Analysis of The Eciency In Sb2Se3 Thin-Film Solar Cells Using Alternative Buffer Layers In n-p and n-i-p Structures By Numerical Simulation.

Antimony Sulde (Sb 2 Se 3 ) Solar Cells are considered a promising emerging photovoltaic devices technology. However, the best reported experimental eciency (9.2%) is well below the theoretical limit of 30%. In this research is demonstrated, by numerical simulation, that using different buffer or electron transport layers (ETL) and device structures (n-p or n-i-p) can signicantly increase the solar cell performance. The study is based on two underlying considerations: the use of inorganic materials to facilitate the manufacturing process and the analysis of the simulation parameters that adjust to the experimental conditions in which the cells can be processed. In the n-p structures, the use of single layers and bilayers as ETL was evaluated and the possible mechanism that explain the electrical parameters of the solar cell were discussed. Especial attention was made in the role of interfacial state density and band alignment in the ETL/Sb 2 Se 3 interface. In addition, the n-i-p structure was studied by adding a hole transport layer (HTL). An improvement in open circuit voltage (Voc) is observed compared with n-p structure. Finally, the behavior of Voc and eciency vs thickness of the ETL and Sb 2 Se 3 layers was analyzed. The results show that using alternative ETLs a signicant improve in Voc and eciency could be achieved for n-p and n-i-p structures. After thickness optimization and taking account a moderate interface defect density, values of Voc and eciency higher than 600 mV and 15 % were respectively obtained. following (a) the substitution of AZO for FTO as TCO and of CdS for V 2 O 5 as ETL layer in the n-p structure. In the FTO/V 2 O 5 /Sb 2 Se 3 /Au conguration, an eciency of 10.2% was obtained; (b) adding a second layer in the form of ETL-bilayer, the best combination turned out to be ZnO/V 2 O 5 with an ulterior increment in the eciency of 11.6%. This result must evaluate in terms of the experimental complexity. ZnO can be deposited by various methods. Between then, the pneumatic chemical spray it has proven to be a suitable technique. (c) The analysis of the n-i-p structure conrms the improvement of the electrical parameter that is more dicult to improve in this type of solar cells, that is, the Voc. Values greater than 600 mV were obtained representing a substantial reduction of the Voc decit. (d)The comparative study of the n-p structure in the FTO/V 2 O 5 /Sb 2 Se 3 /Au conguration and the n-i-p in the FTO/V 2 O 5 /Sb 2 Se 3 /Cu 2 O/Au conguration, showed that by optimizing the layers thickness in both structures eciencies above 14 % could be achieved.(e) No signicant increase in eciency was obtained using n-i-p structure compared to n-p ones, that justied the incorporation of additional layer. (f) It was demonstrated that interface recombination can be counteracted by a good band alignment. Finally, it is important to mention that absorber quality is critical if high eciency wants to be reached. The bulk defect density, the ribbons orientation and doping in the Sb 2 Se 3 material require especial attention. These results could be considered as a guide in the processing of planar solar cells of Sb 2 Se 3 and a step in the purpose of maximize the eciency to guarantee their future at industrial level.


