Towards High-Efficiency Inverted CH3NH3GeI3 Perovskite Solar Cells Authors: Hong-Tao Li, Kang Yan, Jin Wang, Shuang-Shuang Zhang, Peng-An Zong, Xiao-Dong Feng Published: 2026 ABSTRACT The performance of inverted CH3NH3GeI3 (MAGeI3) perovskite solar cells incorporating both a hole transport layer (HTL) and an electron transport layer (ETL) was investigated using the Solar Cell Capacitance Simulator (SCAPS). Three candidate HTLs, including PEDOT:PSS, MoS2, and WS2, along with five ETLs including PCBM, TiO2, IGZO, ZnO, and SnO2, have been systematically evaluated. The analysis shows that WS2 and SnO2 provided the most favorable hole and electron transport, respectively. To improve device efficiency, the absorber layer thickness, defect density in MAGeI3, doping levels of WS2 and SnO2, as well as the interface defect densities and the work function of indium tin oxide (ITO), have been systematically studied. The optimal absorber layer thickness is determined to be approximately 900 nm. The optimal doping density of both WS2 and SnO2 is 1 × 1019 cm−3. The MAGeI3 layer should maintain a defect density as low as 1 × 1015 cm−3, and the defect densities at MAGeI3 interfaces should remain at 1 × 1015 cm−2. Additionally, an ITO work function of at least 5.2 eV is necessary to prevent the formation of a Schottky barrier at the ITO/WS2 interface. The simulated power conversion efficiency (PCE) can reach 22.9% under these optimized conditions. Our simulation results offer a viable route to develop high-efficiency MAGeI3 perovskite solar cells. 1. Introduction Organic–inorganic hybrid perovskites have attracted considerable attention as photovoltaic absorbers because of their excellent optoelectronic properties, such as strong light absorption, efficient carrier transport, long carrier diffusion length, and adjustable band gaps [1,2,3,4]. Since perovskite solar cells (PSCs) were first reported in 2009 [5], continuous advances in material design, film preparation, and device engineering have led to a rapid improvement in their photovoltaic performance, particularly the power conversion efficiency (PCE) [6,7,8,9,10]. During the past 17 years, the PCE of PSCs has improved remarkably, increasing from the initial efficiency of 3.8% to a certified record value of 27.3% [5,11]. Currently, lead-based perovskite materials with an ABX3 crystal structure have been broadly utilized as the light-absorbing layer in high-efficiency PSCs. As monovalent cations, CH3NH3+(MA+) and HC(NH2)2+(FA+) can occupy position A. Position B can be occupied by divalent metal ions, with Pb2+ being a typical example; position X is occupied by halide anions, such as Cl−, Br−, and I− [12,13]. Although the PCE of Pb-based PSCs has been improved substantially, one serious problem of PSCs is that the toxicity of Pb can handicap their potential application by causing serious harm to environmental safety and human well-being [14]. Consequently, developing Pb-free PSCs is essential to achieve photovoltaic performance comparable to that of Pb-based PSCs [15,16]. As elements in the same group as Pb in the periodic table, germanium (Ge) and tin (Sn) are considered viable in the perovskite structure [17,18]. Sn-based perovskites have received extensive attention and have achieved much higher efficiencies than Ge-based perovskites [19,20,21]. However, Sn2+ can be oxidized to Sn4+ during film formation and device operation, which is usually accompanied by tin vacancies, high background carrier concentration, and increased nonradiative recombination [22]. Ge-based perovskites, such as CH3NH3GeI3 (MAGeI3), are also Pb-free and may reduce the toxicity concern associated with Pb-based PSCs [17,23]. Compared with Sn-based perovskites, MAGeI3 is still at a much earlier stage of development. In 2015, Krishnamoorthy et al. first fabricated MAGeI3-based solar cells, and their PCE was only 0.20% [23]. The main challenges of MAGeI3-based PSCs include limited experimental efficiency, difficulty in preparing high-quality films, and possible oxidation of Ge-containing perovskites under ambient conditions [17,23]. Therefore, although Ge oxidation has received less attention than the well-known Sn2+/Sn4+ oxidation in Sn-based perovskites, it should still be regarded as an important stability issue for MAGeI3-based devices. Since only limited experimental progress has been reported for MAGeI3 solar cells, numerical simulation is useful for identifying key parameters and guiding future device optimization. Planar heterojunctions have become the mainstream structure for perovskite solar cells. This structure is fundamentally divided into two types: the conventional structure (n-i-p) and the inverted structure (p-i-n). Compared to the conventional structure, the inverted structure can be fabricated at lower temperatures through a more simplified process, thereby achieving superior stability and reduced current hysteresis effects [24,25,26]. Inverted structures are widely employed in Pb-based perovskite solar cells. Jiang et al. [27] reported an inverted device achieving a maximum PCE of 25.5%. Theoretically, researchers have studied the normal structure of MAGeI3-based