Spintronics element and magnetic memory device

Abstract

A spintronics element (10) comprises: a channel layer (30); a free layer (21) laminated on the channel layer (30); a barrier layer (22) laminated on the free layer (21); and a reference layer (23) laminated on the barrier layer (22). The direction of magnetization in the reference layer (23) is fixed. The direction of magnetization in the free layer (21) changes due to spin-orbit torque, which is generated by a write current that flows through the channel layer (30). The channel layer (30) is formed of a tungsten alloy containing at least one of iron and nickel in an amount of 35at% or less.

Description

Spintronics elements and magnetic memory devices

[0001] The present invention relates to a spintronics element and a magnetic memory device.

[0002] Magnetoresistive Random Access Memory (MRAM) is known as a next-generation nonvolatile memory capable of high-speed operation. Among MRAMs, SOT (Spin Orbit Torque)-MRAM, which is an MRAM that uses spin orbit torque, has attracted attention as a nonvolatile memory that requires high-speed writing (see, for example, Patent Documents 1 and 2).

[0003] Patent No. 6178451 International Publication No. 2020 / 166722

[0004] In SOT-MRAM, data is written by generating a spin current using a write current flowing through the channel layer, and the spin current applies a spin-orbit torque to the free layer of the magnetoresistive element. In nonvolatile memories that write data using this type of spin-orbit torque, it is necessary to efficiently generate a spin current in order to improve the efficiency of data writing.

[0005] The present invention has been made to solve the above-mentioned problems, and has an object to provide a spintronics element and a magnetic memory device that can efficiently generate a spin current.

[0006] In order to achieve the above object, a spintronics element according to a first aspect of the present invention is a spintronics element comprising a channel layer, a free layer stacked on the channel layer, a barrier layer stacked on the free layer, and a reference layer stacked on the barrier layer, wherein the direction of magnetization in the reference layer is fixed, and the direction of magnetization in the free layer is changed by a spin-orbit torque generated by a write current flowing through the channel layer, and the channel layer is formed from a tungsten alloy containing at least one of iron and nickel in a range of 35 at% or less.

[0007] In order to achieve the above object, a magnetic memory device according to a second aspect of the present invention is a magnetic memory device in which a plurality of memory cells are arranged in a matrix, each of the plurality of memory cells having the spintronics element described above and connected to a bit line and a word line.

[0008] According to the present invention, it is possible to provide a spintronics element and a magnetic memory device that can efficiently generate a spin current.

[0009] 7 is a schematic diagram showing the configuration of a spintronics element according to an embodiment. FIG. 8 is a diagram showing how spin current is generated in the channel layer of the spintronics element according to an embodiment. FIG. 9 is a diagram showing the relationship between resistivity and spin current conversion efficiency for a number of known substances. FIG. 10 is a diagram showing the measurement results of resistivity in a first experiment. FIG. 11 is a diagram showing the measurement results of spin Hall conductivity in a first experiment. FIG. 12 is a diagram showing the measurement results of electrical resistivity, spin Hall angle, and spin diffusion length in a second experiment. FIG. 13 is an example of a circuit configuration diagram of a magnetic memory device according to an embodiment. FIG. 14 is an example of a circuit configuration diagram of a memory cell constituting the magnetic memory device shown in FIG. 7. FIG. 15 is a diagram showing the flow of current when writing data in the spintronics element according to an embodiment. FIG. 16 is a diagram showing the flow of current when reading data in the spintronics element according to an embodiment.

[0010] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings, in which the same or corresponding parts are designated by the same reference numerals.

[0011] 1, the spintronics element 10 according to this embodiment includes a magnetoresistive element 20 and a channel layer 30. In the following description, an XYZ orthogonal coordinate system is set, in which the short-side direction of the channel layer 30 is the X-axis direction, the long-side direction (extension direction) of the channel layer 30 is the Y-axis direction, and the height direction of the stack of the magnetoresistive element 20 and the channel layer 30 is the Z-axis direction, and will be referred to as appropriate.

[0012] The magnetoresistive element 20 has a stacked structure in which a free layer 21, a barrier layer 22, and a reference layer 23 are stacked in this order. The stacking direction corresponds to the perpendicular direction. The magnetoresistive element 20 is a memory element that stores data using the magnetoresistive effect. Specifically, the magnetoresistive element 20 is a magnetic tunnel junction element (MTJ element), which is a resistance change type memory element.

[0013] The free layer 21 is stacked on the channel layer 30. The free layer 21 is a magnetic layer and is made of a magnetic material such as CoFeB. The free layer 21 has a magnetization M21 therein. The free layer 21 is also called a memory layer, a recording layer, etc.

[0014] The barrier layer 22 is stacked on the free layer 21. The barrier layer 22 is a very thin non-magnetic insulating layer that allows a tunneling current to flow, and is made of a non-magnetic material such as MgO. The barrier layer 22 is also called a barrier layer, an insulating layer, a tunnel barrier layer, etc.

