Superconducting memory with Josephson phase-based torque

By using a combination of magnetic Josephson junctions and inductors in memory cells to control the magnetization reversal of the magnetic layer, the high power consumption problem of CMOS memory in inactive states is solved, and low-power memory operation is achieved.

CN112703674BActive Publication Date: 2024-11-05MICROSOFT TECHNOLOGY LICENSING LLC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN201980057609.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2018-09-04
Filing Date
2019-06-25
Publication Date
2024-11-05
Estimated Expiration
2039-06-25

AI Technical Summary

Technical Problem

Existing CMOS-based memories still suffer from high power consumption due to leakage current when inactive, especially in data centers where server energy consumption is significant.

Method used

A memory cell based on superconducting logic is used, which forms a circuit using a magnetic Josephson junction and an inductor. The magnetization reversal of the free magnetic layer is controlled by applying current and flux, thereby achieving low-power storage operation.

Benefits of technology

It enables low-power memory operations, reduces current leakage during inactivity, and improves the energy efficiency of data centers.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112703674B_ABST
    Figure CN112703674B_ABST
Patent Text Reader

Abstract

Examples described in the present disclosure relate to a memory cell having a torque based on a Josephson phase. In one example, a memory cell is provided, the memory cell comprising a first inductor and a magnetic Josephson junction (MJJ) coupled to the first inductor to form a loop. The MJJ may include a free magnetic layer formed above a non-magnetic layer, and a fixed magnetic layer below the non-magnetic layer. A first state of the memory cell corresponds to a first magnetization of the free magnetic layer parallel to the magnetization of the fixed magnetic layer, and a second state of the memory cell corresponds to a second magnetization of the free magnetic layer antiparallel to the magnetization of the fixed magnetic layer. The memory cell is configured to switch from a first state to a second state based on whether the MJJ is in a zero state or in a π state.
Need to check novelty before this filing date? Find Prior Art

Description

Background Art

[0001] Semiconductor-based integrated circuits used in electronic devices, such as random access memory, include digital circuits based on complementary metal oxide semiconductor (CMOS) technology. However, CMOS technology is reaching its limits in terms of device size. In addition, leakage currents in CMOS-based memories cause high power consumption even when they are not being accessed.

[0002] For example, servers in data centers are increasingly consuming large amounts of power. Part of the power consumed is power loss due to energy dissipation, even when the CMOS circuits are inactive. This is because even when such circuits (such as random access memory) are inactive and do not consume any dynamic power, these circuits still consume power due to the need to maintain the state of the CMOS transistors. In addition, because CMOS circuits are powered using DC voltages, there is a certain amount of current leakage even when the CMOS circuits are inactive. Therefore, even when such circuits are not processing operations such as read / write, power is wasted due to the need to maintain the state of the CMOS transistors and the current leakage.

[0003] An alternative to memories based on CMOS technology is memories based on superconducting logic. Summary of the invention

[0004] In one example, the present disclosure is directed to a memory cell including a first inductor and a magnetic Josephson junction (MJJ) coupled to the first inductor to form a loop. The MJJ may include at least a first layer formed above a second layer and a third layer formed below the second layer, wherein the first layer is a free magnetic layer, the second layer is a non-magnetic layer, wherein the third layer is a fixed magnetic layer, wherein the memory cell is configured to be in a first state or a second state, and wherein the first state corresponds to a first magnetization configuration of the free magnetic layer and the second state corresponds to a second magnetization configuration of the free magnetic layer, wherein the first magnetization configuration of the free magnetic layer corresponds to a first magnetization that is parallel to a magnetization of the fixed magnetic layer and the second magnetization configuration of the free magnetic layer corresponds to a second magnetization that is antiparallel to the magnetization of the fixed magnetic layer, and wherein the memory cell is configured to switch from the first state of the memory cell to the second state of the memory cell in response to a current flowing through the MJJ based on whether the MJJ is in a zero state or in a π state, wherein the current flowing through the MJJ is configured to apply a torque to the magnetization of the free magnetic layer to cause switching from the first state of the memory cell to the second state of the memory cell, and wherein the current is generated in response to applying flux to the loop via the first inductor.

[0005] In another aspect, a method in a memory cell is provided, the memory cell comprising a first inductor and a magnetic Josephson junction (MJJ) coupled to the first inductor to form a loop, wherein the MJJ comprises at least a first layer formed above a second layer and a third layer formed below the second layer, wherein the first layer is a free magnetic layer, the second layer is a non-magnetic layer, wherein the third layer is a fixed magnetic layer, wherein the memory cell is configured to be in a first state or a second state, and wherein the first state corresponds to a first magnetization configuration of the free magnetic layer and the second state corresponds to a second magnetization configuration of the free magnetic layer, wherein the first magnetization configuration of the free magnetic layer corresponds to a first magnetization parallel to a magnetization of the fixed magnetic layer, and the second magnetization configuration of the free magnetic layer corresponds to a second magnetization antiparallel to the magnetization of the fixed magnetic layer. The method may include receiving a write pulse via a write bit line. The method may further include switching from a first state of the memory cell to a second state of the memory cell based on whether the MJJ is in a zero state or in a π state in response to a current flowing through the MJJ, wherein the current flowing through the MJJ is configured to apply a torque to the magnetization of the free magnetic layer to cause switching from the first state of the memory cell to the second state of the memory cell, and wherein the current is generated in response to applying flux to the loop via the first inductor.

