Atomically Thin Synapse Networks on Van Der Waals Photo-Memtransistors

Area of Science:

• Materials Science
• Nanotechnology
• Neuroscience

Background:

• Artificial neural networks (ANNs) require efficient synaptic devices for computation.
• Existing memristors often lack the precision and integration density needed for complex ANNs.
• Two-dimensional (2D) van der Waals (vdW) heterostructures offer unique electronic properties for novel device applications.

Purpose of the Study:

• To develop a new type of atomically thin photoactive synapse.
• To demonstrate gate-tunable synaptic plasticity in 2D materials.
• To explore the potential of these devices for artificial neural networks.

Main Methods:

• Fabrication of photo-memtransistors using trilayer WS2 on vdW heterostructures.
• Utilizing UV pulses to induce light-lattice interactions and generate dopants.
• Modulating channel conductance via gate tuning for synaptic potentiation and depression.
• Modeling synaptic dynamics using a parallel atomistic resistor network.

Main Results:

• Demonstrated ultrasmall (≈2 nm thick) gate-tunable photoactive synaptic cells.
• Achieved accurate modulation of synaptic conductance through light-induced doping and gate control.
• Showcased the generalizability of the device scheme to other 2D vdW semiconductors (MoS2, MoSe2, MoTe2, WSe2).
• Tuned synaptic weights by controlling atomistic defect density.

Conclusions:

• Atomically thin photo-memtransistor arrays offer a new paradigm for ANNs.
• These devices enable parallel matrix computations with ultrahigh integration density.
• The precise control over synaptic dynamics opens avenues for neuromorphic computing.