A Scalable Artificial Neuron Based on Ultrathin Two-Dimensional Titanium Oxide

Area of Science:

• Materials Science
• Neuroscience
• Nanotechnology

Background:

• Spiking neural networks (SNNs) aim for human-level intelligence using artificial neurons and synapses.
• Fabricating high-performance, large-scale artificial neurons remains a significant challenge due to material and integration limitations.

Purpose of the Study:

• To develop a novel material system and fabrication method for ultrathin, large-scale artificial neurons.
• To demonstrate the potential of these artificial neurons in emulating biological neural activity and enabling 2D SNNs.

Main Methods:

• Controllable assembly of solution-processed, 2D monolayer titanium dioxide (TiO2) nanosheets to create ultrathin (less than 10 nm) artificial neuron systems.
• Fabrication of artificial neuron devices and arrays utilizing these 2D TiO2 films.

Main Results:

• Achieved inch-size, ultrathin artificial neuron devices with a high on/off ratio (10^9) and volatile resistance switching.
• Demonstrated emulation of leaky integrate-and-fire activity with self-recovery capabilities, eliminating the need for external sensing and reset circuits.
• Fabricated uniform artificial neuron arrays, highlighting potential for large-area integration.

Conclusions:

• Presents a viable strategy for fabricating large-scale, ultrathin 2D material-based artificial neurons.
• Offers a pathway towards the development of advanced 2D spiking neural networks with potential for neuromorphic computing applications.