Defect‐Mediated Optical Gating in Phase‐Engineered Ferroelectrics for Machine‐Learning‐Assisted Optical Encryption

ABSTRACT Optical data storage and anti‐counterfeiting technologies require materials that combine high‐contrast photoswitching with computational readout. Herein, we present a phase‐transition‐driven defect engineering strategy in Sm‐doped Pb(Zr,Ti)O 3 ‐based ferroelectric ceramics to achieve defect‐mediated optical gating for high‐contrast luminescence modulation. By manipulating the phase evolution from the rhombohedral ( R ) to the tetragonal ( T ) phase, we precisely tailor the concentration of intrinsic deep‐level oxygen vacancies, driven by the unique lattice distortion and Schottky‐type defect equilibria of the T ‐phase. These abundant defects facilitate the formation of photo‐induced color centers, which function as efficient “optical shutters” to significantly enhance photochromism and luminescent modulation. Time‐resolved spectroscopy shows that the Sm 3+ lifetime remains nearly unchanged after coloration, supporting a predominantly static, reabsorption‐dominated optical gating process rather than dynamic excited‐state quenching, while a photoluminescence modulation depth of over 70% is achieved. Leveraging this robust dual‐mode optical response, we demonstrate a machine‐learning‐assisted polymorphic arithmetic encryption system that integrates computational image readout with a hierarchical triple‐lock encryption protocol. A customized YOLOv8 model achieves more than 97% recognition accuracy within the acquisition variations represented in an independent experimental test set. This work bridges atomic‐scale defect engineering with macroscopic intelligent photonics, paving the way for advanced optical information systems.