WO2001023903A1

WO2001023903A1 - Magnetic field sensor

Info

Publication number: WO2001023903A1
Application number: PCT/SE2000/001828
Authority: WO
Inventors: Olle Eriksson; Peter James; Börje Johansson; Lars NORDSTRÖM; Adrian Taga
Original assignee: Olle Eriksson; Peter James; Johansson Boerje; Nordstroem Lars; Adrian Taga
Priority date: 1999-09-27
Filing date: 2000-09-20
Publication date: 2001-04-05
Also published as: AU7696400A; SE9903481D0

Abstract

A magnetoresistive sensor is disclosed, which comprises of at least set of film layers, which set in turn comprises of a substrate layer (NM1), a first ferromagnetic layer (FM1), an intermediate layer (NM2) and a second ferromagnetic layer (FM4) placed on top of each other in a superlattice structure. The first ferromagnetic layer (FM1), preferably formed of iron, has a magnetisation parallel to the layer, in-plane, while the second ferromagnetic layer (FM4), formed of cobalt, has a magnetisation substantially perpendicular to the layer, out-of-plane. The ferromagnetic layers (FM1, FM4) are preferably tetragonally distorted epitaxially grown crystals, which are grown on the substrate and intermediate layers (NM1, NM2), respectively. The substrate and intermediate layers (NM1, NM2) are preferably crystalline layers of V, Cu, Ag, Au, Pd, Pt, Rh, Ru or Ir, which causes a distortion of the Fe layer corresponding to a c/a ratio of between 0.80 and 0.96 and of the Co layer corresponding to a c/a ratio of between 0.78 and 0.96. All layers are preferably very thin, most preferably less than 6-7 atomic layers.

Description

MAGNETIC FIELD SENSOR

TECHNICAL FIELD

The present invention relates generally to magnetic field sensors and in particular to magnetic field sensors using giant magnetoresistance effects. The present invention also relates to a manufacturing method for such devices.

BACKGROUND

Ever since the discovery of the effect of giant magnetoresistance (GMR), devices exhibiting such effects have been used for detection of magnetic fields. In the U.S. patent 4,949,039, a magnetic field sensor based on the GMR effect is presented. A huge number of other documents have been presented with different modifications and developments of that technique.

One common type of GMR-sensors is built by two or several ferromagnetic layers, separated by intermediate non-magnetic conducting layers, in a so called superlattice geometry. The ferromagnetic layers are arranged to exhibit a magnetisation directed in opposite directions with respect of each other, i.e. antiparallel, in the absence of an applied external magnetic field. This antiparallel ordering is caused by the so called interlayer exchange coupling. If an external magnetic field is present, which is stronger than the interlayer exchange coupling, the magnetisation of one of the layers may change, and the magnetisation in the superlattice becomes parallel. By determining if the magnetisation is parallel or non-parallel, detection of a magnetic field is performed.

When measuring the electrical properties perpendicular to the layers, one discovers that the resistance of the set of layers varies substantially with applied magnetic field. In the absence of an external magnetic field, the resistance is higher than in the presence of a fairly strong magnetic field. The physical explanation of this phenomenon is as follows. Conductivity of a material is generally determined by the electron structure at the Fermi-level. Furthermore, the conductivity in a ferromagnetic material is mainly due to electrons of one spin direction, the majority spin electrons. These electrons experience a potential which is lower than the Fermi-level. However, electrons of opposite spin direction, minority spin electrons, experiences a potential which is above the Fermi level and are therefore not contributing to the conductivity to any large extent. When majority spin electrons from one ferromagnetic layer penetrates the intermediate layer, without loosing their spin orientation, and enter into the anti-parallel ferromagnetic layer, they will appear as minority spin electrons, and the conduction is strongly restricted. Thus, when the distance between the layers is small in comparison with the electron mean free path, all electrons will experience a high potential, in either of the anti-parallel ferromagnetic layers. When an external magnetic field is applied, the magnetisation direction of the ferromagnetic layers are arranged parallel, and the majority spin electrons easily travel through both layers, thus giving a relatively low resistance. A description of the GMR effect can e.g. be found in "Layered magnetic structures: interlayer exchange coupling and giant magnetoresistance" by A. Fert, P. Grunberg, A. Barthelemy, F. Petroff and W. Zinn, Journal of

Magnetism and Magnetic Materials 140- 144 (1995), Elsevier Science B.V., pages 1 - 8.

