JP4987830B2 - Magnetic memory - Google Patents

Description

  The present invention relates to a magnetic memory, for example, a magnetic memory including a magnetoresistive element whose resistance value changes based on a write current.

  A magnetic random access memory (MRAM) includes an MTJ (Magnetic Tunnel Junction) element that uses a tunneling magnetoresistive (TMR) effect as a storage element. The MTJ element is composed of three thin films, a recording layer and a reference layer made of a magnetic material, and an insulating layer sandwiched between them, and stores information according to the magnetization states of the recording layer and the reference layer. In the spin injection MRAM, information is written to the MTJ element by energizing in a direction perpendicular to the MTJ film surface.

  As magnetic layers used in the MTJ element, a perpendicular magnetization film in which the magnetization direction is perpendicular to the film surface and an in-plane magnetization film in which the magnetization direction is in the in-plane direction have been proposed. When the perpendicular magnetization film is employed, the leakage magnetic field generated by the magnetization of the reference layer is oriented in the same direction as the magnetization of the recording layer, and therefore a magnetic field having a large perpendicular component acts on the recording layer. The leakage magnetic field from the reference layer acting on the recording layer acts in a direction that makes the magnetization of the recording layer parallel to the magnetization of the reference layer. Therefore, when the magnetization of the recording layer is reversed from antiparallel to parallel, a small spin injection current may be used. However, when the magnetization is reversed from parallel to antiparallel, a large current is required.

  Further, when the anti-parallel state is unstable due to the leakage magnetic field, if the leakage magnetic field becomes larger than the coercive force of the recording layer, the magnetization state of the MTJ element is reversed in the state where no magnetic field is applied from the outside of the MTJ element. It becomes impossible to hold in a parallel state. In addition, even when the leakage magnetic field is smaller than the coercive force of the recording layer, there is a problem that information cannot be retained because the anti-parallel state is reversed from the anti-parallel state due to thermal disturbance while the anti-parallel state is maintained for a long time. Arise. Therefore, the leakage magnetic field from the reference layer needs to be sufficiently small with respect to the coercive force of the recording layer.

  On the other hand, a double junction structure has been proposed in which a reference layer is disposed above and below a recording layer with a nonmagnetic layer interposed therebetween (Patent Document 1). When the MTJ element using the perpendicular magnetization film has a double junction structure, the magnetization directions of the two reference layers are set to be antiparallel to each other. In this case, since the perpendicular components of the leakage magnetic field generated from the two reference layers and acting on the recording layer are opposite to each other, the saturation magnetization Ms and the thickness of the two reference layers are adjusted to be substantially the same. Thus, the z component (component perpendicular to the film surface) of the leakage magnetic field can be substantially canceled.

However, since the radial component of the leakage magnetic field acts in a direction in which it increases, a strong transverse magnetic field acts particularly on the outer periphery of the recording layer. Due to this transverse magnetic field, there is a problem that the perpendicularity and magnetoresistance characteristics of the magnetization are lowered, and the uniformity of magnetization reversal of the recording layer is also deteriorated. In the double junction structure, it is necessary to make the two reference layers antiparallel to each other. For this purpose, the two reference layers need to be magnetized separately with a sufficient coercive force difference. Limited freedom of process conditions.
JP 2004-193595 A

  The present invention provides a magnetic memory capable of reducing a leakage magnetic field from a reference layer acting on a recording layer.

A magnetic memory according to one embodiment of the present invention includes a first magnetic layer that has a magnetic anisotropy perpendicular to a film surface and whose magnetization direction is changed by spin-polarized electrons, and a magnetic film perpendicular to the film surface. A magnetoresistive element having a second magnetic layer having anisotropy and a magnetization direction unchanged, and a nonmagnetic layer sandwiched between the first magnetic layer and the second magnetic layer; magnetic layer and has a magnetization parallel, and wherein enclose the magnetoresistive element, and comprising a bias magnetic field layer to reduce the leakage magnetic field from the second magnetic layer.

A magnetic memory according to one embodiment of the present invention includes a plurality of first magnetic layers having magnetic anisotropy in a direction perpendicular to a film surface and having a magnetization direction changed by spin-polarized electrons , and the plurality of first magnetic layers. A plurality of nonmagnetic layers provided on one magnetic layer, and provided on the plurality of nonmagnetic layers so as to extend in a plane, and having magnetic anisotropy in a direction perpendicular to the film surface; A second magnetic layer whose magnetization direction is unchanged ; a plurality of selection transistors each having one end of a current path electrically connected to the plurality of first magnetic layers; and a current path of the plurality of selection transistors And a plurality of bit lines electrically connected to the other end of each .

  According to the present invention, it is possible to provide a magnetic memory capable of reducing the leakage magnetic field from the reference layer acting on the recording layer.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description, elements having the same function and configuration are denoted by the same reference numerals, and redundant description will be given only when necessary.

[First Embodiment]
[1. Configuration of storage element]
First, the configuration of the magnetoresistive element [MTJ (magnetic tunnel junction) element] 10 corresponding to the memory element of the MRAM of this embodiment will be described. FIG. 1 is a cross-sectional view showing a configuration of an MTJ element 10 according to the first embodiment of the present invention. The MTJ element 10 is a storage element that stores information according to the relative magnetization directions of two magnetic bodies included therein. The arrow in the figure indicates the magnetization direction.

  The MTJ element 10 includes an underlayer 11 for crystal orientation, a recording layer (also referred to as a storage layer or a free layer) 12, a nonmagnetic layer (tunnel barrier layer) 13, a reference layer (also referred to as a fixed layer) 14, and an upper electrode 15. It has a laminated structure that is laminated in order. In the following description, the portions of the recording layer 12, the tunnel barrier layer 13, and the reference layer 14 are simply referred to as MTJ. In the present embodiment, the configuration in which the crystal orientation underlayer 11 serves as the lower electrode and forms a single layer is illustrated. Of course, the underlayer and the lower electrode may be stacked separately. The upper electrode 15 also functions as a hard mask layer. The recording layer 12 and the reference layer 14 may be stacked in reverse order.

  The recording layer 12 has a variable (reversed) magnetization (or spin) direction. The reference layer 14 has the same magnetization direction (fixed). “The magnetization direction of the reference layer 14 is invariable” means that when a magnetization reversal current used for reversing the magnetization direction of the recording layer 12 is passed through the reference layer 14, the reference layer 14 is energized before and after energization. This means that the magnetization direction does not change. Therefore, in the MTJ element 10, the magnetic layer having a large reversal current is used as the reference layer 14, and the magnetic layer having a reversal current smaller than that of the reference layer 14 is used as the recording layer 12. The MTJ element 10 including the reference layer 14 whose direction is not changed can be realized. When the magnetization reversal is caused by spin-polarized electrons, the reversal current is proportional to the attenuation constant, the anisotropic magnetic field, and the volume. Therefore, the reversal current between the recording layer 12 and the reference layer 14 is adjusted appropriately. A difference can be provided.

  Each of the reference layer 14 and the recording layer 12 has a magnetic anisotropy perpendicular to the film surface, and therefore the easy magnetization directions of the reference layer 14 and the recording layer 12 are perpendicular to the film surface (or the laminated surface) ( Hereinafter referred to as perpendicular magnetization). That is, the MTJ element 10 is a so-called perpendicular magnetization type MTJ element in which the magnetization directions of the reference layer 14 and the recording layer 12 are each perpendicular to the film surface. Note that the easy magnetization direction is a direction in which the internal energy is lowest when the spontaneous magnetization is directed in the absence of an external magnetic field, assuming a macro-sized ferromagnetic material. The difficult magnetization direction is a direction in which the internal energy is maximized when the spontaneous magnetization is directed in the absence of an external magnetic field, assuming a macro-sized ferromagnetic material.

