Machine learning screening and DFT calculations for exploring Cs2Ag0.75X0.25BiCl6 (X=Na, K, Rb) double perovskites as underwater photovoltaic materials

Abstract

Underwater vehicles, sensors, and autonomous systems heavily rely on stable energy supplies, which remain a key limiting factor in their practical applications. Traditional silicon-based solar cells are restricted in underwater applications due to their low light absorption efficiency and poor mechanical stability. In this study, we employed machine learning to screen perovskite materials with suitable bandgaps and constructed a vacuum-ordered Cs₂AgBiCl₆ double perovskite model. Subsequently, through first-principles calculations, we systematically investigated the structural, electronic, optical, and mechanical properties of Cs₂Ag₀.₇₅X₀.₂₅BiCl₆ (X = Na, K, Rb) double perovskites to assess their potential in underwater photovoltaic devices. The results demonstrate that Cs₂Ag₀.₇₅X₀.₂₅BiCl₆ double perovskites exhibit excellent photovoltaic performance, particularly with bandgaps between 1.8 and 2.3 eV, making them ideal for underwater photovoltaic applications. Optical calculations show that these materials have a high light absorption coefficient, reaching up to 1.56×10⁵ cm⁻¹. Cs₂Ag₀.₇₅Na₀.₂₅BiCl₆, in particular, achieved a power conversion efficiency of 19.18 % with a bandgap of 1.83 eV, exhibiting high light absorption capacity. In terms of mechanical performance, Cs₂Ag₀.₇₅Rb₀.₂₅BiCl₆ displayed significant anisotropy, enhancing its mechanical stability under complex water flow conditions. In conclusion, the vacuum-ordered Cs₂Ag₀.₇₅X₀.₂₅BiCl₆ double perovskites exhibit excellent overall performance, providing important theoretical support for the development of efficient and reliable underwater photovoltaic materials.