Introduction
The antimony chalcogenides family Sb 2 (S 1 − x Sex) 3 has been used as a promising absorber material in planar heterojunction or sensitized solar cells. In particular, the Sb 2 Se 3 compounds has been considered an attractive material in the development of second-generation solar cells in thin-lm technology. It is made up of non-toxic elements with relative abundance, with an adequate bandgap value of 1.12 eV, a high absorption coe cient, and p-type conductivity. It is considered as one-dimensional material with the electron density con ned in ribbons (Sb 4 Se 6 )n, with electrically benign grain boundaries in the [001] direction, particularly favorable in exible devices, due to the high tolerance to deformation. It presents an excellent physicochemical stability and low melting point (611 0 C) with high vapor pressure that allows the synthesis of high-quality lms at relatively low temperatures [1]. Despite these excellent properties, the record e ciencies of 7.5 % [2] and 9.2% [3] for n-i-p and n-p structures respectively, are well below the theoretical limit of 30 %. Several aspects must be considered in order to improve the performance in this type of solar cells. The so-called Voc de cit stands out, which according to the detailed balance principle should be approximately 0.24 V for a bandgap about 1.0 eV. The Voc de cit in the record Sb 2 Se 3 based solar cells is greater than 0.7 V [3], which is worse than those of the CIGS (0.42 V), CdTe (0.59 V) and CZTSSe (0.62 V). The most important factors that affect the Voc in these solar cells are the interfaces and bulk absorber recombination and the low free carrier concentration [4]. On the other hand, the orientation of the ribbons also plays an important role in the solar cell performance. The ribbons growing perpendicular to the substrate decreases grain boundary effects favoring the carrier transport.
In the processing of the Sb 2 Se 3 solar cells different routes have been experimentally proposed to improve the device power conversion: different structures and con gurations, different deposition techniques and the use of different buffer or ETL and HTL layers in the n-p and n-i-p structures [5]. An important aspect to considerer when new materials are proposed is to avoid the use of organic compounds. Although, in solar cells such as perovskites have given good results [6] and the record e ciency for Sb 2 (S 1 − x Sex) 3 based solar cells is obtained using an organic HTL [7]; a poor stability and potential di culties in processing on an industrial scale, may limit the use of these materials in photovoltaic applications. Therefore, the study to introduce inorganic layers in the processing of solar cells will be the goal in this work.
Several experimental reports can be found in literature: Cd 1 − x Zn x S was proposed as an alternative buffer layer to CdS [8]; modifying the band alignment in the Cd 1 − x Zn x S/Sb 2 Se 3 interface and increasing the e ciency. Randomly oriented ZnO showed that induces the growth of Sb 2 Se 3 lm with preferred [221] orientation, which results in fewer defects at the interface improving device e ciency [9]. CdSe has been integrated into Sb 2 Se 3 solar cells replacing the CdS with an increase in the e ciency from 4.16% with CdS to 4.51% with CdSe [10]. In these works, the solar cells were made in n-p structure in substrate or superstrate con guration. In simulation studies, where new materials are studied, the greatest emphasis has been placed on n-i-p type structures and focused especially on HTLs with CdS as ETL. In [11] the authors analyze the impact of different HTL in e ciency and concluding that CuO has the best performance, reporting a nal e ciency after parameter optimization of 16.15 %. On the other hand, Cao et al. [12] shown that an inverted con guration in n-i-p structure and using NiO as HTL could improve the e ciency reporting a value of 24.7 % in ideal conditions, i.e. negligible defect states in the Sb 2 Se 3 layer and interface HTL/Sb 2 Se 3 . A power conversion e ciency of 29.35% was calculated in [13] by the addition of the BaSi 2 as a back-surface layer and CdS as ETL in the Al/FTO/CdS/Sb 2 Se 3 /BaSi 2 /Mo con guration. E ciency values of 20% or higher in this type of solar cells, based on simulation models, should be analyzed with caution. Sometimes the calculations overestimate some parameters of the solar cells that are entered as "optimized parameters". However, these studies allow to evaluate the impact of new layers and structures on solar cells performance.
Recently, the authors have studied the impact on the e ciency of Sb 2 Se 3 solar cells, replacing CdS with Cd 1 − x Zn x S in n-i-p structures [14]. In the simulation process the parameters were carefully selected in order to obtain an approach as realistic as possible. Two con gurations were evaluated: the superstrate AZO/Cd 1 − x Zn x S/Sb 2 Se 3 /HTL and inverted HTL/Sb 2 Se 3 /Cd 1 − x Zn x S/AZO, with Cu 2 O as HTL. The impact of molar composition of the ternary compound together with the ETL and Sb 2 Se 3 parameters layers were investigated. The best result was obtained in the superstrate con guration with nal e ciency of 16% after parameters optimization.
Taking in mind the efforts to boost the power conversion e ciencies of solar cells based on antimony chalcogenide compounds, in this work is presented a numerical simulation of Sb 2 Se 3 solar cells with p-n and n-i-p structures. The emphasis will be centered in the buffer layer or ETL using inorganic materials. A solar cell with CdS as buffer layer in n-p structure is used as reference cell. The CdSe, ZnO, V 2 O 5 , and bilayers formed by some of these compounds are evaluated as alternative candidates to the CdS. Finally adding a layer Cu 2 O as HTL the n-i-p structure is analyzed. The roll of lattice mismatch, interface states, the band alignment, and the thickness of the buffer (or ETL) and absorber layers will be also discussed.

