Magnetic tunnel junction device

Abstract

According to one aspect of the inventive concept, there is provided a MTJ device comprising a layer stack, comprising a SOT layer and a first free layer, a second free layer, a reference layer and a tunnel barrier layer arranged between the second free layer and the reference layer, and a spacer layer arranged as an interface layer between the first free layer and the second free layer, wherein the SOT layer is adapted to switch a magnetization direction of the first free layer by SOT, and wherein the first free layer is adapted to generate a magnetic stray field acting on the second free layer such that a magnetization direction of the second free layer is responsive to the magnetization direction of the first free layer.

Description

Technical Field

[0001] The present invention relates to a magnetic tunnel junction device. Background Technology

[0002] With the ever-growing demand for faster, smaller, and more energy-efficient devices, conventional memories such as SRAM and DRAM have reached their expansion limits. New emerging memories are being developed, and magnetic random access memory (MRAM) is considered one of the most promising alternatives. MRAM has the potential to become the next generation of large-scale applications for spintronics, a rapidly growing field of solid-state electronics that aims to utilize the spin of electrons and their charge to achieve new electronic functions.

[0003] MRAM is based on a magnetic tunnel junction (MTJ) structure. An MTJ consists of two ferromagnetic layers (a reference layer and a free layer) and a tunnel barrier disposed between them. The magnetization direction or orientation of the reference layer is fixed or pinned, and is therefore also called a pinned layer. Meanwhile, the magnetization direction of the free layer can switch between two different directions relative to the reference layer, typically corresponding to a "parallel state" (P) and an "antiparallel state" (AP), respectively. The relative orientation of the magnetization of the reference and free layers determines the resistance of the MTJ. When the free layer is in the P state, the MTJ can exhibit a relatively low resistance, while when the free layer is in the AP state, the MTJ can exhibit a relatively high resistance. The tunnel magnetoresistance ratio (TMR) is a measure of the difference in MTJ resistance between the AP and P states. Therefore, the magnetization state of the free layer can be sensed by measuring the resistance of the MTJ. This allows data to be stored in the free layer, which can therefore also be called a sensing layer or storage layer.

[0004] Current MRAM technology typically relies on spin-transfer torsion (STT) as the key switching mechanism for free-layer magnetization. However, alternative switching mechanisms are being investigated to improve switching speed and reduce power consumption.

[0005] One promising candidate is the so-called spin-orbit torsion (SOT)-induced switching, where magnetization dynamics occur via current-induced spin-orbit coupling. Figure 1 schematically illustrates an MTJ configured for SOT-induced free-layer switching. The MTJ comprises a free layer 12, a tunnel barrier layer 14, a reference layer 16, and a magnetized pinned layer 18 that fixes the reference layer 14. The free layer 12 is formed on the SOT layer, which is typically a nonmagnetic metal layer such as Ta, W, or Pt, or a topological insulator such as Bi₂Se₃. By applying an in-plane current I... SOTThrough the SOT layer 10, a vertical spin current can be generated and magnetized and transferred to the free layer 12, thereby producing spin torsion in the free layer 12 and inducing magnetization reversal. The origin of the spin current can be described by the spin Hall effect (in the bulk of the SOT layer 10) and by the Rashba interaction (at the interface between the SOT layer 10 and the free layer 12). Summary of the Invention

[0006] As the inventors recognized, one problem with current SOT-based technologies is the need to compromise between the following: the SOT generation efficiency of the SOT layer, the spin polarization efficiency of the free layer (which contributes to the TMR of the tunnel junction), and the retention performance of the free layer.

[0007] Therefore, the purpose of this invention is to solve this problem. Another objective is to increase flexibility in the design of MTJs used for SOT-induced switching. Further and alternative objectives can be understood from the following.

[0008] According to a first aspect of the present invention, an MTJ device including a layer stack is provided, comprising:

[0009] SOT layer and first free layer,

[0010] The second free layer, the reference layer, and the tunnel barrier layer arranged between the second free layer and the reference layer, and

[0011] A spacer layer arranged as an interface layer between the first free layer and the second free layer.

[0012] The SOT layer is adapted to switch the magnetization direction of the first free layer via SOT, and

[0013] The first free layer is adapted to generate a magnetic stray field acting on the second free layer, such that the magnetization direction of the second free layer responds to the magnetization direction of the first free layer.

[0014] According to a second aspect of the present invention, a memory device is provided comprising at least one memory cell, each memory cell comprising a magnetic tunnel junction device according to the first aspect.

[0015] In a typical conventional design of a SOT-based MTJ device, the same free layer is used for both SOT switching (i.e., storage function) and sensing. In contrast, this invention provides a layer stack comprising first and second free layers separated by a spacer, enabling switching of the first free layer via SOT and sensing of the magnetization direction of the first free layer via the second free layer. Thus, the first free layer can be configured as a "storage layer," and the second free layer can be configured as a "sensing layer." Therefore, the SOT layer and the first free layer can form part of a "storage sub-stack," while the second free layer, the tunnel barrier layer, and the reference layer can form part of an "MTJ sensing sub-stack" (or simply "sensing sub-stack"), wherein the storage sub-stack and the sensing sub-stack are separated by a spacer layer.

[0016] The separation between the storage and sensing sub-stacks provided by the spacing allows each sub-stack to be individually customized (e.g., in terms of material and layer composition) to improve their respective functionality within the stack. This enables performance improvements in SOT switching efficiency, TMR, and retention. Another advantage is improved thermal stability and compatibility with back-to-line (BEOL) processes.

[0017] Compared to traditional single-free-layer SOT designs, dual-free-layer designs can tolerate larger second-free-layer offset fields. Since the second free-layer is used as a sensing layer rather than a write or storage layer, second-free-layer offsets do not cause write asymmetry or retention losses, respectively. This also provides flexibility in pinning layer design, enabling the use of uncompensated synthetic antiferromagnetic (SAF) pinning layers or non-SAF pinning layers, such as synthetic ferromagnetic (SFM) pinning layers.

