Perovskite Silicon Solar Cell Emulation using Multi-Layer Perceptron Deep Neural Network

Abstract

Perovskite silicon solar cells have an unprecedentedly high-power conversion efficiency compared to other solar cell technologies. This manuscript aims to accomplish two specific goals. The first objective is to investigate the impact of perovskite layer thickness and doping concentration on the solar cell's power conversion efficiency. The second one is to conduct a comparative study to identify the best artificial intelligence technique for simulating the complex nonlinear behaviour of the variation of material parameters versus power conversion efficiency. A solar cell capacitance simulator is used to examine the photovoltaic properties of perovskite silicon solar cells. The simulation is conducted in three stages. Firstly, studying the silicon base structure efficiency to determine the absorber layer c-Si(p) thickness and doping, and the buffer layer c-Si (n) thickness and doping. The second stage is the single-junction of solar cell structure in which c-Si (p++) is used as back surface field. Finally, the perovskite silicon solar cells study the impact of perovskite layer thickness and carrier concentration on power conversion efficiency. The efficiency increases linearly from 26.5% to 28.5% with the perovskite layer thickness. Solar cell behaviour is simulated utilizing multi-layer perceptron. It represents satisfied results.