TY - JOUR
T1 - Current-Induced Crystallisation in a Heusler-Alloy-Based Giant Magnetoresistive Junction for Neuromorphic Potentiation
AU - Zhou, Zhenyu
AU - Frost, William James
AU - Lloyd, David
AU - Seki, Takeshi
AU - Kubota, Takahide
AU - Ramos, Rafael
AU - Saitoh, Eiji
AU - Takanashi, Koki
AU - Hirohata, Atsufumi
N1 - © 2023 The Author(s). Published by Elsevier B.V.
PY - 2023/2/27
Y1 - 2023/2/27
N2 - Recent development in neuromorphic computation allows us to achieve low power and highly efficient calculations better than the conventional von Neumann computation. In order to achieve realistic synaptic operation, potentiation to add weighting to strengthen a selected artificial synapse. Such functionality can be achieved by reducing the electrical resistance of the artificial synapse. Recently, a ferromagnetic Heusler alloy used in a magnetoresistive junction has been demonstrated to crystallise via the layer-by-layer mode by introducing an electrical current pulse. In this study, we have extended the current-induced crystallisation to a junction with epitaxially-grown Heusler alloy after post-annealing for crystallisation. By combining this potentiation functionality with the neuromorphic operation, realistic synaptic computation can be developed.
AB - Recent development in neuromorphic computation allows us to achieve low power and highly efficient calculations better than the conventional von Neumann computation. In order to achieve realistic synaptic operation, potentiation to add weighting to strengthen a selected artificial synapse. Such functionality can be achieved by reducing the electrical resistance of the artificial synapse. Recently, a ferromagnetic Heusler alloy used in a magnetoresistive junction has been demonstrated to crystallise via the layer-by-layer mode by introducing an electrical current pulse. In this study, we have extended the current-induced crystallisation to a junction with epitaxially-grown Heusler alloy after post-annealing for crystallisation. By combining this potentiation functionality with the neuromorphic operation, realistic synaptic computation can be developed.
U2 - 10.1016/j.jmmm.2023.170575
DO - 10.1016/j.jmmm.2023.170575
M3 - Review article
SN - 0304-8853
JO - Journal of Magnetism and Magnetic Materials
JF - Journal of Magnetism and Magnetic Materials
ER -