TY - JOUR
T1 - Ultralow Energy Domain Wall Device for Spin-Based Neuromorphic Computing
AU - Kumar, Durgesh
AU - Chung, Hong Jing
AU - Chan, Jian Peng
AU - Jin, Tianli
AU - Lim, Sze Ter
AU - Parkin, Stuart S.P.
AU - Sbiaa, Rachid
AU - Piramanayagam, S. N.
N1 - Funding Information:
The authors gratefully acknowledge the National Research Foundation (NRF), Singapore for the NRF-IIP Grant (NRF2015-IIP003-001) and the NRF-CRP21 Grant (NRF-CRP21-2018-0003). The authors also acknowledge the support provided by the Agency for Science, Technology and Research, A*STAR RIE2020 AME Grant No. A18A6b0057 for this work. DK acknowledges the financial assistance from the NTU research scholarship.
Publisher Copyright:
© 2023 American Chemical Society.
PY - 2023/4/11
Y1 - 2023/4/11
N2 - Neuromorphic computing (NC) is gaining wide acceptance as a potential technology to achieve low-power intelligent devices. To realize NC, researchers investigate various types of synthetic neurons and synaptic devices, such as memristors and spintronic devices. In comparison, spintronics-based neurons and synapses have potentially higher endurance. However, for realizing low-power devices, domain wall (DW) devices that show DW motion at low energies─typically below pJ/bit─are favored. Here, we demonstrate DW motion at current densities as low as 106 A/m2 by engineering the β-W spin-orbit coupling (SOC) material. With our design, we achieve ultralow pinning fields and current density reduction by a factor of 104. The energy required to move the DW by a distance of about 18.6 μm is 0.4 fJ, which translates into the energy consumption of 27 aJ/bit for a bit-length of 1 μm. With a meander DW device configuration, we have established a controlled DW motion for synapse applications and have shown the direction to make ultralow energy spin-based neuromorphic elements.
AB - Neuromorphic computing (NC) is gaining wide acceptance as a potential technology to achieve low-power intelligent devices. To realize NC, researchers investigate various types of synthetic neurons and synaptic devices, such as memristors and spintronic devices. In comparison, spintronics-based neurons and synapses have potentially higher endurance. However, for realizing low-power devices, domain wall (DW) devices that show DW motion at low energies─typically below pJ/bit─are favored. Here, we demonstrate DW motion at current densities as low as 106 A/m2 by engineering the β-W spin-orbit coupling (SOC) material. With our design, we achieve ultralow pinning fields and current density reduction by a factor of 104. The energy required to move the DW by a distance of about 18.6 μm is 0.4 fJ, which translates into the energy consumption of 27 aJ/bit for a bit-length of 1 μm. With a meander DW device configuration, we have established a controlled DW motion for synapse applications and have shown the direction to make ultralow energy spin-based neuromorphic elements.
KW - artificial intelligence
KW - domain wall motion
KW - dual W
KW - neuromorphic computing
KW - pinning field
KW - spin−orbit torque
UR - http://www.scopus.com/inward/record.url?scp=85151257122&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85151257122&partnerID=8YFLogxK
UR - https://www.mendeley.com/catalogue/e9bc94ed-4474-300c-85fb-cef63b8e4a96/
U2 - 10.1021/acsnano.2c09744
DO - 10.1021/acsnano.2c09744
M3 - Article
C2 - 36944594
AN - SCOPUS:85151257122
SN - 1936-0851
VL - 17
SP - 6261
EP - 6274
JO - ACS Nano
JF - ACS Nano
IS - 7
ER -