Introduction

There is an ever-increasing need to find sources of energy as an alternative to fossil fuels, due to their constrained availability and potential to produce pollution. Solar energy has long been seen as one of the most valuable resources to fulfill that requirement. Various technologies, including those based on crystalline silicon (c-Si), amorphous silicon (a-Si), perovskite, and inorganic materials such copper indium gallium selenide (CIGS), copper zinc tin sulfide selenium (CZTSSe), tin sulphide (SnS), gallium arsenide (GaAs), cadmium sulphide (CdS) or cadmium telluride (CdTe), have been tested to achieve higher values of power conversion efficiency (PCE), which remains an important measure of the success of a technology to convert photons into electricity. While with some technologies PCE values as high as 47.1% have been achieved using multijunction cells, others have yielded low values, such as 8.7% for organic cells.1 Apart from PCE, other important parameters to gauge suitability for possible deployment in a solar cell application are cost and stability. Among various materials, silicon is the most widely used due to its abundance, reliability and mature fabrication process.2,3 However, owing to relatively low values of the absorption coefficient of silicon, the demand for thick (~ 250 µm) and high-quality wafers leads to an increase in the cost of cells.4

CZTSSe is an excellent alternative to a silicon-based solar cell due to its composition of earth-abundant elements, large absorption coefficient (104 cm−1), inexpensive production process, suitable bandgap, and nontoxic nature.4,5,6,7,8 Also, the energy bandgap of CZTSSe is tunable (can be varied by changing the [(S)/(S+Se)] ratio), due to which it can absorb a large amount of the solar spectrum, with an absorption coefficient of 104 cm−1.9,10,11,12,13 However, it suffers from very low reported efficiency values of about 12.6%, mainly due to relatively little focus on this technology.1 This value of efficiency is substantially lower than that of other contemporary technologies for the fabrication of solar cells.4,14,15 Yang et al. reported a CZTSSe-based solar cell with efficiency of 12.3%,9 while Wang et al. reported a 12.6% efficient solar cell with CZTSSe as an active layer.16 After a thorough literature survey, it was established that the main reason for the low PCE of CZTSSe material is back surface recombination.14,17,18 This recombination can be reduced by preventing carrier recombination in the backside of the device. This can be achieved by creating a barrier for electrons, i.e., by shifting the bands in an upward direction at the back-contact side. For this purpose, researchers introduced a back surface field (BSF) layer. Kumar et al. simulated a CZTSSe-based solar cell with SnS as a BSF layer and reported an increase in efficiency from 12.3% to 15.7% by the introduction of the BSF,17 whereas Omrani et al. reported a maximum efficiency of 17.25% by the introduction of a SnS BSF in the CZTSSe-based solar cell in their simulation work.14 Other materials have also been used as a BSF, including WSe219 and Earth-abundant Zn3P220 .

However, a detailed collective investigation of calibration, CZTSSe doping optimization in SnS and Sn2S3 BSF layer-based devices, BSF doping optimization, and the impact of interface defect density on CZTSSe/SnS and CZTSSe/Sn2S3 layer interfaces has not been considered in any of the previous literature. These investigations are very important in understanding the carrier dynamics inside the device and the alignment of the energy bands for a better collection of light-generated charge carriers. Therefore, the above-mentioned investigations are carried out in this work to fill the research gap and to make a humble contribution in the field of CZTSSe-based solar cells. Results reported in this work would help in fine tuning the performance of fabrication-feasible CZTSSe solar cells for higher conversion efficiencies. Initially, an appropriate calibration of the device was done in accordance with previously published experimental work9 before actually proposing a device with SnS- and Sn2S3-based BSF layers to mitigate the back surface recombination losses and enhance the conversion efficiency. The SnS is a low-cost thin-film Earth-abundant material having a larger absorption coefficient and non-toxic nature4,21 and Sn2S3 is another thin-film material that is economical, with low electrical resistance and high carrier concentration.22 A comparative study of two BSF layers is also performed where SnS and Sn2S3 materials are used as BSF layers because of their similar material composition as CZTSSe.17 The results reported in this study showed a pathway to realize an efficiency of 16.54% (with SnS BSF) and 17.04% (with Sn2S3 BSF) in CZTSSe-based solar cells. The optimization of the device design and fabrication process-related parameters would be costly using an experimental approach. Therefore, technology computer-aided design (TCAD) simulation results supported by a good understanding of physics-based phenomena can help in achieving the objective, i.e., optimizing the device design for better performance. All the reported investigations are performed using a SCAPS-1D-based device simulation, which is widely adopted by the scientific community23,24,25,26,27,28,29 and the obtained results are very accurate in perceiving the behavior of a fabricated device, which is already validated in some previous literature, where authors performed both experimental as well as simulation studies.30,31,32