[0018] According to an aspect of the invention, the coupling between the free layers depends on the interaction of (magnetic) stray fields. That is, the magnetic stray field generated by the first free layer is used to align the magnetization direction of the second free layer with the direction of the magnetic stray field extending through the second free layer. The direction of the magnetic stray field is determined by the magnetization direction of the first free layer.

[0019] Therefore, the SOT layer can respond to the reversal of the magnetization vector of the first free layer induced by the switching current through the SOT, wherein the reversal of the magnetization vector of the first free layer induces the reversal of the magnetization vector of the second free layer through magnetic stray field coupling.

[0020] The intensity of stray fields tends to increase as the layer critical size (CD) decreases. This is actually advantageous because the overall trend in circuit design and device development is toward smaller and smaller CDs.

[0021] Since the coupling between the magnetizations of the first and second free layers does not depend on magnetic exchange coupling (also known as interlayer exchange coupling (IEC) and Ruderman–Kittel–Kasuya–Yosida (RKKY) exchange coupling), stray field interactions provide considerable freedom in the design of the spacer layers.

[0022] For example, the spacer layer can be formed as a non-magnetic layer, such as a non-magnetic metal layer. Furthermore, the spacer can be adapted to provide texture separation between two free layers and to act as an interface layer between the two free layers, wherein each interface is adapted to enhance the properties of the first and second free layers relative to their intended use.

[0023] As used herein, unless otherwise expressly stated, the term “magnetization” of a layer refers to the net magnetization of the layer, which is equal to the sum of the fundamental magnetic moments of the layer.

[0024] Meanwhile, the term "magnetization vector" for a layer refers to the (net) magnetization vector of that layer. The magnetization direction or orientation of a layer refers to the direction or orientation of its magnetization vector. Furthermore, a given magnetization state of a layer (e.g., P or AP state) refers to the given orientation of the layer's magnetization.

[0025] As used in this article, a "free layer" is a layer with variable magnetization. That is, the direction of the magnetization vector of a free layer can be changed.

[0026] The magnetization direction of the first free layer can be switched between a first direction and a second direction, wherein in the first direction the magnetization direction of the first free layer is aligned with the magnetization direction of the reference layer, and in the second direction the magnetization direction of the first free layer is opposite to the magnetization direction of the reference layer. The first and second directions can be opposite directions. The first direction can be parallel to the magnetization direction of the reference layer, while the second direction can be antiparallel to the magnetization direction of the reference layer. The first free layer can correspondingly switch between a "parallel state" (P state) and an "antiparallel" state (AP state), wherein in the "parallel state" the magnetization direction of the free layer is parallel to the magnetization direction of the reference layer, and in the "antiparallel state" the magnetization direction of the free layer is antiparallel to the magnetization direction of the reference layer. The magnetization vector of the second free layer can then be oriented (through stray field coupling) along or against the magnetization direction of the reference layer. Preferably, the second free layer can be switched to the P state in response to the first free layer (relative to the reference layer) being switched to the P state, and switched to the AP state in response to the first free layer (relative to the reference layer) being switched to the AP state.

[0027] A tunneling barrier layer is a layer disposed between the reference layer and the free layer of an MTJ. The tunneling barrier layer can be adapted to allow tunneling currents to pass through it. The tunneling barrier layer can be a non-magnetic and electrically insulating layer, typically an oxide layer. The tunneling barrier layer can also be adapted to allow electrons to tunnel between the reference layer and the second free layer.

[0028] A reference layer is a layer with fixed or pinned magnetization. That is, the direction of the magnetization vector of the reference layer is fixed or pinned. The reference layer can also be called a fixed layer or a pinned layer. The magnetization of the reference layer can be oriented in-plane (in-plane magnetic anisotropy) or out-of-plane (PMA).

[0029] A pinned layer is a layer that pins the magnetization of a reference layer. A pinned layer can be a hard magnetic layer with high (magnetic) coercivity. A pinned layer can pin the magnetization direction of one or more associated reference layers. By applying a pinning effect to the reference layer, the inversion field of the reference layer can be increased compared to a free layer.

[0030] A SOT-generating layer (or simply "SOT layer") is a layer configured to inject spin current into a free layer (e.g., a first free layer) in response to conducting SOT switching current, thereby inducing magnetization reversal via SOT. The SOT layer can be arranged to contact the associated free layer. The SOT layer can be configured to lie in the plane of the SOT layer and thus conduct current along or parallel to the associated free layer.

[0031] Both single-layer and composite-layer spacer designs are possible. In other words, the spacer layer may include at least a first spacer sublayer.

[0032] The first spacer sublayer can be a SOT-generated layer arranged in contact with the surface of the first free layer, wherein the sign of the SOT coefficient of the first spacer sublayer is opposite to that of the SOT coefficient of the SOT layer. Therefore, both the first spacer sublayer and the SOT layer can contribute positively to switching, thereby effectively minimizing the switching current density of the first free layer.

[0033] The SOT layer and the first spacer sublayer can be formed from different materials exhibiting SOT coefficients with opposite signs.

[0034] The SOT generation layer may include a topological insulator (TI). The topological insulator allows for the formation of layers with strong spin-orbit coupling. However, the first spacer sublayer / SOT generation layer may alternatively include an SOT generation metal, such as a heavy metal.

[0035] The spacer layer may include a second spacer sublayer arranged to contact the first sublayer. Therefore, the first sublayer may be selected with consideration of compatibility with the first free layer, and the second sublayer may be selected with consideration of compatibility with the second free layer.