[0006] In yet another aspect, the present disclosure relates to a memory cell including a first inductor and a magnetic Josephson junction (MJJ). The MJJ may include at least a first layer formed above a second layer and a third layer formed below the second layer, wherein the first layer is a free magnetic layer, the second layer is a non-magnetic layer, and the third layer is a fixed magnetic layer, wherein the memory cell is configured to be in a first state or a second state, and wherein the first state corresponds to a first magnetization configuration of the free magnetic layer and the second state corresponds to a second magnetization configuration of the free magnetic layer, wherein the first magnetization configuration of the free magnetic layer corresponds to a first magnetization parallel to a magnetization of the fixed magnetic layer and the second magnetization configuration of the free magnetic layer corresponds to a second magnetization antiparallel to a magnetization of the fixed magnetic layer, and wherein the memory cell is configured to switch from the first state of the memory cell to the second state of the memory cell based on a torque applied to the magnetization of the free magnetic layer in response to whether the MJJ is in a zero state or in a π state.

[0007] The Summary of the Invention is provided to introduce a series of concepts in a simplified form, which will be further described in the Detailed Description below. The Summary of the Invention is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter. For example, terms such as first layer, second layer, and third layer are merely used to indicate that there is at least one free magnetic layer on one side of at least one fixed magnetic layer and at least one free magnetic layer on the other side of at least one fixed magnetic layer. These terms do not imply a specific arrangement or a particular order in which these layers are created. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The present disclosure is illustrated in an exemplary manner and is not limited by the accompanying drawings, in which like reference numerals indicate like elements. Elements in the accompanying drawings are illustrated for simplicity and clarity, and these elements are not necessarily drawn to scale.

[0009] Figure 1 A diagram showing a memory cell according to an example;

[0010] Figure 2 shows a diagram of another memory cell according to one example;

[0011] Figure 3 is a graph depicting a relationship between a superconducting phase of a MJJ corresponding to a memory cell and a flux applied to a SQUID in a memory cell according to an example;

[0012] Figure 4A and Figure 4B shows a graph depicting the relationship between a switching field of a MJJ corresponding to a memory cell and a flux applied to a SQUID in a memory cell according to an example;

[0013] FIG. 5A to FIG. 5C According to one example, a simulated flipping of the magnetization of an MJJ corresponding to a memory cell based on applying flux via a write word line and a write bit line is shown, and the effect on the critical current of the MJJ is shown;

[0014] FIG. 6A to FIG. 6C According to another example, simulated flipping of the magnetization of an MJJ corresponding to a memory cell based on application of flux via a write word line and a write bit line is shown, and the effect on the critical current of the MJJ is shown; and

[0015] Figure 7 A computing system including a memory coupled to a processor is shown according to one example. DETAILED DESCRIPTION

[0016] Examples described in the present disclosure relate to a memory system based on superconducting logic, which includes a superconducting memory with a torque based on a Josephson phase. Certain examples also relate to a Josephson magnetic random access memory (JMRAM) with a torque based on a Josephson phase. Certain examples also relate to a reciprocal quantum logic (RQL) compatible RAM. Unlike CMOS transistors, RQL circuits are superconductor circuits using devices based on Josephson junctions. An exemplary Josephson junction may include two superconductors coupled via a region that hinders current flow. The region that hinders current flow may be a physical narrowing of the superconductor itself, a metallic region, or a thin insulating barrier. For example, a superconductor-insulator-superconductor (SIS) type Josephson junction may be implemented as part of an RQL circuit. For example, a superconductor is a material that can carry a direct current (DC) in the absence of an electric field. This material has zero resistance. For example, niobium is superconducting at temperatures below Tc (e.g., 9.3K); however, at temperatures above Tc, niobium behaves as a normal metal with resistance. Thus, in a SIS type Josephson junction, the superconductor may be a niobium superconductor, and the insulator may be alumina. In a SIS type junction, superconducting electrons are described by a quantum mechanical wave function. The phase difference of the phase of the superconducting electron wave function between two superconductors in time corresponds to the potential difference between the two superconductors. In an RQL circuit, in one example, an SIS type junction may be part of a superconducting loop. When the time integral of the potential difference between the two superconductors is integrated relative to time over a period of phase change, the magnetic flux passing through the loop changes in integer multiples of a single magnetic flux quantum. A voltage pulse associated with a single magnetic flux quantum is referred to as a single flux quantum (SFQ) pulse. For example, an overdamped Josephson junction may create a separate single flux quantum (SFQ) pulse. In an RQL circuit, each Josephson junction may be part of one or more superconducting loops. The phase difference across the junction may be modulated by the magnetic flux applied to the loop.

[0017] A variety of RQL circuits including transmission lines can be formed by coupling multiple Josephson junctions through inductors or other components as needed. SFQ pulses can be transmitted via these transmission lines under the control of at least one clock. SFQ pulses can be positive or negative. For example, when a sinusoidal bias current is provided to the junction, both positive and negative pulses can be transmitted to the right during opposite clock phases on the transmission line. Since there is no bias resistor, the RQL circuit can advantageously have zero static power dissipation. In addition, the RQL circuit can be powered by using alternating current (AC) power. The AC power supply can also serve as a stable clock reference signal for the RQL circuit. In one example, digital data can be encoded by using a pair of positive and negative (reciprocal) SFQ pulses. For example, one bit of logic can be encoded as a pair of reciprocal SFQ pulses generated in the positive and negative phases of a sinusoidal clock. A logical zero bit can be encoded by having no positive / negative pulse pairs during a clock cycle. A positive SFQ pulse can arrive during the positive portion of the clock, while a negative pulse can arrive during the negative portion of the clock.

[0018] Microwave signals (e.g., SFQ pulses) can be used to control the state of a memory cell. During a read / write operation, word lines and bit lines can be selectively activated by SFQ pulses arriving via an address bus. These pulses can then control word line and bit line drivers that can provide word line and bit line currents to associated memory cells. For example, such a memory cell can be read out using a superconducting quantum interference device (SQUID). The memory cell can include a SQUID and a magnetic Josephson junction (MJJ) device that can be configured so that the memory cell can be in a logic "1" state or a logic "0" state when an appropriate amount of current bias and magnetic flux are applied. In one example, if the memory cell is in a logic "1" state, the SQUID can be converted to a "voltage state" when a current is applied via the word line. A sense amplifier coupled to the memory cell can sense a voltage representing a logic "1" state. The logic "0" state can correspond to a "substantially zero voltage state" so that in a logic "0" state, the SQUID can remain in a "substantially zero voltage state" despite the application of a current via the word line. A sense amplifier can sense this condition which represents a logical "0" state.