[0015] The reference layer 23 is stacked on the barrier layer 22. The reference layer 23 is a magnetic layer and is formed of a magnetic material such as CoFeB. The reference layer 23 has a magnetization M23 therein. The reference layer 23 is also called a fixed layer, a pinned layer, or the like. A first terminal T1 is connected to the reference layer 23.

[0016] The channel layer 30 is a region where a spin induced torque is generated when a write current flows. The channel layer 30 is formed of a material with a large spin-orbit interaction, such as a heavy metal. The channel layer 30 extends in the Y-axis direction and has a plate-like shape that extends across the XY plane. The channel layer 30 is also called a conductive layer, a wiring layer, a non-magnetic layer, etc.

[0017] Here, "the X layer is formed from Y" means that the X layer contains Y as its main component, and is not limited to the case where the X layer is made of only Y, but the X layer may also contain a substance other than Y as long as the substance does not affect the function of the X layer. The same applies hereinafter.

[0018] A second terminal T2 is connected to one end of the channel layer 30 in the Y-axis direction, and a third terminal T3 is connected to the other end of the channel layer 30 in the Y-axis direction. A write current, which will be described later, flows between the second terminal T2 and the third terminal T3. The magnetoresistive element 20 is stacked on the channel layer 30 at a position between the second terminal T2 and the third terminal T3.

[0019] In the magnetoresistive element 20, the orientation of the magnetization M23 in the reference layer 23 is fixed and always points in the same direction, whereas the orientation of the magnetization M21 in the free layer 21 is variable and can be either parallel or anti-parallel to the orientation of the magnetization M23 in the reference layer 23.

[0020] When the direction of magnetization M21 in free layer 21 and the direction of magnetization M23 in reference layer 23 are parallel (parallel state), the electrical resistance between free layer 21 and reference layer 23 via barrier layer 22 is low. In contrast, when the direction of magnetization M21 in free layer 21 and the direction of magnetization M23 in reference layer 23 are antiparallel (antiparallel state), the electrical resistance between free layer 21 and reference layer 23 via barrier layer 22 is high. This phenomenon in which the resistance value changes depending on the magnetization direction of a magnetic material is called the "magnetoresistive effect."

[0021] The magnetoresistive element 20 functions as a memory element by storing the direction of magnetization M21 in the free layer 21 as "0" or "1" and reading the stored information on the direction of magnetization M21 using the magnetoresistive effect. Data is written to and read from the magnetoresistive element 20 by reversing the magnetization M21 in the free layer 21.

[0022] In STT-MRAMs that use spin transfer torque (STT) as a write method, a current is passed from the free layer 21 to the reference layer 23, or from the reference layer 23 to the free layer 21, via the barrier layer 22, causing the spin moment of the reference layer 23 to act on the free layer 21. This switches the direction of magnetization M21 in the free layer 21 to be parallel or anti-parallel to the direction of magnetization M23 in the reference layer 23. Specifically, when a current is passed in the direction from the free layer 21 to the reference layer 23, a spin torque acts in a direction in which the magnetization M21 in the free layer 21 becomes parallel to the magnetization M23 in the reference layer 23. On the other hand, when a current is passed from the reference layer 23 to the free layer 21, a spin torque acts in a direction in which the magnetization M21 in the free layer 21 becomes anti-parallel to the magnetization M23 in the reference layer 23.

[0023] In STT-MRAM, data is written by passing a current through the barrier layer 22, which results in poor rewrite endurance. In contrast, SOT-MRAM, which uses spin orbit torque (SOT), is being researched and developed as a memory that solves these problems associated with STT-MRAM. In SOT-MRAM, the magnetization M21 in the free layer 21 is reversed using the spin Hall effect. The spin Hall effect is a phenomenon in which, when a current is passed through a material with a large spin orbit interaction, such as a nonmagnetic material, a spin current is generated in a direction perpendicular to the current due to scattering caused by the spin orbit interaction. In the spintronics device 10 according to this embodiment, the spin Hall effect is used to apply a spin orbit torque to the free layer 21 using the spin current generated in the channel layer 30, thereby reversing the magnetization M21 in the free layer 21.

[0024] The spin Hall effect in the channel layer 30 will be described with reference to Fig. 2. When an electron current Ie is passed through the channel layer 30 in the positive direction of the Y-axis, i.e., a current is passed through the channel layer 30 in the negative direction of the Y-axis, electrons spin-polarized in the positive direction of the X-axis and electrons spin-polarized in the negative direction of the X-axis are scattered separately in the positive direction of the Z-axis and the negative direction of the Z-axis, respectively. As a result, the electrons scattered in the positive direction of the Z-axis accumulate on the upper surface side (positive direction of the Z-axis) of the channel layer 30, and the electrons scattered in the negative direction of the Z-axis accumulate on the lower surface side (negative direction of the Z-axis) of the channel layer 30. This is called "spin accumulation."

[0025] Due to spin accumulation, electrons with opposite spin angular momentum (hereinafter simply referred to as "spin") are scattered in opposite directions, generating a spin current Is in the perpendicular direction (z-axis direction). When the spin current Is is generated, a spin-orbit torque is generated by the electrons spin-polarized in the in-plane direction.