[0036] For example, when the first sublayer comprises a topological insulator, a second sublayer (and optionally other sublayers) may be selected to provide texture that enhances the magnetic anisotropy of the second free layer. For instance, forming a second free layer with perpendicular magnetic anisotropy (PMA) near a topological insulator layer can be challenging. However, this challenge can be mitigated by forming the second spacer sublayer as a PMA-enhancing layer. Advantageous structures of spacer layers formed from the first and second spacer sublayers include a first spacer sublayer formed from an SOT-generated layer and a second spacer sublayer of a layer of W, Ti, Pd, Pt, Ta, Ru, Mo, or Ti, or a Ru / Mo bilayer.

[0037] The second spacer layer may include at least one metal layer. The second spacer layer may be, for example, a single metal layer or a metal bilayer. The second spacer layer may be formed as a non-magnetic metal layer.

[0038] The spacer layer may include a third spacer sublayer arranged to contact the second sublayer and comprising an oxide layer having a 001 crystal structure. Hereinafter, crystal structures with an index ijk may be indicated in the format (ijk), or equivalently, as “ijk” to avoid confusion with reference numerals in the claims. The spacer layer can therefore be adapted for use in conjunction with a second free layer based on Fe or Co (such as a second free layer comprising Fe, Co, FeB, CoB, CoFe, or CoFeB layers), wherein the (001) crystal texture of the oxide layer can facilitate PMA of the second free layer. MTJs based on Fe and / or Co are capable of achieving high TMR.

[0039] The oxide layer can preferably be a Mg-containing oxide layer. A Mg-containing oxide layer can effectively promote PMA in the second free layer. Another advantage of including an oxide layer in the spacer layer is that it can increase the resistance of the spacer layer, thus more effectively preventing SOT current from flowing into the second free layer.

[0040] The third spacer sublayer further includes a boron-containing layer, wherein the boron-containing layer is arranged to contact the second sublayer, and the oxide layer is arranged to contact the second free layer. The boron-containing layer can promote the (001) crystal structure of the oxide layer. The boron-containing layer can be a CoFeB layer, a CoB layer, or a FeB layer.

[0041] The thickness of the spacer layer can be 5 nm or less. If the spacer layer includes a CoFeB layer, the thickness of the CoFeB layer is preferably 1 nm or less, more preferably 0.4-0.5 nm. Such a thickness allows the CoFeB layer to provide texture-enhancing functionality relative to the oxide layer and to be formed as a (at least substantially) non-magnetic layer.

[0042] Advantageous structures of spacer layers formed by first, second and third spacer sublayers include a first spacer sublayer formed by an SOT-generated layer, a second spacer sublayer of a layer of W, Ti, Pd, Pt, Ta, Ru, Mo or Ti or a Ru / Mo bilayer, and a third spacer sublayer formed by a Mg-containing oxide layer or by a CoFeB layer and a Mg-containing oxide layer.

[0043] The coercivity of the second free layer can be less than the strength of the magnetic stray field acting on it. This allows the magnetization switching of the second free layer to be controlled solely by the stray field generated by the first free layer, without requiring any other field or current to facilitate the switching.

[0044] The strength of the magnetic stray field acting on the second free layer is advantageously at least 10 mT, preferably at least 30 mT.

[0045] The first free layer can be the PMA layer.

[0046] The second free layer can be the PMA layer.

[0047] The layer stack may further include a third free layer and a coupling layer disposed between the second and third free layers and adapted to provide antiparallel coupling of the magnetization of the second and third free layers. The layer stack may thus be configured with a composite antiferromagnetic (SAF) free layer (sub) stack. Magnetic perturbations of the second free layer on the first free layer (e.g., due to the opposite magnetization of the third free layer) can thus be reduced. The second free layer may include a Co-containing ferromagnetic layer. The third free layer may include a Co-containing ferromagnetic layer. The second and third ferromagnetic free layers may each have a (001) crystal structure.

[0048] The magnetic layer stack design of this invention can be applied to both bottom-pinned and top-pinned MTJ configurations. That is, the second free layer, reference layer, and tunnel barrier layer can form part of a top-pinned MTJ disposed above the first free layer and spacer layer. Alternatively, the second free layer, reference layer, and tunnel barrier layer can form part of a bottom-pinned MTJ disposed below the first free layer and spacer layer.

[0049] The device may include a substrate, on which a layer stack is arranged.

[0050] The phrase "a first feature (such as a layer or other structure) is formed above or below a second feature (such as a layer or other structure)" means that the first feature is formed above or below the second feature in the direction of the normal extending in the plane of the main surface or feature (e.g., layer) or correspondingly relative to the normal direction of the substrate (if present) (as shown in the figure).

[0051] The device may include an electrode structure configured to supply SOT switching current to the SOT layer. The electrode structure may be configured to inject / extract write current into / from the SOT layer, such that the write current can pass through the SOT layer in an in-plane direction. Attached Figure Description

[0052] The above and additional objectives, features, and advantages of the invention will be better understood from the following illustrative and non-limiting detailed description, with reference to the accompanying drawings. In the drawings, unless otherwise stated, the same reference numerals will be used for the same elements.

[0053] Figure 1 is a schematic cross-sectional view of a prior art MTJ with SOT-induced free layer switching.

[0054] Figure 2 This is a schematic cross-sectional view of the magnetic layer stack including the top-pinned MTJ.

[0055] Figure 3 This is a schematic cross-sectional view of a magnetic layer stack including a top-pinned MTJ, based on a variant.

[0056] Figure 4 This is a schematic cross-sectional view of a magnetic layer stack of MTJs with bottom pinning, based on a variant.

[0057] Figure 5 This is a schematic view of a memory device. Detailed Implementation

[0058] The MTJ device according to this disclosure will now be described more fully below with reference to the accompanying drawings. In the drawings showing a cross-section of the device, it is conceivable that the device may extend in a direction perpendicular to the plane of the cross-section. The individual layers illustrated in the drawings may also extend laterally / horizontally beyond the illustrated portion, which is for illustrative purposes only. It should also be noted that relative dimensions (e.g., layer thicknesses) are not drawn to scale. In the drawings, arrow Y indicates a vertical direction, i.e., perpendicular to the main surface of the device substrate. The vertical direction corresponds accordingly to the stacking direction of the layers in the depicted layer stack. Meanwhile, arrow X indicates a horizontal direction, i.e., along the main surface of the device substrate.

