Optimization of the Power Conversion Efficiency of CsPbI_xBr_3−x-Based Perovskite Photovoltaic Solar Cells Using ZnO and NiO_x as an Inorganic Charge Transport Layer

Abstract

In this study, we analyzed the maximum power conversion efficiency (PCE) of a photovoltaic cell with an ITO/ZnO/CsPbI_xBr_3−x/NiO_x/Au structure, using ZnO and NiO_x as the inorganic charge transport layers and CsPbI_xBr_3−x as an absorption layer. We optimized the thickness of each layer and investigated the effects of the defect density and interface defect density. To achieve the highest PCE, the optimal thicknesses were 300 nm for the electron transport layer (ZnO), 60 nm for the hole transport layer (NiO_x), and 1000 nm for the absorption layer. The absorber defect density was maintained at approximately 10¹⁵ cm⁻³, and the interface defect density was approximately 10¹¹ cm⁻³. The highest PCE obtained through optimization of each of these factors was 23.07%. These results are expected to contribute to the performance optimization of perovskite solar cells that use inorganic charge carrier transport layers.

1. Introduction

Recently, owing to the depletion of fossil fuel energy and increasing global warming concerns, interest in renewable energy has increased. In particular, photovoltaic solar cells can be used in most places on earth, and their energy is abundant. Thus, many studies on solar cells have been conducted [1,2,3]. Specifically, since perovskite of the ABX₃ structure was first discovered, solar cells have developed more rapidly. Perovskite has a long diffusion length, large absorption coefficient (10⁴ cm⁻¹), and low price [4,5]. In the ABX₃ structure, A contains large cations such as Cs⁺, methylammonium (MA⁺, CH₃NH₃⁺), and formamidine (FA⁺, CH₃CH₂NH₃⁺), B contains metal cations such as Pb²⁺ and Sn²⁺, and X contains anions such as Cl⁻, Br⁻, and I⁻ [6]. Organic and organic-inorganic hybrid perovskite are widely applied in composing solar cell devices including perovskite [7]. Recently, perovskite solar cells (PSCs) have been reported to exhibit a power conversion efficiency (PCE) of nearly 25% [8,9]. However, despite the high PCE, most organic electron transport layers (ETLs) and hole transport layers (HTLs) constituting these PSCs remain sensitive to moisture [10,11]. Owing to these characteristics, encapsulation of the entire device is required to achieve long-term reliability [12]. In addition, organic-inorganic hybrid solar cells have issues such as band alignment and recombination losses and are also a problem in terms of long-term stability [13]. Therefore, interest in all-inorganic solar cell structures, in which inorganic materials are applied to devices, is increasing [14]. Generally, metal oxide semiconductors have high electron mobility (>10 cm²/Vs) so that a separate doping process is not required, as well as a wide band gap (E_g > 3 eV) and visible light region that have excellent transmittance [15]. The p-type metal oxide semiconductors include MoO₃, Cu₂O, CuI, CuS, and NiO_x, and the n-type metal oxide includes SnO₂, ZnO, In₂O₃, and Ga₂O₃ [16,17,18]. Among the n-type materials, ZnO has a wide band gap energy (E_g = 3.26 eV) and excellent mobility, while the p-type NiO_x is chemically stable and has excellent electron-blocking properties [19,20,21]. Therefore, in this study, ZnO and NiO_x were selected as the ETL and HTL materials, respectively, based on these excellent characteristics and the band alignment of the energy band diagram [22]. CsPbI_xBr_3−x was used as the perovskite absorption layer because it has been reported to be superior to other perovskite materials in terms of its moisture, thermal, and light stabilities [23,24].

Although many related papers have reported on all-inorganic PSCs [25,26,27], few studies have systematically analyzed the effect of each factor (e.g., thickness, defect density, and interface defect density) in a solar cell device using ZnO ETL and NiO_x HTL with CsPbI_xBr_3−x as the inorganic absorbing layer on a solar cell’s performance [28]. Therefore, in this study, the effect of each layer’s thickness, defect density, and interface defect density on solar cell performance is systematically analyzed, and an optimized solar cell structure is suggested based on an all-inorganic solar cell with a device structure of inorganic ZnO ETL/absorbing layer CsPbI_xBr_3−x/inorganic NiO_x HTL.

