference cell3 by means of the SCAPS algorithm. Figure 2 illustrates both the simulated and experimental J–V curves, as well as the quantum efficiency (QE) response of the reference CZTS solar cell. The good agreement between the experimental and simulated data supports the accuracy of the SCAPS-1D modeling approach. The QE curve highlights a strong spectral response, confirming efficient photogenerated carrier collection and validating the chosen simulation parameters.

Fig. 2
figure 2

Comparison between simulated and experimental results: (a) J–V characteristics and (b) quantum efficiency (QE) of the reference CZTS solar cell3.

Full size image

Notably, the agreement between the experimental and simulated J–V curves was achieved by using an acceptor density NA of 1016 cm−3 and a bulk defect density of 1012 cm−3, which were calibrated on the experimental data reported by Shin et al.3. The correspondent minority carrier diffusion length is approximately 350 nm. This value is also in good agreement with the modeling results of Frisk et al.32 for high-quality CZTS absorbers. In addition, the series resistance of 4.5 Ω·cm2 was directly extracted from Shin et al.3, while the shunt resistance of 470 Ω·cm2 was fine-tuned to reproduce the experimental fill factor and the overall device performance under realistic leakage conditions. The calculated PV parameters JSC, VOC, FF, and PCE are 0.661 V, 19.5 mA·cm−2, 65.8%, and 8.4% from the experiments, and 0.6604 V, 19.4487 mA·cm−2, 65.39%, and 8.4% from the simulations, respectively. Although slight numerical differences appear, certainly due to the difficulty in determining accurately the true thicknesses as well as the specific shunt resistance of the device under test, the obtained results appear really consistent with each other, leading to the same PCE value.

Results and discussion

In the following, we present our simulation analysis aimed to optimize the PV performance of the solar cell investigated in3, by means of an exhaustive and rigorous approach applied to the device structure in Fig. 1.

Exploring the impact of the thickness and acceptor concentration of the CZTS and MoS2 layers

It is well known that the absorber layer’s physical parameters, e.g., carrier concentration and thickness, play a key role in enhancing the conversion efficiency of a solar cell. In fact, an excessively thick absorber can originate recombination currents into the substrate, whereas an overly thin one may lead to poor light absorption. Consequently, the design of the absorber layer is a critical aspect in determining a significant enhancement or a detrimental effect on the solar cell performance.

The contour plot in Fig. 3 shows the interplay between the CZTS absorber thickness and the MoS2 layer thickness, highlighting how these geometrical parameters affect the performance of the proposed device with a PCE value that can span from 5.84 to 10.83%. Notably, the data reveals that an absorber thickness wider than 0.9 μm improves the device’s performance. Our simulation results show that the PCE begins to surpass the value of the reference cell (8.4%) when the CZTS absorber thickness exceeds approximately 0.9 μm. Below this threshold, insufficient light absorption limits the photocurrent and leads to reduced PCE values. Therefore, CZTS thicknesses below 0.9 μm should be avoided due to their detrimental impact on device performance. Within the 0.9–2.2 μm range, a continuous improvement is observed, although the gains become more gradual beyond 1.8 μm due to increased bulk recombination effects. In contrast, the MoS₂ layer thickness has only a slight influence on device output when it remains below 0.3 μm, with the PCE stabilizing between 10.81% and 10.83%, and reaching its maximum at a thickness of 0.1 μm.

Fig. 3
figure 3

Dependence of the solar cell PV parameters (VOC, JSC, FF, PCE) on the thickness of the CZTS region and MoS2 layer.

Full size image

Taking into account the parasitic resistance of the device, in a successive analysis we have used a sum modulation approach to examine how the carrier concentrations in the CZTS and MoS2 regions impact the device PV parameters. As shown in Fig. 4, maximum conversion efficiency of 12.82%, with VOC = 0.755 V, JSC = 25.95 mA/cm², and FF = 65.44%, are calculated for a carrier concentration of 1016 cm−3 and 1019 cm−3 in the CZTS and MoS2 regions, respectively. These results are consistent with the literature data reported in5.

Fig. 4
figure 4

PV parameters as a function of the doping concentration in the MoS2 and CZTS regions.