[0059] Figure 2 MTJ device 100 is shown. Device 100 includes a layer stack 108 comprising an SOT layer 120 and a first free layer 130. Layer stack 108 includes a second free layer 150, a reference layer 170, and a tunnel barrier layer 160 disposed between the second free layer 150 and the reference layer 170. Layer stack 108 includes a spacer layer 140 disposed as an interface layer between the first free layer 130 and the second free layers 150, 130. The spacer layer will be further described below.

[0060] Layer stack 108 can be arranged above substrate 102 as shown. Substrate 102 can be any conventional substrate or wafer, such as a semiconductor substrate. Examples of semiconductor substrates include Si substrates, Ge substrates, SiGe substrates, SiC substrates, SOI substrates, GeOI substrates, SiGeOI substrates, etc. Although not shown in the figure... Figure 2 As shown, however, substrate 102 may support active device portions or include front-end processing (FEOL) portions of active devices. Active devices may be formed in the semiconductor portion of substrate 102. Devices may form logic circuit devices. Devices may include switching for conducting read and write currents for device 100, as will be described in more detail below. Example active devices include MOSFETs, MISFETs, BJTs, JBTs, FinFETs, nanowire FETs, etc. As further shown, layer stack 108 may be embedded in interconnect structure 104 disposed on substrate 102. Interconnect structure 104 may include multiple dielectric layers, such as SiO2 and / or other conventional back-end processing dielectrics. Interconnect structure 104 may include multiple metal layers embedded with dielectric layers. Metal layers may be formed, for example, of Cu, W, Au, Ag, Al, Co, or Ru. Conductive vertical vias may be provided for interconnecting metal layers. One or more of the metal layers may define read lines, word lines, select lines, etc. The lines of the metal layer, together with the active devices in the FEOL section, can define a circuit system for controlling read and write operations and for driving bit lines, word lines, etc.

[0061] The SOT layer 120 is adapted to switch the magnetization direction of the first free layer 130 via SOT. Simultaneously, the first free layer 130 is adapted to generate a magnetic stray field such that the magnetization direction of the second free layer 150 responds to the magnetization direction of the first free layer 130. As will be further described, this allows the first free layer 130 to function as a storage / write layer of the device 100. On the other hand, the second free layer 150 can function as a sensing / reading layer of the device 100.

[0062] The variable magnetization directions of free layers 130 and 150 and the fixed magnetization direction of reference layer 170 are in Figure 2 The arrows are indicated by a pair of reverse arrows in free layers 130 and 150 and a single arrow in reference layer 170, respectively. Layer stack 108 may include a pinned layer 180, as shown, which is disposed above reference layer 170 and adapted to fix the magnetization direction of reference layer 170. Accordingly, the MTJ defined by the second free layer 150, tunnel barrier layer 160, and reference layer 170 is a top-pinned MTJ. Figure 2Explicit indication of the magnetization direction of pinned layer 180 is omitted. However, for example, in the case of a SAF pinned layer configuration, pinned layer 180 typically has a net magnetization in the same direction as reference layer 170, a net magnetization opposite to reference layer 170, or even zero net magnetization. It should be noted that the upward orientation of the arrows in reference layer 170 is merely an example, and a downward orientation is equally possible. Figure 2 The out-of-plane / vertical orientation of the arrows in the diagram illustrates the PMA of free layers 130, 150, and reference layer 170 (as well as pinned layer 180). However, it can be noted that in-plane magnetic anisotropy of the magnetic layers in the MTJ is also possible.

[0063] The magnetization of the first free layer 130 can be set to the P state, wherein the magnetization of the first free layer 130 presents a first direction parallel to the magnetization direction of the reference layer 170. The magnetization of the first free layer 130 can also be set to the AP state, wherein the magnetization of the first free layer 130 presents a second direction antiparallel to the magnetization direction of the reference layer 170. Therefore, in Figure 2 In this context, the P state corresponds to the upward orientation of the magnetization vector of the first free layer 130, while the AP state corresponds to the downward orientation of the magnetization vector of the first free layer 130.

[0064] The first free layer 130 can achieve a sufficiently large SOT switching current (I) in the plane. SOT The SOT layer 120 can be used to switch between the P and AP states via either the first or second in-plane direction. The critical current required to induce the SOT magnetization state switching of the first free layer 130 depends, for example, on the materials chosen for the SOT layer 120 and the first free layer 130, as well as the layer thickness of the first free layer 130. Figure 2 As indicated, a portion of the current entering SOT layer 120 can flow into other layers of the stack, including layers 130 and 140. Therefore, the SOT current component can flow along free layer 130 in each of layers 120 and 140.

[0065] The first free layer 130 has a non-zero net magnetization in both the P and AP states. This non-zero net magnetization generates a magnetic stray field outside the first free layer 130. The stray field (among others) extends through the second free layer 150. Inside the second free layer 150, the direction of this stray field (i.e., the orientation of the stray field vector within the second free layer 150) is determined by the magnetization direction of the first free layer 130. The stray field can twist the fundamental magnetic moment of the second free layer 150, thereby contributing to its net magnetization such that the magnetization direction of the second free layer 150 is aligned with the direction of the stray field. Assuming that the direction of the stray field within the second free layer 150 is substantially parallel to the magnetization direction of the first free layer 130, it can be concluded that the magnetization direction of the second free layer 150 can switch between the first and second directions together with the first free layer 130. Under this assumption, arrows within the second free layer 150 can also be considered to indicate the direction of the stray field. In other words, the second free layer 150 can switch from the P state to the AP state in response to the first free layer 130 switching from the P state to the AP state, or vice versa. More generally, the reorientation of the magnetization vector of the first free layer 130 can induce a corresponding reorientation of the magnetization vector of the second free layer 150.

