阅读说明：本技术 用于垂直磁化的磁性穿隧结的低电阻MgO盖层 (Low resistance MgO capping layer for perpendicular magnetized magnetic tunneling junctions ) 是由 沙希·帕特尔 真杰诺 童儒颖 维格纳许·桑达 沈冬娜 王郁仁 王伯刚 刘焕龙 于 2018-12-07 设计创作，主要内容包括：本发明实施例公开了一种磁性穿隧结(magnetic tunnel junction,MTJ),其中自由层(free layer,14,FL)与第一金属氧化(metal oxide,17a,MO<Sub>x</Sub>)层及第二金属氧化物(13,穿隧障壁)形成交界以产生垂直磁各向异性(perpendicular magnetic anisotropy,PMA)于FL中。在一些实施例中,贵金属制的导电金属通道(15c)形成于MgO的MO<Sub>x</Sub>中以减少寄生电容。在第二实施例中,具有多个岛状物的不连续MgO层形成为MO<Sub>x</Sub>层且沉积非磁性硬掩模层以填充相邻的岛状物之间的空间且形成穿过MO<Sub>x</Sub>的短路路径。在另一个实施例中,通过沉积还原金属层于MO<Sub>x</Sub>侧壁上,或用组成气体、H<Sub>2</Sub>、或还原性物种进行还原工艺,将中央MO<Sub>x</Sub>部分的侧边及MTJ侧壁之间的端部还原以形成短路路径。(The embodiment of the invention discloses a Magnetic Tunnel Junction (MTJ), wherein a free layer (14, FL) and a first metal oxide (17 a, MO) x ) The layer and the second metal oxide (13, tunnel barrier) form an interface to create Perpendicular Magnetic Anisotropy (PMA) in the FL. In some embodiments, a conductive metal channel (15c) made of noble metal is formed in the MO of MgO x To reduce parasitic capacitance. In the second embodiment, a discontinuous MgO layer having a plurality of islands is formed as MO x Layers and depositing a non-magnetic hard mask layer to fill spaces between adjacent islands and form through MO x The short-circuit path of (2). In another embodiment, the metal layer is reduced by deposition on the MO x On the side wall, or with a forming gas, H 2 Or the reducing species is subjected to a reduction process to convert the central MO x The ends between the sides of the portion and the MTJ sidewalls are reduced to form a short circuit path.)

1. A magnetic element in a memory device, comprising:

(a) a tunnel barrier layer, which is a first metal oxide layer formed between a reference layer and a free layer;

(b) a second metal oxide layer comprising a metal or alloy M1 and contacting a second surface of the free layer opposite a surface of the first free layer contacting the tunnel barrier layer, the second metal oxide layer having a plurality of conductive paths extending from a top surface to a bottom surface of the second metal oxide layer, thereby substantially reducing the resistivity of the second metal oxide layer; and

(c) the Free Layer (FL), wherein contact of the first and second free layer surfaces with the tunnel barrier layer and the second metal oxide layer, respectively, produces an interfacial perpendicular anisotropy resulting in a perpendicular magnetic anisotropy in the free layer.

2. The magnetic element of claim 1 wherein the metal or alloy M1 is one of Mg, MgAl, MgTi, MgGa, Ti, AlTi, MgZn, MgZr, Al, Zn, Zr, Hf, SrTi, BaTi, CaTi, LaAl, V, or MgTa.

3. The magnetic element of claim 1 wherein the conductive paths include the metal or alloy M1.

4. The magnetic element of claim 1 wherein the conductive paths are formed from one or more of Ag, Au, Pt, Au, Pd, Ru, Rh, Ir, Mo, Fe, and Os.

5. The magnetic element of claim 1 further comprising an uppermost layer that is a hard mask formed on the second metal oxide layer and a seed layer having a bottom surface that contacts a bottom electrode and a top surface adjacent the reference layer.

6. The magnetic element of claim 1 further comprising an uppermost layer that is a hard mask formed on the reference layer and a seed layer having a bottom surface contacting a bottom electrode and a top surface adjacent the second metal oxide layer.

7. The magnetic element of claim 1 wherein the memory device is a Magnetic Random Access Memory (MRAM), a spin torque transfer-magnetic random access memory (STT) -MRAM, a sensor, a biosensor, or a spin torque transfer oscillator.

8. A magnetic element in a memory device, comprising:

(a) a tunnel barrier layer, which is a first metal oxide layer formed on a reference layer;

(b) a free layer having a top surface and formed on the tunnel barrier layer;

(c) a discontinuous second metal oxide layer having a plurality of islands on the top surface of the free layer, and a plurality of openings between adjacent islands; and

(d) a non-magnetic hard mask layer formed on the discontinuous second metal oxide layer and on the parts of the top surface of the free layer exposed by the openings.

9. The magnetic element of claim 8 wherein the second metal oxide layer comprises MgO, AlO_x、MgAlO_x、MgTiO_x、TiO_x、AlTiO_x、MgZnO、ZnO、ZrO_x、HfO_xOr MgTaO.

10. The magnetic element of claim 1 wherein the hard mask layer is formed of Ta, Ru, TaN, Ti, TiN, and W, or a conductive oxide which is RuO_x、ReO_x、IrO_x、MnO_x、MoO_x、TiO_xAnd FeO_xOne or more of (a).

11. A magnetic element in a memory device, comprising:

(a) a tunnel barrier layer that is a first metal oxide layer formed between a reference layer and a first surface of a free layer.

(b) A cap layer formed on a second surface of the free layer, comprising:

(1) a central portion having a second metal oxide layer in a first oxidation state and comprising a metal or alloy M1, the second metal oxide layer having a first thickness and two sides vertically aligned to the second surface; and

(2) two conductive end portions having the first thickness adjacent the two sides of the central portion, and wherein the two conductive end portions comprise the metal or alloy M1 or the metal or alloy M1 having an oxidation state substantially lower than the first oxidation state, thereby forming an electrical short path having a first width between each side of the central portion and a sidewall of the magnetic element; and

(c) the free layer, wherein contact of the first and second free layer surfaces with the tunnel barrier layer and the second metal oxide layer, respectively, produces an interfacial perpendicular anisotropy resulting in a perpendicular magnetic anisotropy in the free layer.

