Performance and Stability Comparison of Low-Cost Mixed Halide Perovskite Solar Cells: CH 3 NH 3 PbI 3- x Cl x and CH 3 NH 3 PbI 3- x SCN x

Perovskite solar cell is categorized as a third-generation solar cell which is used for its high-performance and low-cost production. However, device stability is a major problem in the development of perovskite solar cells. Mixed halide perovskite is one of the subjects that have been proposed to improve perovskite solar cell stability. Research about solar cells using mixed halide perovskite is widely reported. However, complex con ﬁ gurations and fabrication using sophisticated equipment were usually used in those reported studies. In this work, the fabrication of solar cells using mixed halide perovskite CH 3 NH 3 PbI 3- x Cl x and CH 3 NH 3 PbI 3- x SCN x was conducted using a simple and low-cost structure. Solution-processed deposition fabrication method using spin coating was used to fabricate the devices. Optimization of the spin coating rate for each layer in the perovskite solar cells was performed to ensure that the devices exhibited decent performance. Stability measurement and analysis of the perovskite solar cells were conducted. Summarily, solar cells with mixed halide perovskite CH 3 NH 3 PbI 3- x Cl x exhibit the highest performance with an e ﬃ ciency of 2.92%. On the other hand, solar cell with mixed halide perovskite CH 3 NH 3 PbI 3- x SCN x has the best stability which only drops its e ﬃ ciency by 39% from its initial value after 13 days.


Introduction
Perovskite material with a general formula of ABX 3 (where A is organic cation, B is divalent metal, and X is halide) has piqued the interest of many researchers, notably in the solar cell field. The first introduction to perovskite solar cell came from Akihiro et al. in 2009 with a device efficiency of 3.8% [1]. Originally, perovskite was used as sensitizer in dyesensitized solar cell where perovskite with the formula CH 3 NH 3 PbX 3 (X was I or Br) was used. As a result, perovskite with CH 3 NH 3 PbI 3 yielded better efficiency. After ten years of its first appearance, the perovskite solar cell has reached its certified highest efficiency at 25.2% [2]. High absorption coefficient, long diffusion length, low cost, and many material possibilities are the several factors behind perovskite solar cell being extensively investigated and gaining its performance boost [3][4][5]. Another effort to achieve high-performance perovskite solar cells is by utilizing an organic transport layer spiro-OMETAD. Those transport layers can easily increase the efficiency of perovskite solar cells by around 15% [6]. Despite the achievement of highperformance devices, material instability is still a major concern in perovskite-based solar cells. Water, including that from humidity, can decompose the material of perovskite and degrade solar cell performance rapidly [7]. Furthermore, the organic transport layer also degrades rapidly in an open environment, thereby lessening the age of perovskite solar cells [8]. This drawback makes it rather difficult for perovskite solar cells to enter commercialization.
Several groups of researchers have suggested an engineering modification in the transport layer that can overcome the instability problem, such as the application of a metal oxide transport layer, porous buffer layer, porous transport layer, and nanocrystal transport layer [9][10][11][12][13]. Additionally, the utilization of inorganic materials replaces the usage of expensive complex organic compounds in the transport layer. Alternatively, other groups proposed an improvement in the perovskite material by partially replacing the Iion in perovskite CH 3 NH 3 PbI 3 , it came to be known as mixed halide perovskite. Zhang et al. provided information about the stability of mixed halide perovskite by incorporating the SCNion into the perovskite material (normally written as CH 3 NH 3 PbI 3-x SCN x ) [14]. Other groups also tried applying the SCNion in perovskite by changing the usage of PbI 2 powder to Pb(SCN) 2 in the perovskite precursor, which resulted in an improvement of device stability [15]. The Clion can also be applied to make mixed halide perovskite CH 3 NH 3 PbI 3-x Cl x . Greater performance and stability can be achieved by using the discussed perovskite rather than pure halide perovskite [16,17].
However, additive and complex structures were applied in the reported perovskite solar cells. In this research, we fabricated a simple structure of mixed halide perovskite solar cells using a solution-processed fabrication technique. Our devices have a n-i-p structure (FTO/TiO 2 /mixed halide perovskite/CuSCN/carbon/FTO) where mixed halide perovskite CH 3 NH 3 PbI 3-x Cl x and CH 3 NH 3 PbI 3-x SCN x were utilized as an active layer and inorganic material was used as transport layers. Use of FTO glass-to-FTO glass device configuration also has shown its potential in producing simple and lowcost PSC. Detailed fabrication methods of the devices were also provided. Further, we investigated the performance and stability of both devices to provide a clear report on the application of mixed halide perovskite.

