阅读说明：本技术 储备池元件和神经形态元件 (Reservoir element and neuromorphic element ) 是由 佐佐木智生 柴田龙雄 于 2019-09-11 设计创作，主要内容包括：本发明的一个方式的储备池元件包括：第1铁磁性层；相对于上述第1铁磁性层位于第1方向,且在从上述第1方向俯视时彼此隔着间隔地配置的多个第2铁磁性层；和位于上述第1铁磁性层与上述多个第2铁磁性层之间的非磁性层。(A reserve tank element according to one embodiment of the present invention includes: a 1 st ferromagnetic layer; a plurality of 2 nd ferromagnetic layers located in a 1 st direction with respect to the 1 st ferromagnetic layer and arranged with a space therebetween in a plan view from the 1 st direction; and a nonmagnetic layer located between the 1 st ferromagnetic layer and the plurality of 2 nd ferromagnetic layers.)

1. A reservoir element, comprising:

a 1 st ferromagnetic layer;

a plurality of 2 nd ferromagnetic layers located in a 1 st direction with respect to the 1 st ferromagnetic layer and arranged with a space therebetween in a plan view from the 1 st direction; and

a non-magnetic layer between the 1 st ferromagnetic layer and the plurality of 2 nd ferromagnetic layers.

2. The reserve pool element of claim 1, wherein:

and a connection wiring electrically connected to a surface of the 1 st ferromagnetic layer opposite to the nonmagnetic layer.

3. The reserve pool element of claim 2, wherein:

the communication wiring is provided in a plurality of numbers,

each of the plurality of via wirings is located at a position overlapping with each of the plurality of 2 nd ferromagnetic layers when viewed from the 1 st direction in plan view.

4. A reservoir element according to any of claims 1 to 3, characterized in that:

the magnetic interference layer is in contact with the surface of the 1 st ferromagnetic layer on the side opposite to the nonmagnetic layer, and the coercive force of the magnetic interference layer is smaller than that of the 1 st ferromagnetic layer.

5. The reserve pool element of claim 4, wherein:

the magnetic interference layer is an alloy containing any one of Fe-Si, Fe-Si-Al, Fe-Co-V, Ni-Fe, Co-Fe-Si-B.

6. A reservoir element according to claim 3, characterized in that:

the semiconductor device further includes a common electrode layer connecting at least two or more of the plurality of via wirings.

7. A reservoir element according to any of claims 1 to 6, characterized in that:

the plurality of 2 nd ferromagnetic layers are arranged in a hexagonal lattice shape when viewed from the 1 st direction.

8. A reservoir element according to any of claims 1 to 6, characterized in that:

the plurality of 2 nd ferromagnetic layers are formed with an aggregate in which the 2 nd ferromagnetic layers are densely arranged when viewed from the 1 st direction,

in the aggregate, the 2 nd ferromagnetic layers are arranged in a hexagonal lattice.

9. A neuromorphic element, comprising:

a reservoir element according to any one of claims 1 to 8;

an input connected to the reservoir element; and

and an output unit connected to the reserve tank element and learning a signal from the reserve tank element.

Technical Field

The present invention relates to a reservoir (reservoir) element and a neuromorphic (neuromorphic) element.

Background

Neuromorphic elements are elements that mimic the brain of a human being through a neural network. Neuromorphic elements artificially mimic the relationship of neurons (neurons) and synapses (synapses) in the human brain.

The hierarchical element is one of the neuromorphic elements. The hierarchical element has chips (neurons in the brain) arranged in a hierarchy and a transfer unit (synapse in the brain) connecting between them. The hierarchical elements increase the correct answer rate of the question by learning the transmission unit (synapse). Learning refers to finding knowledge from information that can be used in the future and weighting the input data in the neuromorphic elements. The hierarchy type element learns in each hierarchy.

However, if the number of chips (neurons) increases in learning in each layer, a large load is imposed on circuit design, which causes an increase in power consumption of neuromorphic devices. Pool computers have been investigated as a way to alleviate this burden.

Reservoir computers are one type of neuromorphic element. The reserve tank computer includes a reserve tank element and an output. The reservoir element comprises a plurality of chips interacting with each other. The plurality of chips output signals by interacting with each other through the input signals. The transfer units connecting the plurality of chips are fixed in weight without learning. The output unit learns the signal from the reserve tank element and outputs the signal to the outside. The pool computer increases the correct answer rate of the question by compressing the data in the pool element and weighting the data in the output section. The learning in the pool computer is performed only by the output section. A reserve pool computer is expected as a method capable of simplifying the circuit design of neuromorphic elements and improving power consumption.

Non-patent document 1 describes a neuromorphic device using a Spin Torque Oscillation (STO) device as a chip (neuron).

Disclosure of Invention

Problems to be solved by the invention

However, using STO elements for the neuromorphic elements of the chip requires matching the resonance frequencies of the respective STO elements. The STO elements may have a variation in resonance frequency due to a manufacturing error or the like, and may not sufficiently interact with each other. In addition, the STO element oscillates by applying a high-frequency current in the stacking direction. Applying a high-frequency current for a long time in the stacking direction of the STO elements having the insulating layers causes a failure of the STO elements.

The present invention has been made in view of the above circumstances, and provides a reservoir element and a neuromorphic element that operate stably.

Means for solving the problems

In order to solve the above problems, the present invention provides the following methods.

