Perov-AccTEST · AI-Driven Discovery of Stable Wide Band Gap Perovskite Materials for Perovskite-Silicon Multi-Junction Solar Cell

The conversion of solar energy into electricity is critical for reducing carbon emissions and expanding renewable energy generation. Photovoltaic (PV) technologies offer a compelling solution for grid-connected and building-integrated applications. Among emerging materials, halide perovskites have shown exceptional promise for next-generation PV devices. In particular, wide bandgap halide perovskites, when integrated in tandem with commercially available silicon solar cells, are revolutionizing the solar industry. These multi-junction architectures enable significantly higher power conversion efficiencies by more effectively harvesting the solar spectrum.However, the long-term operational stability of wide bandgap perovskites remains a key bottleneck in advancing perovskite–silicon tandem technologies. Traditional trial-and-error approaches to materials discovery are slow and labor-intensive, limiting progress. This project proposes a machine learning–guided closed-loop experimental framework to accelerate the identification of stable wide bandgap perovskite compositions. The workflow integrates robotic synthesis, high-throughput optical characterization, rapid data analysis, and predictive modeling to efficiently navigate complex multi-dimensional composition spaces.In Phase I, a Materials Acceleration Platform (MAP) will be deployed to screen diverse perovskite formulations. Machine learning algorithms will generate composition–quality maps, enabling targeted exploration without exhaustive testing. In Phase II, an Accelerated Testing Platform (ATP) will be used to evaluate the long-term stability of ML-predicted compositions under intensified stress conditions. Multiple samples will be tested in a high throughput manner to identify robust candidates. The most stable composition will then be advanced for integration into multi-junction perovskite–silicon solar cell prototypes.