CN1739166A - MRAM memories utilizing magnetic write lines - Google Patents

MRAM memories utilizing magnetic write lines Download PDF

Info

Publication number
CN1739166A
CN1739166A CN 200380105441 CN200380105441A CN1739166A CN 1739166 A CN1739166 A CN 1739166A CN 200380105441 CN200380105441 CN 200380105441 CN 200380105441 A CN200380105441 A CN 200380105441A CN 1739166 A CN1739166 A CN 1739166A
Authority
CN
China
Prior art keywords
magnetic
layer
free layer
bit line
line
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN 200380105441
Other languages
Chinese (zh)
Other versions
CN100524531C (en
Inventor
臧大化
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Applied Spintronics Technology Inc
Original Assignee
Applied Spintronics Technology Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Applied Spintronics Technology Inc filed Critical Applied Spintronics Technology Inc
Publication of CN1739166A publication Critical patent/CN1739166A/en
Application granted granted Critical
Publication of CN100524531C publication Critical patent/CN100524531C/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Landscapes

  • Mram Or Spin Memory Techniques (AREA)
  • Hall/Mr Elements (AREA)
  • Semiconductor Memories (AREA)

Abstract

A method and system for providing and using a magnetic random access memory are disclosed. The method and system include providing a plurality of magnetic memory cells, a first plurality of write lines, and a second plurality of write lines. The first plurality of write lines is a plurality of magnetic write lines. At least one of the plurality of magnetic lines and at least one of the second plurality of write lines each carrying a current for writing to at least one of the plurality of magnetic memory cells. Preferably, the plurality of magnetic write lines have soft magnetic properties and are preferably magnetic bit lines. For magnetic tunneling junction stacks within the magnetic memory cells, the magnetic bit lines are preferably significantly thicker than and closely spaced to the free layers of the magnetic memory cells.

