GB2312088A

GB2312088A - Magnetoresistive devices

Info

Publication number: GB2312088A
Application number: GB9605874A
Authority: GB
Inventors: Jiri Mathon
Original assignee: City University of London
Current assignee: City University of London
Priority date: 1996-03-20
Filing date: 1996-03-20
Publication date: 1997-10-15
Anticipated expiration: 2016-03-20
Also published as: GB9605874D0; GB2312088B

Abstract

The magnetoresistive device comprises a plurality of bi-layers (1,11;2,22;n,nn) wherein each bi-layer includes an electrically conductive magnetic layer (1,2,n) in contact with an electrically conductive non-magnetic layer (11,22,nn), the bi-layers (1,11;2,22;n,nn) being stacked in series. The magnetic layers (1,2,n), forming one set of layers, alternate with the non-magnetic layers (11,22,nn) forming another set of layers. Electrical contact members (10,100) in contact with the outer layers (1,nn)are so positioned that, in operation, electrical current flowing between the electrical contact members (10,100) flows across the interfaces between the layers (1,11,2,22,n,nn). The magnetic layers (1,2,n) are magnetised in alternately opposite senses from one to the next and each layer (1,11,2,22,n,nn) is between a fraction of a nanometre and several tens of nanometres in thickness. One of the two sets of layers (1,2,n) includes a layer (1) differing in thickness from the other layers (2,n) of that set of layers (1,2,n).

Description

MAGNETORESISTIVE DEVICE The invention relates to a magnetoresistive device.

A magnetoresistive device is characterised by having a resistance that is dependent on the strength of its ambient magnetic field.

Magnetoresistive devices in accordance with the invention are intended to be used in magnetic sensors for reading magnetically stored data although they could be used in applications requiring the measurement of magnetic field strength with high spatial resolution.

The invention provides a magnetoresistive device including three bi-layers wherein each bi-layer includes an electrically conductive magnetic layer having a first of its major surfaces in contact with a first major surface of an electrically conductive non-magnetic layer, the bi-layers being so stacked in series that the second major surfaces are in contact with one another, the magnetic layers, forming one set of layers, alternating with the non-magnetic layers forming another set of layers, electrical contact members in contact with the outer layers so positioned that, in operation, electrical current flowing between the electrical contact members flows across the interfaces between the layers, the magnetic layers being magnetised in alternately opposite senses from one to the next and each layer being between a fraction of a nanometre and several tens of nanometres in thickness and one of the two sets of layers including a layer differing in thickness from the other layers of that set of layers.

Preferably, the major surfaces of the layers have dimensions that are no greater than 50 nanometres. Devices including only layers having major surfaces with dimensions no greater than 50 nanometres are expected to be more suitable for operation at around average room temperature than devices including layers having major surfaces with larger dimensions.

Preferably, all of the layers of one of the two sets of layers have thicknesses taken in sequence from a pseudorandom number sequence.

Preferably, the magnetoresistive device includes additional bi-layers stacked in series between the electrical contact members with the second major surfaces in contact with one another, the magnetic layers alternating with the non-magnetic layers and the thicknesses of one set of layers taken in sequence forming a pseudo-random number sequence.

In one arrangement, all of the magnetic layers have thicknesses taken in sequence from a pseudo-random number sequence.

In another arrangement, all of the non-magnetic layers have thicknesses taken in sequence from a pseudo random number sequence.

In a further arrangement, all of the magnetic layers have thicknesses taken in sequence from a pseudo-random sequence and, in addition, all of the non-magnetic layers have thicknesses taken in sequence from a pseudo-random sequence.

Preferably, the magnetoresistive device includes between 5 and 100 bi-layers, both limits included.

Preferably, the ratio of the maximum to the minimum thickness, for either set of layers, has a value of at least 2.

Preferably, the layers range in thickness between a minimum thickness (dmin) and a maximum thickness (may) with dmin being no less than the thickness of one atomic plane (about 0.2 nanometre) and dmax being no more than 50 nanometres.

