Tracing Ion Migration in Halide Perovskites with Machine Learned Force Fields

Area of Science:

• Materials Science
• Solid-State Physics
• Computational Chemistry

Background:

• Halide perovskite optoelectronic devices face challenges from chemical degradation and hysteresis.
• These issues are often caused by mobile charged defects within the material.
• Classical simulations struggle to accurately model the dynamic charge states of these defects.

Purpose of the Study:

• To investigate the diffusion mechanisms of charged halide defects in CsPbI3 perovskites.
• To develop and utilize machine-learned force fields for accurate atomic-scale simulations.
• To understand how defect migration influences the stability of perovskite devices.

Main Methods:

• Density functional theory (DFT) calculations were performed to train machine-learned force fields.
• Atomic scale molecular dynamics (MD) simulations were conducted using these trained force fields.
• The diffusion of various halide interstitial and vacancy defects (charged and neutral) in CsPbI3 was studied.

Main Results:

• Negative iodide interstitials and positive iodide vacancies exhibit similar migration rates at room temperature.
• Neutral iodide interstitials diffuse faster than neutral iodide vacancies.
• Oppositely charged interstitials and vacancies show significantly reduced mobility, especially under device operating conditions.

Conclusions:

• Machine-learned force fields enable accurate simulation of defect dynamics in halide perovskites.
• The mobility of halide defects, particularly charged ones, is a critical factor in perovskite device performance and longevity.
• Understanding defect diffusion is key to mitigating degradation and hysteresis in optoelectronic applications.