Chip-Scale Plasma-Sulfurized Nanocrystalline MoS2-Graphene Heterostructure-Based Optoelectronic Synaptic Transistor
Abstract Optoelectronic synaptic devices based on 2D van der Waals (vdW) materials are very promising for next-generation, energy-efficient data processing and sensing; however, flake-based fabrication limits their practical application. Here, scalable inductively coupled plasma-sulfurized nanocrystalline MoS2 (nc-MoS2) and graphene heterostructure have been investigated for optoelectronic synaptic device applications. The nc-MoS2 with abundant sulfur vacancies, grain boundaries, and defect states facilitates charge trapping at the nc-MoS2/Graphene interface and introduces negative persistent photocurrent (PPC). The PPC state is effective only for ultraviolet (UV) light because of the high recombination of photogenerated charge carriers at the defect sites of nc-MoS2 for higher wavelength light; however, it enables energy-efficient, neuromorphic UV-selective sensing that could serve as a UV exposure alert system. The device demonstrates efficient emulation of fundamental synaptic behaviors, including short-term memory, long-term memory, and paired-pulse facilitation, through modulation of optical stimuli such as light intensity and illumination duration, as well as reconfiguration via the gate voltage. A 96.94% accuracy in MINST image recognition has also been demonstrated through deep neural network-based simulation, which seems potentially applicable in neuromorphic computing. Thus, the nc-MoS2/Gr heterostructure, as a scalable, 2D vdW system, has significant potential for optoelectronic synapses.
Authors
- Sihoon Son (ORCID: https://orcid.org/0009-0002-9052-335X)
- Sukalyan Shyam (ORCID: https://orcid.org/0000-0001-8857-9667)
- Taesung Kim (ORCID: https://orcid.org/0000-0001-6280-7668)
Institutions
- Sungkyunkwan University (KR)
Publication Details
- Journal
- ACS Applied Materials & Interfaces
- Published
- 2026-10-08
- DOI
- https://doi.org/10.1021/acsami.6c15831
- Primary Topic
- Advanced Memory and Neural Computing
- Type
- article
- Field-Weighted Citation Impact
- 0.00