[0019] The memory cells may be arranged in rows and columns such that each row may be activated by a common flux bias (e.g., a read word line signal) and each bit line may form a transmission line that may propagate the output of the memory cell at a voltage state to a sense amplifier at one end of the column. The memory cells in a column may be continuously biased by a common current source (e.g., a flux pump).

[0020] In Josephson magnetic random access memory (JMRAM), memory operation is based on storing information in the relative alignment of magnetic layers in a superconducting spin valve junction. Memory readout is based on the Josephson effect; for example, the parallel or antiparallel alignment of the magnetization of the (fixed and free) layers indicates a "0 junction" in a first state and a "pi junction" in a second state. Alternatively, memory readout is based on the Josephson effect; for example, the parallel or antiparallel alignment of the magnetization of the (fixed and free) layers indicates a "pi junction" in a first state and a "0 junction" in a second state. In JMRAM, write operations are accomplished via a field flip of the free layer relative to the fixed layer. In order to ensure low-power operation, JMRAM may be required to meet certain power efficiencies. The power efficiency requirement results in the flip field being small, for example of the order of 10 Oe. JMRAM magnetic junctions may be required to have a relatively high critical current density; for example, Jc>20 kA / cm2 may be considered acceptable.

[0021] In a magnetic Josephson junction with parameters consistent with energy-efficient operation (e.g., parameters related to the switching field and critical current), the Josephson effect plays an important role in the dynamics of magnetic switching. In particular, the Josephson effect creates a torque on the free magnetic layer that can act to change the orientation of the magnetization of the free magnetic layer. The magnitude and direction of this torque depends on the superconducting phase induced across the magnetic Josephson junction. Certain example memory cells described herein use this Josephson phase-based torque to assist or induce the magnetic switching of a JMRAM bit from one state to another. For example, this can be accomplished by embedding a magnetic Josephson junction in an RF-SQUID circuit and controlling the phase of the Josephson junction via mutual inductance by applying a flux pulse coupled to the RF SQUID.

[0022] Figure 1 1 shows a diagram of a memory cell 100 according to an example. In one example, as Figure 1As shown in , the memory cell 100 may include a memory storage element 102. The memory storage element includes at least one magnetic Josephson junction (MJJ) device 110 and at least two Josephson junctions 112 and 114. In one example, each of the Josephson junctions 112 and 114 may be a superconductor-insulator-superconductor (SIS) type Josephson junction. The memory storage element 102 may also include a transformer 116. The transformer 116 may include two inductors 118 and 120. The memory storage element 102 may also include another transformer 122. The transformer 122 may include inductors 124 and 126. In one example, the Josephson junctions 112 and 114 may form a readout superconducting quantum interference device (SQUID). In one example, the MJJ device 110 may be a superconductor having at least two magnetic barrier layers (e.g., ferromagnetic barriers). In one example, the at least two magnetic barrier layers (e.g., ferromagnetic barriers) may be compatible with circuits based on single flux quantum (SFQ) pulses. In another example, there may be two magnetic barrier layers, one of which has a fixed magnetic state and the other of which has a bistable magnetic state. In one example, the MJJ device 110 may include a non-magnetic layer sandwiched between a free magnetic layer and a fixed magnetic layer. There may be an intermediate layer between any of these layers. In one example, in order to allow the magnetization direction to be flipped in response to a small magnetic field, the free magnetic layer may have very soft magnetic properties. The free magnetic layer may be a ferromagnetic metal or an alloy of several magnetic and non-magnetic materials. The fixed magnetic layer may also be a metal or an alloy. In another example, the fixed magnetic layer may have a lower amount of doping compared to the free magnetic layer. The fixed magnetic layer may have a hysteresis greater than that of the free magnetic layer. The thickness of the free magnetic layer may be selected so that the MJJ device can transition between a "zero state" and a "π state". For example, the magnetic layer may have a thickness of 10 angstroms to 25 angstroms. The magnetic alloy may be a nickel-iron alloy, a nickel-cobalt alloy, an iron-cobalt alloy, or a cobalt-nickel-iron alloy. In another example, the magnetic layer may include a nickel-iron alloy doped with zirconium (Zr). In other examples, the magnetic layer may include a nickel-iron alloy doped with molybdenum or hafnium. The non-magnetic layer may include at least one of vanadium (V), molybdenum (Mo), copper (Cu), aluminum (Al), tantalum (Ta), or chromium (Cr). Any of the layers described as part of the present disclosure may be formed using a physical vapor deposition (PVD) technique such as sputtering. Although a certain number of layers of the MJJ device 110 arranged in a certain manner are described, there may be more or fewer layers arranged in a different manner.

[0023] Continue to refer Figure 1, the memory cell 100 can be coupled to word lines and bit lines for performing various memory operations, such as read and write operations. For example, a read word line (RWL) for performing a read operation can be coupled to the memory cell 100. In this example, the read word line (RWL) can be coupled via a transformer 122. A write word line (WWL) for performing a write operation can be coupled to the memory cell 100. The write word line can be magnetically coupled to the MJJ 110. In addition, a read bit line (RBL) for performing a read operation can be coupled to the memory cell 100. A write bit line (WBL) for performing a write operation can also be coupled to the memory cell 100. In this example, the write bit line (WBL) can be coupled via a transformer 116. The inductive coupling of the write bit line (WBL) can advantageously make the geometry of the memory cell more favorable. In this example, the flux applied to the loop including the MJJ 110 (the loop formed by the inductor 120 and the MJJ 110) via the inductive coupling is used as a control mechanism for changing the state of the memory cell 100. In one example, coupling with the MJJ 110 can enable the magnetization of the magnetic barrier layer of the MJJ to be changed by applying a local write word line current and by applying a write bit line current.