The invention provides a method of fabricating a magnetoresistive device including the step of fabricating three bi-layers wherein each bi-layer includes an electrically conductive magnetic layer having a first major surface in contact with a first major surface of an electrically conductive non-magnetic layer and, during fabrication, so stacking the bi-layers in series that the second major surfaces are in contact with one another, the magnetic layers, forming one set of layers, alternating with the non-magnetic layers forming another set of layers, the fabrication step including the selection of the thickness for one layer in one of the two sets of layers to differ from that of the other layers of that set of layers and the thicknesses of all of the layers being selected from among values within the range extending from a fraction of a nanometre to several tens of nanometres.

Preferably, the method includes the selection, from a pseudo-random number sequence, of thicknesses for all of the layers of one of the two sets of layers taken in sequence.

Preferably, the method includes the fabrication of additional bi-layers, so stacking all of the bi-layers in series that the second major surfaces are in contact with one another and the magnetic layers alternate with the nonmagnetic layers, the thicknesses of all of the layers of one of the two sets of layers taken in sequence being selected from a pseudo-random number sequence.

The method may include the selection, from a pseudorandom number sequence, of thicknesses for all of the magnetic layers taken in sequence.

Alternatively or additionally, the method may include the selection, from a pseudo-random number sequence, of thicknesses for all of the non-magnetic layers taken in sequence.

Preferably, the method provides between 5 and 100 bilayers, both limits included.

Preferably, the method effects the fabrication of either set of layers with a ratio of the maximum thickness to the minimum thickness of a value of at least 2.

Preferably, the method effects the fabrication of layers ranging in thickness between a minimum thickness (dmin) and a maximum thickness (may) with dmin being no less than the thickness of one atomic plane (about 0.2 nanometre) and dmax being no more than 50 nanometres.

Magnetoresistive devices in accordance with the invention will now be described, by way of example only, with reference to the accompanying drawings, in which: Fig. 1 is a diagrammatic representation of a longitudinal cross-section of a first embodiment of the invention, Fig. 2 is a diagrammatic representation of a longitudinal cross-section of a second embodiment of the invention, Fig. 3 is a diagrammatic representation of a longitudinal cross-section of a third embodiment of the invention, Fig. 4 is a diagrammatic representation of the potentials experienced by minority-spin and majority-spin electrons moving along an embodiment of the invention lying in a saturating ambient magnetic field, Fig. 5 is a diagrammatic representation of the potentials experienced by minority-spin and majority-spin electrons moving along an embodiment of the invention in a zero ambient magnetic field, Fig. 6 is a diagrammatic representation first, of the magnetoresistance ratio in an embodiment of the invention having relatively large variations in layer thicknesses as the length of the device varies and, second, the variation of the magnetoresistance with length for an equivalent ordered superlattice device and Fig. 7 represents diagrammatically, first, the variation of the magnetoresistance in an embodiment of the invention having small variations in layer thicknesses as the length of the device varies and, second, the variation in magnetoresistance with length for an equivalent ordered superlattice device.

Referring to Fig. 1 of the accompanying drawings, the first embodiment includes a stack of n bi-layers each consisting of a magnetic and a non-magnetic layer. The first embodiment includes a first bi-layer consisting of a layer 1 of magnetic material and a layer 11 of non-magnetic material, a second bi-layer consisting of a layer 2 of magnetic material and a layer 22 of non-magnetic material and so on up to an end bi-layer consisting of a layer n of magnetic material and a layer nn of non-magnetic material.

The layers are so stacked in series that the major surfaces are in contact with one another and the layers have a common axis, there being a first interface between a first major surface of the magnetic layer 1 and a first major surface of the non-magnetic layer 11, a second interface between a second major surface of the non-magnetic layer 11 and a second major surface of the magnetic layer 2 and so on up to an (n-l)th interface between the magnetic layer n and its preceding non-magnetic layer followed by the nth interface between the magnetic layer n and the non-magnetic layer nn. As is evident, the magnetic layers alternate with the non-magnetic layers. Each layer is electrically conductive in the direction perpendicular to the stack which has a first electrical contact 10 in contact with the first layer 1 and a second electrical contact 100 in contact with the last layer nn for the passage of an electrical current in the direction perpendicular to the layers. For convenience, the passage of an electrical current in the direction perpendicular to the layers will be referred to as the current-perpendicular-to-planes (CPP) structure.

The non-magnetic layers 11, 22 ... nn are all of the same thickness.

