In this study, we present a groundbreaking approach to designing high-efficiency solar cells by combining nanophotonics with machine learning (ML). Our objective is to develop a plasmonic-enhanced solar cell that incorporates 2D material heterostructures such as MoS₂ and graphene. These advanced materials can be coupled with silver (Ag) nanoparticle arrays to boost light absorption through localized surface plasmon resonance (LSPR). To fine-tune the design, we employed AI-assisted simulations that predict the optimal configuration of nanoparticle size, inter-particle spacing, and the thickness of the 2D material layers.

The simulation process is based on finite-difference time-domain (FDTD) modeling, implemented in Python using the pyFDTD library. This approach allowed us to simulate and analyze how light interacts with the plasmon-enhanced solar cell structure. We then trained a neural network using TensorFlow, drawing from a dataset of over 1,000 simulation results. This model predicts the maximum light absorption efficiency for varying design parameters, including nanoparticle radius, particle spacing, and the wavelength of incoming light.

Through this AI-guided process, we discovered that the ideal nanoparticle size is 42 nm with a spacing of 55 nm, which maximizes absorption efficiency in the visible light spectrum. The final design demonstrated an exceptional light absorption efficiency of 92.7% across the 400–700 nm wavelength range, surpassing traditional solar cell performance by more than 30%. Furthermore, integrating machine learning reduced simulation times by approximately 80%, offering a highly efficient and scalable solution for advancing solar energy technology.