Cross-platform simulation DFT–SCAPS–wxAMPS–COMSOL coupled with data-driven modeling of CsInBr3 and CsTlBr3 optoelectronic devices

Engineering notes

Engineering notes will be added by the Power for Solar editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为光伏效率、钙钛矿太阳能电池、储能技术、太阳能热利用、BIPV、并网技术等高价值论文补充中文说明。

Original abstract

The pursuit of stable and environmentally benign alternatives to lead-based perovskite solar cells (PSCs) necessitates the development of efficient all-inorganic absorber materials supported by robust optimization strategies. In this study, the photovoltaic potential of cesium-based bromide perovskites, CsInBr 3 and CsTlBr 3 , is comprehensively evaluated using an integrated multiscale approach combining density functional theory (DFT), numerical device simulations (SCAPS-1D, wxAMPS-1D, and COMSOL), and machine learning (ML) modeling. DFT calculations reveal thermodynamically stable crystal structures with suitable electronic bandgaps and strong optical absorption in the visible region. For CsInBr₃, a favorable electron-dominated transport behavior is observed, with electron and hole mobilities of μₑ ≈ 12.7 cm 2 V⁻ 1 s⁻ 1 and μₕ ≈ 8.9 cm 2 V⁻ 1 s⁻ 1 . Device simulations employing an IGZO/CsInBr 3 /GO architecture yield an initial power conversion efficiency (PCE) of 23.31%, with an open-circuit voltage (Voc ≈ 0.86 V), short-circuit current density (Jsc ≈ 32.1 mA cm⁻ 2 ), and fill factor (FF ≈ 83.4%). Through systematic optimization of absorber thickness (≈ 400–800 nm), defect density (reduced to ~ 10 14 cm⁻ 3 ), doping concentration, interface quality, and parasitic resistances, the optimized device achieves a significantly enhanced PCE of 28.15%. It is important to note that these optimized performance metrics represent upper-bound estimates obtained under near-ideal optical and electrical conditions, including reduced defect densities and simplified recombination mechanisms. Furthermore, CsTlBr 3 -based photodetectors exhibit higher responsivity (~ 0.57 A/W) and detectivity (> 4 × 10 11 Jones) than CsInBr 3 , indicating enhanced light absorption, charge transport, and lower noise levels. Cross-platform validation using wxAMPS-1D and COMSOL demonstrates good agreement in the predicted J–V characteristics and performance metrics. Furthermore, ML regression models (Linear Regression, SVR, Random Forest, and XGBoost) were trained on the simulation dataset, with XGBoost delivering superior accuracy (R 2 > 0.998 and MSE ≈ 0.015 for PCE prediction). Feature-importance analysis identifies defect density, doping concentration, series resistance, and absorber thickness as the dominant parameters governing photovoltaic performance.

Links and sources

Comments