The magnetic layers have thicknesses measured in units of the distance between two neighbouring atomic planes and those thicknesses taken in sequence form a pseudo-random number sequence having values ranging from 0.2 nanometres to several nanometres, the pseudo-random number sequence being obtainable from a pseudo-random number generator restricted to the range 0.2 nanometres to several nanometres. All the non-magnetic layers have the same thickness, measured in units of the distance between two neighbouring atomic planes, ranging from a fraction of a nanometre to several nanometres. Atomic planes are parallel planes in which the atoms of the materials of the layers can be considered to lie.

Suitable materials for the magnetic layers are cobalt, iron, nickel and their alloys (nickel-iron, nickelcobalt, iron-cobalt). Suitable materials for the nonmagnetic layers are copper, gold, silver, chromium, ruthenium, rhodium and vanadium.

The magnetic layers are all so magnetised that in a zero strength ambient magnetic field, the magnetic layers of any two neighbouring bi-layers are magnetised in opposite senses relative to each other. For convenience, this situation will be referred to as an "antiferromagnetic" configuration of the device.

There are several possibilities for achieving the "antiferromagnetic" configuration of the device. One possible arrangement relies on oscillatory interlayer exchange coupling to align the magnetisations of the adjacent ferromagnetic layers antiparallel by the appropriate choice of the thickness of the non-magnetic layer separating neighbouring magnetic layers. For copper or chromium layers, for example, the thickness for which the "antiferromagnetic" coupling is strongest is of the order of one nanometre. It may, however, be desirable to choose other thicknesses in brder to tailor the strength of the coupling to the intended application of the device.

(Mathon et al., Phys. Rev. Lett. 74, 3696, (1995).) Another possible arrangement would include two different magnetic materials, type A and type B, say, with respective coercive forces FA and FB, where FA < FB. The device has magnetic layers that are alternately type A and type B with the thicknesses of the non-magnetic layers so chosen that there is zero interlayer exchange coupling.

The device is initially subjected to a first magnetic field strong enough to align the magnetic moments of all of the magnetic layers in the same direction. The device is then subjected to a magnetic field applied in the opposite direction to the first field and strong enough to reverse the orientation of all the magnetic layers of type A but not strong enough to reverse the magnetic moments of the magnetic layers of type B with the higher coercive force.

The differences in coercive forces required for the magnetic layers may be provided by using different materials for the layers. For example, permalloy would provide a type A layer while iron would provide a type B layer.

The mode of operation of the device is based on a theory advanced by P.W. Anderson, implying that the lowtemperature conductance of a disordered one-dimensional conductor decreases exponentially with its length where disordered refers to the potentials seen by electrons in the conductor. The rate of decrease of the conductance is dependent on the degree of disorder, that is, the greater the degree of disorder, the greater the rate of decrease of the conductance of the conductor. The theory is discussed in an article in Physics Review, B22, 3519 (1980) attributed to P.W. Anderson, D.J. Thouless, E. Abrahams and D.S. Fisher.

The first embodiment of the device exhibits disorder due to the random distribution of the thicknesses of the ferromagnetic layers. The conductance of the device is low in its natural "antiferromagnetic" configuration in zero ambient magnetic field. The device behaves more and more like an insulator as the number of bi-layers is increased.

If, on the other hand, the device is immersed in an ambient magnetic field strong enough to align the magnetisations of all the magnetic layers in substantially the same direction the conductance of the device increases substantially. For convenience, the state with magnetisations of all the magnetic layers aligned in the same direction is referred to as the "ferromagnetic" configuration of the device.

The change in the conductance of the device may be understood by noting that current is carried by two groups of electrons, those having their spin pointing in the positive direction of a quantization axis chosen to be parallel to the direction of the magnetisation and those having spin pointing in the negative direction of the quantization axis. As they move in the structure, both groups of electrons experience quasi-one-dimensional potentials that follow a pseudo-random sequence dependent on the thicknesses of the layers. The degrees of disorder seen by electrons with opposite spin orientations when the device is in its "ferromagnetic1' configuration (with all the film magnetisations aligned parallel) are different whereas, when the device is in its "antiferromagnetic" configuration, the degrees of disorder seen by the two sets of electrons are essentially the same.