[0026] At this time, the spin s, the spin current Is, and the electron current Ie satisfy the relationship of the following formula (1). That is, the generated spin current Is is proportional to the cross product of the spin s and the electron current Ie. Here, the spin Hall angle θ ＳＨ is a parameter indicating the conversion efficiency from the electron current Ie to the spin current Is. The direction of the electron current Ie is opposite to that of the electric current. Is ∝ θ ＳＨ (s × Ie) ... (1)

[0027] As described above, when a current (electron flow Ie) is passed through the channel layer 30 in the longitudinal direction (Y-axis direction), a spin current Is spin-polarized in the in-plane direction (X-axis direction) is generated in the perpendicular direction (Z-axis direction). Here, to determine the rotation direction of the magnetization M21 in the free layer 21, a bias magnetic field Hy must be applied in one direction to slightly tilt the magnetization M21 toward the bias magnetic field Hy. FIG. 2 shows, as an example, a case where the bias magnetic field Hy is applied in the positive direction of the Y-axis. The bias magnetic field Hy is a magnetic field generated by a magnet or an electrically generated external magnetic field. Electrons spin-polarized in the in-plane direction at the interface with the magnetoresistive element 20 exert a spin-orbit torque on the magnetization M21 in the free layer 21. As a result, the magnetization M21 rotates in the direction determined by the bias magnetic field Hy, resulting in magnetization reversal.

[0028] In this way, the spin current Is generated in the channel layer 30 applies a spin orbit torque to the free layer 21 stacked on the channel layer 30, thereby reversing the magnetization M21. Reversing the polarity of the current flowing through the channel layer 30 also reverses the spin orbit torque acting on the free layer 21. In other words, by controlling the current flowing through the channel layer 30, the direction of the magnetization M21 in the free layer 21 can be oriented in a desired direction, either parallel or anti-parallel to the direction of the magnetization M23 in the reference layer 23. In the SOT-MRAM and the spintronics device 10, data is written based on this principle.

[0029] In SOT-MRAM that uses the principle of spin-orbit torque, it is desirable to be able to efficiently obtain a large spin current Is with as small an electron current Ie as possible from the viewpoint of power saving. Therefore, it is an important challenge to find a material with high spin current conversion efficiency to be used in the channel layer 30.

[0030] Spin current conversion efficiency varies depending on the material. Specifically, Figure 3 shows the spin current conversion efficiency of various known materials. Note that Figure 3 is a partially modified version of a figure published in "Lijun Zhu, et al., "Maximizing spin-orbit torque generated by the spin Hall effect of Pt", Applied Physics Reviews Rev. 8, 031308 (2021); AUGUST 13 2021."

[0031] The horizontal axis in Figure 3 shows resistivity ρ ｘｘ (μΩcm) is a parameter indicating the difficulty of electrical conduction in each material. The spin current conversion efficiency on the vertical axis is a value indicating the efficiency of conversion from electron current Ie to spin current Is, and specifically, the spin Hall angle θ ＳＨ To T ｉｎｔ (spin transparency of the magnetic interface) ｊ ＤＬ(damping-like SOT efficiency per unit current density). Spin Hall angle θ ＳＨ is a parameter indicating the conversion efficiency from the electron current Ie to the spin current Is, as shown in the above formula (1). ＳＨ The larger the value, the more efficiently the spin current Is can be generated.

[0032] As shown in Figure 3, the spin current conversion efficiency varies depending on the material. As a general trend, the resistivity ρ ｘｘ The larger the resistivity ρ, the higher the spin current conversion efficiency. ｘｘ As the resistivity ρ increases, the electron current Ie becomes difficult to flow, and therefore, a large amount of power is required to generate a large spin current Is. Therefore, in order to efficiently generate a large spin current Is with a small electron current Ie, the resistivity ρ ｘｘ It is desirable to use a substance that has a small spin current and a high spin current conversion efficiency as the material for the channel layer 30 .

[0033] In the spintronics device 10 according to this embodiment, the channel layer 30 is formed of a tungsten alloy containing at least one of iron (Fe) and nickel (Ni) in a range of 35 at% or less. In other words, the channel layer 30 is formed of a tungsten alloy in which the content x of at least one of iron and nickel in terms of element composition ratio is in the range of 0 < x ≦ 0.35. The reason why the spintronics device 10 according to this embodiment has a channel layer 30 with such a configuration will be explained below. In the following, the content x of metal elements other than tungsten in the tungsten alloy will be explained assuming that the element composition ratio is expressed as a value between 0 and 1 (at% when expressed in %).

[0034] Tungsten (W) is known as a material that is relatively inexpensive, readily available, and exhibits a large spin Hall effect. Specifically, the resistivity of elemental tungsten is ρ ｘｘRegarding the spin current conversion efficiency of tungsten, in "Takahiro Ishikawa et al., "Large intrinsic spin Hall conductivity in orthorhombic tungsten", PHYSICAL REVIEW MATERIALS 7, 026202 (2023)", they explored the crystal structure of tungsten using a computational science method that combined evolutionary algorithms and first-principles electronic structure calculations, and found that certain crystal structures of tungsten exhibit a large spin Hall effect.