[0024] In one example, the MJJ 110 can be in a first state (e.g., corresponding to a first magnetization configuration of at least one magnetic barrier layer) and a second state (e.g., corresponding to a second magnetization configuration of at least one magnetic barrier layer), wherein the first magnetization configuration can be substantially different from the second magnetization configuration. In one example, the MJJ 110 can be in a state when the magnetizations of each of the at least two magnetic barrier layers are opposite to each other. In one example, the state can be a "zero state" of the MJJ 110. The MJJ 110 can be in a different state when the magnetizations of each of the at least two magnetic barrier layers support each other. In one example, the state can be a "π state". In the "π state", MJJ 110 can apply flux to JJs 112 and 114, so that when a word read current is applied, the induced current in the memory storage element 102 - due to the flux generated by MJJ 110 and by the coupling of the word read current via transformer 122 - can combine with the applied bit read current to trigger the Josephson junctions 112 and 114, causing the read SQUID to enter a voltage state. This voltage can generate a current along the read bit line, which can act as a transmission line with a certain impedance. This current can be sensed using a sense amplifier.

[0025] In one example, during a write operation, a write word line (WWL) and a write bit line (WBL) may receive current from respective drivers. When the magnetizations of the magnetic layers support each other, the MJJ 110 may then be in a "π state". Conversely, when the respective magnetizations of the layers are opposite to each other, the MJJ 110 may then be in a "zero state". In one example, the "π state" may correspond to a logical "1" state, and the "zero state" may correspond to a logical "0" state of the memory cell 100.

[0026] like Figure 1 As shown in , MJJ 110 and inductor 120 can form a loop so that any current applied to the write bit line results in inductive coupling of the current to the loop. In other words, in this example, flux is applied to the loop around MJJ 110. The total energy density of the magnetic junction can be written as:

[0027] U=U H +U K +U C +U J

[0028] Among them U H is the energy associated with the applied magnetic field, including the Zeeman energy and the effect of the demagnetization field of the two layers in the spin valve, U K describes the anisotropy energy of the magnetic layer, U C describes the coupling between layers (e.g., magnetostatic layers or exchange layers), and U J is the Josephson energy. In this example, the torque on each of the magnetic layers can be written as:

[0029]

[0030] in is the unit vector in the magnetization direction of the corresponding layer. The change in the magnetization of the magnetic layer is then described by the Landau-Lifshitz-Gilbert equations.

[0031] The Josephson effect in a spin valve depends on the relative orientation of the magnetization in the two layers. In this example, the energy term can be written as:

[0032]

[0033] Where E J is the Josephson energy E J =Φ0I c / 2π, V is the volume of the bit, and δ is the superconducting phase difference across the junction. In addition,

[0034]

[0035] in is related to the evolution of the supercurrent phase in the ferromagnetic layer, d F1(F2) is the thickness of the corresponding magnetic layer, and ξ F1(F2) is the oscillation period of the superconducting phase due to energy exchange in the corresponding magnetic layer. If the layer thickness is exactly An odd multiple of , so the energy term can be written in spherical coordinates as:

[0036]

[0037] where θ 1,2 and φ 1,2 is the polar angle and direction angle of the magnetization vector of the magnetic layer, and results in a zero junction or a pi junction depending on the parallel state, with η being equal to +1 or -1. It follows that the Josephson torque term is (written in spherical coordinates for each magnetic layer in the magnetic layer):

[0038]

[0039] in,

[0040]

[0041] Therefore, by embedding a magnetic junction in the SQUID loop and transmitting the magnetic field to the SQUID through a superconducting transformer (e.g., Figure 1 By applying flux to the loop using a transformer 116 in the junction, the phase of the junction (δ in the above equation) can be controlled and can be used to affect magnetic flipping. This is because a change in the phase of the junction results in a change in the Josephson energy, which in turn changes the magnitude and / or direction of the torque applied to the magnetization of the free magnetic layer.

[0042] To read the state of the memory cell, a change in output voltage, current, or any other parameter of the memory storage element 102 may be sensed by using a sense amplifier. In one example, once amplified by the inductive sensor, the presence or absence of a current pulse may determine the state of the memory cell 100 as a logical "0" or a logical "1". For example, a logical "1" state may correspond to a "voltage state" where the inductive sensor coupled to the memory cell may sense a voltage representing a logical "1" state. A logical "0" state may correspond to a "substantially zero voltage state" such that a sense amplifier may sense such a condition representing a logical "0" state. Although Figure 1 A certain number of components are shown arranged in a certain manner, but the memory cell 100 may include additional or fewer components arranged in a different manner. Figure 1The write bit line (WBL) is shown coupled via an inductor, but the write word line (WWL) may instead be coupled via an inductor.

[0043] Figure 2 2 shows a diagram of a memory cell 200 according to another example. In one example, Figure 2 As shown in , the memory cell 200 may include a memory storage element 202, which includes at least one magnetic Josephson junction (MJJ) device 210 and at least two Josephson junctions 212 and 214. In one example, each of the Josephson junctions 212 and 214 can be a superconductor-insulator-superconductor (SIS) type Josephson junction. The memory storage element 202 may also include a transformer 216. The transformer 116 may include two inductors 218 and 220. The memory storage element 202 may also include another transformer 222. The transformer 222 may include inductors 224 and 226. In one example, the Josephson junctions 212 and 214 may form a readout superconducting quantum interference device (SQUID). In one example, the MJJ device 210 may be a superconductor having at least two magnetic barrier layers (e.g., ferromagnetic barriers).

