Dec 5, 2024
8:00pm - 10:00pm
Hynes, Level 1, Hall A
Gunhoo Woo1,Jinil Cho1,Jinhyoung Lee1,Chanho Park1,Jeongryul Yoo1,Jinsoo Yoon1,Jongwoo Kwon1,Hocheon Yoo2,Taesung Kim1
Sungkyunkwan University1,Gachon University2
Gunhoo Woo1,Jinil Cho1,Jinhyoung Lee1,Chanho Park1,Jeongryul Yoo1,Jinsoo Yoon1,Jongwoo Kwon1,Hocheon Yoo2,Taesung Kim1
Sungkyunkwan University1,Gachon University2
Memristors are emerging as promising candidates for electronic synapses and nonvolatile memory devices due to their high density, analog switching capabilities, and low energy consumption. However, conventional memristors face significant challenges, including device-to-device and cycle-to-cycle variations, because of destructive conductive filament (CF) formation processes. These issues hinder their practical application. This work presents a robust and reliable memristor array utilizing molybdenum disulfide (MoS<sub>2</sub>) nanograins, termed nanoristors. MoS<sub>2</sub> films with 7–10 nm nanograins were synthesized via plasma-enhanced chemical vapor deposition. The resulting structure features tens of thousands of grain boundaries within a 1 μm × 1 μm area, enabling memory operation with remarkable uniformity and endurance exceeding 2300 cycles. Additionally, the vertical grain boundaries and defects facilitate Ag<sup>+</sup> ion diffusion, resulting in a forming-free operation with a low switching voltage of under 500 mV. The MoS<sub>2</sub> nanoristor also successfully emulated potentiation and depression characteristics crucial for online learning in neuromorphic systems. The efficacy of the MoS<sub>2</sub> nanoristor-based synapse device for face recognition was validated through device-to-system simulation. This innovative approach of fabricating nanograined MoS<sub>2</sub> offers highly uniform and robust memristor operation, potentially revolutionizing electronic synapses and memory devices in neuromorphic electronic systems.