Exploration of highly stable and efficient pseudohalide double perovskite doping schemes for solar cells.

Double perovskites offer a chemically versatile platform for photovoltaic applications, yet the systematic design of pseudohalide doping strategies, including the selection of pseudohalide species, host compositions, and substitution levels, remains insufficiently explored. In this work, a machine-learning-assisted screening workflow combined with first-principles calculations was developed to explore pseudohalide doping schemes for double perovskites. By systematically screening combinations of pseudohalide species, host compositions, and substitution coefficients within a large compositional space, four promising BH4--based doping schemes with a low stoichiometric substitution coefficient of x = 0.125 were identified: Cs2AgInCl5.875(BH4)0.125, K2AgInCl5.875(BH4)0.125, Rb2AgInCl5.875(BH4)0.125, and Rb2AgGaCl5.875(BH4)0.125. Compared with their pristine counterparts, BH4--substituted systems exhibit lower formation energies and reduced energies above the convex hull. AIMD simulations further show preservation of the crystal frameworks over the investigated 10 ps time scale at 300 K, while ELF and RMSD analyses indicate modified local bonding environments associated with reduced atomic thermal fluctuations. Electronic-structure calculations reveal band-gap narrowing after substitution, and HSE06 calculations provide more reliable absolute band-gap estimates for the selected candidates. The substituted systems also exhibit enhanced visible-light absorption. COMSOL device simulations yield power conversion efficiencies above 20% for all four absorber materials, with a maximum simulated value of 30.79%. Overall, this work establishes a computational framework for the systematic exploration of pseudohalide doping schemes and identifies low-level BH4- substitution at x = 0.125 as a promising doping strategy for the four selected double-perovskite hosts.