PSCs with different hole transport layers (HTLs) and electron transport layers (ETLs) [28,29,30,31], as summarized in Table 1. However, theoretical investigations of inverted MAGeI3-based perovskite solar cells remain scarce [32]. The previous simulation studies listed in Table 1 mainly focused on normal MAGeI3-based device configurations or on the influence of selected transport layers. However, the performance of inverted MAGeI3 solar cells is not determined by a single layer parameter, but by the combined effects of carrier-selective contacts, interfacial band offsets, absorber quality, transport-layer doping, interface recombination, and electrode work function. Therefore, a more integrated analysis is still required to clarify how these factors jointly govern the device performance of inverted MAGeI3-based PSCs. In this work, an inverted MAGeI3 perovskite solar cell structure is investigated using the Solar Cell Capacitance Simulator (SCAPS). Indium tin oxide (ITO) is used as the front electrode for hole collection, while silver (Ag) serves as the back electrode for electron collection [33]. PEDOT:PSS is considered because it has been widely used as an HTL in inverted PSCs [34]. MoS2 and WS2 are examined as inorganic HTLs because two-dimensional transition metal dichalcogenides (TMDs) have been investigated as functional semiconductor layers in perovskite-related optoelectronic devices [35]. In particular, thin MoS2 films and water-soluble transition metal dichalcogenides have been used as hole transport layers in planar or p-i-n PSCs [36,37]. WS2 is further considered because WS2 thin films show useful optoelectronic properties, including electrical conductivity, which is relevant to carrier transport in solar-cell devices [38]. In addition, WS2/perovskite heterostructures have been reported to influence perovskite film growth and interfacial charge behavior through van der Waals interactions [39,40,41]. For electron extraction, PCBM, TiO2, IGZO, ZnO, and SnO2 are compared as candidate ETLs, since the ETL plays an important role in electron collection and interfacial recombination control [42]. Different from previous studies that mainly optimized limited structural parameters, this work evaluates the effects of HTL/ETL selection, absorber thickness, transport-layer doping density, absorber defect density, interfacial defect density, and ITO work function within the same inverted device framework. This combined optimization provides a clearer understanding of the physical factors limiting MAGeI3-based inverted PSCs and identifies parameter ranges that may guide further experimental development. 2. SCAPS Simulation Methodology and Device Arrangements 2.1. Numerical Method SCAPS (version 3.3.11), developed by Burgelman et al. at Ghent University [43], is a widely utilized tool for simulating solar cells. The software operates based on Poisson’s equation (Formula (1)), the continuity equations (Formula (2)), and the drift-diffusion equations (Formula (3)). By solving these equations under defined boundary conditions, SCAPS establishes a quantitative relationship between current density and voltage, enabling the determination of fundamental performance parameters of solar cells.(1)𝜕2φ𝜕x2=qϵn−p(2)𝜕n𝜕t=1q𝜕Jn𝜕x+G−R 𝜕p𝜕t=−1q𝜕Jp𝜕x+G−R(3)Jn=qDn𝜕n𝜕x−qμnn𝜕φ𝜕x Jp=−qDp𝜕p𝜕x−qμpp𝜕φ𝜕x 2.2. Device Structure The device structure illustrated in Figure 1a is glass substrate/ITO/HTL/MAGeI3/ETL/Ag, where the HTL includes MoS2, WS2, and PEDOT:PSS, and the ETL includes PCBM, IGZO, TiO2, ZnO, and SnO2. Initially, PCBM is selected as an ETL, which is widely used in PSCs, and the corresponding energy level alignment is presented in Figure 1b. 2.3. Material Parameters The parameters adopted for the device simulations were collected from the reported literature and the SCAPS material database [44,45,46,47], as listed in Table 2. Table 3 presents the defect parameters at the MAGeI3 interfaces, while Table 4 lists the initial material parameters of the ETLs. The optical absorption of MAGeI3 was described using the default SCAPS absorption model, in which the absorption coefficient follows a square-root dependence on photon energy above the bandgap, with Eg = 1.90 eV. The default illumination conditions are AM1.5G and a 300 K temperature. To validate the simulation parameters, we simulated a reported MASnI3 device made by Noel et al. [48]. Figure 2 shows that the simulated current–voltage (J-V) characteristics closely match the experimental data, indicating that the initial simulation parameters are reasonable. 