This paper is organized into four sections: the ongoing Introduction section is followed by the device analysis and simulation section, where detailed information of adopted simulation methodology, including device structures and material parameters considered in simulations, are provided. Subsequently, the Results and Discussion section is devoted to the comprehensive analysis of the effect of introducing the BSF layer and a comparison of two different BSF layers on the performance of CZTSSe solar cells. The doping variation of the absorber layer and BSF layer and interface defect variation of the CZTSSe/BSF layers is also discussed. Finally, the work is concluded in the last section, with a complete summary of the work presented.

Device Analysis and Simulation

A SCAPS-1D simulator is used in this work to study the optoelectronic performance of the proposed solar cell. SCAPS is an open-source simulator developed at the Department of Electronics and Information System, University of Gent, Belgium.33,34 First, the simulation of the CZTSSe-based solar cell reported in the literature9 is calibrated. After that, the BSF layer, i.e., SnS and Sn2S3, is introduced in the device. Thus, three structures of the CZTSSe-based solar cell, i.e., without the BSF (named as D0), with the SnS BSF (named as D1) and with the Sn2S3 BSF (named as D2), are analyzed. The configuration of the first calibrated device is shown in Fig. 1a, i.e., ZnO/i-ZnO/CdS/CZTSSe, which is the same as that fabricated by Yang et al.9 Thereafter, the CZTSSe-based solar cell with a BSF layer, i.e., with the SnS BSF (ZnO/i-ZnO/CdS/CZTSSe/SnS) and with the Sn2S3 BSF (ZnO/i-ZnO/CdS/CZTSSe/Sn2S3), are designed as shown in Fig. 1b. The electrical properties of each layer used in this work are listed in Table I.4,6,14,35 The BSF layer should match the bandgap of the CZTSSe to prevent minority carrier movement toward the BSF layer.14 Therefore, a p+ SnS or Sn2S3 material is used as a BSF layer (see Fig. 1b). Individual energy band diagrams for the BSF layers are also shown in Fig. 1c along with CZTSSe to understand the band bending under thermal equilibrium. A detailed discussion of the same is provided in the results section. The optical properties, i.e., the absorption coefficient of the material used in this simulation, are shown in Fig. 2. The CdS (Eg=2.4 eV) layer shows a higher absorption coefficient compared to CZTSSe (Eg=1.1 eV), particularly for lower wavelengths, i.e., below 517 nm, since the cut-off wavelength for CdS, λg (µm) = 1.24/2.4 eV is 0.517 µm or 517 nm, and thereafter CdS is transparent and allows the coupling of photons to the underlying CZTSSe layer. In addition, the net absorption in a particular layer is governed by both the absorption coefficient and the thickness of the layer (Beer-Lambert law).36 The authors utilized only a 50-nm-thick CdS layer compared to 1800-nm-thick CZTSSe (active) layer, and a thin layer of CdS (50 nm) that is almost transparent; therefore, parasitic absorption in CdS is negligible, and the probability of photon coupling to the CZTSSe layer would be high. The descriptions of interface defects at different interfaces are already discussed in our previous work.18 The complete details regarding the fundamental equation used by the SCAPS-1D to model the device, including the recombination mechanism, is provided in our previous works,3,28 and reiteration of the same is avoided in this work.

Fig. 1
figure 1

Device structures of the CZTSSe-based solar cell (a) without BSF, (b) with SnS or Sn2S3 BSF. (c) Energy band diagram of CZTSSe, SnS, and Sn2S3.

