Molecular Dynamics Studies using Machine Learning to predict the effect of nano vacancies on electrical and mechanical performance of Silicon based solar cells

In the literature, significant electric power losses are documented due to defects like cracks, vacancies, and substitutional defects in Silicon-based photovoltaic solar cells. This study delves into power loss and fracture strength of the solar cells made of Silicon at nanoscale considering various factors such as vacancy size, type, inclination, and operating temperatures. Stress at fracture and total power produced by solar cell are predicted by the developed deep machine learning models. The data for machine learning model is adopted from the simulations carried out using LAMMPS software. Simulations are implemented by subjecting the Silicon domain to uniaxial mechanical load and operating temperatures of solar cell. Furthermore, these simulations encompass a range of conditions, including vacancies in different positions within the cell. Vacancies of four different sizes, across four orientations at the center, and at five temperatures are employed for studies. Subsequently, the deep learning models are trained and tested using data from these simulations. Validation against molecular dynamics results shows strong agreement, affirming the utility of the proposed models in estimating total power at fracture and the stress at fracture for photo-voltaic solar cells with vacancies of diverse characteristics and temperatures.