TY - JOUR
T1 - Dynamic NiOx-based memristors for edge computing
AU - Park, Seoyoung
AU - Park, Suyong
AU - Kim, Sungjun
N1 - Publisher Copyright:
© 2025 The Physical Society of the Republic of China (Taiwan)
PY - 2025/6
Y1 - 2025/6
N2 - Resistive random-access memory (RRAM) devices, which leverage resistance state modulation for data storage and retrieval, have garnered considerable interest due to their high-speed performance, low energy consumption, and exceptional scalability. These advanced characteristics make RRAM devices highly suitable for neuromorphic computing, a rapidly emerging paradigm aimed at developing autonomous systems capable of real-time learning, adaptation, and environmental interaction. In neuromorphic architecture, RRAM is increasingly viewed as a promising candidate for computing-in-memory. This research investigates the realization of neuromorphic systems by fine-tuning conductance using the DC sweep and electrical pulse on ITO/NiOX/n+ + Si stacked RRAM devices, based on their distinct resistance states. Key properties crucial for neuromorphic functionality, including Spike Amplitude-Dependent Plasticity (SADP), Spike Number-Dependent Plasticity (SNDP), Spike Duration-Dependent Plasticity (SDDP), were systematically examined. The potentiation and depression dynamics, along with the long-term plasticity characteristics demonstrated by the RRAM device, underscore its promising potential for neuromorphic applications. The demonstrated multi-state operational capability highlights the potential of the device for high-efficiency data processing and storage, which are essential for advanced edge computing architectures.
AB - Resistive random-access memory (RRAM) devices, which leverage resistance state modulation for data storage and retrieval, have garnered considerable interest due to their high-speed performance, low energy consumption, and exceptional scalability. These advanced characteristics make RRAM devices highly suitable for neuromorphic computing, a rapidly emerging paradigm aimed at developing autonomous systems capable of real-time learning, adaptation, and environmental interaction. In neuromorphic architecture, RRAM is increasingly viewed as a promising candidate for computing-in-memory. This research investigates the realization of neuromorphic systems by fine-tuning conductance using the DC sweep and electrical pulse on ITO/NiOX/n+ + Si stacked RRAM devices, based on their distinct resistance states. Key properties crucial for neuromorphic functionality, including Spike Amplitude-Dependent Plasticity (SADP), Spike Number-Dependent Plasticity (SNDP), Spike Duration-Dependent Plasticity (SDDP), were systematically examined. The potentiation and depression dynamics, along with the long-term plasticity characteristics demonstrated by the RRAM device, underscore its promising potential for neuromorphic applications. The demonstrated multi-state operational capability highlights the potential of the device for high-efficiency data processing and storage, which are essential for advanced edge computing architectures.
KW - Edge computing
KW - Neuromorphic computing
KW - Resistive memory
KW - Synaptic plasticity
UR - http://www.scopus.com/inward/record.url?scp=105002243729&partnerID=8YFLogxK
U2 - 10.1016/j.cjph.2025.04.003
DO - 10.1016/j.cjph.2025.04.003
M3 - Article
AN - SCOPUS:105002243729
SN - 0577-9073
VL - 95
SP - 803
EP - 813
JO - Chinese Journal of Physics
JF - Chinese Journal of Physics
ER -