WO2005064357A2 - Flux guides for magnetic field sensors and memories - Google Patents

Abstract

A magnetic sensor has a magneto resistive sensing element (210) having layers of magnetic material and one or more flux guides (120) for concentrating the field onto the sensing element, the flux guide comprising a part of one or more of the same layers used for the sensing element. By using the same layer or layers for the flux guide the guide layer can be formed in the same step as the corresponding layer of the sensing element. Such sensors can be integrated in MRAM chips. Flux guides can be used to rotate the field differently for each of 2 parallel sensing elements, to enable 2D field sensors. As there is now no need for orthogonal exchange bias directions for the two sensors, they can be integrated more easily. Flux guides can also be used to concentrate the field for writing MRAM cells, and hence reduce write current.

Description

Flux guides for magnetic field sensors and memories

This invention relates to sensors for detecting magnetic fields, to magnetoresistive current sensors, to integrated circuits having such sensors, to methods of making such sensors, or such integrated circuits, to methods of using such sensors, and to MRAM (Magnetic Random Access Memory), amongst others. Magnetic field sensors have widely been used for electronic compasses, magnetic read heads, field measurement, metal detection, management of automobile engines and many more applications. Many different types of magnetic field sensors exist, among which magnetoresistive (MR) sensors are preferred in any applications that need compact sizes, such as in magnetic read heads or integrated circuits. There are several types of MR sensors based on different effects such as anisotropic magnetoresistance (AMR), Giant

Magnetoresistance (GMR), Tunnel Magnetoresistance (TMR). Sensors based on the AMR effect have been used in magnetic read heads for several years. They are still used nowadays in relatively small size magnetic sensors for many applications, due to their simple structure and low price. The AMR sensors have a layer of anisotropic magnetic material and the resistance of this layer is influenced by an external magnetic field, which can be sensed by sending a current through this layer. The GMR (Giant MagnetoResistive) sensor has a spacer layer of a conductive and nonmagnetic material being sandwiched in between a layer of magnetic material whose magnetization being fixed in a direction (pinned layer) and a layer of magnetic material of which the magnetic direction can be influenced by an external magnetic field (free layer). The change in the magnetic direction of the free layer causes the change in measured resistance. The GMR change nowadays can reach 15% or even higher. The third generation of MR sensor is based on the Tunnel Magnetoresistance effect (TMR) effect. In a TMR sensor, the conductive and nonmagnetic spacer layer as in the GMR sensor is replaced by a tunnel barrier layer which is made of a nonmagnetic and dielectric material. The MR effect is sensed by sending a current through the stack (in the perpendicular direction to the sensor plane). Nowadays the TMR effect can be achieved up to >50%. The TMR sensor is a promising candidate for the future high-density magnetic read-heads in hard disk drives. Depending on the type and construction, an MR sensor is more sensitive in one direction and less sensitive in another direction in the plane of the sensor. According to the Biot-Savart law there is a relationship between a current and a magnetic field. A current flowing in a straight conductor causes a circular field around itself. Therefore a magnetic sensor placed in the vicinity of the current can be used as a current sensor. Current sensors can find applications in many fields, such as in IC testing. It is required that currents at various locations within an IC chip must be tested and monitored. There are various types of known Built In Current Sensors (BICS) for ICs. Some are described in US 5,963,038, which shows detecting faults in integrated circuits by measurement of current through a conductor in the integrated circuit by means of a magnetic field sensor situated in the vicinity of the conductor. The sensor can be constructed in various ways so as to measure the field produced by the current through the conductor. Examples disclosed include a pick-up-coil sensor, a Hall sensor, an MR (magnetoresistive) sensor and a GMR (giant magnetoresistive) sensor. This can enable testing of conductors which cannot be accessed easily by external test equipment, or for detecting faults in individual ones of parallel paths which would pass a resistivity test even if only one path was conducting. An MR sensor has a resistance that is dependent on an external magnetic field through the plane of the sensor. The current through the conductor causes a circular magnetic field around the conductor through the plane of the sensor and perpendicular to the conductor in the plane of the sensor. The MR sensor is sensitive in this direction, so its resistance is measured along the plane of the sensor, parallel to the conductor to measure the strength of the magnetic field produced by current through the conductor. A traditional MR sensor or a GMR (giant magnetoresistive) sensor can be used. If using the same bias voltage, a GMR sensor is more sensitive than an MR sensor and can more reliably measure the current through the conductor with less power consumption. The resistance of the MR sensor can then be measured inside in the integrated circuit with the detection circuit or outside the integrated circuit with a suitable measurement arrangement. Such sensors are useful for sensing high currents, but are not sufficiently sensitive for applications such as Quiescent IDD current (IDDQ) testing. This testing technique has shown very good coverage of physical defects such as gate oxide shorts, floating gates, and bridging faults which are not very well modeled by classical fault models, or undetectable by conventional logic tests. The demand for high quality and cost effectiveness has led to widespread use of IDDQ testing as a supplementary test to voltage tests. When combined with other test techniques, it has the potential for eliminating the need for burn-in test. However MOSFET leakage currents are rising rapidly with each technology node, narrowing the difference between the IDDQ levels of a faulty and fault-free circuit. One method of IDDQ testing involving switching the power supply off and sampling a decay in voltage at a monitoring point as shown in European patent application EP 0 840 227. Another technique shown in WO 97/18481 involves providing an active output load that can be turned on to test the device. The resulting current can be detected off the chip. Walker et al, "A Practical Built-in Current Sensor for IDDQ Testing",