In the "ferromagnetic" configuration of the device, the degree of disorder seen by the electrons of one spin orientation is large and the disorder seen by the electrons with the opposite spin orientation is small whereas, in the "antiferromagnetic" configuration of the device, both sets of electrons see the same large degree of disorder.

In the operation of the device, on the parallel alignment of the magnetic layer moments by a saturating external magnetic field, the conductance becomes high in a channel seen by electrons with one spin to have a small degree of disorder and remains low in the channel traversed by electrons with the opposite spin orientation. The high-conductance channel shunts the low-conductance channel, making the overall conductance of the device high so that it behaves like a good conductor in the saturating magnetic field.

The resistances of the contacts 10 and 100 contribute towards the total resistance of the device in both its lowconductance and high-conductance states but since the resistances in both spin channels increase exponentially with the number of bi-layers employed, the device can be made to have a substantially larger resistance than that provided by the contact resistances both in its lowconductance and its high-conductance state.

Figs. 4 and 5 of the accompanying drawings show an example of disordered potentials seen by electrons carrying current in the device. All of the potentials in Figs. 4 and 5 are measured in dimensionless energy units in which the width of the electron conduction band in the non-magnetic layers is 6.

Referring to Fig. 4 of the accompanying drawings, in the "ferromagnetic" state of the device the barrier potentials experienced by the minority-spin electrons and represented by the broken lines are seen to be of the order of 2.8, which results in substantial confinement of the minority-spin electrons. On the other hand, the barrier potentials experienced by the majority-spin electrons and represented by the unbroken lines are seen to be 2.4 and result in substantially less confinement of the majorityspin electrons than of the minority-spin electrons. The simulated relationships represent the conditions in a device having a total thickness of the order of 60 atomic planes and indicate that the device has a relatively high conductance in the "ferromagnetic" state. The nonmagnetic layers are shown as presenting a barrier potential of 2.2 to both sets of electrons, the non-magnetic layers being, as shown, all of one thickness and the magnetic layers being, as shown, of differing thicknesses chosen from a pseudo-random number sequence.

The number of atomic planes included in each layer depends on the crystalline structure of the material and the crystal orientation in the layer but there will be between one and several tens of atomic planes in each layer.

Referring to Fig. 5 of the accompanying drawings, in the "antiferromagnetic" state of the device the barrier potentials experienced by the minority-spin electrons are the same as those experienced by the majority-spin electrons and are represented by the unbroken lines. As shown, both sets of electrons experience a barrier potential of 2.4 in each odd-numbered magnetic layer and experience a barrier potential of 2.8 in each even-numbered magnetic layer, resulting in the substantial confinement of both sets of electrons. The simulated relationships represent the conditions in a device having the same thickness as before (60 atomic planes) and indicate that the device has a relatively low conductance in the "antiferromagnetic state. As is the case for Fig. 4, the barrier potential for both sets of electrons is 2.2 in the non-magnetic layers.

Referring to Fig. 6 of the accompanying drawings, the variation of the parameter (Raf-Rfm)/Rfm, known in the art as giant magnetoresistance (GMR), with thickness is plotted for one device having the structure shown in Flg. 1 against the variation of GMR for an equivalent ordered superlattice device of equal thickness, where Raf is the resistance of the device in its "antiferromagnetic" state and Rfm is the resistance of the device in its "ferromagnetic" state; an ordered superlattice is a device with uniform thicknesses of magnetic and non-magnetic layers throughout it. The GMR is plotted on a natural logarithmic scale, its variation with device thickness being marked by circles for a device such as is shown in Fig. 1. Squares indicate the variation in GMR for an equivalent ordered superlattice device.

The thickness of the device shown in Fig. 6 is measured in the number of repeats Nrpt of a cell consisting of two neighbouring bi-layers each containing one magnetic and one non-magnetic layer.

The equivalent ordered superlattice device includes magnetic layers all having the same thickness which is the mean thickness of the layers of a device such as is shown in Fig. 1 and the equivalent ordered superlattice device has the same potentials as those of a device such as is shown in Fig. 1.

