JPWO2006129725A1 - MRAM - Google Patents

Abstract

The MRAM according to the present invention includes a substrate, a magnetization fixed layer having fixed magnetization, a magnetization free layer having reversible magnetization, and a nonmagnetic barrier interposed between the magnetization fixed layer and the magnetization free layer. A layer. The magnetization free layer includes a plurality of ferromagnetic layers and a nonmagnetic layer formed so as to exhibit an antiferromagnetic RKKY interaction. The nonmagnetic layer located closest to the substrate has a film thickness in a range corresponding to the first-order antiferromagnetic peak of the RKKY interaction, and the nonmagnetic layer located farthest from the substrate exhibits the RKKY interaction. The film thickness is in a range corresponding to the αN-th order antiferromagnetic peak. The α1 and the αN satisfy the following relationship: α1 <αN.

Description

  The present invention relates to an MRAM (Magnetic Random Access Memory), and more particularly to an MRAM using a magnetoresistive element using a SAF (synthetic antiferromagnet) as a magnetization free layer as a memory cell.

  An MRAM (Magnetic Random Access Memory) is a nonvolatile memory capable of high-speed writing / reading, and research and development for practical use has been actively conducted in recent years.

  Most typically, an MRAM includes a magnetoresistive element composed of a magnetization free layer in which magnetization can be reversed, a magnetization fixed layer in which magnetization is fixed, and a nonmagnetic layer interposed therebetween. Use as Data is stored as the magnetization direction of the magnetization free layer. When the nonmagnetic layer is made of an extremely thin insulator, the magnetoresistive element exhibits a TMR (tunnel magnetoregistance) effect, and the magnetoresistive element thus configured is often called an MTJ (magnetic tunnel junction) element. . On the other hand, when the nonmagnetic layer is made of a nonmagnetic conductor, the magnetoresistive element exhibits a GMR (giant magnetoregistance) effect, and the thus configured magnetoresistive element is a CPP-GMR (current perpendicular). to-plane giant magnetoresistive) element.

  Data is written by applying an external magnetic field to the magnetization free layer, thereby reversing the magnetization of the magnetization free layer in a desired direction.

  On the other hand, data reading uses the magnetoresistive effect exhibited by the magnetoresistive element. Regardless of whether the TMR effect or the GMR effect is used, the resistance of the magnetoresistive element changes according to the magnetization direction of the magnetization free layer. Using the change in resistance of the magnetoresistive element, the magnetization direction of the magnetization free layer, that is, the written data is determined.

  One problem with MRAM is the selectivity of memory cells in the write operation. In the most conventional MRAM, the reversal field of the magnetization in the direction of the easy axis becomes smaller as a strong magnetic field is applied in the direction of the hard axis of the ferromagnetic layer by utilizing the fact that the ferromagnetic material shows the asteroid characteristic. By using this, selective writing is performed. Specifically, word lines and bit lines that are orthogonal to each other are provided in the vicinity of each magnetoresistive element, and current flows through the word lines and bit lines that intersect the selected memory cell. The magnetization of the magnetization free layer of the magnetoresistive element of the memory cell in which current is passed through both the corresponding word line and bit line is selectively inverted in the desired direction, thereby completing the selective writing of data. To do. In a memory cell in which a current is supplied only to one of the corresponding word line and bit line, the magnetization of the magnetization free layer is not reversed, that is, data is not written. However, selective writing by such an operation is not stable because there may be bits that cause magnetization reversal even when a magnetic field is applied only to the easy axis or the difficult axis, and the selectivity of the memory cell is not good.

  As one method for improving the selectivity of the write operation of the memory cell, a toggle write method is known (see US Pat. No. 6,545,906). The toggle write method is a technique for performing a write operation with high selectivity by using SAF for the magnetization free layer; where SAF is composed of a plurality of ferromagnetic layers, and adjacent ferromagnetic layers It is a magnetically antiferromagnetically coupled structure.

FIG. 1 is a plan view showing a typical configuration of an MRAM 101 that employs a toggle writing method. In the memory array of the MRAM 101, a word line 103 and a bit line 102 orthogonal to the word line 103 are extended. An MTJ element 101 used as a memory cell is provided at each position where the word line 103 and the bit line 102 intersect. As shown in FIG. 2, the MTJ element 101 typically includes a lower electrode layer 111, an antiferromagnetic layer 112, a magnetization fixed layer 113, a barrier layer 114, a magnetization free layer provided above the substrate 100. A layer 115 and an upper electrode layer 116 are provided. As shown in FIG. 1, the MTJ element 101 is configured such that the easy axes of the magnetization fixed layer 113 and the magnetization free layer 115 form an angle of 45 ° with the word line 103 and the bit line 102, that is, the MTJ element 101. Are arranged such that the longitudinal direction forms an angle of 45 ° with the word line 103 and the bit line 102.

Referring to FIG. 2 again, the magnetization free layer 115 is composed of SAF. More specifically, the magnetization free layer 115 includes ferromagnetic layers 121 and 122 and a nonmagnetic layer 123 interposed therebetween. The nonmagnetic layer 123 is configured to cause an antiferromagnetic RKKY (Rudermann, Kittel, Kasuya, Yoshida) interaction between the ferromagnetic layers 121 and 122. The overall residual magnetization of the magnetization free layer 115 (that is, the magnetization of the magnetization free layer 115 as a whole when the external magnetic field is 0) is as close to 0 as possible. This condition can be satisfied, for example, by forming the two ferromagnetic layers 121 and 122 with the same material and the same film thickness.

In the toggle writing method, selective data writing is performed using the property that SAF develops a spin flop. FIG. 3 shows a magnetization curve of SAF that develops a spin flop. Even if an external magnetic field is applied in the easy axis direction of the SAF, if the external magnetic field is small, the magnetization of the SAF remains zero. When the external magnetic field is increased and reaches the magnetic field H _flop , magnetization suddenly appears in the SAF. At this time, the magnetizations of the two ferromagnetic layers are arranged so that they are magnetically coupled so that they form an angle smaller than 180 °, and the resultant magnetization is in the direction of the external magnetic field. This phenomenon is called a spin flop, and the magnetic field H _flop that the spin flop develops is called a flop magnetic field. Note that the spin flop appears only if the overall remanent magnetization of the SAF is sufficiently small. When a magnetic field is further applied, the magnetizations of the two ferromagnetic layers will eventually be arranged completely in parallel. This magnetic field is called a saturation magnetic field H _s . The flop magnetic field H _flop and the saturation magnetic field H _s are expressed by equations (1b) and (2b), respectively.
H _flop = 2 / M · K [(J _saf / t−K)] ^0.5 (1a)
= (H _s · H _k ) ^0.5 , ... (1b)
H _s = 2J _saf / (M · t) −2K / M, (2a)
= 2J _saf / (M · t) −H _k , (2b)
As is clear from the equation (1), the flop magnetic field H _flop is uniquely determined by the saturation magnetic field H _s and the anisotropic magnetic field H _k .

FIG. 4 is a conceptual diagram for explaining the procedure of the toggle writing method. FIG. 5 is a graph showing waveforms of currents flowing through the word line 103 and the bit line 102 when data writing is performed by toggle writing. It should be noted in FIG. 3 that the magnetizations of the ferromagnetic layers 121 and 122 of the magnetization free layer 115 are referred to by symbols M ₁ and M ₂ , respectively.

In the data writing by the toggle writing method, the direction of the magnetic field applied to the magnetization free layer 115 is rotated in the plane to direct the magnetization of the ferromagnetic layers 121 and 122 constituting the magnetization free layer 115 in a desired direction. Is done by. Specifically, first, a write current is supplied to the word line 103, thereby generating a magnetic field _{HWL in} a direction perpendicular to the word line 103 (time t ₁ ). Subsequently, the write current is supplied to the bit line 102 while the write current is supplied to the word line 103 (time t ₂ ). As a result, a magnetic field H _WL + H _BL is generated in a direction that forms an angle of 45 ° with both the word line 103 and the bit line 102. Subsequently, the supply of the write current to the word line is stopped while the write current is supplied to the bit line (time t ₃ ). As a result, a magnetic field _HBL is generated in a direction perpendicular to the bit line 102 (that is, a direction parallel to the word line 103). By applying a write current to the word line 103 and the bit line 102 in such a procedure, the magnetic field applied to the magnetization free layer 115 is rotated, whereby the ferromagnetic layers 121 and 122 constituting the magnetization free layer 115 are rotated. Can be rotated 180 °.

  The advantage of toggle writing is that the selectivity of the memory cell is high in principle. As shown in FIG. 6 which is a graph showing a magnetization reversible region with respect to the write current magnetic field of the bit line and the word line, in the toggle write, in principle, only the word line 103 or the bit line 102 is written. Even when a current is applied, the magnetization of the SAF does not reverse. In other words, the magnetization of the half-selected memory cell does not reverse undesirably in principle. This effectively improves the reliability of the operation of the MRAM.

In the toggle write method, the magnetic field applied to the magnetization free layer 115 when a write current is passed through the word line 103 and the bit line 102 must be larger than the above-described flop magnetic field H _{flop and} smaller than the saturation magnetic field H _s. I must. Otherwise, the magnetic field of the magnetization free layer 115 cannot be reversed in a desired direction.

Therefore, the write margin of the toggle writing method is larger as the flop magnetic field H _flop is smaller and as the saturation magnetic field H _s is larger. In other words, the larger the ratio H _s / H _flop of the saturation magnetic field H _s to the flop magnetic field H _{flop, the} larger the write margin. An increase in the write margin is important for improving the reliability of the operation of the MRAM.

The spin flop indicated by SAF is also effective for writing methods other than toggle writing. For example, when an external magnetic field is applied in the easy axis direction, SAF first causes magnetization reversal in a specific direction while maintaining the mutual magnetic field. This is called direct inversion. When the magnetic field is further increased, SAF develops a spin _flop in the flop magnetic field H _flop . This direct inversion is associated with the spin flop and can be used for data writing. The magnetic field H _{dir at} which direct inversion occurs is substantially proportional to the flop magnetic field H _flop . Therefore, reducing the flop magnetic field H _flop is important because it leads to a reduction in the magnetic field H _dir .

Increasing the remanent magnetization can also reduce the magnetic field H _dir , but this is not preferred. When the residual magnetization increases, the SAF will cause direct reversal even if a magnetic field is applied in a direction other than the easy axis, and eventually the SAF will not cause a spin flop. The write characteristic to SAF which does not express a spin flop is exactly the same as the write characteristic of data write using the asteroid characteristic described above. This means that write selectivity is lost. Further, the increase in remanent magnetization is a problem because the shape magnetic anisotropy is greatly increased and the switching magnetic field is increased.

Also, according to equation (1b), it may seem that reducing the saturation magnetic field H _s is effective for reducing the flop magnetic field H _flop . However, when the saturation magnetic field H _s decreases, it becomes difficult to cause a spin flop. If the saturation field H _s is reduced to a magnitude close to the anisotropy field H _k , the SAF will no longer cause a spin flop. This is a problem because the operation rate of the MRAM is reduced by writing using a spin flop.

  In principle, the toggle writing method can also be applied to the case where the number of ferromagnetic layers included in the SAF is 3 or more. In the above-mentioned US Pat. No. 6,545,906, the strong force included in the SAF is described. It is disclosed that the number of magnetic layers may be three or more. In addition, US Pat. No. 6,714,446 discloses a configuration in which each of the two ferromagnetic layers included in the SAF is formed of two ferromagnetic films separated by a nonmagnetic film. ing.

  Japanese Patent Laid-Open No. 2005-86015 discloses that the number of ferromagnetic layers is 4, and the antiferromagnetic interaction between the first and second ferromagnetic layers is strong. Is approximately equal to the strength of the antiferromagnetic interaction between the third ferromagnetic layer and the fourth ferromagnetic layer, and the first ferromagnetic layer and the fourth ferromagnetic layer. In which the second and third ferromagnetic layers are substantially equal in magnetization amount.