[0044] Continue to refer Figure 2 , the memory cell 200 can be coupled to word lines and bit lines for performing various memory operations, such as read and write operations. For example, a read word line (RWL) for performing a read operation can be coupled to the memory cell 200. In this example, the read word line (RWL) can be coupled via a transformer 222. A write word line (WWL) for performing a write operation can be coupled to the memory cell 200. The write word line can be magnetically coupled to the MJJ 210. In addition, a read bit line (RBL) for performing a read operation can be coupled to the memory cell 200. Two write bit lines (WBL1 and WBL2) for performing a write operation can also be coupled to the memory cell 200. As shown in FIG. Figure 2 , one of the write bit lines (WBL1) can be magnetically coupled to MJJ 210. In this example, the other write bit line (WBL2) can be coupled via transformer 216. In this example, the flux applied to the loop including MJJ 210 (the loop formed by inductor 220 and MJJ 210) via this inductive coupling is used as a control mechanism for changing the state of memory cell 200. In one example, coupling with MJJ 110 can enable the magnetization of the magnetic barrier layer of the MJJ to be changed by applying a local write word line current and by applying a write bit line current via WBL1 and WBL2.

[0045] In one example, the MJJ 210 can be in a first state (e.g., corresponding to a first magnetization configuration of at least one magnetic barrier layer) and a second state (e.g., corresponding to a second magnetization configuration of at least one magnetic barrier layer), wherein the first magnetization configuration can be substantially different from the second magnetization configuration. In one example, the MJJ 210 can be in a state when the magnetizations of each of the at least two magnetic barrier layers are opposite to each other. In one example, the state can be a "zero state" of the MJJ 210. The MJJ 210 can be in a different state when the magnetizations of each of the at least two magnetic barrier layers support each other. In one example, the state can be a "π state". In the "π state", MJJ 210 can apply flux to JJs 212 and 214, so that when a word read current is applied, the induced current in the memory storage element 202 - due to the flux generated by MJJ 210 and by the coupling of the word read current via transformer 222 - can combine with the bit read current to trigger the Josephson junctions 212 and 214, causing the read SQUID to enter a voltage state. This voltage can generate a current along the read bit line, which can act as a transmission line with a certain impedance. This current can be sensed using a sense amplifier.

[0046] In one example, during a write operation, the write word line (WWL) and the write bit line (WBL1 and WBL2) can receive current from corresponding drivers. When the magnetizations of the magnetic layers support each other, the MJJ 210 can then be in a "π state". Conversely, when the corresponding magnetizations of the layers are opposite to each other, the MJJ 210 can then be in a "zero state". In one example, the "π state" can correspond to a logical "1" state, and the "zero state" can correspond to a logical "0" state of the memory cell 200.

[0047] Figure 3 300 is a graph depicting the relationship between the junction phase of an MJJ (e.g., MJJ 110 or MJJ 210) corresponding to a memory cell and the flux applied to the SQUID in the memory cell according to one example. In this example, the applied flux (Φ0) is shown along the horizontal axis of the graph 300, and the change in the junction phase (δ / π) of the MJJ is shown along the vertical axis. The graph shows the change in the junction phase for both the 0 junction (solid line 314) and the π junction (dashed line 312). At a phase of Φ0 / 2 for the 0 junction and at a phase of 0 for the π junction, the SQUID becomes bistable.

[0048] Figure 4A A diagram 400 is shown and Figure 4B Graph 450 is shown, and graphs 400 and 450 depict the switching field (H) of the MJJ corresponding to the memory cell. SW) and the flux (Φ0) applied to the SQUID in a memory cell according to one example. Graph 400 shows this relationship for a junction that is a 0 junction when the magnetizations of both the free layer and the fixed layer are parallel to each other, and graph 450 shows this relationship for a junction that is a π junction when the magnetizations of both the free layer and the fixed layer are parallel to each other. As shown in both graphs 400 and 450, in this example, the flux applied to the junction (e.g., applied via a write bit line inductively coupled to the loop) changes not only the width of the flip loop, but also the center of the flip loop.

[0049] FIG. 5A to FIG. 5C According to one example, a simulated flipping of the magnetization of an MJJ corresponding to a memory cell based on applying flux via a write word line and a write bit line is shown, and the effect on the critical current flowing through the MJJ is shown. In this example, the junction in the MJJ is in a 0 state when the magnetization of the free layer is parallel to the magnetization of the fixed layer. Panel 510 shows a profile of a write signal applied to the memory cell. The solid line 512 in panel 510 shows a write pulse applied to the MJJ via the write word line (as a function of time) in response to applying current via the write word line. The dashed line 514 in panel 510 shows the flux induced in the MJJ (as a function of time) as a result of applying current via the write bit line inductively coupled to a loop including the MJJ. Panel 520 ( Figure 5B ) shows the change in the junction phase of the MJJ during the same duration as panel 510. Curve 522 in panel 520 shows the change in the easy axis magnetization (M) of the free layer of the MJJ during the same duration as panel 510. x Curve 524 shows the hard axis magnetization (M) of the free layer of the MJJ during the same duration as the panel 510. y ). In this example, in response to a combined flux pulse applied word line pulse (applied via a write bit line), the easy axis magnetization (M x ) can be controlled from M x = -1 (parallel alignment of magnetization of free and fixed layers) flips to M x = +1 (antiparallel alignment of magnetizations of the free and fixed layers), and then back to M x = -1. Panel 530 ( Figure 5C ) shows the time domain evolution of the critical current of the MJJ during the same duration as panel 510. Curve 532 shows the critical current when the junction is in the 0 state, and curves 534 and 536 show the critical current when the junction is in the π state.