Materials and Methods
2.1. Device Structure. Perovskite solar cell was made by following the structure shown in Figure 1. The structure type was n-i-p, where incoming light was passed through the electron transport layer (ETL), which was TiO 2 in this research, before reaching the perovskite layer. This makes the TiO 2 side become the top of the device. TiO 2 as ETL was used because it shows commendable stability as compared to other ETLs, such as ZnO. Although TiO 2 has a more acidic nature than ZnO, it is more thermally stable when used together with perovskite [18]. On the other hand, a wide bandgap (Eg = 3:2 eV) makes TiO 2 transparent, and thus, more light can directly pass through the perovskite layer. For the perovskite layer, mixed halide perovskite CH 3 NH 3 PbI 3-x Cl x and CH 3 NH 3 PbI 3-x SCN x were used. The hole transport layer (HTL) in this device was CuSCN. The advantage of using CuSCN was its wide bandgap (Eg = 3:6 eV) and high hole mobility (0.01-0.1 cm 2 V -1 s -1 ) [19]. Additionally, CuSCN could improve PSC stability [20]. Carbon powder was used to fill the gap or buffer layer between CuSCN and the FTO glass back-contact. In addition, carbon can be used as an electrode and an alternative electrode to Au or Ag [21]. In this device, the carbon powder combined with FTO glass backcontact acted as an electrode [22].
In perovskite solar cells, it is essential to take into account the energy band diagram of each layer. Mismatched energy level from one of the layers hold the potential to block the movement of the electron or hole generated from the perovskite layer. Figure 2 shows an energy level diagram of fabricated devices collected from various research. CH 3 NH 3 PbI 3-x Cl x has a bandgap of 1.55 eV with an energy level of -3.75 eV for the conduction band and -5.43 eV for the valence band [23]. CH 3 NH 3 PbI 3-x SCN x has a bandgap of 1.5 eV with a conduction band of -3.93 eV and a valence band of -5.43 eV, similar to pure perovskite CH 3 NH 3 PbI 3 [15]. Wang et al. have provided energy level information for TiO 2 and carbon [24]. TiO 2 has an energy level of -4.1 eV for the conduction band and -7.3 eV for the valence band. Moreover, carbon has an energy level of -5 eV. The energy level for CuSCN was provided as -1.8 eV for the conduction band and -5.4 eV for the valence band [25]. As metal contact, FTO has an energy level of -4.4 eV [7].