  However, according to the experiments by the inventors, MRAMs using a SAF including three or more ferromagnetic layers as a magnetization free layer often do not actually operate in the toggle writing system. Further, in the case where the number of ferromagnetic layers is four, in order to make the magnitude of the antiferromagnetic interaction equal, a nonmagnetic layer between the first ferromagnetic layer and the second ferromagnetic layer, Even when the material and film thickness of the nonmagnetic layer between the ferromagnetic layer and the fourth ferromagnetic layer were set to be equal, the operation with the toggle writing method was rare. This was thought to be because SAF including three or more ferromagnetic layers did not develop a spin flop for some reason. If a SAF including three or more ferromagnetic layers can be operated by the toggle writing method, an MRAM with a large writing margin should be realized.

  The fact that SAF does not indicate a spin flop is fatal not only to the toggle writing method but also to all MRAMs that employ a writing method using a spin flop (for example, an MRAM that performs data writing using direct inversion). The absence of spin flop means that various advantages of SAF (such as improved thermal disturbance resistance) cannot be used. Therefore, a technique for causing the SAF to develop a spin flop is important in terms of improving the performance of the MRAM device using writing using the spin flop.

Accordingly, an object of the present invention is to provide a technique for causing a spin flop to appear in a SAF including three or more ferromagnetic layers.
Another object of the present invention is to enable an MRAM using the SAF as a magnetization free layer to operate in a toggle writing system by developing a spin flop in an SAF including three or more ferromagnetic layers.
Still another object of the present invention is to provide a technique for increasing the write margin of an MRAM operating in the toggle write mode, in other words, increasing the ratio H _s / H _flop of the saturation magnetic field H _s to the flop magnetic field H _flop . It is in.

  The inventor of the present invention has the reason that the spin flop does not appear in the SAF including three or more ferromagnetic layers because of the RKKY interaction that the lowermost nonmagnetic layer and the uppermost nonmagnetic layer develop. It was found that the strength was different due to the difference in crystallinity. The uppermost nonmagnetic layer (the last nonmagnetic layer) is positioned at the bottom because crystallization is promoted by the ferromagnetic layer and the nonmagnetic layer formed therebelow. It has better crystallinity (particularly crystal orientation) than the nonmagnetic layer (that is, the first nonmagnetic layer). For this reason, when the uppermost nonmagnetic layer and the lowermost nonmagnetic layer are formed with the same material and film thickness, the strength of the RKKY interaction differs. When the strength of the RKKY interaction is different, the magnetization free layer does not develop a spin flop.

  In order to reduce the difference in the strength of the RKKY interaction due to the difference in crystallinity, in the MRAM of the present invention, the uppermost nonmagnetic layer has a higher order than the lowermost nonmagnetic layer. The film thickness is in a range corresponding to the antiferromagnetic peak. Thereby, the difference in the strength of the RKKY interaction due to the difference in crystallinity is canceled, and it becomes possible to cause the spin flop to appear in the SAF including three or more ferromagnetic layers.

In one aspect, the MRAM according to the present invention includes a substrate, a magnetization fixed layer having a fixed magnetization, a magnetization free layer having a reversible magnetization, and the magnetization fixed layer and the magnetization free layer. And a nonmagnetic barrier layer provided. The magnetization free layer includes first to (N + 1) th ferromagnetic layers (N is an integer of 2 or more) and first to Nth nonmagnetic layers formed to exhibit an antiferromagnetic RKKY interaction. Including. The kth nonmagnetic layer (k is an arbitrary integer between 1 and N) is provided between the kth ferromagnetic layer and the (k + 1) th ferromagnetic layer. The first nonmagnetic layer is located closest to the substrate among the first to Nth nonmagnetic layers, and the Nth nonmagnetic layer is among the first to Nth nonmagnetic layers. Located farthest from the substrate. The first nonmagnetic layer has a thickness corresponding to the α _first- order antiferromagnetic peak of the RKKY interaction, and the Nth nonmagnetic layer has the α _Nth order antiferromagnetic peak of the RKKY interaction. The film thickness is in a range corresponding to the ferromagnetic peak. The α ₁ and the α _N have the following relationship:
α ₁ <α _N ,
Is satisfied. Here, “the k-th nonmagnetic layer has a film thickness in a range corresponding to the k-th antiferromagnetic peak of the RKKY interaction” means that the film thickness t _k of the k-th nonmagnetic layer is in the following range:
t _{k_min} <t _k <t _{k_max}
Refers to that it is _in; t k_min is smaller than the alpha _k Next antiferromagnetic peak, and, among the thickness, such as the strength of the RKKY interaction is 0, the next first alpha _k The thickness closest to the antiferromagnetic peak, t _{k_max} is larger than the α _{k -th} order antiferromagnetic peak, and the α th of the thicknesses where the strength of the RKKY interaction is zero _The film thickness is closest to the _kth- order antiferromagnetic peak.

  In such an MRAM, the difference in the strength of the RKKY interaction due to the difference in crystallinity can be canceled by the difference in the strength of the RKKY interaction due to the film thickness of the nonmagnetic layer. This makes it possible to develop a spin flop in a SAF including three or more ferromagnetic layers.

The α ₁ and the α _N have the following relationship:
α _N = α ₁ +1,
It is preferable to satisfy

  The first nonmagnetic layer is a ruthenium layer having a thickness of 1.8 nm to 2.5 nm, and the Nth nonmagnetic layer is a ruthenium layer having a thickness of 3.2 nm to 3.8 nm. It is preferred that

To increase the saturation magnetic field H _s of the magnetization free layer is expressed, at least one non-magnetic layer of said second to (N-1) non-magnetic layer is lower than the first N nonmagnetic layer It is preferable to have a film thickness in a range corresponding to the next antiferromagnetic peak.

  The at least one nonmagnetic layer is preferably located at the center of the magnetization free layer. Specifically, when the number N of nonmagnetic layers is an odd number, the at least one nonmagnetic layer is the ([N + 1] / 2) nonmagnetic of the first to Nth nonmagnetic layers. A layer is preferred. On the other hand, when N is an even number, the at least one nonmagnetic layer includes the (N / 2) -th nonmagnetic layer of the first to N-th nonmagnetic layers ([N / 2] +1) A nonmagnetic layer is preferred.

In this case, it is preferable that the strength of the RKKY interaction expressed by the first to Nth nonmagnetic layers (31 to 35) is stronger toward the center of the magnetization free layer (15). Specifically, the second to (N-1) nonmagnetic layers have film thicknesses in a range corresponding to the α _{2 to} α _N-1 order antiferromagnetic peaks of the RKKY interaction, respectively. In the case where N is an odd number, the α _{1 to} α _N satisfy the following conditions:
α ₁ ≧ α ₂ ≧ α ₃ ≧... ≧ α _{[N + 1] / 2-1} ≧ α _{[N + 1] / 2} ,
α _{[N + 1] / 2} ≦ α _{[N + 1] / 2 + 1} ≦ ・ ・ ・ ≦ α _N−1 ≦ α _N ,
α ₁ <α _N ,
α ₂ = α _N−1 ,
α ₃ = α _N-2 ,
...
α _p = α _{N−p + 1} , (p is an integer of 2 or more and (N + 1) / 2-1 or less)
...
α _{[N + 1] / 2-1} = α _{[N + 1] / 2 + 1} ,
Is preferably satisfied.

On the other hand, when N is an even number, the α _{1 to} α _N satisfy the following conditions:
α ₁ ≧ α ₂ ≧ α ₃ ≧ ・ ・ ・ ≧ α _{N / 2-1} ≧ α _{N / 2} ,
α _{N / 2 + 1} ≦ α _{N / 2 + 2} ≦ ・ ・ ・ ≦ α _N−1 ≦ α _N ,
α ₁ <α _N ,
α ₂ = α _N−1 ,
...
α _p = α _{N−p + 1} , (p is an integer of 2 or more and (N / 2) or less)
...
α _{N / 2-1} = α _{N / 2 + 2} ,
α _{N / 2} = α _{N / 2 + 1} ,
Is preferably satisfied.

  Most preferably, the first nonmagnetic layer is composed of a ruthenium layer having a thickness of 1.8 nm to 2.5 nm, and each of the second to (N-1) th nonmagnetic layers has a thickness of 0. The Nth nonmagnetic layer is formed of either a ruthenium layer having a thickness of 7 nm to 1.2 nm or a ruthenium layer having a thickness of 1.8 nm to 2.5 nm. It is composed of a ruthenium layer having a thickness of 9 nm.

  Such a configuration of the magnetization free layer is particularly effective when the barrier layer is an amorphous layer and the magnetization free layer is formed on the upper surface of the barrier layer.

  In the broadest sense, the magnetization free layer of the present invention has the first nonmagnetic layer and the first nonmagnetic layer so that the strength of the RKKY interaction via the Nth nonmagnetic layer is substantially equal. The nonmagnetic layer and the Nth nonmagnetic layer have a structure that is not the same. Specifically, the first nonmagnetic layer and the Nth nonmagnetic layer are designed to have different film thicknesses. The first nonmagnetic layer and the Nth nonmagnetic layer have different crystal orientations. For example, when the first nonmagnetic layer to the Nth nonmagnetic layer are formed of ruthenium, the film surface of the ruthenium HCP (001) surface of the Nth nonmagnetic layer as compared with the first nonmagnetic layer. High crystal orientation in the straight direction. At that time, a difference is often seen also in the crystal orientation of the first ferromagnetic layer and the Nth ferromagnetic layer, which are the foundations of the first nonmagnetic layer and the Nth nonmagnetic layer. For example, when the first to Nth ferromagnetic layers are composed of permalloy, the Nth ferromagnetic layer is more perpendicular to the FCC (111) plane than the first ferromagnetic layer. High crystal orientation.

  In order to offset the difference in RKKY interaction due to the difference in crystallinity, the composition of the ferromagnetic layer in contact with the first nonmagnetic layer and the ferromagnetic layer in contact with the Nth nonmagnetic layer are not the same. A configuration is also effective (see FIGS. 7D and 7E). For example, the composition of the first surface of the film in contact with the first nonmagnetic layer on the first surface of the first nonmagnetic layer is in contact with the Nth nonmagnetic layer on the second surface of the Nth nonmagnetic layer. It is also possible to offset the difference in RKKY interaction by differing from the composition on the second side of the film. In this case, at least one of the first ferromagnetic layer, the second ferromagnetic layer, the Nth ferromagnetic layer, and the (N + 1) th ferromagnetic layer is a stacked layer in which a plurality of films having different compositions are stacked. By changing the ratio of the effective film thicknesses of the plurality of films constituting the laminated film, the interaction via the first nonmagnetic layer and the Nth nonmagnetic layer are used. It is preferable that the effective strength of the interaction is set to be approximately equal. The laminated film is preferably composed of a laminated film of a NiFe film and a CoFe film.

  When the strength of the RKKY interaction is controlled by the composition of the film in contact with the first nonmagnetic layer and the Nth nonmagnetic layer, the first nonmagnetic layer and the Nth nonmagnetic layer may have the same thickness. Is possible. In this case, it is preferable that the Co composition on the first surface of the first nonmagnetic layer is higher than the Co composition on the second surface of the Nth nonmagnetic layer.

  Further, for the first time by the inventors' experiment, even in the case where N is an even number of 2 or more in the magnetization free layer composed of the first to (N + 1) th ferromagnetic layers and the first to Nth nonmagnetic layers. By making the effective strength of the interaction through the first nonmagnetic layer and the interaction through the Nth nonmagnetic layer substantially equal, a favorable toggle operation and an increase in write margin are possible. It was shown that. In addition, it is preferable that a magnetization film thickness product of the (N / 2 + 1) th ferromagnetic layer among the first to (N + 1) th ferromagnetic layers is larger than a magnetization film thickness product of other ferromagnetic layers. Furthermore, it is preferable that the magnetization film thickness products of the first ferromagnetic layer and the (N + 1) th ferromagnetic layer are substantially equal. In particular, an element in which N is 2 (that is, an element including three or more ferromagnetic layers) has been demonstrated to perform well.

