Simulation of CdTe, CIGS and CZTS Solar Cells using WxAMPS Software

Solar cells made of Cadmium telluride (CdTe), copper indium gallium selenide (CIGS) and copper zinc tin sulde (CZTS) are currently the most widely studied thin lm technologies.To increase the performance and for better understanding of the behavior of CdTe, CIGS and CZTS solar cell simulations have been performed using WxAMPS software. Moreover, all the solar cells have been simulated with different buffer layers and transparent conductive oxide (TCO) layers such as Cadmium Sulphide (CdS), Zinc Sulphide (ZnS), Aluminum Zinc Oxide (AZO) and Indium Tin Oxide (ITO).Variations in the thickness and doping concentrations of TCO layers, buffer layers, and absorber layers have been done to test the performance of the solar cells.The effects of using a Back-Surface Reector (BSR) layer made of Zinc Telluride (ZnTe) have also been studied.Furthermore, the simulation work is exceptional in this regard since all of the layers of CdTe, CIGS, and CZTS solar cells were modeled using optical parameters (absorption coecients) from the literature. All the solar cell's open circuit voltage (Voc), short circuit current (Isc), maximum power (Pm), ll factor (FF), and photovoltaic eciencies have been represented in this work. The simulation results may provide valuable insight in developing and better understanding of high-eciency thin lm solar cells. Some of the parameters can be changed but in limited versions. The thickness of the layer and doping concentrations of N-type or P-type material can be changed. By varying these parameters, an improved result can be obtained. possible combinations done CdTe, CIGS alternative Zinc sulde (ZnS) have been used as buffer layers. Their thickness have been changed form 0.01–0.1 µm in the simulation. Apart for CZTS solar cell (as simulation could not be done due to numerical failure), ZnTe with thickness of 1 µm have been used as BSR for CdTe and CIGS solar cells. For all the materials the reference values of donor and acceptor concentrations (As tabulated in Table 2) have been used at the beginning of the simulations, later those values have been changed for optimization. In this simulation work, all the combinations used for CdTe, CIGS and CZTS solar cells and their respective outcomes have been disclosed in the results and discussion section.

layers have been altered in the simulation for optimization of CdTe, CIGS and CZTS solar cells. All the observed results including Voc, Jsc, ll factor (FF), and e ciency (η) and performance analysis of different solar cells have been discussed in this paper.

