Charge Loss in Hybrid Metal Halide Perovskites from Hydrogen Defects: Machine Learning Aided Nonadiabatic Molecular Dynamics

Synergy between ion motion and nonradiative recombination constitutes an instability pathway that undermines the optoelectronic performance of metal halide perovskites, with prior studies largely centered on rapidly diffusing halide ions. However, the interplay between rotational and torsional motions of organic cations, such as methylammonium (MA = CH 3 NH 3 + ), and hydrogen defects in hybrid organic–inorganic perovskites has remained unexplored. Here, we employed machine learning force-field-assisted nonadiabatic molecular dynamics to investigate, on the nanosecond time scale, the coupling between MA rotation and nonradiative recombination mediated by carbon-site hydrogen vacancies (V H–C ). The results reveal that fluctuations of V H–C defect state levels are strongly modulated by MA rotation and torsion, primarily governed by the φ(H–C–N–H) dihedral angle defined by the two residual hydrogens at the C site and the C–N bond. At low φ(H–C–N–H) values, the V H–C defect state resides near the conduction band minimum, whereas increasing φ(H–C–N–H) drives it toward the valence band maximum. On the nanosecond time scale, MA motion places the defect state in the midgap region. The coupling between MA rotation and V H–C defect state enhances electron–vibrational interactions and accelerates charge losses. The demonstrated synergy between the organic cation motions and hydrogen defects represents a previously unrecognized pathway for nonradiative charge recombination, highlighting a critical challenge to the optoelectronic performance of hybrid organic–inorganic perovskites.