A general requirement for magnetic field sensors using the GMR effect is that the external magnetic field, necessary to flip the magnetisation direction of one of the ferromagnetic layers, should be as small as possible. This influences the sensitivity of the sensor. The magnetic coupling between the ferromagnetic layers therefore has to be reduced in most cases. This is normally done by adjusting the width of the intermediate layer. A typical width of the intermediate layer is between 20 and 40 A. However, an increased width of the intermediate layer will also increase the scattering probability for electrons travelling through the intermediate layer, and an increased portion of the electrons will undergo scattering processes, in which their spin direction is not conserved. When the intermediate layer thickness is a substantial part of the mean free path of the electrons, a non-neglectable portion of the electrons has changed the spin direction, and the GMR effect is strongly attenuated.

Thus, in GMR sensors according to the prior art, there is a severe trade-off between magnetic field sensitivity and magnitude of the GMR effect. Furthermore, when using a relatively thick intermediate layer, the total set of layers becomes relatively thick, and the density of ferromagnetic layers that contribute to the GMR effect is low.

SUMMARY

An object of the present invention is thus to provide a GMR device, with a weakened coupling between adjacent ferromagnetic layers. Another object of the present invention is to reduce the thickness of the intermediate layer. A further object of the present invention is to provide a manufacturing method for such improved GMR devices.

The above objects are achieved by a magnetoresistive sensor and a manufacturing method therefore, according to the enclosed claims.

In general terms, a magnetoresistive sensor is provided, which comprises of at least one set of film layers, which set in turn comprises of a substrate layer, a first ferromagnetic layer, an intermediate layer and a second ferromagnetic layer placed on top of each other in a superlattice structure. The first ferromagnetic layer, preferably formed of iron, has a magnetisation parallel to the layer, in-plane, while the second ferromagnetic layer, preferably formed of cobalt, has a magnetisation substantially perpendicular to the layer, out-of-plane. The ferromagnetic layers are preferably tetragonally distorted epitaxially grown crystals, which are grown on the substrate and intermediate layers, respectively. The substrate and intermediate layers are preferably crystalline layers of V, Nb, Ta, Cu, Ag, Au, Pd, Pt, Rh, Ru, Os or Ir or alloys containing these elements, which causes a distortion of the Fe layer corresponding to a c/a ratio of between 0.80 and

0.96 and preferably between 0.82 and 0.86 and of the Co layer corresponding to a c/a ratio of between 0.78 and 0.96 and preferably between 0.78 and 0.82. All layers are preferably very thin, most preferably less than 6 atomic layers.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention, together with further objects and advantages thereof, may best be understood by making reference to the following description taken together with the accompanying drawings, in which:

FIG. 1 is a diagram illustrating effective potentials for electrons in a GMR superlattice exhibiting parallel magnetisation; FIG. 2 is a diagram illustrating effective potentials for electrons in a GMR superlattice exhibiting antiparallel magnetisation;

FIG. 3 is a diagram illustrating effective potentials for electrons in a GMR superlattice according to the present invention in the absence of any external magnetic field; FIG. 4 is a schematic drawing illustrating a face centred cubic crystal structure;

FIG. 5 is a schematic drawing illustrating a body centred cubic crystal- structure;

FIG. 6 is a diagram illustrating calculated values of magneto crystalline anisotropy energies for tetragonally distorted Fe an Co;

FIG. 7 is a schematic drawing of a GMR device according to prior art; FIG. 8 is a schematic drawing of a GMR device according to the present invention;

FIG. 9 is a schematic drawing of epitaxial growth of distorted crystals; and FIG. 10 is a flow diagram illustrating a manufacturing process according to the present invention. DETAILED DESCRIPTION

In the following description, the term "layer" will refer to a volume of atoms with substantially equal element composition. Such a "layer" should not be interpreted as a "monolayer", a single covering coating of atoms, i.e. a one atomic distance thick coating. A "layer" in the present description may thus comprise material with a thickness from one up to several atomic "monolayers".