(1) The reserve tank element according to mode 1 includes: a 1 st ferromagnetic layer; a plurality of 2 nd ferromagnetic layers located in a 1 st direction with respect to the 1 st ferromagnetic layer and arranged with a space therebetween in a plan view from the 1 st direction; and a nonmagnetic layer located between the 1 st ferromagnetic layer and the plurality of 2 nd ferromagnetic layers.

(2) The reservoir element according to the above aspect may further include a communication wiring electrically connected to a surface of the 1 st ferromagnetic layer opposite to the nonmagnetic layer.

(3) In the reservoir element according to the above aspect, the communication wiring may be provided in plurality, and each of the plurality of communication wirings may be located at a position overlapping with each of the plurality of 2 nd ferromagnetic layers when viewed in a plan view from the 1 st direction.

(4) The reservoir element according to the above aspect may further include a magnetic interference layer that is in contact with a surface of the 1 st ferromagnetic layer opposite to the nonmagnetic layer and has a smaller coercive force than the 1 st ferromagnetic layer.

(5) In the reservoir element of the above aspect, the magnetic interference layer may be an alloy containing any one of Fe-Si, Fe-Si-Al, Fe-Co-V, Ni-Fe and Co-Fe-Si-B.

(6) The reservoir element according to the above aspect may further include a common electrode layer that connects at least two or more of the plurality of communication wirings.

(7) In the reservoir element according to the above aspect, the plurality of 2 nd ferromagnetic layers may be arranged in a hexagonal lattice shape when viewed from the 1 st direction.

(8) In the reservoir element according to the above aspect, a plurality of 2 nd ferromagnetic layers may be formed as an aggregate in which the 2 nd ferromagnetic layers are densely arranged when viewed in a plan view from the 1 st direction, and in the aggregate, the 2 nd ferromagnetic layers may be arranged in a hexagonal lattice shape.

(9) The neuromorphic element of claim 2 comprising: a reservoir element in the manner described above; an input part connected with the reserve pool element; and an output unit connected to the reservoir element and learning a signal from the reservoir element.

ADVANTAGEOUS EFFECTS OF INVENTION

The reservoir element and the neuromorphic element of the present embodiment can stably operate.

Drawings

Fig. 1 is a conceptual diagram of a neuromorphic element of embodiment 1.

Fig. 2 is a perspective view of the reservoir member of embodiment 1.

Fig. 3 is a side view of the reservoir member of embodiment 1.

Fig. 4 is a plan view of the reservoir member of embodiment 1.

Fig. 5 is a plan view of another example of the reservoir member according to embodiment 1.

Fig. 6 is a plan view of another example of the reservoir member according to embodiment 1.

Fig. 7 is a plan view of another example of the reservoir member according to embodiment 1.

Fig. 8A is a sectional view showing a method of manufacturing a reservoir element according to embodiment 1.

Fig. 8B is a sectional view showing a method of manufacturing a reservoir element according to embodiment 1.

Fig. 8C is a sectional view showing a method of manufacturing a reservoir element according to embodiment 1.

Fig. 8D is a sectional view showing a method of manufacturing a reservoir element according to embodiment 1.

Fig. 9 is a perspective view of the reservoir member of embodiment 2.

Fig. 10 is a side view of the reservoir member of embodiment 3.

Fig. 11 is a side view of the reservoir member of embodiment 4.

Fig. 12 is a schematic diagram for explaining an example of the operation of the neuromorphic device.

Fig. 13 is a schematic diagram for explaining another example of the operation of the neuromorphic element.

Fig. 14 is a schematic diagram for explaining another example of the neuromorphic device.

Fig. 15 is a schematic diagram for explaining another example of the neuromorphic device.

Fig. 16 is a plan view of the reservoir member of the 5 th embodiment.

Description of the reference numerals

1 st ferromagnetic layer

2. 2A, 2B, 2C, 2D, 2E, 2F, 2G 2 nd ferromagnetic layer

2' ferromagnetic layer

3. 3' nonmagnetic layer

4 connected wiring

5 common electrode layer

6 magnetic interference layer

10. Reservoir elements 10A, 10B, 10C, 11, 12, 13, 14

20 input part

21. 22, 23, 24 input terminal

30 output part

31. 32 output terminal

40 2 nd output part

41. 42, 43 terminal

100 neuromorphic element

Aggregate A

Cp chip

HM hard mask (hard mask)

I interlayer insulating film

Sb substrate

Sp synapse

Detailed Description

Hereinafter, the present embodiment will be described in detail with reference to the accompanying drawings as appropriate. The drawings used in the following description may be enlarged to show the characteristic portions for easier understanding of the characteristics, and the dimensional ratios of the respective components may be different from those in reality. The materials, dimensions, and the like exemplified in the following description are examples, and the present invention is not limited thereto, and can be modified as appropriate within a range in which the effects of the present invention are achieved.

"embodiment 1

Fig. 1 is a conceptual diagram of a neuromorphic element of embodiment 1. The neuromorphic element 100 has: an input 20, a reservoir element 10 and an output 30. The input 20 and the output 30 are connected to the reservoir element 10.

The neuromorphic device 100 compresses the signal input from the input unit 20 in the reservoir device 10, weights (learns) the signal compressed in the output unit 30, and outputs the signal to the outside.

The input section 20 transmits a signal input from the outside to the reservoir member 10. The input unit 20 includes, for example, a plurality of sensors. The plurality of sensors sense information outside the neuromorphic element 100, and input the information as a signal to the reservoir element 10. The signal may be input to the reservoir unit 10 continuously with time, or may be input to the reservoir unit 10 divided into predetermined time zones.