[0035] In view of these results, in this embodiment, attention is focused on tungsten as the material for the channel layer 30. Specifically, in this embodiment, a tungsten alloy, which is an alloy in which tungsten is combined with a metal element other than tungsten, is produced. Then, as shown by the arrow in FIG. 3, the resistivity ρ ｘｘ A tungsten alloy that increases the spin-current conversion efficiency without increasing the .DELTA. as much as possible was searched for through the first and second experiments described below.

[0036] <First Experiment> In the first experiment, the resistivity ρ of a plurality of types of tungsten alloys was measured when the tungsten content was changed in a plurality of ways. ｘｘ and the spin Hall conductivity σ ｚ ｘｙ 4 and 5 show the resistivity ρ ｘｘ and the spin Hall conductivity σ ｚ ｘｙ Here, the spin Hall conductivity σ ｚ ｘｙ is a parameter that indicates the ease of generation of the spin current Is due to the spin Hall effect, and the spin Hall angle θ ＳＨ Similarly, it is one of the indicators of spin current conversion efficiency.

[0037] In FIGS. 4 and 5, “W （１－ｘ） A ｘ" represents a tungsten alloy containing a metal element A at a content rate x (0<x<1). Here, the metal element A is any of chromium (Cr), manganese (Mn), iron (Fe), cobalt (Co), nickel (Ni), copper (Cu), nickel-iron alloy (Py), and terbium (Tb). The nickel-iron alloy (Py) is also called permalloy, and is an alloy in which iron (Fe) and nickel (Ni) are combined in a composition ratio of 1:4 (20 at%:80 at%). For each of these multiple types of metal element A, the resistivity ρ of the tungsten alloy in which the content rate x is changed to various values ​​between 0 and 1 is ｘｘ and the spin Hall conductivity σ ｚ ｘｙ was measured.

[0038] The measurement method used in the first experiment was the same as that described in "Junyeon Kim et al., "Spin Hall Magnetoresistance in Metallic Bilayers", PHYSICAL REVIEW LETTERS, 116, 097201 (2016)" (hereinafter referred to as "Kim et al."). Specifically, a measurement system was adopted in which CoFeB was used as the material for the free layer 21, MgO was used as the material for the barrier layer 22, and no reference layer 23 was provided. As described above, multiple types of tungsten alloys were used as the material for the channel layer 30, with the content x of each of multiple types of metal element A varying between 0 and 1. In the first experiment, the spin diffusion length λ Ｎ In such a measurement system, when each of the plurality of types of tungsten alloys was used for the channel layer 30, the resistivity ρ ｘｘ and the spin Hall conductivity σ ｚ ｘｙ More specifically, the potential difference that occurs when a current is applied to the channel layer 30 is measured, and the resistivity ρ is calculated from the ratio of the applied current to the measured potential difference. ｘｘ The spin Hall conductivity σ ｚ ｘｙ is the measured resistivity ρ ｘｘand the spin Hall angle θ based on equation (1) of Kim et al. ＳＨ Calculate the calculated spin Hall angle θ ＳＨ is the spin Hall conductivity σ ｚ ｘｙ In the first experiment, the thickness of the channel layer 30 was fixed to one value. In the first experiment, the spin Hall conductivity σ ｚ ｘｙ Since the content x is measured for various values, the measurement accuracy per measurement is lower than in the second experiment described below.

[0039] Resistivity ρ ｘｘ As shown in FIG. （１－ｘ） Cu ｘ , W （１－ｘ） Mn ｘ , and W （１－ｘ） Tb ｘ In other words, the resistivity ρ of these three types of tungsten alloys was small when the content x was small, and tended to increase as the content x increased. ｘｘ Although the value is small for tungsten alone, it tends to increase as the ratio of tungsten in the tungsten alloy decreases. On the other hand, the other five types of tungsten alloys, W （１－ｘ） Cr ｘ , W （１－ｘ） C ｘ , W （１－ｘ） Fe ｘ , W （１－ｘ） Ni ｘ , and W （１－ｘ） Py ｘ For resistivity ρ ｘｘ did not show a significant increase even when the content x increased.

[0040] Spin Hall conductivity σ ｚ ｘｙ As shown in Figure 5, the spin Hall conductivity σ is generally large when the content x is small and tends to decrease as the content x increases. ｚ ｘｙis large for tungsten alone, but tends to become smaller as the ratio of tungsten in the tungsten alloy decreases. （１－ｘ） Ni ｘ Then, the spin Hall conductivity σ ｚ ｘｙ became larger than near x = 0. And, W （１－ｘ） Py ｘ However, the spin Hall conductivity σ ｚ ｘｙ is slightly larger than near x = 0. （１－ｘ） Fe ｘ In comparison with the five tungsten alloys other than Ni, Fe, and Py, the spin Hall conductivity σ ｚ ｘｙ The decline was gradual.