Full size image

Investigation of alternative BSF

In order to show the importance of the BSF layer in improving the solar cell performance, both in terms of separation and collection mechanisms, the resulting J-V characteristics calculated assuming alternative BSF materials (PTAA, Zn3P2, SnS, MoOx, CuI, CNTS, V2O5, and Cu2O) are reported in Fig. 5. In particular, we observed a significant improvement in the PCE escalating from 12.82% in the CZTS solar cell with a MoS2 layer (see Sect. 3.1) to 18.64%, 18.58%, 17.55%, 17.39%,16.88%, 15.73%, 15.44%, and 14.09%, with the use Cu2O, V2O5, CuI, CNTS, MoOx, SnS, Zn3P2, and PTAA respectively. For a more exhaustive analysis, Table 5 compares different output parameters for various BSF configurations integrated with the CZTS absorber.

Fig. 5
figure 5

J-V curves (simulated) for different structures of the investigated CZTS solar cells.

Full size image
Table 5 Output parameters for various BSF configurations integrated with the CZTS absorber.
Full size table
$$\:CBO=\:{\:{\upchi\:}}_{abs}-{\:{\upchi\:}}_{BSF}\:$$
(8)
$$\:VBO=-CBO+\left(\:{\:E}_{g,BSF}-{\:E}_{g,abs}\right)$$
(9)

where χ is the electron affinity and Eg is the bandgap energy of the relevant layer.

When VBO becomes negative, the BSF valence band lies below that of the absorber, creating a “cliff” that eases hole transport but also lowers the recombination barrier. When VBO is positive, the band edge forms a “spike” that can hinder hole extraction.

As well known, an improper band alignment at the interface between the BSF and the absorber layer can significantly impact the carrier transport and recombination dynamics in thin-film solar cells. In more detail, a large negative VBO can facilitate the hole flow but it reduces the activation energy for carrier recombination affecting, therefore, the device performance. On the other hand, a large positive VBO can impede hole collection, leading to charge accumulation and higher recombination losses. A careful tuning of VBO is crucial for an optimized device performance24. Table 6 presents the values of χ, Eg, CBO, and VBO for the BSF candidates that were explored in this study.

Table 6 Energy band alignment parameters of BSF materials with the CZTS absorber.
Full size table

Based on Tables 5 and 6, the optimal device performance is obtained when the back-surface field (BSF) layer has a conduction band offset > 0.3 eV and a valence band offset in the range from − 0.28 eV to + 0.05 eV. These results are in good agreement with previously published studies27,39,40,41,42. Accordingly, Cu2O, V2O5, CNTS, and CuI gain interest as they exhibit excellent CBO (0.8–2.1 eV) and VBO in the ideal range. Cu2O, especially, exhibits a high CBO (1.0 eV) in conjunction with a moderately negative VBO (−0.28 eV), bolstering its position as an ultra-selective BSF layer. In a similar manner, V2O5 (CBO = 0.8 eV, VBO = −0.05 eV) and CNTS (CBO = 0.33 eV, VBO = −0.04 eV) have band offsets favorably aligned for efficient separation of carriers. In contrast, materials with deeper valence band positions, such as PTAA (VBO = −0.39 eV) and MoOx (VBO = −0.15 eV), may limit the hole extraction (unless compensated by other interfacial effects). Although SnS and Zn3P2 display suitable VBOs (−0.15 eV and + 0.05 eV, respectively), their CBOs of zero hinder them from performing an effective electron reflection with additional recombination losses in the device structure.

The presented theoretical band alignment analysis was further corroborated by SCAPS-1D simulations of CZTS solar cells containing the investigated BSF materials. Energy band diagrams in Fig. 6 further confirm the selective nature of the resulting interfaces. Interestingly, the optimal BSFs show suitable band bending close to the back contact, suggesting the promotion of hole extraction with recombination suppression.

Fig. 6
figure 6

Energy band diagram for various BSF configurations integrated with the CZTS absorber.

Full size image

Impact of the defect density in the absorber CZTS region for various BSFs

The defect density Nt in the CZTS absorber significantly influences the device performance by increasing carrier recombination. In this section, combining the use of BSF layers in the device structure, we assumed a defect density in the CZTS absorber ranging from 1010 to 1018 cm−3.