As is evident from Fig. 6, the GMR for the equivalent ordered superlattice device has a value the natural logarithm of which is about 2 while the GMR for the device having the structure shown in Fig. 1 increases from a value the natural logarithm of which is about 2 for a thickness Nrpt = 5 to a value the logarithm of which is about 25 for a thickness of Nrp = 50. At a thickness of about Nrpt = 20, the GMR is of the order of 5000 times that of the equivalent ordered superlattice device.

The device having the structure shown in Fig. 1 has magnetic layers which vary in thickness between 2 and 4 atomic planes, both limits included and non-magnetic layers having all the same thickness of 4 atomic planes.

Referring to Fig. 7 of the accompanying drawings, the variation with thickness of the GMR for a further device having the same heights of potential barriers as in Fig. 1 is compared with the GMR of its equivalent ordered superlattice device. The GMR is plotted on a natural logarithmic scale as in Fig. 6.

The further device having potential barriers of the same heights as those shown in Fig. 1 includes bi-layers in which the magnetic layers vary in thickness between 5 and 6 atomic planes, both limits included and all of the non-magnetic layers have the same thickness of 5 atomic planes. That is, the range of fluctuations of thicknesses of the magnetic layers of the further device having the structure shown in Fig. 1 is not as great as the range of fluctuations of thicknesses of the magnetic layers of the first device.

As is evident from Fig. 7, the GHR for the equivalent ordered superlattice device has a value the natural logarithm of which is about 1 while the GMR of the further device having the structure shown in Fig. 1 increases from a value the logarithm of which is about 1 for a thickness Nrpt = 5 to a value the logarithm of which is about 3 for a thickness Nrpt = 50.

In comparison with Fig. 6, Fig. 7 shows that the further device having the structure of Fig. 1 does not provide the huge increase in GHR provided by the first device having the structure of Fig. 1 but that, nevertheless, the further device having the structure of Fig. 1 provides a significant improvement in GMR in comparison with its equivalent ordered superlattice device.

Referring to Fig. 2 of the accompanying drawings, the second embodiment includes a stack of n bi-layers each consisting of a magnetic and a non-magnetic layer. The first bi-layer includes the magnetic layer 201 and the nonmagnetic layer 211, the second bi-layer includes the magnetic layer 202 and the non-magnetic layer 222 and so on, in a similar manner to the embodiment shown in Fig. 1, up to and including the n th. bi-layer consisting of the magnetic layer 20n and the non-magnetic layer 2nn.

The second embodiment differs from the first embodiment in that, in the second embodiment, the magnetic layers 201,202 ... 20n are all of one thickness. The nonmagnetic layers 211,222 ... 2nn have thicknesses measured in units of the distance between two neighbouring atomic planes and those thicknesses are selected in sequence to correspond to the output values from a pseudo-random number generator limited to a range of values for which the interlayer exchange coupling is "antiferromagnetic. The magnetic layers 201,202 ... 20n all have the same thickness which can range from a fraction of a nanometre to several nanometres.

In the second embodiment, as in the first embodiment, the bi-layers have a common axis and each bi-layer is electrically conductive in the axial direction of the stack which has a first electrical contact 10 in contact with the first bi-layer and a second electrical contact 100 in contact with the last bi-layer for the passage of an electrical current in the axial direction perpendicular to the layers. As before, the arrangement with a current passing in the axial direction is referred to as the CPP structure.

As in the case for the first embodiment, suitable materials for the magnetic films are cobalt, iron, nickel and their alloys (nickel-iron, nickel-cobalt, iron-cobalt) and suitable materials for the non-magnetic films are copper, gold, silver, chromium, ruthenium, rhodium and vanadium.

As indicated above, all the magnetic layers are so magnetised that the magnetisations of the magnetic layers in any two neighbouring bi-layers have directions opposite to one another when in a zero ambient magnetic field. As in the case of the first embodiment, that condition is the "antiferromagnetic" configuration of the device.

The antiferromagnetic configuration of the magnetic layers is achieved using oscillatory interlayer exchange coupling which aligns the magnetisations of the adjacent ferromagnetic films antiparallel when the thicknesses of all the non-magnetic interlayers are chosen from an appropriate interval. For copper interlayers between cobalt layers, for example, one appropriate interval has the thickness of 5-8 atomic planes of copper but other intervals for which the coupling is antiferromagnetic can also be chosen.

