Simulation, optimization, and machine learning strategies for CH₃NH₃PbBr₃ perovskite solar cells

Perovskite solar cells (PSCs) combine outstanding optoelectronic properties with low fabrication cost, with methylammonium lead bromide (CH₃NH₃PbBr₃) offering superior thermal stability, a 2.2 eV band gap, and a high absorption coefficient (10⁵–10⁶ cm⁻¹). This study employs SCAPS-1D simulations under AM1.5 G illumination to analyze an FTO/BaTiO₃/CH₃NH₃PbBr₃/Cu₂O/Ni device, achieving a 17.00% power conversion efficiency (PCE), 1.8515 V open-circuit voltage (V OC ), 9.923 mA cm⁻² short-circuit current density (J SC ), and 92.51% fill factor (FF), enabled by optimal band alignment and reduced recombination. Quantum efficiency (QE) reached ∼100% in the visible range, confirming strong light-harvesting. Parametric optimization identified optimal operation at 300 K with a shunt resistance of 10⁵ Ω·cm². Machine learning (ML) models; artificial neural networks (ANN) and k-nearest neighbors (k-NN) were applied to assess the influence of material properties on device performance. The results offer guidelines for fabricating cost-effective, high-performance Pb–based PSCs and reinforce CH₃NH₃PbBr₃’s role as a benchmark absorber for device optimization.