  The underlayer 11 is a layer necessary for growing a flat perpendicular magnetization magnetic layer. For example, a compound layer such as titanium nitride (TiN) and a metal such as tantalum (Ta) or platinum (Pt) are used. It is a laminated structure in which a layer and thin magnesium oxide (MgO) having a thickness of 0.5 nm or less are laminated in order.

  As a material of the recording layer 12 and the reference layer 14, a ferromagnetic material having an L10 structure such as FePt or FePd, a ferrimagnetic material such as TbCoFe, or a laminated structure of a magnetic material such as NiFe and a nonmagnetic material such as Cu An artificial lattice made of

Examples of the tunnel barrier layer 13 include an insulating material such as magnesium oxide (MgO) or aluminum oxide (Al ₂ O ₃ ). Examples of the upper electrode (hard mask layer) 15 include metals such as tantalum (Ta) and titanium nitride (TiN).

  In the MTJ element 10 configured as described above, information is written as follows. When writing information, the MTJ element 10 is energized bidirectionally in a direction perpendicular to the film surface.

  When a write current is passed from the recording layer 12 to the reference layer 14, the electron flow is from the reference layer 14 to the recording layer 12. In this case, the magnetization of the recording layer 12 is subjected to spin torque in a direction that is aligned with the magnetization of the reference layer 14. For this reason, when the magnetization of the recording layer 12 is antiparallel to the reference layer 14, the magnetization of the recording layer 12 is reversed and becomes parallel to the reference layer 14.

  On the other hand, when a write current is passed from the reference layer 14 to the recording layer 12, the electron flow is from the recording layer 12 to the reference layer 14. In this case, the magnetization of the recording layer 12 is subjected to spin torque in a direction that is antiparallel to the reference layer 14. Therefore, when the magnetization of the recording layer 12 is parallel to the reference layer 14, the magnetization of the recording layer 12 is reversed and becomes antiparallel to the reference layer 14.

  The resistance value when a vertical read current flows through the MTJ element 10 changes depending on the relative directions of the two magnetic layers due to the magnetoresistive effect. For example, the resistance value of the MTJ element 10 is low when the magnetization directions of the recording layer 12 and the reference layer 14 are parallel, and high when it is antiparallel. In the example of FIG. 1, the upward state of the magnetization of the recording layer 12 is a parallel state, and the downward state is an antiparallel state. For example, by defining the low resistance state as data “0” and the high resistance state as data “1”, the MTJ element 10 can store 1-bit information.

  When the resistance value in the parallel state is R0 and the resistance value in the antiparallel state is R1, the value defined by “(R1−R0) / R0” is called a magnetoresistance ratio (MR ratio). The magnetoresistance ratio varies depending on the material constituting the MTJ element 10 and process conditions, but can take a value of several tens to several hundreds. The MRAM reads information stored in the MTJ element 10 using the magnetoresistive effect. The read current that flows through the MTJ element 10 during the read operation is set to a current value sufficiently smaller than the current at which the magnetization of the recording layer 12 is reversed by spin injection.

[2. Configuration of Bias Magnetic Field Layer 16]
Next, the configuration of the bias magnetic field layer 16 having a function of canceling the leakage magnetic field from the reference layer 14 acting on the recording layer 12 will be described. FIG. 2 is a perspective view showing the configuration of the MRAM including the bias magnetic field layer 16.

  The MRAM includes a plurality of MTJ elements 10 that are two-dimensionally arranged in a matrix. The configuration of each MTJ element 10 is the same as in FIG. In FIG. 2, four of the plurality of MTJ elements 10 are extracted and shown.

  The bias magnetic field layer 16 is disposed at a position overlapping with a space obtained by extending the MTJ in the film surface direction. The bias magnetic field layer 16 is effectively spread in a two-dimensional plane, and a plurality of holes are formed in a matrix. The MTJ element 10 is disposed so as to pierce into the hole of the bias magnetic field layer 16. In other words, the bias magnetic field layer 16 extends in the same plane as the MTJ and is configured to surround each MTJ element 10. “The bias magnetic field layer 16 extends in the same plane as the MTJ” means that the central position (center line) in the thickness direction of the bias magnetic layer 16 is any of the recording layer, tunnel barrier layer, and reference layer constituting the MTJ. Typically, it is between the center position (center line) of the recording layer constituting the MTJ and the center position (center line) of the reference layer (coincides with the positions of these centers). Means to be located).

  The bias magnetic field layer 16 has a magnetic anisotropy in a direction (perpendicular direction) parallel to the reference layer 14, that is, its easy magnetization direction is perpendicular to the film surface. Therefore, the bias magnetic field layer 16 is magnetized so that its magnetization direction is parallel to the magnetization direction of the reference layer 14. The saturation magnetization Ms of the bias magnetic field layer 16 is set to be equal to or higher than the saturation magnetization of the reference layer 14.

  The bias magnetic field layer 16 is insulated from the MTJ element 10 and is separated by an insulating film (not shown) having a thickness of several nm or more that can ensure insulation reliability. The upper electrode 15 is electrically connected to the upper wiring. The lower electrode 11 is electrically connected to a selection transistor or a lower wiring orthogonal to the upper wiring. FIG. 2 illustrates the case where the planar shape of the MTJ element 10 is a circle. The planar shape of the MTJ element 10 is not particularly limited, and any of an ellipse, a square, a rectangle, and the like other than a circle may be used. Further, a square or a rectangular shape with rounded corners may be used. The shape of the hole in the bias magnetic field layer 16 is set to be the same as the planar shape of the MTJ element 10. Then, the distance between the periphery of the MTJ element 10 and the bias magnetic field layer 16 is set at equal intervals through the insulating film.

  2 shows an example of a structure in which the recording layer 12 is disposed on the lower side (substrate side) with respect to the reference layer 14, but conversely, the recording layer 12 is on the upper side with respect to the reference layer 14 (opposite to the substrate). Even in the structure arranged on the side), the leakage magnetic field from the reference layer 14 acting on the recording layer 12 can be substantially canceled by arranging the bias magnetic field layer 16 as in FIG.

  FIG. 3 is a diagram for explaining how the bias magnetic field layer 16 cancels the leakage magnetic field from the reference layer 14 acting on the recording layer 12. When the magnetization of the reference layer 14 is upward, the magnetization of the bias magnetic field layer 16 is also set upward.

  The reference layer 14 applies an upward leakage magnetic field to the recording layer 12 as indicated by a solid arrow in FIG. Further, the reference layer 14 applies a stronger leakage magnetic field to the end portion in the in-plane direction of the recording layer 12 than to the central portion. On the other hand, the bias magnetic field layer 16 applies a downward leakage magnetic field to the recording layer 12 as indicated by a dotted arrow in FIG. As described above, the leakage magnetic field generated from the reference layer 14 and the leakage magnetic field generated from the bias magnetic field layer 16 are opposite to each other, and the external magnetic field acting on the recording layer 12 is canceled out. Further, since the leakage magnetic field acting on the recording layer 12 acts not only in the vertical direction component but also in the radial direction component, the radial direction component of the leakage magnetic field is canceled out.

