An improvement in solar cell characteristics of simple structured p-Si based solar cells has been achieved with an optimal diffusion process determined using Bayesian optimization. The simple solar cells possess a nanocrystalline Si (nc-Si) layer on the surfaces, which is formed with the Surface Structure Chemical Transfer method. A pn-junction and a boron doped back-surface-field (B-BSF) are formed by co-diffusion of phosphorus and boron atoms from phosphosilicate glass (PSG) and boron-containing compounds on the front and the rear surfaces, respectively. During the heat treatment of the diffusion process, the surfaces of the nc-Si layer are passivated with PSG. Bayesian optimization (BO) of the diffusion process has been performed to improve implied open-circuit voltage values which mainly depend on the pn-junction, the B-BSF, and PSG passivation qualities. Using BO, three experimental parameters of the diffusion process have been optimized in just 15 experiments. The best cell characteristics have been obtained by an improvement of PSG passivation.