Reservoir element 10 has a plurality of chips Cp. The plurality of chips Cp interact with each other. The signal input to the reservoir unit 10 contains a large amount of information. The signal contains a large amount of information which is compressed into the desired information by the interaction of the chips Cp. The compressed signal is transmitted to the output section 30. The reserve pool element 10 does not learn. That is, the plurality of chips Cp interact with each other only, and signals transmitted between the plurality of chips Cp are not weighted.

Output 30 receives a signal from chip Cp of reservoir element 10. The output unit 30 performs learning. The output unit 30 weights each signal from each chip Cp by learning. The output unit 30 includes, for example, a nonvolatile memory. The nonvolatile memory is, for example, a magnetoresistive effect element. The output unit 30 outputs a signal to the outside of the neuromorphic device 100.

The neuromorphic element 100 increases the correct answer rate of the question by compressing the data in the reservoir element 10 and weighting the data in the output unit 30.

In addition, the neuromorphic device 100 is excellent in power consumption. Only the output unit 30 learns in the neuromorphic device 100. Learning is to adjust the weight of the signal transferred from each chip Cp. The weight of the signal is determined according to the importance of the signal. If the weight of the signal is adjusted at any time, the circuit between the chips Cp is activated (active). The more circuits that are activated, the more power the neuromorphic element 100 consumes. The neuromorphic device 100 is excellent in power consumption as long as it is learned only by the output unit 30 at the final stage.

Fig. 2 is a perspective view of the reservoir member 10 of embodiment 1. Fig. 3 is a side view of the reservoir member 10 of embodiment 1. Fig. 4 is a plan view of the reservoir member 10 of embodiment 1.

The reserve tank element 10 includes: a 1 st ferromagnetic layer 1, a plurality of 2 nd ferromagnetic layers 2, a nonmagnetic layer 3, and a plurality of via wirings 4. The plurality of 2 nd ferromagnetic layers 2 correspond to the chip Cp in fig. 1.

Here, the direction is defined. An arbitrary direction in the expanded plane of the 1 st ferromagnetic layer 1 is defined as an x-direction, a direction intersecting with (e.g., substantially orthogonal to) the x-direction in the expanded plane of the 1 st ferromagnetic layer 1 is defined as a y-direction, and a direction intersecting with (e.g., substantially orthogonal to) the expanded plane of the 1 st ferromagnetic layer 1 is defined as a z-direction. The z direction is an example of the 1 st direction.

The 1 st ferromagnetic layer 1 extends continuously in the xy plane. The 1 st ferromagnetic layer 1 may be a perpendicular magnetization film in which the easy magnetization axis of magnetization is oriented in the z direction, or an in-plane magnetization film in which the easy magnetization axis is oriented in the xy in-plane direction.

The 1 st ferromagnetic layer 1 contains a ferromagnetic material. The 1 st ferromagnetic layer 1 contains, for example, a metal selected from Cr, Mn, Co, Fe, and Ni, an alloy containing one or more of these metals, and an alloy containing these metals and B, C, and at least one or more elements of N. The 1 st ferromagnetic layer 1 is, for example, Co-Fe-B, Ni-Fe, Co-Ho alloy (CoHO)₂) Sm-Fe alloy (SmFe)₁₂). When the 1 st ferromagnetic layer 1 contains a Co-Ho alloy (CoHO)₂) Sm-Fe alloy (SmFe)₁₂) In this case, the 1 st ferromagnetic layer 1 is likely to be an in-plane magnetization film.

The 1 st ferromagnetic layer 1 may also be a heusler alloy. The heusler alloy is of XYZ or X₂YZ, X is a transition metal element or a noble metal element of Co, Fe, Ni, or Cu group of the periodic table, Y is a transition metal or an element species of X of Mn, V, Cr, or Ti group, and Z is a typical element from III group to V group. The heusler alloy being, for example, Co₂FeSi、Co₂FeGe、Co₂FeGa、Co₂MnSi、Co₂Mn_1-aFe_aAl_bSi_1-b、Co₂FeGe_1-cGa_c. The heusler alloy has high spin polarizability and exhibits a magnetoresistive effect more strongly.

The 1 st ferromagnetic layer 1 preferably has at least one element selected from the group consisting of Co, Ni, Pt, Pd, Gd, Tb, Mn, Ge, Ga. Examples thereof include a laminated film of Co and Ni, a laminated film of Co and Pt, a laminated film of Co and Pd, a MnGa-based material, a GdCo-based material, and a TbCo-based material. The saturation magnetization of ferrimagnetic materials such as MnGa-based materials, GdCo-based materials, and TbCo-based materials is small, and the threshold current required for moving a magnetic domain wall is small.

The 2 nd ferromagnetic layer 2 is formed on one surface of the nonmagnetic layer 3. The 2 nd ferromagnetic layer 2 protrudes in the z direction, and a plurality of the layers are present apart from each other in the xy plane. A plurality of the 2 nd ferromagnetic layers 2 exist with respect to one 1 st ferromagnetic layer 1. The adjacent 2 nd ferromagnetic layer 2 is insulated by, for example, an interlayer insulating film.

The plurality of 2 nd ferromagnetic layers 2 are arranged in a hexagonal lattice shape when viewed from the z direction, for example (see fig. 4). A signal input to the 2 nd ferromagnetic layer 2 propagates in the 1 st ferromagnetic layer 1. When the 2 nd ferromagnetic layers 2 are arranged in a hexagonal grid, a signal input from the 2 nd ferromagnetic layer 2 easily interferes with a signal input from another 2 nd ferromagnetic layer 2.