The second embodiment operates in the same manner as does the first embodiment except that disorder now arises because of the random distribution of the thicknesses of the non-magnetic layers. The relationships shown in Figs.

4 to 7 apply equally to the second embodiment. It is to be understood that there could be a first form of the second embodiment with large variations in the thicknesses of the non-magnetic layers 211,222 ... 2nn giving rise to relationships similar to those shown in Fig. 6 and a further form of the second embodiment with relatively small variations in the thicknesses of the non-magnetic layers 211,222 ... 2nn giving rise to relationships similar to those shown in Fig. 7.

As is the case with the first embodiment, arrangements giving rise to the relationships shown in Fig.

7 represent an improvement over known devices although the improvement is not as great as that provided by arrangements giving rise to the relationships shown in Fig. 6.

Referring to Fig. 3 of the accompanying drawings, the third embodiment includes a stack of n bi-layers each consisting of a magnetic and a non-magnetic layer. The first bi-layer includes the magnetic layer 301 and the nonmagnetic layer 311, the second bi-layer includes the magnetic layer 302 and the non-magnetic layer 322 and so on, in a similar manner to the embodiment shown in Fig. 1, up to and including the n th. bi-layer consisting of the magnetic layer 30n and the non-magnetic layer 3nn.

The third embodiment differs from the first embodiment in that, in the third embodiment, not only do the magnetic layers 301, 302 ... 30n differ in thickness from one another but, also, the non-magnetic layers 311,322 ... 3nn differ in thickness from one another.

As is the case for the first embodiment, the bilayers of the third embodiment have a common axis and each layer is electrically conductive in the axial direction of the stack which has a first electrical contact 10 in contact with the first layer and a second electrical contact 100 in contact with the last layer for the passage of an electrical current in the axial direction perpendicular to the layers, that is, the third embodiment is a CPP arrangement.

For the third embodiment the thicknesses of the magnetic layers are selected from the same interval as is used in the first embodiment and the thicknesses of the non-magnetic layers are selected from the same interval as is used in the second embodiment.

As is the case for the first embodiment, suitable materials for the magnetic layers are cobalt, iron, nickel and their alloys (nickel-iron, nickel-cobalt, iron-cobalt) and suitable materials for the non-magnetic layers include copper, gold, silver, chromium, ruthenium, rhodium and vanadium.

In the third embodiment, all of the magnetic layers are so magnetised that the magnetisations in any two neighbouring bi-layers have directions opposite to one another when the ambient magnetic field strength is zero, that is, the device has the "antiferromagnetic" configuration in a zero strength ambient magnetic field.

As before, the "antiferromagnetic" configuration of the magnetic layers is achieved using oscillatory interlayer exchange coupling which aligns the magnetisations of the adjacent ferromagnetic layers antiparallel by choosing the thicknesses of the nonmagnetic interlayers from an appropriate interval. As indicated above, for copper interlayers between cobalt layers, for example, one appropriate interval has the thickness of 5-8 atomic planes of copper and other intervals in which the coupling is antiferromagnetic can also be chosen.

The third embodiment operates in the same manner as do the first and second embodiments, disorder now arising because both the magnetic and the non-magnetic layer thicknesses taken in sequence are pseudo-random. The relationships shown in Figs. 4 to 7 apply equally to the third embodiment. It is to be understood that there could be a first form of the third embodiment with large variations in the thicknesses both of the magnetic layers 301,302 ... 30n and the non-magnetic layers 311,322 ... 3nn giving rise to relationships similar to those shown in Fig. 6 and a further form of the third embodiment with relatively small variations of those thicknesses giving rise to relationships similar to those shown in Fig. 7.

As is the case with the first embodiment, arrangements giving rise to the relationships shown in Fig. 7 represent an improvement over known devices although the improvement is not as great as that provided by arrangements giving rise to the relationships shown in Fig. 6.

Irrespective of the materials used, the devices shown in Figs. 1, 2 and 3 show enhanced performance in comparison with known magnetoresistive devices using magnetic layers alternating with non-magnetic layers in the following respects: First, the ratio (Raf-Rfm)/Rfm is significantly greater than for the known devices, Raf being the resistance of the device in its "anti ferromagnetic" conf (GMR), increases exponentially with the number of bi-layers N for the devices shown in Figs. 1, 2 and 3, while it increases, at best, linearly with N for the known devices.