The detailed description will start with a description of the physical process behind the giant magnetoresistance (GMR). In the upper part, fig. 1 schematically illustrates a superlattice of two ferromagnetic layers FM 1 , FM2 separated by a non-magnetic conducting layer NM. The magnetisation direction is indicated by arrows, which reveals that the illustrated superlattice is in a parallel state. The non-magnetic conducting layer NM is narrow enough to let the majority of the electrons passing, keep their spin direction.

In a ferromagnetic material, the electrical current is carried by majority electrons (spin-up) and minority electrons (spin-down) in two approximately independent channels. The conductivities of the two channels is normally very different. The majority and minority electrons thus experiences different- efficient potentials in the ferromagnetic material. This is illustrated in the two diagrams in the lower part of fig. 1.

In the upper diagram, denoted "spin-up", the efficient potential for majority electrons is sketched. In the ferromagnetic materials FM l, FM2, the potential is situated a certain level below the Fermi level, which means that the electrons are more or less freely movable to be used as conducting electrons. The magnitude of the potential of the non-magnetic layer NM is not very essential, since it is situated below the Fermi level (conducting material). It can be higher than the potentials in the ferromagnetic layers or as illustrated in the diagram, lower. The potential differences at the interfaces may give rise to some scattering processes, but the three layer superlattice is substantially conducting, when only considering majority electrons.

In the bottom diagram denoted "spin-down", the efficient potential for minority electrons is sketched. In the ferromagnetic materials FMl, FM2, the potential, for spin down electrons is situated a certain level above the Fermi level, which means that the electrons can not be used as conducting electrons. Some tunnelling effects may be present, but the current conducted by minority electrons is substantially zero. The magnitude of the potential of the non-magnetic layer NM is not very essential, since it is situated below the Fermi level (conducting material). It can be higher than the potentials in the ferromagnetic layers or as illustrated in the diagram, lower. The non-magnetic layer NM has potentials that are equal for both spin directions.

From fig. 1 , it is easily concluded that a system with a superlattice of parallel ferromagnetic layers is conducting, and the conducting electrons consist mainly of majority electrons. The majority conducting channel is enough to give rise to a good electrical conduction.

In the upper part, fig. 2 schematically illustrates a superlattice of two- ferromagnetic layers FMl, FM3 separated by a non-magnetic conducting layer NM. The magnetisation directions of the layers FMl, FM3 are indicated by arrows, which reveals that the illustrated superlattice is in an antiparallel state. The ferromagnetic layer FM l has a magnetisation in one direction and the ferromagnetic layer FM3 has a magnetisation in the opposite direction. The non-magnetic conducting layer NM is narrow enough to let the majority of the electrons passing, keeping their spin direction. The efficient potentials for minority and majority electrons are illustrated in the two diagrams in the lower part of fig. 2. In the upper diagram, denoted "spin-up", the efficient potential for majority electrons of the first ferromagnetic layer FMl, is sketched. In the ferromagnetic material FMl, the potential is situated a certain level below the Fermi level, which means that the electrons are more or less freely 5 movable to be used as conducting electrons, just as in fig. 1. When the majority electrons of FMl travel through the non-magnetic layer, they will still experience the same potential as in fig. 1. However, when the electrons enter the second ferromagnetic layer FM3, they will now be considered as minority electrons, since the relative direction of the spin and the .0 magnetisation is reversed. The electrons will thus experience a potential above the Fermi level, and are substantially prohibited to be used as conducting electrons.

In the bottom diagram denoted "spin-down", the efficient potential for

15 minority electrons of the first ferromagnetic layer FMl is sketched. The situation is just opposite the conditions for the upper diagram of fig. 2. In the first ferromagnetic layer, the electrons experience a high potential and are prohibited from conducting, while the potential in the second ferromagnetic layer is low and allows for conduction.