The arrangement of the plurality of 2 nd ferromagnetic layers 2 is not limited to the case of fig. 4. Fig. 5 to 7 are plan views of another example of the reserve tank element according to embodiment 1.

The plurality of 2 nd ferromagnetic layers 2 of the reserve cell element 10A shown in fig. 5 are arranged in a square lattice shape. The distances between the adjacent 2 nd ferromagnetic layers 2 are equal to homogenize the input signal.

The plurality of 2 nd ferromagnetic layers 2 of the reservoir element 10B shown in fig. 6 are densely packed in a hexagonal lattice shape. Since the density of the 2 nd ferromagnetic layer 2 becomes high, signals input to different 2 nd ferromagnetic layers 2 easily interfere with each other. Further, even in this case, the 2 nd ferromagnetic layers 2 are insulated from each other.

The reservoir element 10C shown in fig. 7 forms an aggregate a in which a plurality of 2 nd ferromagnetic layers 2 are densely packed. In the aggregate a, the 2 nd ferromagnetic layers 2 are arranged in a hexagonal lattice shape. The adjacent 2 nd ferromagnetic layers 2 are insulated from each other. In the reservoir element 10C, the mutual interference of signals input to the 2 nd ferromagnetic layer 2 constituting one aggregate a and the mutual interference of signals input to the 2 nd ferromagnetic layer 2 constituting a different aggregate a are different from each other, and the mutual interference conditions are different. In the reservoir element 10C, by adjusting the condition of mutual interference, the reservoir element 10C emphasizes a specific signal and transmits it to the output section 30.

Each 2 nd ferromagnetic layer 2 has, for example, a columnar shape (see fig. 1). The shape of the 2 nd ferromagnetic layer 2 is not limited to the cylindrical shape. The shape of the 2 nd ferromagnetic layer 2 may be, for example, an elliptic cylinder, a quadrangular prism, a cone, an elliptic cone, a truncated cone, a quadrangular frustum, or the like.

The 2 nd ferromagnetic layer 2 contains a ferromagnetic body. The ferromagnetic body used for the 2 nd ferromagnetic layer 2 is the same as the material used for the 1 st ferromagnetic layer 1.

The 2 nd ferromagnetic layer 2 is referred to as a magnetization pinned layer because it is difficult to change the magnetization direction compared to the 1 st ferromagnetic layer 1. The magnetization of the 2 nd ferromagnetic layer 2 is fixed with respect to the magnetization of the 1 st ferromagnetic layer 1, for example, by adjusting the materials used for the 1 st ferromagnetic layer 1 and the 2 nd ferromagnetic layer 2, and the layer structure adjacent to the 2 nd ferromagnetic layer 2. For example, if a material having a larger coercive force than the material constituting the 1 st ferromagnetic layer 1 is used for the 2 nd ferromagnetic layer 2, the magnetization of the 2 nd ferromagnetic layer 2 is fixed with respect to the magnetization of the 1 st ferromagnetic layer 1. For example, an antiferromagnetic layer is laminated on the surface of the 2 nd ferromagnetic layer 2 opposite to the nonmagnetic layer 3, and when the antiferromagnetic layer and the 2 nd ferromagnetic layer 2 are antiferromagnetically coupled, the magnetization of the 2 nd ferromagnetic layer 2 is fixed to the magnetization of the 1 st ferromagnetic layer.

The nonmagnetic layer 3 is located between the 1 st ferromagnetic layer 1 and the 2 nd ferromagnetic layer 2. The nonmagnetic layer 3 extends continuously in the xy plane, for example. The nonmagnetic layer 3 may be dispersed in the xy plane only at a position between the 1 st ferromagnetic layer 1 and the 2 nd ferromagnetic layer 2.

The nonmagnetic layer 3 is made of a nonmagnetic material.

In the non-magnetic layer 3 for insulationIn the case of bulk (as a tunnel barrier layer), the nonmagnetic layer 3 is, for example, Al₂O₃、SiO₂、MgO、MgAl₂O₄And the like. The nonmagnetic layer 3 may Be formed of a material in which Al, Si, and Mg are partially substituted with Zn, Be, or the like. MgO, MgAl₂O₄A coherent tunneling phenomenon can be achieved between the 1 st ferromagnetic layer 1 and the 2 nd ferromagnetic layer 2, and spins can be efficiently injected from the 1 st ferromagnetic layer 1 to the 2 nd ferromagnetic layer 2. When the nonmagnetic layer 3 is a metal, the nonmagnetic layer 3 is, for example, Cu, Au, Ag, or the like. In the case where the nonmagnetic layer 3 is a semiconductor, the nonmagnetic layer 3 is, for example, Si, Ge, CuInSe₂、CuGaSe₂、Cu(In,Ga)Se₂And the like.

The via wiring 4 is electrically connected to the surface of the 1 st ferromagnetic layer 1 opposite to the nonmagnetic layer 3. The communication wiring 4 may be connected directly to the 1 st ferromagnetic layer 1 or may be connected via another layer. The through wiring 4 shown in fig. 1 and 3 protrudes from the 1 st ferromagnetic layer 1 in the-z direction, and a plurality of the through wirings are present apart from each other in the xy plane.

The communication wiring 4 includes a conductor. The interconnection line 4 is, for example, Cu, Al, or Au. The adjacent communication wiring 4 is insulated.

