阅读说明：本技术 包括不连续互连区段的磁阻元件和包括多个磁阻元件的磁存储器 (Magnetoresistive element comprising discontinuous interconnection segments and magnetic memory comprising a plurality of magnetoresistive elements ) 是由 S·马丁 J·罗切 M·德鲁阿尔德 于 2021-03-17 设计创作，主要内容包括：包括不连续互连区段的磁阻元件和包括多个磁阻元件的磁存储器。本公开涉及一种用于制造磁阻元件的方法,该磁阻元件包括磁性隧道结MTJ,该磁性隧道结MTJ包括隧道势垒层、第一铁磁层和第二铁磁层；写入电流层；以及互连层,其被配置用于向写入电流层供应写入电流。在互连层中提供间隙,使得互连层包括沿着层平面延伸并串联连接写入电流层的两个不连续互连区段。该方法包括：沉积互连层(50)、写入电流层(30)、第二铁磁层(23)、隧道势垒层(22)和第一铁磁层(21)；在互连层(50)中形成间隙(34)；用间隙材料填充间隙；以及通过执行单个蚀刻步骤来形成支柱(40),直到达到充当停止层的互连层(50)。(Magnetoresistive elements including discontinuous interconnected segments and magnetic memories including a plurality of magnetoresistive elements. The present disclosure relates to a method for manufacturing a magnetoresistive element comprising a magnetic tunnel junction MTJ comprising a tunnel barrier layer, a first ferromagnetic layer and a second ferromagnetic layer; writing a current layer; and an interconnect layer configured to supply a write current to the write current layer. A gap is provided in the interconnect layer such that the interconnect layer includes two discontinuous interconnect segments extending along the layer plane and connecting the write current layer in series. The method comprises the following steps: depositing an interconnect layer (50), a write current layer (30), a second ferromagnetic layer (23), a tunnel barrier layer (22), and a first ferromagnetic layer (21); forming a gap (34) in the interconnect layer (50); filling the gap with a gap material; and forming the pillars (40) by performing a single etching step until reaching the interconnect layer (50) acting as a stop layer.)

1. A method for manufacturing a magnetoresistive element (10), the magnetoresistive element (10) comprising a magnetic tunnel junction, MTJ (20), the magnetic tunnel junction, MTJ (20) comprising a tunnel barrier layer (22) sandwiched between a first ferromagnetic layer (21) having a first magnetization (210) and a second ferromagnetic layer (23) having a second magnetization (230); a write current layer (30) extending in a layer Plane (PL) substantially parallel to the layers (21, 22, 23) of the MTJ (20) and contacting the second ferromagnetic layer (23), the write current layer (30) being configured for passing a write current (31), the write current (31) being adapted to switch the second magnetization (230) by spin-orbit torque (SOT) interaction; and an interconnect layer (50) configured for supplying a write current (31) to the write current layer (30); wherein the write current layer (30) has substantially the same lateral dimensions as the lateral dimensions of the MTJ (20) in the layer Plane (PL) such that the MTJ (20) and the write current layer (30) form a pillar (40); and wherein a gap (34) is provided in the interconnect layer (50) such that the interconnect layer (50) comprises two discontinuous interconnect segments (51) extending along the layer Plane (PL) and connecting the write current layer (30) in series;

the method comprises the following steps:

depositing an interconnect layer (50), a write current layer (30), a second ferromagnetic layer (23), a tunnel barrier layer (22), and a first ferromagnetic layer (21);

characterized in that the method further comprises the steps of:

forming a gap (34) in the interconnect layer (50);

filling the gap with a gap material;

the pillars (40) are formed by performing a single etching step until an interconnect layer (50) acting as a stop layer is reached.

2. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

wherein filling the gap comprises depositing a gap layer comprising a gap material on top of the interconnect layer (50).

3. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

further comprising planarizing the gap layer to release an upper surface of the interconnect layer (50).

4. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

wherein the gap (34) is formed by using a photolithography and etching step.

5. The method of claim 4, wherein the first and second light sources are selected from the group consisting of,

wherein the etching step is performed using a multi-angle etch.

6. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

wherein the gap (34) has a lateral dimension (D) in the write current layer (30)₃₀) 0.9 and 0.1 (W)_G）。

7. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

wherein the struts (40) have a geometrically isotropic shape.

8. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

wherein the pillars (40) have a geometrically anisotropic shape.

9. The method of claim 8, wherein the first and second light sources are selected from the group consisting of,

wherein the long axis (41) of the strut (40) is parallel to the gap width (W)_G) Perpendicular to the gap width (W)_G) Or relative to the gap width (W)_G) Oriented at an angle (θ).

10. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

wherein the gap (34) comprises a magnetic material having a bias magnetization (60), the bias magnetization (60) being configured to provide a magnetic bias field (61) adapted to interact with the second magnetization (230) to provide a deterministic switching of the second magnetization (230) when the write current (31) is passed.

11. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

wherein the interconnect layer (50) comprises a ferromagnetic material having a bias magnetization (60), the bias magnetization (60) providing a magnetic bias field (61) adapted to interact with the second magnetization (230) so as to provide a deterministic switching of the second magnetization (230) when the write current (31) is delivered.

12. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

wherein the interconnect layer (50) comprises a conductive material such as Cu, W, Au, Ag, Fe, Pt, Al, Co, Ru, Mo, NiSi, carbon nanotubes, graphene or alloys of these elements.

Technical Field

The present invention relates to a magnetoresistive element, and more particularly, to a magnetoresistive element having increased writing current efficiency and being simpler to manufacture. A method for manufacturing a magnetoresistive element is also disclosed. The invention further relates to a magnetic memory comprising a plurality of magnetoresistive elements.