[0041] From the measurement results in the first experiment, W （１－ｘ） Ni ｘ and W （１－ｘ） Py ｘ When x is around 0.2, the resistivity ρ ｘｘ It can be predicted that it is possible to increase the spin current conversion efficiency while suppressing the increase in W. （１－ｘ） Fe ｘ For the spin Hall conductivity σ ｚ ｘｙ The decrease in the temperature is gradual, and the W containing iron （１－ｘ） Py ｘ Spin Hall conductivity σ at ｚ ｘｙ Considering the measurement results of resistivity ρ ｘｘ It is thought that there is a possibility that the spin-current conversion efficiency can be increased while suppressing the increase of . Therefore, in this embodiment, a second experiment was carried out to measure the spin-current conversion efficiency in more detail for a tungsten alloy that combines tungsten with at least one of iron and nickel.

[0042] <Second Experiment> In the second experiment, an alloy containing 80 at% tungsten and 20 at% iron was used. 0.8 Fe 0.2 and W, an alloy containing tungsten at a content of 80 at% and nickel at a content of 20 at%. 0.8 Ni 0.2 For each of and, the resistivity ρ ｘｘ and the spin Hall angle θ ＳＨ and were measured.

[0043] As in the first experiment, the measurement method in the second experiment was the same as that described in Kim et al. However, in the second experiment, the content x was fixed at 0.2 and the spin diffusion length λ Ｎ In this experiment, the measurement was performed with higher accuracy than in the first experiment, without fixing the value of tungsten to a single value. Specifically, a measurement system was adopted in which CoFeB was used as the material for the free layer 21, MgO was used as the material for the barrier layer 22, and the reference layer 23 was not present. In this measurement system, two cases were used for the material of the channel layer 30: (a) tungsten alone, and (b) W 0.8 Fe 0.2 and (c) W 0.8 Ni 0.2 When using and when using, the resistivity ρ ｘｘ and the spin Hall angle θ ＳＨ and the spin diffusion length λ Ｎ More specifically, in the second experiment, the thickness of the channel layer 30 was changed in several ways, and the resistivity ρ ｘｘ and measured the spin Hall magnetoresistance ΔR expressed by equation (1) of Kim et al. ＳＭＲ ＸＸ The dependence of the spin Hall angle θ on the thickness of the channel layer 30 (for example, the characteristics shown in Fig. 3 of Kim et al.) was derived. Then, similar to Kim et al., the spin Hall angle θ was calculated based on the peak value of the derived dependence. ＳＨ and the spin diffusion length λ ＮIn this way, the content x was fixed at 0.2, and the thickness of the channel layer 30 was changed in several ways for measurement. Therefore, in the second experiment, when the content x was 0.2, the spin Hall angle θ , which is an index of the spin-current conversion efficiency, was measured with higher accuracy than in the first experiment. ＳＨ can be measured.

[0044] As a result of the measurement in the second experiment, as shown in FIG. ｘｘ (a) When the channel layer 30 was made of tungsten alone, the resistance was 137 μΩcm. (b) When the channel layer 30 was made of W 0.8 Fe 0.2 When the channel layer 30 is made of W, the resistance is 136 μΩcm. 0.8 Ni 0.2 In other words, when the channel layer 30 is formed of W, the resistance is 134 μΩcm. 0.8 Fe 0.2 Or W 0.8 Ni 0.2 When the channel layer 30 is formed of tungsten alone, the resistivity ρ ｘｘ On the other hand, the spin Hall angle θ ＳＨ was 0.24 when the channel layer 30 was made of tungsten alone, whereas 0.8 Fe 0.2 (c) When the channel layer 30 is formed of W, the value is 0.43. 0.8 Ni 0.2 In other words, when the channel layer 30 is formed of W, the value is 0.33. 0.8 Fe 0.2 Or W 0.8 Ni 0.2 When the channel layer 30 is formed of tungsten alone, the spin Hall angle θ ＳＨ The spin diffusion length λ Ｎ were 0.96 nm, 0.37 nm, and 0.56 nm in the cases (a) to (c), respectively.

[0045] From the measurement results of the second experiment, it was found that the W 0.8 Fe 0.2 Or W 0.8 Ni 0.2 When tungsten is used for the channel layer 30, the resistivity ρ ｘｘ While maintaining the spin Hall angle θ ＳＨ In other words, for the tungsten alloy containing at least one of iron and nickel obtained in the first experiment, when x is around 0.2, the resistivity ρ ｘｘ The prediction that it is possible to increase the spin current conversion efficiency while suppressing the increase in 0.8 Fe 0.2 and W 0.8 Ni 0.2 It was possible to confirm this in the case of

[0046] In the first experiment, （１－ｘ） Fe ｘ Spin Hall conductivity σ ｚ ｘｙ In the second experiment, W 0.8 Fe 0.2 Spin Hall angle θ ＳＨ is larger than that of tungsten alone. The measurement results of the first and second experiments seem to be inconsistent, but this is thought to be mainly due to the fact that the measurement accuracy of the first experiment was lower than that of the second experiment. In particular, the W in (b) in the second experiment 0.8 Fe 0.2 The spin diffusion length λ Ｎ The measured value of is smaller than that in the cases of (a) and (c). In contrast, in the first experiment, the spin diffusion length λ Ｎ is fixed to a specific value, the spin diffusion length λ in the first experiment Ｎ The way in which the conditions were set can be explained as the cause of the difference in the measurement results between the first experiment and the second experiment.

