A SmNiO₃ memristor with artificial synapse function properties and the implementation of Boolean logic circuits.

Area of Science:

• Materials Science
• Neuroscience
• Computer Engineering

Background:

• The demand for efficient data processing in artificial intelligence necessitates novel computing paradigms.
• Memristor devices simulating neural networks are crucial for AI applications, but face challenges in biological function simulation and circuit design.

Purpose of the Study:

• To present a perovskite rare earth nickelate memristor with advanced electrical properties.
• To demonstrate the device's capability in mimicking biological synaptic functions.
• To integrate the memristor into logic gate circuits for AI applications.

Main Methods:

• Fabrication of a memristor device using perovskite rare earth nickelates (RNiO3).
• Characterization of electrical performance, including switching ratio and repeatability.
• Experimental simulation of synaptic behaviors like EPSC, PPF, STDP, STP, and LTP.
• Integration of synaptic and neuron devices to form logic gate circuits (AND, OR, NOT).

Main Results:

• The memristor exhibited a three-orders-of-magnitude current switching ratio and good repeatability.
• Bidirectional conductance regulation mimicking bio-synaptic weight modulation was achieved.
• Comprehensive synaptic characteristics, including plasticity and refractory period, were successfully mimicked.
• Logic gate functions (AND, OR, NOT) were realized using the developed devices.

Conclusions:

• The perovskite rare earth nickelate memristor shows significant potential for AI applications.
• The device's ability to emulate synaptic functions and form logic circuits paves the way for high-density AI hardware.
• This work addresses challenges in biological function simulation and circuit design for neuromorphic computing.