Device Fabrication.
In this research, device fabrication was carried out in ambient air. Prior to the deposition using spin coating, a precursor of TiO 2 , perovskite, and CuSCN was prepared. The TiO 2 precursor was synthesized by mixing 0.2 M titanium(IV) isopropoxide (TTIP) (97%, Sigma-Aldrich) and 0.1 M hydrochloric acid (HCl) (37%, Merck) into 5 ml of ethanol then stirred for 15 minutes at room temperature. CH 3 NH 3 I perovskite precursor was synthesized by mixing 8.66 ml of CH 3 NH 2 (40 wt. % in H 2 O, Sigma-Aldrich) and 13.2 ml of HI (57 wt. % in H 2 O, Sigma-Aldrich). The mixed solution was then evaporated with a rotary evaporator for 2 hours until a white dry crystal of CH 3 NH 3 I formed.
Perovskite precursor for both Cland SCNmixed halide was synthesized by mixing PbCl 2 or Pb(SCN) 2 (Sigma-Aldrich) powder and CH 3 NH 3 I with a molar ratio of 1 : 3, followed by dissolving it in 1 ml of N,N-dimethylformamide (DMF) (≥99.9%, Sigma-Aldrich). Both perovskite precursors were stirred for 1 hour at a temperature of 80°C. As HTL, the CuSCN precursor was made by dissolving 6 mg CuSCN powder (Sigma-Aldrich) in dipropyl sulfide (DPS) (97%, Sigma-Aldrich) and then mixed for 8 hours at room temperature.
The FTO glass (2 × 2 cm 2 ) was cleaned by an ultrasonic cleaner using three different solutions: deionized water, ethanol, and acetone. Solutions were used to clean the FTO glass for 5 minutes each. Then, the FTO glass was allowed to dry before the deposition using spin coating. The first layer to be deposited was TiO 2 by dropping the TiO 2 precursor on top of the clean FTO glass. The spin coater was spun for 30 seconds, followed by an annealing process using a hotplate with a temperature of 425°C for 30 minutes and left to cool to room temperature. The next deposition was a perovskite layer on top of the FTO/TiO 2 substrate. The perovskite precursor was heated at a temperature of 60°C before the deposition in order to lower the viscosity of the International Journal of Photoenergy precursor such that the solution could spread evenly. The FTO/TiO 2 substrate was placed in the spin coater and the perovskite precursor was dropped on top of it, followed by spin coating for 30 seconds. Then, the sample was dried using a hotplate at a temperature of 130°C for 15 minutes. When the annealing process was conducted, the yellow solution of perovskite became grey, implying that the perovskite layer was formed. The CuSCN precursor was applied to cover the surface of the perovskite layer by spin coating it for 30 seconds. Afterwards, the annealing process was performed to remove the DPS solution and this resulted in the formation of the CuSCN layer. Finally, carbon powder was placed onto the surface of the CuSCN layer as a buffer layer from the main substrate to the FTO glass back-contact. After the two FTO glasses were clipped together, the active area of the perovskite solar cell was 2 × 1:5 cm 2 . The spin coating rate used in this research was varied to obtain an optimized spin coating rate for each layer. Varied spin coating rate can produce different thicknesses of thin film. The general rule is that when the spin coating rate is high, the thickness of film will decrease because the majority of solvent will be flung off the side. Contrarily, a low spin coating rate will form a thick film because the force is not adequate to throw the solvent off the side as much as the force of a high spin coating rate. For TiO 2 , we varied spin the coating rate from 1000-4000 rpm,; for the perovskite layer; it was 500-2000 rpm, and for the CuSCN layer, it was 1000-4000 rpm. After optimization, a fixed spin coating rate was obtained from the highest performance out of all the variations. Afterwards, it was applied to fabricate the perovskite solar cells for performance and stability comparison.

Measurement.
The fabricated solar cells were measured for their dark I-V performance using the Semiconductor Parameter Analyser HP 4145B. In order to determine solar cell performance under illumination, the solar cell was placed outside on a sunny day and then the generated current (I L ) was measured using multimeter. The intensity from the sun was measured using a solar power meter, and the generated current measurement of the solar cell was carried out when the sun intensity was averagely 900 W/m 2 . By using Equation (1), information from dark I-V will shift below, corresponding to the generated current [26].
The shifted dark I-V will become a typical I-V curve for solar cells, and therefore, photovoltaic parameters such as open-circuit voltage (V OC ), short-circuit current (I SC ), maximum power voltage (V mp ), and maximum power current (I mp ) will be visible. Moreover, fill factor (FF) and efficiency (η) can be calculated by using Equations (2) and (3).
Series resistance (R SH ) and shear resistance (R S ) were also provided using the calculation. Before calculating parasitic Since the fill factor is related to parasitic resistance, normalised R S (r s ) and normalised R SH (r sh ) can be written as Equations (6) and (7), respectively.
If the characteristic resistance (R CH ) of solar cell is defined in series resistance and shear resistance are given as follows: By substituting Equation (6) into Equation (9), series resistance expression would be given as follows: Similarly, for shunt resistance, by the substitution of Equation (7) into Equation (10), the following can be obtained: 3. Results and Discussion 3.1. Film Characterization. In order to examine whether the deposition process was successful using the prepared precursor, characterization was performed to the fabricated film. Figure 3 shows the X-ray diffraction (XRD) pattern of CH 3 NH 3 PbI 3-x Cl x and CH 3 NH 3 PbI 3-x SCN x film on top of the FTO glass. Each perovskite has apparent peaks (2θ) located in 14°, 28°, and 32°, which named planes (110), (220), and (310), respectively (marked with a star). The three peaks correspond to the three lattices of perovskite that are also apparent in perovskite CH 3 NH 3 PbI 3 . This shows that the application of PbCl 2 or Pb(SCN) 2 does not change the crystal structure of perovskite. The other peaks appeared to correspond to the application of those materials [27]. Figure 4 shows the result of energy-dispersive X-ray (EDX) analysis for each material. The EDX was used to demonstrate the presence of ETL, perovskite, and HTL material. TiO 2 was successfully formed with the presence of titanium and oxygen elements (Figure 4(a)). EDX analysis also gave a dominant signal of Pb and I elements, implying that perovskite material exists after the deposition (Figures 4(b) and 4(c)). Furthermore, CuSCN material was also confirmed with the presence of Cu, S, C, and N elements in EDX analysis with a high signal of copper and sulphur elements (Figure 4(d)).