The communication wirings 4 shown in fig. 1 and 3 are disposed at positions corresponding to the plurality of 2 nd ferromagnetic layers 2, respectively. That is, the 2 nd ferromagnetic layers 2 and the communication lines 4 overlap each other when viewed from the z direction.

Next, an example of a method for manufacturing the reservoir device 10 in the neuromorphic device 100 will be described. Fig. 8A to 8D are sectional views showing a method of manufacturing the reservoir element 10 according to embodiment 1.

First, a hole (ホール) is formed in the substrate Sb, and the hole is filled with an electric conductor (fig. 8A). The substrate Sb is, for example, a semiconductor substrate. The substrate Sb is preferably Si or AlTiC, for example. Si and AlTiC easily provide a surface having excellent flatness. For example, Reactive Ion Etching (RIE) is used to form the holes. The conductor charged in the hole serves as a communication wiring 4.

Next, the surfaces of the substrate Sb and the communication wiring 4 are planarized by Chemical Mechanical Polishing (CMP). The 1 st ferromagnetic layer 1, the nonmagnetic layer 3, and the ferromagnetic layer 2' are laminated in this order on the planarized substrate Sb and the interconnection 4 (fig. 8B). For example, the 1 st ferromagnetic layer 1, the nonmagnetic layer 3, and the ferromagnetic layer 2' are laminated using a Chemical Vapor Deposition (CVD) method.

Next, a hard mask HM is formed at a predetermined position on the surface of the ferromagnetic layer 2' (fig. 8C). The portions not covered by the hard mask HM of the ferromagnetic layer 2' are removed by RIE or ion milling. By removing unnecessary portions, the ferromagnetic layer 2' becomes a plurality of 2 nd ferromagnetic layers 2. Finally, the space between the 2 nd ferromagnetic layers 2 is protected by an interlayer insulating film I (fig. 8D). The neuromorphic device 100 of embodiment 1 was obtained through the above-described procedure.

Next, the function of the neuromorphic device 100 will be described. Fig. 12 is a schematic diagram for explaining an example of the operation of the neuromorphic device 100. The input unit 20 has a plurality of input terminals 21, 22, 23, and 24 …. The plurality of input terminals 21, 22, 23, and 24 … constituting the input unit 20 are connected to the respective 2 nd ferromagnetic layers 2 of the reservoir element 10. An input signal is input to the input unit 20 from the outside. The input signal is divided, for example, for each time domain, and is input to the plurality of input terminals 21, 22, 23, and 24 … of the input unit 20 as a plurality of signals S1, S2, S3, S4, S5, and S6 …. Although an example in which the input signal is not processed and input is shown, a signal subjected to Fast Fourier Transform Analysis (FFT Analysis) may be input. The FFT analysis has the effect of extracting frequency features. In addition, the FFT analysis can filter signals of small amplitude caused by noise. The input terminals 21, 22, 23, and 24 … that receive an external signal pass a write current from the corresponding 2 nd ferromagnetic layers 2A, 2B, 2C, and 2D … to the via wiring 4. The input signal is divided in time series in the order of the signal S1, the signal S2, the signal S3, and the signal S4, for example. A write current flows in the order of the 2 nd ferromagnetic layer 2A to which the signal S1 is input, the 2 nd ferromagnetic layer 2B to which the signal S2 is input, the 2 nd ferromagnetic layer 2C to which the signal S3 is input, and the 2 nd ferromagnetic layer 2D to which the signal S4 is input. When the communication lines 4 are arranged at positions corresponding to the respective plural 2 nd ferromagnetic layers 2, the write current flows in the z direction in many cases.

The write current is spin-polarized by the 2 nd ferromagnetic layers 2A, 2B, 2C, and 2D, and reaches the 1 st ferromagnetic layer 1. The spin-polarized current imparts Spin Transfer Torque (STT) to the magnetization of the 1 st ferromagnetic layer 1. The magnetization of the 1 st ferromagnetic layer 1 near the 2 nd ferromagnetic layer 2 where the write current flows is inverted by the STT. Magnetization reversal propagates to the surroundings in accordance with the application time and the application amount of the write current in a manner that water droplets are spread in a ripple manner. Therefore, a domain wall is formed between the portion of the 1 st ferromagnetic layer 1 where the magnetization is inverted and the portion of the 1 st ferromagnetic layer 1 where the magnetization is not inverted, and a domain wall expansion structure is formed due to the magnetization inversion propagating in the 1 st ferromagnetic layer 1. As a result, the direction of magnetization of the 1 st ferromagnetic layer 1 differs between the vicinity of the 2 nd ferromagnetic layers 2A, 2B, 2C, and 2D where the write current flows and other portions where the write current does not flow, and a plurality of magnetic domains are formed in the 1 st ferromagnetic layer 1.

The range of the magnetic domains in the vicinity of the 2 nd ferromagnetic layers 2A, 2B, 2C, and 2D to which the write current is applied varies depending on the application time and the application amount of the write current so that the range of the ripple spread varies depending on the size and the speed of the water droplets dropped on the water surface. When the amount of write current applied is large, the range of the magnetic regions formed in the vicinity of the 2 nd ferromagnetic layers 2A, 2B, 2C, 2D is expanded. The domain wall moves in a direction of expanding from the 2 nd ferromagnetic layers 2A, 2B, 2C, 2D according to the expansion of the magnetic region.