Description

Utilize the storer of the MRAM of magnetic write lines
Mutual reference to related application
The application is the right of priority of the 60/431st, No. 741 temporary patent application in the sequence number that 35 USC 119 (e) require down to submit on Dec 9th, 2002.
The application and exercise question be " HIGH DENSITY AND HIGH PROGRAMMINGEFFICIENCY MARM DESIGN ", on February 5th, 2003 assignee that submit to and that transfer the application unsettled U.S. Patent application the 60/444th, No. 881 (2817P) relevant.The application and exercise question be " MRAM ARCHITECTURE AND A MAETHOD ANDSYSTEM FOR FABRICATING MRAM MEMORIES UTILIZING THEARCHITECTURE ", _ _ assignee's that submit to and that transfer the application unsettled U.S. Patent application the _ _ number (2818P) relevant.The application and exercise question be " MRAMARRAY WITH MAGNETIC WRITE LINES ", _ _ assignee's that submit to and that transfer the application unsettled U.S. Patent application the---number (2780P) is relevant.
Technical field
The present invention relates to magnetic storage, and more specifically relate to a kind of method and system that is used to provide magnetic RAM (MRAM), wherein this magnetic RAM is preferably high density, non-volatile, and it combines has the improved line of writing of writing efficient, being easy to make and have the electromigratory better reliability of opposing.
Background technology
DRAM, FLASH and SRAM are three kinds of main semiconductor memories on the market.Although the production cost of DRAM is minimum, DRAM has some shortcomings.DRAM is a volatibility, therefore, and obliterated data when turning off power supply.In addition, DRAM needs to upgrade, and has low relatively speed and has high power consumption.The FLASH storer provides non-volatile, but its speed is very low.In addition, the write cycle time permanance of FLASH storer is less than the cycle usually 1000,000 times.These drawbacks limit the application of FLASH storer in some high data rate markets.SRAM is a short-access storage, but it is a volatibility, and each unit occupies a large amount of relatively silicon areas.In order to seek to provide the universal random access memory of high-speed, non-volatile, junior unit area and good durability, developed thin film magnetic random access memories (MRAM).
Traditional thin film magnetic random access memories can comprise: anisotropic magnetoresistive (AMR) unit, giant magnetoresistance (GMR) unit and MTJ (MTJ) unit with multiple traditional memory cell types manufacturing.Because the easiest production in traditional MTJ unit and use, it will be used as and run through this disclosed basic example.Yet those of ordinary skills will understand these notions and also be applicable to other mram cell and array.These traditional MTJ unit comprise that in essence a MTJ piles up.This MTJ piles up and comprises pair of magnetic layers, accompanies an insulator layer between the magnetosphere.One of them magnetosphere, promptly bolt layer (pinned layer) has a fixed magnetic vector (fixed magnetization intensity).Another magnetosphere (free layer) has a changeable magnetic vector (variable magnetization intensity), and it stably is arranged in parallel or in fact antiparallel with the fixed magnetic vector in the bolt layer.When magnetic vector overlapped, traditional MTJ piled up and the resistance of traditional thus MTJ unit, promptly for the resistance of electric current between two magnetospheres, is minimum value.When magnetic vector is when overlapping on the contrary or not, the resistance of traditional MTJ unit is maximal value.
By applying a magnetic field for traditional MTJ unit with data storage in traditional MTJ unit.The magnetic field that is applied has a preferential direction and moves to a selected direction with the changeable magnetic vector with free layer.In other words, write traditional MTJ unit by applying the magnetic field that can change the magnetic vector direction of free layer usually.Usually, the direction of coincidence can be designed to logical one or 0, and the direction that does not overlap is opposite, promptly is respectively logical zero or 1.By making electric current flow through the data that traditional MTJ unit reads or sensing has been stored from a magnetosphere to another magnetosphere.Flow through the magnitude of current of traditional MTJ unit, the voltage drop at two ends, perhaps traditional MTJ unit will change according to the direction of changeable magnetic vector.
Usually be applied to the magnetic field of the direction that is used to change changeable magnetic vector with two leads that in fact are perpendicular to one another.When electric current flows through these two leads simultaneously, produce two with these two leads in the relevant magnetic field of electric current.These two the action of a magnetic fields are in the changeable magnetic vector of the free layer direction with directed this changeable magnetic vector.
Fig. 1 has described the part of conventional magnetic, and it comprises traditional mutually perpendicular lead 10 and 12, conventional magnetic storage cell 11 and conventional transistor 13.Conventional magnetic storage cell 11 is between conventional wires 10 and 12 and be positioned at the point of crossing place of two leads.The magnetic memory cells of describing among Fig. 1 is one and is piled up the traditional MTJ unit that constitutes by traditional MTJ.Traditional wire 10 and traditional wire 12 are called as word line and bit line usually respectively.Yet these names are interchangeable.Also can use other name, such as line, alignment, digital line and data line.
Tradition MTJ11 piles up and mainly comprises the free layer 1104 of (clearly not illustrating) that has changeable magnetic vector, the bolt layer 1102 with fixed magnetic vector (clearly not illustrating) and the insulator 1103 between two magnetospheres 1104 and 1102.Insulator 1103 has an enough thin thickness usually to allow the tunnelling of the electric charge carrier between the magnetosphere 1102 and 1104.Layer 1101 normally a kind of synthetic material that a plurality of inculating crystal layers and an anti-ferromagnetic layer are arranged, wherein anti-ferromagnetic layer firmly is coupled to the bolt magnetosphere.
During writing, the electric current I that flows in the conventional bit line 12 1With the electric current I that flows in the conventional word line 10 2On free layer 1104, produce two magnetic fields.In response to this magnetic field, the direction that magnetic vector is directed in the free layer 1104 depends on I 1And I 2Direction and the characteristic and the shape of amplitude and free layer 1104.Need I when usually, writing zero (0) 1Perhaps I 2Direction with different when writing one (1).During reading, this conventional transistor 13 is opened and a little tunnelling current is flowed through traditional MTJ unit.Measurement is flow through the voltage drop at the magnitude of current of traditional MTJ unit 11 or 11 two ends, traditional MTJ unit to determine the state of memory cell.In some designs, conventional transistor 13 is replaced by diode, perhaps omits fully, and traditional MTJ unit 11 directly contacts with conventional word line 10.
Although above-mentioned traditional MTJ unit 11 can write with conventional word line 10 and conventional bit line 12, those of ordinary skills will readily appreciate that the I in the great majority design 1Perhaps I 2Amplitude be about several milliamperes of levels.Therefore, those of ordinary skills also will recognize little write current of lots of memory application need.