[0066] To facilitate stray-field-based switching of the magnetization of the second free layer 150, the first free layer 130 can be adapted to generate a magnetic stray field with an intensity exceeding that of the reverse field of the second free layer 150. In the case of a non-zero offset of the second free layer (its hysteresis loop), the reverse field can be larger in one switching direction and smaller in the other. Therefore, the intensity of the magnetic stray field should exceed the maximum intensity of the reverse field of the second free layer 150. For a zero-second free layer offset (corresponding to a perfectly centered hysteresis loop), stray-field-based switching of the magnetization of the second free layer 150 can be achieved by making the second free layer 150 exhibit a magnetic coercivity less than the intensity of the magnetic stray field acting on it. In either case, a magnetic stray field intensity of at least 10 mT acting on the second free layer 150 may be sufficient to ensure the reorientation of the magnetization vector of the second free layer 150. A stronger stray field of at least 30 mT may also allow stray field coupling to obtain a greater second free layer coercivity. The coercivity of the second free layer 150 is influenced / determined by its magnetic anisotropy and its magnetic moment. The stray field acting on the second free layer is influenced / determined by the saturation magnetization and thickness of the first free layer, the thickness of the spacer layer, and the critical size (CD) or diameter of the first free layer 130.

[0067] In the foregoing, the first and second directions of magnetization of the free layer are referred to as opposite directions. However, as those skilled in the art will understand, such statements should not be interpreted literally as requiring completely opposite magnetization directions. In practice, the magnetization direction of the free layer may deviate to some extent from completely opposite directions depending on factors such as the degree of PMA, switching efficiency, inhomogeneity of the local magnetic field, and layer composition. This applies accordingly to statements regarding magnetization vector reversal, and therefore does not need to correspond to a complete 180-degree reorientation. Preferably, during magnetization vector reversal, at least the principal component of the magnetization vector should be reoriented by 180 degrees. Furthermore, it is envisioned that the second free layer 150 need not have the same magnetic anisotropy direction as the first free layer 130. In fact, the above operating principle will work as long as the switching of the first free layer causes the magnetization of the second free layer 150 to deterministically switch between states that can be distinguished during readout.

[0068] Equipment 100 can be used as Figure 2 The diagram shows a configuration to provide SOT switching current or write current I to SOT layer 120. SOT The bottom electrode structure 110. The electrode structure 110 is configured to inject / extract write current into / from the SOT layer 120, such that the write current can be in an in-plane direction (e.g., in...). Figure 2 (From left to right, or vice versa) through the SOT layer 120. Portions of the SOT current can further propagate in the plane through the first free layer 130. The electrode structure 110 includes first and second electrode portions 110a, 110c spaced apart by a middle portion 110b of insulating material. Therefore, as... Figure 2 The write current I is shown schematically. SOT Conduction can be achieved between electrode portions 110a and 110c via the SOT layer 120. The first electrode portion 110a can be connected to a write line or bit line via a write switch (e.g., the FEOL portion of substrate 102). The switch can be controlled by applying a control voltage to the select line. The other terminal of the switch can be connected to the write line. The state of the switch can be controlled by applying a control voltage to the select line.

[0069] SOT layer 120 has a larger footprint (i.e., a larger lateral dimension) than the first free layer 130, on the first and second sides of the first free layer 130 that are opposite to each other (e.g., Figure 2The left and right sides of the first free layer 130 define the lateral protrusions. A first electrode portion 110a can be connected to the protrusion on the first side, while a second electrode portion 110c can be connected to the protrusion on the second side. This design of the SOT layer 120 can improve SOT switching efficiency because the current path can extend along the entire length of the interface between the SOT layer 120 and the first free layer 130. However, a layer stack design in which the SOT layer 120 and the first free layer 130 coexist is also possible.

[0070] Device 100 may further include a top electrode 190, as shown. The top electrode 190 is disposed above the pinned layer 180. The top electrode 190 may be connected directly or via vias to read lines in the interconnect structure to allow read current to conduct along a read path extending through the layer stack 108 between the top electrode 190 and the first and / or second electrode portions 110a, 110c. Read switches, such as FETs, may be arranged in the read path to allow selective activation of the read path.

[0071] The write operation of device 100 may include causing the write current I... SOT Through SOT layer 120, as disclosed above, the first free layer 130 can switch to a P-state or an AP-state depending on the direction of the write current. In response, the second free layer 150 can switch to a P-state or an AP-state depending on the direction of the magnetic stray field generated by the first free layer 130 and acting on the second free layer 150. Read operations of device 100 may include measuring the resistance across layer stack 108. The resistance can be measured by providing a read current across layer stack 108 via the read line through the MTJ between the top electrode 190 and either electrode portion 110a, 110c and sensing the resulting voltage drop. The P-state of the second free layer 150 will result in a lower resistance than the AP state of the second free layer 150, the difference being given by the TMR of the MTJ. A logic "1" can be associated with a lower resistance, and a logic "0" can be associated with a higher resistance, and vice versa. Therefore, data can be written to the MTJ device 100 by switching the magnetization state of the first free layer 130, and data can be read from the MTJ device 100 by detecting the resistance of the MTJ to sense the state of the second free layer 150.

