Exploring radiation damage in halide perovskites through computational methods

Understanding the behavior of materials under irradiation is crucial for the design and safety of nuclear reactors, spacecraft, and other radiation environments. One key parameter in assessing radiation damage is the average threshold displacement energy (Ed), which quantifies the minimum transferred kinetic energy required to displace an atom from its lattice site, generating a permanent defect. We calculate Ed using ab initio molecular dynamics (AIMD) for lead halide perovskites MAPbI₃, FAPbI₃, and CsPbI₃. The Ed values we obtain, which are considerably lower than those commonly assumed in the literature for several species, hold significant implications for predicting radiation damage to these materials. We perform Monte Carlo simulations using the software Stopping and Range of Ions in Matter (SRIM) with the AIMD-obtained Ed and compare them to simulations using the default Ed. Our results show an increase in specific vacancies in halide perovskites that is not captured in simulations using the default Ed values. This suggests that the default Ed values may not accurately predict the radiation damage in these materials, potentially leading to underestimation of the damage. These results are discussed in the context of the radiation hardness of the materials and suggest they get damaged when exposed to radiation, but they are resilient due to self-healing processes. Additionally, to complement this work we calculate the electronic stopping power (Sₑ) curves for a hydrogen projectile traveling through two perovskites, FAPbI₃ and FA(Pb₀.₈₇₅Sr₀.₁₂₅)I₃, using first-principles time-dependent density functional theory (TD-DFT) and SRIM. Both methods yield similar results, indicating that 12.5% Sr doping in FAPbI₃ does not significantly impact the unchanneled Sₑ. TD-DFT simulations provide detailed insight into the initial interactions between the projectile and the perovskite material, enabling a comparison of the roles of anions and cations in radiation processes for both pristine and Sr-doped FAPbI₃. Finally, since traditional methods for determining Ed involve computationally expensive simulations or complex experiments. This project proposes a novel approach leveraging machine learning techniques, specifically the Sure Independence Screening and Sparsifying Operator (SISSO) method, to develop analytic expressions for predicting the average Ed in a wide variety of materials.