Structures And Device Simulation Parameters
The structures studied in this work are shown in Fig. 1. The numerical simulation of the Sb 2 Se 3 solar cells was performed using SCAPS [15]. Performance parameters are obtained by solving Poisson's equation, electron continuity equation, and continuity equation, as previously described [14]. Table 1 summarize the semiconductor layer parameters and Table 2 those related to the interfaces. Lattice mismatch between two different crystals is an important factor to consider in the solar cell design. The differences in crystallographic structures and lattice constants are the principal cause, but not the only one, in defect formation at the interfaces. The lattice mismatch is expressed as [21]: (1) Where a 2 and a 1 are the lattice parameters of the substrate lm and the lm deposited on the substrate, respectively. The deposited lm is in expansion if f > 0 or in tension if f < 0.
Increasing the absolute value of f determines a higher density of states in the metallurgical interface. The Sb 2 Se 3 has orthorhombic structure with α = 11.7 Å, b = 3.6 Å and c = 11.6Å . The values of lattice mismatch are shown in Table 3. Table 3. Lattice mismatch between the interfaces of the simulated solar cells As can be seen the lowest lattice mismatch is obtained for the V 2 O 5 /Sb 2 Se , which will be analyzed later Results And Discussion 4.1 Effect of different TCOs and different single-layers on n-p device photovoltaic performance.
In the rst step, the FTO and ZnO as TCOs with different single buffer layers were studied. Table 4 shows the electrical parameters of the Sb 2 Se 3 solar cells whose con gurations are shown in Figure 1 a), b), c) and d). In the case of V 2 O 5 the calculations were made considering the indirect and direct interband transitions [19]. According to the results, the following considerations can be made: (a) the conversion e ciency of solar cells is generally improved when the FTO is used as TCO, compared to AZO. The increase in e ciency is due to a slight increase in Jsc, which can be justi ed by the higher bandgap value of the FTO; (b) the device photovoltaic performance is enhanced replacing the CdS with the alternative layers, especially the V 2 O 5 . The improvement is caused by a signi cant increase in the Voc. The Voc is mainly in uenced by two factors: the band alignment and defects densities in the interfaces and the bulk absorber. The bulk defect density in the Sb 2 Se 3 layer was xed in 10 15 cm -3 for all simulations, in order to focus the attention on the factors that depend on the buffer layer selected. Then the discussion will be centered in the band alignment and interface defects density. The results reported in Table 4 were obtained using an interface defect density 10 15 cm -2 in all cases. The idea is to evaluate rst, the roll of band alignment. Following this criterion, in Figure 2 the band diagrams of the four solar cells are presented and the valence and conduction band offset (VBO and CBO respectively) are pointed. It is common to take the spike-like band offset as positive and cliff-like as negative. It was demonstrated in references [22][23][24]  In Figure 3 the electrical parameters versus interface defect density are shown. It can be seen how the Voc in solar cell with CdS and ZnO as buffer layers is more sensible to the increase of interface defects density. Notice that ZnO, have the higher VBO that acts as a barrier for holes diffusion throw the interface, lowering the recombination in the interface and increasing the electric eld in the depletion region. This results in a lower de cit of Voc comparing with CdS ones. Taking account, that this buffer layers have the worst lattice mismatch (see Table 3) therefore special attention would be paid in the interface quality specially in the CdS/Sb 2 Se 3 interface. In the case of CdSe, the high lattice mismatch is compensated with the good band alignment that reduces the impact of the interface recombination, maintaining the Voc almost constant. Due to V 2 O 5 has the best band alignment the effect of interface defects on Voc is almost nulled. Also, this buffer layer has the best lattice mismatch then a low defect density is expected.
The Jsc depends on the bandgap, the interface defects and the thickness of the buffer layer. As can be seen in Figure 3 b) , the bandgap plays the most important role, been the cells with ZnO as buffer layer, the one that exhibits the best Jsc. A wide band gap allows a better response in the short wavelengths, improving the external quantum e ciency (see Figure 4) and therefore the Jsc. The CdS and V 2 O 5 have similar band gaps but a cliff like CBO in the CdS/Sb 2 Se 3 interface favored the Jsc in this case. In all cases, an increase in the interface defect density implies an increase of recombination and consequently losses in Jsc.
As we have pointed out before, a good band alignment reduces the impact of interface recombination and signi cantly improve the Voc and the solar cell e ciency. In Figure 3d it could be notice that the solar cells with buffer layers that have the best band alignment, are less affected by the recombination losses. Finally, the V 2 O 5 shows the best performance as buffer layer due to the improve in FF and Voc resulting in the best e ciency behavior.

