A heterointerface effect of Mo1-xWxS2-based artificial synapse for neuromorphic computing

Jinwoo Hwang, Junho Sung, Eunho Lee, Wonbong Choi

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

Two-dimensional transition metal dichalcogenides (2D TMDs) are attracting significant interest as materials for next-generation electronic devices due to their unique properties, including ultra-fast switching, high density, and low energy consumption at atomic thickness. 2D TMDs as channel layers have potential for exhibiting memristive behavior whose operation is analogous to the movement of neurotransmitters such as Ca2+ between adjacent neurons. While 2D TMDs-based memristive devices have been studied, the precise operation mechanism of metal ion diffusion and its interaction within the 2D TMDs microstructure remain unclear. In this study, a novel memristor was designed using a Mo1-xWxS2 alloy as the channel layer fabricated by our two-step method of sputtering and post-chemical vapor deposition (CVD) process. By controlling the interfacial structure of MoS2 and WS2, we were able to elucidate the interaction mechanism of metal ions with the channel layer and enhance its stability and efficiency by forming conductive filaments at low voltages. Furthermore, the fabricated devices demonstrate a high dynamic range, long-term potentiation/depression (LTP/D), and paired-pulse facilitation (PPF), which emulate the behavior of chemical synapses, thus enabling them to function as synaptic devices. Our novel approach paves the way for developing artificial synapses using diffusive memristors applied in neuromorphic computing.

Original languageEnglish
Article number161622
JournalChemical Engineering Journal
Volume510
DOIs
StatePublished - 15 Apr 2025

Keywords

  • Artificial synapse
  • Heterointerface
  • Memristor
  • Neuromorphic computing
  • Two-dimensional materials

Fingerprint

Dive into the research topics of 'A heterointerface effect of Mo1-xWxS2-based artificial synapse for neuromorphic computing'. Together they form a unique fingerprint.

Cite this