12. The magnetic element of claim 11 wherein the metal or alloy M1 is one of Mg, MgAl, MgTi, MgGa, Ti, AlTi, MgZn, MgZr, Al, Zn, Zr, Hf, SrTi, BaTi, CaTi, LaAl, V, or MgTa.

13. The magnetic element of claim 11 further comprising an uppermost layer which is a hard mask formed on the second metal oxide layer and the two conductive ends, and a seed layer having a bottom surface contacting a bottom electrode and a top surface adjacent the reference layer.

14. The magnetic element of claim 11 further comprising an uppermost layer that is a hard mask formed on the reference layer and a seed layer having a bottom surface contacting a bottom electrode and a top surface adjacent the second metal oxide layer and two conductive ends.

15. The magnetic element of claim 11 wherein the second metal oxide layer further comprises a plurality of conductive paths extending from a top surface to a bottom surface of the second metal oxide layer.

16. The magnetic element of claim 15 wherein the conductive paths are formed from a noble metal selected from the group consisting of Ag, Au, Pt, Au, Pd, Ru, Rh, Ir, Mo, Fe, and Os, the noble metal being different from the metal or alloy M1.

17. A method of forming a magnetic element, comprising:

(a) providing a free layer having a top surface and a bottom surface contacting a tunnel barrier layer;

(b) forming a metal oxide layer comprising a metal or alloy (M1) on the top surface of the free layer, the metal oxide layer having a plurality of conductive channels extending from a top surface to a bottom surface thereof to allow a plurality of conductive paths from the free layer to the metal oxide layer, and wherein perpendicular magnetic anisotropy in the free layer is established through contact with the tunnel barrier layer and the metal oxide layer; and

(c) depositing a conductive non-magnetic layer on the metal oxide layer, wherein the conductive non-magnetic layer is a hard mask layer in the magnetic element and contacts the conductive channels.

18. The method of claim 17, wherein the metal oxide layer is formed by first depositing a metal or alloy (M1) on the top surface of the free layer and then oxidizing such that a plurality of metal or alloy (M1) atoms are not oxidized and form the conductive vias.

19. The method of claim 17, wherein the metal oxide layer and the plurality of conductive vias are formed by a process comprising:

(a) depositing a first metal or alloy (M1) layer on the top surface of the free layer;

(b) depositing a non-magnetic (NM) layer on the first M1 layer such that some M1 atoms are re-sputtered to form a first sublayer of M1 and NM atoms on the free layer and a second sublayer including M1 and NM atoms on the first sublayer;

(c) depositing a second M1 layer on the second sublayer such that some of the NM atoms in the second sublayer are resputtered to form a third sublayer having M1 and NM atoms on the second sublayer; and

(d) an oxidation is performed such that a plurality of M1 atoms in the first, second, and third sub-layers are oxidized to form the metal oxide layer, and the NM atoms are not oxidized, thereby forming the conductive vias.

20. The method of claim 17, wherein the metal or alloy M1 is one of MgAl, MgTi, MgGa, Ti, AlTi, MgZn, Al, Zn, Zr, Hf, SrTi, BaTi, CaTi, LaAl, V, or MgTa, and the conductive vias comprise the metal or alloy M1.

21. The method of claim 19, wherein the metal or alloy M1 is one of Mg, Ti, AlTi, MgZn, Al, Zn, Zr, Hf, or MgTa, and the NM atoms in the conductive channels comprise one or more of Ag, Au, Pt, Au, Pd, Ru, Rh, Ir, Mo, Fe, and Os.

22. The method of claim 19, wherein the oxidizing is performed in a natural oxidation process of a reaction chamber and comprises an oxygen flow rate for a time period of 5 seconds to 5 minutes.

23. A method of forming a magnetic element, comprising:

(a) providing a free layer having a top surface and a bottom surface contacting a tunnel barrier layer;

(b) depositing a discontinuous metal or alloy (M1) layer comprising a plurality of islands contacting a majority of the top surface of the free layer;

(c) oxidizing the metal or alloy (M1) layer to a metal oxide layer having a stoichiometric oxidation state or a non-stoichiometric oxidation state, the islands being separated by openings that expose a portion of the top surface of the free layer; and

(d) depositing a conductive non-magnetic layer on the metal oxide layer and on the exposed portion of the top surface of the free layer, the conductive non-magnetic layer forming a plurality of conductive vias through the metal oxide layer, and the metal oxide layer generating perpendicular magnetic anisotropy in the free layer.

24. The method of claim 22 wherein the metal or alloy M1 is one of MgAl, MgTi, MgGa, Ti, AlTi, MgZn, Al, Zn, Zr, Hf, SrTi, BaTi, CaTi, LaAl, V, or MgTa.

25. The method of claim 23 wherein the conductive non-magnetic layer is formed of Ta, Ru, TaN, Ti, TiN, and W, or a conductive oxide, the conductive oxide being RuO_x、ReO_x、IrO_x、MnO_x、MoO_x、TiO_xAnd FeO_xOne or more of (a).

26. The method of claim 21 wherein the discontinuous metal or alloy (M1) layer is deposited by a physical vapor deposition process that includes a temperature between room temperature and 350 ℃.

27. A method of forming a magnetic element, comprising:

(a) sequentially forming a reference layer, a tunnel barrier layer, a free layer, a cap layer and a hard mask layer on a substrate, wherein the cap layer is a metal oxide layer;

(b) performing a first etching process to form a sidewall on the hard mask layer and the cap layer having a first oxidation state;

(c) applying a reducing agent such that ends of the cap layer are reduced to a metal or alloy, or a non-stoichiometric oxidation state having an oxygen content substantially lower than the first oxidation state in a central metal oxide portion of the cap layer; and

(d) a second etch process is performed to create a sidewall over the free layer, the tunnel barrier layer, and the reference layer.

28. The method of claim 27 wherein the metal oxide layer comprises MgO, AlO_x、MgTiO_x、MgAlO_x、TiO_x、AlTiO_x、MgZnO、AlO_x、ZnO、ZrO_x、HfO_xOr MgTaO.

