Dynamic Monolayer WSe₂ Electrolyte-Gated Transistor with Coexistent Double Relaxation Timescale for Enhanced Physical...

Area of Science:

• Materials Science
• Condensed Matter Physics
• Computational Neuroscience

Background:

• Physical reservoir computing (PRC) offers efficient solutions for complex time series tasks, but single timescale dynamics limit feature extraction.
• Existing memristor-based reservoirs often lack the necessary complexity for multi-timescale analysis.

Purpose of the Study:

• To develop a novel physical reservoir computing element with enriched dynamics for improved time series prediction.
• To investigate the impact of coexistent double relaxation timescales on information processing capabilities.

Main Methods:

• Fabrication of a volatile, electrolyte-gated monolayer (ML) WSe2 transistor.
• Characterization of the ion-electron coupling effect responsible for the double relaxation timescale (DRT).
• Evaluation of the WSe2 transistor's performance in PRC for various time series prediction tasks.

Main Results:

• Demonstrated a WSe2 transistor exhibiting a coexistent DRT due to ion-electron coupling.
• The DRT significantly enhanced reservoir dynamics compared to single relaxation timescale (SRT) systems.
• Achieved superior performance in chaotic, multi-scale, and traffic trajectory time series prediction.

Conclusions:

• The WSe2 transistor with DRT provides enriched dynamics for high-performance physical reservoir computing.
• This work enables the development of advanced dynamic devices for complex timescale-based computing networks.
• The findings pave the way for next-generation AI hardware with enhanced predictive capabilities.