[0050] FIG. 6A to FIG. 6CAccording to another example, a simulated flipping of the magnetization of an MJJ corresponding to a memory cell based on applying flux via a write word line and a write bit line is shown, and the effect on the critical current flowing through the MJJ is shown. In this example, the junction in the MJJ is in a π state when the magnetization of the free layer is parallel to the magnetization of the fixed layer. Panel 610 shows a profile of a write signal applied to the memory cell. The solid line 612 in panel 610 shows a write pulse applied to the MJJ via the write word line (as a function of time) in response to applying a current via the write word line. The dashed line 614 in panel 610 shows the flux induced in the MJJ (as a function of time) as a result of applying a current via a write bit line inductively coupled to a loop including the MJJ. Panel 620 ( Figure 6B Curve 626 in panel 620 shows the change in the junction phase of the MJJ during the same duration as panel 610. Curve 622 in panel 620 shows the change in the easy axis magnetization (M) of the free layer of the MJJ during the same duration as panel 610. x Curve 624 shows the hard axis magnetization (M) of the free layer of the MJJ during the same duration as the panel 610. y ). In this example, in response to a combined flux pulse applied word line pulse (applied via a write bit line), the easy axis magnetization (M x ) can be controlled from M x = -1 (parallel alignment of magnetization of free and fixed layers) flips to M x = +1 (antiparallel alignment of magnetizations of the free and fixed layers), and then back to M x = -1. Panel 630 ( Figure 6C ) shows the time domain evolution of the critical current flowing through the MJJ during the same duration as panel 610. Curve 632 shows the critical current when the junction is in the 0 state, and curves 634 and 636 show the critical current when the junction is in the π state.

[0051] Figure 7 A computing system including a memory coupled to a processor is shown according to one example. Figure 7A computing system 700 including a processor 710 coupled to a memory 720 according to one example is shown. The processor 710 can perform read or write operations on the memory 720 in a manner as explained above. In addition, the processor 710 and the memory 720 can be used together with other superconducting logic-based devices. Generally, any superconducting device that operates in a cryogenic environment and requires storage of instructions or data may include the memory 720. In addition, the processor 710 does not need to be in a cryogenic environment; on the contrary, the processor 710 can operate at a non-cryogenic temperature. In this example, the memory 720 can be in a separate cryogenic environment and can be coupled to the processor 710 via a connector in a manner that can maintain the cryogenic environment. The memory 720 can be used as part of the storage in a data center for providing cloud-based services (such as software as a service, platform as a service) or other services. The memory system 720 may include an array 702 of memory cells arranged in rows and columns. In one example, the array 702 may be an array of memory cells, for example, the memory cells 100 and 200 shown above. The memory system 720 may also include a row decoder 718, which may be configured to decode row control / address signals. The row decoder 718 may also be coupled to a word line driver 722. The word line driver 722 may include circuitry to provide a word line read / write current to a subset or all of the memory cells associated with the selected word line for any read or write operation. The word line driver 722 may provide such current via a word line 724.

[0052] The memory system 720 may also include a bus interface 730, which may be configured to receive control signals or other signals from the processor 710. The bus interface 730 may also be coupled to a bit line driver 732. The bit line driver 732 may include a circuit device to provide a bit line read current to a subset or all of the memory cells associated with the selected bit line for any read or write operation. The bit line driver 732 may provide such current via a bit line 734. The bit line 734 may include both a read bit line and a write bit line. In other words, different bit lines may be used to provide current to the selected memory cell for a read or write operation. By using row and column addresses, any memory cell in the memory cell may be accessed by using an address. Each bit line in the bit line (e.g., bit line 734) may also be coupled to a sense amplifier 704 for sensing the bit line to determine the logic state of each memory cell in the array 702 of memory cells.

[0053] The coupling between the array 702 of memory cells and the sense amplifier 704 may include a radio frequency (RF) transmission line. The memory cells in each column may be continuously current biased by a common current source (e.g., a flux pump). As previously described, the bit line 734 may be used to couple this current to each of the memory cells in the column. Although Figure 7 A certain number of components of computing system 700 are shown arranged in a certain manner, but there may be a greater or lesser number of components arranged in a different manner.

[0054] In summary, the present disclosure is directed to a memory cell including a first inductor and a magnetic Josephson junction (MJJ) coupled to the first inductor to form a loop. The MJJ may include at least a first layer formed above a second layer and a third layer formed below the second layer, wherein the first layer is a free magnetic layer, the second layer is a non-magnetic layer, wherein the third layer is a fixed magnetic layer, wherein the memory cell is configured to be in a first state or a second state, and wherein the first state corresponds to a first magnetization configuration of the free magnetic layer and the second state corresponds to a second magnetization configuration of the free magnetic layer, wherein the first magnetization configuration of the free magnetic layer corresponds to a first magnetization that is parallel to a magnetization of the fixed magnetic layer and the second magnetization configuration of the free magnetic layer corresponds to a second magnetization that is antiparallel to the magnetization of the fixed magnetic layer, and wherein the memory cell is configured to switch from the first state of the memory cell to the second state of the memory cell in response to a current flowing through the MJJ based on whether the MJJ is in a zero state or in a π state, wherein the current flowing through the MJJ is configured to apply a torque to the magnetization of the free magnetic layer to cause switching from the first state of the memory cell to the second state of the memory cell, and wherein the current is generated in response to applying flux to the loop via the first inductor.

[0055] The first inductor may be inductively coupled to a write bit line. The write bit line may be configured to receive a write current as part of a write operation associated with the memory cell. The MJJ may be magnetically coupled to the write bit line.

[0056] The current flowing through the MJJ is configured to apply a torque to the magnetization of the free magnetic layer, the torque being caused by the change in the superconducting phase of the MJJ from the zero state to the π state. The memory cell can be configured to switch from a first state of the memory cell to a second state of the memory cell in response to the torque being applied to the magnetization of the free magnetic layer. The magnitude of the torque and the direction of the torque can depend on the superconducting phase of the MJJ.