Full size image
Table I Electrical parameters used in simulation
Full size table
Fig. 2
figure 2

The wavelength-dependent absorption coefficient of the materials (ZnO, CdS, SnS, CZTSSe) is used in the simulation. Data derived from references.4,6,14,37,38,39

Full size image

Results and Discussion

The Carrier Dynamics in Three Different Structures

This section is devoted to understanding the carrier dynamics in CZTSSe solar cells having two different BSF layers, i.e., SnS and Sn2S3 , and comparing them with conventional devices. To analyze the effect of adopting SnS or Sn2S3 as the back surface layer in the proposed CZTSSe-based solar cell structure, individual energy band diagrams of all three materials are shown in Fig. 1c. The electron affinity of CZTSSe, Sn2S3 and SnS as taken from the literature have been clearly marked in the figure as 4.0 eV, 4.26 eV and 3.59 eV, respectively. Additionally, the energy band diagram of the proposed solar cell structure without the BSF, with the SnS BSF and with the Sn2S3 BSF is shown in Fig. 3a, b, and c under dark and Fig. 4a, b, and c under illumination, respectively. These energy band diagrams could be very well appreciated owing to the difference in work function values between CZTSSe and SnS/Sn2S3. With Sn2S3 as the BSF, movement of electrons is bound to happen from the CZTSSe toward the Sn2S3, which leads to a bending of the conduction band of CZTSSe away from the Fermi level close to the interface between two materials. The same can be clearly seen from the magnified image explaining the movement of carriers at the interface in Figure 4c. In the case of the adoption of SnS as the BSF, movement of electrons will happen from SnS toward CZTSSe because of its low work function value. The resulting energy band diagram clearly shows the bending of the conduction band of CZTSSe toward the Fermi level. From Fig. 4b and c, it can be clearly understood that in both cases, the presence of the BSF somehow discourages the movement of electrons generated from the active layer of the solar cell structure toward the BSF, which eventually will lead to higher chances of return of these carriers in the active layer and facilitate their movement toward the top layer.14,17 It can be further appreciated from Fig. 4b and c that, on the way toward the top layer, electrons generated by photons are bound to come across some energy potential kinks close to the interfaces between CZTSSe/CdS, CdS/i-ZnO and i-ZnO/ZnO, but would have enough thermal energy to surpass them and contribute to enhanced efficiency.

Fig. 3
figure 3

Energy band diagram of CZTSSe-based solar cell (a) D0, (b) D1, (c) D2. The data are obtained under thermal equilibrium without illumination and bias voltage.

Full size image
Fig. 4
figure 4

Energy band diagram of CZTSSe-based solar cell (a) D0, (b) D1, (c) D2. The data are obtained under illumination and without bias voltage. The representation of electrodes is avoided, as they are aligned at 0 eV on the y-axis.

Full size image

Further, the J–V curve of all the devices under consideration is depicted in Fig. 5. The experimental data as reported in9 is also reproduced in Fig. 5, as well as in Table II for the fitting process of the calibrated device, i.e., D0, which showed that the simulated results are consistent with experimental data. Further, it can be clearly observed from Fig. 5 that the devices D1 and D2 deliver approximately the same increase in VOC and JSC in comparison with CZTSSe-based solar cell devices, i.e., D0. While the VOC value increased from 0.52 V (for D0) to 0.59 V (for both D1 and D2), values of JSC for D1 and D2 improved from 34.44 mA/cm2 (for D0) to 37.74 mA/cm2 and 37.68 mA/cm2 for D1 and D2 respectively. Similarly, while the device D2 with Sn2S3 as the BSF has a fill factor (FF) value of 76.46%, the same is found to be equal to 73.36% for D1 (with SnS as the BSF) and 69.51% for device D0 (without any BSF). This observed improvement in the FF value of the Sn2S3-BSF-based device is mainly due to the enhanced collection of carriers leading to relatively lower interface resistance at the CZTSSe/Sn2S3 interface. These results are summarized Table II. In totality, the PCE of both devices, D1 and D2, have been found to improve from 12.57% (for D0) to 16.34% and 17.04%, respectively, as brought forward in Table II.