Fig. 2 has described a part that has than the conventional magnetic of low writing current.At United States Patent (USP) the 5th, 659, No. 499, the 5th, 940, No. 319, the 6th, 211, No. 090, the 6th, 153, in No. the 2002/0127743rd, No. 443 and the U.S. Patent application publication similar system has been described.These with reference in disclosed legacy system and the being used to classic method of making these legacy systems with three surface encapsulation not facing MTJ unit 11 ' of soft magnetism overlayer with bit line and word line.Those that describe among a lot of parts of the legacy memory of describing among Fig. 2 and Fig. 1 are similar, thereby mark similar.
The system of describing among Fig. 2 comprises traditional MTJ unit 11 ', conventional word line 10 ' and bit line 12 '.Conventional word line 10 ' is made up of two parts: copper core 1001 and soft magnetism overlayer 1002.Equally, conventional bit line 12 ' is made up of two parts: copper core 1201 and soft magnetism overlayer 1202.
With respect to the design among Fig. 1, soft magnetism overlayer 1002 and 1202 can will with I 1And I 2Relevant flux concentration is to MTJ unit 11 ', and the lip-deep magnetic field of MTJ unit 11 ' is not faced in minimizing.Therefore, soft magnetism overlayer 1002 and 1202 to the MTJ that forms MTJ unit 11 ', makes free layer 1104 be easier to programming flux concentration.Although respond well on this theoretical method, those of ordinary skills will readily appreciate that the magnetic of the soft overlayer 1002 on the vertical sidewall of traditional wire 10 ' and 12 ' and 1202 parts is difficult to control respectively.Those of ordinary skills will recognize that also the technology of making conventional word line 10 ' and conventional bit line 12 ' is very complicated.Complicated manufacture method give demarcate (scaling) more high density significant challenge has been proposed.Therefore, be desirable to provide very much a kind of MRAM architecture, it can demarcate and be easy to make, and provides height to write efficient.
In addition, conventional write lines 10,10 ', 12,12 ' the restriction demarcation property of the traditional design of describing among Fig. 1 and Fig. 2.In these traditional design, conventional write lines 10,10 ', 12,12 ' major part are made of either aluminum or copper.The current density limit of aluminium and copper is about 1 * 10 6A/cm 2Level or lower.When reducing live width with the increase storage density, electromigration current density limit has proposed severe challenge to demarcation.
Other conventional systems attempt has proposed different solutions, and wherein each has its shortcoming.For example, the U.S. Patent application publication proposes for No. 2002/0080643 to apply an inverse current to prevent electromigration to writing line after a write operation.But these classic methods have been damaged performance because of having reduced storage speed and having increased complicacy.Therefore, also wish very much to have by what the material that high reliability is arranged in electromigration was made and write line, it will allow the easy demarcation property to high density memory arrays.
May be used for conventional thin bit lines, which might littler or more efficient memory and have some shortcomings.Thinner conventional bit line has higher resistance.This influences the performance of whole memory arrays unfriendly.Yet, the classic method that much overcomes this problem is arranged.A common practice is that the long bit line in the memory array is divided into the global bit line of being made by crude metal, and these global bit line are connected to by thinner metal local bitline, thereby has high electrical resistance.The example of this design is in U.S. Patent No. 6,335, in No. 890 and the U.S. Patent application publication No.2002/0034117 number instruction arranged.Yet above-mentioned other problem does not still overcome such as electromigration.
Therefore, need one to be used to provide system and method that can demarcate, effective, low current magnetic memory.The present invention is devoted to address that need.
Summary of the invention
The present invention discloses a kind of method and system that is used to provide and use a kind of magnetic RAM.This method and system comprises provides a plurality of magnetic memory cells, more than first group write line and more than second group write line.This more than first group is write line is many magnetic write lines.Each bar that in many magnetic line more than at least one and second group writes at least one in the line all carries at least one the electric current that is used to write a plurality of magnetic memory cells.Preferably, many magnetic write lines have soft magnetic characteristic, and magnetic bit line preferably.Pile up for MTJ in the magnetic memory cells, magnetic bit line is preferably obviously than the free bed thickness of magnetic memory cells, and and the free layer of magnetic memory cells between tight spacing.
According to system and method disclosed herein, the invention provides a kind of magnetic storage that improves efficient, improved opposing electromigration reliability, is easier to make that has.
Description of drawings
Fig. 1 is the 3-D view of the part of a conventional magnetic, and this storer comprises a MTJ unit that is positioned at bit line and place, word line point of crossing.
Fig. 2 is the 3-D view of the part of a conventional magnetic, and this storer comprises a MTJ unit that is positioned at bit line and word line point of crossing place, and wherein this bit line and word line have the magnetic overlayer and write efficient with improvement.
Fig. 3 a, Fig. 3 b and Fig. 3 c are respectively side view, viewgraph of cross-section and the planimetric maps according to the part of an embodiment of magnetic storage of the present invention, and this storer comprises that a MTJ who is arranged in the mram cell with magnetic bit line piles up.
Fig. 4 is according to the MTJ unit of an embodiment of magnetic storage of the present invention and the planimetric map of bit line, and the diagram of the magnetic bit line of the magnetic vector of the free layer of MTJ unit and static state.
Fig. 5 is according to the MTJ unit of an embodiment of magnetic storage of the present invention and the planimetric map of bit line, and the diagram of the magnetic vector of the free layer of MTJ unit and the magnetic bit line when a write current flows in bit line.
Fig. 6 is according to the MTJ unit of an embodiment of magnetic storage of the present invention and the viewgraph of cross-section of bit line, and it has shown the field that is produced by the current segment in the metal spacer layer.
Fig. 7 is the viewgraph of cross-section according to the part of the embodiment of the MRAM that comprises a memory cell of the present invention.
Fig. 8 is the viewgraph of cross-section according to the part of another embodiment of the MRAM that comprises a memory cell of the present invention.
Embodiment
The present invention relates to the improvement in a kind of magnetic storage.Following description is introduced so that those of ordinary skill in the art can make and use the present invention, and this description is provided between the context of patented claim and its requirement.To those skilled in the art, all be clearly to the various modifications of preferred embodiment, and can be applied to other embodiment at this general principle.Therefore, unintentionally the present invention is limited to the embodiment that illustrates, and has a mind to make the present invention with consistent with the corresponding to wide region of principle described here and feature.