Effect of bilayers on n-p device photovoltaic performance
In the next step a bilayer is used as buffer layer in the n-p structure. The interdiffusion between CdS and Sb 2 Se 3 is considered as a possible degradation mechanism in this solar cell. Cd act as donor dopant of Sb 2 Se 3 creating an additional rectifying junction in the spatial charge region [25]; while based on the second one process, the CdS is not a suitable emitter to partner Sb 2 Se 3 due to Se diffusion, that contributes to the interface defects formation and compensation of donor states in CdS layer [26]. In reference [27] the CdSe was used with the intention to suppress the interdiffusion of Se across the interface. However, in this work no signi cant improvement in the e ciency was obtained.
Considering the above-mentioned results, two bilayers were evaluated in the present work: CdS/CdSe and the ZnO/V2O5. The last one was introduced since single-layers shown the best e ciency results and not contain Cd. Table 5 list the electrical parameters for the solar cells using the bilayers as emitters. From these results, the fallowing is concluded: CdS/CdSe bilayers improve the electrical parameters respect to the CdS and CdSe single-layers and using a bilayer without Cd an ulterior increase in the e ciency was achieved. A best band alignment between V 2 O 5 and Sb 2 Se 3 and high bandgap of ZnO results in better Voc and Jsc values respectively compared with CdS-CdSe bilayer, then a best solar cell e ciency was obtained. Taking in to account that e ciency improvement in 0.5 % for the records cells based in Sb 2 (SSe) 3 for 10% to 10.5% implied a technological effort, an increase from 5,0% (with CdS) to 11.6% (with ZnO/V 2 O 5 ) may justify adding another layer in the n-p structures. Table 5. J-V parameters for solar cells in n-p con gurations with ETL-bilayers.