When a write current flows from the plurality of 2 nd ferromagnetic layers 2A, 2B, 2C, 2D to the via wiring 4, a magnetic domain is formed in the vicinity of each of the 2 nd ferromagnetic layers 2A, 2B, 2C, 2D. For example, in the case where magnetization inversion transferred from the 2 nd ferromagnetic layer 2A and magnetization inversion transferred from the 2 nd ferromagnetic layer 2B interfere with each other, a magnetic region reflecting the interference is formed between the 2 nd ferromagnetic layer 2A and the 2 nd ferromagnetic layer 2B. Therefore, the magnetic domains formed in the 1 st ferromagnetic layer 1 serve as magnetic domains reflecting interference of magnetization inversion from the 2 nd ferromagnetic layers 2A, 2B, 2C, and 2D. The magnetization reversal transferred from the 2 nd ferromagnetic layer 2A and the magnetization reversal transferred from the 2 nd ferromagnetic layer 2B interfere with each other more easily than the magnetization reversal transferred from the 2 nd ferromagnetic layer 2A and the magnetization reversal transferred from the 2 nd ferromagnetic layer 2C. This is because the distance between the 2 nd ferromagnetic layer 2A and the 2 nd ferromagnetic layer 2B is shorter than the distance between the 2 nd ferromagnetic layer 2A and the 2 nd ferromagnetic layer 2C. That is, the closer the distance between the 2 nd ferromagnetic layers 2 is, the more easily the input signals S1, S2, S3, S4, S5, and S6 … interfere with each other. Further, the signals S1, S2, S3, S4, S5, and S6 … tend to interfere with each other as the time series becomes closer. Therefore, it is preferable that the signals S1, S2, S3, S4, S5, and S6 … in the time series are input to the 2 nd ferromagnetic layer 2 in a shorter distance. For example, the distance between the 2 nd ferromagnetic layer 2A to which the signal S1 is input and the 2 nd ferromagnetic layer 2B to which the signal S2 is input is preferably shorter than the distance between the 2 nd ferromagnetic layer 2A to which the signal S1 is input and the 2 nd ferromagnetic layer 2C to which the signal S3 is input. When the write current is stopped from being applied to the reservoir element 10, the magnetic state of the 1 st ferromagnetic layer 1 is stored non-volatilely.

Finally, the signal is output from the reservoir element 10 to the output section 30. The output unit 30 has, for example, a plurality of output terminals 31, 32, …. The plurality of output terminals 31 and 32 … are connected to any 2 nd ferromagnetic layer 2. Fig. 12 shows an example in which the second ferromagnetic layers 2E and 2F different from the 2 nd ferromagnetic layers 2A, 2B, 2C, and 2D to which the input terminals 21, 22, 23, and 24 … for input are connected. The input and output can be switched, and the 2 nd ferromagnetic layers 2A, 2B, 2C, and 2D to which the input terminals 21, 22, 23, and 24 … for input are connected may be connected to the output terminals 31 and 32 … for output. A signal is output by flowing a read current from the 2 nd ferromagnetic layer 2 to the via wiring 4. The read current has a lower current density than the write current, and does not invert the magnetization of the 1 st ferromagnetic layer 1.

When a readout current flows in the reservoir element 10, the difference in relative angle between the magnetization of the 2 nd ferromagnetic layer 2 and the magnetization of the 1 st ferromagnetic layer 1 at the position overlapping with the 2 nd ferromagnetic layer 2 is output as a change in resistance value. The direction of magnetization of the 1 st ferromagnetic layer 1 at a position overlapping with the 2 nd ferromagnetic layer 2 is affected by a magnetic domain spreading from the vicinity of the other 2 nd ferromagnetic layer. That is, a signal read from one 2 nd ferromagnetic layer 2 contains information written to the other 2 nd ferromagnetic layer 2, and the information is compressed.

The finally compressed signal is transmitted to the output section 30 via the plurality of output terminals 31, 32, …. The output unit 30 weights the signals read from the 2 nd ferromagnetic layer 2 by learning. Fig. 13 is a schematic diagram for explaining another example of the operation of the neuromorphic device 100. Fig. 13 shows a method of dividing an input signal, and the connection positions of the input terminals 21, 22, 23, and 24 and the output terminals 31 and 32 … are different from the example shown in fig. 12. In the example shown in fig. 13, the input signal is not overlapped with the signals S1, S2, S3, and S4 … and is time-sequentially divided. In the neuromorphic device shown in fig. 13, the 2 nd ferromagnetic layers 2A, 2B, 2C, and 2D of the signals S1, S2, S3, and S4 …, which are separated in time series, are separated from each other by a larger distance.

Fig. 14 is a schematic diagram for explaining another example of the neuromorphic device. The neuromorphic device shown in fig. 14 differs from the example shown in fig. 13 in that it has the 2 nd output unit 40. The terminals 41, 42, 43 of the 2 nd output unit 40 are connected to the output terminals 31, 32 of the output unit 30 via synapses Sp. When information is transmitted from the output terminals 31 and 32 of the output unit 30 to the terminals 41, 42, and 43 of the 2 nd output unit 40, data is weighted by the synapse Sp. The neuromorphic element shown in fig. 14 learns between the output unit 30 and the 2 nd output unit 40. That is, the neuromorphic element shown in fig. 14 is a structure of a deep neural network. By having the neuromorphic element as a deep neural network structure, more complex information can be identified.