29. The method of claim 27, wherein the applying the reducing agent comprises performing a reduction process in which a forming gas, hydrogen gas, or a reactive species is applied to the sidewall of the cap layer, the reactive species being hydrogen radicals (H), H⁺、OH^-CO, C, or CH₃Or a C-H radical (CH. cndot., CH 2. cndot., CH 3. cndot., CH 4. cndot.), or a hydroxyl radical (OH. cndot.).

30. The method of claim 27, wherein the applying the reducing agent comprises depositing a reduced metal layer on the hard mask and the sidewalls of the hard mask and cap layer, the reduced metal layer absorbing oxygen from the ends of the cap layer.

31. The method of claim 29 wherein the second etch forms a sidewall on the free layer, the tunnel barrier layer, and the reference layer, the sidewall forming a continuous surface with sidewalls on the hard mask and the cap layer.

32. The method of claim 30 wherein the second etch forms a sidewall on the free layer, the tunnel barrier layer, and the reference layer, the sidewall and the sidewall on the reduced metal layer forming a continuous surface.

Technical Field

The present disclosure relates to Magnetic Tunnel Junctions (MTJs) that include a free layer that interfaces with a tunnel barrier layer and a metal oxide cap layer, and in particular to reducing the observed parasitic resistance by a cap layer having a fully oxidized state, while reducing the cap layer resistance to mitigate the reduction in magnetoresistance ratio.

Background

Perpendicular magnetization magnetic tunneling junctions (p-MTJs) are a significant emerging technology for embedded Magnetic Random Access Memory (MRAM) applications, as well as stand-alone MRAM applications. Perpendicular magnetization magnetic tunneling magnetic random access memory (STT-MRAM) technology used to write memory bits is described by j.c. slonczewski in "Current drive excitation of magnetic multilayers", j.magn.magn.mater.v. 159, L1-L7(1996) and is highly competitive with existing semiconductor memory technologies such as static random access memory, dynamic random access memory, and flash memory.

Magnetic random access memory and spin torque transfer-magnetic random access memory have MTJ elements based on Tunneling Magnetoresistive (TMR) effect, in which the MTJ film stack has a configuration in which two ferromagnetic layers are separated by a thin insulating tunnel barrier layer. When the planes of each film are arranged in the x-axis and y-axis directions, one of the ferromagnetic layers, referred to as the pinned layer, has a magnetic moment pinned in an out-of-plane direction (e.g., + z direction). The second ferromagnetic layer has an out-of-plane magnetization direction that can be freely rotated to either + z (parallel or P-state) or-z (non-parallel or AP-state). The difference between the P-state resistance (Rp) and the AP-state resistance (Rap) is analyzed by the equation (Rap-Rp)/Rp, (Rap-Rp)/Rp is also referred to as DRR. For MTJ devices, it is important to have a large DRR value, preferably above 1, because DRR is directly related to the read margin of a memory bit or how easily it is to distinguish between the P and AP states (0 or 1 bit).

When the free layer has a magnetization direction perpendicular to the plane of the film, the critical current (I) required for switching the magnetic element_c) Proportional to the perpendicular anisotropy field shown in formula (1), where e of formula (1) is the electron charge, α is the Gibbe damping constant, Ms is the saturation magnetization of the free layer, h is the approximate Planck constant, g is the gyromagnetic ratio,is the out-of-plane anisotropy field of the magnetic region to be switched, andv is the volume of the free layer:

Δ＝kV/k_Bthe value of T is a measure of the thermal stability of the magnetic element, where kV is also referred to as E_bOr energy barrier, k, between two magnetic states (P and AP)_BIs the bauzmann constant and T is temperature. For functional MRAM products, the free layer (information storage layer) must have a sufficiently high E_bTo resist switching caused by thermal or magnetic environmental perturbations. The energy barrier for this random switching is related to the strength of the Perpendicular Magnetic Anisotropy (PMA) of the free layer. One practical way to achieve strong PMA is through the interface PMA at the interface between the CoFeB free layer and the MgO tunnel barrier. Higher PMA is achieved by forming a second MgO interface on the opposite side of the free layer relative to the tunnel barrier for additional interfacial PMA. Thus, the overall PMA in the free layer is to utilize the MgO/CoFeB/MgO stack in the p-MTJ to increase the E_bTo optimize.

Fig. 1 illustrates a conventional p-MTJ1 as an example, wherein an optional seed layer 11, a pinned layer 12, a tunneling barrier 13, a free layer 14, a metal oxide cap 17, and a hard mask 16 are sequentially formed on a substrate, which is a bottom electrode 10 in an MRAM architecture. Unfortunately, the consequence of using the MgO capping layer is the addition of parasitic resistance to the p-MTJ device. Equation (2) shows the effect of the cap layer resistance contributing to the total MTJ resistance, and equation (3) shows the effect on DRR.

WhereinAnd is

Because of the fact that

In summary, the series resistance caused by the metal oxide cap layer (And) This results in a reduced DRR, which effectively reduces the read margin of the MRAM bit, and also increases the write voltage of the bit by adding a series resistor. Due to the need for MgO cap layers or the like to achieve strong PMA for enhanced thermal stability, improved p-MTJ structures are needed such that the series resistance contributing to the self-cap layer is significantly reduced while maintaining strong PMA.

Disclosure of Invention

It is an object of the present disclosure to reduce the product of the cap layer resistance and the resistance x area (RA) of the p-MTJ nanopillar, where the metal oxide layer forms an interface with the top and bottom surfaces of the free layer, thereby boosting DRR and maintaining a high PMA for thermal stability while reducing the write voltage.

A second object is to provide a method of forming the p-MTJ of the first object, and the above method is easily implemented in the fabrication of a memory device.

There are several ways in which this can be achieved in accordance with the present disclosure. In the first embodiment in which the free layer is formed between the tunnel barrier layer and the metal oxide cap layer, the metal oxide cap layer preferably has a large number of unoxidized metal atoms such that a metal (conductive) channel is between the top and bottom surfaces of the metal oxide layer to reduce the resistance therein. Thus, for example, unoxidized Mg paths may be formed in the MgO layer. In alternative embodiments, a Noble Metal (NM) may be co-deposited with Mg followed by oxidation of Mg, or a Mg/NM/Mg stack may be formed prior to oxidation, or a process including deposition and oxidation of Mg, redeposition of NM, and redeposition and oxidation of Mg may be used to create a NM conductive path in the MgO layer.

