Reproducible Ultrathin Ferroelectric Domain Switching for High-Performance Neuromorphic Computing

Area of Science:

• Materials Science
• Computer Science
• Neuroscience

Background:

• Current computing architectures face limitations.
• Neuromorphic computing offers a promising alternative.
• Electronic devices mimicking synaptic plasticity are key.

Purpose of the Study:

• To investigate ferroelectric tunnel junctions for electronic synapses.
• To emulate synaptic plasticity using these devices.
• To evaluate their performance in a neural network.

Main Methods:

• Utilized robust ferroelectric tunnel junctions.
• Demonstrated memristor function with ≈200 reproducible states via domain switching.
• Emulated short- and long-term plasticity by tuning pulse parameters.
• Achieved analog conductance switching with high linearity and symmetry.

Main Results:

• Electronic synapses exhibited high linearity and symmetry.
• Reproducible states (≈200) were achieved through gradual ferroelectric domain switching.
• A simulated neural network achieved 96.4% accuracy on MNIST handwritten recognition.

Conclusions:

• Ferroelectric tunnel junctions are suitable for high-performance electronic synapses.
• These devices can emulate biological synaptic plasticity.
• The developed neuromorphic computing approach shows significant potential for pattern recognition tasks.