The human brain has a completely different structure and information processing mode from integrated circuits used in personal computers and mobile phones, and realizes advanced information processing functions such as recognition, judgment, learning, and memory with extremely small energy.Therefore, efforts to improve the performance by incorporating the structure and operation mechanism of the neural network of the brain into the information processing device have been active.

　The research group is studying "spiking neural networks" that use spike-like outputs (action potentials) generated by neurons as signals.This requires a new concept hardware unit that can reproduce the temporal responses of neurons and synapses, which are the basic components of the brain, but since they have very different functions, they have been similar to them with completely different materials. The elements that show the behavior were developed separately.

　The research group has previously developed a material system in which antiferromagnetic materials and ferromagnetic materials are stacked.This time, we have shown that by microfabrication, elements that express each function of neurons and synapses can be simultaneously formed by the principle of spintronics that applies the characteristics of electron charge and spin.In addition, using quantum relativistic effects, we succeeded in realizing movements similar to the typical functions of neurons and the characteristic functions of synapses in learning and memory.

　This element enables artificial realization of the neural circuit function of the living body, flexible recognition and judgment similar to the human brain, learning and memory, and a completely new computer with excellent adaptability to the constantly changing environment and energy efficiency. Is expected to be realized.

Paper information:[Advanced Materials] Artificial neuron and synapse realized in an antiferromagnet / ferromagnet heterostructure using dynamics of spin-orbit torque switching