20

Since the non-magnetic layer is sufficiently thin, not to alter the spin directions of electrons travelling through the layer, the spin is conserved- between the two ferromagnetic layers. It is now easily understood, that neither spin-up or spin-down electrons have any free conducting channels,

"5 why the conduction through the superlattice is low, i.e. the resistance increases considerably. This is the origin of the giant magnetoresistance effect, used in conventional GMR devices. Fig. 1 corresponds to the conditions, when an external magnetic field is present, where the resistance through the superlattice is low. Fig 2 corresponds to the conditions, when no 0 external magnetic fields are present and the interlayer exchange coupling causes the magnetisation to be antiparallel. The resistance through the superlattice is high. In the upper part, fig. 3 schematically illustrates a superlattice of two ferromagnetic layers FM 1 , FM4 according to the present invention, separated by a non-magnetic conducting layer NM. The magnetisation directions of the layers FMl , FM4 are indicated by arrows, which reveals that the illustrated superlattice has magnetisation directions which are substantially perpendicular to each other. The first ferromagnetic layer has an in-plane magnetisation and the second ferromagnetic layer FM4 has an out-of-plane magnetisation. The non-magnetic conducting layer NM is also here narrow enough to let the majority of the electrons passing, keep their spin direction. The efficient potential for minority and majority electrons are illustrated in the two diagrams in the lower part of fig. 3.

In the upper diagram, denoted "spin-up", the efficient potential for majority electrons of the first ferromagnetic layer FMl, is sketched. In the ferromagnetic material FMl , the potential is situated a certain level below the Fermi level, which means that the electrons are more or less freely movable to be used as conducting electrons, just as in fig. 1. When the majority electrons of FMl travel through the non-magnetic layer, they will still experience the same potential as in fig. 1. However, when the electrons enter the second ferromagnetic layer FM4, the magnetisation is different. The electrons will here experience a potential above the Fermi level, and are substantially prohibited to be used as conducting electrons. However, the magnitude of this potential is somewhat smaller than for the antiparallel case, so the prohibition of conduction is not quite as efficient as in the antiparallel case. For electrons of opposite spin, the situation in the bottom diagram is valid. Since the out-of-plane magnetisation is symmetric concerning spin-up and spin-down electrons, the efficient potential in the second ferromagnetic layer FM4 is the same for both spins

The Fermi-level of the isolated atomic species is different, but when the atoms are combined to a multi-layer, they will adopt a common Fermi-level that is branched by the Fermi-levels of the isolated elements, as a result of redistribution of charge. According to calculations, the efficient potential in systems of the present invention will normally be placed above the Fermi- level. Furthermore, in conventional magnetic sensor devices, the position of the Fermi-level with respect to the effective potential of the conducting electrons is adjusted by means of the number of layers of different atomic species. This is also valid for the present multi- layer system, so that the effective potential always lies above the Fermi-level for layer FM4. If necessary, the tuning of the Fermi-level with respect to the effective potential of an individual layer can be realised by changing the relative thickness of the different layers, since the Fermi-level of the multi-layer will approach the Fermi-level of any of the individual atomic levels, provided that the thickness of that atomic layer dominates.

Since the non -magnetic layer is enough thin, not to alter the spin directions of electrons travelling through the layer, the spin is conserved between the two ferromagnetic layers. It is now easily understood, that neither spin-up or spin-down electrons have any free conducting channels, why the conduction through the superlattice is rather low, i.e. the resistance increases somewhat. This is the origin of the giant magnetoresistance effect, used in GMR devices according to the present invention. In a first embodiment, fig. 1 corresponds to the conditions, when an external magnetic field is present, where the resistance through the superlattice is low. Fig 3 then corresponds to the conditions, when no external magnetic fields are present and the- magnetisation of the different ferromagnetic layers are essentially perpendicular, one out-of-plane, one in-plane. The resistance through the superlattice is rather high. In a second embodiment, fig. 1 corresponds to the conditions, when no external magnetic field is present, where the resistance through the superlattice is low. Fig 3 then corresponds to the conditions, when external magnetic fields are present and turns the magnetisation of the different ferromagnetic layers essentially perpendicularly to each other, one out-of-plane, one in-plane. The resistance through the superlattice is rather high. It is of course understood by someone skilled in the art, that the sequence of ferromagnetic layers may be repeated, in order to enhance the sensitivity and total GMR effect. A preferred embodiment of a device according to the present invention thus comprises a superlattice of ferromagnetic and non- magnetic layers, where every second ferromagnetic layer has a magnetisation perpendicular to the plane and the others have a magnetisation in-plane. The structure may also be described as a multitude of sets of layers, which sets of layers comprise one first ferromagnetic layer and one second ferromagnetic layer, separated by an intermediate layer. On top of one of the ferromagnetic layers, another non-magnetic layer is formed, forming the substrate for next set of layers.