  There is no gap between the reference layer 14 and the bias magnetic layer 16, the thickness of the reference layer 14 and the bias magnetic layer 16 is equal to the saturation magnetization, and the outer diameter of the bias magnetic layer 16 is sufficient with respect to the thickness (for example, , About 100 times or more), the magnetic field between the reference layer 14 and the bias magnetic field layer 16 is almost completely canceled, and the leakage magnetic field acting on the recording layer 12 becomes almost zero. In this case, the reference layer 14 and the bias magnetic field layer 16 can be regarded as a single magnetic film. If the magnetic film is sufficiently wide, it can be said that the leakage magnetic field does not leak out from the magnetic film. The reason for this will be described with reference to FIG.

  FIG. 4 shows a part of the magnetic film having perpendicular magnetization and extending in the in-plane direction. This magnetic film can be regarded as a combination of the bias magnetic field layer 16 and the recording layer 12. Positive and negative magnetic charges are induced on the top and bottom surfaces of the magnetic film, respectively, and it can be considered that the magnetic field is generated by these magnetic charges. Here, since the magnetic film is sufficiently wide, the magnetic field is oriented in the direction perpendicular to the film surface, and the magnetic flux density is preserved, so the magnetic field strength does not depend on the distance from the film surface. For this reason, for example, the magnetic field created by the positive magnetic charge at point A (solid arrow) and the magnetic field created by the negative magnetic charge (broken arrow) are reversed with the same intensity, so the combined magnetic field is canceled. Become zero.

  According to the principle of FIGS. 3 and 4, when the distance between the reference layer 14 and the bias magnetic field layer 16 is close to zero, the bias magnetic field layer 16 is disposed directly beside the reference layer 14 (in the same plane as the reference layer 14). In addition, by making the saturation magnetization of the bias magnetic field layer 16 and the reference layer 14 the same, the leakage magnetic field from the reference layer 14 acting on the recording layer 12 can be canceled by the bias magnetic field layer 16. Therefore, basically, it is preferable to configure the bias magnetic field layer 16 in this way.

  However, since there is a distance corresponding to the thickness of the insulating film between the reference layer 14 and the bias magnetic field layer 16, in order to completely cancel the leakage magnetic field from the reference layer 14, the magnetization of the insulating film It is necessary to compensate. As this method, as described later, the saturation magnetization of the bias magnetic field layer 16 is made slightly larger than the saturation magnetization of the reference layer 14, or the bias magnetic field layer 16 is brought a little closer to the recording layer 12. This adjustment width depends on the distance between the reference layer 14 and the bias magnetic field layer 16. The saturation magnetization of the bias magnetic field layer 16 can be adjusted by the magnetic material and its thickness.

  FIG. 5 is a perspective view showing another configuration example of the MRAM including the bias magnetic field layer 16. The MRAM in FIG. 5 is an example in which the bias magnetic field layer 16 is applied to the MTJ element 10 formed by a barrier stopping process in which an etching process for junction isolation is stopped by an insulating layer (tunnel barrier layer). FIG. 5 shows two extracted MTJ elements 10 two-dimensionally arranged in a matrix.

  The MTJ element 10 has a laminated structure in which a lower electrode 11, a reference layer 14, a tunnel barrier layer 13, a recording layer 12, and an upper electrode 15 are laminated in this order. That is, the reference layer 14 is disposed on the lower side (substrate side) with respect to the recording layer 12 due to process restrictions. The planar shapes of the recording layer 12 and the upper electrode 15 are the same as those of the MTJ element 10 of FIG. The area of the reference layer 14 is larger than the area of the recording layer 12. For example, the reference layer 14 has a planar shape in which rectangular corners are rounded. The bias magnetic field layer 16 is disposed at a position overlapping with a space obtained by extending the reference layer 14 in the film surface direction. The bias magnetic field layer 16 is effectively spread in a two-dimensional plane, and a plurality of holes are formed in a matrix. The reference layer 14 is disposed so as to pierce into the hole of the bias magnetic field layer 16. In other words, the bias magnetic field layer 16 extends in the same plane as the reference layer 14 and is configured to surround each reference layer 14.

  The bias magnetic field layer 16 has a magnetic anisotropy in a direction (perpendicular direction) parallel to the reference layer 14, that is, its easy magnetization direction is perpendicular to the film surface. The saturation magnetization of the bias magnetic field layer 16 is set to be equal to or higher than the saturation magnetization of the reference layer 14. The bias magnetic field layer 16 is insulated from the MTJ element 10 and is separated by an insulating film (not shown) having a thickness of several nm or more that can ensure insulation reliability. The upper electrode 15 is electrically connected to the upper wiring. The lower electrode 11 is electrically connected to a selection transistor or a lower wiring orthogonal to the upper wiring through a contact.

  Even when the MRAM is configured as shown in FIG. 5, the bias magnetic field layer 16 can substantially cancel the leakage magnetic field from the reference layer 14 acting on the recording layer 12 as in the MRAM shown in FIG.

  FIG. 6 is a diagram showing a model used for the simulation calculation for confirming the effect of the bias magnetic field layer 16. FIG. 6 shows a cross section of one MTJ element 10 and a bias magnetic field layer 16 disposed around the MTJ element 10.

  The center of the recording layer 12 is the origin, and the z axis is set in the direction perpendicular to the film surface. The reference layer 14 had a thickness of 10 nm, the recording layer 12 had a thickness of 2.2 nm, the tunnel barrier layer had a thickness of 1 nm, and the MTJ had a disk shape with a diameter of 60 nm. The bias magnetic field layer 16 has a ring shape with an inner diameter of 100 nm for easy calculation. A 20 nm insulating region was provided between the MTJ and the bias magnetic field layer 16. The outer diameter R of the ring was 20 μm. When the bias magnetic field layers 16 are connected to each other as shown in FIG. 2, the calculation is performed on the assumption that the magnetic field acting on the MTJ element 10 of interest hardly changes due to the existence of the adjacent MTJ element 10. It was. Further, the thickness of the bias magnetic field layer 16 is “d”, the difference between the height of the bias magnetic field layer 16 and the height of the recording layer 12 is “h”, and the height of the bias magnetic field layer 16 is the same as the height of the recording layer 12. If h = 0. The “height of the bias magnetic field layer 16” refers to the height at the center position of the bias magnetic layer 16 in the thickness direction. Similarly, “the height of the recording layer 12” refers to the height at the center position of the recording layer 12 in the thickness direction. The saturation magnetizations of the reference layer 14 and the bias magnetic field layer 16 are both about 1000 emu / cc.

  FIG. 7 is a graph showing the leakage magnetic field distribution acting on the center position in the thickness direction of the recording layer 12 when the bias magnetic field layer 16 is arranged at the same height as the recording layer 12 (h = 0). The horizontal axis in FIG. 7 indicates the radial distance from the center of the recording layer 12, and the vertical axis indicates the z component [Hz (Oe)] of the combined magnetic field applied to the recording layer 12 from the reference layer 14 and the bias magnetic field layer 16. ing. FIG. 7 shows the respective results when the thickness d of the bias magnetic field layer 16 is changed in the range of 10 to 20 nm.

  Since the radius of the reference layer 14 is 30 nm, in a region where the distance is 30 nm or less, the magnetic field generated by the reference layer 14 and the magnetic field generated by the bias magnetic field layer 16 cancel each other, and the magnetic field Hz is near zero. . That is, the leakage magnetic field applied to the recording layer 12 can be canceled by adjusting the thickness d of the bias magnetic field layer 16. In the model of FIG. 6, the effect is large when the thickness d of the bias magnetic field layer 16 is 15 to 17 nm.