FIG. 1 is a plan view showing a typical configuration of an MRAM corresponding to the toggle writing method. FIG. 2 is a cross-sectional view showing a typical configuration of an MTJ element incorporated in an MRAM that supports the toggle writing method. FIG. 3 is a graph showing a typical magnetization curve of SAF expressing a spin flop. FIG. 4 is a conceptual diagram showing a data writing procedure by the toggle writing method. FIG. 5 is a graph showing a waveform of a write current that flows in the bit line and the word line when data writing is performed by the toggle writing method. FIG. 6 is a graph showing an operation area of the MRAM in which the toggle writing method is adopted. FIG. 7A is a cross-sectional view showing the configuration of the MTJ element of the MRAM according to the first embodiment of the present invention. FIG. 7B is a cross-sectional view showing another configuration of the MTJ element of the MRAM according to the first exemplary embodiment of the present invention. FIG. 7C is a cross-sectional view showing still another configuration of the MTJ element of the MRAM according to the first exemplary embodiment of the present invention. FIG. 7D is a cross-sectional view showing still another configuration of the MTJ element of the MRAM according to the first exemplary embodiment of the present invention. FIG. 7E is a cross-sectional view showing still another configuration of the MTJ element of the MRAM according to the first exemplary embodiment of the present invention. FIG. 7F is a cross-sectional view showing still another configuration of the MTJ element of the MRAM according to the first exemplary embodiment of the present invention. FIG. 7G is a cross-sectional view showing still another configuration of the MTJ element of the MRAM according to the first exemplary embodiment of the present invention. FIG. 8 is a graph showing the dependence of the binding energy of the RKKY interaction on the nonmagnetic layer thickness. FIG. 9A is a cross-sectional view showing the configuration of the MTJ element of the MRAM according to the second exemplary embodiment of the present invention. FIG. 9B is a cross-sectional view showing another configuration of the MTJ element of the MRAM according to the second exemplary embodiment of the present invention. FIG. 9C is a cross-sectional view showing still another configuration of the MTJ element of the MRAM according to the second exemplary embodiment of the present invention. FIG. 9D is a cross-sectional view showing still another configuration of the MTJ element of the MRAM according to the second exemplary embodiment of the present invention. FIG. 10 is a table showing the configurations of the magnetization free layers of the samples of Comparative Examples 1 to 3 and Examples 1 to 3. FIG. 11 is a frequency distribution table showing the saturation magnetic field H _s and the flop magnetic field H _flop of the sample of Comparative Example 1. 12A is a frequency distribution table showing the saturation magnetic field H _s and the flop magnetic field H _flop of the sample of Example 1. FIG. FIG. 12B is a frequency distribution table showing the saturation magnetic field H _s and the flop magnetic field H _flop of the sample of Example 2. FIG. 12C is a frequency distribution table showing the saturation magnetic field H _s and the flop magnetic field H _flop of the sample of Example 3. FIG. 13 is a table showing the configuration of the SAF for evaluating the strength of antiferromagnetic coupling and the characteristics of the SAF. FIG. 14 is a table showing variations in the saturation magnetic field H _s , the flop magnetic field H _flop , and the flop magnetic field H _flop of the samples of Examples 1 and 4 to 6. FIG. 15 is a table showing the configuration of SAF for evaluating the strength of antiferromagnetic coupling and the characteristics of the SAF.

(First embodiment)
FIG. 7A is a cross-sectional view showing a configuration of the MTJ element 1 employed in the memory cell of the MRAM according to the first embodiment of the present invention. The MTJ element 1 includes a lower electrode layer 11, an antiferromagnetic layer 12, a magnetization fixed layer 13, a barrier layer 14, a magnetization free layer 15, a cap layer 16, and an upper electrode layer 17.

  The MTJ element 1 is arranged so as to correspond to the toggle writing method. Specifically, similar to the MTJ element 101 of the conventional MRAM shown in FIG. 1, the MTJ element 1 has an angle of 45 ° with respect to the word line (and the bit line perpendicular thereto). It is arranged to make. Thereby, the easy axes of the ferromagnetic layers constituting the magnetization fixed layer 13 and the magnetization free layer 15 are oriented in a direction that forms an angle of 45 ° with respect to the word line (and the bit line perpendicular thereto). Hereinafter, the configuration of the MTJ element 1 will be described in detail.

  The lower electrode layer 11 is formed on a substrate 10 on which MOS transistors (not shown) are integrated, and functions as a path for providing an electrical connection to the magnetization fixed layer 13. The lower electrode layer 11 is made of, for example, Ta, TaN, Cu, or Al.

  The antiferromagnetic layer 12 is formed of an antiferromagnetic material such as PtMn, IrMn, or NiMn, for example, and has a role of fixing the magnetization of the magnetization fixed layer 13.

  The magnetization fixed layer 13 is formed of a magnetically hard ferromagnetic material such as CoFe. The magnetization of the magnetization fixed layer 13 is fixed by the exchange interaction in which the antiferromagnetic layer 12 acts. The magnetization fixed layer 13 may be configured by the above-described SAF. For example, the magnetization fixed layer 13 can be composed of two CoFe films and a Ru film inserted between them. In this case, the Ru film is formed so as to have a film thickness that exhibits an antiferromagnetic RKKY interaction.

The barrier layer 14 is often an amorphous insulator film that is thin enough to allow a tunnel current to flow. The fact that the barrier layer 14 is amorphous greatly affects the crystallinity of the film constituting the magnetization free layer 15 as will be described later. More specifically, the barrier layer 14 includes, for example, alumina (AlO _x ), magnesium oxide (MgO), zirconium oxide (ZrO ₂ ), hafnium oxide (HfO ₂ ), silicon oxide (SiO ₂ ), aluminum nitride (AlN). ). However, the barrier layer 14 should not be limited to be amorphous. The barrier layer 14 may be, for example, single crystal MgO having a NaCl structure.

  The magnetization free layer 15 is composed of SAF having four ferromagnetic layers. More specifically, the magnetization free layer 15 includes ferromagnetic layers 21 to 24 and nonmagnetic layers 31 to 33 interposed therebetween. More specifically, the ferromagnetic layer 21 is formed on the barrier layer 14, and the nonmagnetic layer 31 is formed on the ferromagnetic layer 21. On the nonmagnetic layer 31, the ferromagnetic layer 22, the nonmagnetic layer 32, the ferromagnetic layer 23, the nonmagnetic layer 33, and the ferromagnetic layer 24 are sequentially formed in this order. The overall residual magnetization of the magnetization free layer 15 (that is, the magnetization of the magnetization free layer 15 as a whole when the external magnetic field is 0) is as close to 0 as possible. This condition can be satisfied, for example, by forming the ferromagnetic layers 21 to 24 with the same material and the same film thickness.

  In addition, the ferromagnetic layer as used in this specification means the layer which expresses ferromagnetism as a whole, and it should not be limited and interpreted that it is comprised with the single ferromagnetic film | membrane. For example, the ferromagnetic layer referred to in the present specification includes a laminate composed of two ferromagnetic films and a nonmagnetic film interposed between the two ferromagnetic films. It must be interpreted.

Each of the nonmagnetic layers 31 to 33 of the magnetization free layer 15 is configured to cause antiferromagnetic RKKY interaction to act on the ferromagnetic layers bonded to the upper and lower sides thereof and to couple them antiferromagnetically. ing. It is well known to those skilled in the art that by appropriately selecting the material and film thickness of the nonmagnetic layer, the ferromagnetic layers bonded to each other can be antiferromagnetically coupled by the RKKY interaction. FIG. 8 is a graph showing the dependence of the strength of the binding energy due to the RKKY interaction on the film thickness of the nonmagnetic layer. Note that the graph of FIG. 8 defines that the binding energy is positive when the ferromagnetic layers are antiferromagnetically coupled. The binding energy _Jsaf due to the RKKY interaction oscillates with an increase in the thickness of the nonmagnetic layer, exhibits an antiferromagnetic RKKY interaction in a certain range, and exhibits a strong ferromagnetic strength in the other range. Magnetic RKKY interaction is expressed. The film thicknesses of the nonmagnetic layers 31 to 33 are selected so that they exhibit an antiferromagnetic RKKY interaction.

The cap layer 16 is a layer for protecting the magnetization fixed layer 13, the barrier layer 14, and the magnetization free layer 15. The cap layer 16 is made of Ta or Ru, for example. The cap layer 16 can also be formed of AlO _x that is extremely thin to the extent that a tunnel current flows.

  The upper electrode layer 17 functions as a path that provides an electrical connection to the magnetization free layer 15. The upper electrode layer 17 is made of Ta, TaN, Cu, or Al, for example.

  The main feature of the MRAM of this embodiment is that the nonmagnetic layer 31 located closest to the substrate 10 among the nonmagnetic layers 31 to 31 has a film thickness in a range corresponding to a relatively low-order antiferromagnetic peak. However, the nonmagnetic layer 33 located farthest from the substrate 10 has a film thickness in a range corresponding to a relatively high-order antiferromagnetic peak. Such a configuration is effective to make the strength of the antiferromagnetic RKKY interaction expressed by the nonmagnetic layer 31 and the nonmagnetic layer 33 as close as possible. As described above, the strength of the antiferromagnetic RKKY interaction expressed by the nonmagnetic layer 31 and the nonmagnetic layer 33 is effective to increase the write margin of the MRAM.

  As shown in FIG. 8, since the strength of the RKKY interaction depends on the material and film thickness of the nonmagnetic layer, the antiferromagnetic RKKY interaction developed by the nonmagnetic layer 31 and the nonmagnetic layer 33. In order to have the same strength, it may be considered that the nonmagnetic layer 31 and the nonmagnetic layer 33 may be formed with the same material and the same film thickness. However, in an actually integrated MRAM, even if the nonmagnetic layer 31 and the nonmagnetic layer 33 are formed with the same material and the same film thickness, the RKKY interaction appears in the nonmagnetic layer 31 and the nonmagnetic layer 33. The strengths of are not the same. This is because the nonmagnetic layer 31 and the nonmagnetic layer 33 have different crystallinity. The magnetization free layer 15 includes a ferromagnetic layer 21, a nonmagnetic layer 31, a ferromagnetic layer 22, a nonmagnetic layer 32, a ferromagnetic layer 23, a nonmagnetic layer 33, and a ferromagnetic layer 24 sequentially on the barrier layer 14. Therefore, the nonmagnetic layer 33 formed later has better crystallinity than the nonmagnetic layer 31 formed first. Since the strength of the RKKY interaction is stronger as the crystallinity is better, when the nonmagnetic layer 31 and the nonmagnetic layer 33 are formed with the same material and the same film thickness, the nonmagnetic layer 33 has a higher thickness than the nonmagnetic layer 31. Will develop a stronger RKKY interaction.

  In order to make the strength of the antiferromagnetic RKKY interaction expressed by the nonmagnetic layer 31 and the nonmagnetic layer 33 as close as possible, rather, the nonmagnetic layer 31 is in a range corresponding to a relatively low order peak. It is preferable to form the nonmagnetic layer 33 so as to have a film thickness in a range corresponding to a relatively high-order peak. Thereby, the effect that the strength of the RKKY interaction expressed by the nonmagnetic layer 33 is strengthened by good crystallinity, and the effect that it is weakened by having a film thickness corresponding to a relatively high-order peak Is canceled, and the strength of the antiferromagnetic RKKY interaction expressed by the nonmagnetic layer 31 and the nonmagnetic layer 33 becomes almost the same.

  Specifically, in this embodiment, the nonmagnetic layer 31 has a film thickness in a range corresponding to the second-order antiferromagnetic peak (antiferromagnetic 2nd peak) of the RKKY interaction, and the nonmagnetic layer 32 and 33 have a film thickness in a range corresponding to the third antiferromagnetic peak (antiferromagnetic 3rd peak). More specifically, when the nonmagnetic layers 31 to 33 are made of ruthenium, the nonmagnetic layer 31 is formed so that the film thickness is more than 1.8 nm and less than 2.5 nm. The magnetic layers 32 and 33 are formed so that the film thickness is more than 3.1 nm and less than 3.9 nm. Most preferably, the nonmagnetic layer 31 is formed to have a thickness of 2.1 nm corresponding to the antiferromagnetic 2nd peak, and the nonmagnetic layers 32 and 33 correspond to the antiferromagnetic 3rd peak. It is formed to have a film thickness of 5 nm. Such a combination of film thicknesses is effective for making the strength of the antiferromagnetic RKKY interaction expressed by the nonmagnetic layer 31 and the nonmagnetic layer 33 closer to the same.