[0047] From the continuity of the data obtained when the content x is changed in the first experiment, it can be inferred that the same results can be obtained not only when x is 0.2 but also when x is at least in the vicinity of 0.2. Furthermore, from the results of the first experiment, it can be seen that when tungsten is combined with only iron or only nickel (W （１－ｘ） Fe ｘ Or W （１－ｘ） Ni ｘ ), but also when tungsten is combined with nickel-iron alloy (W （１－ｘ） Py ｘ ) it can be inferred that similar results will be obtained.

[0048] Specifically, the W shown in FIG. （１－ｘ） Ni ｘ and W （１－ｘ） Py ｘ Spin Hall conductivity σ ｚ ｘｙ does not show a large decrease when the content x is in the range of 0.4 or less, but shows a large decrease when the content x exceeds 0.4. （１－ｘ） Fe ｘ Regarding W, it can be said that the degree of decrease is gentle when the content x is in the range of 0 to about 0.4. （１－ｘ） Fe ｘ , W （１－ｘ） Ni ｘ and W （１－ｘ） Py ｘ Resistivity ρ ｘｘ Therefore, taking into consideration the experimental error, if the content x is in the range of 0<x≦0.35, the resistivity ρ ｘｘ It can be inferred that the spin current conversion efficiency can be increased while suppressing the increase in

[0049] Considering the above experimental results, the channel layer 30 of the spintronics device 10 according to this embodiment is formed of a tungsten alloy containing at least one of iron and nickel in a range of 35 at % or less, which makes it possible to generate the spin current Is more efficiently than when tungsten alone is used as the material for the channel layer 30.

[0050] <Magnetic Memory Device 100> Next, the magnetic memory device 100 according to this embodiment will be described with reference to Figures 7 and 8. The magnetic memory device 100 corresponds to an SOT-MRAM.

[0051] 7, the magnetic memory device 100 includes a memory cell array 110, an X driver 120, a Y driver 130, and a controller 140. The X driver 120 and the Y driver 130 are connected to the memory cell array 110. The controller 140 is connected to the X driver 120 and the Y driver 130. Here, "connected" means electrically connected, that is, the two objects may be directly connected, or a conductor may be disposed between them. The same applies below.

[0052] The memory cell array 110 includes a plurality of memory cells MC. In the memory cell array 110, the plurality of memory cells MC are arranged in an m×n matrix. Each of the plurality of memory cells MC includes the spintronics element 10 shown in FIG. 1. Each of the plurality of memory cells MC is connected to first bit lines BL1_1 to BLm_1 and second bit lines BL1_2 to BLm_2. Each of the memory cells MC is also connected to word lines WL1 to WLn and ground lines GND1 to GNDn.

[0053] The X driver 120 is connected to a plurality of word lines WLj (j=1, 2, . . . , n). Under the control of the controller 140, the X driver 120 drives the word line WLj to be accessed to an active level (for example, H level).

[0054] The Y driver 130 is connected to a plurality of first bit lines BLi_1 and a plurality of second bit lines BLi_2 (i=1, 2, ..., m). The first bit line BLi_1 and the second bit line BLi_2 form a pair of bit lines. Under the control of the controller 140, the Y driver 130 sets the voltage levels of the pair of bit lines to be accessed, that is, the first bit line BLi_1 and the second bit line BLi_2, to an H level or an L level.

[0055] Each of the ground lines GNDj (j=1, 2, . . . , n) is set to the ground voltage. Note that each ground line GNDj may be set to a reference voltage other than the ground voltage.

[0056] The controller 140 controls the X driver 120 and the Y driver 130 when writing data to the magnetoresistive element 20 and when reading data from the magnetoresistive element 20 .

[0057] 8, each memory cell MC includes a spintronics element 10. The spintronics element 10 is a three-terminal device in which a first terminal T1 is connected to the reference layer 23 of the magnetoresistive element 20, a second terminal T2 is connected to one end of the channel layer 30, and a third terminal T3 is connected to the other end of the channel layer 30.

[0058] In each memory cell MC, a transistor Tr1 is connected to the second terminal T2, and a transistor Tr2 is connected to the third terminal T3. The transistors Tr1 and Tr2 are, for example, NMOS (N-channel Metal Oxide Semiconductor) transistors.

[0059] The first terminal T1 is connected to the ground line GNDj, the second terminal T2 is connected to the drain of the transistor Tr1, and the third terminal T3 is connected to the drain of the transistor Tr2. The gates of the transistors Tr1 and Tr2 are connected to the word line WLj. The source of the transistor Tr1 is connected to the first bit line BLi_1, and the source of the transistor Tr2 is connected to the second bit line BLi_2.

[0060] Next, the operation of writing data to the magnetoresistive element 20 and the operation of reading data from the magnetoresistive element 20 will be described.