According to a second embodiment, a metal, for example Mg, is formed as a discontinuous layer in the form of islands on the free layer by Physical Vapor Deposition (PVD). Thereafter, an oxidation step is performed to convert the irregular islands into a discontinuous MgO layer. Next, a non-magnetic metal hard mask layer is deposited to form a metal path through the discontinuous MgO layer, thereby effectively reducing the resistive contribution of the MgO cap layer to the total RA product of the p-MTJ.

According to a third embodiment, the outer portion of the continuous MgO capping layer formed proximate to the MTJ nanocolumn sidewalls is reduced to provide a conductive path around the central portion of the MgO layer. The metal oxide can be externally etched by applying a forming gas, H, during the p-MTJ etching process₂Or a reducing agent species. In some embodiments, the MgO layer is reduced to an unoxidized state outside of the p-MTJ sidewalls by contact with a reducing metal in a subsequently deposited encapsulation layer. For example, oxygen in the MgO layer may diffuse into the encapsulation layer, thereby reducing the oxygen content in the exterior of the MgO layer in the p-MTJ.

The present disclosure also includes methods of forming a metal oxide layer having a lower resistivity, as can be seen in the above embodiments. For example, the outer portion of the MgO cap layer in the third embodiment may be partially reduced to a non-stoichiometric oxidation state or fully reduced to Mg by breaking the etch of the MTJ after the sidewalls are formed along the cap layer. Then, a forming gas treatment, H, is performed before continuing the p-MTJ etch to form sidewalls along the free layer, the tunnel barrier layer, and the underlying film layer₂Treating, exposing to a reducing species, or depositing a reducing metal layer. As a result, the MgO tunnel barrier layer maintains a substantially stoichiometric oxidation state to optimize DRR.

In other embodiments of the present disclosure relating to top spin valve configurations of p-MTJs, a free layer is formed between an underlying metal oxide layer, referred to as a k enhancing layer, and an overlying tunnel barrier layer. As in the previous embodiments of the bottom spin valve configuration, the metal oxide layer has a conductive via extending between its top and bottom surfaces to provide an electrical short path as a means of reducing parasitic resistance.

Drawings

FIG. 1 is a cross-sectional view showing a perpendicular magnetization magnetic tunneling junction (p-MTJ) according to the prior art, where a free layer is formed between two uniform metal oxide layers.

FIG. 2 is a cross-sectional view of a p-MTJ according to a first embodiment of the disclosure where a conductive channel is formed in the metal oxide layer and contacts the top surface of the free layer.

Fig. 3a depicts the sputter deposition of a noble metal on a metal or metal oxide layer, and fig. 3b depicts the sputter deposition of a metal or metal oxide layer on the film stack from fig. 3 a.

FIG. 4 is a cross-sectional view of a p-MTJ according to a second embodiment of the present disclosure where a discontinuous layer of metal oxide islands is formed on the free layer to reduce the RA product in the p-MTJ nanocolumns.

Fig. 5 is a top view of the metal islands in fig. 4 and shows the spaces between the islands filled with a subsequently deposited metal layer.

FIG. 6 is a cross-sectional view of a p-MTJ and depicts a central portion of a metal oxide layer having a substantially stoichiometric oxygen content and an outer portion having a substantially less than stoichiometric oxygen content in accordance with a third embodiment of the present disclosure.

Fig. 7-12 are cross-sectional views showing a process of forming a p-MTJ depicted in the bottomed spin valve configuration of fig. 6, in accordance with an embodiment of the present disclosure.

Fig. 13-15 are cross-sectional views depicting alternative methods of forming the p-MTJ structure in fig. 6.

FIG. 16 is a cross-sectional view of a p-MTJ with a top spin valve configuration, representing a modification to the p-MTJ in FIG. 2 according to an embodiment of the present disclosure.

FIG. 17 is a cross-sectional view of a p-MTJ with a top spin valve configuration, representing a modification to the p-MTJ in FIG. 6 according to an embodiment of the present disclosure.

Fig. 18 is a cross-sectional view depicting an unpatterned p-MTJ film stack in which a free layer is formed between a tunnel barrier layer and a metal oxide layer having a plurality of conductive vias therebetween.

FIG. 19 is a cross-sectional view of the p-MTJ stack of FIG. 18 after using a process flow according to the present disclosure to produce a p-MTJ nanorod, where the metal oxide layer has a center portion with a conductive via and two end portions with a substantially reduced oxidation state.

Fig. 20 is a table listing the free energies of various elements for forming oxides.

Detailed Description

The present disclosure relates to p-MTJ nano-pillars and their fabrication, in which a free layer forms a first interface with a tunnel barrier layer and a second interface with a metal oxide layer, and a conductive channel is formed in or around the metal oxide layer to mitigate parasitic resistance and reduce DRR associated with a uniform metal oxide layer that does not have tunnel magnetoresistance. While the illustrative embodiments depict p-MTJ nano-pillars with a bottom spin valve configuration and a top spin valve configuration, the present disclosure also encompasses p-MTJs with dual spin valve structures as understood by those skilled in the art. The p-MTJ may be incorporated into MRAM, STT-MRAM, sensors, biosensors, or other spintronic devices, such as Spin Torque Oscillators (STOs). Only one p-MTJ nano-pillar is depicted in the drawing, but typically millions of p-MTJ nano-pillars are formed in an array of rows and columns on a substrate when manufacturing a memory device. The "oxidation state" and "oxygen content" may alternatively be used in describing the condition of the metal oxide layer. The conductive channel may have a width of a plurality of metal atoms in a direction parallel to a plane of the metal oxide layer. The top surface of the film is defined as the surface facing away from the substrate, while the bottom surface faces the substrate. The interface is a boundary region comprising a bottom surface of one layer and a top surface of an adjacent second layer.