  In the stacked magnetization free layer shown in FIG. 7A, not only the nonmagnetic layers 31 to 33 but also the ferromagnetic layers 21 to 24, the crystallinity is improved as the upper layer is increased. For example, when permalloy is used for the ferromagnetic layers 21 to 24 and ruthenium is used for the nonmagnetic layers 31 to 33, the permalloy of the ferromagnetic layer 23 is FCC (face center cubic) (compared to the ferromagnetic layer 21). The ruthenium of the nonmagnetic layer 31 and the nonmagnetic layer 33 that have a high degree of crystal orientation in the direction perpendicular to the film surface of the (111) plane and that respectively grow directly above them is also higher than that of the nonmagnetic layer 31. Ruthenium has a higher degree of crystal orientation in the direction perpendicular to the HCP (hexagonal close packed) (001) plane.

Such a technique is also applicable when the number of ferromagnetic layers included in the SAF constituting the magnetization free layer is different from that in FIG. 7A. For example, as shown in FIG. 7B, the magnetization free layer 15A of the MTJ element 1A is formed of three ferromagnetic layers 21 to 23 and nonmagnetic layers 31 and 32 inserted therebetween. It is also possible. In this case, the nonmagnetic layer 32 is formed to have a film thickness in a range corresponding to a higher-order peak than the nonmagnetic layer 31. For example, the nonmagnetic layer 31 is formed to have a film thickness in a range corresponding to the secondary antiferromagnetic peak, and the nonmagnetic layer 32 has a film thickness in a range corresponding to the third antiferromagnetic peak. Formed to have. The ferromagnetic layers 21 to 23 are formed so that the residual magnetization of the entire magnetization free layer 15A is close to zero. More specifically, the magnetization film thickness product of the ferromagnetic layers 21 and 23 (that is, the product of the film thickness of the ferromagnetic layers 21 and 23 and the film thickness) is made the same, and the magnetization film thickness product of the ferromagnetic layer 22 is It is larger than the magnetization film thickness product of the ferromagnetic layers 21 and 23, most preferably doubled. Such conditions include, for example, that the ferromagnetic layers 21 to 23 are formed of the same material, the ferromagnetic layers 21 and 23 are formed to have the same film thickness, and the ferromagnetic layer 22 is formed of the ferromagnetic layer 21, This can be achieved by forming the film so as to have a film thickness twice as large as 23.

  The number of ferromagnetic layers included in the magnetization free layer may be an odd number other than three. In this case, the magnetization film thickness product of the ferromagnetic layer located at the center (that is, the number of nonmagnetic layers is N and the number of ferromagnetic layers is N + 1, the (N / 2 + 1) th ferromagnetic layer). It is preferable that the magnetization film thickness product of the other ferromagnetic layers be larger than that of the two ferromagnetic layers positioned at both ends. This is because, in such a configuration, a plurality of ferromagnetic layers arranged in the vertical direction with respect to the magnetization of the centermost ferromagnetic layer of the laminated magnetization free layer take a certain magnetic field while taking a symmetrical magnetization arrangement. At the same time, it is easy to cause spin flop. This is important for realizing a good toggle operation.

  As shown in FIG. 7C, the magnetization free layer 15B of the MTJ element 1B may be formed of six ferromagnetic layers 21 to 26 and nonmagnetic layers 31 to 35 inserted therebetween. Is possible. In this case, the nonmagnetic layer 35 is formed to have a film thickness in a range corresponding to a higher-order peak than the nonmagnetic layer 31. For example, the nonmagnetic layer 31 is formed to have a film thickness in a range corresponding to the secondary antiferromagnetic peak, and the nonmagnetic layers 32 to 35 are films in a range corresponding to the third antiferromagnetic peak. It is formed to have a thickness. The ferromagnetic layers 21 to 26 are formed so that the residual magnetization of the entire magnetization free layer 15A is close to zero. Such a condition can be achieved, for example, by forming the ferromagnetic layers 21 to 26 with the same material and the same film thickness.

  In order to make the strength of the RKKY interaction more uniform, in addition to the film thickness of the nonmagnetic layer, the strength of the interaction may be controlled by the composition of the ferromagnetic layer in contact therewith. Control of the strength of the RKKY interaction by the film thickness of the nonmagnetic layer has a small degree of freedom in the film thickness of the nonmagnetic layer, and thus the strength of the RKKY interaction may not be completely uniform. In such a case, it is effective to control the strength of interaction by the composition of the ferromagnetic layer. Since the strength of the RKKY interaction depends on the composition of the ferromagnetic layer on the surface in contact with the nonmagnetic layer, each nonmagnetic layer is manifested by appropriately selecting the composition of the ferromagnetic layer on the surface in contact with the nonmagnetic layer. The strength of the RKKY interaction can be made closer to the same. For example, since cobalt exhibits a stronger RKKY interaction than nickel, if the composition of cobalt on the surface in contact with a certain nonmagnetic layer is increased on the surface in contact with another nonmagnetic layer, the nonmagnetic layer It is possible to increase the strength of RKKY interaction via

  FIG. 7D is a cross-sectional view showing an example of the configuration of the MTJ element 1J in which the strength of the RKKY interaction is controlled by the composition of the ferromagnetic layer on the surface in contact with the nonmagnetic layer. The magnetization free layer 15J of the MTJ element 1J includes three ferromagnetic layers 21 to 23 and two nonmagnetic layers 31 and 32. The upper nonmagnetic layer 32 is formed to have a thickness in a range corresponding to a higher-order peak than the nonmagnetic layer 31. For example, the nonmagnetic layer 31 is formed to have a film thickness in a range corresponding to the secondary antiferromagnetic peak, and the nonmagnetic layer 32 has a film thickness in a range corresponding to the third antiferromagnetic peak. Formed to have. As described above, the strength of the RKKY interaction expressed by the nonmagnetic layers 31 and 32 is made more uniform by optimizing the film thicknesses of the nonmagnetic layers 31 and 32.

  The ferromagnetic layers 21 and 22 are composed of a laminated film of a NiFe film and a CoFe film. The effective film thickness (that is, the average film thickness) of the CoFe films 21b and 22b of the ferromagnetic layers 21 and 22 is extremely thin, and thus is formed in an island shape, and the NiFe film 21a and NiFe film positioned therebelow are formed. 22 b is in partial contact with the nonmagnetic layers 31 and 32. On the other hand, the ferromagnetic layer 23 is formed of a single layer of NiFe film.

  In the MTJ element 1J of FIG. 7D, the strength of the RKKY interaction via the nonmagnetic layers 31 and 32 is made uniform in addition to the optimization of the film thickness of the nonmagnetic layers 31 and 32, and the effective effect of the CoFe films 21b and 22b. Is achieved by optimal control of the film thickness. The upper surfaces of the nonmagnetic layers 31 and 32 are in contact with the NiFe film, whereas the lower surfaces of the nonmagnetic layers 31 and 32 are in contact with the CoFe films 21b and 22b having different effective film thicknesses. Since the CoFe films 21b and 21b cover the NiFe film 21a and the NiFe film 22b only incompletely, the composition of the ferromagnetic layers 21 and 22 on the lower surfaces of the nonmagnetic layers 31 and 32 is the same as that of the CoFe films 21b and 22b. It depends on the effective film thickness. The lower surface of the nonmagnetic layer 32 in contact with the CoFe film 22b having a large effective film thickness has a larger Co composition than the lower surface of the nonmagnetic layer 31 in contact with the CoFe film 21b having a small effective film thickness. The RKKY interaction via 32 is relatively strengthened. As described above, by controlling the Co composition on the lower surfaces of the nonmagnetic layers 31 and 32, it is possible to achieve a uniform level of interaction that cannot be completely achieved by optimizing the film thickness of the nonmagnetic layers 31 and 32.

  The RKKY interaction expressed by the nonmagnetic layers 31 and 32 can be weakened by forming another nonmagnetic film thin instead of the CoFe film. For example, by forming an Al film that is thin enough to form an island instead of the CoFe film 21b, the RKKY interaction expressed by the nonmagnetic layer 31 is relatively weakened. Thus, by controlling the composition of the surfaces of the nonmagnetic layers 31 and 32 in contact with the ferromagnetic layer, the strength of the interaction that cannot be completely achieved by optimizing the film thickness of the nonmagnetic layers 31 and 32 can be made uniform. Can be achieved.

  In the MTJ element 1J of FIG. 7D, the strength of the RKKY interaction is controlled by the composition of the ferromagnetic layer on the lower surface of the nonmagnetic layers 31 and 32, but the composition on the upper surface of the nonmagnetic layers 31 and 32 is also the same as that of the RKKY mutual. It can be used to control the strength of action.

  Control of the composition of the ferromagnetic layer on the surface in contact with the nonmagnetic layer can also be achieved by forming the ferromagnetic layer with a single layer of ferromagnetic film and making the composition different. However, such a configuration necessitates the formation of ferromagnetic layers having various compositions, which increases the manufacturing costs. It is preferable to use a laminated film for the ferromagnetic layer and to control the composition by the effective film thickness of the film constituting the laminated film because the composition can be controlled at a low cost.

  When the strength of interaction is controlled by the composition of the ferromagnetic layer on the surface in contact with the nonmagnetic layer, the thickness of the nonmagnetic layer may be the same. As described above, when the film thickness of the nonmagnetic layer is the same, the strength of the RKKY interaction expressed by them becomes stronger as the nonmagnetic layer located above. However, the strength of the RKKY interaction can be made uniform by optimizing the composition of the ferromagnetic layer on the surface in contact with the nonmagnetic layer.

  FIG. 7E is a cross-sectional view showing an example of the configuration of the MTJ element 1K having such a configuration. The magnetization free layer 15K of the MTJ element 1K includes three ferromagnetic layers 21 to 23 and two nonmagnetic layers 31 and 32. The nonmagnetic layers 31 and 32 have the same film thickness. Since the nonmagnetic layer 32 has higher crystallinity even if it has the same film thickness, the nonmagnetic layer 32 basically has stronger RKKY interaction than the nonmagnetic layer 31. It can be expressed.

  In order to make the strength of the RKKY interaction expressed by the nonmagnetic layers 31 and 32 uniform, the composition of the ferromagnetic layers 21 to 23 on the surface in contact with the nonmagnetic layers 31 and 32 is controlled. More specifically, the ferromagnetic layer 21 is composed of a laminated film of a NiFe film and a CoFe film, while the ferromagnetic layers 22 and 23 are composed of a single layer of NiFe film. The effective film thickness (that is, the average film thickness) of the CoFe film 21b of the ferromagnetic layer 21 is extremely thin. Therefore, the CoFe film 21b is formed in an island shape, and the NiFe film 21a located thereunder is partially a nonmagnetic layer. 31 is in contact.

  The upper surfaces of the nonmagnetic layers 31 and 32 are both in contact with the NiFe film, whereas the Co composition of the ferromagnetic layer on the lower surfaces of the nonmagnetic layers 31 and 32 is larger in the nonmagnetic layer 31. An increase in the Co composition on the lower surface of the nonmagnetic layer 31 causes the nonmagnetic layer 31 to exhibit a strong RKKY interaction. As described above, in the magnetization free layer 15K of FIG. 7E, the difference in the strength of the RKKY interaction due to crystallinity is offset by appropriately controlling the Co composition on the lower surfaces of the nonmagnetic layers 31 and 32, and the RKKY interaction The strength is uniform.

  It should be noted that the ferromagnetic layer referred to in the present invention should not be interpreted as being limited to a single-layered ferromagnetic film. Even if a ferromagnetic structure includes a plurality of ferromagnetic films, the structure is defined as a single ferromagnetic layer as long as they are ferromagnetically coupled and behave as a single ferromagnetic body. . A nonmagnetic film for ferromagnetically coupling them may be provided between the ferromagnetic films.

