Publication date: 15th April 2025
Physical reservoir computing (RC) is considered as an efficient method for solving complex time-series tasks due to its outstanding advantages such as easy training and hardware implementation. However, memristor-based reservoir with only single timescale carrier/ion dynamics limits the performance of multiple timescale feature extraction. Here, we report a fully volatile, electrolyte-gated monolayer (ML) WSe2 transistor with coexistent double relaxation timescale (DRT), which enriches the reservoir dynamics and has excellent time series prediction capability in physical reservoir computing. The coexistence of DRT is achieved by ion-electron coupling effect between the electrolyte dielectric and the WSe2 channel. Compared with single relaxation timescale (SRT), the DRT in WSe2 transistor leads to enhanced information processing capability in chaotic time series prediction, multi-scale time series prediction and traffic trajectory prediction applications. This work paves the way for the development of complex timescale based dynamic devices for high-performance reservoir computing networks.