[0072] The free layer, reference layer, and pinned layer of the layer stack 108 can be formed as layers with a PMA. The first and second free layers 130, 150 can include corresponding ferromagnetic layers. Examples of materials for free layers 130, 150 include Fe, Co, FeB, CoB, CoFe, CoFeB, WCoFeB, and CoFeBTa. Multilayer Co / X (where X = Ni, Pt, or Pd) is also possible. It is envisioned that free layers 130, 150 can have a multilayer structure including combinations of the materials mentioned above. For example, the corresponding thicknesses of free layers 130, 150 can be in the range of 0.5 to 2 nm. The reference layer 170 can include a ferromagnetic layer. Examples of materials for the reference layer 170 include Fe, Co, FeB, CoB, CoFe, CoFeB, or WCoFeB. Other suitable materials may include, for example, Ni, FePt, CoGd, CoFeGd, CoFeTb, and CoTb. It is envisioned that the reference layer 170 can have a multilayer structure including combinations of the materials mentioned above. The barrier layer 160 may include a non-magnetic and electrically insulating layer such as an oxide layer, for example, MgO, AlOx, MgAlOx, MgGaOx, or MgTiOx.

[0073] Pinning layer 180 may include one or more hard magnetic layers. Pinning layer 180 may, for example, include a repeating sequence or a bilayer superlattice laminate. Pinning layer 180 may also include repeating sequences of Co layers, Fe layers, or CoFe layers or superlattice laminates or [CoFe / X], where X represents Pd, Pt, Ni, Tb, or Gd. Pinning layer 180 may also include alloying compositions such as alloys of Co, Fe, or CoFe with Pt, Pd, Ni, Tb, or Gd, or ternary alloys of CoXCr, where X represents Pt, Ni, Pd, Tb, or Gd. Parallel and antiparallel coupling of the magnetization of pinning layer 180 and reference layer 170 is possible. Pinning layer 180 may be coupled to reference layer 170, for example, via a nonmagnetic spacer layer (such as a thin Ru layer) or other RKKY coupling metal layers (such as Ir, Os, or Rh layers having a thickness that provides the desired coupling), in parallel or antiparallel fashion. A transition layer (such as Ta, W, Mo, or their alloys, or CoX, FeX, CoFeX, CoFeB of Ta, W, or Mo) may also be present between the reference layer 170 and the pinning layer 180. The pinning layer 180 may be formed as a synthetic ferromagnetic pinning layer (SFM) or a SAF pinning layer. The SAF-type pinning layer 180 may include two hard magnetic layers (e.g., each composed of the components listed above) separated by a non-magnetic spacer layer (such as a Ru or Ir layer) of such thickness that antiparallel coupling is achieved.

[0074] SOT layer 120 may include a conductive material layer exhibiting relatively large spin-orbit coupling. SOT layer 120 may be a non-magnetic SOT-formed metallic layer. Metallic SOT layers with negative SOT coefficients include, for example, Ta, W, Hf, Ir, IrMn, W, and WO. x WN, W(O,N), TaN, and TaB. Metallic SOT layers with a positive SOT coefficient include, for example, Pt, Pd, Hf, Au, AuPt, PtHf, PtMn, FeMn, and NiMn. SOT-forming layers also include topological insulator layers, such as Bi. x Se 1-x Bi x Sb 1-x And (Bi,Sb)₂Te₃. Topological insulators typically exhibit a positive SOT coefficient. SOT layers of transition metal chalcogenides (TMD or MX₂) are also possible, such as MoS₂ and WTe₂. SOT layer 120 can also have a multilayer structure, for example, comprising a combination of two or more of the materials mentioned above. SOT layer 120 can be formed with a thickness of, for example, 10 nm or less (preferably 5 nm or less).

[0075] The electrode portions 110a and 110c of the bottom electrode structure 110 can be formed of W, Ta, or TaN, or some other conventional electrode materials. The insulating intermediate portion 110b can be formed of an oxide such as SiO2, or some other conventional dielectric.

[0076] To facilitate the formation of the SOT layer 120 above the bottom electrode structure 110, the layer stack 108 may include a metal seed / adhesion layer 115 forming an interface seed layer between the electrode structure 110 and the SOT layer 120. The seed layer 115 may be formed of a metal nitride layer XN (where X = Ta, Ti, W, Si, Mo). For example, a seed layer 115 having a thickness in the range of 1-2 nm may be formed.

[0077] The top electrode 190 can be a single layer of, for example, Ru or Ta or some other conventional electrode material.

[0078] Spacer 140 physically separates the first free layer 130 from the second free layer 150. The first and second free layers 130 and 150 can therefore be structurally separated from each other. Since the switching of the second free layer 150 is facilitated by stray field intervening generated by the first free layer 130, spacer 140 does not need to provide strong IEC. Single-layer and composite-layer configurations of the spacer are possible, such as... Figure 2 As shown:

[0079] Spacer layer 140 can be formed from a first spacer sublayer 140a. The first spacer sublayer 140a can be arranged accordingly to contact free layers 130, 150 to form an interface layer between free layers 130, 150. The first spacer sublayer 140a can be a non-magnetic metallic layer. A "non-magnetic" layer refers to a non-ferromagnetic layer that does not exhibit or substantially does not exhibit net magnetization in the absence of an external magnetic field. The first spacer sublayer 140a can be formed, for example, as a layer of Ru, Mo, or Ti. The first spacer sublayer 140a can advantageously be formed with a thickness in the range of 1 to 5 nm. Therefore, the first spacer sublayer 140a can provide an IEC intensity that is negligible or at least considerably low compared to stray fields. The first spacer sublayer 140a can also be formed as a SOT-generated layer. Examples of possible SOT-generated layers are provided above in conjunction with SOT layer 120. Preferably, if the first spacer sublayer 140a is formed as an SOT generation layer, then the first spacer sublayer 140a can be formed as an SOT generation layer having SOT coefficients with opposite signs to those of the SOT layer 120.