Layer Optimization.
In this research, we optimized the spin coating rate of every layer to obtain decent performance levels for both perovskite solar cells and a comparable sample. A summary of the optimized spin coating rate for this configuration can be found in Table 1.

International Journal of Photoenergy
It is worth noting that the spin coating rate can affect the performance of devices in terms of the layer thickness and uniformity formed. In TiO 2 , if the layer is too thick from using a low spin coating rate, the electron will travel a greater distance to the metal contact, resulting in increased resistance. Furthermore, the thick TiO 2 layer increases transmittance, which reduces incoming light to the perovskite layer [30]. Those reports are in good agreement with the experiment results where a spin coating rate of 1000 rpm was used in TiO 2 , the performance was the lowest among other spin coating rates. On the other hand, a high spin coating rate forms a thinner layer as compared to a slow spin coating rate. If the spin coating rate for TIO 2 is too high, the TiO 2 layer will be too thin, which causes the FTO layer below it to not be covered uniformly. This will create a pinhole and decrease the overall performance of the devices [30]. From the experiment, it was obtained that 3000 rpm is an optimized spin coating rate for TiO 2 .   6 International Journal of Photoenergy The thickness of the perovskite layer played a key role in the performance of perovskite solar cells. The spin coating rate optimization was performed for both perovskites with Cland SCNto achieve the optimum thickness. It was observed from the experiment that when the lowest spin coating rate (500 rpm) was used, the perovskite layer did not completely transform into grey when the annealing process was finished, which might be caused by an excess of perovskite precursor left in the substrate which in turn might have caused some spot to be too thick; thus, the DMF solvent did not evaporate completely. Wang et al. showed that there was an increase in the performance of the perovskite solar cell when the perovskite thickness was increased. However, if the thickness was too high, it showed a decrease in performance caused by the rough surface of the perovskite [31]. The annealing process can be performed with a higher temperature in order to evaporate the solvent completely. However, higher annealing temperature affected the quality of the perovskite film, thereby decreasing the device's performance [32]. A thin perovskite layer can decrease the performance of perovskite solar cells because there is not enough material to generate electron-hole pairs from the incoming light. In the experiment, the optimum spin coating rate for the perovskite layer in this configuration was found to be 1000 rpm.
Spin coating rate optimization for the CuSCN layer was performed for both layers deposited on top of perovskite with Clas well as SCN -. Both had an optimum spin coating rate of 2000 rpm. In the case of CuSCN thin film, when the thickness is too thin, the CuSCN surface becomes rough and inhomogeneous [25]. These results have a negative impact on the efficiency of the charge transport.