3. Results and Discussion 3.1. Impact of Different HTLs Figure 3 compares the current density voltage (J-V) characteristics of devices using PCBM as the ETL and WS2, MoS2, or PEDOT:PSS as the HTL. Table 5 shows that the WS2-based device gives the highest simulated PCE of 9.46% among the three investigated HTLs. Its open-circuit voltage (VOC), short-circuit current density (JSC), and fill factor (FF) are 0.89 V, 13.28 mA/cm2, and 80.04%, respectively. These results indicate that the HTL strongly affects the photovoltaic performance through band alignment and carrier selectivity at the MAGeI3/HTL interface. Previous band offset studies have shown that proper valence-band offset (VBO) regulation at the perovskite/HTL interface is important for efficient hole extraction and recombination suppression [49,50]. Among the investigated HTLs, the WS2 device shows the highest simulated VOC and PCE. The higher VOC can be related to reduced interfacial recombination, while the comparable FF values indicate that carrier extraction remains effective at the MAGeI3/WS2 interface. The JSC values of the three devices are relatively close, suggesting that the HTL mainly influences carrier collection rather than optical absorption under the same absorber thickness. As shown in Figure 1b, WS2 forms a favorable VBO with MAGeI3, which facilitates hole transport. In addition, the larger conduction band barrier between WS2 and MAGeI3 helps block electrons and suppress interfacial recombination. Therefore, WS2 is selected as the HTL for the following analysis under the present simulation conditions. 3.2. Impact of Different ETLs The J-V curves in Figure 4 compare the devices employing different ETLs, including PCBM, TiO2, IGZO, ZnO, and SnO2, and their band alignment diagrams are shown in Figure 5. Table 6 shows that the SnO2-based device gives the highest simulated PCE of 9.53% among the five investigated ETLs. As shown in Table 6, the JSC values of the devices with different ETLs remain close, indicating that the ETL has a limited influence on photocurrent generation under the same MAGeI3 absorber thickness. In contrast, VOC, FF, and PCE show more obvious variations, suggesting that the ETL mainly affects device performance by modulating electron extraction and interfacial recombination at the ETL/MAGeI3 interface. The conduction-band offset (CBO) across the ETL/MAGeI3 junction can influence carrier recombination [51,52]. CBO is determined by the conduction-band energy difference across the ETL/perovskite junction. Previous studies have shown that CBO optimization is important for selecting suitable transport layers in MAGeI3 solar cells and for improving JSC, VOC, FF, and PCE through band alignment control [50]. As shown in Figure 6a, a negative CBO results in a cliff-like band alignment when the conduction-band minimum of the ETL is lower than that of the perovskite absorber. In contrast, a positive CBO corresponds to a spike-like band alignment, as illustrated in Figure 6b. Typically, a cliff structure can accumulate more electrons at the interface, accelerating carrier recombination, leading to a decrease in VOC and PCE [53]. The ETL of ZnO shows the lowest efficiency because the largest cliff-like structure (−0.28 eV) is formed at the interface. When the barrier of a low spike structure (e.g., CBO = 0.1 eV) is formed, the interfacial carrier recombination can be diminished, and the device performance will be improved [28]. At the SnO2/MAGeI3 interface, a small spike-like barrier of 0.08 eV is established, which helps suppress interfacial recombination and contributes to the improved PCE. Moreover, SnO2 exhibits a relatively large VBO with respect to MAGeI3 and thus acts as a hole reflector, which suppresses recombination and improves device performance. Therefore, SnO2 is selected as the ETL for the subsequent simulations under the present simulation conditions. 3.3. Performance Dependence on Absorber Thickness To evaluate the role of absorber thickness, the device performance was analyzed under different MAGeI3 thicknesses, as shown in Figure 7. When the absorber thickness varies from 100 to 1500 nm, the VOC shows only a slight decrease from 1.05 to 1.01 V, whereas the JSC changes much more obviously and reaches 17.80 mA/cm2 at 1300 nm. This trend is mainly attributed to the stronger light-harvesting capability of a thicker perovskite layer, which contributes to an increase in JSC. However, once the absorber layer becomes excessively thick, the transport path of carriers becomes longer, which increases the probability of recombination and causes JSC to decline beyond 1300 nm. As shown in Figure 7c, the FF gradually decreases over the absorber thickness range of 200 to 1500 nm, which can be attributed to the longer carrier transport path and enhanced recombination loss in the thicker absorber layer within the SCAPS model. Finally, according to Figure 7d, the optimum efficiency of 14.29% is obtained when the absorber thickness is 900 nm. 3.4. Impact of the Doping Density of ETL and HTL For device optimization, controlling the doping density in ETL (SnO2) and HTL (WS2) is essential. In our work, their effects are examined over the range of 1 × 1014 to 1 × 1019 cm−3, as presented in Figure 8. At relatively low doping levels, the device PCE increases as the density rises. At a doping density of 1 × 1019 cm−3 for SnO2 and WS2, the PCE reaches a maximum efficiency of 14.83%. Fundamentally, as the doping densities of SnO2 and WS2 increase, the built-in electric field near the MAGeI3 interfaces is strengthened, which facilitates electron–hole pair separation and improves device performance. Furthermore, the results indicate that, compared to the doping concentration of SnO2, an increase in the doping concentration of WS2 has a more significant impact on device performance. 