The present invention discloses a category of materials which are of interest in GMR devices. The materials have distorted crystal structures, compared with the naturally occurring bulk crystal structures. In order to understand the nature of the crystal distortions, definitions of different concepts in connection with these crystal structures will be described.

Fig. 4 shows the structure of a face centred cubic (fee) crystal. Lattice positions are indicated by circles. In the materials discussed in this document, each lattice point is associated with one atom. Eight atoms are placed in the corners of a cube, indicated by filled circles. Six additional- atoms, indicated as open circles, are placed in the centre position of each of side of the cube. Lattice parameters, a and c are defined as the distance along the x (or y) axis and the z axis, respectively, between two corner atoms.

It is obvious that in a undistorted fee crystal, the lattice parameter c is equal to the lattice parameter a, i.e. the c/a ratio is equal to 1.

Fig. 5 shows the structure of a body centred cubic (bcc) crystal. Lattice positions are indicated by circles. In the materials discussed in this document, each lattice point is associated with one atom. Eight atoms are placed in the corners of a cube, indicated by filled circles. One additional atom, indicated by an open circle, is placed in the centre of the cube. Lattice parameters, a and c are defined as the distance along the x (or y) axis and the z axis, respectively, between two corner atoms. It is obvious that in an undistorted bcc crystal, the lattice parameter c is equal to the lattice parameter a, i.e. the c/a ratio is equal to 1.

The materials of the present invention have tetragonally distorted crystal structures. Such structures are closely related to the cubic crystal structures, shown in figs. 4 and 5. If the lattice parameter c is different from the lattice parameter a, i.e. if the cubic structures in figs. 4 and 5 are compressed or stretched in the z direction, the structure is called a tetragonal structure. The c/a ratio is then smaller or larger than unity.

Since the choice of lattice parameters for e.g. tetragonal structures are not unique, it is convenient to relate them to the choice of lattice parameters for a cubic structure. By stretching or compressing cubic or tetragonal structures, one may pass from one type of structure to another. If the bcc structure, see fig. 5, is stretched in the z direction by about 41 %, the diagonal plane, indicated as a shadowed plane, will become quadratic. The structure may then be described as an fee structure, where the lattice parameters are directed along and perpendicular to the diagonal plane, respectively. It is thus possible to go from one cubic structure to another by a simple one axis expansion or compression. Such path is referred to as the- Bain path.

In bulk crystals, Fe adopts the bcc structure and Co adopts the hep structure (not described). However, in thin layers, these bulk crystal structures may be modified. How to achieve this is described further below. Co may for instance exist in thin films in either fee or bcc structures. In particular, Fe and Co may be produced in tetragonally distorted structures.

Now consider a preferred embodiment of the present invention, where a layer is built up by a substrate layer of V, a first ferromagnetic layer of tetragonally distorted Fe, an intermediate layer of V and a second ferromagnetic layer of tetragonally distorted Co. In a crystal of ferromagnetic material, electron interactions determine the properties of the material. Within the layer, the electronic contribution and the dipole-dipole interaction competes, and when the electronic contribution dominates an out-of-plane magnetisation is achieved.

The Fe layer exhibits a magnetisation parallel to the layer, i.e. an in-plane magnetisation. This is an effect of the relativistic spin orbit coupling in combination with the structural strain. The relativistic spin orbit coupling follows mainly the magneto crystalline anisotropy. The energies for the magnetic crystalline anisotropy for tetragonally distorted Fe crystals are calculated by ab-initio methods and presented in fig. 6. A Fe layer grown on V gives a mismatch producing a c/a ratio of approximately 0.84. As seen in fig. 6, the magneto crystalline anisotropy energies (MAE) are negative, which means that the magnetisation is placed in-plane. A c/a ratio interval between 0.80 and 0.96 is of interest, and in particular 0.82 to 0.86.