  FIG. 8 is a graph showing the average area of the leakage magnetic field acting on the central position of the recording layer 12 in the thickness direction. The horizontal axis of FIG. 8 is the thickness d (nm) of the bias magnetic field layer 16, and the vertical axis is the area average [Hz (Oe) in the z component of the combined magnetic field applied from the reference layer 14 and the bias magnetic layer 16 to the recording layer 12. ] Is shown. Increasing the thickness d of the bias magnetic field layer 16 increases the downward magnetic field, and the leakage magnetic field of the reference layer 14 can be canceled when the thickness d is about 15 nm.

  FIG. 9 is a graph showing the average area of the leakage magnetic field acting on the center position in the thickness direction of the recording layer 12 when the outer diameter R of the bias magnetic field layer 16 is fixed to 170 nm. The horizontal axis of FIG. 9 is the thickness d (nm) of the bias magnetic field layer 16, and the vertical axis is the area average [Hz (Oe) in the z component of the combined magnetic field applied from the reference layer 14 and the bias magnetic layer 16 to the recording layer 12. ] Is shown. The parameters and calculation method other than the outer diameter of the bias magnetic field layer 16 are the same as those in FIG.

  Under the conditions of FIG. 9, a magnetic field of about 870 Oe remains even if the thickness of the bias magnetic field layer 16 is increased to 20 nm. Under the condition of R = 20 μm in FIG. 8, the outer peripheral portion of the bias magnetic field layer 16 is sufficiently separated from the region of 30 nm or less from the center of the recording layer 12. The effect on 12 can be ignored. However, under the condition of R = 170 nm in FIG. Since the leakage magnetic field generated from the outer peripheral portion acts on the recording layer 12 and works in the direction of increasing the leakage magnetic field from the reference layer 14, an external magnetic field acting on the recording layer 12 remains. By extrapolating FIG. 9, the thickness d of the bias magnetic field layer 16 for completely canceling the leakage magnetic field becomes about 35 nm. Since it is difficult to make the magnetization of the magnetic film having the thickness d perpendicular, it is preferable that the outer peripheral portion of the bias magnetic field layer 16 be sufficiently separated from the recording layer 12.

  On the other hand, in order to increase the capacity of the MRAM, it is not preferable to increase the pitch of the MTJ element array. As shown in FIG. 2 or FIG. 5, the bias magnetic field layer 16 is shared by the plurality of MTJ elements 10, so that the outer peripheral portion of the bias magnetic field layer 16 can be sufficiently separated from the MTJ elements without increasing the pitch of the MTJ element array. Can be released. In this case, the MTJ elements arranged in the outer peripheral area of the array also need to be separated from the end of the bias magnetic field layer 16.

  FIG. 10 shows the calculation of the leakage magnetic field acting on the center position in the thickness direction of the recording layer 12 when the thickness d of the bias magnetic field layer 16 is fixed to 15 nm and the outer diameter R of the bias magnetic field layer 16 is used as a parameter. It is a graph which shows a result. The horizontal axis of FIG. 10 is the outer diameter R (nm) of the bias magnetic field layer 16, and the vertical axis is the area average [Hz (Oe) in the z component of the combined magnetic field applied from the reference layer 14 and the bias magnetic layer 16 to the recording layer 12. ] Is shown.

  According to the result of FIG. 10, since the outer diameter R is required to be about 2 μm or more when the allowable amount of the magnetic field Hz is 100 Oe, the distance between the outer peripheral portion of the bias magnetic field layer 16 and the MTJ element 10 is about 1 μm or more, which is a half thereof. I need it. Therefore, it is necessary to set the end of the bias magnetic field layer 16 with a margin of about 1 μm or more with respect to the MTJ elements arranged on the outermost periphery in the MTJ element array sharing the bias magnetic field layer 16.

  FIG. 11 is a diagram for explaining a margin at the end of the bias magnetic field layer 16. “A” and “B” are MTJ elements arranged on the outermost periphery of the MTJ element array. The distances dA and dB from these MTJ elements to the end of the bias magnetic field layer 16 require a margin of about 1 μm or more. However, a dummy MTJ element that is not used as a storage element may be arranged in the margin area. By using the dummy MTJ element in this way, it is possible to increase the distance between the MTJ element actually used as a storage element and the end of the bias magnetic field layer 16.

  FIG. 12 is a graph showing the average area of the leakage magnetic field acting on the center position in the thickness direction of the recording layer 12 when the bias magnetic field layer 16 is arranged at the same height as the reference layer 14 (h = 7.1 nm). It is. The horizontal axis of FIG. 12 is the thickness d (nm) of the bias magnetic field layer 16, and the vertical axis is the area average [Hz (Oe) in the z component of the composite magnetic field applied from the reference layer 14 and the bias magnetic layer 16 to the recording layer 12. ] Is shown. The calculation is performed with the outer diameter R of the bias magnetic layer 16 being 20 μm.

  As shown in FIG. 12, when the thickness d of the bias magnetic layer 16 is set to 16 nm, the leakage magnetic field applied to the recording layer 12 from the reference layer 14 and the bias magnetic layer 16 is canceled. Comparing the results of FIG. 12 with the results of FIG. 8, the efficiency of canceling the leakage magnetic field of the z component applied to the recording layer 12 is recorded when the height of the bias magnetic field layer 16 is the same as that of the reference layer 14. Slightly worse than when layer 12 is the same. However, when the height of the bias magnetic field layer 16 is the same as that of the reference layer 14, the leakage magnetic field of the radial direction component is also canceled. For this reason, it is desirable that the height of the bias magnetic field layer 16 be the same as that of the reference layer 14. However, when the saturation magnetization or thickness of the bias magnetic field layer 16 cannot be sufficiently increased, the height of the bias magnetic field layer 16 is set to the reference layer 14. And the recording layer 12.

  When the thickness of the recording layer 12 is dF, the thickness of the tunnel barrier layer 13 is dT, and the thickness of the reference layer 14 is dP, the height of the bias magnetic field layer 16 is the height of the recording layer 12 and the reference layer 14. When the height satisfies the condition “0 <h ≦ (dF / 2) + (dP / 2) + dT”, the radial component of the leakage magnetic field between the reference layer 14 and the bias magnetic field layer 16 is It acts in the direction of reducing each other. Therefore, the radial component of the leakage magnetic field applied to the recording layer 12 can be reduced by setting the height of the bias magnetic field layer 16 to satisfy the range h.

[3. Manufacturing method of MRAM]
Next, an example of a method for manufacturing the MRAM including the bias magnetic field layer 16 will be described with reference to the drawings.

An interlayer insulating layer 21 is deposited on a MOS transistor or FEOL (Front End Of Line) formed on a semiconductor substrate, and a contact 22 electrically connected to a lower wiring or a MOS transistor is formed in the interlayer insulating layer 21. FIG. 13 shows a cross section in which the upper surface is flattened by CMP (Chemical Mechanical Polishing) and etch back after the formation. For example, silicon oxide (SiO ₂ ) is used as the interlayer insulating layer 21, and tungsten (W) is used as the contact 22, for example.

  Subsequently, as illustrated in FIG. 14, the base layer 11, the MTJ film, and the hard mask layer 15 are sequentially formed on the contact 22 by, for example, sputtering. The underlayer 11 is a layer necessary for growing a flat perpendicular magnetization magnetic layer and is made of the above-described material. The MTJ film is composed of a recording layer 12, a tunnel barrier layer 13, and a reference layer 14 from the bottom. As the material of the recording layer 12 and the reference layer 14, for example, FePt having an L10 structure is used, and as the tunnel barrier layer 13, for example, magnesium oxide (MgO) is used. As the hard mask layer 15, for example, tantalum (Ta) is used. A thin cap layer such as magnesium oxide (MgO) or ruthenium (Ru) may be sandwiched between the MTJ film and the hard mask layer 15.

