Text-Embedding-Assisted Design of Rigid Molecular Cations for Suppressing Ion Migration in Hybrid Single-Crystal X-ray Detectors

Hybrid single crystals exhibit remarkable optoelectronic properties that make them highly promising for photovoltaic devices and radiation detectors. However, ion migration-induced instability represents a critical barrier to their commercial viability. By integrating large language models (LLMs) with k-Nearest Neighbor (kNN) algorithms, we develop a machine-learning model that identifies rigid organic cations as effective modulators for perovskite crystal stiffness, thereby suppressing ion migration. Guided by the analysis, we replaced the flexible alkyl chains in (HDA)BiI 5 (HDA = 1,6-hexanediamine) with rigid carbon rings to synthesize a highly stable (CHDA)BiI 5 single crystal (CHDA = trans-1,4-diaminocyclohexane). Density functional theory (DFT) calculations revealed that the rigid CHDA molecule exhibits ordered vibrations and stronger interactions with the inorganic framework compared to the disordered vibrations of HDA. Solid-state nuclear magnetic resonance (SSNMR) spin–lattice relaxation measurements further confirmed enhanced lattice rigidity, with the relaxation rate decreasing from 1.29 s –1 to 0.15 s –1, enhancing lattice rigidity. Consequently, the ion migration activation energy increased substantially from 0.42 to 0.61 eV. The resulting (CHDA)BiI 5 X-ray detector achieved an exceptional sensitivity of 8209 μC·Gy air –1 ·cm –2 at 175 V/mm and a low detection limit of 4.7 nGy s –1 . This study underscores the pivotal role of organic cation rigidity in optimizing the structural stability and functional performance of low-dimensional hybrid semiconductors.