The results shown in Fig. 7 demonstrate a consistent decrease in the photovoltaic parameters as the defect density increases. In particular, PCE gradually stays constant up to 1014 cm−3 but it decreases substantially afterwards due to the raised recombination. The best performance in the low-defect regime is achieved by Cu2O, V2O5, and CuI. At 1017cm−3, the PCE of Cu2O drops from 18.64% to 10.24%. PTAA offers an added advantage by retaining 70% of its initial efficiency, indicating a better defect tolerance. These results suggest that in order to minimize the defect density and to ensure optimal device performance, a value less than 1014 cm3 is needed. For the sake of simplicity, we assumed an absorber region defect density of 1014 cm−3 in the successive calculations.

Fig. 7
figure 7

Photovoltaic performance parameters as a function of the defect density in the absorber region.

Full size image

Influences of interface defects on PV performance

Defects at the bsf/absorber interface

This section analyzes the effect of interfacial defect density (Nt) at the BSF/CZTS interface from 108 to 1018 cm⁻2. Figure 8 presents the simulated performance, which unequivocally illustrates a gradual deterioration in the photovoltaic output as Nt expands, whereby each BSF material manifests its own sensitivity curve.

At low values (Nt ≤ 1011 cm−2), the majority of BSF materials maintain a high and large performance stability. These materials were used to produce impressive device performance up to 18.5% (PCE), 0.982 V (Voc), 26.37 mA/cm2 (Jsc), and FF > 71% with notable examples such as Cu2O, V2O5, and CuI. The exceptionally low recombination losses were ascribed to the step-graded, optimal energy band alignment and the high carrier selectivity at the BSF/CZTS interface43.

Fig. 8
figure 8

Effect of the defect density at the BSF/CZTS interface on the photovoltaic performance of CZTS solar cells with different BSF materials.

Full size image

When Nt is further increased from 1012 cm−2, a gradual performance degradation is observed. Cu2O maintains the best durability so far, with only a slight efficiency loss from 18.64 to 15.63% at Nt =1018 cm−2, with relatively high values for Voc (0.943 V) and FF (69.84%). The same trend is observed in the case of V2O5, where the efficiency drops to 16.42%, affirming a good defect tolerance. In comparison, CuI is more sensitive to the interfacial defects, where the FF decreases from 71.48 to 69.62% and PCE decreases to 15.44%, which suggests higher interfacial recombination phenomena44,45.

Under defect stress, CNTS, Zn3P2, and MoOx demonstrate intermediate behaviors. CNTS maintains almost constant JSC but shows a slow decrease in Voc and PCE, which reaches 16.06% at high Nt. The Voc of Zn3P2 is a little bit unchanged (from 0.905 V to 0.895 V) but the PCE is reduced from 15.44 to 14.92%. MoOx shows the higher degradation in defects impact with the PCE dropped from 16.88 to 15.61% and the FF declining from 66.82 to 65.96%, implying the more proven recombination and the lower interfacial selectivity. Conversely, SnS exhibits significant consistency in Jsc (23.84 mA/cm²) and Voc (0.944 V) but displays a little fluctuation in the PCE, exhibiting therefore a limited response to the interfacial variations. Finally, PTAA shows a wide defect tolerance range, where a slight decrease in efficiency from 17.12 to 14.07% is still accompanied by good Voc and FF values, making it a strong candidate under defect stress conditions.

These results highlight the need for a proper selection of the BSF material, not only for beneficial energy band alignment but also for establishing chemically stable and low-defect interfaces with CZTS.

Defects at the absorber/buffer interface

The absorber/buffer interface is essential for defining the PV performance of thin-film solar cells as it affects the carrier recombination dynamics. In this section, we explore how the performance of CZTS-based devices, which use the different BSF layers taken into account, degrades their performance with an increased defect density at the CZTS/CdS interface. In particular, the interface defect density was changed from 108 to 1018 cm−2, and the key PV parameters (PCE, FF, JSC, and VOC) were extracted for each case.