Subsequently, as shown in FIG. 15, element separation is performed by lithography and etching to form a plurality of MTJ elements 10. Subsequently, as shown in FIG. 16, an insulating film 23 is deposited on the MTJ element 10 and the exposed interlayer insulating layer 21. For example, silicon oxide (SiO ₂ ) or silicon nitride (SiN) is used as the insulating film 23. The insulating film 23 is for insulating the MTJ element 10 and the bias magnetic field layer 16. Therefore, the insulating film 23 is also formed on the side surface of the MTJ element 10, and the thickness of the insulating film 23 is thick enough to ensure the insulation reliability. At the same time, the bias magnetic field layer 16 forms the recording layer 12 of the MTJ element 10. It is necessary to satisfy a condition that the leakage magnetic field from the acting reference layer 14 can be canceled. For example, the thickness of the insulating film 23 is about 5 to 30 nm.

  Subsequently, as shown in FIG. 17, the base layer 24 and the bias magnetic field layer 16 are sequentially formed on the insulating film 23 by, for example, sputtering. Since the distance between the bias magnetic field layer 16 and the MTJ element 10 is controlled by self-alignment according to the thickness of the insulating film 23 formed on the side surface of the MTJ element 10, the variation becomes small. The variation of can be suppressed. Sputtering of the bias magnetic field layer 16 is performed under a high directivity condition so that the magnetic layer attached to the side surface of the hard mask layer 15 is reduced.

  As the magnetic material for the bias magnetic field layer 16, a material whose magnetization direction is perpendicular and whose saturation magnetization is about the same as that of the reference layer 14 is used. As the bias magnetic field layer 16, for example, FePt having the same L10 structure as that of the reference layer 14 is used. Also, by providing an appropriate underlayer 24 under the bias magnetic field layer 16, the perpendicularity of the magnetization of the bias magnetic field layer 16 is maintained, and the height of the bias magnetic field layer 16 and the height of the MTJ are substantially the same. adjust. Thereafter, a cap layer (not shown) for protecting the bias magnetic layer 16 is formed on the bias magnetic layer 16. A material similar to that of the underlayer 11 is used for the underlayer 24.

Subsequently, as shown in FIG. 18, an interlayer insulating layer 25 made of, for example, silicon oxide (SiO ₂ ) or silicon nitride (SiN) is deposited on the entire surface of the sample. Subsequently, as shown in FIG. 19, the interlayer insulating layer 25 is planarized by CMP. At the same time, the bias magnetic field layer 16 and the base layer 24 deposited on the upper surface of the MTJ element 10 are also scraped to expose the hard mask layer 15 on the MTJ film.

  Subsequently, as shown in FIG. 20, an upper wiring 26 that is electrically connected to the hard mask layer 15 is formed on the MTJ element 10. As the upper wiring 26, for example, aluminum (Al) or copper (Cu) is used. In this way, the MRAM of this embodiment is formed.

Next, another method of manufacturing the MRAM that prevents the bias magnetic field layer 16 from adhering to the side surface of the hard mask layer 15 will be described. As shown in FIG. 21, after the MTJ film is formed, a first hard mask layer 15 made of metal and a second hard mask layer 15A made of an insulator are sequentially formed on the MTJ film. As the first hard mask layer 15, for example, tantalum (Ta) is used. For example, silicon oxide (SiO ₂ ) is used as the second hard mask layer 15A.

Subsequently, as shown in FIG. 22, the second hard mask layer 15A is patterned into the same shape as the planar shape of the MTJ element 10 by lithography and RIE (Reactive Ion Etching). Subsequently, the first hard mask layer 15 is etched by RIE under strong isotropic conditions (high temperature, low bias, chlorine gas (Cl ₂ ), etc.), so that an under layer is formed under the second hard mask layer 15A. Make a cut.

  Subsequently, as shown in FIG. 23, the MTJ and the base layer 11 are etched by the oblique incidence milling or the like using the first hard mask layer 15 as a mask. At this time, the MTJ has a forward tapered shape. Subsequently, an insulating film 23 made of, for example, silicon nitride is deposited on the entire surface of the sample.

  Subsequently, as shown in FIG. 24, a base layer (not shown) and a bias magnetic field layer 16 are formed on the insulating film 23 by sputtering under a highly directional condition. At this time, since an undercut is present under the second hard mask layer 15A, the side surface of the first hard mask layer 15 is hidden under the overhang. This prevents the bias magnetic field layer 16 from adhering to the side surfaces of the first hard mask layer 15 during the formation of the bias magnetic field layer 16.

Subsequently, as shown in FIG. 25, an interlayer insulating layer 25 made of, for example, silicon oxide (SiO ₂ ) is deposited on the entire surface of the sample. Subsequently, the bias magnetic field layer 16, the underlayer, the insulating film 23, and the second hard mask layer 15A deposited on the upper surface of the first hard mask layer 15 are removed by CMP to expose the first hard mask layer 15. Let Subsequently, as shown in FIG. 26, the upper wiring 26 is formed on the hard mask layer 15.

  The point of this process is that an overhang is formed in the second hard mask layer 15A in order not to attach the bias magnetic field layer 16 to the side surface of the first hard mask layer 15 (in other words, the second hard mask layer 15A Undercut). As a result, the side surface of the first hard mask layer 15 becomes a shadow of the second hard mask layer 15A when the bias magnetic field layer 16 is formed, so that the magnetic layer does not adhere. One method for this is to form an undercut by isotropically etching the lower layer of the two hard masks under conditions that facilitate side etching, as described with reference to FIG.

In addition to the example of FIG. 22, as shown in FIG. 27, on the first hard mask layer 15, a second hard mask layer 15B made of, for example, silicon nitride (SiN), for example, a second hard mask layer made of, for example, silicon oxide (SiO ₂ ). There is also a method of forming an undercut under the third hard mask layer 15A by sequentially forming the third hard mask layer 15A and etching the second hard mask layer 15B by, for example, CDE (Chemical Dry Etching). . Alternatively, the lower second hard mask layer 15B is made of silicon oxide (SiO ₂ ), and the upper third hard mask layer 15A is made of silicon nitride (SiN). Then, the second hard mask layer 15B made of silicon oxide (SiO ₂ ) may be etched using hydrofluoric acid (HF) so as to make an undercut under the third hard mask layer 15A. . Thereafter, the first hard mask layer 15, the MTJ, and the base layer 11 are patterned into desired shapes.

Next, another method for forming an overhang (or undercut) in the hard mask layer will be described. As shown in FIG. 28, after forming the MTJ film, an insulating layer 27 made of, for example, silicon oxide (SiO ₂ ) is deposited on the MTJ film. Subsequently, a hole 28 exposing the MTJ film is formed in the insulating layer 27 by lithography and etching so that the etching surface of the insulating layer 27 becomes a forward taper.

  Subsequently, as shown in FIG. 29, after the hard mask layer 15 made of, for example, tantalum (Ta) is embedded in the hole 28, the upper surface of the hard mask layer 15 is planarized by CMP. Then, the hard mask layer 15 having an overhang can be formed by removing the insulating layer 27.

  As described in detail above, in the first embodiment, an MRAM having a plurality of MTJ elements 10 two-dimensionally arranged in a matrix is shared by the plurality of MTJ elements 10 and surrounds each MTJ element 10. A new layer 16 is provided. The bias magnetic field layer 16 is effectively spread in a two-dimensional plane. The bias magnetic field layer 16 has the same perpendicular magnetic anisotropy as the reference layer 14, and the saturation magnetization thereof is set to be equal to or higher than the saturation magnetization of the reference layer 14.

