Contact Engineering Toward High‐Performance MoS 2 Neuromorphic Devices
ABSTRACT Two‐dimensional transition metal dichalcogenides (2D TMDs) offer atomically thin, phase‐tunable platforms to obtain energy‐efficient neuromorphic devices. However, high contact resistance severely limits device performance by narrowing the memory window, increasing power consumption, and compromising long‐term reliability. To address these challenges, semi‐metallic bismuth (Bi) is employed as a metal electrode. The Bi‐contact configuration shows significantly reduced contact resistance, arising from electron doping at the MoS 2 –Bi interface, as confirmed by systematic I–V characterization, Raman, and XPS analysis. After lithiation, the Bi‐contact MoS 2 device demonstrates pronounced synaptic performance enhancement, achieving enhanced linearity and symmetry in synaptic weight modulation along with a high I on /I off ratio. Such synaptic memory modulation is attributed to ion‐induced localized phase change in the MoS 2 structure. Moreover, our Bi‐contact device achieves nanosecond switching speed with ultra‐low power consumption of 5.4 fJ/µm per event, along with 97% pattern recognition accuracy, outperforming the performance of conventional metal contacts. Localized phase modulation driven by efficient Li‐ion migration in an ohmic contact MoS 2 enables high‐accuracy synaptic memory operation with ultra‐low power consumption, demonstrating the viability of this approach for neuromorphic computing applications.
Authors
- Wonbong Choi (ORCID: https://orcid.org/0000-0002-7896-7655)
- Dongseok Suh (ORCID: https://orcid.org/0000-0002-0392-3391)
- Rifat Hasan Rupom (ORCID: https://orcid.org/0009-0007-0753-9817)
- Anh D. Vu (ORCID: https://orcid.org/0009-0005-6913-4992)
- Shinoj Sridharan Nair (ORCID: https://orcid.org/0000-0002-9868-7992)
- Jeongmin Park
- Oliver Chyan
Institutions
- University of North Texas (US)
- Ewha Womans University (KR)
- Sungkyunkwan University (KR)
Publication Details
- Journal
- Advanced Functional Materials
- Published
- 2026-10-07
- DOI
- https://doi.org/10.1002/adfm.78824
- Primary Topic
- Advanced Memory and Neural Computing
- Type
- article
- Field-Weighted Citation Impact
- 0.00