As shown in Fig. 9, a near-constant performance for low defect levels (Nt ≤ 1015 cm−2) demonstrates that minor interfacial trap densities cause almost no impact on all configurations. However, after this limit, performance metrics start to decrease (most significantly the PCE and VOC values) as a consequence of Shockley–Read–Hall recombination mechanisms. Overall, Cu2O, V2O5, and CuI all showed a robust tolerance for increasing interface trap densities across all investigated BSF structures. Even at the highest defect level (1018 cm−2), Cu2O still retained an 18.73% PCE, which was very closely followed by the performance of V2O5 (18.67%) and CuI (18.60%). This resilience relates to an optimal energy band alignment leading to low recombination losses and high VOC values (above 0.98 V). In comparison, the performance of the PTAA, Zn3P2, and MoOx devices decreased more sharply. The PCE of PTAA decreased from 20.24 to 17.20% and that of Zn3P2 from 17.65 to 15.27%, demonstrating their incapability in resisting interfacial degradation. MoOx, while also performing well when defects are at low levels, suffers a PCE of 16.95% at 1018 cm−2, while the FF drops from 68.03 to 66.83%. CNTS and SnS showed intermediate behaviors. CNTS had a more stable JSC but the trend was down for FF and VOC, finishing with a PCE of 17.46%. In comparison, SnS showed the highest stability with respect to JSC (25.5 mA/cm²) but the lowest VOC (0.94 V), which constrained its overall efficiency improvement under defect stress. Accordingly, the interfacial defect density is taken to be 1012 cm−2 for the CZTS/CdS interface.

These results highlight the significance of the absorber/buffer interface engineering in the performance of CZTS solar cells. But while highly performant BSF materials are important, they can only deliver efficiency gains if interfacial defects are suppressed; if not, efficiency improvements can be compromised. High-temperature processing, robust native oxides, and interface passivation or buffer layer optimization (i.e., alternatives to CdS) should be considered for long-term device stability and performance enhancements.

Fig. 9
figure 9

Effect of the CZTS/CdS interface defect density on the photovoltaic performance of different solar cells with a BSF layer.

Full size image

Effect of generation rate and recombination rate

The generation and recombination rates are the most significant factors affecting the overall device performance, especially the power conversion efficiency of thin-film CZTS solar cells. The generated number of electron-hole pairs per unit volume and per unit time at a given wavelength λ and position x is related as follows46:

$$\:\text{G}\left({\uplambda\:},\text{x}\right)={\upalpha\:}\left({\uplambda\:},\text{x}\right).\:{N}_{phot}\left({\uplambda\:},\text{x}\right)$$
(10)

with α(λ,x) and Nphot(λ,x) denoting the absorption coefficient and the incident photon flux, respectively. Both quantities are functions of the wavelength and depth within the solar cell.

As shown in Fig. 10a and b, the generation and recombination rates at the BSF/CZTS interface are strongly dependent on the selected BSF material. Among the investigated materials, SnS showed the maximum carrier generation rate of ∼9.99 × 1020 cm−3s−1, which was attributed to the strong photon absorption of SnS and the efficient generation of electron-hole pairs47. This advantage, however, was offset by an incredibly large recombination rate of 2.40 × 1021 cm−3s−1, likely due to a poor band alignment which caused a significant carrier loss at the interface that lowered the PCE to 17.06%. Conversely, MoOₓ exhibited an exceptionally low recombination rate (~ 9.25 × 105 cm−3s−1) despite a modest production rate (~ 1.05 × 1011 cm−3s−1). This led to improved interface quality and increased carrier collecting efficiency, resulting in an overall PCE of 19.48%.

Fig. 10
figure 10

Effect of the (a) generation and (b) recombination rates at the BSF/CZTS interface for different BSFs.

Full size image

Cu2O and V2O5 achieved a well-balanced situation with low recombination rates (1.32 × 1012 cm−3s−1 and 2.49 × 1013 cm−3s−1) and appropriate generation rates (1.64 × 1017 cm−3s−1 and 1.30 × 1017cm⁻³s⁻¹), resulting in the highest efficiencies of 21.41% and 21.34%, respectively. These results are due to their ideal valence band offset (VBO) and conduction band offset (CBO) greatly disconnecting carrier recombination at the absorber interface. Intermediate BSF materials like Zn3P2 (17.65%), CNTS (20.02%), PTAA (20.24%), and CuI (21.31%) also showed a balance between generation and recombination rates, indicating partial passivation of interface defects. The generation rate of Zn3P2 is very high (3.97 × 1019 cm−3s−1) and exhibits significant recombination (7.19 × 1018 cm−3s−1). CuI and CNTS showed generation rates of 1.64 × 1013 and 3.81 × 1018 cm−3s−1 with recombination values of 1.36 × 1014 cm3s1 and 3.46 × 108 cm3s1, respectively. Finally, despite the moderate generation seen with PTAA (2.35 × 1014 cm−3s−1), the recombination is relatively higher (1.31 × 1012 cm−3s−1) resulting in a poor carrier collection efficiency.