2. Simulation Procedure

In this study, as shown in Figure 1, ITO/ZnO/CsPbI_xBr_3−x/NiO_x/Au PSC devices with an n-i-p structure were employed. The simulation was performed using SCAPS-1D simulation software (Burgelman et al., 2000) [29]. This tool is based on Poisson’s equation (Equation (1)) and the continuity equation for each electron and hole (Equations (2) and (3)), and it helps to calculate the short-circuit current density (J_sc), open circuit voltage (V_oc), power conversion efficiency (PCE), fill factor (FF), quantum efficiency (QE), and recombination current, among others [30]. It is also based on the drift–diffusion equation (Equation (4)), which causes carrier transport in semiconductors [31]:

$${𝒹}^2\mathsf{\phi }\left(𝓍\right)/𝒹{𝓍}^2=ℯ/{ℰ}_0{ℰ}_{\mathrm{r}}(p\left(𝓍\right)-𝓃\left(𝓍\right)+{\mathrm{N}}_{\mathrm{D}}-{\mathrm{N}}_{\mathrm{A}}+{\rho }_{\mathrm{p}}-{\rho }_{𝓃})$$

(1)

In Poisson’s equation, we have the following:

• φ = the electrostatic potential;
• e = electrical charge;
• ε_r = relative permittivity;
• ε₀ = vacuum permittivity;
• N_D = charged impurities of the donor;
• N_A = charged impurities of the acceptor;
• ${\rho }_{\mathrm{p}}$ = distribution of holes;
• ${\rho }_{𝓃}$ = distribution of electrons.

$$𝒹{\mathrm{J}}_{\mathrm{n}}/𝒹𝓍=\mathrm{G}-\mathrm{R}$$

(2)

$$𝒹{\mathrm{J}}_{\mathrm{p}}/𝒹𝓍=\mathrm{G}-\mathrm{R}$$

(3)

In the continuity equation, we have the following:

• R = the recombination rate;
• J_n = the electron current density;
• J_p = the hole current density;
• G = the generation rate.

$${\mathrm{J}}_{\mathrm{n}}\left(𝓍\right)={\mathrm{q}\mathsf{\mu }}_{n}n\left(𝓍\right)ℰ\left(𝓍\right)+{\mathrm{qD}}_n\frac{dn\left(x\right)}{dx},{\mathrm{J}}_{\mathrm{p}}\left(𝓍\right)={\mathrm{q}\mathsf{\mu }}_{p}p\left(𝓍\right)ℰ\left(𝓍\right)-{\mathrm{qD}}_p\frac{dp\left(x\right)}{dx}$$

(4)

In the above drift–diffusion model, we have the following:

• q = the charge of the electron;
• ${\mathsf{\mu }}_n$ = the mobility of the electrons;
• ${\mathsf{\mu }}_{\mathrm{p}}$ = the mobility of the holes;
• $n\left(𝓍\right)$ = electron distribution;
• $p\left(𝓍\right)$ = hole distribution;
• $ℰ\left(𝓍\right)$ = the electric field;
• D_n = the electron diffusion coefficient;
• D_p = the hole diffusion coefficient.

Figure 1a shows the basic structure of the all-inorganic multi-layer perovskite solar cell (PSC) in the study. ITO, ZnO, CsPbI_xBr_3−x, NiO_x, and Au were used as the front metal contact, ETL, perovskite active layer (PAL), HTL, and back metal contact, respectively. The energy band alignment of this multi-layer perovskite solar cell (PSC) is shown in Figure 1b. The work functions of ITO (used as a front contact) and Au (used as a back contact) were −4.52 eV and −5.1 eV, respectively [32,33]. The lowest unoccupied molecular orbital (LUMO) levels and the highest occupied molecular orbital (HOMO) levels of the ZnO, CsPbI_xBr_3−x, and NiO_x were −4.1 eV, −3.95 eV, and −1.46 eV and −7.4 eV, −5.73 eV, and −5.26 eV, respectively [34,35,36,37]. Through this optimized energy band alignment, it can be seen that the electrons and holes generated in the CsPbI_xBr_3−x absorption layer can efficiently move to the ITO electrode and the Au electrode, respectively. This device analysis was conducted under AM1.5G spectrum (1000 W/m, T = 300 K) conditions.

Table 1. Physical parameters of ZnO, CsPbI_xBr_3−x, and NiO_x used in the ITO/ZnO/CsPbI_xBr_3−x/NiO_x/Au PSC.

