Mixed-Dimensional Cu-Based Perovskites for Stable and Energy-Efficient Neuromorphic Memristors

Abstract

Synaptic memristors, which mimic the fundamental functional units of the human brain, offer a promising strategy for implementing brain-inspired (neuromorphic) computing. Among various materials, high-performance perovskites have emerged as key candidates for artificial neuromorphic devices due to their favorable electronic properties. However, the integration of perovskites into microelectronic systems faces significant challenges, primarily stemming from their intrinsic limitations, including high defect densities, environmental instability, and toxicity. Here, a controlled phase-modulation approach using low-dimensional (0D and 1D) copper halide perovskites (CHP) via solvent dripping is presented, which facilitates the formation of highly ordered mixed phases on a single thin film through regulated nucleation kinetics. This strategy effectively reduces interfacial recombination and enhances ion migration. The resulting device exhibits ultralow energy consumption in the picojoule regime, along with an on/off ratio of approximate to 10-5, enabling robust logic-state separation and supporting reliable non-volatile memory and multilevel data storage functionalities. These findings position lead-free perovskite memristors not only as sustainable alternatives but also as superior candidates capable of surpassing 2D materials in terms of scalability, power efficiency, and resistive switching reliability.