摘要:

Organic-inorganic metal halide perovskites have drawn a great deal of attention due to their supreme optical and electrical properties and their potential in future application in optoelectronic devices. Here, we carry out finite-difference time-domain (FDTD) simulation on different experimentally realistic structures of perovskite solar cells (PSC) and optimize their parameters with assistance of neural network (NN). We find an optimized structure with 30.48% enhancement comparing to planar structure and the fact that with properly design, 300-nm-thick nano-textured structure can outperform 900-nm-thick planar structure. We believe that light trapping structure is essential in thin film PSCs and also has a great potential in lead-free PSCs.