Stacked magnetic devices
Techniques for improving magnetic device performance are provided. In one aspect, a magnetic device, e.g., a magnetic random access memory device, is provided which comprises a plurality of current carrying lines; and two or more adjacent stacked magnetic toggling devices sharing at least one of the plurality of current carrying lines in common and positioned therebetween. The magnetic device is configured such that at least one of the adjacent magnetic toggling devices toggles mutually exclusively of another of the adjacent magnetic toggling devices. In an exemplary embodiment, the magnetic device comprises a plurality of levels with each of the adjacent stacked magnetic toggling devices residing in a different level.
Latest IBM Patents:
The present invention relates to techniques for improving magnetic device performance and, more particularly, to improved magnetic device structures and methods for use thereof
BACKGROUND OF THE INVENTIONMagnetic devices, such as magnetic random access memories (MRAM), commonly employ a plurality of magnetic tunnel junction (MTJ) devices to store information, each MTJ device comprising one or more magnetic films, or layers. Typically, one of these magnetic layers is a pinned magnetic layer and another of these magnetic layers is a free magnetic layer.
Information is stored in such MTJs as an orientation of the magnetization of the free magnetic layer as compared to an orientation of the magnetization of the pinned magnetic layer. Namely, whereas the pinned magnetic layer has an orientation of magnetization that is fixed, e.g., by a variety of techniques, the free magnetic layer has an orientation of magnetization that is programmed to be either parallel or anti-parallel to that of the pinned magnetic layer, e.g., during a “WRITE” operation of the device.
The free magnetic layer and the pinned magnetic layer typically have a tunnel barrier therebetween. The resistance of this tunnel barrier depends on the orientation of the magnetization of the free magnetic layer relative to the orientation of the magnetization of the pinned magnetic layer. Namely, the resistance of the magnetic tunnel junction is higher when the free magnetic layer and the pinned magnetic layer have orientations of magnetizations that are anti-parallel, as compared to when they are parallel. If the anti-parallel and parallel magnetization states of the magnetic layers are coded to represent binary bits, the state of the magnetic tunnel junction, e.g., either a logic “1” or a “0,” is detected in the “READ” operation by measuring the resistance of the MTJ.
The orientation of the magnetization of a given layer (pinned or free) may be represented by an arrow which, by way of example only, can in some configurations be represented as pointing either to the left or to the right. When the MTJ is sitting in a zero applied magnetic field, the magnetization of the MTJ is stable, pointing either left or right. The application of a magnetic field, however, can toggle the magnetization of the free layer from left to right, and vice versa, to write information to the MTJ. One of the important requirements for data storage is that the magnetization of the MTJ not change orientation unintentionally during the writing process or when there is a zero applied field, or only a small applied field.
However, in many conventional MTJs, the free magnetic layer is typically composed of a single magnetic layer, which can possess a net magnetic dipole moment. The net magnetic dipole moment can undesirably cause a dipole field outside of the MTJ that interferes with the write operation of neighboring MTJs. In addition, this net external dipole field couples strongly to applied fields used to “WRITE” neighboring MTJs and can cause additional MTJs to switch undesirably.
For example, U.S. patent application Ser. No. US2003/0161180 by Bloomquist et al., entitled “Shared Bit Lines In Stacked MRAM Arrays,” (hereinafter “Bloomquist”), the disclosure of which is incorporated by reference herein, discloses cross-point arrays which include stacked magnetic storage elements, each element comprising a single-layer free magnetic layer. Such cross-point arrays have several notable disadvantages. As mentioned above, a dipole field may undesirably be generated outside of a given MTJ that interferes with the write operations of neighboring MTJs. As such, device density must be greatly reduced, by increasing the separation between adjacent devices, to compensate for this effect. Further, cross-point arrays undesirably generate parasitic current paths, which are competing paths for each given magnetic toggling device. These parasitic paths reduce signal strength during the READ operation. A reduced signal requires that signal averaging be done to read the logical states of the memory cells. As such, inaccuracies and loss of speed surely result.
Therefore, techniques for increasing the density of MTJs in magnetic devices while at the same time decreasing error rates, e.g., in the WRITE operation, would be desirable.
SUMMARY OF THE INVENTIONThe present invention provides techniques for improving magnetic device performance. In one aspect of the invention, a magnetic device, e.g., a magnetic random access memory device, is provided which comprises a plurality of current carrying lines; and two or more adjacent stacked magnetic toggling devices sharing at least one of the plurality of current carrying lines in common and positioned therebetween. The magnetic device is configured such that at least one of the adjacent magnetic toggling devices toggles mutually exclusively of another of the adjacent magnetic toggling devices. In an exemplary embodiment, the magnetic device comprises a plurality of levels with each of the adjacent stacked magnetic toggling devices residing in a different level.