[0061] The magnetoresistive element 20 is assigned one bit of data, either "0" or "1", depending on the resistance state. In the following description, it is assumed that the low resistance state and high resistance state of the magnetoresistive element 20 represent "0" and "1", respectively. Note that the assignment of data to the magnetoresistive element 20 may be reversed.

[0062] 9, when writing data to the magnetoresistive element 20, the controller 140 applies a voltage corresponding to the data to be written between the second terminal T2 and the third terminal T3, and passes a write current through the channel layer 30. Then, a spin-orbit torque acts, and the direction of the magnetization M21 of the free layer 21 changes according to the direction of the write current.

[0063] Specifically, first, a low resistance state will be described in which the magnetoresistive element 20 of the memory cell MC located at row i and column j of the n×m memory cells MC stores data "0," i.e., the magnetization M23 of the reference layer 23 and the magnetization M21 of the free layer 21 are in the same direction. When writing data "1" to the memory cell MC in the low resistance state, the controller 140 controls the X driver 120 to set the word line WLj to the H level. At the same time, the controller 140 controls the Y driver 130 to set the first bit line BLi_1 to the H level and the second bit line BLi_2 to the L level.

[0064] This turns on the transistors Tr1 and Tr2, and a write current flows in the channel layer 30 from the first bit line BLi_1 side to the second bit line BLi_2 side, i.e., from the second terminal T2 to the third terminal T3. When the write current flows, a spin current Is is generated in the perpendicular direction due to the magnetic spin Hall effect. When the spin current Is is generated, a spin-orbit torque acts on the magnetization M21 in the free layer 21 due to the generated spin current Is, reversing the magnetization M21 and writing data "1".

[0065] Second, a description will be given of a high resistance state in which the magnetoresistive element 20 of the memory cell MC located in the ith row and jth column of the n×m memory cells MC stores data "1," i.e., the case in which the magnetization M13 of the reference layer 23 and the magnetization M11 of the free layer 21 are in opposite directions. When writing data "0" to the memory cell MC in the high resistance state, the controller 140 controls the X driver 120 to set the word line WLj to the H level. At the same time, the controller 140 controls the Y driver 130 to set the first bit line BLi_1 to the L level and the second bit line BLi_2 to the H level.

[0066] This turns on the transistors Tr1 and Tr2, and a write current flows in the channel layer 30 from the second bit line BLi_2 side to the first bit line BLi_1 side, i.e., from the third terminal T3 to the second terminal T2. When the write current flows, a spin current Is is generated in the perpendicular direction due to the magnetic spin Hall effect. When the spin current Is is generated, a spin-orbit torque acts on the magnetization M21 in the free layer 21 due to the generated spin current Is, reversing the magnetization M21 and writing data "0".

[0067] Note that when the controller 140 passes a current to write data "0" through the channel layer 30 while the magnetoresistive element 20 is storing data "0," the angle between the direction of magnetization M21 and the spin polarization direction at the interface of the channel layer 30 is small. Therefore, the spin-orbit torque acting on the magnetization M21 is small, so the magnetization M21 does not reverse and data is not rewritten. The same is true when the controller 140 passes a current to write data "1" through the channel layer 30 while the magnetoresistive element 20 is storing data "1."

[0068] When reading data from the magnetoresistive element 20, the controller 140 applies a read voltage between the first terminal T1 and the second terminal T2 or the third terminal T3, as shown in FIG. 10, and outputs data corresponding to the magnitude of the read current flowing.

[0069] Specifically, when reading data stored in a memory cell MC located in the i-th row and j-th column of n×m memory cells MC, the controller 140 controls the X driver 120 to set the word line WLj to the H level. At the same time, the controller 140 controls the Y driver 130 to set one of the first bit line BLi_1 and the second bit line BLi_2 to the H level and leave the other in an open state.

[0070] This turns on the transistors Tr1 and Tr2, and a read current flows from the first bit line BLi_1 or the second bit line BLi_2 at the H level through the second terminal T2 or the third terminal T3, the channel layer 30, the free layer 21, the barrier layer 22, the reference layer 23, the first terminal T1, and the ground line GNDj in this order. By measuring the magnitude of this read current, it is possible to determine whether the resistance state of the magnetoresistive element 20 is a high resistance state or a low resistance state. This provides information on whether the data stored in the memory cell MC is "0" or "1."

[0071] As described above, in the spintronics element 10 and magnetic memory device 100 according to this embodiment, the channel layer 30 is formed of a tungsten alloy containing at least one of iron and nickel in a range of 35 at% or less. This makes it possible to efficiently generate spin current Is in the channel layer 30, which is primarily made of tungsten, which is relatively inexpensive and readily available. As a result, data can be written with a small write current, thereby improving the efficiency of data writing.

[0072] Although the embodiments of the present invention have been described above, the above embodiments are merely examples, and the scope of application of the present invention is not limited to these. In other words, the embodiments of the present invention are applicable to various applications, and all embodiments are included in the scope of the present invention.