Device Performance and Stability.
For stability tests, the devices were kept in a dry box with humidity of 25-30% and temperature of 28°C. The devices only placed outside the box when the measurement is conducted. Table 2 shows the performance of two mixed halide perovskite solar cells.
The perovskite solar cells were made by using a fixed spin coating rate from the previous experiment. It shows that the perovskite solar cell with CH 3 NH 3 PbI 3-x Cl x yields higher efficiency than the SCN --doped perovskite solar cell. The results are consistent with research conducted by Chen et al. [27]. The external quantum efficiency (EQE) of CH 3 NH 3 PbI 3-x Cl x is higher at almost all wavelengths, compared to CH 3 NH 3 PbI 3-x SCN x . Higher EQE indicates that more electron-hole pairs will be generated when the incoming light is absorbed by the materials. Furthermore, short-circuit current of the two devices shows compatible results with the aforementioned research. CH 3 NH 3 PbI 3-x Cl x with short-circuit current of 4.32 mA/cm 2 generates more charge than CH 3 NH 3 PbI 3-x SCN x with short-circuit current of 2.99 mA/cm 2 .
To analyze performance stability by time, the sample was stored in a dry box with a humidity level of 18%-20% and measured periodically for 13 days. Perovskite solar cells were taken outside the dry box only for measurement. The changes in the efficiency of fabricated devices are shown in Figure 5.
The results show that the solar cell CH 3 NH 3 PbI 3-x SCN x possessed better stability compared to the Cl --based perovskite solar cell. On day two, there was a slight increase in the efficiency for solar cells using CH 3 NH 3 PbI 3-x SCN x . This might have been caused by the safe condition where the humidity was lower or drier from ambient. Research by Leguy et al. reveals that humidity in perovskite solar cells can be removed by drying the devices [33]. Contrarily, the solar cell using CH 3 NH 3 PbI 3-x Cl x shows a decrease in performance by 5% from its initial efficiency. Days 6 and 13 show a performance decrease for both samples. However, on day 13, the SCN --based perovskite solar cell exhibits greater stability with only 39% degradation in efficiency, compared to solar cell CH 3 NH 3 PbI 3-x Cl x with 61% degradation. The changes of photovoltaic parameters for 13 days are summarized in Table 3.
Furthermore, the decrease in the fill factor of solar cells using CH 3 NH 3 PbI 3-x Cl x is also shown in the I-V curve of   Figure 5: The stability test of perovskite solar cell using CH 3 NH 3 PbI 3-x Cl x (circle) and CH 3 NH 3 PbI 3-x SCN x (square) for 13 days. 7 International Journal of Photoenergy the stability test (Figure 6(a)). Decrease in the fill factor makes the I-V curve's shape more linear. On the other hand, SCN-based perovskite solar cells did not show a significant decrease in the fill factor, which leads to a consistent I-V curve shape (Figure 6(b)).
Perovskite solar cell decomposition can be seen visually through the changes of colour in the perovskite layer from dark grey to yellow. This degradation is shown in Figure 7. On day 13 days, it is visible that the colour of Cl-based perovskite solar cell turned back into yellow. This finding is significantly in agreement with the stability test data where the performance degradation of CH 3 NH 3 PbI 3-x Cl x is much faster and greater compared to CH 3 NH 3 PbI 3-x SCN x . Perovskite CH 3 NH 3 PbI 3-x SCN x has good stability      International Journal of Photoenergy because SCNhas an ionic bond with Pb 2+ and a hydrogen bond with CH 3 NH 2 + [15]. Compared to pure halide perovskite (CH 3 NH 3 PbI 3 ), Clbased perovskite also has greater stability [16]. Moreover, its stability was provided by its better crystallinity and morphology. The results show that chemically stable perovskite CH 3 NH 3 PbI 3-x SCN x exhibits better stability than perovskite CH 3 NH 3 PbI 3-x Cl x does.

Conclusions
In conclusion, the performance and stability of solar cells using mixed halide perovskite CH 3 NH 3 PbI 3-x Cl x and CH 3 NH 3 PbI 3-x SCN x were demonstrated. It was found that solar cells using CH 3 NH 3 PbI 3-x Cl x have a better performance rate as compared to SCN --based perovskite solar cells with V OC of 1.16 V, I SC of 4.32 mA/cm 2 , FF of 0.52, and an efficiency of 2.92%. The stability test shows that perovskite solar cells using CH 3 NH 3 PbI 3-x SCN x yield better stability with only 39% degradation from its initial efficiency in comparison with another mixed halide perovskite solar cell.

Data Availability
All data used to support the findings of this study are included within the article.

Conflicts of Interest
The authors declare that they have no conflict of interest.