Sub-Millisecond and Energy-Efficient Electrochemical Synaptic Transistors with a Partially Reduced Graphene Oxide...

Area of Science:

• Materials Science
• Neuroscience
• Computer Science

Background:

• Artificial synaptic transistors emulate biological synapses for advanced AI.
• Existing devices face challenges in efficiency and performance.

Purpose of the Study:

• To propose and characterize a novel synaptic transistor based on Nafion and partially reduced graphene oxide (prGO).
• To evaluate its potential for neuromorphic computing and artificial intelligence.

Main Methods:

• Fabrication of a synaptic transistor with a prGO channel and Nafion electrolyte.
• Electrical characterization of conductance states, linearity, and plasticity.
• Implementation in a two-layer perceptron artificial neural network for image recognition.

Main Results:

• The Nafion/prGO transistor exhibited over 200 distinct conductance states with low energy consumption (10-50 pJ).
• Achieved 90% image recognition accuracy, increasing to 94% with current pulses, due to linear and symmetric potentiation/depression.
• Demonstrated short-term plasticity, including paired-pulse facilitation and spike-timing-dependent plasticity.

Conclusions:

• The Nafion/prGO synaptic transistors show superior properties for neuromorphic computing.
• These devices offer a promising pathway for developing next-generation AI architectures.
• The electrochemical operation based on prGO and proton-assisted Nafion electrolyte enables efficient synaptic emulation.