[0057] In another aspect, a method in a memory cell is provided, the memory cell comprising a first inductor and a magnetic Josephson junction (MJJ) coupled to the first inductor to form a loop, wherein the MJJ comprises at least a first layer formed above a second layer and a third layer formed below the second layer, wherein the first layer is a free magnetic layer, the second layer is a non-magnetic layer, wherein the third layer is a fixed magnetic layer, wherein the memory cell is configured to be in a first state or a second state, and wherein the first state corresponds to a first magnetization configuration of the free magnetic layer and the second state corresponds to a second magnetization configuration of the free magnetic layer, wherein the first magnetization configuration of the free magnetic layer corresponds to a first magnetization parallel to a magnetization of the fixed magnetic layer, and the second magnetization configuration of the free magnetic layer corresponds to a second magnetization antiparallel to the magnetization of the fixed magnetic layer. The method may include receiving a write pulse via a write bit line. The method may further include switching from a first state of the memory cell to a second state of the memory cell based on whether the MJJ is in a zero state or in a π state in response to a current flowing through the MJJ, wherein the current flowing through the MJJ is configured to apply a torque to the magnetization of the free magnetic layer to cause switching from the first state of the memory cell to the second state of the memory cell, and wherein the current is generated in response to applying flux to the loop via the first inductor.

[0058] The first inductor may be inductively coupled to a write bit line. The write bit line may be configured to receive a write current as part of a write operation associated with the memory cell. The MJJ may be magnetically coupled to the write bit line.

[0059] The current flowing through the MJJ is configured to apply a torque to the magnetization of the free magnetic layer, the torque being caused by the change in the superconducting phase of the MJJ from the zero state to the π state. The memory cell can be configured to switch from a first state of the memory cell to a second state of the memory cell in response to the torque being applied to the magnetization of the free magnetic layer. The magnitude of the torque and the direction of the torque can depend on the superconducting phase of the MJJ.

[0060] In yet another aspect, the present disclosure relates to a memory cell, the memory cell comprising a first inductor and a magnetic Josephson junction (MJJ) coupled to the first inductor to form a loop. The MJJ may include at least a first layer formed above a second layer and a third layer formed below the second layer, wherein the first layer is a free magnetic layer, the second layer is a non-magnetic layer, wherein the third layer is a fixed magnetic layer, wherein the memory cell is configured to be in a first state or a second state, and wherein the first state corresponds to a first magnetization configuration of the free magnetic layer, and the second state corresponds to a second magnetization configuration of the free magnetic layer, wherein the first magnetization configuration of the free magnetic layer corresponds to a first magnetization parallel to a magnetization of the fixed magnetic layer, and the second magnetization configuration of the free magnetic layer corresponds to a second magnetization antiparallel to a magnetization of the fixed magnetic layer, and wherein the memory cell is configured to switch from the first state of the memory cell to the second state of the memory cell based on a torque applied to the magnetization of the free magnetic layer in response to whether the MJJ is in a zero state or in a π state.

[0061] The memory cell may further include a first inductor, wherein the first inductor is inductively coupled to a write bit line. The first inductor may be inductively coupled to the write bit line. The write bit line may be configured to receive a write current as part of a write operation associated with the memory cell. The MJJ may be magnetically coupled to the write word line.

[0062] Due to the change of the superconducting phase of the MJJ, a torque can be applied to the magnetization of the free magnetic layer. The magnitude of the torque and the direction of the torque depend on the superconducting phase of the MJJ.

[0063] It should be understood that the methods, modules, and components described in the present disclosure are merely exemplary. Alternatively, or additionally, the functions described herein may be performed at least in part by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that can be used include: field programmable gate array (FPGA), application specific integrated circuit (ASIC), application specific standard product (ASSP), system-on-chip system (SOC), complex programmable logic device (CPLD), etc. In an abstract but still definite sense, any arrangement of components that implement the same function is effectively "associated" so that the desired function is achieved. Therefore, any two components that are combined to achieve a specific function herein can be regarded as "associated" with each other so that the desired function is achieved, without considering architecture or intermediate components. Similarly, any two components so associated can also be regarded as "operably connected" or "coupled" to each other to achieve the desired function.

[0064] Functions associated with the examples described in this disclosure may also include instructions stored in non-transient media. As used herein, the term "non-transient media" refers to any medium that stores data and / or causes a machine such as processor 710 to operate in a specific manner. Exemplary non-transient media include non-volatile media and / or volatile media. Non-volatile media, for example, include hard disks, solid-state drives, disks or tapes, optical or optical tapes, flash memory, EPROM, NVRAM, PRAM, or other such media, or networked versions of such media. Volatile media, for example, include dynamic memory (such as DRAM, SRAM, cache) or other such media. Non-transient media are different from transmission media, but can be used in combination with transmission media. Transmission media are used to transmit data and / or instructions to or from a machine. Exemplary transmission media include coaxial cables, fiber optic cables, copper wires, and wireless media, such as radio waves.

[0065] In addition, those skilled in the art will recognize that the boundaries between the functions of the operations described above are merely illustrative. The functions of multiple operations can be combined into a single operation, and / or the functions of a single operation can be distributed in additional operations. In addition, alternative embodiments can include multiple instances of specific operations, and in various other embodiments, the order of operations can be changed.

[0066] Although the present disclosure provides specific examples, various modifications and changes may be made without departing from the scope of the present disclosure as set forth in the appended claims. Therefore, the specification and drawings should be regarded as illustrative rather than restrictive, and all such modifications are intended to be included within the scope of the present disclosure. Any advantages, advantages, or solutions to problems described herein with respect to specific examples are not intended to be interpreted as key, required, or essential features or elements of any or all claims.

[0067] Furthermore, the terms "a" or "an" as used herein are defined as one or more than one. Likewise, the use of introductory phrases such as "at least one", "one or more" in the claims should not be interpreted as meaning that another claim element introduced by the indefinite article "a" or "an" limits any particular claim including such claim element introduced into the invention to only include one such element, even when the same claim includes the introductory phrases "one or more" or "at least one" and an indefinite article such as "a" or "an". The same is true for the use of definite articles.