  FIG. 7F is a cross-sectional view showing the structure of the MTJ element 1M in which one ferromagnetic layer includes a plurality of ferromagnetic films that are ferromagnetically coupled. The magnetization free layer 15M of the MTJ element 1M includes three ferromagnetic layers 21 to 23 and two nonmagnetic layers 31 and 32. The upper nonmagnetic layer 32 is formed to have a thickness in a range corresponding to a higher-order peak than the nonmagnetic layer 31. For example, the nonmagnetic layer 31 is formed to have a film thickness in a range corresponding to the secondary antiferromagnetic peak, and the nonmagnetic layer 32 has a film thickness in a range corresponding to the third antiferromagnetic peak. Formed to have. As described above, the optimization of the film thickness of the nonmagnetic layers 31 and 32 is effective to make the strength of the RKKY interaction expressed by the nonmagnetic layers 31 and 32 close to uniform.

  The ferromagnetic layer 22 includes ferromagnetic films 22c and 22d and a nonmagnetic film 22e sandwiched between them. The nonmagnetic film 22e couples the ferromagnetic films 22c and 22d by the ferromagnetic RKKY interaction. Therefore, the ferromagnetic films 22c and 22d behave as a single ferromagnetic body as a whole. The nonmagnetic film 22e is preferably formed of a ruthenium film having a thickness of 1.2 to 1.8 nm.

  Further, like the MTJ element 1C shown in FIG. 7G, the magnetization fixed layer 13 is formed on the upper surface of the barrier layer 14, and the magnetization free layer is in contact with the lower surface of the barrier layer 14 (that is, the substrate). FIG. 7D shows a magnetization free layer 15A with three ferromagnetic layers. Although the degree of crystal growth of each layer in the magnetization free layer varies depending on the underlayer, even in such a configuration, the crystallinity of the nonmagnetic layer 32 to be formed later is the nonmagnetic layer 31 formed first. Higher than the crystallinity. Therefore, by forming the nonmagnetic layer 32 so as to have a film thickness in a range corresponding to a higher order peak than the nonmagnetic layer 31, the nonmagnetic layer 31 and the nonmagnetic layer 32 are developed in an antiferromagnetic manner. The strength of the RKKY interaction can be made closer to the same.

  However, it should be noted that the above technique is particularly effective when the magnetization free layer is formed on the upper surface of the barrier layer. The barrier layer is often amorphous, and therefore the ferromagnetic layer formed immediately above it does not have good crystallinity. Therefore, the nonmagnetic layer closest to the substrate (that is, closest to the barrier layer) among the nonmagnetic layers of the magnetization free layer also has poor crystallinity. The above technique for canceling the difference in RKKY interaction due to crystallinity by the effect of the film thickness of the nonmagnetic layer is most effective in such a case.

  In actual devices, the configuration of the underlayer for crystal growth of the magnetization free layer and the nonmagnetic layer with a suitable orientation plane is often limited by other functions required for the actual device. For example, when using an amorphous material for the tunnel barrier described above, and when arranging a magnetization free layer below the tunnel barrier, it is necessary to obtain a microcrystal on the underlayer of the magnetization free layer in order to obtain flatness of the underlayer. Alternatively, an amorphous material may be used. In addition, there may be a case where preparation of a sufficient base layer is restricted due to the restriction of the film thickness. Even in such a case, the present technology is effective.

  Further, even when the crystallinity of the underlayer of the magnetization free layer is good, the magnetization free layer to be grown thereon does not always grow while having a suitable crystal orientation. Rather, in general, undesired orientation on crystal planes or uneven growth occurs due to lattice matching or mismatching. Even in that case, the present technology is particularly effective. This is because the lowermost nonmagnetic layer of the magnetization free layer cannot avoid the influence of crystal growth or uneven growth on an undesired crystal plane, and therefore has undesired crystallinity and weakens the RKKY interaction. However, before the uppermost nonmagnetic layer grows, it is expected that the uppermost nonmagnetic layer recovers the desired crystallinity and strengthens the RKKY interaction by the ferromagnetic / nonmagnetic multilayer substrate.

(Second Embodiment)
FIG. 9A is a cross-sectional view showing a configuration of an MTJ element 1E according to the second embodiment of the present invention. As in the MTJ element 1 shown in FIG. 7A, the MTJ element 1E shown in FIG. 9A includes four ferromagnetic layers 21 to 24 and three nonmagnetic layers 31 to 31. 33.

The difference is that, among the nonmagnetic layers 31 to 33, the nonmagnetic layer 32 located in the middle is configured to develop a stronger RKKY interaction than the other nonmagnetic layers 31 and 33. Such a configuration is effective to increase the saturation magnetic field H _s while keeping the flop magnetic field H _flop of the magnetization free layer 15E low, that is, to increase the write margin. This is because the flop magnetic field H _flop of the magnetization free layer 15E strongly depends on the strength of the antiferromagnetic RKKY interaction expressed by the nonmagnetic layers 31 and 33 rather than the nonmagnetic layer 32, and the saturation magnetic field H _s is This is because it strongly depends on the strength of the RKKY interaction expressed by the nonmagnetic layer 32.

  Specifically, in the MTJ element 1E shown in FIG. 9A, the nonmagnetic layer 33 located farthest from the substrate 10 has a higher order peak than the nonmagnetic layer 31 located closest to the substrate 10. The nonmagnetic layer 32 located at the center is formed to have a film thickness in a range corresponding to a lower-order peak than the nonmagnetic layer 33. Thereby, the RKKY interaction stronger than the other nonmagnetic layers 31 and 33 can be expressed in the nonmagnetic layer 32.

  More specifically, the nonmagnetic layer 31 is formed to have a thickness corresponding to the secondary antiferromagnetic peak of the RKKY interaction, and the nonmagnetic layer 33 is the tertiary antiferromagnetic peak. It is formed to have a film thickness in a range corresponding to. The nonmagnetic layer 32 located in the center is formed so as to have a film thickness in a range corresponding to the primary or secondary antiferromagnetic peak. When the nonmagnetic layers 31 to 33 are formed of ruthenium, the nonmagnetic layer 31 is formed so that the film thickness exceeds 1.8 nm and less than 2.5 nm, and the nonmagnetic layer 33 includes It is formed so that the film thickness is more than 3.1 nm and less than 3.9 nm. The film thickness of the nonmagnetic layer 32 is greater than 0.7 nm and less than 1.2 nm, or greater than 1.8 nm and less than 2.5 nm. Most preferably, the nonmagnetic layer 31 is formed to have a film thickness of 2.1 nm corresponding to the antiferromagnetic 2nd peak, and the nonmagnetic layer 33 is formed of 3.5 nm corresponding to the antiferromagnetic 3nd peak. It is formed to have a film thickness. The nonmagnetic layer 31 is formed to have a thickness of 0.9 nm corresponding to the antiferromagnetic 1st peak or a thickness of 2.1 nm corresponding to the antiferromagnetic 2nd peak. Such a combination of film thicknesses brings the strength of the antiferromagnetic RKKY interaction expressed by the nonmagnetic layer 31 and the nonmagnetic layer 33 closer to the same, and further, the nonmagnetic layer 32 is in contact with other nonmagnetic layers. It is suitable for expressing RKKY interaction stronger than 31,33.

Such a technique can also be applied when the magnetization free layer has four or more nonmagnetic layers. When N layers (N is an integer of 4 or more) of nonmagnetic layers are provided in the magnetization free layer, the nonmagnetic layer located farthest from the substrate is higher in order than the nonmagnetic layer located closest to the substrate. The nonmagnetic layer is formed so as to have a film thickness corresponding to the antiferromagnetic peak, and at least one of the intermediate (N-2) nonmagnetic layers is located farthest from the substrate. It is formed so as to have a film thickness in a range corresponding to a lower order antiferromagnetic peak. According to such a configuration, the RKKY interaction expressed by the nonmagnetic layer positioned in the middle is made stronger than the RKKY interaction expressed by the uppermost layer and the lowermost nonmagnetic layer, and thereby the flop magnetic field H _flop is reduced. The saturation magnetic field H _s can be increased while keeping it low.

  FIG. 9B is a cross-sectional view showing an example of the configuration of the MTJ element 1F according to the present embodiment. In the MTJ element 1F, the magnetization free layer 15F is composed of six ferromagnetic layers 21 to 26 and five nonmagnetic layers 31 to 35. In the MTJ element 1F of FIG. 9B, the nonmagnetic layer 35 located farthest from the substrate 10 has a film thickness in a range corresponding to a higher order peak than the nonmagnetic layer 31 located closest to the substrate 10. Further, the nonmagnetic layer 33 located at the center is formed so as to have a film thickness in a range corresponding to a lower-order peak than the nonmagnetic layer 35. The remaining nonmagnetic layers 32 and 34 may have a film thickness in the range corresponding to the same antiferromagnetic peak as the nonmagnetic layer 35 (that is, the same film thickness), as shown in FIG. 9C. Further, it may be formed to have a film thickness in a range corresponding to a lower-order antiferromagnetic peak than the nonmagnetic layer 35.

  Specifically, in the MTJ element 1F illustrated in FIG. 9B, the lowermost nonmagnetic layer 31 is formed to have a film thickness in a range corresponding to the second antiferromagnetic peak of the RKKY interaction. The uppermost nonmagnetic layer 35 is preferably formed to have a film thickness in a range corresponding to the third antiferromagnetic peak. The nonmagnetic layer 33 located at the center is formed so as to have a film thickness in a range corresponding to the secondary antiferromagnetic peak. The remaining nonmagnetic layers 32 and 34 are formed so as to have a film thickness in a range corresponding to the third antiferromagnetic peak, similar to the nonmagnetic layer 35.

  On the other hand, in the MTJ element 1G shown in FIG. 9C, the nonmagnetic layer 31 located at the bottom of the magnetization free layer 15G has a film thickness in a range corresponding to the second antiferromagnetic peak of the RKKY interaction. It is preferable that the nonmagnetic layer 35 formed on the uppermost layer is formed to have a film thickness in a range corresponding to the third antiferromagnetic peak. The nonmagnetic layer 33 located at the center is formed so as to have a film thickness in a range corresponding to the secondary antiferromagnetic peak. The remaining nonmagnetic layers 32 and 34 are formed so as to have a film thickness in a range corresponding to the secondary antiferromagnetic peak, similar to the nonmagnetic layer 33.

  In the most preferred embodiment, the magnetization free layer has a RKKY interaction that is strongest in a nonmagnetic layer located in the center thereof, weakens as the distance from the center of the magnetization free layer decreases, and RKKY interaction. Is configured to be vertically symmetrical with respect to the center of the magnetization free layer. “Nonmagnetic layer located in the center” means a (N + 1) / 2th nonmagnetic layer from the substrate when the number N of nonmagnetic layers included in the magnetization free layer is an odd number. It is. On the other hand, when the number N of the ferromagnetic layers is an even number, it means two (N / 2) th and ([N / 2] +1) th ferromagnetic layers located from the substrate.

More precisely, when the magnetization free layer includes the first to (N + 1) th ferromagnetic layers and the first to Nth nonmagnetic layers, the jth nonmagnetic layer has α _{j of the} RKKY interaction, respectively. Assuming that the film thickness is in a range corresponding to the next antiferromagnetic peak, it is preferable that the magnetization free layer is formed so as to satisfy the following conditions. However, the k-th ferromagnetic layer means the k-th ferromagnetic layer closest to the substrate among the first to (N + 1) -th ferromagnetic layers, and the j-th non-magnetic layer means the first to first (N + 1) -th ferromagnetic layers. Note that this means the jth nonmagnetic layer closest to the substrate among the Nth nonmagnetic layers:
(1) When the number N of nonmagnetic layers is an odd number α ₁ ≧ α ₂ ≧ α ₃ ≧... ≧ α _{[N + 1] / 2-1} ≧ α _{[N + 1] / 2} , (1a)
α _{[N + 1] / 2} ≦ α _{[N + 1] / 2 + 1} ≦ ・ ・ ・ ≦ α _N−1 ≦ α _N , (1b)
α ₁ <α _N , (1c)
α ₂ = α _N−1 ,
α ₃ = α _N-2 ,
...
α _p = α _{N−p + 1} , (p is an integer of 2 or more and (N + 1) / 2-1 or less) (1d)
...
α _{[N + 1] / 2-1} = α _{[N + 1] / 2 + 1} .
(2) When the number N of nonmagnetic layers is an even number α ₁ ≧ α ₂ ≧ α ₃ ≧... ≧ α _{N / 2-1} ≧ α _{N / 2} , (2a)
α _{N / 2 + 1} ≦ α _{N / 2 + 2} ≦ ・ ・ ・ ≦ α _N−1 ≦ α _N , (2b)
α ₁ <α _N , (2c)
α ₂ = α _N−1 ,
...
α _p = α _{N−p + 1} , (p is an integer of 2 or more and N / 2 or less) (2d)
...
α _{N / 2-1} = α _{N / 2 + 2} ,
α _{N / 2} = α _{N / 2 + 1.}

  In such a configuration, since the strength of the RKKY interaction expressed by the nonmagnetic layer gradually changes, a part of the ferromagnetic layer of the magnetization free layer may behave independently of the remaining ferromagnetic layers. Is effectively prevented. This is effective for stably writing to the magnetization free layer.