In related us patent 9,006,704, we disclose a dielectric constant promoting layer having a metal oxide composition that promotes Perpendicular Magnetic Anisotropy (PMA) in the adjacent free layer, with the metal oxide embedded with conductive particles of Co, Fe, Ni, etc. to reduce resistivity in the metal oxide. In addition, in related U.S. patent 9,230,571, we disclose an STO structure in which the cap layer has a Closed Current Path (CCP) configuration that includes a metal channel in a metal oxide matrix. Typically, the metal channels are formed of Cu, which is a different metal than the metal in the metal oxide matrix.

Additional approaches have been found to reduce the parasitic resistance of the metal oxide layer that interfaces with the free layer and is formed on the opposite side of the free layer relative to the tunnel barrier layer. All embodiments described herein include forming one or more conductive paths through or around a metal oxide layer that abuts a top or bottom surface of the free layer. While each embodiment contains key features to allow for lower resistivity, the present disclosure contemplates that concepts from one aspect may be incorporated into another aspect to further promote the reduction in resistivity through a metal oxide layer as described below.

As mentioned above, the p-MTJ nano-pillar 1 is manufactured by the inventors according to a process of record (POR) scheme and includes a film stack in which the free layer 14 is sandwiched between the tunnel barrier layer 13 and the metal oxide cap layer 17 to promote PMA and thermal stability in the free layer. However, there is a need for improved p-MTJ nano-pillars to reduce the capping layer resistanceThe impact of lower DRR due to the large contribution to the sum in the denominator on the right side of equation (3).

According to the first embodiment of the present disclosure as shown in p-MTJ2 in fig. 2, we find that the resistivity in metal oxide layer 17a is significantly reduced by forming metal via 15c or a conductive path in metal oxide layer 17 a. P-MTJ2 has optional seed layer 11 formed on bottom electrode 10, and reference layer 12, tunneling barrier 13, free layer 14, metal oxide layer 17a, and hard mask 16 sequentially formed on the seed layer. The seed layer comprises one or more of NiCr, Ta, Ru, Ti, TaN, Cu, Mg, or other materials commonly used to promote a smooth and uniform grain structure in the upper coating layer.

The reference layer 12 may have a synthetic anti-parallel (SyAP) configuration, represented by AP2/Ru/AP1, in which the antiferromagnetic coupling formed by, for example, Ru, Rh, or IrThe layer is sandwiched between an AP2 magnetic layer and an AP1 magnetic layer (not shown). An AP2 layer, also referred to as an outer pinned layer, is formed on the seed layer, while AP1 is an inner pinned layer and typically contacts the tunneling barrier. The AP1 and AP2 layers may comprise CoFe, CoFeB, Co, or combinations thereof. In other embodiments, the reference layer may be a build-up stack with an intrinsic PMA, e.g. (Co/Ni)_n、(CoFe/Ni)_n、(Co/NiFe)_n、(Co/Pt)_n、(Co/Pd)_nEtc., where n is the number of layers. In addition, a transition layer, such as CoFeB or Co, may be interposed between the uppermost layer of the build-up stack and the tunnel barrier layer 13.

The tunnel barrier layer 13 is preferably a metal oxide, which is MgO, TiO_x、AlTiO、MgZnO、Al₂O₃、ZnO、ZrO_x、MgAlO_x、MgGaO_x、HfO_xOr MgTaO, or a stack of one or more of the foregoing metal oxides. More preferably, MgO is selected as the tunnel barrier because MgO provides the highest magnetoresistance ratio (DRR), especially when sandwiched between, for example, two CoFeB layers.

The free layer 14 may be Co, Fe, CoFe, or alloys thereof, with one or both of B and Ni, or a multi-layer stack comprising a combination of the foregoing. In another embodiment, the free layer may have a non-moment diluting layer, such as Ta or Mg, interposed between two ferromagnetically coupled CoFe or CoFeB layers. In an alternative embodiment, the free layer has a SyAP configuration, such as FL1/Ru/FL2, where FL1 and FL2 are antiferromagnetically coupled two magnetic layers, or a lamination stack with intrinsic PMA as described above with respect to the reference layer composition.

The hard mask 16 is non-magnetic and typically includes one or more conductive metals or alloys, including but not limited to Ta, Ru, TaN, Ti, TiN, and W. It should be understood that other hard mask materials may be selected, including MnPt, thereby providing high etch selectivity with respect to the underlying MTJ film layer during the etch process that forms the MTJ nano-pillar stopping at the sidewall of the bottom electrode. In other embodiments, the hard mask is a conductive nonmagnetic layer that is RuO_x、ReO_x、IrO_x、MnO_x、MoO_x、TiO_xAnd FeO_xOne or more of (a).

Preferably, the metal oxide layer 17a has a material selected from MgO, AlO_x、TiO_x、MgTiO_x、AlTiO、MgZnO、MgAlO_x、ZnO、ZrO_x、MgZrO_x、HfO_x、SrTiO_x、BaTiO_x、CaTiO_x、LaAlO_x、VO_xAnd MgTaO, wherein one or more of the metals in the metal oxide is M1 metal or alloy. In some embodiments, the metal via 15c formed therein comprises the same M1 metal or alloy as the metal oxide layer, unlike CCP structures. Therefore, Mg channels are formed in the MgO layer by first depositing a Mg layer on the free layer 14. Subsequently, Natural Oxidation (NOX) is performed, whereby a large amount of Mg atoms remain unoxidized (non-stoichiometric oxidation state) by one or both of reducing the flow rate of oxygen and shortening the reaction time relative to a standard NOX process in which the Mg layer is converted to a MgO layer in a stoichiometric oxidation state (substantially without the remaining unoxidized Mg atoms). It is noted that the channels do not have to have a substantially perpendicular alignment, or uniform width in the x-axis and y-axis directions, but each channel contacts the top surface of the free layer and the bottom surface of hard mask 16 in p-MTJ2 with a bottom spin valve configuration.