Thin Film Solar Cell (Tfsc)
A solar cell (also called a photovoltaic cell, as shown in Fig. 1) is an electrical device that converts the energy of light directly into electricity by the photovoltaic effect [27]. Where, the photovoltaic effect is the process that generates voltage in a photovoltaic cell when it is exposed to light or other radiant energy [28].
A thin lm solar cell is a solar cell whose thicknesses varies from few micrometers to few nanometers and it is constructed by depositing some thin layers consecutively. A general structure of thin lm solar cell is depicted in Fig. 2. A thin lm solar cell consists of different layers. Such as front contact / Transparent Conducting Oxide (TCO), window layer, buffer layer, absorber layer, back surface re ector (BSR), back contact and substrate. The TCO and window layer can be separate or can be the same as shown in Fig. 2 (a) and (b). Each layer has a speci c function and these layers are described brie y in the following section.
Front contact is the topmost layer of the thin lm solar cell. The main function of this layer is to collect the current produced by the cell and serve it to an outer circuit or load. 1st generation solar cell has opaque grid nger front contacts. Whereas, in 2nd generation solar cell (Thin Film) the front contact is designed differently. The front contact is a transparent conductive oxide (TCO) layer. The conductive oxide provides the same function as 1st generation contact and for being transparent, light can easily enter the TFSC. Thus, the front contact is also known as transparent conductive oxide (TCO) layer. The TCO being transparent, can be assumed it behaves like a window for the solar cell by which most of the light enters into the solar cell. Thus, it is also called window layer and both TCO and window layer can be the same (As shown in Fig. 2 (a)). However, when the transparency of the TCO layer increases, the resistance also increases proportionally. The resistance of TCO layer needs to be low because of the current collection from the buffer layer. To tradeoff between transparency and conductance, sometimes TCO and window layers are separated (As shown in Fig. 2 (b)) to improve the overall performance of the solar cell. The bandgap of the window layer must be high for greater light trapping and absorption of high energy photon. In this layer, a great amount of light is absorbed to the cell.
Under the window layer is the buffer layer. The buffer layer in general is an N-type semiconductor material, which along with the P-type absorber layer form the P-N junction of TFSC. The buffer layer is named so like this, because it adjusts the bandgap matching between the absorber layer and the window layer. The doping concentration of a buffer layer must be high so that the number of minority carrier is reduced and as a result recombination can be minimized.
Recombination on this layer degrade the working ability of the solar cell. Absorber layer absorb the low energy photon as the bandgap of this semiconductor material is low. In general the absorber layer is a P-type semiconductor material and have a higher contribution from photo-generated electron-hole (e-h) pairs [29]. Furthermore, the thickness of this layer is much higher than that of the other layers used in the solar cell. Not all of the TFSC architectures has the Back-Surface Re ector (BSR) layer. In the recent years, to improve the performance of back contact and to reduce the recombination in the structure, a back surface re ector (BSR) layer has been used. The essential role of this layer is to con ne the photogenerated minority carriers and make sure is close enough to reach the P-N junction so that current could be pro ciently collected [26,30]. By using back contact layer a full path for the current carrying circuit is made and is used for the purpose of current collection from the cell. It is deposited on the substrate layer and it does make contact with the back surface re ector (BSR) layer or absorber layer. On the substrate, the consecutive layers are deposited. The substrate can be of glass, plastic etc. Soda Lime Glass (SLG) is generally used as substrate in TFSC.

Types of Thin Film Solar Cell (TFSC)
In TFSC technology, there are different types of solar cells. These solar cells are named after the material used in the absorber layer. The rst type of TFSC is amorphous silicon solar cell. This type of TFSC uses bulk amount of silicon. However, to reduce the cost, the thin lm technology has been evolved and many solar cells like CdTe, CIGS, CZTS etc. have been designed successively. In this research CdTe, CIGS and CZTS TFSCs have been simulated and optimized.
Thus, some information of each of these cells are discussed in the following section.

Cadmium Telluride (CdTe) Solar Cell
Cadmium Telluride (chemical formula: CdTe) is the second most utilized solar cell technology today. The rst is still 1st generation silicon solar cell. CdTe is a suitable material for solar cell operation. Because, it has a direct bandgap of 1.45 eV for AM 1.5 solar spectrum and is nearly optimal for converting sunlight into electricity [31]. Furthermore, one of the great advantage of CdTe solar cell is, it is a low cost manufacturing technology. Nonetheless, the main problem with CdTe solar cell is its toxicity. Cd is harmful for environment. By using Cadmium sul de (CdS) as buffer layer the toxicity of CdTe solar cell is reduced as CdS material is less toxic than the Cd material alone. Currently many research is running for nding the alternative of CdS buffer layer to reduce the toxicity effect. At present, prominent research is going on the following buffer material: Zinc selenide (ZnSe), Zinc cadmium sul de (ZnCdS), Indium sul de (In 2 S 3 ), Zinc sul de (ZnS) and Indium selenide (In 2 Se 3 ). In addition to solve the toxicity problem, research on Window / TCO layers and Back-Surface Re ectors are still going on for the improvement of the e ciency of CdTe solar cell. The popular current research materials with thickness for Window / TCO layers and Back-Surface Re ectors have been tabulated in Table 1.