  FIG. 9D is a cross-sectional view showing an example of the configuration of the MTJ element 1H in which the number N of nonmagnetic layers integrated in the magnetization free layer is five. The magnetization free layer 15H includes six ferromagnetic layers 21 to 26 and five nonmagnetic layers 31 to 35. The lowermost nonmagnetic layer 31 is formed to have a film thickness in a range corresponding to the second-order antiferromagnetic peak of the RKKY interaction, and the uppermost nonmagnetic layer 35 is a third layer. The film is formed to have a film thickness in a range corresponding to the next antiferromagnetic peak. The nonmagnetic layers 32 and 34 are both formed to have a film thickness in a range corresponding to the secondary antiferromagnetic peak. The nonmagnetic layer 33 located in the center is formed so as to have a film thickness in a range corresponding to the first antiferromagnetic peak. It will be easily understood that the magnetization free layer 15H configured in this way satisfies the above formulas (1a) to (1d).

  In the above-described embodiment, an MRAM that employs the toggle writing method is described, but it should be noted that the above-described configuration of the magnetization free layer can be applied to all MRAMs using spin flops.

1. Evaluation of flop magnetic field, saturation magnetic field, and write margin By forming the uppermost nonmagnetic layer to have a film thickness in a range corresponding to higher-order peaks than the lowermost nonmagnetic layer The effect of increasing the write margin was confirmed by preparing MTJ elements having a magnetization free layer formed of SAFs having various structures and measuring the characteristics. Specifically, 200 to 224 MTJ elements were prepared for each of Comparative Examples 1 to 3 and Examples 1 to 3 shown in FIG. The planar shape of the MTJ element is an oval of 0.4 × 0.8 μm ² . Further, the multilayer structure as a whole of the MTJ element is as follows:
Substrate / Ta (20 nm) / PtMn (20 nm) / CoFe (2.5 nm) / Ru (0
. 9 nm) / CoFe (2.5 nm) / Al (0.9 nm) O _x / magnetization free layer / Al (0
. 7 nm) O _x / Ta (100 nm)

Here, Al (αnm) O _x means an AlO _x film formed by oxidizing an Al film of α (nm). In addition, a Ta film, a PtMn film, a CoFe / Ru / CoFe laminate, and an Al (0.9 nm) O _x film sequentially formed on the substrate are a lower electrode layer 11, an antiferromagnetic layer 12, and a magnetization, respectively. Al (0.7 nm) O _x film and Ta film formed on the magnetization free layer correspond to the fixed layer 13 and the barrier layer 14, and correspond to the cap layer 16 and the upper electrode layer 17. Please note that.

  FIG. 10 is a table showing the configuration of the magnetization free layer of each sample. The number of ferromagnetic layers included in the magnetization free layer is 2 to 4, and the material and film thickness of the ferromagnetic layer are selected such that the residual magnetization as a whole of the magnetization free layer is zero. All the nonmagnetic layers of the magnetization free layer are made of ruthenium. In the following, the characteristics of each sample are schematically described.

  The magnetization free layer of the MTJ element of Comparative Example 1 is formed of SAF including two ferromagnetic layers and one nonmagnetic layer. The film thickness of the nonmagnetic layer is 2.1 nm corresponding to the antiferromagnetic 2nd peak.

  The magnetization free layer of the MTJ element of Comparative Example 2 is formed of SAF including three ferromagnetic layers and two nonmagnetic layers, and the magnetization free layer of the MTJ element of Comparative Example 3 is four ferromagnetic layers. And SAF including three nonmagnetic layers. In Comparative Examples 2 and 3, all the nonmagnetic layers included in the magnetization free layer have the same thickness of 3.5 nm.

  The sample of Example 1 is an MTJ element in which a magnetization free layer is formed by SAF including three ferromagnetic layers and two nonmagnetic layers. The nonmagnetic layer located away from the substrate has a film thickness corresponding to a higher-order antiferromagnetic peak than the nonmagnetic layer located near the substrate. Specifically, the thickness of the nonmagnetic layer located near the substrate is 2.1 nm corresponding to the antiferromagnetic 2nd peak, and the thickness of the nonmagnetic layer located away from the substrate is antiferromagnetic. It is 3.5 nm corresponding to the 3rd peak.

  The samples of Examples 2 and 3 are all MTJ elements in which a magnetization free layer is formed by SAF including four ferromagnetic layers and three nonmagnetic layers. The nonmagnetic layer located farthest from the substrate has a film thickness corresponding to a higher-order antiferromagnetic peak than the nonmagnetic layer located closest to the substrate. Specifically, the thickness of the nonmagnetic layer located near the substrate is 2.1 nm corresponding to the antiferromagnetic 2nd peak, and the thickness of the nonmagnetic layer located away from the substrate is antiferromagnetic. It is 3.5 nm corresponding to the 3rd peak.

  The difference between the samples of Examples 2 and 3 is the film thickness of the nonmagnetic layer located in the middle. In the sample of Example 2, the film thickness of the nonmagnetic layer located in the middle is 3.5 nm corresponding to the antiferromagnetic 3rd peak, and in the sample of Example 3, the film thickness of the nonmagnetic layer located in the middle is It is 2.1 nm corresponding to the antiferromagnetic 2nd peak. It should be noted that the sample of Example 3 is formed such that the nonmagnetic layer located in the middle expresses stronger RKKY interaction than the nonmagnetic layers located above and below it.

  The samples of Comparative Examples 2 and 3 in which the three-layer and four-layer ferromagnetic layers were integrated in the magnetization free layer did not show spin flops, and therefore could not be written by the toggle writing method. This is because if the nonmagnetic layer has the same film thickness, the strength of the RKKY interaction is not uniform, and the magnetization of the ferromagnetic layer included in the magnetization free layer is independently reversed. It is estimated to be.

  On the other hand, the MTJ elements of Comparative Example 1 and Examples 1 to 3 all exhibited spin flops and could be written by the toggle writing method. However, there is a difference in the size of the write margin between Comparative Example 1 and Examples 1 to 3. Below, the characteristic of the MTJ element of the comparative example 1 and Examples 1-3 is demonstrated in detail.

FIG. 11 is a frequency graph showing the flop magnetic field H _flop and saturation magnetic field H _s of the MTJ element of Comparative Example 1, and FIGS. 12A to 12C are the flop magnetic field H _flop and saturation magnetic field H _{s of} Examples 1 to 3, respectively. It is a frequency graph which shows. As shown in FIGS. 11 and 12A to 12C, the flop magnetic fields H _flop of the MTJ elements of Comparative Example 1 and Examples 1 to 3 are 50 (Oe), 55 (Oe), and 51.5 ( Oe) and 51.5 (Oe). The MTJ elements of Comparative Example 1 and Examples 1 to 3 showed no significant difference in the flop magnetic field H _flop .

On the other hand, the saturation magnetic field H _s, a significant difference appeared between Comparative Example 1 and Examples 1-3. Specifically, the saturation magnetic field H _s of the MTJ element of Comparative Example 1 is distributed around 147 (Oe), whereas the saturation magnetic field H _s of the MTJ elements of Examples 1 to 3 is respectively It was distributed around 246 (Oe), 326 (Oe), and 535 (Oe).

As a result, in Comparative Example 1 and Examples 1 to 3, a significant difference also appeared in the ratio H _s / H _flop representing the size of the write margin. In the MTJ element of Comparative Example 1, the ratio H _s / H _flop was 2.9, whereas in the MTJ elements of Examples 1 to 3, the ratio H _s / H _flop was 4.5 and 6 respectively. .3, 10.4. This indicates that the MTJ elements of Examples 1 to 3 are superior to Comparative Example 1 in terms of the write margin.

When Example 1 was compared with Example 2, Example 2 showed a larger saturation magnetic field H _s , that is, a larger write margin than Example 1. This indicates that increasing the number of ferromagnetic layers of the magnetization free layer is effective for increasing the write margin.

Further, when Example 2 and Example 3 were compared, Example 3 showed a larger saturation magnetic field H _s than Example 2, that is, a larger write margin. This indicates that the saturation magnetic field H _s , that is, the write margin can be increased by forming the nonmagnetic layer positioned in the middle so as to express a relatively strong RKKY interaction.

2. Evaluation of antiferromagnetic coupling by RKKY interaction Antiferromagnetism in which nonmagnetic layer closest to substrate (lowermost nonmagnetic layer) and nonmagnetic layer farthest from substrate (uppermost nonmagnetic layer) are developed The strength of the bond was evaluated using the SAF with the configuration shown in FIG. The material constituting the nonmagnetic layer is ruthenium. For the lowermost nonmagnetic layer, a nonmagnetic layer having a thickness of 3.5 nm corresponding to the antiferromagnetic 3rd peak, a nonmagnetic layer having a thickness of 2.1 nm corresponding to the antiferromagnetic 2nd peak, and Was evaluated. On the other hand, for the uppermost nonmagnetic layer, a nonmagnetic layer having a thickness of 3.5 nm corresponding to the antiferromagnetic 3rd peak was evaluated.

Samples (samples 1 and 2 in FIG. 13) for evaluating the antiferromagnetic coupling strength expressed by the lowermost nonmagnetic layer are two layers formed on an amorphous AlOx layer. A SAF having a ferromagnetic layer. A ruthenium layer having a thickness of 3.5 nm (sample 1) or 2.1 nm (sample 2) is interposed between the two ferromagnetic layers. The saturation magnetic field H _{s of the} SAF was measured, and the binding energy J of the RKKY interaction expressed by the lowermost nonmagnetic layer was calculated from the measured saturation magnetic field H _s . The calculated binding energy J represents the strength of the RKKY interaction expressed by the lowermost nonmagnetic layer.

On the other hand, a sample (sample 3) for evaluating the strength of antiferromagnetic coupling expressed by the uppermost nonmagnetic layer is a four-layer ferromagnetic layer formed on an amorphous AlOx layer. And a SAF having three non-magnetic layers interposed therebetween. In this sample, the lowermost nonmagnetic layer and the second nonmagnetic layer are formed of a ruthenium layer having a thickness of 3.1 nm, and the uppermost nonmagnetic layer is a ruthenium layer having a thickness of 3.5 nm. Been formed. Since the ruthenium layer having a thickness of 3.1 nm has an RKKY interaction strength of almost 0, in this sample, the uppermost nonmagnetic layer and the ferromagnetic layer sandwiching it are substantially Functions as SAF. The saturation magnetic field H _{s of the} SAF was measured, and the binding energy J of the RKKY interaction expressed by the lowermost nonmagnetic layer was calculated from the measured saturation magnetic field H _s . The calculated binding energy J represents the strength of the RKKY interaction expressed by the uppermost nonmagnetic layer.