Second, the resistances of the contacts contribute towards the total resistance of any magnetoresistive device using alternating magnetic and non-magnetic layers. For the effective operation of any magnetoresistive device, the ratio of the device resistance to the contact resistance in both its low-conductance and high-conductance state should be large. That condition is difficult to achieve for the known devices using the CPP structure but since the total resistance in each spin channel of any one of the three embodiments of the present device increases exponentially with the number N of bi-layers, the device can be made to have a resistance in either its low-conductance or its high-conductance state that is significantly larger than conventional known CPP devices and, consequently, the contact resistance problem of the CPP-geometry device is minimised.

The number of bi-layers included in any one of the three devices shown in Figs. 1, 2 and 3 can be in the range 3 to 100, both limits included. In practice, a large number of bi-layers is preferred but the maximum number of bi-layers that could be included in any one of the three devices would depend on the maximum number of layers that could be fabricated with substantially no impurities, since impurities would reduce the electron mean free path in the device. The dimensions transverse to the current flow would be comparable to or smaller than the electron mean free path in the current flow direction.

The fabrication of the layers can be by molecular beam epitaxy (MBE), chemical vapour deposition (CVD), sputtering or electrodeposition using one of the electrical contacts 10,100 as a substrate, the other electrical contact being fabricated, in due course by MBE, CVD, sputtering or electrodeposition.

Multilayer devices as described above can be fabricated as wires by the electrodeposition of the layers into pores in polycarbonate foils and the diameters of those wires are of the order of several tens of nanometres.

It would be possible, following the fabrication of the device by other methods, to reduce the diameter by chemical etching (lithography methods).

Magnetoresistive devices in accordance with the invention may be referred to as magnetoresistive spin valves. The magnetoresistive spin valves can consist of bi-layers in the shapes of convex polygons, for example, rectangles, circles or ellipses.

The results shown in Figs. 6 and 7 for devices in accordance with the invention relate to their simulated performance at temperatures approaching 0 Kelvin, the devices including only layers having major surfaces with dimensions smaller than the electron mean free path at those temperatures.

In devices in accordance with the invention, the magnetisation of the magnetic layers may be either parallel to the planes of the layers or perpendicular to the planes of the layers.

The predicted enhancement of the GMR for devices in accordance with the invention under conditions permitting the existence of a "localisation length", as suggested by Anderson, referred to above, is expected to provide an exponential increase with length that would give an enhancement factor of the GMR of the order of 1000 in comparison with conventional devices.

If, above some particular temperature, the "localisation length" becomes smaller than the dimensions of the device, then the GMR is no longer expected to increase exponentially with length but, in those circumstances, devices in accordance with the invention may be expected to provide a GMR enhancement which increases linearly with the length of the device giving a still significant improvement over conventional devices.

The temperature at which the GMR ceases to increase exponentially with length can be above average room temperature for devices in accordance with the invention including only layers with major surfaces having dimensions no greater than 50 nanometres and such devices would then show an exponential increase in GMR with length at average room temperature.

Claims

CLAIMS:

1. A magnetoresistive device including three bilayers wherein each bi-layer includes an electrically conductive magnetic layer having a first major surface in contact with a first major surface of an electrically conductive non-magnetic layer, the bi-layers being so stacked in series that the second major surfaces are in contact with one another, the magnetic layers, forming one set of layers, alternating with the non-magnetic layers forming another set of layers, electrical contact members in contact with the outer layers so positioned that, in operation, electrical current flowing between the electrical contact members flows across the interfaces between the layers, the magnetic layers being magnetised in alternately opposite senses from one to the next and each layer being between a fraction of a nanometre and several tens of nanometres in thickness and one of the two sets of layers including a layer differing in thickness from the other layers of that set of layers.

2. A magnetoresistive device as claimed in claim 1, wherein the major surfaces of the layers have dimensions no greater than 50 nanometres.

3. A magnetoresistive device as claimed in claim 1 or 2, wherein all of the layers of one of the two sets of layers have thicknesses taken in sequence from a pseudo-random number sequence.