Fig. 15 is a schematic diagram for explaining another example of the neuromorphic device. The neuromorphic element shown in fig. 15 differs from the example shown in fig. 14 in that reservoir elements 10 are arranged in parallel and an output unit 30 connected to each reservoir element 10 is shared. With such a configuration, it is possible to simultaneously recognize signals having different signal speeds and outputs from a plurality of sensors, and to realize a multi-mode (multimodal) reservoir element.

As described above, in the reservoir element 10 according to embodiment 1, magnetization inversions from the respective 2 nd ferromagnetic layers 2 interfere with each other in the 1 st ferromagnetic layer 1, and the magnetic regions formed therebetween reflect the interaction thereof. The signals input from the input section 20 interact with each other in the 1 st ferromagnetic layer 1, and one magnetic state is generated in the 1 st ferromagnetic layer 1. That is, a signal input from the input unit 20 is compressed to one magnetic state in the 1 st ferromagnetic layer 1. Thus, the neuromorphic element 100 of embodiment 1 utilizes reservoir element 10 to properly compress the signal. By compressing the signal, only the output unit 30 needs to learn, and the power consumption of the neuromorphic device 100 is reduced. The magnetic state of the 1 st ferromagnetic layer 1 can be maintained in a nonvolatile manner unless a new write current is applied.

By maintaining the information in a nonvolatile manner, the reserve pool element 10 is not limited in time. When the time-series data is input to the input unit 20 of the reservoir element 10 and is taken out from the output unit 30 to be subjected to information processing, it is necessary to match the time interval between input and output with the time interval to be detected by the reservoir element 10. The magnetization reversal and domain wall driving operations of the reservoir element 10 are generally completed in a time of 1nsec to 1 μ sec or less. However, since the motion of a person or an object generally occurs in a time unit of about 1sec, the motion speed of one terminal of the reservoir element 10 and the motion speed of the person or the object have a large time difference. In order to cause the reservoir element 10 to function, it is preferable that at least the influence of the operation of one terminal of the reservoir element 10 is left between the operations of a person or thing. Since the reservoir element 10 can hold the input information in a nonvolatile manner, interference between the input signals in the reservoir element 10 can be maintained even if there is a large time difference between the operating speed of one terminal of the reservoir element 10 and the operating speed of a person or object.

"embodiment 2

Fig. 9 is a sectional view of the reservoir member of embodiment 2. The reservoir element 11 of embodiment 2 is different from the reservoir element 10 of embodiment 1 in that it does not have a plurality of communication wires 4. The other structures are the same as those of the reservoir member 10 of embodiment 1, and description thereof is omitted. In fig. 9, the same components as those in fig. 1 are denoted by the same reference numerals.

In the reservoir element 11, the communication wiring 4 is one. The via wiring 4 is electrically connected to the 1 st ferromagnetic layer 1.

As shown in fig. 1, when the via wirings 4 are arranged at positions corresponding to the plurality of 2 nd ferromagnetic layers 2, respectively, the write current flows in the z direction in many cases. On the other hand, when there is one through wiring 4, a part of the write current flows in the 1 st ferromagnetic layer 1 in the xy plane. The spin-polarized write current moves a magnetic domain wall that is a boundary of different magnetic regions. That is, if there is one communication line 4 provided in the reservoir element 11, the domain wall moves efficiently in the 1 st ferromagnetic layer 1, and interaction between a signal input to one 2 nd ferromagnetic layer 2 and a signal input to the other 2 nd ferromagnetic layer 2 is promoted.

In addition, when there is one connection wiring 4, the distance between the 2 nd ferromagnetic layer 2 and the connection wiring 4 is different for each 2 nd ferromagnetic layer 2. The amount of write current flowing in the xy plane of the 1 st ferromagnetic layer 1 differs depending on which 2 nd ferromagnetic layer 2a signal is input to. That is, the ease of movement of the domain wall is not uniform depending on which of the 2 nd ferromagnetic layers 2a signal is input to. In other words, the pool element 11 can preferentially output the predetermined information in the signal input from the input section 20 and give a weight to the necessary information in advance.

The reservoir member 11 of embodiment 2 can be applied to the neuromorphic member 100. Further, the reserve tank element 11 according to embodiment 2 achieves the same effects as those of the reserve tank element 10 according to embodiment 1. The reservoir element 11 of embodiment 2 has different characteristics in the ease of flowing the write current, and gives priority to the signal.

The reservoir element 11 of embodiment 2 can be variously modified. For example, the via wiring 4 need not be formed downward from one surface of the 1 st ferromagnetic layer 1, and may be provided as a wiring on a side surface of the 1 st ferromagnetic layer 1.

"embodiment 3

Fig. 10 is a sectional view of the reservoir member of embodiment 3. The reservoir element 12 of embodiment 3 differs from the reservoir element 10 of embodiment 1 in that it has the common electrode layer 5. The other structures are the same as those of the reservoir member 10 of embodiment 1, and description thereof is omitted. In fig. 10, the same components as those in fig. 1 are denoted by the same reference numerals.

The common electrode layer 5 connects at least two or more of the plurality of via wirings 4. The common electrode layer 5 is continuously extended in the xy plane, for example. The common electrode layer 5 is made of the same material as the via wiring 4.

If the reservoir element 12 has the common electrode layer 5, a part of the write current flows in the 1 st ferromagnetic layer 1 in the xy plane. The domain wall effectively moves within the 1 st ferromagnetic layer 1, facilitating the interaction of a signal input to one 2 nd ferromagnetic layer 2 and a signal input to the other 2 nd ferromagnetic layer 2, and can exhibit a more complicated phenomenon.