  Therefore, according to the first embodiment, the leakage magnetic field from the reference layer 14 acting on the recording layer 12 can be substantially canceled by the bias magnetic field layer 16. Thereby, it is possible to prevent the magnetization state of the recording layer 12 from being reversed by the leakage magnetic field. In particular, the stability when the magnetizations of the recording layer 12 and the reference layer 14 are antiparallel is improved. As a result, the retention characteristic of the MTJ element 10 as a memory element can be improved.

  Further, by adjusting the thickness and saturation magnetization of the bias magnetic field layer 16 or adjusting the height of the bias magnetic field layer 16, the leakage magnetic field acting on the recording layer 12 can be canceled with high accuracy.

  Further, since the magnetization directions of the bias magnetic field layer 16 and the reference layer 14 are the same, it is not necessary to provide a coercive force difference between the bias magnetic field layer 16 and the reference layer 14. Further, the bias magnetic field layer 16 and the reference layer 14 can be magnetized simultaneously. This facilitates the selection of materials and reduces process condition constraints.

[Second Embodiment]
In the second embodiment, the bias magnetic field layer 16 is formed so as to extend in the same direction as the upper wiring 26, and this bias magnetic field layer 16 is electrically connected to the reference layer 14 directly or via the upper wiring 26. Like to do.

[1. Configuration of Bias Magnetic Field Layer 16]
FIG. 30 is a perspective view showing the configuration of the MRAM according to the second exemplary embodiment of the present invention. The MRAM includes a plurality of MTJ elements 10 that are two-dimensionally arranged in a matrix. Each MTJ element 10 has a stacked structure in which a base layer 11, a recording layer 12, a tunnel barrier layer 13, a reference layer 14, and a hard mask layer 15 are stacked in this order. The recording layer 12 is electrically connected to the selection transistor or the lower wiring orthogonal to the upper wiring through the base layer 11. In FIG. 30, the base layer 11 and the hard mask layer 15 are not shown for easy understanding of the drawing, that is, only the MTJ portion is shown in FIG. FIG. 30 shows 12 (4 × 3) extracted from a plurality of MTJs two-dimensionally arranged in a matrix.

  The MRAM includes a plurality of bias magnetic field layers 16 corresponding to the number of MTJ columns, and the plurality of bias magnetic field layers 16 are arranged adjacent to each other in the X direction. Each bias magnetic field layer 16 has a structure extending in the same direction (Y direction) as the upper wiring, and is formed so as to surround each of the reference layers 14 included in one row of MTJs in the same plane. ing. That is, the bias magnetic field layer 16 has a number of holes corresponding to the MTJ. The bias magnetic field layer 16 has magnetic anisotropy in the same direction (vertical direction) as that of the reference layer 14, and the saturation magnetization of the bias magnetic field layer 16 is set to the same level as the saturation magnetization of the reference layer 14.

  The bias magnetic field layer 16 is in direct contact with the reference layer 14 and forms at least a part of the upper wiring. As shown in FIG. 30, the bias magnetic field layer 16 may also function as the upper wiring without providing the upper wiring. Actually, an upper wiring 26 extending in the same direction as the bias magnetic field layer 16 is provided on the bias magnetic layer 16, and the bias magnetic layer 16 and the reference layer 14 are electrically connected to the upper wiring 26. It is desirable to reduce electrical resistance. When the upper wiring 26 is provided in this way, the bias magnetic field layer 16 and the reference layer 14 do not need to be in direct contact, and the bias magnetic field layer 16 and the reference layer 14 are electrically connected to the upper wiring 26, respectively. It only has to be. That is, in FIG. 30, an insulating region may be provided between the bias magnetic field layer 16 and the reference layer 14. This configuration example will be described below.

  FIG. 31 is a plan view showing another configuration example of the MRAM according to the second embodiment. FIG. 32 is a cross-sectional view of the MRAM along the line AA ′ shown in FIG. In FIG. 31, 9 (3 × 3) pieces of a plurality of MTJs arranged in a two-dimensional matrix are extracted and shown.

  The MRAM includes a plurality of bias magnetic field layers 16 corresponding to the number of MTJ columns, and each of the plurality of bias magnetic field layers 16 extends in the Y direction. Each bias magnetic field layer 16 is configured to surround each row of MTJ elements 10 in the same plane, and is insulated from the MTJ element 10 by an insulating film 23. The bias magnetic field layer 16 has magnetic anisotropy in the same direction (vertical direction) as that of the reference layer 14, and the saturation magnetization of the bias magnetic field layer 16 is set to be equal to or higher than the saturation magnetization of the reference layer 14. Is done.

  An upper wiring 26 extending in the Y direction is provided on the bias magnetic field layer 16 and the hard mask layer 15. That is, the bias magnetic field layer 16 and the reference layer 14 are electrically connected via the upper wiring 26. Similar to the bias magnetic field layer 16, the upper wiring 26 extends in the Y direction. In this way, the MRAM of the second embodiment may be configured.

  The plurality of bias magnetic field layers 16 and the reference layer 14 arranged adjacent to each other in the X direction can be regarded as one magnetic film spreading in the same plane by reducing the distance between the bias magnetic field layers 16. it can. Therefore, when the bias magnetic field layer 16 shown in the second embodiment is used, the leakage magnetic field acting on the recording layer 12 from the reference layer 14 and the bias magnetic field layer 16 can be almost canceled. The principle that the leakage magnetic field is canceled is the same as that in FIGS. 3 and 4 shown in the first embodiment.

  FIG. 33 is a diagram illustrating a model used for simulation calculation for confirming the effect of the bias magnetic field layer 16 under the condition that the bias magnetic field layer 16 and the reference layer 14 are in contact with each other. FIG. 33 corresponds to a cross-sectional view of the configuration of FIG. 30 cut along the X direction. In this model, the bias magnetic field layer 16 and the reference layer 14 have a width of 50 nm and a thickness of 10 nm, the tunnel barrier layer 13 has a thickness of 1 nm, the recording layer 12 has a width of 50 nm and a thickness of 2 nm. The distance between the adjacent bias magnetic field layers 16 was 50 nm. In addition, the calculation was performed on the assumption that the structure of FIG.

  FIG. 34 is a graph showing the leakage magnetic field distribution acting on the center position (the position of the alternate long and short dash line) in the thickness direction of the recording layer 12. The horizontal axis in FIG. 34 indicates the distance (nm) from the end of the bias magnetic field layer 16, and the vertical axis indicates the z component [Hz (Oe)] of the leakage magnetic field at the center position in the thickness direction of the recording layer 12. . FIG. 34 shows the z component of the leakage magnetic field generated from the reference layer 14, the z component of the leakage magnetic field generated from the bias magnetic field layer 16, and the z component of these combined magnetic fields.

  As shown in FIG. 34, the leakage magnetic field between the reference layer 14 and the bias magnetic field layer 16 acts in a direction to cancel each other out at the center position in the thickness direction of the recording layer 12. Since the distance between the bias magnetic field layers 16 is not zero, in order to completely cancel the leakage magnetic field from the reference layer 14, it is necessary to supplement the magnetization for this distance. In the model of FIG. 33, the saturation magnetization of the reference layer 14 is set to 1000 emu / cc, and the saturation magnetization of the bias magnetic field layer 16 is set to 1200 emu / cc, so that the leakage acting on the recording layer 12 from the reference layer 14 and the bias magnetic field layer 16 is set. The magnetic field can be made almost zero.