4. A magnetoresistive device as claimed in any one of claims 1 to 3, including additional bi-layers so stacked in series between the electrical contact members that the second major surfaces are in contact with one another, the magnetic layers alternating with the non-magnetic layers and the thicknesses of one set of layers taken in sequence forming a pseudo-random number sequence.

5. A magnetoresistive device as claimed in any one of claims 1 to 4, wherein all of the magnetic layers have thicknesses taken in sequence from a pseudo-random number sequence.

6. A magnetoresistive device as claimed in any one of claims 1 to 4, wherein all of the non-magnetic layers have thicknesses taken in sequence from a pseude-random number sequence.

7. A magnetoresistive device as claimed in any one of claims 1 to 4, wherein all of the non-magnetic layers have thicknesses taken in sequence from a pseudo-random number sequence and all of the magnetic layers have thicknesses taken in sequence from a pseudo-random number sequence.

8. A magnetoresistive device as claimed in any one of claims 1 to 7, including between 3 and 100 bi-layers, both limits included.

9. A magnetoresistive device as claimed in any one of claims 1 to 8, wherein the ratio of the maximum to the minimum thickness in either set of layers has a value of at least 2.

10. A magnetoresistive device as claimed in any one of claims 1 to 9, wherein the layers range in thickness between a minimum thickness (dmin) and a maximum thickness (may) with dmin being no less than the thickness of one atomic plane (about 0.2 nanometre) and dmax being no more than 50 nanometres.

11. A magnetoresistive device substantially as herein described with reference to and as shown by Fig. 1 of the accompanying drawings.

12. A magnetoresistive device substantially as herein described with reference to and as shown by Fig. 2 of the accompanying drawings.

13. A magnetoresistive device substantially as herein described with reference to and as shown by Fig. 3 of the accompanying drawings.

14. A method of fabricating a magnetoresistive device including the step of fabricating three bi-layers wherein each bi-layer includes an electrically conductive magnetic layer having a first major surface in contact with a first major surface of an electrically conductive nonmagnetic layer and, during fabrication, so stacking the bilayers in series that the second major surfaces are in contact with one another, the magnetic layers, forming one set of layers, alternating with the non-magnetic layers forming another set of layers, the fabrication step including the selection of the thickness for one layer in one of the two sets of layers to differ from the thicknesses of the other layers in that set of layers and the thicknesses of all of the layers being selected from among values within the range extending from a fraction of a nanometre to several tens of nanometres.

15. A method as claimed in claim 14, including the selection, from a pseudo-random number sequence, of the thicknesses of all of the layers of one of the two sets of layers taken in sequence.

16. A method as claimed in claim 14 or 15, including the fabrication of additional bi-layers and so stacking all of the bi-layers in series that the second major surfaces are in contact with one another and the magnetic layers alternate with the non-magnetic layers, the thicknesses of all of the layers of one of the two sets of layers taken in sequence being selected from a pseudorandom number sequence.

17. A method as claimed in any one of claim 14 to 16, including the selection, from a pseudo-random number sequence, of thicknesses for all of the magnetic layers, taken in sequence.

18. A method as claimed in any one of claims 14 to 16, including the selection, from a pseudo-random number sequence, of thicknesses for all of the non-magnetic layers taken in sequence.

19. A method as claimed in any one of claims 14 to 16, including the selection, from a pseudo-random number sequence, of thicknesses for all of the non-magnetic layers taken in sequence and the selection, from a pseudo-random number sequence, of thicknesses for all of the magnetic layers taken in sequence.

20. A method as claimed in any one of claims 14 to 19, providing between 3 and 100 bi-layers, both limits included.

21. A method as claimed in any one of claims 14 to 20, effecting the fabrication of either set of layers with a value of at least 2 for the ratio of the thickest to the thinnest layer.

22. A method as claimed in any one of claims 14 to 21, effecting the fabrication of layers ranging in thickness between a minimum thickness (dmin) and a maximum thickness (may) with dmin being no less than the thickness of one atomic plane (about 0.2 nanometre) and dmax being no more than 50 nanometres.

23. A method as claimed in any one of claims 14 to 22, effecting the fabrication of layers having major surfaces with dimensions no greater than 50 nanometres.

24. A method of fabricating a magnetoresistive device substantially as herein described with reference to the accompanying drawings.