Impact of various BSFs on the electric field profile

The electric field at the absorber/BSF interface strongly influences charge separation and carrier collection in CZTS solar cells. Although the generation rate is an important performance metric, the spatial electric field profile critically determines the efficiency with which photogenerated carriers are extracted before recombining. Therefore, even BSFs with moderate or low generation can contribute positively to the overall performance if they sustain a sufficiently strong and well-oriented electric field at the CZTS rear interface.

As shown in Fig. 11, the electric field distribution at the BSF/CZTS interface indicates that PTAA and CuI give rise to the most substantial built-in electric fields of the evaluated BSF materials, at roughly 4.25 × 105 V/cm and 7.5 × 104 V/cm, respectively. Such strong electric fields at the BSF/CZTS interface play a crucial role in efficiently driving the photogenerated holes toward the rear contact. This effect enhances the fill factor and compensates, at least partially, for performance losses due to the interfacial recombination. In the case of PTAA, despite its relatively high recombination rate, the strong internal electric field significantly facilitates charge carrier extraction, thus maintaining an acceptable device performance.

Fig. 11
figure 11

Electric field distribution at the BSF/CZTS interface for various BSF materials.

Full size image

In contrast, BSF materials such as MoOₓ, CNTS, SnS, V2O5, and Zn3P2 exhibit considerably weaker-occasionally even negative-electric field values near the interface. Low-field or field-inverted regions are indicative of inversely bent band profiles, which reduce the driving force for the separation and transport of minority carriers, ultimately enhancing the probability of recombination events48. This phenomenon is responsible for the enhanced recombination rates for SnS and Zn3P2 and thus their significantly smaller FF and PCE compared to BSFs that produce stronger internal fields.

Although Cu2O shows the least bilinear electric field, it has a flat and moderate electric field profile in the range of ∼38–78 V/cm. Coupled with their low interfacial recombination characteristics and consistency, they provide balanced device operation and a high fill factor, leading to a competitive power conversion efficiency.

Back contact work function effect

Fig. 12
figure 12

PV parameters as a function of the back contact work function for CZTS solar cells with various BSF layers.

Full size image

The hole extraction and recombination at the CZTS/metal interface are critically governed by the back contact work function. Results of numerical modeling demonstrate that increasing the metal work function from 4.6 eV to 5.6 eV enhances all photovoltaic parameters, in particular VOC and FF, by promoting a transition from Schottky to ohmic behaviors. At low work functions, MoOx and SnS experience significant losses in performance, while Cu2O and CuI show strong output due to a favorable band alignment. Figure 12 shows that the efficiency-dependent regime exhibits saturation for a work function on the order of 5.0 eV in the case of Cu2O, V2O5, and CuI (typically corresponding to a design where PCE is around 26.19%, FF > 88% and VOC is close to 1.0495 V), while for the other configurations including CNTS, Zn3P2, and MoOx higher values of work function are required (~ 5.2 eV). By contrast, SnS displays high sensitivity to contact energetics, whereas PTAA maintains steadiness across the spectrum. These findings show the importance of matching the metal-work function and BSF choice to minimize the rear interface recombination for the best-performing CZTS solar cells.

Impact of the series resistance and shunt resistance

Fig. 13
figure 13

Influence of series resistance (Rs) on the PV performance of CZTS solar cells with different BSF materials.