  In the second embodiment, for the same reason as in the first embodiment, the distance from the MTJ element arranged at the outermost periphery of the MTJ element array to the end of the bias magnetic field layer 16 is about 1 μm or more. A margin is required. Therefore, also in the second embodiment, it is necessary to set the end of the bias magnetic field layer 16 with a margin of about 1 μm or more with respect to the MTJ element arranged on the outermost periphery in the MTJ element array.

  The height of the bias magnetic field layer 16 is adjusted between the reference layer 14 and the recording layer 12 so that the leakage magnetic field applied to the recording layer 12 is canceled.

[2. Manufacturing method of MRAM]
Next, an example of a method for manufacturing the MRAM including the bias magnetic field layer 16 will be described with reference to the drawings. The manufacturing steps from FIG. 13 to FIG. 17 are the same as those in the first embodiment.

  As shown in FIG. 35, after a cap layer (not shown) made of, for example, ruthenium (Ru) is deposited on the bias magnetic field layer 16, an upper wiring 26 made of, for example, aluminum (Al) is formed on the entire surface of the sample by, for example, sputtering. accumulate. Subsequently, as shown in FIG. 36, planarization is performed to the middle of the hard mask layer 15 by CMP to expose the upper surface of the hard mask layer 15.

  Subsequently, as shown in FIG. 37, the upper wirings 26 are stacked by sputtering, for example. Subsequently, the upper surface of the upper wiring 26 is planarized by CMP. Subsequently, as shown in FIG. 38, the upper wiring 26 is patterned by lithography and etching, and the bias magnetic field layer 16 and the base layer 24 are cut. Thereby, the bias magnetic field layers 16 adjacent to each other in the X direction can be electrically separated, and each bias magnetic field layer 16 can be electrically connected to the corresponding upper wiring 26. In this way, the MRAM of this embodiment is formed.

  As described above in detail, in the second embodiment, a plurality of bias magnetic field layers 16 corresponding to the number of columns are provided in an MRAM having a plurality of MTJ elements 10 two-dimensionally arranged in a matrix. Are shared by the MTJ elements 10 arranged in a row, and are configured to surround the MTJ elements 10 in a row. The bias magnetic field layer 16 is electrically connected to the reference layer 14 via the upper wiring 26 extending in the same direction as this, or directly connected to the reference layer 14 not via the upper wiring 26. ing.

  Therefore, according to the second embodiment, the leakage magnetic field from the reference layer 14 acting on the recording layer 12 can be almost canceled by the bias magnetic field layer 16. Thereby, it is possible to prevent the magnetization state of the recording layer 12 from being reversed by the leakage magnetic field.

  In the structure of FIG. 30, it is not necessary to insulate the bias magnetic field layer 16 and the reference layer 14. Therefore, the step for preventing the magnetic layer from adhering to the side surface of the reference layer 14 when forming the bias magnetic field layer 16 as described in the first embodiment is unnecessary. As a result, the film forming process of the bias magnetic field layer 16 can be facilitated.

  Further, since the bias magnetic field layer 16 is used as a part of the upper wiring, a structure similar to a conventional device structure having no bias magnetic field layer can be used. That is, it is not necessary to make a large design change of the MRAM, and an increase in manufacturing process due to the application of the bias magnetic field layer 16 to the MRAM can be suppressed.

[Third Embodiment]
In the third embodiment, the reference layer 14 is formed in a planar shape, and the reference layer 14 is shared by a plurality of MTJ elements 10. Thereby, the leakage magnetic field acting on the recording layer 12 from the reference layer 14 is eliminated.

[1. Configuration of MRAM]
FIG. 39 is a plan view showing the configuration of the MRAM according to the third embodiment of the present invention. FIG. 40 is a cross-sectional view of the MRAM along the line AA ′ shown in FIG.

  The MRAM includes a plurality of MTJs in which a recording layer 12, a tunnel barrier layer 13, and a reference layer 14 are sequentially stacked. In FIG. 39, 6 (2 × 3) pieces of a plurality of MTJs arranged two-dimensionally in a matrix are extracted and shown. The reference layer 14 of each MTJ is electrically connected. That is, the reference layer 14 extends in a planar shape and is shared by a plurality of MTJs.

  In other words, the present embodiment has a structure in which the reference layer 14 and the bias magnetic field layer are integrated or electrically connected. Since this structure is equivalent to the absence of a gap between the reference layer 14 and the bias magnetic field layer, the reference layer 14 is formed in a planar shape (that is, the reference layer 14 and the bias magnetic layer have the same thickness and saturation magnetization). Therefore, the leakage magnetic field that acts on the recording layer 12 from the reference layer 14 can be eliminated.

  Under each recording layer 12, an underlayer (not shown) is provided as in the first embodiment. On the planar reference layer 14, for example, an upper wiring (not shown) having the same planar shape as the reference layer 14 is provided.

  In the structure of FIG. 40, since the MTJ reference layers are electrically connected to each other, it is necessary to select an element only on the lower side (substrate side) of the MTJ. For this reason, the recording layer 12 is electrically connected to the bit line BL via the selection transistor 31. The gate terminal of the selection transistor 31 is electrically connected to the word line WL.

  FIG. 41 is a plan view showing another configuration example of the MRAM according to the third embodiment. FIG. 42 is a cross-sectional view of the MRAM along the line AA ′ shown in FIG.

  The bias magnetic field layer 16 extends in the same plane as the MTJ, and is configured to surround each of the plurality of MTJs. The bias magnetic field layer 16 is insulated from the MTJ element 10 by the insulating film 23. On the bias magnetic field layer 16 and the reference layer 14, an upper wiring 26 extending in a planar shape is provided. That is, the bias magnetic field layer 16 and the reference layer 14 are electrically connected via the upper wiring 26. In this manner, the MRAM of the third embodiment can be configured.

[2. Manufacturing method of MRAM]
Next, an example of a method for manufacturing the MRAM according to the third embodiment will be described with reference to the drawings.

  As shown in FIG. 43, a reference layer 14, a tunnel barrier layer 13, and a recording layer 12 are sequentially formed on an underlayer (not shown) to form an MTJ film. Subsequently, as shown in FIG. 44, element separation is performed by patterning the recording layer 12 and the tunnel barrier layer 13 into a desired planar shape by lithography and etching.

  Subsequently, as shown in FIG. 45, a circuit for element selection (including the selection transistor 31) is formed on another substrate. Then, the two substrates are bonded together so that the recording layer 12 faces the circuit substrate. In this way, the MRAM shown in FIGS. 39 and 40 is formed.

  Next, another method for manufacturing the MRAM according to the third embodiment will be described. The manufacturing steps from FIG. 13 to FIG. 17 are the same as those in the first embodiment. Subsequently, as shown in FIG. 46, the hard mask layer 15 is removed by CMP, and the upper surfaces of the reference layer 14 and the bias magnetic field layer 16 are exposed.

  Subsequently, as shown in FIG. 47, the upper wiring 26 made of, for example, aluminum (Al) is deposited on the reference layer 14 and the bias magnetic field layer 16 by, for example, sputtering. Thereafter, the upper surface of the upper wiring 26 is planarized by CMP. In this way, the MRAM shown in FIGS. 41 and 42 is formed.

  As described above in detail, in the third embodiment, the reference layer 14 is formed in a planar shape so that the leakage magnetic field acting on the recording layer 12 from the reference layer 14 is eliminated. Thereby, it is possible to prevent the magnetization state of the recording layer 12 from being reversed by the leakage magnetic field.

[Example]
Examples of the MRAM shown in the first and second embodiments will be described below.