Fig. 5
figure 5

Illuminated current density voltage curve for CZTSSe solar cell without BSF (D0), with SnS BSF (D1), and with Sn3S3 BSF (D2). The data for the experimental device without BSF is obtained from the literature9 to show the consistency of simulated results with experimental data.

Full size image
Table II Photovoltaic parameters of CZTSSe-based solar cell without BSF, with SnS BSF and with Sn2S3 BSF
Full size table

After a detailed study of the devices and adoption of either SnS or Sn2S3 as a BSF layer, the devices are further analyzed on the basis of doping variation of the active layer, i.e., CZTSSe and BSF layer. The results of these analyses are presented in subsections devoted to the impact of CZTSSe doping variation and the impact of BSF Layer doping variation as follows.

Impact of CZTSSe Doping Variation

Values of PV parameters like VOC, JSC, FF and PCE for all the three devices D0, D1 and D2 as obtained from Fig. 7 for acceptor doping concentration varying from 1 × 1015 cm−3 to 1 × 1018 cm−3 in the CZTSSe active layer are presented in Table III. A quick comparison of these results is presented graphically in Fig. 6. As can be clearly seen from Fig. 6a, b, and c, at a high acceptor doping concentration of 1 × 1018 cm−3, while the value of VOC is close to 0.64 V for all the three devices, JSC values are equal to 28.4 mA/cm2, 29.6 mA/cm2 and 29 mA/cm2 for D0, D1 and D2, respectively. Further, energy band diagrams at a minimum (1 × 1015 cm−3) and maximum (1 × 1018 cm−3) doping level of CZTSSe are also obtained to understand the trend in the PV parameters and are reported in Fig. 8a, b, and c. The overall PCE depends collectively on JSC, VOC and FF, where the magnitude of JSC depends on the net photogenerated current density and the probability of the collection of generated carriers. At the same time, the magnitude of VOC is governed by the offset in the quasi-Fermi level and the strength of the electric field across the absorber layer. The remaining parameters, i.e., FF, depend on the overall series resistance of the device, including bulk as well as interfacial resistance. In addition, the strength of the electric field across the absorber layer strongly affects the FF. The trend in the PV parameters as depicted in Figs. 6 and 7 shows that in the case of D0, increasing the doping in CZTSSe increases the built-in potential and corresponding electric field (Fig. 8) as well as offset in the quasi-Fermi level, which resulted in enhancement in VOC and FF. The improvement in FF is also credited to the improved conductivity (less resistivity) in the CZTSSe layer with higher doping density.14,16 However, the opposite trend, i.e., a reduction in JSC, is also observed, as noted in Table III and Figs. 5 and 6 with an increase in doping concentration mainly due to the increased probability of recombination of generated carriers before they eventually manage to reach the end metal contacts of the device. This fact is endorsed by the flattening of energy bands near the vicinity of the CdS/ CZTSSe interface (Fig. 8), leading to the possible absence of any active electric field in the CZTSSe layer after the increase in the doping concentration to 1 × 1018 cm−3. This leads to a reduction in collection probability for the charge carriers generated far away from the collecting interfaces. The reduction in JSC counterbalances the improvement in VOC and FF, and hence the overall conversion efficiency remains almost stagnant until doping at 1 × 1017 cm−3 thereafter, significant improvement in PCE is observed for doping, 1 × 1018 cm−3 when considerable enhancement in VOC and FF mitigated the reduction in JSC.

Table III Impact of doping variations in an active layer on PV parameters
Full size table
Fig. 6
figure 6

Impact of CZTSSe layer doping variation on PV parameters (a) VOC, (b) JSC, (c) FF and (d) PCE of devices D0, D1, D2.

Full size image
Fig. 7
figure 7

Impact of CZTSSe layer doping on J–V characteristics of device (a) D0, (b) D1, (c) D2.

Full size image
Fig. 8
figure 8

Impact of CZTSSe layer acceptor doping on band diagram of device (a) D0, (b) D1, (c) D2.