The Co layer exhibits a magnetisation perpendicular to the layer, i.e. an out- of-plane magnetisation. This is an effect of the relativistic spin orbit coupling in combination with the structural strain. The relativistic spin orbit coupling follows mainly the magneto crystalline anisotropy. The energies for the magnetic crystalline anisotropy for tetragonally distorted Co crystals are- calculated by ab-initio methods and also presented in fig. 6. A Co layer grown on V gives a mismatch producing a c/a ratio of approximately 0.80. As seen in fig. 6, the magneto crystalline anisotropy energies (MAE) are positive, which means that the magnetisation is placed out-of-plane. A c/a ratio interval between 0.78 and 0.96 is of interest, and in particular 0.78 to 0.82.

When placing the two ferromagnetic layers close to each other, only separated by a thin intermediate layer, interlayer interactions are introduced. The exchange interaction between the two ferromagnetic layers tends to turn the magnetisation in order to arrange them parallel in a ferromagnetic arrangement. This interaction thus counteracts the effect of the magneto crystalline anisotropy. If the thickness of the intermediate layer is small, the interlayer exchange splitting is in the same order of magnitude as the magneto crystalline contribution, but of the opposite sign. The two competing interactions can thus be made to produce a tiny energy difference between a perpendicular and parallel magnetic orientation.

In table 1, typical values of the interlayer exchange interactions are shown for some pertinent device geometries. It may e.g. be observed that for 3 Fe layers separated from 3 Co layers with 5 V layers, the exchange interaction produces an energy difference between the ferromagnetic and perpendicular magnetisation configurations of about 1.1 meV. At the same time, from fig. 6, it may be observed that the energy cost due to the MAE of turning two layers of Fe in the same direction as Co costs an energy of about the same amount. The two interactions thus cancel out and the energy difference between perpendicular and parallel magnetisation direction is only marginally more stable in one of the configurations. Hence, for a device with a geometry that favours a perpendicular configuration, even a very weak magnetic field may easily flip the configuration to ferromagnetic. Alternatively for slightly different geometries, where a ferromagnetic configuration is stable, by a small energy, a tiny applied field may stabilise the perpendicular configuration. Further candidates could e.g. be^~ 6Fe/4V/2Co or 6Fe/6V/2Co.

Geometry Energy (me VI

3Fe/4V/3Co 3.4

3Fe/3V/3Co 14.2

3Fe/5V/3Co 1.1

3Fe/6V/3Co 3.5

TABLE 1. Interlayer exchange interaction energies. Although only the combination of Fe and Co has been discussed so far, also other combinations of metals and alloys are conceivable. A number of possible candidates could e.g. be Co/Pt/Ni, Coo.9Feo.ι/Pto.9Rho.ι/Nio.9Coo.ι, Coo.ιFeo.9/Vo.9Nbo.ι/Feo.5Coo.5 and Feo.9Sio.ι/Vo.9Nbo.ι/Coo.ιFeo.9. The two first examples exhibit a distorted fee structure (fct) and the two last ones exhibit a distorted bcc structure (bet). Anyone skilled in the art understands that a large number of combinations are possible.

One important advantage with devices according to the present invention is that the magnetic sensitive layers, i.e. the ferromagnetic layers which may change the direction of magnetisation, is placed much closer than for conventional devices. In fig. 7, a device according to prior art is shown. The distance between the ferromagnetic layers FM is big, in typical cases 10-20 atomic layers. Thick intermediate layers give rise to scattering which does not conserve the electron spin, so called incoherent scattering, which reduces the GMR effect. The distance D between two consecutive ferromagnetic layers of the same type, i.e. the length of the device repetition unit RU, is in a quantum mechanical point of view very long. The set of layers, discussed above, hence constitutes such a repetition unit RU.