3.5. Impact of MAGeI3 Absorber Defect Density To evaluate the impact of bulk defects on device performance, the defect density in the MAGeI3 absorber is varied from 1 × 1014 to 1 × 1018 cm−3. As shown in Figure 9, increasing the defect density beyond 1015 cm−3 causes a significant deterioration in photovoltaic performance, and the PCE drops from 16.03% to 2.84% at 1 × 1018 cm−3. By comparison, reducing the defect density below 1015 cm−3 leads to only a slight improvement. This trend is attributed to the increased trap-assisted Shockley–Read–Hall recombination in the MAGeI3 absorber layer, since a higher defect density introduces more recombination-active trap states. Therefore, 1 × 1015 cm−3 is adopted as the optimized bulk defect density for the subsequent simulations. 3.6. Influence of Defect Densities at MAGeI3 Interfaces Figure 10 shows that the defect densities at the SnO2/MAGeI3 and MAGeI3/WS2 interfaces have a significant effect on device efficiency. As the defect density decreases from 1 × 1019 to 1 × 1015 cm−2, the efficiency increases from 14.35% to 16.03%. However, when the defect density falls below 1 × 1015 cm−2, the improvement in efficiency becomes negligible. Moreover, the MAGeI3/WS2 interface has a stronger influence on device performance than the SnO2/MAGeI3 interface. This difference can be attributed to different carrier distributions, interfacial defect states, band alignment, and recombination processes at the two interfaces. At the MAGeI3/WS2 interface, interfacial defects can act as recombination centers for photogenerated carriers, which increases recombination loss and degrades device performance. Accordingly, the defect density of the MAGeI3 interfaces is set to 1 × 1015 cm−2 for the subsequent simulations. 3.7. Impact of ITO Work Function In addition to transport layer parameters, the work function of ITO plays an important role in device performance. The work function of ITO can be effectively tuned through surface treatments, including plasma treatment, surface molecular modification, and chlorination [54,55,56]. As shown in Figure 11, increasing the ITO work function from 4.8 to 5.2 eV enhances the efficiency from 16.03% to 22.9%. Once the ITO work function exceeds 5.2 eV, its influence on device performance becomes negligible. When the work function falls below 5.2 eV, a pronounced energy level mismatch emerges at the ITO/WS2 interface, which induces a Schottky barrier and suppresses hole extraction. Therefore, the ITO work function should be tuned to 5.2 eV or larger, and the device can reach the optimum efficiency of 22.9%. 4. Conclusions This research presents a simulation study of inverted MAGeI3 solar cells with a configuration of glass substrate/ITO/HTL/MAGeI3/ETL/Ag using SCAPS. Three HTLs and five ETLs have been carefully investigated. Under the simulation conditions investigated, WS2 provides the highest simulated performance among the considered HTLs, which is related to its favorable VBO with MAGeI3 and electron blocking ability. SnO2 provides the highest simulated performance among the considered ETLs, mainly because the band alignment at the SnO2/MAGeI3 interface helps suppress interfacial recombination and improve carrier selectivity. The effects of MAGeI3 absorber thickness, ETL and HTL doping densities, absorber defect density, interfacial defect density, and ITO work function were then systematically studied. The results indicate that the MAGeI3 layer achieves its optimum thickness at 900 nm. The optimal doping density of both WS2 and SnO2 is 1 × 1019 cm−3. It is important to maintain a low defect density of 1 × 1015 cm−3 in the absorber layer, and the defect densities at the MAGeI3 interfaces should be kept as 1 × 1015 cm−2 or lower. When the ITO work function is maintained at 5.2 eV, Schottky barrier formation at the ITO/WS2 interface is suppressed, allowing the device to achieve a peak efficiency of 22.9%. Beyond these parameter optimizations, the simulation results provide design guidance for inverted MAGeI3 PSCs. The device performance is governed by the coupled effects of transport layer band alignment, absorber quality, interface recombination, and electrode contact properties. Therefore, further improvement requires simultaneous optimization of carrier-selective contacts, MAGeI3 film quality, and interface passivation. Compared with the previously reported experimental MAGeI3 solar cell with a PCE of 0.20%, the simulated efficiency obtained in this work suggests that inverted MAGeI3 PSCs still have considerable potential for performance improvement. It should be noted that the PCE of 22.9% was obtained from SCAPS simulation under the adopted material parameters and investigated parameter ranges. Therefore, this value should be regarded as the theoretical potential of the proposed device structure rather than a direct prediction of experimentally achievable efficiency. Further experimental studies are still needed to examine the practical feasibility of the proposed device structure and optimize the MAGeI3 film quality, interface quality, and relevant device parameters.