The present invention discloses a kind of method and system that is used to provide and use a kind of magnetic RAM.This method and system comprises provides a plurality of magnetic memory cells, more than first group write line and more than second group write line.This more than first group is write line is many magnetic write lines.Each bar that in many magnetic line more than at least one and second group writes at least one in the line all carries at least one a electric current that is used to write a plurality of magnetic memory cells.Preferably, many magnetic write lines have soft magnetic characteristic, and magnetic bit line preferably.Pile up for MTJ in the magnetic memory cells, magnetic bit line is preferably obviously than the free bed thickness of magnetic memory cells, and with the free layer tight spacing of magnetic memory cells.
To construct according to magnetic memory cells, certain material and the particular element of particular type and describe the present invention.Yet those of ordinary skills will readily appreciate that this method and system are for will effectively implementing with reconcilable other magnetic memory cells of the present invention, other material and structure.For example, comprise single magnetosphere, do not stop and use other material, other alloy and synthetic layer although MTJ piles up to be described as.Although those of ordinary skills also will readily appreciate that according to magnetic bit line the present invention is described, this method and system compatible magnetic word lines, digit line or only write line.
In order to describe the method according to this invention and system in more detail,, the part of magnetic storage according to the present invention such as the embodiment of MRAM described referring now to Fig. 3 a, 3b and 3c.Fig. 3 a has described the side view according to the part of magnetic storage of the present invention.This MRAM comprises such as many word line (not shown), the bit line of magnetic bit line 32 and the mram cell that one of them illustrates in the drawings.The mram cell 31 that shows comprises that a MTJ piles up 31.This MTJ piles up 31 and comprises two magnetospheres: bolt layer 3101 and free layer 3103.Free layer 3103 has a changeable magnetic vector, and bolt layer 3101 has a fixed magnetic vector.Magnetosphere 3101 and 3103 is preferably by making such as magnetic materials such as nickel, cobalt, iron or its alloys.Also can add other element, produce the magnetic that needs such as boron, nitrogen.Though each is shown as an individual course in the drawings free layer 3103 and bolt layer 3101, those of ordinary skills should know that each layer also can be a plurality of magnetospheric synthetic.Bolt layer 3101 is fixed with an anti-ferromagnetic layer below 3101 usually, and is not shown here, perhaps uses the alternate manner that those of ordinary skills were familiar with.
Insulator layer 3102 separates two magnetospheres.Insulator 3102 is preferably the synthetic of aluminium oxide, other oxide or two or more oxide skin(coating)s.Insulator 3102 is also preferably enough thin to allow the tunnelling of the electric charge carrier between the magnetosphere 3101 and 3103.Free layer 3103 is preferably covered by a metal level 3104.In diagram, overlayer is preferably made such as copper, gold, silver, rhodium, ruthenium, aluminium and tantalum by high conductivity metal.If copper is used as overlayer, cobalt-iron (not shown) is often used as diffusion barrier.In the case, the top of free layer 3103 should comprise at least that a thick thin cobalt-iron layer of several dusts is diffused into free layer 3103 to prevent copper from overlayer 3104.When Ni-Fe was free layer 3101 a part of, this diffusion barrier was particularly useful.
Magnetic bit line 32 is piled up 31 with MTJ and is intersected, and preferably contacts with metal cladding 3104.In the present invention, bit line 32 is a magnetic.The major part of magnetic bit line 32 is preferably all, is made by magnetic material, and these magnetic materials are such as nickel, cobalt, iron or its alloy, such as NiFe.The magnetic material that constitutes magnetic bit line 32 is a kind of soft magnetic material.In one embodiment, the core of magnetic bit line 32 (core) is a magnetic.In a preferred embodiment, most or all of magnetic bit line 32 are magnetic, except the layer such as inculating crystal layer.Magnetic bit line 32 can further be several magnetospheric synthetic.If copper is used as metal cladding 3104, magnetic bit line 31 should comprise at least that a cobalt iron layer (not illustrating respectively) is as the diffusion barrier between copper metal cladding 3104 and the magnetic bit line 32.Though magnetic bit line 32 is preferably mainly made by magnetic material, magnetic bit line 32 also can have non-magnetic metal seed layer.Yet because the reason of discussing below, the interval between the magnetic part of magnetic bit line 32 and the free layer 3103 is preferably about 300 dusts or littler.In a preferred embodiment, MTJ piles up 34 metal seed layer (not shown) and the synthetic thickness of non magnetic overlayer 3104 remains about 300 dusts or littler.
Fig. 3 b has shown the viewgraph of cross-section that piles up 31 structure according to magnetic bit line 32 of the present invention and MTJ.Magnetic bit line 32 has width W 32.MTJ piles up 31 and has width W 31.In the present invention, magnetic bit line width W 32 preferably is equal to, or greater than MTJ and piles up width W 31.In Fig. 3 b, magnetic bit line width W 32 is piled up width W 31 greater than MTJ.Yet, can pile up 31 cardinal principle equal widths and desirable substantially alignings between the two with magnetic bit line 32 of autoregistration Patternized technique generation and MTJ.In this technology, use a kind of etch process, grind (ion milling) such as ion and determine that magnetic bit line width W 32 and MTJ pile up width W 31.
Fig. 3 c is the planimetric map that piles up 31 embodiment according to magnetic bit line 32 of the present invention and MTJ.In Fig. 3 C, shown that MTJ piles up 31 and is rectangle, its major axis is arranged as basically with magnetic bit line 32 and aims at.Therefore, MTJ piles up 31 and has anisotropic shape so that the magnetic vector of free layer 3103 basically along magnetic bit line 32 vertically.In the following description, we will continue this shape anisotropy of supposition.Yet, it should be noted that other shape and other direction of the magnetic vector of free layer 3103 also is consistent with the present invention.For example, it should be noted that the present invention also works with the vertical vertical easy magnetizing axis of bit line to the direction of free layer.
Fig. 4 is the planimetric map of embodiment of a part of the MRAM of the free layer 3103 described in Fig. 3 a, 3b and 3c and magnetic bit line 32.In Fig. 4, magnetic vector M321, the M322 of magnetic bit line 32 and schematically showing of M323 have been shown.According to the present invention, these magnetic vectors are orientated and are arranged essentially parallel to bit line in the vertical.In Fig. 4, also show the magnetic vector M311/M312 of free layer 3103.Under static state, the magnetic vector M311/M312 of free layer will be in easy axis.Therefore, for the purpose of example, the free layer vector is shown as or M311 or M312, two logic states of expression memory cell.In addition, Fig. 4 has described because MTJ piles up 31 shape, and the easy magnetizing axis that MTJ piles up 31 free layer 3103 is parallel to the major axis of aiming at magnetic bit line 32 basically.Yet, can form the easy magnetizing axis of free layer 3103 in another way, as because inherent and/or stress anisotropy.In a preferred embodiment, by the anisotropy of shape, inherence and the anisotropic easy magnetizing axis that is combined to form free layer that stress forms.