Copper Indium Gallium Selenide (CIGS) Solar Cell
Copper Indium Gallium Selenide (CIGS) is another kind of TFSC where CIGS is used as absorber layer. The CIGS material is a solid solution of copper indium selenide (CIS) and copper gallium selenide. It has a chemical formula of CuIn (1−x) Ga (x) Se 2 . Where the value of x can vary from 0 (pure copper indium selenide) to 1 (pure copper gallium selenide) The bandgap of CIGS material varying continuously with x from about 1.0 eV (for copper indium selenide) to about 1.7 eV (for copper gallium selenide). As CIGS material is used instead of CdTe as absorber layer, CIGS has advantage over CdTe in toxicity aspects. CIGS lm acts as a direct bandgap semiconductor. Nevertheless, the toxic effect is still has not been completely removed as in general CdS is used as buffer layer. However, alternative materials are used as a substitute of CdS nowadays. They are zinc sul de (ZnS), Zinc selenide (ZnSe), Indium sul de (In 2 S 3 ), Zinc Oxide (ZnO) and magnesium zinc oxide (MgZnO). By varying the thickness of materials in different layers and also by utilizing different combinations of materials, research on CIGS TFSC is still continuing. Summarized thickness, materials and associated layers are given away in Table 1.

Copper Zinc Tin Sul de (CZTS) Solar Cell
One of the critical issues with CdTe and CIGS based solar cells are, the less availability of tellurium and indium on earth. To solve this problem, Copper Zinc Tin Sul de (CZTS) material has drawn the attention of the researchers as an alternative absorber layer. Furthermore, CZTS is a non-toxic, low cost, earthabundant material having reasonable electrical and optical properties. The chemical formula of CZTS is Cu 2 ZnSnS 4 . CZTS also has a direct and tunable band gap (Eg ∼ 1.45 eV-1.6 eV). Yet detoxi cation of Cd from CdS buffer layer is still an issue for its advancement. To avoid this problem, other options include using zinc sul de (ZnS) / Zinc selenide (ZnSe) / Zinc cadmium sul de (ZnCdS) / Indium sul de (In 2 S 3 ) / zinc sul de (ZnS) or Indium selenide (In 2 Se 3 ) as buffer layer. CZTS solar cell is comparatively new than CdTe and CIGS TFSC. Thus, research on Window / TCO layer is also underway. Parameters of associated layers and materials of CZTS solar cell are presented in Table 1.

Wxamps Simulation Software Operating Procedure [32] WxAMPS (Widget Provided Analysis of Microelectronic and Photonic Structures) is an improved version of AMPS (Analysis of Microelectronic and Photonic
Structures) software. The AMPS software was developed by Stephen Fonash and his group at the Pennsylvania State University, U.S.A. and wasreleased in 1997 [33]. Then in 2012, AMPS has been improved to WxAMPS software at the University of Illinois at Urbana Champaign, in collaboration with Nankai University of China [34]. It is a popular simulation software for modeling of thin lm solar cells. In WxAMPS there is no limit to the number of layers allowed.
The dimension and input power of the solar cells are considered 1cm ×1cm and 100 mW/cm 2 respectively in the sofware. The pictorial diagram of the graphical user interface is given in Fig. 3. To make the simulation software operational, the simulation procedure needs to be carried out in three steps. In the rst step, the ambient operational environment is selected. Where, the standard value of room temperature 300K (25 0 c) and solar spectrum air mass 1.5G is considered. For proper simulation click the light on and load the AM1_5G 1 sun.spe and Light0_0.7_1.vol bias voltage le. Secondly, the materials operational environment is selected. Here, for each layer of the TFSC, different electrical and optical properties of the materials are entered. Finally, the simulation process is executed by pressing the run button and the outputs are shown in the results section.

Physical Parameters of Different Layers:
In WxAMPS simulation software's materials operational environment (also known as the "Material" section), speci ed electrical parameters values of every layers must be included to make the simulation to work properly. Some of the electrical parameters are bandgap, electron a nity, hole and electron mobility.
The values of the electrical parameters for some materials are stated in a Table 2.

