Tunable Resistive Switching in CsPbBr3 Nanocrystal-Based Memristors for Artificial Synapse and Neuromorphic Applications

Abstract

The growing demand for energy-efficient, brain-inspired computing has driven interest in memristors for neuromorphic hardware. All-inorganic halide perovskite cesium lead bromide (CsPbBr₃) is a promising material for memristor-based artificial synapses due to its mixed ionic-electronic conductivity, low activation energy of bromide vacancy, and superior defect tolerance. This study demonstrates tunable resistive switching properties of a forming-free memristor with CsPbBr₃ nanocrystals, achieving both digital (abrupt) and analog (gradual) switching for neuromorphic applications. The fabricated device exhibits stable non-volatile digital switching behavior with an ON/OFF ratio of 10³, endurance of 500 cycles, and a high retention time of 4000 s with relatively low SET and RESET voltage, along with displaying gradual conductance states with appropriate voltage pulses. The device replicates various key biological synaptic functionalities, including short-term plasticity, long-term plasticity, and paired-pulse facilitation, spike rate-dependent plasticity, which can be controlled by the amplitude and duration of the applied bias. The potentiation and depression characteristics are utilized to train an artificial neural network, achieving 93.2% classification accuracy for handwritten digit recognition. This work highlights a reliable method to control switching dynamics in CsPbBr₃ nanocrystal-based memristors, making them suitable for data storage and in-memory computing applications.