In fig. 8, a corresponding illustration of a device according to the present invention is shown. Here, the distance between neighbouring ferromagnetic- layers is much smaller, on the order of 2 to 7 atomic layers. The distance d between two consecutive ferromagnetic layers of the same type, i.e. the length of the device repetition unit RU, is in a quantum mechanical point of view moderate. This has the advantage of decreasing the disturbances caused by incoherent scattering, and at the same time, several repetition units according to the present invention may be placed within the same volume as one single repetition unit according to prior art. This leads to a better sensitivity of the GMR device.

In fig. 7 and 8, the resistance sensing means are schematically illustrated by a voltage supply unit 101 and a current meter 100. Any such equipment according to prior art may be used for detecting the changes in resistance. Since this particular feature is not the scope for the present invention, it is not further discussed, and references are made to standard literature in this field. 5

Materials according to the present invention are possible to produce by different known techniques. The presently preferred method would be by using epitaxial crystal growth of thin films.

u.0 Referring to fig. 9, when a crystal surface of one element, A, is covered by atoms of another element, X, it is quite common that the X atoms, at least initially (in some cases up to 50 atomic layers), adopts to the crystal lattice of element A. If X in bulk material has larger lattice constants than A, the X atoms are forced to be located closer to each other than under normal bulk

15 condition. This is equivalent to a compression in x and y directions (c.f. figs.

4 and 5). When several layers of the X element is built up, the compression in the x-y plane is compensated by an expansion in the z direction, since the volume tends to be almost constant. Thus, if the A element has a cubic or tetragonal structure, the element X may in a thin layer adopt to a 0 tetragonally structure, where the lattice parameter a is adjusted to the lattice parameter a of the A element. The c/a ratio will in this case be larger than 1.

If X in bulk material has smaller lattice constants than A, the X atoms are forced to be located more distant to each other than under normal bulk 5 condition. This is equivalent to an expansion in x and y directions (c.f. figs. 4 and 5). When several layers of the X element is built up, the expansion in the x-y plane is compensated by a compression in the z direction, since the volume tends to be almost constant. Thus, if the A element has a cubic or tetragonal structure, the element X may in a thin layer adopt to a 0 tetragonally structure, where the lattice parameter a is adjusted to the lattice parameter a of the A element. The c/a ratio will in this case be smaller than 1. Such distorted crystal structures are for some combinations of substrate and overlayer elements possible to grow up to ten, twenty or even more atomic layers. Generally speaking, the larger the discrepancy between the bulk lattice parameters are, the fewer layers are possible to grow, before the overlayer adopts its bulk crystal structure. The substrate lattice parameters may be varied continuously, by using alloys of different elements, and vary the ratio between the alloying elements. Most alloy systems show continuously changing lattice parameters, at least in some composition intervals (Vegards law). In this manner it is possible to vary the c/a ratio of an overlayer structure.

The difficulty to grow thin films of heavily distorted structures implies a practical limit of the c/a intervals that are possible to use. Generally speaking, distortions of above 25 % are less likely to be producable by epitaxial growth methods of today, but since such methods are improving rapidly, distortions of 30% might be possible in a near future.

A preferred method of fabrication is illustrated by the flow diagram of fig. 10. The procedure starts in step 200. In step 202, a substrate layer of one of the elements V, Nb, Ta, Cu, Ag, Au, Pd, Pt, Rh, Ru Os, and Ir or an alloy containing one of these elements is provided. In step 204, a first ferromagnetic layer is epitaxially grown onto the substrate layer. In step 206,- an intermediate layer is epitaxially grown onto the first ferromagnetic layer. In step 208, a second ferromagnetic layer is epitaxially grown onto the intermediate layer. One of the ferromagnetic layers is formed by Fe, the other by Co. The order of deposition in not essential, as anyone skilled in the art understands. The epitaxial growth cycle is then in step 210 repeated until a suitable number of repetitive units have been formed on top of each other. The provision of the substrate layer comprises in the repetitions providing of a non-magnetic layer on top of the second ferromagnetic layer. The procedure is ended in step 212. Suitable materials for the intermediate layers are preferably selected in the following list: V, Nb, Ta, Cu, Ag, Au, Pd, Pt, Rh, Ru, Os and Ir or an alloy containing one of these elements. The different elements may give rise to different optimum combination of ferromagnetic layer thicknesses and intermediate layer thickness, since the different elements give rise to different distortions of the Fe and Co crystals. However, in all cases, Co will be magnetised out-of-plane and Fe will be magnetised in-plane.