[0080] Spacer layer 140 may be formed of a first spacer sublayer 140a and a second spacer sublayer 140b arranged in contact with the first spacer sublayer 140a. The first spacer sublayer 140a may be arranged in contact with a first free layer 130, and the second spacer sublayer 140b may be arranged in contact with a second free layer 150. The first spacer sublayer 140a may be a non-magnetic metal layer. The first and second spacer sublayers 140a and 140b may each be formed from corresponding layers of non-magnetic metal layers such as Ru, Mo, or Ti. Advantageously, the first spacer sublayer 140a may be formed as an SOT-generated layer as mentioned above, and the second spacer sublayer 140b may be formed from a non-magnetic metal layer or a non-magnetic metal bilayer. For example, the second spacer sublayer 140b may be a layer of W, Ti, Pd, Pt, Ta, Ru, Mo, or Ti, or a Ru / Mo bilayer. Spacer layer 140 with a thickness in the range of 1 to 5 nm may be advantageously formed.

[0081] A third spacer sublayer 140c can be used to supplement the combination of the first and second spacer sublayers of spacer layer 140. Therefore, spacer layer 140 can be formed by a first spacer sublayer 140a, a second spacer sublayer 140b arranged to contact the first spacer sublayer 140a, and a third spacer sublayer 140c arranged to contact the second spacer sublayer 140b. The first spacer sublayer 140a can be arranged to contact the first free layer 130. The third spacer sublayer 140c can be arranged to contact the second free layer 150. The second spacer sublayer 140b can be arranged between the first and third spacer sublayers 140a and 140c to form an interface layer between them. The combined thickness of the first, second, and third spacer sublayers 140a, 140b, and 140c can be 5 nm or less, preferably in the range of 1-3 nm.

[0082] The third spacer sublayer 140c can be a single or composite layer having a (001) crystal structure. This can facilitate the formation of a second free layer 150 based on PMA Fe or Co on the spacer layer 140. The third spacer sublayer 140c can be formed from an oxide layer having at least a (001) crystal structure. Mg-containing oxides such as MgO, MgAlO, MgGaO, and MgTiO can form an effective PMA-induced textured layer; however, other oxides such as AlOx and TiOx are also possible. The oxide layer of the third spacer sublayer 140c can also be supplemented with a CoFeB layer, wherein the third spacer sublayer 140c can be formed as a bilayer of CoFeB and oxide layers. The B content of the CoFeB layer can promote (001) crystal texture in the oxide layer and further extend into the upper layers of the stack. The CoFeB layer can be arranged to contact the second sublayer 140b, while the oxide layer can be arranged to contact the second free layer 150. The thickness of the CoFeB layer is preferably 1 nm or less, more preferably 0.4-0.5 nm. The CoFeB layer can more generally be a layer containing B, such as a FeB or CoB layer. The B content can promote a (001) crystal structure in the oxide layer during device fabrication. The B-containing layer can be initially deposited as an amorphous layer and induce a (001) crystal structure in the oxide layer during annealing, and may also exhibit a (001) crystal structure itself.

[0083] The electrode structure 110 can be formed by depositing and patterning the respective materials for the electrode portions 110a, 110c and the insulating portion 110b using techniques known in the art. Additional layers of the device 100 and the layer stack 108 can be formed using conventional deposition techniques such as evaporation or sputtering, or MBE (molecular beam epitaxy), or ALD (atomic layer deposition), or MOCVD (metal-organic chemical vapor deposition). After layer deposition, a hard mask, such as TiN, TaN, TiTaN, or spin-coated carbon / spin-coated glass material, can be defined on top of the layer stack. The hard mask may be rectangular, elliptical, or circular, for example. Subsequently, the deposited layers can be patterned into “MTJ pillars” by etching using the hard mask as an etching mask. Possible etching techniques include, for example, anisotropic etching processes such as reactive ion etching (RIE) or ion beam etching (IBE). To increase the footprint of the SOT layer 120, the layer stack can be patterned in two steps, including first etching the layers above the SOT layer 120 using a first, smaller footprint hard mask, and then stopping above the SOT layer 120. Subsequently, the SOT layer 120 (and optionally the seed layer 115) can be patterned using a second, larger footprint hard mask. By forming multiple hard masks, any number of such MTJ pillars can be patterned in parallel within the same deposited layer stack. The fabrication process may further include one or more annealing steps to crystallize the magnetic layers while applying a magnetic field to the structure to set the desired magnetization. Annealing can be performed before or after pillar patterning. Annealing can be performed in a furnace at an ambient temperature of approximately 400°C. During annealing, the magnetization directions of the reference layer 170 and the first and second free layers 130, 150 can be aligned with each other.

[0084] Figure 3 A variant of MTJ device 100' is illustrated. MTJ device 100' is combined with... Figure 2 The disclosed MTJ device 100 is identical, but differs in that the magnetic tunnel junction of the layer stack 108' includes a double-free-layer design. More specifically, the layer stack 108' further includes a third free-layer 154 and a coupling layer 152 disposed between the second and third free-layers 150, 154 and adapted to provide antiparallel coupling of the magnetization of the second and third free-layers. The opposite magnetization directions of the second and third free-layers 150, 154 are determined by... Figure 3The opposing orientations of the arrow pairs in the layers are indicated. The third free layer 154 can be formed of any material mentioned above in conjunction with the second free layer 150. The coupling layer 152 can be formed of a non-magnetic layer (such as a thin Ru layer) or other RKKY coupling metal layers (such as Ir, Os, or Rh layers with a thickness that provides antiparallel coupling). Due to the antiparallel coupling, the net magnetization of the free layer system including the second free layer 150, the coupling layer 152, and the third free layer 154 (i.e., the combined net magnetization seen in the first free layer 130) can be reduced compared to a free layer system including only the second free layer 150.

