Robust 2D MoS₂ Artificial Synapse Device Based on a Lithium Silicate Solid Electrolyte for High-Precision Analogue...

Area of Science:

• Materials Science
• Electrical Engineering
• Computer Science

Background:

• Neuromorphic hardware requires high-precision artificial synaptic devices for parallel analogue computation.
• Existing artificial synaptic devices face challenges in reliability and variability, hindering practical implementation.
• Von Neumann architecture limitations necessitate advancements in computing paradigms.

Purpose of the Study:

• To propose a robust three-terminal two-dimensional (2D) MoS2 artificial synaptic device.
• To address nonlinearity and asymmetric weight updates in artificial synaptic devices.
• To enhance the performance of neuromorphic hardware systems.

Main Methods:

• Fabrication of a three-terminal 2D MoS2 artificial synaptic device integrated with a lithium silicate (LSO) solid-state electrolyte.
• Characterization of device linearity, symmetry, and cycle-to-cycle variations during potentiation and depression.
• Simulation of the device's performance in Modified National Institute of Standards and Technology (MNIST) classification tasks.

Main Results:

• The proposed MoS2 synaptic device demonstrated excellent linearity and symmetry due to reversible Li ion intercalation.
• Achieved extremely low cycle-to-cycle variations (3.01%) over 500 pulses, enabling statistical analogue discrete states.
• Attained a high MNIST classification accuracy of 96.77%, closely approaching the software baseline of 98%.

Conclusions:

• The developed 2D MoS2 artificial synaptic device offers a robust solution for high-precision analogue neuromorphic computing.
• The combination of 2D materials and solid-state electrolytes presents a promising pathway for next-generation computing hardware.
• This work provides a future perspective for designing reliable synaptic devices for advanced AI systems.