[0068] Unless otherwise stated, terms such as "first" and "second" are used to arbitrarily distinguish between the elements such terms describe. Therefore, these terms are not necessarily intended to indicate spatial priority, temporal priority, or other priority of such elements.

Claims

1. A memory cell, comprising: a first inductor; as well as a magnetic Josephson junction MJJ coupled to the first inductor to form a loop, wherein the MJJ includes at least a first layer formed above a second layer and a third layer formed below the second layer, wherein the first layer is a free magnetic layer, the second layer is a non-magnetic layer, wherein the third layer is a fixed magnetic layer, wherein the memory cell is configured to be in a first state or a second state, and wherein the first state corresponds to a first magnetization configuration of the free magnetic layer, and the second state corresponds to a second magnetization configuration of the free magnetic layer; wherein the first magnetization configuration of the free magnetic layer corresponds to a first magnetization parallel to the magnetization of the fixed magnetic layer, and the second magnetization configuration of the free magnetic layer corresponds to a first magnetization parallel to the magnetization of the fixed magnetic layer a magnetization of a fixed magnetic layer that is antiparallel to a second magnetization, and wherein the memory cell is configured to switch from the first state of the memory cell to the second state of the memory cell in response to a current flowing through the MJJ based on whether the MJJ is in a zero state or in a π state, wherein in response to a change in a superconducting phase of the MJJ from the zero state to the π state, the current flowing through the MJJ is configured to apply a torque dependent on a Josephson phase to the magnetization of the free magnetic layer to cause the switching from the first state of the memory cell to the second state of the memory cell, and wherein the current is generated in response to applying a flux to the loop via the first inductor.

2. The memory cell of claim 1, wherein the first inductor is inductively coupled to a write bit line.

3. The memory cell of claim 2, wherein the write bit line is configured to receive a write current as part of a write operation associated with the memory cell. The memory cell of claim 2 , wherein the MJJ is magnetically coupled to a write word line.

5. The memory cell of claim 1, wherein a magnitude of the Josephson phase-dependent torque and a direction of the Josephson phase-dependent torque depend on the superconducting phase of the MJJ.

6. The memory cell of claim 1, wherein the memory cell is configured to switch from the first state of the memory cell to the second state of the memory cell in response to the Josephson phase-dependent torque being applied to the magnetization of the free magnetic layer.

7. A method in a memory cell, the memory cell comprising a first inductor and a magnetic Josephson junction (MJJ) coupled to the first inductor to form a loop, wherein the MJJ comprises at least a first layer formed above a second layer and a third layer formed below the second layer, wherein the first layer is a free magnetic layer, the second layer is a non-magnetic layer, wherein the third layer is a fixed magnetic layer, wherein the memory cell is configured to be in a first state or a second state, and wherein the first state corresponds to a first magnetization configuration of the free magnetic layer, and the second state corresponds to a second magnetization configuration of the free magnetic layer, wherein the first magnetization configuration of the free magnetic layer corresponds to a first magnetization parallel to the magnetization of the fixed magnetic layer, and the second magnetization configuration of the free magnetic layer corresponds to a second magnetization antiparallel to the magnetization of the fixed magnetic layer, the method comprising: receiving a write pulse via a write bit line; as well as and switching from the first state of the memory cell to the second state of the memory cell based on whether the MJJ is in a zero state or in a π state in response to a current flowing through the MJJ, wherein in response to a change in a superconducting phase of the MJJ from the zero state to the π state, the current flowing through the MJJ is configured to apply a torque dependent on a Josephson phase to the magnetization of the free magnetic layer to cause the switching from the first state of the memory cell to the second state of the memory cell, and wherein the current is generated in response to applying flux to the loop via the first inductor. The method of claim 7 , wherein the first inductor is inductively coupled to a write bit line.

9. The method of claim 8, wherein the write bit line is configured to receive a write current as part of a write operation associated with the memory cell.

10. The method of claim 8, wherein the MJJ is magnetically coupled to a write word line.

11. The method of claim 7, wherein the magnitude of the Josephson phase-dependent torque and the direction of the Josephson phase-dependent torque depend on the superconducting phase of the MJJ.

12. The method of claim 7, wherein the memory cell is configured to switch from the first state of the memory cell to the second state of the memory cell in response to the Josephson phase-dependent torque being applied to the magnetization of the free magnetic layer.

13. A memory cell comprising: A magnetic Josephson junction (MJJ) includes at least a first layer formed above a second layer and a third layer formed below the second layer, wherein the first layer is a free magnetic layer, the second layer is a non-magnetic layer, wherein the third layer is a fixed magnetic layer, wherein the memory cell is configured to be in a first state or a second state, and wherein the first state corresponds to a first magnetization configuration of the free magnetic layer, and the second state corresponds to a second magnetization configuration of the free magnetic layer, wherein the first magnetization configuration of the free magnetic layer corresponds to a first magnetization parallel to the magnetization of the fixed magnetic layer, and the second magnetization configuration of the free magnetic layer corresponds to a second magnetization antiparallel to the magnetization of the fixed magnetic layer, and wherein the memory cell is configured to switch from the first state of the memory cell to the second state of the memory cell based on a torque depending on a Josephson phase applied to the magnetization of the free magnetic layer in response to whether the MJJ is in a zero state or a π state.

14. The memory cell of claim 13, wherein the MJJ is coupled to a first inductor to form a loop, and wherein the first inductor is inductively coupled to a write bit line.

15. The memory cell of claim 14, wherein the write bit line is configured to receive a write current as part of a write operation associated with the memory cell.

16. The memory cell of claim 14, wherein the MJJ is magnetically coupled to a write word line.

17. The memory cell of claim 13, wherein the Josephson phase-dependent torque is applied to the magnetization of the free magnetic layer due to a change in a superconducting phase of the MJJ.

18. The memory cell of claim 17, wherein a magnitude of the Josephson phase-dependent torque and a direction of the Josephson phase-dependent torque depend on the superconducting phase of the MJJ.