Fig. 5 has described an embodiment of the part of the magnetic storage during writing.Fig. 5 has shown the magnetic vector of magnetic bit line 32 and free layer 3103, as Fig. 4 described, have in magnetic bit line 32 the write current I that flows 32Suppose that this logic state is corresponding to free layer magnetic vector M312.Electric current I 32Induce the magnetic field of the magnetic vector M312 of rotation free layer according to the right-hand rule.Therefore, as shown in Figure 5, the magnetic vector M312 of free layer 3103 is rotated down.Because the magnetic vector M322 of magnetic bit line 32 is closely near free layer 31, the magnetic vector M312 of M322 and free layer has powerful magnetostatic coupling.As a result, M322 also rotates up as shown in Figure 5, to have formed flux closure on the direction perpendicular to magnetic bit line 32.
At a preferred embodiment, the magnetic moment of magnetic bit line 32 is more a lot of greatly than the magnetic moment of free layer 3103.Preferably, this thickness by making magnetic bit line 32 is than thick assurance of thickness of free layer 3103.Because the moment of the torque ratio free layer 3103 of magnetic bit line 32 is a lot of greatly, so the rotation angle θ 322 of bit line magnetic vector M322 is little more a lot of than the rotation angle θ 312 of free layer magnetic vector M312.In a preferred embodiment, run through whole write operation, therefore the magnetic vector M322 of magnetic bit line 32 remains on the longitudinal direction of magnetic bit line 32 substantially.In other words, reverse during the free layer magnetic vector, the magnetic vector M322 of magnetic bit line 32 only departs from a very little angle from the bit line longitudinal direction.In a preferred embodiment, the sine of the sine of the rotation angle θ 322 of bit line magnetic vector M322 and the rotation angle θ 312 of free layer magnetic vector M312 more as described below than roughly,
Sin (θ 322)/sin (θ 312)~(resultant couple of free layer 3103)/(resultant couple of magnetic bit line 32).
At least 10 times of magnetic moments by guaranteeing magnetic bit line 32 allow bit line magnetic vector that little corner is arranged to the magnetic moment of free layer 3103.In fact, this difference of magnetic moment is preferably by providing line thickness to realize at least 300 dusts or bigger bit line 32.
Expectation improves the writing efficient of magnetic bit line 32 with the little rotational angle theta 322 of the magnetic vector M322 of magnetic bit line 32 and need not give a discount to the efficient of writing of word line 30.As shown in Figure 5, magnetic storage also comprises word line 30, and in a preferred embodiment, word line 30 is vertical with magnetic bit line 32 basically.During writing, electric current I 30Flow to word line 30, simultaneously electric current I 32Flow to magnetic bit line 32.Synthetic (the I of two electric currents 30And I 32) and word line current I 30The last direction of magnetic vector (M311/M312) of polarity decision free layer 3103.In other words, two electric current I 30And I 32Combine and write MTJ and pile up 31.During writing, I 30Produce basically along magnetic bit line 32 magnetic field F30 longitudinally.The polarity of magnetic field F30 or left or to the right, as shown in Figure 5, this is by electric current I 30Polarity determine.The polarity of magnetic field F30 determines to be written to the logic state of MTJ unit 31.
Magnetic field F30 not only also produces a torque at bit line magnetic vector M322 on free layer magnetic vector M312.When θ 312 is more a lot of greatly than θ 322, more a lot of greatly in the torque that free layer magnetic vector M312 produces than the torque on bit line magnetic vector M322 by magnetic field F30.In addition, when θ 322 little can ignore the time, the torque on the M322 also can be ignored.As a result, it is vertical along magnetic bit line 32 that bit line magnetic vector M322 keeps substantially, and keep flux closure.Like this, by flux closure improved greatly magnetic bit line 32 write efficient simultaneously the efficient of writing of word line 30 do not give a discount.
Except the thickness of magnetic bit line 32 than free layer 3103 thick a lot of, in order to obtain more effective flux closure between magnetic bit line magnetic vector M322 and free layer magnetic vector M311/M312, what the spacing between magnetic bit line 32 and the free layer 3103 will be enough is little.Therefore, in a preferred embodiment, the thickness of magnetic bit line 32 is more a lot of greatly than the thickness of free layer 3103, and the spacing between magnetic bit line 32 and the free layer 3103 is enough little.Yet, in another embodiment, can omit the one or more of these features.For example, the spacing between magnetic bit line 32 and the free layer 3103 is enough little, but thickness difference also can be little between magnetic bit line 32 and the free layer 3103.Use traditional characteristic length that is used for two magnetospheric flux closure can assess desired spacing.Owing to be preferably magnetic bit line 32 thick a lot of than free layer 3103, so characteristic length probably is (μ gt/2) 0.5, wherein μ is the magnetic permeability of free layer, g is the spacing between magnetic bit line 32 and the free layer 3103, and t is the thickness of free layer 3103.It is preferably a lot of greatly than feature length that MTJ piles up 31 width W 31, to avoid tangible edge roll curved wall.As a result, can determine desired spacing from following relation:
(μgt/2) 0.5<W31。
In the application of today, the meaning of this relation is that spacing should be in 300 dusts or littler magnitude for such little spacing as expected.Therefore, in a preferred embodiment, the spacing between free layer 3103 and the magnetic bit line 32 is less than 300 dusts, and the thickness of magnetic bit line 32 is greater than 300 dusts.
Fig. 6 has shown the viewgraph of cross-section as the embodiment of magnetic bit line of describing 32 and MTJ31 in Fig. 3 b.During writing, the electric current I in the magnetic bit line 32 32A part shunted by metal cladding 3104.Electric current I in the metal cladding 61In magnetic bit line 32 and free layer 3103, produce magnetic field F61 and F62 respectively.These two magnetic field F61 and F62 are opposite on direction, have further strengthened the flux closure between magnetic bit line 32 and the free layer 3103.Therefore, wish that overlayer 3104 strengthens flux closure between magnetic bit line 32 and the free layer 3103 for high conductivity metal.In a preferred embodiment, overlayer 3104 comprises copper, gold, silver, rhodium, ruthenium, aluminium and tantalum, other known good conductor and its alloy.