A more complete understanding of the present invention, as well as further features and advantages of the present invention, will be obtained by reference to the following detailed description and drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
Non-magnetic spacer layer 104 may be configured to have a composition and a thickness suitable for loosely anti-parallel coupling magnetic layers 102 and 106. For example, in an exemplary embodiment, non-magnetic spacer layer 104 comprises ruthenium and has a thickness of from about 50 angstroms to about 60 angstroms. As mentioned above, magnetic toggling devices having a free layer comprising at least two loosely anti-parallel coupled, e.g., ferromagnetic, layers will be referred to herein as magnetic toggling devices.
U.S. Pat. No. 5,640,343 issued to Gallagher et al., entitled “Magnetic Memory Array Using Magnetic Tunnel Junction Devices in the Memory Cells,” the disclosure of which is incorporated by reference herein, describes an array of MTJs, each MTJ being uniquely addressed by two wires that intersect over the corresponding MTJ. Several technology demonstrators (e.g., experimental embodiment for understanding the parameters of a technology) have been accomplished embodying this principle. See, for example, W. Reohr et al., Memories of Tomorrow, IEEE C
U.S. Pat. No. 6,545,906 issued to Savtchenko et al., entitled “Method of Writing to Scalable Magnetoresistance Random Access Memory Element,” (hereinafter “Savtchenko”), the disclosure of which is incorporated by reference herein, describes an MTJ having a free magnetic layer comprising two loosely anti-parallel coupled ferromagnetic films with a non-magnetic spacer layer therebetween. Savtchenko teaches that the balance between two loosely anti-parallel coupled free magnetic layers can be perturbed by an external magnetic field to produce a net transient moment. The techniques described in Savtchenko help avoid the unintentional toggling of half-selected MTJs. See, for example, D. C. Worledge, Spin Flop Switching for Magnetic Random Access Memory, A
The graph in
According to another exemplary embodiment, a positive Hw is first employed which generates a transient net moment, followed in time by a positive Hb which toggles the magnetic toggling device by exploiting the transient net moment. Upon removal of both fields, balance is restored between the two magnetic layers making up the free layer and the transient net moment vanishes.
As further shown in
The diagram in
Therefore, if the time domain sequencing of magnetic fields as shown illustrated in
Importantly, magnetic toggling device 508 experiences the same polarity of magnetic field H1 but the opposite polarity of magnetic field H2 when compared with magnetic toggling device 510. Further, while magnetic toggling device 508 experiences the same polarity of magnetic field H1 as magnetic toggling device 510, it is of a weaker magnitude due to increased spatial separation of magnetic toggling device 508 from current carrying line 506, e.g., as compared to the close proximity of magnetic toggling device 510 and current carrying line 506. Also, the magnetic field H2 experienced by magnetic toggling device 508, the polarity of which being opposite to that experienced by magnetic toggling device 510, is of a comparable magnitude since magnetic toggling device 508 is positioned a similar distance to, but on the opposite side of, current carrying line 504.
Therefore, the magnetic field trajectory traced by magnetic fields H1 and H2 uniquely lies in the fourth quadrant as experienced by magnetic toggling device 508. Magnetic fields H1 and H2 thus will not toggle magnetic toggling device 508. As such, the present techniques allow for mutually non-interfering addressability of adjacent stacked magnetic toggling devices.
By a similar argument one can also prove that the time domain sequencing of magnetic fields H2 and H3, as shown in
Therefore, the magnetic field sequence generated by a pair of orthogonally oriented current carrying lines that each belong to two immediately adjacent wiring levels in a stacked magnetic toggling device array can be operated to uniquely address only the magnetic toggling devices sandwiched between them and not any other magnetic toggling devices, as long as the magnetic toggling devices are designed to have substantially parallel axes of anisotropy.
As mentioned above, the magnetic toggling devices may reside in distinct, different, levels within magnetic device 500, e.g., magnetic toggling device levels. According to an exemplary embodiment, the present magnetic device comprises exactly two magnetic toggling device levels, e.g., N=2, wherein N is the number of magnetic toggling device levels. According to another exemplary embodiment, the magnetic device comprises more than two magnetic toggling device levels, e.g., N>2. In these embodiments, the number of current carrying lines should be one greater than the number of magnetic toggling device levels, e.g., N+1. Further, the differences in potential layouts and designs based on these two exemplary embodiments can be significant and should be noted.