  FIG. 48 is a circuit diagram illustrating a configuration of the MRAM according to the embodiment. The MRAM includes a memory cell array 32 having a plurality of memory cells MC arranged in a matrix. In the memory cell array 32, a plurality of bit line pairs BL, / BL are arranged so as to extend in the column direction. In the memory cell array 32, a plurality of word lines WL are arranged so as to extend in the row (row) direction.

  A memory cell MC is arranged in an intersection region between the bit line BL and the word line WL. Each memory cell MC includes an MTJ element 10 and a selection transistor 31 composed of an N-channel MOS transistor. One end of the MTJ element 10 is connected to the bit line BL. The other end of the MTJ element 10 is connected to the drain terminal of the selection transistor 31. The gate terminal of the selection transistor 31 is connected to the word line WL. The source terminal of the selection transistor 31 is connected to the bit line / BL.

  A row decoder 33 is connected to the word line WL. A write circuit 35 and a read circuit 36 are connected to the bit line pair BL, / BL. A column decoder 34 is connected to the write circuit 35 and the read circuit 36. Each memory cell MC is selected by the row decoder 33 and the column decoder 34.

  Data is written to the memory cell MC as follows. First, in order to select a memory cell MC for data writing, the word line WL connected to the memory cell MC is activated by the row decoder. As a result, the selection transistor 31 is turned on. Further, the bit line pair BL, / BL connected to the connection memory cell MC is selected by the column decoder 34.

  Here, a bidirectional write current is supplied to the MTJ element 10 in accordance with the write data. Specifically, when a write current is supplied to the MTJ element 10 from left to right, the write circuit 35 applies a positive voltage to the bit line BL and applies a ground voltage to the bit line / BL. When supplying a write current to the MTJ element 10 from right to left, the write circuit 35 applies a positive voltage to the bit line / BL and applies a ground voltage to the bit line BL. In this way, data “0” or data “1” can be written in the memory cell MC.

  Next, data reading from the memory cell MC is performed as follows. First, as in the case of writing, the selection transistor 31 of the selected memory cell MC is turned on. The read circuit 36 supplies the MTJ element 10 with a read current that flows from right to left, for example. This read current is set to a value smaller than a threshold value at which magnetization is reversed by spin injection. The read circuit 36 detects the resistance value of the MTJ element 10 based on the read current. In this way, data stored in the MTJ element 10 can be read.

  The present invention is not limited to the above-described embodiment, and can be embodied by modifying the components without departing from the scope of the invention. In addition, various inventions can be configured by appropriately combining a plurality of constituent elements disclosed in the embodiments. For example, some constituent elements may be deleted from all the constituent elements disclosed in the embodiments, or constituent elements of different embodiments may be appropriately combined.

1 is a cross-sectional view showing a configuration of an MTJ element 10 according to a first embodiment. FIG. 4 is a perspective view showing a configuration of an MRAM including a bias magnetic field layer 16. The figure explaining a mode that the bias magnetic field layer 16 cancels a leakage magnetic field. The figure which shows a part of magnetic film which has perpendicular magnetization and spreads in an in-plane direction. The perspective view which shows the other structural example of MRAM provided with the bias magnetic field layer 16. FIG. The figure which shows the model used for the simulation calculation for confirming the effect of the bias magnetic field layer. 6 is a graph showing a leakage magnetic field distribution acting on the recording layer 12. The graph which shows the area average of the leakage magnetic field which the recording layer 12 acts. The graph which shows the area average of the leakage magnetic field which acts on the recording layer 12 at the time of fixing outer diameter R = 170nm. 6 is a graph showing a leakage magnetic field acting on the recording layer 12 when the thickness d of the bias magnetic field layer 16 is fixed to 15 nm. FIG. 6 is a diagram for explaining a margin at an end of a bias magnetic field layer 16. 6 is a graph showing an average area of a leakage magnetic field acting on the recording layer 12 when the bias magnetic field layer 16 is disposed at the same height as the reference layer 14. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. The perspective view which shows the structure of MRAM which concerns on 2nd Embodiment. The top view which shows the other structural example of MRAM. FIG. 32 is a cross-sectional view of the MRAM along the line AA ′ shown in FIG. 31. The figure which shows the model used for the simulation calculation for confirming the effect of the bias magnetic field layer. 6 is a graph showing a leakage magnetic field distribution acting on the recording layer 12. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. The top view which shows the structure of MRAM which concerns on 3rd Embodiment. FIG. 40 is a cross-sectional view of the MRAM along the line AA ′ shown in FIG. 39. The top view which shows the other structural example of MRAM which concerns on 3rd Embodiment. FIG. 42 is a cross-sectional view of the MRAM along the line AA ′ shown in FIG. 41. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. Sectional drawing which shows 1 process of the other manufacturing method of MRAM. The circuit diagram which shows the structure of MRAM which concerns on an Example.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... MTJ element, 11 ... Underlayer, 12 ... Recording layer, 13 ... Tunnel barrier layer, 14 ... Reference layer, 16 ... Bias magnetic field layer, 21 ... Interlayer insulating layer, 22 ... Contact, 23 ... Insulating film, 24 ... Bottom Base layer, 25 ... interlayer insulating layer, 26 ... upper wiring, 27 ... insulating layer, 28 ... hole, 31 ... select transistor, 32 ... memory cell array, 33 ... row decoder, 34 ... column decoder, 35 ... write circuit, 36 ... read Circuit, BL ... bit line, WL ... word line, MC ... memory cell.

Claims (10)

A first magnetic layer having a magnetic anisotropy perpendicular to the film surface and having a magnetization direction changed by spin-polarized electrons, and a magnetic anisotropy perpendicular to the film surface and the magnetization direction being unchanged. A magnetoresistive element having a second magnetic layer and a nonmagnetic layer sandwiched between the first magnetic layer and the second magnetic layer;
A bias magnetic field layer to reduce the leakage magnetic field from the second having a magnetization parallel and the magnetic layer, and wherein enclose the magnetoresistive element, and said second magnetic layer,
A magnetic memory comprising:   The magnetic memory according to claim 1, wherein the bias magnetic field layer surrounds each of the plurality of magnetoresistive elements arranged in two dimensions.   3. The magnetic memory according to claim 1, wherein the bias magnetic field layer is disposed on the same plane as the second magnetic layer.   The magnetic memory according to claim 1, wherein the bias magnetic field layer is insulated from the magnetoresistive element by an insulating region.   The magnetic memory according to claim 1, wherein the bias magnetic field layer is electrically connected to the second magnetic layer.   The wiring board further includes a wiring layer provided on a side opposite to the nonmagnetic layer of the second magnetic layer and electrically connected to the second magnetic layer and the bias magnetic field layer. The magnetic memory according to any one of 1 to 3.   7. The magnetic memory according to claim 1, wherein the saturation magnetization of the bias magnetic field layer is the same as that of the second magnetic layer.   8. The magnetic memory according to claim 1, wherein the bias magnetic field layer has the same thickness as that of the second magnetic layer.   The magnetic memory according to claim 1, wherein a distance between an end of the bias magnetic field layer and the magnetoresistive element is 1 μm or more. A plurality of first magnetic layers having magnetic anisotropy perpendicular to the film surface and having a magnetization direction changed by spin-polarized electrons ;
A plurality of nonmagnetic layers provided on the plurality of first magnetic layers;
A second magnetic layer that is provided on the plurality of nonmagnetic layers so as to extend in a plane, has a magnetic anisotropy in a direction perpendicular to the film surface, and has an invariable magnetization direction;
A plurality of select transistors having one end of a current path electrically connected to each of the plurality of first magnetic layers;
A plurality of bit lines respectively electrically connected to the other ends of the current paths of the plurality of select transistors;
A magnetic memory comprising:

Images