Effect of n-i-p structure
on device photovoltaic performance.
As indicated above, the Voc de cit of 0.7 eV in antimony chalcogenide cells, is the parameter to be overcome. The Jsc and the FF reach their Shockley-Queisser (S-Q) limits faster than the Voc in this type of solar cells, which determines a lower Voc growth compared to e ciency increase. So, an increase in e ciency requires a decrease in the Voc de cit.
The n-i-p structure provides a more effective electric eld to separate the photoexcited carriers, compared to the traditional p-n structure in substrate or superstrate con gurations. The electric eld responsible for charge separation is formed throughout the entire thickness of the absorber in the n-i-p structure, unlike the eld created in the region of space charge, in the p-n junction. The n-i-p structure is formed adding a Cu 2 O layer (see Figure 1f). In order to compare the impact of both structures on the properties of solar cells, Table 6 shows the electrical parameters of the structures using the V 2 O 5 as ETL in both cases. As can be seen, the e ciency value of 11% is reached in the FTO/V 2 O 5 /Sb 2 Se 3 /Cu 2 O/Au con guration. For this comparative analysis, the same thicknesses (100 nm) were set for the ETL, 50 nm for Cu 2 O and 500 nm for the absorber. The value of 10 15 cm -2 was used for Sb 2 Se 3 /Cu 2 O interface defect density. In Figure   5 the dependences of Voc, Jsc, FF, and on the ETL/Sb 2 Se 3 interface defects density are shown, for n-p and n-i-p structures. The increase in e ciency in n-i-p structure is due to the reduction in bulk recombination in the absorber layer resulting in a higher Voc and FF with respect to the n-p structures. The Jsc was not improved by the addition of an HTL. In terms of interface defect density, small changes in Voc and FF are observed when defects increase. On the other hand, the Jsc is more affected by interface recombination in both cases. The good band alignment between V 2 O 5 and Sb 2 Se 3 determines that above a certain value of the interface defect density, the electrical parameters remain constant. In other words, the better the band alignment, the less the effect of the interfacial states on solar cell e ciency. Values of 10 11 cm -2 was reported in [9] for ZnO/Sb 2 Se 3 interface, then value of 10 12 cm -2 could be considering the optimal for both structures taking account that for lower values all parameters show saturation. Table 6. J-V parameters for solar cells in n-p and n-i-p con gurations using V2O5 as ETL Device structure Voc of the carriers must travel a greater distance to reach the electrodes, and by series resistance that also increase with the layer thickness. As can be seen for absorber thickness higher than 0.6 high e ciencies could be achieve using V 2 O 5 as emitter or ETL in n-p and n-i-p structures. In the case of V 2 O 5 layer the idea is reduce the thickness as much as possible, but a thinner layer could create pinholes and short circuit the device, besides the technological limitation. Nevertheless for 70 nm of layer thickness e ciencies above the 14 % in both cases could be achieved.

Conclusion
According to the results presented in this work the following conclusions are derived: The record e ciency achieved of 9.2% in Sb 2 Se 3 solar cells can be improved by the following proposals: (a) the substitution of AZO for FTO as TCO and of CdS for V 2 O 5 as ETL layer in the n-p structure. In the FTO/V 2 O 5 /Sb 2 Se 3 /Au con guration, an e ciency of 10.2% was obtained; (b) adding a second layer in the form of ETL-bilayer, the best combination turned out to be ZnO/V 2 O 5 with an ulterior increment in the e ciency of 11.6%. This result must evaluate in terms of the experimental complexity. ZnO can be deposited by various methods. Between then, the pneumatic chemical spray it has proven to be a suitable technique. (c) The analysis of the n-i-p structure con rms the improvement of the electrical parameter that is more di cult to improve in this type of solar cells, that is, the Voc. Values greater than 600 mV were obtained representing a substantial reduction of the Voc de cit. (d)The comparative study of the n-p structure in the FTO/V 2 O 5 /Sb 2 Se 3 /Au con guration and the n-i-p in the FTO/V 2 O 5 /Sb 2 Se 3 /Cu 2 O/Au con guration, showed that by optimizing the layers thickness in both structures e ciencies above 14 % could be achieved.(e) No signi cant increase in e ciency was obtained using n-i-p structure compared to n-p ones, that justi ed the incorporation of additional layer. (f) It was demonstrated that interface recombination can be counteracted by a good band alignment. Finally, it is important to mention that absorber quality is critical if high e ciency wants to be reached. The bulk defect density, the ribbons orientation and doping in the Sb 2 Se 3 material require especial attention. These results could be considered as a guide in the processing of planar solar cells of Sb 2 Se 3 and a step in the purpose of maximize the e ciency to guarantee their future at industrial level.

Declarations
Figures Figure 1 Solar cell structures simulated in this works.

Figure 2
Band diagram for solar cell in n-p structure.

Figure 3
Electrical parameters versus interface defects density for each solar cell in n-p structure using different buffer layers.

Figure 4
External quantum e ciency for solar cells in n-p structure with single buffer layer. Figure 5