Prior Art

Fig. 1 shows a side view of a conventional magnetoresistive memory cell comprising a magnetic tunnel junction MTJ 20, the magnetic tunnel junction MTJ 20 comprising a tunnel barrier layer 22 sandwiched between a first ferromagnetic layer 21 having a first magnetization 210 and a second ferromagnetic layer 23 having a free second magnetization 230. Layer 24 may comprise a conductive capping or antiferromagnetic layer exchange coupled with first ferromagnetic layer 21 to pin first magnetization 210. Alternatively, the capping layer 24 may include an SAF structure including a metallic nonmagnetic spacer layer and a ferromagnetic layer. The magnetoresistive memory cell further comprises a current line 30, which current line 30 extends in a plane substantially parallel to the layers 21, 22, 23 of the MTJ 20 and contacts the second ferromagnetic layer 23. By passing a write current 31 in the current line 30, the second magnetization 230 can be switched (magnetization reversal). The switching of the second magnetization 230 may be performed by Spin Orbit Torque (SOT) interaction. The spin current generated by the spin hall effect and/or the Rashba-eldstein (Rashba-Edelstein) effect by the write current 31 exerts a torque in the initial orientation of the second magnetization 230 such that the orientation of the second magnetization 230 can be changed, for example, from being parallel to the first magnetization 210 to being anti-parallel to the first magnetization 210

Switching based on SOT allows the use of separate read and write paths, which is beneficial for the long life of the memory cell. In case the magnetizations 210 and 230 are perpendicular to the plane of the layers 21, 23 (as shown in fig. 1), in addition to the write current 31 an in-plane magnetic field (not shown) has to be applied in order to switch the second magnetization 230 deterministically.

Fig. 2 shows a top view of the conventional cell of fig. 1. Current line 30 extends in the "x-y" plane around MTJ 20. As represented in FIG. 2, the write current 31 flowing in the current line 30 does not all flow under the MTJ 20. The effective current of write current 31 for switching second magnetization 230 increases because write current 31 is evenly distributed across the width of current line 30 and is used for switching only by a portion of write current 31 under MTJ 20. It is difficult to fabricate a conventional cell in which current line 30 has a width similar to the lateral dimensions of MTJ 20. One reason is that it is complicated to achieve alignment of the MTJ 20 with the width of the current line 30.

On the other hand, the conventional magnetoresistive memory cell shown in fig. 1 and 2 is typically fabricated by etching the MTJ 20 until the current line 30 is reached. Here, the current line 30 functions as an etching stopper. Since the current line 30 is typically very thin, between 1 nm and 10 nm, the etching process needs to be very precise.

Document US2019326353 discloses an SOT memory device comprising an SOT electrode at the upper end of the MTJ device. The MTJ device includes a free magnet, a fixed magnet, and a tunnel barrier between the free magnet and the fixed magnet, and is coupled with a conductive interconnect at a lower end of the MTJ device. The SOT electrode has a footprint substantially the same as the footprint of the MTJ device. The SOT device includes a first contact and a second contact on an upper surface of an SOT electrode. The first contact and the second contact are laterally spaced apart by a distance no greater than a length of the MTJ device.

Disclosure of Invention

The present disclosure relates to a method for manufacturing a magnetoresistive element, comprising: an MTJ comprising a tunnel barrier layer sandwiched between a first ferromagnetic layer having a first magnetization, a second ferromagnetic layer having a second magnetization, and a write current layer in contact with the second ferromagnetic layer and configured to pass a write current adapted to switch the second magnetization through SOT interaction; an interconnect layer contacting the write current layer and configured to supply a write current to the write current layer; the interconnect layer includes a gap configured such that the interconnect layer includes two discontinuous interconnect segments extending along a layer plane substantially parallel to the MTJ layers, the interconnect segments connecting the write current layers in series. The method comprises the following steps: depositing an interconnect layer, a write current layer, a second ferromagnetic layer, a tunnel barrier layer, and a first ferromagnetic layer; forming a gap in the interconnect layer; filling the gap with a gap material; and forming the pillars by performing a single etching step until reaching the interconnect layer serving as a stop layer.

The present disclosure relates to a magnetoresistive element obtained by the method.

As is known in the art, a magnetoresistive element provides increased write current efficiency, since the write current for switching the second magnetization by SOT interaction flows only below the second ferromagnetic layer. Because a portion of the write current layer not under the MTJ is substantially free from shunting the write current, the write current can be reduced. Furthermore, the gaps between the interconnect segments force the write current to flow entirely in the write current layer.

Due to the larger size of the interconnect layers relative to the size of the MTJ, the dimensional constraints along the gap width can be relaxed. The MTJ only needs to be precisely positioned in a direction perpendicular to the gap width. Depending on the shape of the MTJ, the constraint along the direction perpendicular to the gap width may be relaxed.

During the fabrication process, the MTJ may be etched with the write current layer until the interconnect layer is reached. The interconnect layer is thus used as an etch stop layer. Since the interconnect layer may be much thicker than the thin write current layer, the control of the etching process may be simplified and variations between differently manufactured magneto-resistive elements may be reduced. The magneto-resistive element is then easier to manufacture. Furthermore, oxidation or degradation of the write current layer during the next process step can be avoided.

Another advantage of the magnetoresistive element disclosed herein is that the interconnect layer may have a lower resistance than the resistance of the write current layer, such that less heating occurs during write/read operations. The lower resistance along the current path allows to reduce the total energy required for applying a sufficiently high current to switch the second magnetization.

Yet another advantage is that the thickness of the write current layer can be reduced compared to the thickness of the write current layer of a conventional magnetoresistive element. In fact, the thickness of the current layer does not need to be controlled for the etching process and can be optimized for the delivered write current.