Bayesian Optimization–Driven Design of Hydrogen Concentration in A-Si/H Multilayers for Silicon Heterojunction Solar Cells

Multilayer thin-film passivation structures provide high design flexibility but are inherently challenging to optimize due to the large number of process parameters and their coupled effects. In hydrogenated amorphous silicon (i-a-Si/H) passivation layers, hydrogen concentration plays a critical role in interfacial passivation and film quality, making it a physically meaningful parameter for systematic investigation. In this study, a three-layer i-a-Si/H passivation structure was examined as a representative example to verify an optimization approach that combines physical insight with Bayesian optimization. Rather than treating Bayesian optimization as a black-box search method, hydrogen concentration was explicitly incorporated into the experimental design. A comparative experiment was first conducted on the interfacial layer by varying hydrogen concentration over a wide range, followed by Bayesian optimization of the upper layers using hydrogen flow rate as the control variable. The results demonstrate that a graded hydrogen concentration profile is effective for passivation and validate the proposed approach as a practical framework for optimizing multilayer thin films.