Harnessing Selective Defect Formation for Filament Control in Reliable and Energy‐Efficient Perovskite Resistive Random Access Memory

Abstract

Halide perovskite memristors are emerging as compelling platforms for in‐memory computing because their low ionic migration energies and compositional versatility enable tunable conductance states. However, previous studies have primarily focused on defect passivation to improve device reliability, with limited exploration of designs that harness defects as preferential pathways for controlled filament growth. Here, a defect‐engineering strategy based on the differential modulation of vacancy formation energetics is introduced wherein the incorporation of poly(quaternized dimethylaminoethyl methacrylate) iodide (PAQI), a quaternary‐ammonium polymer, into formamidinium lead triiodide (FAPbI 3 ) facilitates conductive filament formation. PAQI‐induced Pb–I bond weakening decreases the formation energy of iodine vacancies (V I ), selectively activating V I that contribute to the percolation pathways for metal cation transport during resistive switching. Furthermore, the functional groups of PAQI passivate the undercoordinated Pb 2+ in the perovskite, enabling an exceptionally low switching voltage ( V SET = 0.28 V) and SET energy consumption (3.64 pJ), along with stable retention (<4% current fluctuation). Using a 7 × 7 crossbar array, the device demonstrated selective recognition of optoelectronic input patterns and high‐fidelity edge detection, comparable to software convolution. These results highlight a transformative approach that exploits defects as functional elements in halide perovskites and offers scalable pathways for high‐performance memristive systems.