Defect and Ion-Migration Regulation in MAPbI3 Memristors via ZIF-67 Modification toward Stable Resistive Switching and Neuromorphic Computing

High-performance resistive random-access memory (RRAM) based on organic–inorganic hybrid perovskites remains limited by intrinsic vacancy defects and uncontrolled ion migration, which usually induce random conductive filaments (CFs) formation and poor device stability. Herein, a defect-engineering strategy is proposed by using a cobalt-based zeolitic imidazolate framework (ZIF-67) as a multifunctional additive to synergistically regulate the MAPbI3 film through Co ion incorporation and ligand-assisted defect passivation. Structural, morphological, optical/electrical, and conduction analyses reveal that the introduction of ZIF-67 improves film quality by enlarging grains and reducing grain boundaries, while regulating Pb-vacancy-related defect states and introducing a stronger Co–I bond. These effects suppress disordered vacancy migration and promote stable CFs formation during resistive switching (RS). Consequently, the Cu/ZIF67-MAPbI3/p-Si device exhibits enhanced RS performance, including a high resistance ratio (>103), low operating power consumption (≈1.01 mW), endurance over 2300 cycles, and improved switching uniformity with reduced coefficients of variation for both high-resistance state and low-resistance state. Benefiting from its continuously tunable conductance, the device successfully emulates synaptic plasticity behaviors, including paired-pulse facilitation (PPF), spike-timing-dependent plasticity (STDP), and long-term potentiation/depression (LTP/LTD). Moreover, an experimentally calibrated multilayer perceptron achieves Modified National Institute of Standards and Technology (MNIST) handwritten digit recognition with an accuracy of 97.54% and a last loss of 3.35%, demonstrating the potential of ZIF67-MAPbI3-based RRAM device for artificial synapses and neuromorphic computing.