The present disclosure also contemplates that metal via 15c may comprise a different metal than in metal oxide layer 17 a. According to another embodiment, in which the metal oxide is chosen to be MgO, a first Mg layer is deposited on the free layer 14, followed by a Noble Metal (NM) layer on the first Mg layer, and then a second Mg layer on the NM layer. All layers are preferably RF or DC sputtered using conventional processes. Next, an oxidation step is performed to oxidize most of the deposited Mg atoms, resulting in a MgO matrix in which NM channels are formed. It is noted that another M1 metal or alloy may replace the Mg in the first a layer of fig. 3a, and may replace the second a layer when deposited in fig. 3 b.

Referring to fig. 3a, a first step of depositing an NM layer on a metal (a) layer is shown. It should be appreciated that when depositing NM atoms on the first a (M1 metal or alloy) layer, a certain number of a atoms are dislocated (resputtered) and co-deposited with the NM atoms. As a result, the bottom layer SP1 may contain multiple NM atoms, and the second layer SP2 may contain multiple a atoms after depositing NM. Thereafter, in fig. 3b, some NM atoms may be displaced from the second layer and co-deposited with a atoms in the third or uppermost layer SP3 when depositing the second a layer. Thus, the stack of NM atoms from the bottom SP1, the second SP2, and the top SP3 may form a first channel 15c1, and the second stack of NM atoms forms a second conductive channel 15c 2. Only two channels are shown to simplify the drawing. However, it is preferred that the sequence of steps be through sputter deposition as described herein to create a plurality of channels greater than two. It is also understood that the width of one or more channels may be determined by the number of NM atoms in one or more of the layers SP1-SP 3.

Referring to fig. 20, Mg or M1 metal or alloy has a larger negative energy to form oxide than noble metals and is selectively oxidized during the oxidation step that follows the deposition sequence described above. Thus, after the oxidation step, NM metal of one or more of Ag, Au, Pt, Au, Pd, Ru, Rh, Ir, Os, Mo, Fe, or the like provides a plurality of conductive channels 15c in the MgO (or M1 metal or alloy) matrix to provide an electrical short path through oxide matrix 17a in p-MTJ 2. Preferably, the vast majority of the metal vias contact the hard mask 16 and the free layer 14. However, the presence of isolated NM atoms surrounded by an oxide matrix also reduces the resistivity within the metal oxide layer.

In some embodiments (not shown), a first Mg (or another M1 metal or alloy) layer is oxidized prior to depositing the NM layer, and then a second Mg (or another M1 metal or alloy) layer is deposited on the NM layer, followed by a second oxidation step. Each step of depositing an NM layer on the MgO layer, and depositing a second Mg layer on the NM layer, may redeposit a portion of the MgO and NM layers, respectively, to create an NM channel in the second oxidized MgO matrix. In other embodiments, NM metal and Mg (or other M1 metal or alloy used to form a metal oxide matrix) are co-sputtered in a sputtering tool from a single target or separate targets before an oxidation step is performed to convert Mg atoms into a MgO matrix with NM channels therein. In yet another embodiment, a first MgO layer (layer A in FIG. 3 a) may be sputtered on the free layer 14, followed by deposition of an NM layer, and a second A (MgO) layer sputtered on the NM layer to form the stack of layers SP1-SP3 in FIG. 3 b. Upon deposition of the second MgO layer, some of the NM atoms in the SP2 layer are effectively resputtered, and then co-sputtered with the second MgO layer to create a MgO matrix (stack of a atoms) with multiple NM channels (including 15c1, 15c2 paths) therein.

Referring to FIG. 4, a second embodiment (p-MTJ3) is depicted in which a conductive path is formed in the metal oxide layer to electrically connect the free layer 14 to the metal layer in the hard mask 16. Here, a discontinuous layer comprising M1 metal or metal alloy is formed on the free layer by, for example, a low pressure PVD process. We previously disclosed in us patent 8,981,505 a PVD method for depositing a discontinuous Mg layer 1 to 3 angstroms thick on a free layer. Here, as previously described in the first embodiment, the M1 metal or alloy is preferably one of Mg, MgAl, MgTi, MgGa, Ti, AlTi, MgZn, MgZr, Al, Zn, Zr, Hf, SrTi, BaTi, CaTi, LaAl, V, or MgTa. Then, a Natural Oxidation (NOX) or the like is performed to oxidize the metal or metal alloy, thereby providing a plurality of metal oxide islands 15i and spaces exposing a portion of the top surface of the free layer 14, as shown in fig. 5. Then, a hard mask of nonmagnetic conductive material is deposited on the metal oxide islands and fills the space, so that the hard mask contacts the exposed surface of the free layer. In some embodiments, the hard mask has multiple film layers (not shown), but at least the bottom most layer in the hard mask stack fills the spaces between adjacent metal oxide islands 15 i.

From the top view of fig. 5 with the hard mask and overlying film layers removed, the metal oxide islands 15i have an irregular shape and tend to cover most of the top surface of the free layer 14. It is noted that PVD deposition of Mg islands, for example, typically involves a temperature in the reaction chamber between room temperature and 400 ℃. In particular, increasing the temperature of the PVD reduces the surface area coverage of Mg islands. However, the upper temperature limit for PVD in this case is typically 400 ℃ in order not to negatively impact the thermal stability of the free layer. We have found that PVD deposition of a discontinuous Mg layer is preferably performed between room temperature and 350 ℃ to optimize adhesion to the free layer and form islands on the top surface of the free layer. In other words, the deposition temperature of Mg can be used to control the surface coverage and adhesion of subsequently formed MgO islands, thereby adjusting the MgO/free layer contact area and the magnitude of the interfacial perpendicular anisotropy at the MgO/free layer interface. Therefore, the series resistance contribution of the cap layer (MgO islands) is directly related to the percentage of the free layer PMA and the top surface of the free layer contacted by the MgO islands and the percentage of the top surface of the free layer contacted by the conductive hard mask 16. The trade-off is that higher metal oxide surface coverage results in larger PMA, but at the expense of higher series resistance. It is understood that the oxidation state of the MgO islands may be reduced to further reduce the series resistance contribution of the metal oxide cap layer 15 i.