If with copper as overlayer 3104, usually with cobalt ions (not shown) or other suitable material as the diffusion barrier between metal cladding 3104 and free layer 3103.In a such embodiment, the thin cobalt-iron layer (not shown) that free layer 3103 preferably includes several at least dusts that are positioned at its top diffuses into free layer 3103 to prevent copper.When nickel ion was free layer 3103 a part of, this diffusing barrier was particularly useful.Because similar factor, a cobalt-iron layer (not shown) can be placed between copper capping layer 3104 and the magnetic bit line 32, particularly when nickel ion being used as magnetic bit line 32 a part of.
For the resistance of the bit line 32 that deperms, also a high conductivity nonmagnetic metal thin layer can be placed as the inculating crystal layer (not shown) of magnetic bit line 32.Yet the thickness of this inculating crystal layer is preferably enough little so that the merging thickness of non magnetic inculating crystal layer and overlayer 3104 is approximately 300 dusts or thinner.
Describe the method according to this invention and system though use the MTJ that has the free layer 3103 that is positioned on the thin insulator layer 3102 to pile up 31, other magnetic texure is worked yet those of ordinary skills will be easy to recognize also.These magnetic texures include but are not limited to one to have the MTJ that is positioned at the free layer under the insulator tunnel layer and piles up (not shown) and one and have two insulator tunnel layers and pile up (not shown) with the MTJ that is clipped in the free layer of two insulator layers.In each structure, preferably the spacing between free layer and the magnetic bit line is to be in 300 dusts or littler magnitude.In different structures, this preferred thickness has been considered the random layer between magnetic bit line 32 and free layer 3103, as overlayer, bolt layer or other layer.
In addition, those of ordinary skills recognize easily that also magnetic bit line does not need to be placed on MTJ and piles up 31 top.For example, magnetic bit line 32 also can be positioned under the mtj structure.Need between magnetic bit line 32 and MTJ pile up 31, electrical connection not arranged yet.In such an embodiment, this magnetic line also is called writes line.In addition, in another embodiment, writing line and bit line can be magnetic.Yet, in all embodiment, even if magnetic write lines and to have spacing between the free layer of variable magnetic vector approximately be 300 dusts or still less preferably, and the thickness of magnetic write lines is greater than 300 dusts.
Fig. 7 has described an embodiment according to the part of mram memory of the present invention.A lot of parts are similar to the magnetic bit line 32, the memory cell/MTJ that describe and pile up 31 and word line 30 in Fig. 3 a-3c, 4,5 and 6.Therefore, similar to the mark of a lot of parts.For example, the MTJ among Fig. 7 piles up the 31 ' MTJ corresponding to description in Fig. 3 a-3c, 4,5 and 6 and piles up 31.With reference to Fig. 7, pile up 31 ' with MTJ except aforesaid magnetic bit line 32 ', the transistor 81 in word line 30 ', bypass connection 78, conductive layer 79, ground wire 77 and the substrate 72 has also been described.Word line 30 ' is shown as vertical with magnetic bit line 32 '.MTJ piles up 31 ' and is connected to the source electrode 74 of transistor 81 by conductive layer 79 and bypass connection 78.Transistor is a FET transistor preferably.The drain electrode 73 of FET transistor 81 is connected to ground wire 77 by a contact 75.The door 76 of FET transistor 81 is connected to the readout word line (not shown).Although the structure that illustrates is preferred, yet those of ordinary skills will readily appreciate that much other the different structure that combines according to magnetic bit line of the present invention also is feasible.For example, available diode (not shown) replaces transistor 81.The mram cell (not shown) does not have transistor or diode also is feasible.
Fig. 8 has shown another embodiment according to the part of MRAM of the present invention.A lot of parts are similar to the magnetic bit line of describing 32 in Fig. 3 a-3c, 4,5,6 and 7 ", memory cell/MTJ pile up 31 and 31 ' with word line 30 and 30 ".Therefore, similar to the mark of a lot of parts.For example, the MTJ among Fig. 8 piles up 31 " pile up 31 and 31 ' corresponding to the MTJ that in Fig. 3 a-3c, 4,5,6 and 7, describes.Similarly, transistor 81 ' is corresponding to the transistor among Fig. 7 81.Word line 80 is corresponding to the word line 30 and 30 ' among Fig. 3 a-3c, 4,5,6 and 7.With reference to Fig. 8, MRAM comprises magnetic bit line 32 ", MTJ piles up 31 ", the transistor 81 ' in bypass connection 78 ', ground wire 77 ' and the substrate 72 '.Magnetic bit line 32 ", MTJ piles up 31 " with word read line 80 basically with reference Fig. 7 describe identical.Yet, do not resemble the word line 30 among Fig. 7, the word line 80 ' among Fig. 8 is placed on magnetic bit line 32 " on.In order to keep the efficient of writing of word line 80 ', magnetic bit line 32 " thickness preferably thin as much as possible.Make traditional thin like this bit line and be difficult in normal design, its neutrality line is made of aluminum or copper.Yet the thickness thin like this in the present invention is practicable.
As mentioned above, traditional thin bit line has two shortcomings.The first, thin conventional bit line has high resistance.This performance to whole memory arrays produces adverse influence.Yet, the classic method that much overcomes this problem is arranged.A common practice is that the long bit line in the memory array is divided into the global bit line of being made by crude metal, and these global bit line are connected to by thinner metal local bitline, thereby has high electrical resistance.Yet these classic methods are not used magnetic bit line.In addition, these classic methods still suffer the puzzlement of other defective, electromigration as discussed below.
The second, the electromigration of conventional thin metal lines will bring adverse influence to reliability.For example, copper is for about 1 * 10 6A/cm 2Or current density still less has relatively low electromigration limit.Aluminum steel is anti-, and the ability of loaded current is not lower because electromigration does not affect adversely.Comparatively speaking, the ability of loaded current is not higher because electromigration does not affect adversely for magnetic material such as nickel ion.For example, nickel-iron films is used as the magnetic resistance read transducer in hard disk drive, and demonstrates to have and surpass 10 8A/cm 2The ability of loaded current.Therefore this allow to be used for thinner and/or narrower bit line much larger than aluminium and copper.
Owing to the excellent electromigration characteristic of soft magnetic film such as nickel ion, just had the thin bit line of hundreds of dust magnitude in the practice.This will strengthen the efficient of word line 80 greatly.Although word line 80 is shown as single line in Fig. 8, yet those of ordinary skills are interpreted as the present invention includes many effective word line structures, as using overlayer for word line 80.
Disclose a kind of be used to provide have improved efficient, the method and system of reliability and the magnetic storage that is easy to make preferably write.Although described the present invention according to the embodiment that shows, yet one of skill in the art will recognize that the change that has embodiment, and those changes will be within the spirit and scope of the present invention.Therefore, do not break away from the spirit and scope basis of appended claim, the field those of ordinary skill can be made a lot of modifications.

