Scholarlink

<p>This study provides a detailed investigation into the structural, electronic, optical, and photovoltaic properties of the lead-free double perovskite Cs<sub>2</sub>NaBiI<sub>6</sub> by integrating DFT, multi-platform device simulations (SCAPS-1D, wxAMPS, and COMSOL), and data-driven machine learning modeling. DFT calculations confirm a stable cubic Fm̅<sub>3</sub>m phase with a lattice constant of 12.586 Å, and energy bandgap of 2.20&#xa0;eV, indicating strong optical absorption in the visible region. The electronic band analysis reveals a lighter electron effective mass (mₑ = 0.617&#xa0;m₀) and higher mobility (u<sub>n</sub> = 28.5&#xa0;cm<sup>2</sup>/V·s) compared with that of holes (mₕ = 1.26 m<sub>o</sub>, u<sub>h</sub> = 14.0&#xa0;cm<sup>2</sup>/V·s), confirming electron-dominated transport. The effective density of states at 300&#xa0;K was found to be Nc = 1.21 × 10<sup>19</sup> cm<sup>−3</sup>&#xa0;and Nv = 3.54 × 10<sup>19</sup> cm<sup>−3</sup>. SCAPS-1D device simulation of the optimized AZO/Cs<sub>2</sub>NaBiI<sub>6</sub>/CuI configuration achieved a power conversion efficiency (PCE) of 23.31% with Voc = 1.23&#xa0;V, Jsc = 21.80&#xa0;mA/cm<sup>2</sup>, and fill factor (FF) = 86.63%. Further optimization of the absorber thickness (800&#xa0;nm), defect density (10<sup>15</sup> cm<sup>−3</sup>), and interface parameters resulted in an efficiency enhancement of 25.22%. Cross-validation with wxAMPS and COMSOL showed excellent agreement, confirming the model’s robustness. Machine learning-based regression models (Linear, SVM, Random Forest, and XGBoost) were trained on simulation datasets; XGBoost achieved superior accuracy (R<sup>2</sup> ≈0.9993, MSE = 0.010) for PCE prediction. Feature-importance analysis identified defect density, doping concentration, and active layer thickness as the most critical determinants of PV performance. These combined theoretical, simulation, and machine learning findings establish Cs<sub>2</sub>NaBiI<sub>6</sub> as a high-potential, and environmentally benign material for next-generation PSCs.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Unraveling the photovoltaic behavior of Cs2NaBiI6 double perovskite: a combined DFT, SCAPS-1D, wxAMPS, COMSOL and machine learning approach

  • 期刊论文

  • 发表日期：

    2025年12月12日

查看全文
  • Ghazi Aman Nowsherwan

摘要

This study provides a detailed investigation into the structural, electronic, optical, and photovoltaic properties of the lead-free double perovskite Cs2NaBiI6 by integrating DFT, multi-platform device simulations (SCAPS-1D, wxAMPS, and COMSOL), and data-driven machine learning modeling. DFT calculations confirm a stable cubic Fm̅3m phase with a lattice constant of 12.586 Å, and energy bandgap of 2.20 eV, indicating strong optical absorption in the visible region. The electronic band analysis reveals a lighter electron effective mass (mₑ = 0.617 m₀) and higher mobility (un = 28.5 cm2/V·s) compared with that of holes (mₕ = 1.26 mo, uh = 14.0 cm2/V·s), confirming electron-dominated transport. The effective density of states at 300 K was found to be Nc = 1.21 × 1019 cm−3 and Nv = 3.54 × 1019 cm−3. SCAPS-1D device simulation of the optimized AZO/Cs2NaBiI6/CuI configuration achieved a power conversion efficiency (PCE) of 23.31% with Voc = 1.23 V, Jsc = 21.80 mA/cm2, and fill factor (FF) = 86.63%. Further optimization of the absorber thickness (800 nm), defect density (1015 cm−3), and interface parameters resulted in an efficiency enhancement of 25.22%. Cross-validation with wxAMPS and COMSOL showed excellent agreement, confirming the model’s robustness. Machine learning-based regression models (Linear, SVM, Random Forest, and XGBoost) were trained on simulation datasets; XGBoost achieved superior accuracy (R2 ≈0.9993, MSE = 0.010) for PCE prediction. Feature-importance analysis identified defect density, doping concentration, and active layer thickness as the most critical determinants of PV performance. These combined theoretical, simulation, and machine learning findings establish Cs2NaBiI6 as a high-potential, and environmentally benign material for next-generation PSCs.