The total RA value of p-MTJ3 is determined by the contribution from each metal oxide layer and is given by the formula RA_TOTAL＝(RA_barrier+RA_cap) Is represented by, wherein RA_barrierAnd RA_capIn addition, in, for example, layers 13, 15i, reducing the oxidation state from a stoichiometric MgO to a non-stoichiometric MgO would desirably reduce RA, but would also undesirably reduce the magnitude of the interfacial perpendicular anisotropy at the interface with the free layer, and thus reduce the PMA in the free layer 14_capFor RA_TOTALThe contribution of (c).

In another embodiment depicted in fig. 6, a bottom spin valve configuration of p-MTJ4 is shown, in which an optional seed layer 11, reference layer 12, tunnel barrier layer 13, free layer 14, a capping layer having a metal oxide center portion 17c and conductive end portions 15e, and hard mask 16 are sequentially formed on bottom electrode 10. One key feature is that the metal oxide portion of the cap layer is confined to the center of the cap layer between the sides 17e and is separated from the sidewalls 4s by a conductive end portion formed of the same M1 metal or alloy as the center portion. The M1 metal or alloy is one of Mg, MgAl, MgTi, MgGa, Ti, AlTi, MgZn, MgZr, Al, Zn, Zr, Hf, SrTi, BaTi, CaTi, LaAl, V, or MgTa. The conductive end portions have a width of about 1 to 10 nanometers and a thickness t the same as the central portion. In other words, the present disclosure contemplates that a uniform metal oxide cap layer may be deposited on the free layer, but then reduced to a metal or non-stoichiometric oxidation state with a large number of metal atoms at the conductive end near the sidewall 4 s.

In FIGS. 7-12, a sequence of steps is depicted according to a process flow for forming conductive end portion 15e and central metal oxide portion 17c in the cap layer of p-MTJ 4. Referring to fig. 7, a seed layer 11, a reference layer 12, a tunnel barrier layer 13, and a free layer 14 are sequentially formed on a bottom electrode 10. In a preferred embodiment, an M1 metal or alloy 15, such as Mg, is deposited on the free layer and then subjected to a natural oxidation step 30 in which oxygen is flowed into the reaction chamber for a period of about 5 seconds to 5 minutes.

FIG. 8 shows an intermediate step in the fabrication of p-MTJ4 after the Mg layer has been converted to MgO capping layer 17 by a NOX process, where hard mask 16 is formed on the MgO layer and a photoresist layer is coated and patterned on the hard mask to create photoresist mask layer 40 with sidewalls 40 s. Width d is the critical dimension in the y-axis direction of subsequently formed p-MTJ 4. The longitudinal dimension in the x-axis direction is not shown but may be equal to or greater than d to produce a substantially circular or elliptical shape, respectively, from a top view. A Reactive Ion Etch (RIE) step is used to transfer the photoresist sidewalls through the hard mask and then stop on the top surface 17t of the cap layer. As a result, the portion of sidewall 4s above p-MTJ4 is formed as a continuation of sidewall 40s and the critical dimension d is copied into the hard mask. The photoresist mask layer is then stripped by conventional methods.

According to one embodiment in FIG. 9, a second process is includedA second RIE of gas is used to transfer the pattern in the hard mask 16 through the cap layer 17 and thereby extend the sidewalls 4s from the top surface 16t to the free layer top surface 14 t. The second process gas comprises methanol, CO, and NH₃And one or more of Ar or another inert gas.

Referring to fig. 10, a reduction process 31 with a forming gas or hydrogen gas is used in the reaction chamber to reduce the cap layer near the portion of the side wall 4s, thereby forming the end portion 15e containing Mg. In some embodiments, substantially all of the MgO in the cap layer 17 near the sidewalls of the p-MTJ is reduced to Mg. In other embodiments, a substantial amount of MgO is converted to Mg, such that a plurality of conductive paths are established in the end portions of the cap layer. It is to be noted that depending on the forming gas or H₂By the rate of horizontal diffusion of the capping layer, and the reaction conditions in terms of time and pressure, a lower concentration of Mg atoms can be formed with increasing distance from the side wall 4 s. The present disclosure contemplates that other reducing gases or agents known in the art may be used in place of the forming gas or hydrogen. For example, the plasma may act as a reducing agent in forming the sidewalls 4s along the cap layer 17 and in subsequent etching steps to extend the sidewalls to the substrate top surface 10 t. The plasma contains hydrogen radicals (H), H⁺、OH^-Or hydroxyl radicals (OH.) and are reactive species generated in the reaction chamber during the second RIE step in the presence of MeOH and Ar. Alternatively, the reactive species may comprise a radical of CO, a radical of C, or CH₃Or a C-H radical (CH, CH)₂·、CH₃·、CH₄One or more of (a).

Referring to fig. 11, a third etching step is performed after the end portion 15e is formed in the cap layer, and the third etching step may be RIE with MeOH and Ar. A third etch step transfers a pattern having a critical dimension d through the free layer, the tunnel barrier layer 13, the reference layer 12, the seed layer 11, and stops on the top surface 10t of the bottom electrode 10. As a result, the sidewall 4s is a continuous surface from the hard mask top surface 16t to the bottom electrode top surface 10 t. Although sidewall angle α is shown as vertical in the exemplary embodiment, angle α may be greater than 90 ° in other embodiments where the width of seed layer 11 is greater than the width of hard mask 16. An optional sputter clean step including Ion Beam Etching (IBE) may be utilized to remove any residue on the sidewalls 4s before continuing with the subsequent steps.

Referring to FIG. 12, an encapsulation layer 19 is deposited on bottom electrode top surface 10t and abutting sidewalls 4s to fill the gap between p-MTJ4 and the adjacent p-MTJ. A Chemical Mechanical Polish (CMP) step may then be utilized to form a top surface 19t on the encapsulation layer that is coplanar with the top surface 16t of the hard mask. Thereafter, a top electrode layer including the top electrode 20 is formed on the top surface of the hard mask by conventional methods. It is noted that in embodiments where the ends are in a low oxidation state (under oxidized state) and do not fully convert to a metal or alloy upon early exposure to a reducing agent, particularly when temperatures up to 400 ℃ are applied for annealing purposes during or after the encapsulation process, the encapsulation layer may absorb oxygen from the ends 15e of the cap layer. In some embodiments, the encapsulation layer includes a strongly reducing metal including, but not limited to, Ta, Al, Mg, Ti, Hf, La, Y, Zr, Fe, and B.