Parameter	ZnO [34]	CsPbIxBr3−x [35]	NiOx [36]
Thickness (nm)	100	800	50
Band gap (eV)	3.3	1.78	3.8
Electron affinity (eV)	4.1	3.95	1.46
Dielectric permittivity (relative)	9	6	10.7
CB effective density of states (cm−3)	4.0 × 1018	1.1 × 1020	2.8 × 1019
VB effective density of states (cm−3)	1.0 × 1019	8.0 × 1019	1.0 × 1019
Electron thermal velocity (cm/s)	1.0 × 107	1.0 × 107	1.0 × 107
Hole thermal velocity (cm/s)	1.0 × 107	1.0 × 107	1.0 × 107
Electron mobility (cm2/Vs)	100	16	12
Hole mobility (cm2/Vs)	25	16	2.8
Shallow uniform donor density ND (cm−3)	1.0 × 1018	-	-
Shallow uniform acceptor density NA (cm−3)	-	1.0 × 1015	1.0 × 1018
Defect density Nt (cm−3)	1.0 × 1014	1.0 × 1014	1.0 × 1014

summarizes the parameters of the PSC layers used in this study with SCAPS simulation software. For the numerical values of the ETL, HTL, and absorption layer, the previously published literature was referenced [34,35,36,37]. For the initial PSC device configuration, the thickness of each layer of ZnO, CsPbI_xBr_3−x, and NiO_x was 100 nm, 800 nm, and 50 nm, respectively. After that, optimization simulation was performed for each thickness to increase the power conversion efficiency first. The band gaps of ZnO, CsPbI_xBr_3−x, and NiO_x were 3.3, 1.78, and 3.8 eV, respectively, and the electron affinity was 4.1, 3.95, and 1.46 eV, respectively. Relative dielectric permittivity was set to 9, 6, and 10.7, respectively. Refer to Table 1 for the remaining conduction band (CB) effective densities of the states, valence band (VB) effective densities of the states, electron and hole thermal mobilities, electron and hole mobilities, shallow uniform donor densities, shallow uniform acceptor densities, and defect densities. For the interface defect between ZnO/CsPbI_xBr_3−x, the defect type was set to neutral, the capture cross-section of electrons and holes was set to 1.0 × 10⁻¹⁵ cm², and the total density value was 1.0 × 10¹¹ cm⁻² [30]. For the interface defect between CsPbI_xBr_3−x and NiO_x, the defect type was set to neutral, the capture cross-sections of the electrons and holes were set to 1.0 × 10⁻¹⁸ cm² and 1.0 × 10⁻¹⁶ cm², respectively, and the total density value was set to 1.0 × 10¹² cm⁻² [37].

3. Results and Discussion

3.1. Performance of the Perovskite Solar Cell (PSC) under the Initial Conditions

Figure 2 shows the current density–voltage (J–V) and QE of the ITO/ZnO/CsPbI_xBr_3−x/NiO_x/Au PSC device under the initial conditions. The V_oc, J_sc, and FF were measured to be 1.03 V, 23.86 mA/cm², and 67.27%, respectively. The FF and PCE values were calculated using Equations (5) and (6) [38]. Accordingly, the PCE was determined to be 16.61%. Based on these basic performance indicators, from Section 3.2, the effect of the layer thickness, defect density, and interface defect density on the PCE were systematically analyzed:

$$\mathrm{FF}=\frac{{\mathrm{V}}_{\max }{\mathrm{J}}_{\max }}{{\mathrm{V}}_{\mathrm{oc}}{\mathrm{J}}_{\mathrm{sc}}}$$

(5)

$$\mathrm{PCE}=\frac{{\mathrm{V}}_{\mathrm{oc}}{\mathrm{J}}_{\mathrm{sc}}\mathrm{FF}}{{\mathrm{P}}_{\mathrm{in}}}$$

(6)

In these equations, we use the following:

• FF = the fill factor;
• V_max = the maximum voltage;
• J_max = the maximum current density;
• V_oc = the open-circuit voltage;
• J_sc = the short-circuit current density;
• PCE = power conversion efficiency;
• P_in = the incident power of the sun.

3.2. Effect of the ZnO Thickness

ZnO applied as an ETL in an n-i-p device as the first layer that absorbed the incident light. Therefore, the thickness of the ZnO layer was optimized in the range of 100–1000 nm. The current density–voltage (J–V) curves with respect to the ZnO thickness are shown in Figure 3. At a ZnO thickness of 300 nm, the highest values of V_oc, J_sc, FF, and PCE were 1.04 V, 23.3 mA/cm², 80.2%, and 19.4%, respectively. Then, as the ZnO thickness increased, each output parameter values showed a saturated behavior. The reason that the PCE increased with the ETL thickness from 100 nm to 300 nm may be attributed to the increase in dissociation of the photo-generated holes and electrons and reduced recombination in the photovoltaic solar cells [39]. When the thickness exceeded 300 nm, as recombination also increased as the thickness of ZnO increased, the PCE did not significantly increase and was considered to show a behavior of saturation. A similar behavior can be found in [39].

