Machine Learning Driven Strategies for Improving $\text{CsFASnI}_{3}$ Perovskite Solar Cell Efficiency
Year: 2026
Type: conference-paper
Abstract: Recently, photovoltaic devices utilizing CsFASnI${ }_{3}$are highly studied and popularized. They are highly effective, flexible and simple to fabricate. In addition, the material in question has a direct bandgap and exhibits strong absorption of light (approximately$\mathbf{1 0}^{\mathbf{5}} \mathbf{c m}^{\mathbf{- 1}}$), making it quite suitable for production of panels. By involving AFORS-HET simulations in our project, it is possible to examine the performance of a device having CeZnO as a buffer layer and$\mathbf{C r}_{\mathbf{2}} \mathbf{O}_{\mathbf{3}}$as hole transport debilitation (HTL). In particular, we arranged the device as follows: Al/FTO/CeZnO/CsFASnI${ }_{3} / \text{Cr}_{2} \mathrm{O}_{3} / \text{Au}$. What do we seek to achieve here? First of all, it is important to reduce the level of carrier recombination and provide additional stimulation. Then, we varied the thickness of the absorb layer, doping level and operational temperature to find the most suitable conditions. After performing a number of optimizations, we got the following results: Voc equals 1281 mV, Jsc amounts to 23.58$\text{mA} / \text{cm}^{2}, \text{FF}$reaches 80.88%, and power conversion efficiency makes 24.43%. However, further machine learning predictions allowed us to push power conversion efficiency up to 25.36%. It is superior compared to the usual solar cell.
Institutions: SR University, KIIT University, JSPM University
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7203769108