According to an exemplary embodiment wherein magnetic device 500 comprises more than two magnetic toggling device levels, vias and split (or offset) current carrying lines are employed to stack multiple magnetic toggling devices yet use only one transistor for reading all of them.
In
In an exemplary embodiment, self-aligned vias are employed. With this particular design, magnetic device areas approximately equal to 15 F2/n, wherein n is the number of stacking levels and F is the minimum feature size in the technology, are quite plausible. F refers to the ground rules of the design of the transistor complimentary metal oxide semiconductor (CMOS) front end. For example, if a 0.18 micron technology transistor is being used, the F=0.18 microns. Thus, F typically refers to the smallest dimension in the transistor, e.g., typically the gate length. The configuration shown illustrated in
The configuration shown in
As is illustrated in, and described in conjunction with the description of,
According to another exemplary embodiment, the present magnetic device comprises two magnetic toggling devices and two semiconductor transistors.
A further advantage of the configuration shown in
The field created by these two currents resides in the appropriate quadrant such that magnetic toggling device 510 can be written. As is illustrated in, and described in conjunction with the description of,
To write magnetic toggling device 508, for example, an electrical current of appropriate magnitude is flowed through WWL2(M4) and another electrical current of appropriate magnitude is flowed through BL1(M3). As above, the time domain sequencing of the magnetic fields is performed. The direction of the two currents is such that either both current directions are identical or both current directions are opposite to what are used to write magnetic toggling device 510. The magnetic fields created by these two currents will then also reside in the appropriate quadrant to write magnetic toggling device 508 but will also not affect magnetic toggling device 510.
To read magnetic toggling devices 508 and 510, the read word line (labeled “RWL(PC)”) is activated so that semiconductor transistors 509 and 512 (labeled “T1”) are conductive. Further, BL1(M3) is held at a potential different than that of the ground to cause a sense current (which is typically less than the current used to write the magnetic toggling devices) to flow across magnetic toggling device 510. BL2(M1) and WWL2(M4) will be held at different potentials to cause a sense current to flow across magnetic toggling device 508. The resistance of magnetic toggling devices 508 and 510 can then be measured based on the size of the two sense currents. In comparison with the configuration shown in, and described in conjunction with the description of,
Larger write current requirements lead to larger transistors being employed for steering the currents. These large transistors, in turn, lead to reduced array efficiency (e.g., a reduction in the percentage of chip area actually used for magnetic toggling devices). For most conventional applications, each layer of stacked magnetic toggling devices requires a word line driver and a bit line driver. However, according to the present techniques, for n layers of stacked magnetic toggling devices, only n/2 word line drivers and n/2+1 bit line drivers (or vice versa) are required, since the write lines are shared for neighboring layers. For example, if n equals four, then only five drivers are needed (e.g., three bit line drivers plus two word line drivers; or three word line drivers plus two bit line drivers) in total, instead of the eight (four word line drivers plus four bit line drivers) that would be needed with a conventional system. This exemplary configuration significantly reduces the amount of chip area required, at least on a comparable level as the reduced cell footprint. An additional benefit to this configuration includes reduced chip complexity.
Magnetic device 1200 further comprises magnetic toggling device 510 between BL1 and WWL1. One end of magnetic toggling device 510 is electrically connected to BL1, e.g., on the third metal level (M3). Local interconnect 1203 and contacts 1205, 1206 and 1207 electrically connect the other end of magnetic toggling device 510 to the drain of transistor 512. The source of transistor 512 is connected to system ground. The gate of both transistors 509 and 512 is the RWL (read word line).
Similar to bit lines 1002, 1004 and 1006, and word lines 1008 and 1010, as described above in conjunction with the description of
Reading of magnetic device 1300 may involve generating different sense-current levels for different combination of states of said magnetic toggling devices 1301 and 1302. Suitable methods for generating different sense-current levels include, but are not limited to, designing a different resistance change for magnetic toggling devices 1301 and 1302. In this embodiment, it is important to note that the bit lines and word lines are interchangeable. It is also important to note that this exemplary embodiment, as well as all the embodiments described herein, should not be limited to having any particular number of magnetic toggling devices. By way of example only, the exemplary configuration depicted in
Although illustrative embodiments of the present invention have been described herein, it is to be understood that the invention is not limited to those precise embodiments, and that various other changes and modifications may be made by one skilled in the art without departing from the scope or spirit of the invention.
Claims
1. A magnetic device comprising:
- a plurality of current carrying lines; and
- two or more adjacent stacked magnetic toggling devices sharing at least one of the plurality of current carrying lines in common and positioned therebetween, the magnetic device being configured such that at least one of the adjacent magnetic toggling devices toggles mutually exclusively of another of the adjacent magnetic toggling devices.
