A 2D Heterostructure-Based Multifunctional Floating Gate Memory Device for Multimodal Reservoir Computing.

Area of Science:

• Materials Science
• Condensed Matter Physics
• Computer Engineering

Background:

• Neuromorphic computing demands efficient data processing, driving interest in 2D materials.
• Tellurium (Te), a p-type Weyl semiconductor, shows promise for electronics but lacks exploration in memory devices.
• Floating gate (FG) memory is crucial for information processing applications.

Purpose of the Study:

• To explore the application of 2D tellurium in electronic/optoelectronic floating gate memory devices.
• To develop a multimodal reservoir computing (RC) system using a Te-based vdW heterostructure.
• To investigate the device's memory characteristics under varying electrical and optical stimuli.

Main Methods:

• Fabrication of a 2D van der Waals (vdW) heterostructure device utilizing tellurium.
• Characterization of nonvolatile electronic and optoelectronic memory behaviors under intense stimuli.
• Demonstration of a multimodal reservoir computing system for handwritten digit recognition.

Main Results:

• The Te-based FG memory device exhibited an extinction ratio of ≈10^8, switching speed of ≈100 ns, and stability over >4000 cycles/seconds.
• Nonvolatile multibit optoelectronic programmable characteristics were observed.
• The device transitioned from nonvolatile to volatile memory as input stimuli weakened.
• A multimodal RC system achieved 90.77% accuracy in handwritten digit recognition.

Conclusions:

• Te-based 2D vdW heterostructures are promising for advanced electronic/optoelectronic memory devices.
• The device's tunable memory properties (nonvolatile to volatile) are beneficial for multimodal reservoir computing.
• This work opens new avenues for 2D materials in neuromorphic computing and information processing.