Tactile tribotronic reconfigurable p-n junctions for artificial synapses

Area of Science:

• Materials Science
• Nanotechnology
• Neuromorphic Engineering

Background:

• Neuromorphic computing systems aim to emulate biological synapses for enhanced learning and memory.
• Developing artificial synapses with versatile plasticity is crucial for advancing these systems.

Purpose of the Study:

• To demonstrate a robust, continuously adjustable mechanoplastic transistor for neuromorphic applications.
• To construct a reconfigurable artificial synapse capable of simulating synaptic plasticity and dynamic control correlations.

Main Methods:

• Fabrication of a van der Waals heterostructure transistor using graphene, hexagonal boron nitride, and tungsten diselenide.
• Integration with a triboelectric nanogenerator (TENG) for mechanical modulation of transistor states.
• Characterization of synaptic plasticity, including short-/long-term plasticity and learning-experience behavior.

Main Results:

• Demonstrated a mechanoplastic transistor whose working states are controlled by TENG-derived triboelectric potential.
• Constructed a reconfigurable artificial synapse exhibiting ultra-low energy consumption (74.2 fJ/event) and extended synaptic weights.
• Observed mechanical behavior-derived synaptic plasticity, including facilitation, depression, and learning-experience phenomena.

Conclusions:

• The developed mechanoplastic transistor and artificial synapse offer a promising platform for energy-efficient, real-time interactive neuromodulation.
• These features provide insights for future artificial intelligent systems beyond the von Neumann architecture.