The reservoir member 11 of embodiment 3 can be applied to the neuromorphic element 100. Further, the reserve tank element 12 according to embodiment 3 achieves the same effects as those of the reserve tank element 10 according to embodiment 1.

"embodiment 4

Fig. 11 is a sectional view of the reserve cell element of embodiment 4. The reserve cell element 13 of embodiment 4 is different from the reserve cell element 10 of embodiment 1 in that it has a magnetic interference layer 6. The other structures are the same as those of the reservoir member 10 of embodiment 1, and description thereof is omitted. In fig. 11, the same components as those in fig. 1 are denoted by the same reference numerals.

The magnetic interference layer 6 is in contact with the surface of the 1 st ferromagnetic layer 1 opposite to the nonmagnetic layer 3. The magnetic interference layer 6 extends continuously in the xy plane.

The coercive force of the magnetic interference layer 6 is smaller than that of the 1 st ferromagnetic layer 1, and the soft magnetic characteristics are excellent. That is, the magnetization of the magnetic interference layer 6 is more easily inverted than the magnetization of the 1 st ferromagnetic layer 1. The magnetic interference layer 6 is an alloy containing any of Fe-Si, Fe-Si-Al, Fe-Co-V, Ni-Fe, Co-Fe-Si-B, for example.

When a write current flows in the reservoir element 13, a different magnetic region is formed in the magnetic interference layer 6, as in the 1 st ferromagnetic layer 1, and a magnetic domain wall is generated. The domain walls of the magnetic interference layer 6 move more easily than the domain walls of the 1 st ferromagnetic layer 1. The magnetic interference layer 6 is responsible for the magnetic correlation over long distances.

The freedom of choice of the material of the 1 st ferromagnetic layer 1 is increased if the reservoir element 13 has a magnetic interference layer 6.

The magnetic resistance change is caused by a change in the magnetic state of the two magnetic bodies (the 1 st ferromagnetic layer 1 and the 2 nd ferromagnetic layer 2) sandwiching the nonmagnetic layer 3. The 1 st ferromagnetic layer 1 preferably contains a material (e.g., MgO, MgAl) that easily obtains a coherent tunneling effect with the 2 nd ferromagnetic layer 2₂O₄)。

On the other hand, a material which easily obtains a coherent tunneling effect is not necessarily a material in which a magnetic domain wall easily moves. The 1 st ferromagnetic layer 1 of the reservoir element 12 is responsible for the magnetoresistance change, and the magnetic interference layer 6 is responsible for the magnetic correlation over a long distance. That is, the 1 st ferromagnetic layer 1 does not require a material for facilitating the movement of the domain wall, and the degree of freedom in selecting the material of the 1 st ferromagnetic layer 1 is improved.

The reservoir member 13 of embodiment 4 can be applied to the neuromorphic member 100. Further, the reserve tank element 13 according to embodiment 4 achieves the same effects as those of the reserve tank element 10 according to embodiment 1. In addition, the reservoir element 13 according to embodiment 4 can exhibit a more complicated phenomenon by sharing functions between the 1 st ferromagnetic layer 1 and the magnetic interference layer 6, and promoting the interaction between a signal input to one 2 nd ferromagnetic layer 2 and a signal input to the other 2 nd ferromagnetic layer 2.

Fig. 16 is a plan view of the reservoir member of the 5 th embodiment. The reservoir element 14 of embodiment 5 differs from the reservoir element 10 of embodiment 1 in that the 1 st ferromagnetic layer and the nonmagnetic layer 3' are annular, and the plurality of 2 nd ferromagnetic layers 2 are dispersed along the annular 1 st ferromagnetic layer. The other structures are the same as those of the reservoir member 10 of embodiment 1, and description thereof is omitted. In fig. 16, the same components as those in fig. 1 are denoted by the same reference numerals.

The input signal is divided, for example, for each time domain, and is input to the pool element 14 as a plurality of signals S1, S2, S3, and S4 …. For example, the signal S1 is input to the 2 nd ferromagnetic layer 2A, the signal S2 is input to the 2 nd ferromagnetic layer 2B, and the signal S3 is input to the 2 nd ferromagnetic layer 2C. The signals S1, S2, S3, S4 … invert the magnetization of a portion of the 1 st ferromagnetic layer. The magnetization inversions spreading from the respective 2 nd ferromagnetic layers 2A, 2B, 2C propagate in the circumferential direction, and interfere with each other.

The magnetization state of the 1 st ferromagnetic layer is output from, for example, the 2 nd ferromagnetic layers 2E, 2F, 2G. If the 2 nd ferromagnetic layers 2A, 2B, 2C to which a signal is input and the 2 nd ferromagnetic layers 2E, 2F, 2G to which a signal is output are made different terminals, a part of the write current flows in the circumferential direction along the 1 st ferromagnetic layer. The spin-polarized current flowing in the circumferential direction moves the domain wall, and promotes interference of magnetization reversal spreading from each of the 2 nd ferromagnetic layers 2A, 2B, and 2C.

In addition, the input position of the signal to the 2 nd ferromagnetic layer and the output position of the signal from the 2 nd ferromagnetic layer may be sequentially changed.

Although a preferred embodiment of the present invention has been described in detail, the present invention is not limited to the embodiment, and various modifications and changes can be made within the scope of the gist of the present invention described in the scope of the claims.

For example, the features of the reserve cell elements 10 to 14 according to embodiments 1 to 5 may be combined.