Claims (37)

1. magnetic storage, it comprises:
Many magnetic memory cells;
Write line for more than first group, described more than first group to write line be many magnetic write lines; With
Write line for second group, at least one in described many magnetic line and described more than second group each bar of writing at least one in the line all carry at least one the electric current that is used to write described a plurality of magnetic memory cells.
2. magnetic storage as claimed in claim 1 is write line and is comprised many magnetic bit line that are electrically coupled to described a plurality of magnetic memory cells for wherein said more than first group.
3. magnetic storage as claimed in claim 2, wherein said a plurality of magnetic memory cells comprises that a plurality of MTJ pile up, described a plurality of MTJ pile up each comprise free layer, insulator layer and bolt layer, free layer and bolt layer are ferromagnets, insulator layer between described free layer and described bolt layer and its thickness allow electric charge carrier tunnelling between free layer and bolt layer.
4. magnetic storage as claimed in claim 3, wherein said many magnetic bit line and described free layer are separated into and are less than or equal to 300 dusts.
5. magnetic storage as claimed in claim 4, wherein said a plurality of MTJ pile up each comprise nonmagnetic spacer layer between free layer and corresponding magnetic bit line, this nonmagnetic spacer layer conduction.
6. magnetic storage as claimed in claim 1, wherein this nonmagnetic spacer layer is the metal of high conductivity.
7. magnetic storage as claimed in claim 1, the metal of wherein said high conductivity comprise a kind of alloy of gold, copper, aluminium, rhodium, ruthenium, tantalum and/or gold, copper, aluminium, rhodium, ruthenium, tantalum.
8. magnetic storage as claimed in claim 5, wherein this nonmagnetic spacer layer comprises copper, and wherein said corresponding magnetic bit line comprises that one is in copper and respective magnetic and writes diffusion barrier between the remainder of line.
9. magnetic storage as claimed in claim 8, wherein said free layer comprise second diffusion barrier that is between described copper and the described free layer.
10. magnetic storage as claimed in claim 9, wherein this diffusion barrier and described second diffusion barrier comprise cobalt ions.
11. magnetic storage as claimed in claim 3, wherein said magnetic bit line have first thickness thicker than second thickness of described free layer.
12. magnetic storage as claimed in claim 11, wherein first thickness is 10 times of second thickness.
13. magnetic storage as claimed in claim 11, wherein first thickness at least 300 dusts.
14. magnetic storage as claimed in claim 1, wherein said many magnetic write lines comprise soft magnetic material.
15. magnetic storage as claimed in claim 14, wherein said soft magnetic material comprises the alloy of cobalt, nickel, iron and/or cobalt, nickel, iron.
16. magnetic storage as claimed in claim 1, wherein said many magnetic write lines comprise magnetospheric a plurality of synthetic.
17. magnetic storage as claimed in claim 1, each bar of wherein said many magnetic write lines has an easy magnetizing axis, and this easy magnetizing axis is oriented the longitudinal direction of each bar that is parallel to many magnetic write lines basically.
18. magnetic storage as claimed in claim 5, wherein said nonmagnetic spacer layer and write line accordingly and be placed on described MTJ piles up, and wherein second group of multiple bit lines is placed on many magnetic write lines and is oriented and be basically perpendicular to described many magnetic write lines.
Therefore 19. magnetic storage as claimed in claim 18, wherein the thickness of many magnetic write lines makes more than second group the interval minimum of writing between line and the free layer between 300 dusts and 3000 dusts.
20. magnetic storage as claimed in claim 5, wherein MTJ is piled up and is further comprised second insulator layer and the second bolt layer, and free layer is clipped between the insulator layer and second insulator layer, and second insulator layer is between the free layer and the second bolt layer.
21. magnetic storage as claimed in claim 1 further comprises a plurality of selecting arrangements, in a plurality of selecting arrangements each is corresponding in a plurality of magnetic memory cells each, in wherein a plurality of selecting arrangements each is a FET who comprises a door, and this is connected to an other readout word line.
22. magnetic storage as claimed in claim 1 further comprises a plurality of selecting arrangements, each in a plurality of selecting arrangements is corresponding in a plurality of magnetic memory cells each, and each in wherein a plurality of selecting arrangements is a diode.
23. magnetic storage as claimed in claim 1, wherein more than second group to write line be second group of many magnetic write lines.
24. a method of utilizing magnetic storage, the method comprising the steps of:
(a) under WriteMode, write the first of a plurality of memory cells, a plurality of memory cells be coupled to more than first group write line and more than second group write line, more than first group is write line is many magnetic write lines; Each bar that in many magnetic line more than at least one and second group writes at least one in the line all carries at least one the electric current that is used to write a plurality of magnetic memory cells;
(b) under reading mode, read from the second portion of a plurality of memory cells.
25. method as claimed in claim 24 is wherein write line and is comprised many magnetic bit line that are electrically coupled to a plurality of magnetic memory cells for more than first group.
26. method as claimed in claim 25, wherein a plurality of magnetic memory cells comprise that a plurality of MTJ pile up, a plurality of MTJ pile up each comprise free layer, insulator layer and bolt layer, free layer and bolt layer are ferromagnets, insulator layer between free layer and the bolt layer and its thickness allow electric charge carrier tunnelling between free layer and bolt layer.
27. method as claimed in claim 26, wherein many magnetic bit line and free layer are separated for being less than or equal to 300 dusts.
28. method as claimed in claim 27, wherein a plurality of MTJ pile up each comprise nonmagnetic spacer layer between free layer and corresponding magnetic bit line, this nonmagnetic spacer layer is conducted electricity.
29. method as claimed in claim 24, wherein this nonmagnetic spacer layer is the metal of high conductivity.
30. method as claimed in claim 28, wherein this nonmagnetic spacer layer comprises copper, and wherein corresponding magnetic bit line comprises a diffusion barrier between the remainder that is in copper and corresponding magnetic write lines.
31. method as claimed in claim 30, wherein free layer comprises second diffusion barrier that is between copper and the free layer.
32. method as claimed in claim 26, wherein magnetic bit line has first thickness thicker than second thickness of free layer.
33. method as claimed in claim 32, wherein first thickness is at least 300 dusts.
34. method as claimed in claim 24, wherein said a plurality of magnetic write lines comprise soft magnetic material.
35. method as claimed in claim 24, wherein many magnetic write lines comprise magnetospheric a plurality of complex.
36. method as claimed in claim 25, wherein each bar of many magnetic write lines has an easy magnetizing axis, and this easy magnetizing axis is oriented the longitudinal direction of each bar that is parallel to many magnetic write lines basically.
37. method as claimed in claim 24, wherein more than second group to write line be many magnetic write lines of folded group.
CNB2003801054418A 2002-12-09 2003-12-09 MRAM memories utilizing magnetic write lines Expired - Fee Related CN100524531C (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US43174202P 2002-12-09 2002-12-09
US60/431,742 2002-12-09
US10/459,133 2003-06-11