Full size image

The value of the series resistance (Rs) is an important limiting factor in CZTS solar cells since it restricts the charge extraction and lowers FF and PCE49. In particular, by simulating Rs from 0.1 to 15 Ω·cm², the performance decreased as Rs increased. As shown in Fig. 13, the PCE and FF show a rapid decrease beyond Rs = 1.5 Ω·cm², while VOC is relatively normalized. Following this, the lowest PCEs (∼24–24.3%) and the highest FF (> 81.5%) are achieved with Cu2O, V2O5, and CuI at Rs = 0.1 Ω·cm². In contrast, in the case of SnS and Zn3P2, these parameters are lower and degrade faster (e.g. PCE is lower than 12.02% for SnS at Rs = 15 Ω·cm²). An overall small effect with resistance is shown for JSC, whereas it is evident that the FF collapses. Furthermore, Cu2O, V2O5, and CuI all demonstrate exceptional stability, achieving PCEs considerably above 17% even in the face of high Rs values. Zn3P2 and SnS are the most sensitive, while PTAA, MoOx, and CNTS show moderate performance under resistive stress. These results underscore the necessity to decrease Rs through optimized contacts, conductive BSFs, and interface engineering for the high-efficiency operation of CZTS solar cells.

In this context, in CZTS solar cells, shunt resistance (Rsh) is necessary to reduce leakage. On the other hand, the output power can be penalized if the Rsh value is too low. In fact, as seen in Fig. 14, simulations show that all configurations are in very poor shape with PCE below 2% and FF of 25% at Rsh = 10 Ω·cm2, indicating the presence of excessive shunt pathways. High values of Rsh (e.g., 100 Ω·cm²) give rise to instantaneous jump recovery but the PCE is restrained to 13.67% for SnS and 17.32% for CuI. Above 103 Ω·cm², further improvements are noted with the photoelectrical parameters starting to stabilize.

Fig. 14
figure 14

Influence of shunt resistance (Rsh) on the PV performance of CZTS solar cells with different BSF materials.

Full size image

At Rsh = 105–106 Ω·cm2, peak PCEs of Cu2O, V2O5, and CuI reach 26.19%, with FF > 88% and VOC close to 1.05 V, while SnS remains the least active due to VOC < 0.955 V and a max PCE of 20.85%.

Most benefits are already achieved for JSC across the Rsh range, while VOC and FF are sensitive at low Rsh. These findings reveal the importance of keeping Rsh greater than 104 Ω·cm² to avoid leakage losses and to optimize performance. Interface engineering and the introduction of high-resistivity layers are important to the long-term stability of CZTS solar cells under the condition of high shunt resistance.

Role of the temperature

Fig. 15
figure 15

Temperature effect on the PV performance of CZTS solar cells with various BSF materials.

Full size image

To investigate the influence of the operational temperature on assessing the thermal robustness of the proposed device, a dedicated set of simulations was conducted. Specifically, the responses of VOC, JSC​, FF, and PCE to a temperature ranging from 280 to 420 K are presented in Fig. 15. It can be seen that VOC​, FF, and PCE exhibit a continuous decrease with the increasing value of T. The higher temperature generates more electron-hole pairs, which subsequently enhances the transition rates between the energy bands36. Such an increase in internal recombination leads to an increase in reverse saturation current and thus it decreases VOC. On the other hand, JSC ​ only marginally increases with temperature, which is in agreement with37. This is due to an overall narrowing of the material bandgap. These results of VOC ​ and JSC ​ imply, however, a decrease in FF and PCE.

From Fig. 15, we can observe a PCE reduction from 27.22 to 19.96% for Cu2O-based devices when the temperature spans a relatively wide range from 280 K to 420 K, while SnS-based structures crumble most drastically from 21.86 to 14.34%. These results further support the use of Cu2O and V2O5 as BSFs in CZTS solar cells.

J-V and QE curves

The current-voltage (J–V) and quantum efficiency (QE) characteristics of a CZTS solar cell with and without a BSF material were evaluated for an optimized Al: ZnO/ZnO/CdS/CZTS/BSF/Mo solar cell as shown in Fig. 16a and b. Notably, the devices, including a BSF based on Cu2O, V2O5, and CuI, exhibit drastically improved PV parameters, with PCEs over 26% corresponding to a very high FF (> 88%) and VOC (~ 1.0495 V). These improvements are directly connected to the reduced interfacial recombination​ and the well-adjusted energy band alignment at the CZTS/BSF interface. On the contrary, the reference structure without BSF showed much poorer performance (PCE ≈ 12.82%, VOC = 0.755 V, FF = 65.44%), which suggested that BSF layers are significant in increasing carrier collection and reducing the recombination losses. Specifically, Cu2O exhibited the best performance under electrical bias, demonstrating an efficient hole extraction and effective electron blocking thanks to a favorable band alignment of its conduction and valence bands.