[0085] Figure 4 Another MTJ device 200 is illustrated. MTJ device 200 is similar to MTJ device 100 and includes a layer stack 208 similar to layer stack 108. However, MTJ device 200 differs from device 100 in that it is a bottom-pinned configuration. That is, the reference layer 170 and pinned layer 180 are arranged below the second free layer 150. Figure 4 As indicated in the diagram, the spacer layers 140 of the layer stack 208 can be formed in a manner corresponding to the spacer layers 140 of the layer stack 108. However, due to the bottom pinning configuration, Figure 4 The relative order of sublayers 140a-c shown is... Figure 2 Conversely, another difference compared to device 100 is the provision of a top electrode structure 210, which is disposed above SOT layer 120 and adapted to provide write current to SOT layer 120. Electrode structure 210 includes first and second electrode portions 210a, 110c spaced apart by an intermediate portion 210b of insulating material, similar to electrode structure 110 of MTJ device 100. Meanwhile, layer stack 208 includes a bottom electrode 290 disposed below pinned layer 180. Bottom electrode 290 can be directly or via via connected to read lines in the interconnect structure to allow read current to conduct along a read path extending through layer stack 208 between bottom electrode 290 and the first and / or second electrode portions 210a, 210c of electrode structure 210. The above discussion of bottom electrode structure 110 and top electrode 190 in relation to layer stack 108 corresponds to top electrode structure 210 and bottom electrode 290, respectively. The layer stack 208 may include a seed layer 282 disposed between the bottom electrode 290 and the pinned layer 180, as shown. The seed layer 282 may include, for example, a Pt layer, a Ru layer, or a Ta layer, or a stack of two or more of these layers. Such a layer may induce a (111) texture of the pinned layer 180.

[0086] Figure 5A schematic block diagram of circuit device 300 is provided. As schematically indicated, circuit device 300 includes a plurality of circuit cells 310, for example, arranged as an array including a plurality of cell rows and cell columns. Each circuit cell includes a corresponding MTJ device of any of the foregoing designs, such as device 100, 100', or 200. Circuit device 300 may include circuitry 320 for addressing, reading, and writing data to / from the respective write layers 130, 230 of the respective MTJ devices of the circuit cell. Circuitry 320 may include, for example, word lines, bit lines, select lines, row and column drivers, sense amplifiers, etc., which are known in the art. Circuit device 300 may be a magnetic random access memory (MRAM) device, wherein each circuit cell forms a memory cell of the MRAM device. However, other applications are also possible, such as non-volatile static random access memory (nvSRAM) devices, wherein each SRAM bit cell is connected to a respective memory cell including MTJ devices configured to provide non-volatile storage of the logical state of the SRAM bit cell.

[0087] The concept of the invention has been described above primarily with reference to a limited number of examples. However, as will be readily apparent to those skilled in the art, other examples besides those disclosed above are also possible within the scope of the concept of the invention as defined in the appended claims.

Claims

1. A magnetic tunnel junction (MTJ) device (100, 200) comprising a layer stack (108, 208), comprising: SOT layer (120) and first free layer (130), A second free layer (150), a reference layer (170), and a tunnel barrier layer (160) disposed between the second free layer (150) and the reference layer (170), and A spacer layer (140) is arranged as an interface layer between the first free layer (130) and the second free layer (150). The SOT layer (120) is adapted to switch the magnetization direction of the first free layer (130) via SOT, and The first free layer (130) is adapted to generate a magnetic stray field acting on the second free layer (150), such that the magnetization direction of the second free layer (150) responds to the magnetization direction of the first free layer (130), and the intensity of the magnetic stray field acting on the second free layer (150) exceeds the maximum intensity of the reverse field of the second free layer (150).

2. The device as described in claim 1, characterized in that, The spacer layer (140) includes a first spacer sublayer (140a).

3. The device as described in claim 2, characterized in that, The first spacer sublayer (140a) is a generated layer arranged to contact the surface of the first free layer (130), wherein the sign of the SOT coefficient of the first spacer sublayer (140a) is opposite to the sign of the SOT coefficient of the SOT layer (120).

4. The device as described in any one of claims 2-3, characterized in that, The spacer layer (140) further includes a second spacer sublayer (140b) arranged to contact the first spacer sublayer (140a).

5. The device as described in claim 4, characterized in that, The second spacer sublayer (140b) includes at least one metal layer.

6. The device as described in claim 5, characterized in that, The spacer layer (140) further includes a third spacer sublayer (140c) arranged to contact the second spacer sublayer (140b), wherein the third spacer sublayer (140c) includes an oxide layer having a 001 crystal structure.

7. The device as claimed in claim 6, characterized in that, The third spacer sublayer (140c) further includes a B-containing layer, wherein the B-containing layer is arranged to contact the second spacer sublayer (140b), and the oxide layer is arranged to contact the second free layer (150).

8. The device as claimed in claim 6, characterized in that, The thickness of the spacer layer (140) is 5 nm or less.

9. The device as described in any one of claims 1-3, characterized in that, The intensity of the magnetic stray field acting on the second free layer (150) is at least 10 mT.

10. The device as claimed in claim 9, characterized in that, The intensity of the magnetic stray field acting on the second free layer (150) is at least 30 mT.

11. The device as described in any one of claims 1-3, characterized in that, The layer stack further includes a third free layer (154) and a coupling layer (152), the coupling layer (152) being disposed between the second and third free layers and adapted to provide antiparallel coupling of the magnetization of the second and third free layers.

12. The device as described in any one of claims 1-3, characterized in that, The second free layer (150), the reference layer (170), and the tunnel barrier layer (160) constitute part of a top pinned magnetic tunnel junction disposed above the first free layer (130) and the spacer layer (140); or, wherein the second free layer (150), the reference layer (170), and the tunnel barrier layer (160) constitute part of a bottom pinned magnetic tunnel junction disposed below the first free layer (130) and the spacer layer (140).

13. The device as described in any one of claims 1-3, characterized in that, It further includes electrode structures (110, 210) configured to provide SOT switching current to the SOT layer (120).

14. A circuit device (300) comprising at least one circuit unit (310), each circuit unit comprising a magnetic tunnel junction device according to any one of the preceding claims.

Images