Footnote
Thickness -W (µ m ), Bandgap -Eg (eV), Relative Permittivity -ε r , Electron a nity -χe (eV), Electron mobility -µ n (cm 2 /vs), Hole mobility -µ p (cm 2 /vs), Some of the parameters can be changed but in limited versions. The thickness of the layer and doping concentrations of N-type or P-type material can be changed. By varying these parameters, an improved result can be obtained.
Alongside electrical parameters one has to input the optical values like the wavelengths and their respective absorption coe cients. Every material has different action on different wavelength. Thus, the wavelengths of different materials and associated absorption coe cients are distinct with each other. In Table 3, the absorption coe cients for all of the materials used in the simulation are listed. By giving emphasis on the visible spectrum, the wavelengths from 320-800 nm have been taken into consideration for simulation in this work. The values of absorption coe cients are compulsory, otherwise the simulation will not run. To include the absorption coe cient value click the From AB button in the optical tab of the "Material" section. Then nm and m-1 will appear in the lower portion. Include relevant wavelength and absorption coe cient accordingly e.g inputting the absorption coe cient value of 1.6587e + 006 alongside 400 nm wavelength. By default, a lot of wavelengths will appear after clicking the From AB button. So the easy process is to save the needed wavelengths and associated absorption coe cients data by clicking To XML(.absx). Then open the ABSX le with note pad and reduce the number of unnecessary wavelength slots, if necessary put additional absorption coe cients and associated wavelength values, rename the le and then load the renamed .absx data by clicking From File button in the optical tab of the "Material" section. From  CdS [43] α (m − 1 ) CdTe [43] α (m − 1 ) ZnTe [44] α (m − 1 ) AZO [43] α (m − 1 ) CZTS [45] α (m − 1 ) ZnS [46] α (m − 1 ) CIGS [47] α (m − 1 ) ITO [48][49][50] α (m − 1 ) 320 The materials whose electrical and optical values are well established in the literature (as shown in Table 2 and Table 3)  From the simulation, it is observed that 0.1 µm thickness has the highest e ciency in all the cases of TCO. Whenever, the thickness of TCOs is increased from 0.1 to 1 µm the consequences are always associated with the reduction in the e ciency. This is due to the fact that, when the thickness is increased the sheet resistance is increased and thus the overall conductivity is decreased. From the experimental study, it is in agreement that if the sheet resistance increases, then the overall conductivity also decreases [51]. However, it is also seen from the experimental study that low thickness does not always have the best value.
Because when a layer thickness is minimized a lot of complications arises and in this simulation those factors have not been taken into consideration. Also from Fig. 4 it is seen that the transparent conductive Al-doped ZnO (AZO) thin lms have less variations with e ciency due to thickness variations than transparent conductive indium tin oxide (ITO). The reason is, due to the alteration of the thickness the variation in sheet resistance is low for AZO in compare to ITO, where sheet resistance changes considerably. Highest e ciency of 21.15% has being achieved by AZO_ZnS_CZTS solar cell. Closely followed by AZO_CdS_CdTe_ZnTe solar cell with 20.18% e ciency. Later in the simulation when doping concentrations have been optimized (As shown in Table 4) it is seen that, in all the cases AZO's performance is better than ITO layer. It may be due to the reason that, AZO layer has better optical transparency and thermal stability [52]. Nevertheless, the debate between the values of AZO vs. ITO for thin lm solar cell is still ongoing [53]. Currently both TCO materials are used by the manufacturers.
Simulation has been carried out with ZnTe as BSR layer with 1 µm thickness. Unfortunately, numerical error has been obtained for some of the parameters for CZTS solar cell. Thus, ZnTe layer has been used for CdTe and CIGS solar cell only. It has been found that for all the combinations the e ciency of thin lm solar cells with ZnTe BSR layer are more than without BSR layer. This is because the BSR layer re ects the incident light back to the absorber layer of the solar cell, thus extending the light path and causing the "light trapping effect" [54]. Reducing the thickness of ZnTe layer increases the e ciency of solar cell, but has not been done elaborately. So it is proposed as the future work of this research.