[0073] For example, in the above embodiment, the channel layer 30 is formed of a tungsten alloy containing at least one of iron and nickel in a range of 35 at % or less. However, the range of the content x is not limited to 0<x≦0.35, and the resistivity ρｘｘ Any range may be used as long as it is possible to infer that the spin-current conversion efficiency can be increased while suppressing an increase in the tungsten content. For example, the tungsten alloy forming the channel layer 30 may contain at least one of iron and nickel in a range of 30 at% or less, or in a range of 10 at% or more and 35 at% or less.

[0074] More specifically, the upper limit of the range of the content x may be, for example, x≦0.3 or x≦0.25. It is predicted that the closer the upper limit is to 0.2, the closer the results to the results of the second experiment (x=0.2) will be obtained. Furthermore, the lower limit of the range of the content x may be, for example, 0.05≦x, 0.10≦x, or 0.15≦x. It is predicted that the closer the lower limit is to 0.2, the more significantly the effect of increasing the spin current conversion efficiency compared to tungsten alone will be obtained. The range of the content x can be set by freely combining such upper and lower limits. Furthermore, values ​​other than those exemplified may be set as the upper or lower limit.

[0075] In the first and second experiments, the nickel-iron alloy (Py) used was an alloy in which iron and nickel were combined in a composition ratio of 1:4 (20 at%:80 at%). （１－ｘ） Fe ｘ , W （１－ｘ） Ni ｘ and W （１－ｘ） Py ｘ Considering the experimental results in （１－ｘ） Py ｘ It can be inferred that the same results can be obtained even if the ratio of iron and nickel combined is changed. （１－ｘ） Py ｘ In the above, the ratio of iron and nickel to be combined may be any ratio.

[0076] In the above embodiment, the magnetization M21 in the free layer 21 and the magnetization M23 in the reference layer 23 are oriented in the Z-axis direction, i.e., perpendicular to the film surface. In other words, the above embodiment has been described using an SOT-MRAM with a structure known as Type-Z as an example. However, the structure of the SOT-MRAM is not limited to Type-Z, and may be Type-X or Type-Y. For example, in a Type-X SOT-MRAM, the orientation of the magnetizations M21 and M23 in the equilibrium state is the Y-axis direction, i.e., parallel to the current flow. In this case, the magnetic field required for SOT-induced magnetization reversal is the magnetic field Hz in the Z-axis direction, not the magnetic field Hy in the Y-axis direction. On the other hand, in a Type-Y SOT-MRAM, the orientation of the magnetizations M21 and M23 in the equilibrium state is the X-axis direction, i.e., parallel to the film surface and perpendicular to the current flow. In other words, the direction of the spin of the spin current Is injected into the free layer 21 is parallel to the direction of the magnetizations M21 and M23. In this case, no magnetic field is required for the SOT-induced magnetization reversal.

[0077] The present invention allows various embodiments and modifications without departing from the broad spirit and scope of the present invention. Furthermore, the above-described embodiments are intended to illustrate the present invention and do not limit the scope of the present invention. That is, the scope of the present invention is defined not by the embodiments but by the claims. Various modifications made within the scope of the claims and the meaning of the invention equivalent thereto are considered to be within the scope of the present invention.

[0078] This application is based on Japanese Patent Application No. 2024-87879 filed on May 30, 2024. The entire specification, claims, and drawings of Japanese Patent Application No. 2024-87879 are incorporated herein by reference.

[0079] REFERENCE SIGNS LIST 10 Spintronics element 20 Magnetoresistance element 21 Free layer 22 Barrier layer 23 Reference layer 30 Channel layer 100 Magnetic memory device 110 Memory cell array 120 X driver 130 Y driver 140 Controller MC Memory cell BLi_1 First bit line BLi_2 Second bit line GNDj Ground line MC Memory cell T1, T2, T3 Terminals Tr1, Tr2 Transistors WLj Word line

Claims

1. A spintronics element comprising a channel layer, a free layer stacked on the channel layer, a barrier layer stacked on the free layer, and a reference layer stacked on the barrier layer, wherein the direction of magnetization in the reference layer is fixed, and the direction of magnetization in the free layer changes due to a spin-orbit torque generated by a write current flowing through the channel layer, and the channel layer is formed from a tungsten alloy containing at least one of iron and nickel in a range of 35 at% or less.

2. The spintronics element according to claim 1, wherein the tungsten alloy contains the iron in a range of 35 at % or less.

3. The spintronics element according to claim 1, wherein the tungsten alloy contains nickel in a range of 35 at % or less.

4. The spintronics device according to claim 1, wherein the tungsten alloy contains a nickel-iron alloy in a range of 35 at % or less.

5. The spintronics element according to claim 1, wherein the tungsten alloy contains at least one of the iron and the nickel in an amount of 30 at % or less.

6. The spintronics element according to claim 1, wherein the tungsten alloy contains at least one of the iron and the nickel in a range of 10 at % or more and 35 at % or less.

7. A magnetic memory device in which a plurality of memory cells are arranged in a matrix, each of the plurality of memory cells comprising a spintronics element according to any one of claims 1 to 6, and connected to a bit line and a word line.

Images