It will be understood by those skilled in the art that various modifications and changes may be made to the present invention without departure from the scope thereof, which is defined by the appended claims.

Claims

1. A magnetoresistive sensor comprising at least one set (RU) of film layers, said set (RU) of film layers comprising: a substrate layer (NMl); a first ferromagnetic layer (FMl), formed on said substrate layer (NMl) and having a magnetisation direction parallel to said first ferromagnetic layer (FMl); an intermediate layer (NM; NM2), formed on said first ferromagnetic layer (FMl); and a second ferromagnetic layer (FM4), formed on said intermediate layer (NM; NM2), whereby an interlayer interaction is formed between said first ferromagnetic layer (FM l) and said second ferromagnetic layer (FM4), said magnetoresistive sensor is characterised in that said second ferromagnetic layer (FM4) is formed of a magnetic material of a tetragonally distorted crystal structure, having a magnetisation direction substantially perpendicular out-of-plane, in the absence of external magnetic fields applied thereto.

2. The magnetoresistive sensor according to claim 1, characterised in that said first and second ferromagnetic layers (FM 1 , FM4) are epitaxial layers grown on crystalline surfaces of said substrate layer (NMl) and said intermediate layer (NM; NM2), respectively.

3. The magnetoresistive sensor according to claim 1 or 2, characterised in that said substrate layer (NMl) and said intermediate layer (NM; NM2) is formed of one of the elements V, Nb, Ta, Cu, Ag, Au, Pd, Pt, Rh, Ru, Os, Ir or an alloy of at least one of said elements.

4. The magnetoresistive sensor according to claim 1, 2 or 3, characterised in that said magnetic material includes Co.

5. The magnetoresistive sensor according to any of the claims 1 to 4, characterised in that said first layer (FMl) is formed of iron of a tetragonally distorted crystal structure.

6. The magnetoresistive sensor according to claim 5, characterised in that the distortion of the crystal structure of said first ferromagnetic layer (FM l), expressed as the c/a ratio of the tetragonal lattice compared with body centred cubic Fe, is between 0.80 and 0.96.

7. The magnetoresistive sensor according to claim 5 or 6, characterised in that the distortion of the crystal structure of said second ferromagnetic layer (FM2), expressed as the c/a ratio of the tetragonal lattice compared with body centred cubic Co, is between 0.78 and 0.96.

8. The magnetoresistive sensor according to any of the preceding claims, characterised in that said intermediate layer (NM; NM2) is less than 7 atomic layers thick.

9. The magnetoresistive sensor according to any of the preceding claims, characterised in that said first and second ferromagnetic layers (FMl, FM4) are less than 6 atomic layers thick.

10. The magnetoresistive sensor according to any of the preceding claims, characterised in that the number of sets (RU) of film layers is larger than one, whereby the substrate layer (NMl) of one of said sets is arranged against the second ferromagnetic layer (FM4) in an adjacent one of said sets (RU), and in that said substrate layer (NMl) is less than 7 atomic layers thick.

11. A manufacturing method for a magnetoresistive sensor comprising the steps of: providing a crystalline substrate layer (NMl) of one of the elements V, Nb, Ta, Cu, Ag, Au, Pd, Pt, Rh, Ru, Os, Ir or an alloy of at least of said elements; growing a first ferromagnetic layer (FMl; FM4) epitaxially onto said substrate layer; growing an intermediate layer (NM; NM2) of the element of said substrate layer (NMl) epitaxially onto said first ferromagnetic layer (FM l ; FM4); growing a second ferromagnetic layer (FMl; FM4) epitaxially onto said intermediate layer (NM; NM2), forming an interlayer interaction between said first ferromagnetic layer (FMl) and said second ferromagnetic layer (FM4); whereby one of said first and second ferromagnetic layer (FM l) is formed by Fe and the other (FM4) by Co.