Impacts of Doping Concentrations and Thickness Variations of Buffer Layers and Absorber Layers
One of the key factors to increase the e ciency of the solar cell is to optimize the doping concentration. The doping concentration of the P-type and N-type  Table 4.
The highlighted blue colours in Table 4  Tinedert et al. also found using Silvaco-Atlas simulation software that CdTe thicknesses lower than 2 µm strongly penalize the solar cell e ciency [8]. For CIGS absorber layers, the lower the thickness the higher the e ciency is. Optimum thickness of CIGS absorber layer has been found 1 µm.
CdS buffer layer has been used as the counterpart to Zns buffer layer in this simulation. Both the buffer layers thickness have been varied from 0.01 to 1 µm. For ZnS layer the optimum thickness has been obtained at 0.02 µm (20 nm). Whereas, for CdS layer the optimum thickness has been acquired at 0.01 µm (10 nm). When the thickness of buffer layers increases from 0.01 µm then the e ciency of the solar cells decreases. Nykyruy et al. investigations also tells that the decrease of CdS window layer thickness leads to an increase in the e ciency and 10 nm thickness is the best [9]. Because of decreasing the thickness of CdS layer, it leads directly to the increase in the performance of the absorber layer of solar cells through decreasing the absorption losses that take place in the CdS layer and thus increase in higher short circuit current [55]. In practical situation achieving such low thickness of 0.01 µm (10 nm) can be of some di culty. Because in the nanometer range there may be some complications due to miniaturization. So far, the CdS layer thickness of 50 nm is the technologically minimal limit for the open evaporation method [55].
It has been observed from the solar cell simulations that, for most of the cases CdS Buffer layer has better e ciency than its counterpart ZnS buffer layer. Although, the best e ciency has been obtained with CZTS solar cell having ZnS as buffer layer. Highest e ciency of 23   Footnote D c is donor concentration, Jsc is short circuit current density, Voc is open circuit voltage, FF is ll factor, η is e ciency and t is thickness.

Conclusion
In this work, simulations have been performed for CdTe, CIGS and CZTS solar cells using WxAMPS software. In addition, all the solar cells have been simulated with different buffer layers and TCO layers such as CdS, ZnS, AZO and ITO. The e ciency of the solar cells have been observed by varying the thickness and doping concentrations of TCO layers, buffer layers, and absorber layers. Moreover, (except CZTS solar cells) the impacts of with and without Zinc telluride (ZnTe) Back-Surface Re ector (BSR) layer has also been investigated. Furthermore, the simulation work is unique in this perspective that, optical parameters (absorption coe cients) values found in the literature have been used for all the layers of CdTe, CIGS and CZTS solar cells. From the simulation, it is observed that, for all the cases of TCOs thickness ranging from 0.1 to 1 µm, the maximum e ciency has been achieved for 0.1 µm. While, the dispute over AZO vs. ITO as TCO is still ongoing and both TCO materials are at present used by the manufacturers, AZO's output outperforms ITO's in this simulation. The results show that (except CZTS solar cell) a thin lm solar cell with a ZnTe BSR layer performs better than one without. Solar cell simulations have shown that the CdS buffer layer has a higher e ciency than the ZnS buffer layer in the vast majority of cases. Although, the best e ciency has been obtained with CZTS solar cell having ZnS as buffer layer. The best e ciency of CdTe, CIGS, and CZTS solar cells have been found respectively for AZO_ZnS_CZTS, AZO_CdS_CdTe_ZnTe, and AZO_CdS_CIGS_ZnTe solar cells. The e ciency of these solar cells is 23.67%, 23.47% and 22.53% e ciency respectively. After optimization, all three solar cells shows promising results. Thus, it is hoped that this study will help the novices, scientistis, researchers and manufacturers to understand the behavior of CdTe, CIGS and CZTS thin lm solar cells and to fabricate high-e ciency thin lm solar cells in the near future. Figure 1 Basic structure of a silicon solar cell.   Thickness versus E ciency Column Chart for AZO.

Figure 5
Thickness versus E ciency Column Chart for ITO. I-V Curves of best CdTe, CIGS and CZTS solar cell