Computational study of defect engineered K₂GeI₆ based perovskite solar cell using multiscale modeling and machine learning

K₂GeI₆ is introduced as a novel, lead-free halide double perovskite designed to address the increasing demand for stable and eco-friendly photovoltaic technologies. K₂GeI₆ exhibits favourable electronic and optical properties as confirmed through DFT-based analyses, which highlight its suitable bandgap, strong absorption, and favourable other optical properties. Using a multiscale optimization strategy that integrates SCAPS-1D simulations with machine-learning-assisted defect engineering, we refined the FTO/WS₂/K₂GeI₆/MoO₃/Au architecture to suppress recombination losses and enhance carrier transport. SCAPS-1D results validate the strong optoelectronic performance of the absorber, while ML-guided defect introduction proves lesser degradation even at elevated interface defect densities, demonstrating remarkable defect resilience and operational robustness under real environmental conditions. With an excellent PCE of 27.83%, FF of 85.78%, Jsc of 30.58 mA/cm², and Voc of 0.98 V, the device establishes K₂GeI₆ as a promising lead-free candidate for next-generation high-performance photovoltaic applications, where defect tolerance, thermal stability, and long-term operational reliability are critical. The novelty of this work lies in the synergistic integration of DFT insights, SCAPS-1D device optimization, and ML-driven defect engineering to deliver a comprehensive computational framework for designing defect-resilient K₂GeI₆ perovskite solar cells.