3.3. Effect of the Perovskite Absorption Layer (PAL) Thickness

In a photovoltaic solar cell, the absorption layer absorbs light and creates a hole-electron pair. Thickness optimization is important because the amount of absorbed light can vary depending on the thickness of the absorption layer. In this study, such analysis was conducted in the PAL thickness range of 100–1000 nm. Figure 4 presents the current density–voltage (J–V) curves for various PAL thicknesses. In Figure 5, each output parameter value (i.e., V_oc, J_sc, FF, and PCE) is shown according to the thickness of the PAL.

As the PAL thickness increased, the V_oc and J_sc exhibited increasing behavior. However, the FF value decreased at PAL thicknesses of above 500 nm. This behavior was somewhat different from those of the V_oc and J_sc. This can be determined from the relationship between the thickness of the absorption layer and resistance and is expressed as shown in Equations (7) and (8) [40].

As the thickness of the absorption layer increased, the resistance of the absorption layer also increased, owing to an increase in the series resistance. This increased series resistance led to a decrease in the FF value, as given by Equation (9) [41,42]. Therefore, the FF value was considered to decrease above a PAL thickness of 500 nm:

$${\mathrm{R}}_{\mathrm{absorber}}=\frac{\rho \mathrm{t}}{{\mathrm{A}}^2}$$

(7)

$${\mathrm{R}}_{\mathrm{series}}\cong \rho \mathrm{t}+\frac{{\mathrm{R}}_{\mathrm{s}}{\mathrm{L}}^2}{2}$$

(8)

$${\mathrm{FF}}_{\mathrm{series}}={\mathrm{FF}}_0\left(1-{\mathrm{r}}_{\mathrm{series}}\right)$$

(9)

Here, we use the following:

• R_absorber = the absorber resistance;
• $\rho$ = the absorber layer resistivity;
• t = the absorber layer’s bulk region thickness;
• A = the base area;
• R_series = the series resistance;
• L = length;
• R_s = sheet resistance;
• FF_series = the FF affected by the series resistance;
• F₀ = the initial FF;
• r_series = normalized series resistance (R_series J_sc/V_oc).

In the case of a PCE above 800 nm, it appeared to be saturated. At a PAL thickness of 1000 nm, the PCE showed the highest value of 19.7%. The more light was absorbed, the more hole-electron pairs which could be created, which increased each output parameter [43]. However, if the thickness was too thick, the carrier movement path and recombination current increased. Thus, it was necessary to select an appropriate thickness [44]. Accordingly, 1000 nm was selected as the optimal PAL thickness.

3.4. Effect of the NiO_x Thickness

In this study, NiO_x was used as a hole transport layer (HTL), and its thickness was analyzed in the range of 30–60 nm. The current density–voltage (J–V) data for varying NiO_x thicknesses are depicted in Figure 6. The V_oc did not change significantly until 60 nm, whereas the J_sc increased from 22.8 to 25.6 mA/cm². These results indicate that the PCE increased from 18.5% to 20.3%.

3.5. Device Performance Optimization of a PSC

Based on the optimization of the ETL, HTL, and perovskite layer thicknesses, the thicknesses for each layer were 300 nm for ZnO, 1000 nm for PAL, and 60 nm for NiO_x. As shown in Figure 7a,b, the optimal V_oc, J_sc, FF, and PCE values simulated using each of these optimized layers increased from 1.03 to 1.05 V, 23.86 to 25.56 mA/cm², 67.27% to 75.75%, and 16.61% to 20.26%, respectively, when compared with those of the initial reference device structure. With this optimized structure, the effects of the PAL defect density and the interface defect density between ZnO/CsPbI_xBr_3−x and CsPbI_xBr_3−x/NiO_x are also analyzed in the following section.

3.6. Effect of the PAL Defect Density

The defect density in the absorption layer is one of the major factors affecting the device performance because it can act as a trap when the hole and electron carriers move to each electrode [45]. A corresponding analysis was performed in the range of 10¹¹–10¹⁸ cm⁻³, and the results are shown in Figure 8 and Figure 9. Figure 8 shows each output parameter, such as the V_oc, J_sc, FF, and PCE, according to the defect density. All output parameters showed a constant value until the defect density was 10¹⁵/cm³ but showed a sharp decrease when it exceeded 10¹⁵/cm³. Figure 9 presents the current density–voltage (J–V) curves for the various PAL defect densities. As the defect density increased, the V_oc, J_sc, FF, and PCE decreased significantly from 1.05 to 0.99 V, 25.6 to 11.9 mA/cm², 76.1% to 45.8%, and 20.4% to 5.44%, respectively. These results were thought to be caused by Shockley–Read–Hall (SRH) recombination [46]. According to the SRH recombination, as the defect density increases, the carrier lifetime and diffusion length decrease. This causes an increase in SRH recombination, as shown in Figure 9. Thus, the PCE of this device decreased.