  As understood from FIG. 13, when the uppermost nonmagnetic layer is a ruthenium layer having a film thickness of 3.5 nm corresponding to the antiferromagnetic 3rd peak, the lowermost nonmagnetic layer has an antiferromagnetic 3rd. A binding energy J close to that of the uppermost nonmagnetic layer was expressed not in the case of having a film thickness corresponding to the peak but in the case of having a film thickness corresponding to the antiferromagnetic 2nd peak. In order to make the antiferromagnetic coupling strength expressed by the lowermost nonmagnetic layer and the uppermost nonmagnetic layer close to the same level, the uppermost nonmagnetic layer is changed to the lowermost nonmagnetic layer. It shows that it is effective to form a film having a film thickness corresponding to a higher-order antiferromagnetic peak than the layer.

  This experimental result shows that a SAF having three or more ferromagnetic layers is formed such that the uppermost nonmagnetic layer has a film thickness corresponding to a higher-order antiferromagnetic peak than the lowermost nonmagnetic layer. As a result, the phenomenon of spin flops came to be due to the fact that the strength of antiferromagnetic coupling between the nonmagnetic layer at the bottom and the nonmagnetic layer at the top became close to the same. Suggests that

3. Control of antiferromagnetic coupling strength by the composition of the ferromagnetic layer In order to verify the effect of controlling the strength of the antiferromagnetic coupling by the composition of the ferromagnetic layer, the effects of the examples 4 to 6 were examined. A sample was created. In the samples of Examples 4 and 5, three layers of ferromagnetic films and two layers of nonmagnetic films are alternately stacked, and in the remanent magnetization state, each ferromagnetic film is coupled antiparallel through the nonmagnetic film. Has a free layer. In the sample of Example 6, four layers of ferromagnetic films and three layers of nonmagnetic films are alternately stacked, and only the nonmagnetic film at the center of the layer is set to exhibit ferromagnetic coupling. In the magnetized state, the lowermost and uppermost ferromagnetic films are coupled antiparallel to the two layers of ferromagnetic films existing between them, and the inner two layers of ferromagnetic films are coupled in parallel. It has a magnetization free layer. Although the sixth embodiment includes four layers of ferromagnetic films, the intermediate two layers of ferromagnetic films and the nonmagnetic film between them function as a single layer of ferromagnetic layers. Therefore, it should be noted that each of the magnetization free layers of Examples 4 to 6 functions as a SAF in which three ferromagnetic layers are antiparallel coupled to each other in the residual magnetization state.

The overall laminated structure of the MTJ element is as follows:
Substrate / Ta (20 nm) / PtMn (20 nm) / CoFe (2.5 nm) / Ru (0.9 nm) / CoFe (2.5 nm) / Al (0.9 nm) Ox / magnetization free layer / Al (0.7 nm ) Ox / Ta (100 nm)

The configuration of the magnetization free layer of the samples of Examples 4 to 6 is as follows.
Example 4:
Tunnel barrier / Ni ₈₁ Fe ₁₉ (3 nm) / CoFe (0.35 nm) / Ru (2.1 nm) / Ni ₈₁ Fe ₁₉ (6 nm) / CoFe (0.5 nm) / Ru (3.5 nm) / Ni ₈₁ Fe ₁₉ (3.7 nm)
Example 5:
Tunnel barrier / Ni ₈₁ Fe ₁₉ (3 nm) / CoFe (0.35 nm) / Ru (2.1 nm) / Ni ₈₁ Fe ₁₉ (7.3 nm) / Ru (2.1 nm) / Ni ₈₁ Fe ₁₉ (3.7 nm) )
Example 6:
Tunnel barrier / Ni ₈₁ Fe ₁₉ (3 nm) / CoFe (0.35 nm) / Ru (2.1 nm) / Ni ₈₁ Fe ₁₉ (3 nm) / CoFe (0.35 nm) / Ru (1.4 nm) / Ni ₈₁ Fe ₁₉ (3 nm) / CoFe (0.35 nm) / Ru (3.5 nm) / Ni ₈₁ Fe ₁₉ (3.7 nm)

  In each of Examples 1, 4, 5, and 6, the magnetization film thickness product of the first ferromagnetic layer (lowermost ferromagnetic layer) and the third ferromagnetic layer (uppermost ferromagnetic layer) is 3.15 T. It is equal to nm, and the magnetization film thickness product of the second ferromagnetic layer (intermediate ferromagnetic layer) is 6.3 T · nm, twice those. The configuration of Example 4 is the same as that of Example 3 except for the second ferromagnetic layer, and the configuration of Example 5 is the same as that of Example 1 except for the second and third ferromagnetic layers. It is. In Examples 1, 4, and 5, the magnetization film thickness product of each ferromagnetic layer is constant, and the antiparallel coupling force of the first nonmagnetic layer (lower nonmagnetic layer) is also constant, Only the antiparallel coupling force of the magnetic layer (upper nonmagnetic layer) increases in this order. As a means for increasing the antiparallel coupling force of the second nonmagnetic layer, in Example 4, the ratio of NiFe constituting the second ferromagnetic layer was adjusted to effectively reduce the CoFe film existing at the second nonmagnetic layer side interface. This is achieved by increasing the film thickness (average film thickness). The thickness of the CoFe film is as thin as 0.35 to 0.5 nm. Note that the NiFe film is in partial contact with the Ru film (2.1 nm). In such a case, the effective film thickness of the CoFe film is equivalent to the Co composition on the surface in contact with the Ru film. That is, the greater the effective film thickness of the CoFe film, the greater the Co composition on the surface in contact with the Ru film. In Example 5, conversely, the thickness of the CoFe film existing at the second nonmagnetic layer side interface is set to zero, and the antiparallel coupling force depending on the ferromagnetic film material is weakened, but the second nonmagnetic layer is made of the Ru film ( By increasing the antiparallel coupling force depending on the nonmagnetic layer using the 2nd antiferromagnetic peak of 2.1 nm), the antiparallel coupling force of the second nonmagnetic layer is the same as that in the three-layer SAF of this example. The strongest binding force is achieved.

200 to 220 oval MTJ devices of 0.4 × 0.8 μm ² having these magnetization free layers were fabricated on an 8-inch substrate, and their toggle writing characteristics were examined.

FIG. 14 shows 0.4 × 0.8 μm ² MTJ flop magnetic field H _flop composed of the three-layer SAF magnetization free layers of Examples 1 and 4 to 6, its variation (standard deviation), and saturation magnetic field Hs. It is a comparison. The devices according to Example 1 and Examples 4 to 6 all showed a good toggle operation.

The flop magnetic field H _flop of the MTJ elements of Examples 1, 4, 5, and 6 was distributed around 55 (Oe), 56 (Oe), 49 (Oe), and 54 (Oe), respectively. Slightly Example 5 showed the smallest flop magnetic field. On the other hand, the saturation magnetic field H _s was distributed around 246 (Oe), 27 1 (Oe), 268 (Oe), and 238 (Oe), respectively. As compared with the typical two-layer SAF of Comparative Example 1, the write margin of each of the three-layer SAFs of this example achieved a large increase in the saturation magnetic field and an effect of increasing the write margin, while keeping the flop magnetic field almost constant. is doing. Furthermore, regarding the variation (standard deviation) of the flop magnetic field of Examples 1, 4, 5, and 6, 8.7 (Oe), 4.0 (Oe), 3.3 (Oe), and 6.6 (Oe), respectively. Met. Compared to Example 1, the variation in the flop magnetic field of Examples 4 and 5 is less than half, and a better toggle writing characteristic is obtained. Further, even if the second ferromagnetic layer as in Example 6 is a ferromagnetic film / nonmagnetic film / ferromagnetic film and the nonmagnetic film is set to have ferromagnetic coupling, it is equivalent to that in Example 1. Toggle writing characteristics are obtained.

Further, the antiparallel coupling force of the first nonmagnetic layer and the second nonmagnetic layer of the elements of Examples 1, 4, and 5 was examined by evaluating the magnetization curve of the sample shown in FIG. The sample has the following configuration, and is substantially the same as the elements of Examples 1, 4, and 5 except for the magnetization free layer.
Substrate / Ta (20 nm) / PtMn (20 nm) / CoFe (2.5 nm) / Ru (0.9 nm) / CoFe (2.5 nm) / Al (0.9 nm) Ox / magnetization free layer / Al (0.7 nm ) Ox / Ta (10 nm)

  In FIG. 15, the sample name of each evaluated sample, the configuration of the magnetization free layer, the nonmagnetic layer evaluated for antiparallel coupling force, the SAF composed of the nonmagnetic layer and two ferromagnetic layers interposed therebetween are shown. The anti-parallel coupling energy J of the nonmagnetic layer obtained from the saturation magnetic field and the saturation magnetic field is shown.

As in the case of evaluating the antiparallel coupling force of the nonmagnetic layer in the four-layer SAF in FIG. 13, when evaluating the first nonmagnetic layer (lower nonmagnetic layer), the layers up to the second nonmagnetic layer were formed. When evaluating the second nonmagnetic layer (upper nonmagnetic layer), the sample (sample 4) is formed up to the third ferromagnetic layer, and the ruthenium thickness of the first nonmagnetic layer is 2.5 nm. Thus, samples (samples 5 to 7) in which the binding force of the first nonmagnetic layer was set to zero were prepared. The antiparallel bond energy was calculated from the saturation magnetic field of the obtained sample. In this case, since the antiparallel coupling energy of SAF having a difference in the magnetization film thickness product is evaluated, the following equation (2c) is used.
H _s = J _saf [1 / (M ₁ · t ₁ ) + 1 / (M ₂ · t ₂ )], (2c)

As can be understood from FIG. 15, the antiparallel coupling energy of the first nonmagnetic layer common to Examples 1, 4, and 5 is 0.0121 erg / cm ² , and the second non-magnetic properties of Examples 1, 4, and 5 are the same. The antiparallel coupling energy of the magnetic layer is 0.0097 erg / cm ₂ , 0.011 erg / cm ₂ , and 0.0138 erg / cm ² , respectively. As for the second ferromagnetic layer composed of the NiFe film and the CoFe film from Samples 5 and 6 in FIG. 15, the antiparallel coupling force of the second nonmagnetic layer in contact therewith increases as the CoFe film at the interface increases. I understand. In sample 7, the second nonmagnetic layer is made of the same material and film thickness as the first nonmagnetic layer. However, the interface material of the ferromagnetic layer in contact with the second nonmagnetic layer is only NiFe, whereas the ferromagnetic layer interface material in contact with the first nonmagnetic layer is a CoFe film having an effective film thickness of 0.35 nm. Exists. If the first nonmagnetic layer and the second nonmagnetic layer have the same film quality, the antiparallel coupling force of the second nonmagnetic layer should naturally decrease, but here the crystal orientation of the second nonmagnetic layer is high. Due to the effect, a higher antiparallel coupling force is developed, and as a result, the antiparallel coupling energy of the first nonmagnetic layer and the second nonmagnetic layer is set to be approximately equal.

  In Examples 1, 4, and 5, depending on the difference in the configuration of the second ferromagnetic layer and the second nonmagnetic layer, the antiparallel coupling force of the second nonmagnetic layer in the order of Examples 1, 4, and 5 Has increased. Compared with Example 1, Examples 4 and 5 are closer in antiparallel coupling force between the second nonmagnetic layer and the first nonmagnetic layer. In FIG. 14, the variation in the flop magnetic field in Examples 4 and 5 is smaller than that in Example 1 because the antiparallel coupling force between the second nonmagnetic layer and the first nonmagnetic layer approaches, and ferromagnetic layer 1 and strong This is considered to be an effect that the magnetic layer 3 flops simultaneously. In the multilayer SAF, the antiparallel coupling force between the lowermost nonmagnetic layer and the uppermost nonmagnetic layer is set under the condition that the magnetization film thickness products of the lowermost ferromagnetic layer and the uppermost ferromagnetic layer are set equal. It was shown by this experiment that the toggle writing operation rate is improved and the variation of the flop magnetic field is improved by making it more precise.