Fig. 16
figure 16

(a) J–V curves and (b) QE curves as a function of the incident wavelength for an optimized Al: ZnO/ZnO/CdS/CZTS/BSF/Mo solar cell, assuming alternative BSF materials.

Full size image

The QE curves display that QE across the entire spectral range from 400 nm to 800 nm remains high (QE > 90%) for all BSF-integrated devices, with peak values approaching 100% around 500–550 nm. The enhanced QE suggests an effective photogenerated carrier production and collection, which correlates well with the increased JSC. The JSC values achieved through Cu2O and V2O5 were found to be the most significant due to the low defects near the interface and the efficient charge separation through the applied field. Finally, CNTS, MoOₓ, and PTAA devices also show high photovoltaic response, with efficiencies in the range of 23.97–24.78%, indicating well-balanced interfacial properties and a good spectral response. Despite the high Jsc achieved with Zn3P2 (28.32 mA/cm²), the resulting FF (75.84%) brings down the ultimate efficiency (22.25%), suggesting recombination losses at the back interface. In comparison, the SnS-based device showed an acceptable visible-light QE but had a steeper drop-off in QE at longer wavelengths, as well as a lower Jsc (25.70 mA/cm2). This behavior indicates more significant recombination at the back interface and shorter carrier diffused lengths, resulting in a lower overall efficiency of 17.27%.

Comparison of SCAPS-1D results with previous works

To evaluate the effectiveness of the BSF materials explored in this study, we compared our simulation results with previously reported experimental and theoretical works on CZTS solar cells. Table 7 summarizes the PV parameters of different structures of CZTS solar cells. The best efficiency of conventional CZTS devices reported to date is 11.01% (experimental), which is certainly limited by back-side recombination and poor power level. The present work demonstrates significant improvements through the incorporation of optimized BSF layers between the CZTS absorber and Mo contact. In particular, from the simulations, the Al: ZnO/ZnO/CdS/CZTS/Cu2O/Mo design showed the best results among all the configurations (26.19% PCE, 1.0495 V Voc, 28.27 mA/cm² JSC, and 88.27% FF). Similar results were also achieved by other BSFs, including V2O5, CuI, CMTS, and PTAA, to achieve PCEs over 24%, indicating that these BSF layers could be promising alternatives to Cu2O, particularly when a suitable back contact is employed as discussed in Sect. 3.7.

Table 7 PV parameters of different CZTS solar cell designs.
Full size table

Conclusion

In this study, we explored a well-referenced experimental CZTS solar cell implementing different design modifications and integrating alternative back surface field layers in the device structure. Specifically, we considered alternative BSF layers in PTAA, Zn3P2, SnS, MoOx, CNTS, CuI, V2O5, and Cu2O. During the simulations, we improved the photovoltaic parameters of the reference cell by analyzing different factors such as the energy band alignments, generation-recombination rates, electric field distribution, electrical parameters, and current-voltage and quantum efficiency characteristics. Also, these analyses took into account the role of the thickness, doping level, and defect density of the absorber layer, as well as the back contact work function, operating temperature, and shunt and series resistances of the device. Our findings show that the careful incorporation of an ultrathin Cu2O BSF layer at the CZTS/Mo heterojunction effectively boosts photon absorption and curbs recombination at the back-contact, generating an open-circuit voltage of 1.0495 V, a fill factor of 88.27%, and a power conversion efficiency of 26.19%. Other BSFs like V2O5, CuI, CMTS, and PTAA achieved efficiencies above 24%, thus affirming that appropriate conduction- and valence-band offsets are critical for blocking electrons and extraction of holes. Performance was found to be robust to absorber defect densities near 10¹⁴cm−3 and interface trap densities up to 10¹¹ cm−2, emphasizing the defect tolerance enabled by an optimized BSF integration. The series resistance was kept to a minimum below 1.5 Ω·cm², and the shunt resistance was above 10⁴ Ω·cm², which aids in maintaining high fill factors. Finally, temperature-dependent studies revealed that Cu2O-based devices exhibit only minimal efficiency losses (≤ 8%) at elevated temperatures up to 420 K. These findings not only underscore the potential for the use of an effective BSF layer but also highlight the importance of strategic material selection and device engineering in the development of high-efficiency thin-film solar cells.