Full size image

Furthermore, in the case of BSF-layer-based devices, i.e., D1 and D2, the reduction in all the PV parameters such as VOC, FF, and JSC is simulated while increasing the CZTSSe doping from 1 × 1015 cm−3 to 1 × 1017 cm−3. The root of the reduction in JSC is the same as that of the D0 device discussed above; however, reduction in VOC and FF is governed by the presence of the BSF layer, which modulates the overall energy bands of CZTSSe and hence the offset in the quasi-Fermi level as well as the strength of the electric field across the CZTSSe absorber layer. For D1 and D2, the PCE shows a dip at 1 × 1017 cm−3 since all the PV parameters reduce until that point. However, in both D1 and D2, while increasing the doping beyond 1 × 1017 cm−3 reflected an improvement in VOC and FF similar to D0; however, this improvement is not sufficient to mitigate the reduction in JSC. Therefore, overall conversion efficiency remains below the initial conversion efficiency achieved at 1 × 1015 cm−3. This section concludes that improvement in conversion efficiency could be obtained in the case of D0 while increasing the CZTSSe doping from 1 × 1015 cm−3 to 1 × 1018 cm−3, whereas for D1 and D2, the optimum efficiency is observed at the initial doping of 1 × 1015 cm−3.

Effect of variation in values of VOC, JSC and FF with the increase in doping concentration can also be observed from the noted PCE values for D0, D1 and D2. While the PCE value for D1 and D2 decreases from 16.3% and 17.0% to 15.3% and 14.9%, the same is observed to increase from 12.57% to 14.64% for D0. For devices D1 and D2, the increase in the values of VOC and FF is not sufficiently high to compensate for the decrease in the value of JSC. Thus, PCE is reduced. However, in the case of D0, the increase in values of VOC and FF is sufficient to mitigate the reductions related to the fall in JSC, and hence the overall PCE is increased. A reduction in the values of external quantum efficiency (EQE) at wavelengths ranging from 0.55 µm to 1.2 µm is also observed with the change in doping concentration from 1 × 1015 cm−3 to 1 × 1018 cm−3 as shown in Fig. 9 due to reduction in space charge region width. This result is consistent with earlier reported work by Khattak et al.40

Fig. 9
figure 9

Impact of CZTSSe layer doping on EQE of the device (a) D0 (b) D1 (c) D2.

Full size image

Impact of BSF Layer Doping Variation

The previous subsection revealed that D1 and D2 delivered the optimum conversion efficiency of 16.3% and 17.0%, with the lowest examined value of doping concentration (1 × 1015 cm−3) in CZTSSe. Therefore D1 and D2 with CZTSSe doping of 1 × 1015 cm−3 are considered in this subsection for further investigation. In this subsection, a detailed analysis has been carried out to understand the impact of variation in the strength of doping for BSF layers, SnS and Sn2S3. For that, the acceptor doping concentration values for both these BSF layers under consideration are varied from 1 × 1016 cm−3 to 3 × 1018 cm−3, and the impact on all possible PV parameters is studied. The results of the same have been tabulated in Table IV and graphically represented in Fig. 10. Figs. 10 and 11 show that for device D1, as the values of doping concentration are increased, the values of VOC are also increased from 0.50 V to 0.59 V, while the JSC shows negligible variation. Similarly, while for device D2, there is no significant effect of Sn2S3 doping on the JSC and VOC. The reason for this may be found by having a close look at the energy band diagram for the devices D1 and D2 shown in Fig. 12. For device D1, when the doping concentration was as high as 3 × 1018 cm−3, the energy bands at the interface of CZTSSe/SnS are shifted away from the Fermi level, as shown in the inset of Fig. 12a. These prevent more carriers from recombining at the rear side of the device, and thus the VOC value is increased for the higher doping concentration value, i.e., 3 × 1018 cm−3. However, the thickness of the BSF layer is very small (~ 10 nm); thus, the doping of the BSF layer has no significant impact on the JSC. Similarly, the FF values are also observed to increase with the increase in doping concentration from 1 × 1016 cm−3 to 3 × 1018 cm−3 for device D1. The reason behind this may be a reduction in series resistance and barrier height for the holes, as discussed in the subsection devoted to the impact of CZTSSe doping variation.14,16

Table IV Impact of doping variations in BSF layer on PV parameters
Full size table
Fig. 10
figure 10

The impact of BSF layer doping on PV parameters of devices D1 and D2.

