Constrained Bayesian Optimization for the Deposition Conditions of Doped a-Si:H with High Passivation Performance for SHJ Solar Cells

Ultrahigh conversion efficiency of silicon heterojunction (SHJ) solar cells requires high-quality passivation performance in both intrinsic hydrogenated amorphous silicon (i-a-Si:H) for chemical passivation and doped (n-type and p-type) a-Si:H for field-effect passivation. In this study, we report the effective determination of the deposition conditions for doped a-Si:H deposited by catalytic chemical vapor deposition using Bayesian optimization (BO) to maximize the passivation performance. The BO scheme proposed in this study comprises multiple prediction models to determine the deposition conditions that maximize effective minority carrier lifetime (τeff) while maintaining a-Si:H film thickness and conductivity within a certain range under the experimental equipment capabilities. In n-a-Si:H, the BO was started with 10 initial samples performed with random conditions, and the deposition conditions were optimized in 21 BO cycles, leading to a τeff of approximately 5.4 ms. In p-a-Si:H, BO was started with 13 initial samples, and deposition conditions were optimized in 7 BO cycles, resulting in a τeff of approximately 1.2 ms. The SHJ solar cells fabricated using a-Si:H films deposited under the optimized conditions had a τeff of 2.4 ms and an open-circuit voltage (VOC) of 0.701 V. The a-Si:H/crystalline Si samples with symmetric passivation stacks exhibited an implied VOC of > 0.7 V, indicating sufficient passivation performance.