Moreover, it can be seen from Figure 10 that the recombination current increased rapidly above 10¹⁶ cm⁻³. Therefore, in this study, the PAL defect density was kept below 10¹⁵ cm⁻³.

3.7. Effect of the Interface Defect Density

Because the holes and electron carriers generated by light must pass through the interface to reach each electrode, the interface defect density is important for device performance. The effects of the ZnO/PAL (CsPbI_xBr_3−x) interface and PAL (CsPbI_xBr_3−x)/NiO_x interface defect density were investigated in the interface defect density range of 10⁸–10¹⁴ cm⁻³. Figure 11a,b illustrates the current density–voltage (J–V) curves for the ZnO/PAL and PAL/NiO_x interfaces, respectively.

In the case of the ZnO/PAL interface, as the interface defect density increased from 10⁸ cm⁻³ to 10¹⁴ cm⁻³, each of the output parameters (V_oc, J_sc, FF, and PCE) decreased from 1.06 V, 25.6 mA/cm², 76.0%, and 20.5% to 0.96 V, 25.0 mA/cm², 72.4%, and 17.3%, respectively.

In the case of PAL/NiO_x, as the interface defect density increased from 10⁸ cm⁻³ to 10¹⁴ cm⁻³, the V_oc, J_sc, FF, and PCE decreased from 1.13 V, 25.6 mA/cm², 81.8, and 23.7% to 1.01 V, 23.8 mA/cm², 56.0, and 13.4%, respectively.

Even if the carriers have a diffusion length that can reach each the electrode without recombination, severe recombination can occur because of an increase in the interface defect density [47]. Therefore, when stacking layers, a reduction in the interface defect density is recommended.

In this analysis, corresponding values of 10¹¹ cm⁻³ or less were considered because both interface defect densities exhibited sharp decreases in efficiency above 10¹² cm⁻³. Therefore, the interface defect density was fixed at 10¹¹ cm⁻³ for both the ZnO/PAL and PAL/NiO_x interfaces. Following this optimization process, the V_oc, J_sc, and FF obtained were 1.09 V, 25.58 mA/cm², and 82.89%, respectively, and as a result, a maximum PCE of 23.07% was realized.

3.8. The Effect of Back Contact Materials

The work function is defined as the minimum energy of the photon required for an electron to come out of a metal surface [48]. In order to find out the effect of the work function of the back contact, as summarized in

Table 2. Work function of various back contact materials for PSC.

Material	Cu	Ag	Fe	Au
Work function (eV)	4.6	4.7	4.8	5.1
PCE (%)	17.36	19.65	21.62	23.07

, various electrodes such as Cu, Ag, Fe, and Au were used [33].

Figure 12 shows the current density–voltage (J–V) curves for various back contact materials. It can be seen that the output parameter value increased as the work function value increased. The V_oc increased from 0.87 V to 1.04 V, and the PCE increased from 17.36% to 23.07%. This is consistent with the results of the previous literature, which reported that solar cell efficiency increased as the work function value increased [49]. As the work function value increased, the barrier height value of the majority carriers decreased, which made the electrode more of an ohmic type. Therefore, the V_oc and PCE values increased. Accordingly, the best results were obtained when Au, which had the largest work function of 5.1 eV, was used as the back electrode in this experiment.

4. Conclusions

In this study, the layer thickness, defect density, and interface defect density for an all-inorganic PSC device were optimized using CsPbI_xBr_3−x, ZnO, and NiO_x as the inorganic absorber layer, ETL, and HTL, respectively. The optimal thicknesses were determined to be 300 nm for ZnO, 1000 nm for the absorption layer, and 60 nm for NiO_x. Based on this optimized structure, a V_oc of 1.05 V, J_sc of 25.56 mA/cm², FF of 75.75%, and PCE of 20.26% were achieved. In addition, the effects of the absorber defect density and interface defect density between ETL/PAL and PAL/NiO_x were analyzed. Based on this, maintaining the absorber layer defect density at approximately 10¹⁵ cm⁻³ and the interface defect density at 10¹¹ cm⁻³ or less is recommended. Through optimization, we achieved a V_oc of 1.09 V, J_sc of 25.58 mA/cm², FF of 82.89%, and PCE of 23.07%. Thus, this study on the optimization of each factor of a solar cell is expected to improve the performance of inorganic PSC devices.