Full size image
Fig. 11
figure 11

Impact of BSF layer doping on J–V characteristics of (a) D1 and (b) D2 device.

Full size image
Fig. 12
figure 12

Impact of BSF layer doping on band diagram of (a) D1, (b) D2.

Full size image

On the other hand, for device D2, the energy bands are already (at a lower doping value of 1 × 1016 cm−3) away from the Fermi level to prevent the recombination shown in the inset of Fig. 12b. Thus, for this device, doping variation has little impact on the performance of the device. The values of VOC, JSC, and FF for the device D1 increased with the change in doping of the SnS layer. Thus, the PCE is also increased from 12.2% to 16.3%. However, a marginal increase from 16.3% to 17.0% is observed for device D2.

Impact of CZTSSe/SnS and CZTSSe/Sn2S3 Interface Defects

In this subsection, after optimizing the doping of CZTSSe and BSF layer, i.e., 1 × 1015 cm−3 and 3 × 1018 cm−3, respectively, the effect of CZTSSe/BSF interface defects (from 5 × 1010 cm−2 to 5 × 1013 cm−2) on the device performance is analyzed while keeping all the other parameters constant, as stated in Table I. Further, the impact of interface defect variation on the PV parameters for the device D1 and D2 are shown in Table V. A quick comparison of these results is presented graphically in Fig. 13. Moreover, Fig. 14 depicts the variation of CZTSSe/SnS and CZTSSe/Sn2S3 interface defects on the J–V curve of the device. The results show that the increase in interface defects degrades the device D1, as shown in Fig. 14a. This is due to enhanced carrier recombination at the back surface of the device, and the electric field at the interface is not sufficient to prevent this recombination. Therefore, the VOC and JSC were reduced from 16.34 V to 13.48 V and 37.74 mA/cm2 to 36.52 mA/cm2, respectively, as the interface defects of the device D1 increased from 5 × 1010 cm−2 to 5 × 1013 cm−2, respectively. Whereas the FF decreases from 73.36% to 70.98%, and PCE reduces from 16.34% to 13.48%, as shown in Fig. 13c and d. Whereas for device D2, the interface field is sufficiently large to prevent the motion of minority carrier electrons, and hence negligible performance degradation is observed even at an interface defect density of 5 × 1013 cm−3. Consequently, the PV parameter remains unaltered with interface defects for the Sn2S3 BSF CZTSSe solar cell.

Table V Impact of CZTSSe/BSF interface defects density variations on PV parameters
Full size table
Fig. 13
figure 13

Influence of interface defect from 5 × 1010 cm−2 to 5 × 1013 cm−2 of CZTSSe/BSF on PV parameters of CZTSSe-based solar cells.

Full size image
Fig. 14
figure 14

Influence of interface defects from 5 × 1010 cm−2 to 5 × 1013 cm−2 of CZTSSe/BSF on J–V curve of CZTSSe-based solar cells (a) with SnS BSF and (b) with Sn2S3 BSF.

Full size image

Conclusion

In the present work, numerical simulations are performed to analyze the effect of the back surface field layer on the CZTSSe-based solar cell. Two back surface layers, i.e., SnS and Sn2S3, are examined for CZTSSe-based solar cells, and the results are simultaneously compared with the CZTSSe solar cell without BSF. The results show that the presence of the BSF significantly improves the overall device performance owing to reduced minority carrier recombination at the back surface. Both BSF layers showed similar enhancement in VOC and JSC compared to devices without the BSF layer. However, FF enhancement in the case of Sn2S3 surpasses the performance of SnS-based devices. The device with Sn2S3 and SnS-based BSF showed conversion efficiency of 16.34% and 17.04%, respectively, which is 29.9% and 35.5% higher compared to devices without the BSF layer. Further, the effect of doping variation on the CZTSSe layer and BSF layer is also analyzed. The results showed direct/inverse reliance of VOC/JSC and the doping of CZTSSe. The VOC increased while increasing the doping concentration; however, the JSC decreased for all the reported devices. Additionally, the influence of BSF doping, as well as the CZTSSe/BSF interface defect density, is also studied on the PV performance of the device. The results reported in this manuscript could open a path for the development of CZTSSe solar cells for low-cost and high-efficiency applications.