2. The device of claim 1, comprising a magnetic random access memory device.
3. The device of claim 1, wherein at least one of the magnetic toggling devices comprises a magnetic tunnel junction.
4. The device of claim 1, comprising a plurality of levels and wherein each of the two or more adjacent stacked magnetic toggling devices resides in a different one of the plurality of levels from each other.
5. The device of claim 4, comprising N+1 current carrying lines, wherein N is a number of levels in the device.
6. The device of claim 1, wherein at least one of the magnetic toggling devices has at least one free magnetic layer separated from at least one pinned magnetic layer by at least one barrier layer.
7. The device of claim 6, wherein a given one of the at least one free magnetic layer comprises two or more magnetic layers anti-parallel coupled by at least one non-magnetic spacer layer positioned therebetween.
8. The device of claim 6, wherein a given one of the at least one free magnetic layer comprises two or more ferromagnetic layers anti-parallel coupled by at least one non-magnetic spacer layer positioned therebetween.
9. The device of claim 7, wherein the non-magnetic spacer layer comprises ruthenium.
10. The device of claim 7, wherein the non-magnetic spacer layer has a thickness of between about 50 angstroms to about 60 angstroms.
11. The device of claim 1, comprising two magnetic toggling devices disposed in a stacked arrangement relative to one another.
12. The device of claim 1, comprising greater than two magnetic toggling devices disposed in a stacked arrangement relative to one another.
13. The device of claim 1, wherein at least one of the plurality of current carrying lines is oriented orthogonal to one or more other of the plurality of current carrying lines.
14. The device of claim 1, wherein the two or more adjacent stacked magnetic toggling devices have axes of anisotropy that are substantially parallel to each other.
15. A magnetic device comprising:
- a plurality of current carrying lines; and
- two or more adjacent stacked magnetic toggling devices sharing at least one of the plurality of current carrying lines in common and positioned therebetween, wherein one or more of the magnetic toggling devices comprises at least one free magnetic layer separated from at least one pinned magnetic layer by at least one barrier layer, the at least one free magnetic layer comprising two or more magnetic layers anti-parallel coupled by at least one non-magnetic spacer layer positioned therebetween so as to permit toggling mutually exclusive of another of the adjacent magnetic toggling devices.
16. A method of toggling a magnetic device comprising a plurality of current carrying lines and two or more adjacent stacked magnetic toggling devices sharing at least one of the plurality of current carrying lines in common and positioned therebetween, the method comprising the step of:
- passing current through one or more of the current carrying lines to generate a magnetic field sufficient to toggle at least one of the adjacent magnetic toggling devices mutually exclusively of another of the adjacent magnetic toggling devices.
17. The method of claim 16, wherein the magnetic device further comprises a plurality of levels and wherein each of the two or more adjacent stacked magnetic toggling devices resides in a different one of the plurality of levels from each other.
18. The device of claim 17, comprising N+1 current carrying lines, wherein N is a number of levels in the device.
19. The method of claim 16, wherein the magnetic device comprises two magnetic toggling devices disposed in a stacked arrangement relative to one another.
20. The method of claim 16, wherein the magnetic device comprises greater than two magnetic toggling devices disposed in a stacked arrangement relative to one another.
21. An integrated circuit including at least one magnetic device comprising:
- a plurality of current carrying lines; and
- two or more adjacent stacked magnetic toggling devices sharing at least one of the plurality of current carrying lines in common and positioned therebetween, the magnetic device being configured such that at least one of the adjacent magnetic toggling devices toggles mutually exclusively of another of the adjacent magnetic toggling devices.
22. A magnetic random access memory device, comprising a plurality of magnetic devices configured in an array, at least one of the plurality of magnetic devices comprising:
- a plurality of current carrying lines; and
- two or more adjacent stacked magnetic toggling devices sharing at least one of the plurality of current carrying lines in common and positioned therebetween, the magnetic device being configured such that at least one of the adjacent magnetic toggling devices toggles mutually exclusively of another of the adjacent magnetic toggling devices.
Type: Application
Filed: Oct 29, 2004
Publication Date: May 4, 2006
Applicant: International Business Machines Corporation (Armonk, NY)
Inventors: Sivananda Kanakasabapathy (Hopewell Junction, NY), Yu Lu (Hopewell Junction, NY), Michael Gaidis (Wappingers Falls, NY)
Application Number: 10/977,792
International Classification: G11C 11/00 (20060101);