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

Area of Science:

• Materials Science
• Neuroscience
• Computer Engineering

Background:

• Perovskite memristors show promise for neuromorphic computing due to their properties.
• All-inorganic CsPbBr3 perovskites offer excellent stability but face fabrication challenges.
• Low solubility of CsBr hinders the creation of high-quality CsPbBr3 films.

Purpose of the Study:

• To develop a high-performance CsPbBr3-based memristor for neuromorphic applications.
• To address the solubility issue of CsBr in CsPbBr3 film fabrication.
• To demonstrate the potential of additive engineering in enhancing memristor performance.

Main Methods:

• A facile one-step spin-coating approach was used.
• A carbohydrazide (CBH) additive was incorporated into the perovskite precursor.
• The fabricated memristors were tested for synaptic functions and in a convolutional neural network.

Main Results:

• The modified memristor exhibited improved ON/OFF ratio, endurance, and retention time.
• The device successfully emulated synaptic functions like EPSC, PPF, LTP/LTD, and learning-forgetting behaviors.
• An associative learning experiment demonstrated memory formation and extinction.
• High recognition accuracy (89.07%) was achieved in Fashion MNIST classification using a CNN.

Conclusions:

• Additive engineering with CBH is an effective strategy for high-performance perovskite memristors.
• The developed CsPbBr3 memristor shows significant potential for neuromorphic computing.
• This work paves the way for advanced memory devices mimicking biological synapses.