Additive-Engineered CsPbBr₃-Based Perovskite Memristors for Neuromorphic Computing and Associative Learning Applications

Perovskite memristors have emerged as promising candidates for neuromorphic computing due to their simple fabrication process and mixed ionic and electronic properties. Among them, all-inorganic CsPbBr₃perovskites have garnered significant interest due to their excellent stability. However, the low solubility of cesium bromide (CsBr) in most common solvents poses a major challenge in fabricating high-quality, pinhole-free CsPbBr₃films for memory device applications using a convenient one-step solution method. In this work, a facile one-step spin-coating approach was employed to fabricate CsPbBr₃-based memristors, incorporating a carbohydrazide (CBH) additive into the perovskite precursor to enhance device performance. The modified device exhibited an improved ON/OFF ratio, enhanced endurance, and longer retention time. Furthermore, it successfully emulated key synaptic functions, including excitatory postsynaptic current, paired-pulse facilitation, long-term potentiation/depression, and learning–forgetting–relearning behaviors, effectively mimicking biological synapses. Additionally, an associative learning experiment inspired by Pavlov’s dog experiment was conducted, demonstrating memory formation and extinction under optical and electrical stimuli. The fabricated perovskite memristor was further evaluated in a convolutional neural network for Fashion MNIST classification, achieving a high recognition accuracy of 89.07%, confirming its potential for neuromorphic computing applications. This study highlights the effectiveness of additive engineering as a strategy for developing high-performance perovskite-based neuromorphic electronics.