The demand for bio-inspired neuromorphic systems drives research on artificial neuronal devices with dual excitatory/inhibitory capabilities. Although perovskite memristors show promise, interfacial barriers and rapid ion migration hinder modulation. This study addresses these challenges through atomic-scale cathode engineering, developing perovskite-based memristor (Au/Ni1-rGO/MAPbI3/ITO) with nickel single-atom modified reduced graphene oxide (Ni1-rGO) cathode. The Ni–O configuration of Ni1-rGO with a single Ni atom anchored on rGO, modulates the electronic properties of rGO, reducing Schottky barrier via energy band alignment for bipolar current symmetry and spatially confining iodide ion migration through a high diffusion energy barrier (2.912 eV), conferring 780 ms relaxation. The device achieves 1,000 distinct conductance states and emulates interlayer connections and bidirectional signal transmission in feedback neuronal networks. We demonstrate neuromorphic functionalities: unsupervised competitive learning exceeds 50% clustering accuracy, while cooperative learning solves NP-hard problem 6× faster than simulated annealing, establishing an atomic-engineered approach for high-efficiency neuromorphic hardware. Li et al. report feedback neurons based on perovskite memristors with a nickel single-atom modified reduced graphene oxide cathode. The device successfully implements an unsupervised learning network with over 50% clustering accuracy and cooperative learning for solving NP-hard combinatorial optimisation problem.
Li et al. (Mon,) studied this question.