According to another embodiment, illustrated in FIGS. 13-15, of forming the end portions 15e on each side of the central metal oxide portion 17c of the cap layer, the reduction process begins with the intermediate p-MTJ structure illustrated in FIG. 9 and includes depositing the metal layer 18 on the top surface 16t and sidewalls 4s after the initial etch step stops on the free layer top surface 14 t. Preferably, the deposition provides a conformal metal layer having a substantially uniform thickness. The metal may be a reducing agent, is conductive, and preferably comprises one of Ta, Al, Mg, Ti, Hf, La, Y, Zr, Fe, and B, and also serves as an effective mask layer in the subsequent etching of the transferred sidewall 4 s' through the underlying film layer.

Referring to fig. 14, RIE with MeOH and Ar, for example, extends sidewalls 4 s' through free layer 14, tunneling barrier 13, reference layer 12, and seed layer 11 and stops on top surface 10 t. The critical dimension d1 is greater than d in the previous embodiments. Since the RIE plasma is generally directed substantially vertically toward the substrate, the exposed top surface 18t1 above top surface 14t is also removed, but a substantial portion of metal layer 18 remains on sidewalls 4s along hard mask 16 and metal oxide cap layer 17. In some embodiments, the metal layer of the portion above the hard mask having the top surface 18t is also removed by the RIE step. The metal layer is used as a hard mask layer and a packaging layer of the metal oxide cover layer.

FIG. 15 depicts p-MTJ4 after forming an encapsulation layer 19 on sidewalls 4 s' to fill the opening between p-MTJ4 and the adjacent p-MTJ, and a CMP process is used to form a top surface 19t on the encapsulation layer that is coplanar with top surface 18t of metal layer 18, and top surface 16t of hard mask 16. The present disclosure contemplates that a solid state reaction occurs between the reduced metal in the metal layer 18 and the metal oxide layer 17, wherein oxygen is absorbed into the metal layer from the metal oxide end, compared to the original oxidation state in the cap layer, thereby forming the end 15e having a substantially reduced oxidation state. The solid state reaction may occur during deposition of the metal layer, as well as at any subsequent process step (including annealing the p-MTJ4 after forming the sidewall 4 s' along the entire film stack), or during formation of the encapsulation layer 19. Annealing can occur at the time of packaging, where the reaction temperature can be near 400 ℃ to accelerate the solid state reaction rate. In embodiments where the metal encapsulation layer 18 does not absorb oxygen from the metal oxide layer 17, the encapsulation layer 18 provides the conductive path required to cause an electrical short of the metal oxide layer 17.

The present disclosure also encompasses an embodiment wherein the reduction process shown in fig. 7-12 (or fig. 13-15) may be applied to a metal oxide layer that already has a plurality of conductive channels therein. For example, the metal oxide layer 17 in the p-MTJ film stack may be replaced with a metal oxide layer 17a having multiple conductive vias 15c to provide the p-MTJ film stack of FIG. 16.

The process steps represented in FIGS. 8-11 are then followed to produce p-MTJ 4' with sidewall 4x, as shown in FIG. 17. In particular, the p-MTJ has a metal oxide center portion 17 c' with a conductive via 15c therein, and the center portion is separated from the sidewalls by end portions 15e having an oxidation state substantially lower than the oxidation state in the metal oxide center portion. In some embodiments, the end portion is an effective conductive path of a first metal or alloy M1 in the metal oxide layer 17c, while the conductive path 15c contains a second metal of one of the aforementioned Noble Metals (NM), such that the first metal or alloy has a different composition than the second metal.

In fig. 18, another embodiment of the present disclosure is depicted that retains all of the film layers from the first embodiment in fig. 2 except for the order of the film layers. A top spin valve configuration of p-MTJ5 is shown in which seed layer 11, metal oxide layer 17a in which metal via 15c is formed, free layer 14, tunneling barrier 13, reference layer 12, and hard mask 16 are sequentially formed on bottom electrode 10. The metal via extends from a top surface of the seed layer to a bottom surface of the free layer. As previously mentioned, the metal channel 15c may comprise the same metal or alloy as in the metal oxide layer, or may be one of the aforementioned NM metals or alloys. One key feature is to provide the electrical short circuit path 15c in the metal oxide layer to reduce the parasitic resistance of the metal oxide layer 17a for RA of the p-MTJ_TOTALThe contribution of (c). The metal oxide layer is referred to herein as the Hk promoting layer rather than the cap layer because the PMA in the free layer is increased due to the interfacial perpendicular anisotropy created at the interface of the bottom surface of the free layer and the underlying metal oxide layer.

In another top spin valve embodiment in FIG. 19, the bottom spin valve configuration in FIG. 6 is modified by retaining all the p-MTJ film layers, but changing the deposition order to produce p-MTJ6, where seed layer 11, metal oxide layer 17c with end 15e, free layer 14, tunneling barrier 13, reference layer 12, and hard mask 16 are sequentially formed on bottom electrode 10. The metal oxide layer 17c is considered to be a Hk promoting layer because it is no longer a capping layer over the free layer. The end portion 15e contains the same metal or metal alloy as in the metal oxide layer and provides a conductive path around the central metal oxide portion 17 c. As previously described, the termination is reduced by exposing the uniform metal oxide layer to a reducing agent or reducing species, which may be in the form of a metal layer (not shown) deposited on the sides of the etched metal oxide layer, or by performing a process with a forming gas or H₂Or reduction of one or both of the free radicals and the ions.

All of the embodiments described herein may be incorporated with the mastering toolAnd in the manufacturing scheme of the process. Significantly reduced RA in p-MTJ nano-pillars_TOTALWhile substantially maintaining DRR and free layer thermal stability, it facilitates 64Mb and 256Mb STT-MRAM technology, and related spintronic devices in which switching current, RA value, DRR, and thermal stability are all critical parameters.

While the present disclosure has been particularly shown and described with reference to preferred embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present disclosure.