Publications (2)

Publication Number Publication Date
CN1739166A true CN1739166A (en) 2006-02-22
CN100524531C CN100524531C (en) 2009-08-05

Family

ID=36081269

Family Applications (1)

Application Number Title Priority Date Filing Date
CNB2003801054418A Expired - Fee Related CN100524531C (en) 2002-12-09 2003-12-09 MRAM memories utilizing magnetic write lines

Country Status (1)

Country Link
CN (1) CN100524531C (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101304038B (en) * 2007-05-10 2010-09-29 旺宏电子股份有限公司 Magnetic memory and memory cell thereof and method of manufacturing the memory cell
CN110890458A (en) * 2018-09-07 2020-03-17 上海磁宇信息科技有限公司 Method for improving write efficiency of magnetic random access memory

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101304038B (en) * 2007-05-10 2010-09-29 旺宏电子股份有限公司 Magnetic memory and memory cell thereof and method of manufacturing the memory cell
CN110890458A (en) * 2018-09-07 2020-03-17 上海磁宇信息科技有限公司 Method for improving write efficiency of magnetic random access memory
CN110890458B (en) * 2018-09-07 2024-04-12 上海磁宇信息科技有限公司 Method for improving writing efficiency of magnetic random access memory

Also Published As

Publication number Publication date
CN100524531C (en) 2009-08-05

Similar Documents

Publication Publication Date Title
CN100350495C (en) Magneto-resistive device having magnetically soft reference
CN1409321A (en) Reading method for magnet resistance device with soft reference layer
US6477077B2 (en) Non-volatile memory device
CN1729537A (en) Magnetoresistive random access memory with reduced switching field
US6940749B2 (en) MRAM array with segmented word and bit lines
CN1405777A (en) Multiple-location storage unit
CN1427416A (en) Magnetic storage device with soft reference layer
CN1610949A (en) Writing of scalable MRAM elements
CN101036195A (en) Spin-transfer based MRAM using angular-dependent selectivity
US6909630B2 (en) MRAM memories utilizing magnetic write lines
US6870759B2 (en) MRAM array with segmented magnetic write lines
CN1777956A (en) Simultaneous reading from and writing to different memory cells
US7095648B2 (en) Magnetoresistive memory cell array and MRAM memory comprising such array
US7206220B2 (en) MRAM-cell and array-architecture with maximum read-out signal and reduced electromagnetic interference
CN1957423A (en) Reversed magnetic tunneling junction for power efficient byte writing of MRAM
JP2004103125A (en) Magnetic random access memory
CN1276435C (en) Magnetic random access stroage and its producing method
CN100524531C (en) MRAM memories utilizing magnetic write lines
US6933550B2 (en) Method and system for providing a magnetic memory having a wrapped write line
CN1901087A (en) Magnetic random access storage based on circular magnetic multilayer film and its control method
WO2021237497A1 (en) Magnetic random access memory, data read-write method, and electronic device
CN1901088A (en) Magnetic random access storage based on closed magnetic multilayer film and control method
TW478146B (en) Magneto-resistance random access memory array having pseudo spin valve
CN1748317A (en) Mram cells having magnetic write lines with a stable magnetic state at the end regions

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
C17 Cessation of patent right
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20090805

Termination date: 20100111