Tdk Corporation

Keita Suda

Tomoyuki Sasaki

Eiji Komura

Yohei Shiokawa

Yugo Ishitani

Provided is a magnetoresistance effect element that suppresses re-adhesion of impurities during preparation and allows a write current to easily flow. The magnetoresistance effect element includes a first ferromagnetic layer, a second ferromagnetic layer; and a nonmagnetic layer interposed between the first ferromagnetic layer and the second ferromagnetic layer. In the magnetoresistance effect element, the nonmagnetic layer is a tunnel barrier layer constituted by an insulator, a side surface of the first ferromagnetic layer, a side surface of the second ferromagnetic layer and a side surface of the nonmagnetic layer form a continuous inclined surface in any side surface, and a thickness of inside the nonmagnetic layer is thicker than a thickness of outside the nonmagnetic layer.

Magnetoresistance effect element

A spin-orbit-torque magnetization rotational element includes: a ferromagnetic metal layer, a magnetization direction of the ferromagnetic metal layer being configured to change; a spin-orbit torque wiring which extends in the first direction intersecting a lamination direction of the ferromagnetic metal layer and is joined to the ferromagnetic metal layer; and two via wires, each of which extends in a direction intersecting the spin-orbit torque wiring from a surface of the spin-orbit torque wiring opposite to a side with the ferromagnetic metal layer and is connected to a semiconductor circuit, wherein a via-to-via distance between the two via wires in the first direction is shorter than a width of the ferromagnetic metal layer in the first direction.

Spin-orbit-torque magnetization rotational element and magnetic memory

A neural network arithmetic processing device is capable of implementing a further increase in speed and efficiency of multiply-accumulate arithmetic operation, suppressing an increase in circuit scale, and performing multiply-accumulate arithmetic operation with simple design. A neural network arithmetic processing device includes a first multiply-accumulate arithmetic unit, a register connected to the first multiply-accumulate arithmetic unit, and a second multiply-accumulate arithmetic unit connected to the register. The first multiply-accumulate arithmetic unit has a first memory, a second memory, a first multiplier, a first adder, and a first output unit. The second multiply-accumulate arithmetic unit has an input unit, a third memory, second multipliers, second adders, and second output units.

Neural network arithmetic processing device and neural network arithmetic processing method

Provided is a spin-orbit-torque magnetization rotational element that suppresses re-adhesion of impurities during preparation and allows a write current to easily flow. The spin-orbit-torque magnetization rotational element includes a spin-orbit torque wiring that extends in a first direction, and a first ferromagnetic layer that is located on a side of one surface of the spin-orbit torque wiring. A side surface of the spin-orbit torque wiring and a side surface of the first ferromagnetic layer form a continuous inclined surface in any side surface.

Spin-orbit-torque magnetization rotational element, spin-orbit-torque magnetoresistance effect element, and magnetic memory

Provided is a spin current magnetization rotational element including: a spin-orbit torque wiring that extends in a first direction and is configured to generate a spin current; a first ferromagnetic layer that is laminated in a second direction intersecting the spin-orbit torque wiring and is configured for magnetization direction to be changed; and a first perpendicular magnetic field applying layer that is disposed to be separated from the spin-orbit torque wiring and the first ferromagnetic layer, the first perpendicular magnetic field applying layer being configured to apply an assistant magnetic field assisting a magnetization rotation of the first ferromagnetic layer.

Growing community of inventors

Tokyo, Japan

Keita Suda