Claims (37)

A substrate,
A magnetization fixed layer having a fixed magnetization;
A magnetization free layer having reversible magnetization;
A nonmagnetic barrier layer interposed between the magnetization fixed layer and the magnetization free layer,
The magnetization free layer is
First to (N + 1) th ferromagnetic layers (N is an integer of 2 or more);
Including first to Nth nonmagnetic layers formed to exhibit antiferromagnetic RKKY interaction,
The kth nonmagnetic layer (k is an arbitrary integer of 1 or more and N or less) of the first to Nth nonmagnetic layers is the kth ferromagnetism of the first to (N + 1) th ferromagnetic layers. Between the layer and the (k + 1) th ferromagnetic layer,
The first nonmagnetic layer is located closest to the substrate among the first to Nth nonmagnetic layers, and the Nth nonmagnetic layer is among the first to Nth nonmagnetic layers. Located farthest from the substrate,
The first nonmagnetic layer has a film thickness in a range corresponding to the α _first- order antiferromagnetic peak of the RKKY interaction,
The Nth nonmagnetic layer has a thickness in a range corresponding to the α _Nth antiferromagnetic peak of the RKKY interaction,
The α ₁ and the α _N have the following relationship:
α ₁ <α _N ,
MRAM that satisfies An MRAM according to claim 1, wherein
The α ₁ and the α _N have the following relationship:
α _N = α ₁ +1,
MRAM that satisfies An MRAM according to claim 2,
The first nonmagnetic layer is composed of a ruthenium layer having a thickness of 1.8 nm to 2.5 nm,
The Nth nonmagnetic layer is an MRAM composed of a ruthenium layer having a thickness of 3.1 nm to 3.9 nm. An MRAM according to claim 2,
At least one nonmagnetic layer of the second to (N-1) nonmagnetic layers has a thickness in a range corresponding to a lower order antiferromagnetic peak than the Nth nonmagnetic layer. An MRAM according to claim 4, wherein
N is an odd number;
The at least one nonmagnetic layer is the ([N + 1] / 2) nonmagnetic layer of the first to Nth nonmagnetic layers. MRAM. An MRAM according to claim 5, wherein
Each of the second to (N-1) nonmagnetic layers has a thickness corresponding to the α _{2 to} α _N-1 order antiferromagnetic peaks of the RKKY interaction,
The α _{1 to} α _N are as follows:
α ₁ ≧ α ₂ ≧ α ₃ ≧... ≧ α _{[N + 1] / 2-1} ≧ α _{[N + 1] / 2} ,
α _{[N + 1] / 2} ≦ α _{[N + 1] / 2 + 1} ≦ ・ ・ ・ ≦ α _N−1 ≦ α _N ,
α ₁ <α _N ,
α ₂ = α _N−1 ,
α ₃ = α _N-2 ,
...
α _p = α _{N−p + 1} , (p is an integer of 2 or more and (N + 1) / 2-1 or less)
...
α _{[N + 1] / 2-1} = α _{[N + 1] / 2 + 1} ,
MRAM that satisfies An MRAM according to claim 4, wherein
N is an even number;
The at least one nonmagnetic layer is an (N / 2) th nonmagnetic layer and a ([N / 2] +1) th nonmagnetic layer of the first to Nth nonmagnetic layers. An MRAM according to claim 7, wherein
Each of the second to (N-1) nonmagnetic layers has a thickness corresponding to the α _{2 to} α _N-1 order antiferromagnetic peaks of the RKKY interaction,
The α _{1 to} α _N are as follows:
α ₁ ≧ α ₂ ≧ α ₃ ≧ ・ ・ ・ ≧ α _{N / 2-1} ≧ α _{N / 2} ,
α _{N / 2 + 1} ≦ α _{N / 2 + 2} ≦ ・ ・ ・ ≦ α _N−1 ≦ α _N ,
α ₁ <α _N ,
α ₂ = α _N−1 ,
...
α _p = α _{N−p + 1} , (p is an integer of 2 or more and (N / 2) or less)
...
α _{N / 2-1} = α _{N / 2 + 2} ,
α _{N / 2} = α _{N / 2 + 1} ,
MRAM that satisfies An MRAM according to claim 4, wherein
The first nonmagnetic layer is composed of a ruthenium layer having a thickness of 1.8 nm to 2.5 nm,
Each of the second to (N-1) nonmagnetic layers is a ruthenium layer having a thickness of 0.7 nm to 1.2 nm or a ruthenium layer having a thickness of 1.8 nm to 2.5 nm. Formed with
The Nth nonmagnetic layer is an MRAM composed of a ruthenium layer having a thickness of 3.1 nm to 3.9 nm. An MRAM according to claim 1, wherein
An MRAM, wherein the magnetization film thickness products of the first ferromagnetic layer and the (N + 1) th ferromagnetic layer are substantially equal. An MRAM according to claim 1, wherein
The barrier layer is an amorphous layer,
The magnetization free layer is an MRAM formed on an upper surface of the barrier layer. An MRAM according to claim 1, wherein
The first to Nth nonmagnetic layers are made of ruthenium;
The MRAM having higher crystal orientation in the direction perpendicular to the film surface of the ruthenium HCP (001) plane of the Nth nonmagnetic layer than the first nonmagnetic layer. An MRAM according to claim 1, wherein
The first to Nth ferromagnetic layers are composed of a ferromagnetic alloy mainly composed of Ni having an FCC structure, or a laminated film,
The MRAM having higher crystal orientation in the direction perpendicular to the FCC (111) plane of the Nth ferromagnetic layer than the first ferromagnetic layer. A magnetization fixed layer having a fixed magnetization;
A magnetization free layer having reversible magnetization;
A nonmagnetic barrier layer interposed between the magnetization fixed layer and the magnetization free layer,
The magnetization free layer is
First to (N + 1) th ferromagnetic layers (N is an integer of 2 or more);
Including first to Nth nonmagnetic layers formed to exhibit an antiferromagnetic interaction,
The kth nonmagnetic layer (k is an arbitrary integer of 1 or more and N or less) of the first to Nth nonmagnetic layers is the kth ferromagnetism of the first to (N + 1) th ferromagnetic layers. Between the layer and the (k + 1) th ferromagnetic layer,
The first nonmagnetic layer and the Nth nonmagnetic layer have a structure that is not the same, and the interaction through the first nonmagnetic layer and the interaction through the Nth nonmagnetic layer are effective. MRAM with almost the same strength. An MRAM according to claim 14, wherein
An MRAM in which the first nonmagnetic layer and the Nth nonmagnetic layer have different film thicknesses. An MRAM according to claim 14, wherein
The MRAM having different crystal orientations of the first nonmagnetic layer and the Nth nonmagnetic layer. An MRAM according to claim 16, wherein
The MRAM in which the Nth nonmagnetic layer has higher crystal orientation than the first nonmagnetic layer. An MRAM according to claim 16, wherein
The first to Nth nonmagnetic layers are made of ruthenium;
The MRAM having higher crystal orientation in the direction perpendicular to the film surface of the ruthenium HCP (001) plane of the Nth nonmagnetic layer than the first nonmagnetic layer. An MRAM according to claim 14, wherein
The MRAM, wherein the first ferromagnetic layer and the Nth ferromagnetic layer have different crystal orientations. An MRAM according to claim 19, comprising:
The MRAM in which the crystal orientation of the Nth ferromagnetic layer is higher than that of the first ferromagnetic layer. An MRAM according to claim 19, comprising:
The first to Nth ferromagnetic layers are composed of a ferromagnetic alloy mainly composed of Ni having an FCC structure, or a laminated film,
The MRAM having higher crystal orientation in the direction perpendicular to the FCC (111) plane of the Nth ferromagnetic layer than the first ferromagnetic layer. An MRAM according to claim 14, wherein
The first nonmagnetic layer is in contact with one of the first and second ferromagnetic layers on a first surface, and the Nth nonmagnetic layer is on the second surface with the Nth and In contact with one of the (N + 1) th ferromagnetic layers,
The MRAM in which the composition of the film in contact with the first nonmagnetic layer on the first surface is different from the composition of the film in contact with the Nth nonmagnetic layer on the second surface. An MRAM according to claim 22, wherein
At least one of the first ferromagnetic layer, the second ferromagnetic layer, the Nth ferromagnetic layer, and the (N + 1) th ferromagnetic layer is a laminated film in which a plurality of films having different compositions are laminated. By changing the effective film thickness ratio of the plurality of films constituting the laminated film, the interaction through the first nonmagnetic layer and the mutual through the Nth nonmagnetic layer are changed. MRAM in which the effective strength of action is set approximately equal. An MRAM according to claim 23, wherein
The multilayer film is an MRAM composed of a multilayer film of a NiFe film and a CoFe film. An MRAM according to claim 14, wherein
At least one of the first to Nth ferromagnetic layers has a multilayer structure in which at least two ferromagnetic films are ferromagnetically coupled by ferromagnetic RKKY coupling via a nonmagnetic film. Have MRAM. An MRAM according to claim 25, wherein
The non-magnetic film included in the multilayer structure is an MRAM configured by a ruthenium layer having a thickness of 1.2 nm to 1.8 nm. A magnetization fixed layer having a fixed magnetization;
A magnetization free layer having reversible magnetization;
A nonmagnetic barrier layer interposed between the magnetization fixed layer and the magnetization free layer,
The magnetization free layer is
First to (N + 1) th ferromagnetic layers (N is an even number of 2 or more);
Including first to Nth nonmagnetic layers formed to exhibit an antiferromagnetic interaction,
The kth nonmagnetic layer (k is an arbitrary integer of 1 or more and N or less) of the first to Nth nonmagnetic layers is the kth ferromagnetism of the first to (N + 1) th ferromagnetic layers. Between the layer and the (k + 1) th ferromagnetic layer,
The MRAM having substantially the same effective strength of the interaction through the first nonmagnetic layer and the interaction through the Nth nonmagnetic layer. An MRAM according to claim 27, wherein
The magnetic film thickness product of the (N / 2 + 1) th ferromagnetic layer among the first to (N + 1) th ferromagnetic layers is larger than the magnetic film thickness product of the other ferromagnetic layers. An MRAM according to claim 27, wherein
An MRAM in which the magnetization film thickness products of the first ferromagnetic layer and the (N + 1) th ferromagnetic layer are substantially equal. An MRAM according to claim 27, wherein
MRAM where N is 2. An MRAM according to claim 27, wherein
The first nonmagnetic layer is in contact with one of the first and second ferromagnetic layers on a first surface, and the Nth nonmagnetic layer is on the second surface with the Nth and In contact with one of the (N + 1) th ferromagnetic layers,
The MRAM in which the composition of the film in contact with the first nonmagnetic layer on the first surface is different from the composition of the film in contact with the Nth nonmagnetic layer on the second surface. An MRAM according to claim 31, comprising:
At least one of the first ferromagnetic layer, the second ferromagnetic layer, the Nth ferromagnetic layer, and the (N + 1) th ferromagnetic layer is a laminated film in which a plurality of films having different compositions are laminated. By changing the effective film thickness ratio of the plurality of films constituting the laminated film, the interaction through the first nonmagnetic layer and the mutual through the Nth nonmagnetic layer are changed. MRAM in which the effective strength of action is set approximately equal. An MRAM according to claim 32, comprising:
The multilayer film is an MRAM composed of a multilayer film of a NiFe film and a CoFe film. An MRAM according to claim 27, wherein
The Nth nonmagnetic layer has a higher crystal orientation than the first nonmagnetic layer, and the film thickness thereof is set equal.
The first nonmagnetic layer is in contact with one of the first and second ferromagnetic layers on a first surface, and the Nth nonmagnetic layer is on the second surface with the Nth and In contact with one of the (N + 1) th ferromagnetic layers,
The MRAM in which the composition of the film in contact with the first nonmagnetic layer on the first surface is different from the composition of the film in contact with the Nth nonmagnetic layer on the second surface. An MRAM according to claim 34, wherein
The MRAM, wherein a Co composition on the first surface of the first nonmagnetic layer is higher than a Co composition on the second surface of the Nth nonmagnetic layer. An MRAM according to claim 27, wherein
At least one of the first to Nth ferromagnetic layers has a multilayer structure in which at least two ferromagnetic films are ferromagnetically coupled by ferromagnetic RKKY coupling via a nonmagnetic film. Have MRAM. An MRAM according to claim 36, wherein
The non-magnetic film included in the multilayer structure is an MRAM configured by a ruthenium layer having a thickness of 1.2 nm to 1.8 nm.

Images