ITC2001, paper 14.3 describes some recent work on IDDQ testing. It summarizes the requirements for a BICS as follows: < 2 μA resolution, < 1000 transistors in size, - < 1 ms/vector test time, Does not cause performance degradation, Measures current level and direction, Control and digital readout via scan chain, Parallel operation, - Self-calibration, Low power dissipation in use, none when idle. To try to meet these, Walker uses MagFET devices, which are based on the Hall effect. To overcome the difficulty of low sensitivity and high noise levels, filtering, amplification and test repetition are used. However, these techniques have the disadvantages of taking up space and power, which are scarce resources for many applications such as mobile devices. For magnetic sensor applications, such as magnetic read-head for hard disk, it is known to use flux guides (or flux concentrators) to concentrate more magnetic flux into the sensing layer of the sensor, thus making the sensor more sensitive. An example of a sensor with flux guides 120 is shown in Fig. 2. Thanks to the flux guides, the magnetic flux is more concentrated at the sensing layer of the sensing element 210. The flux guide is usually made of a thin film of a high permeability magnetic material, such as permalloy, which is patterned into a suitable shape. The flux guide layer is typically 5 to a few tens of n thick and isolated from the sensor with a distance of about a few tens of nm. MRAM development: Another application of magneto resistive materials is MRAM. During recent years, research into Magnetic RAM (MRAM) has been intense. The integration of magnetic materials with CMOS technology has become less of a problem. Commercial MRAM production is planned in 2004- 2005. A survey of MRAM technology is presented by K..- M.H. Lenssen et al, "Expectations of MRAM in comparison.", Non- Volatile Memory Technology Symposium 2000, (Nov. 15-16,2000, Arlington VA, U.S.A.). This shows that first generation Magnetic Random Access Memory (MRAM) were based on AMR. After 1988 the discovery of a larger magnetoresistance effect called giant magnetoresistance (GMR), enabled the realization of smaller elements with a higher resistance and a larger MR effect (5 to 15%), and therefore a higher output signal. This enabled, in principle, MRAMs for general applications. A decade after its discovery the GMR effect is already applied in commercial products like HDD read heads and magnetic sensors. A breakthrough in the field of magnetic tunnel junctions around 1995 improved the prospects of MRAM even further, when a large tunnel magnetoresistance (TMR) effect was demonstrated at room temperature. Since then TMR effects with amplitudes up to >50% have been shown, but because of the strong bias-voltage dependence, the useable resistance change in practical applications is at present around 35%. MTJ memory cell: An example of part of a known MRAM showing an array of integrated memory cells using such TMR technology is shown in perspective view in Fig. 1. This structure and how to manufacture it is well known and need not be described again in detail here. To summarize, such a TMR-based MRAM contains cells which are magnetic tunnel junctions (MTJs). MTJs basically contain a free magnetic layer 100, an insulating layer (tunnel barrier 102), a pinned magnetic layer 104, and an antiferromagnetic AF layer 106 which is used to "pin" the magnetization of the pinned layer to a fixed direction, which is usually called the exchange bias direction. In the example shown in this Figure, there is also an underlayer 108. For simplicity, only 4 active layers are shown in the magnetic tunnel junction (MTJ) stack shown in Fig. 1. In practice there may be more layers, which are not relevant to the principle of operation. The MRAM cells store information (1/0) in the directions of magnetization of the free magnetic layer, which can be relatively free to rotate between two opposite directions. The resistance of the MTJ is small if the magnetization directions of the free layer is parallel with that of the pinned layer and is large when they are antiparallel. For reading information on a certain cell, a small current is sent through the MTJ stack (vertically) of the selected cell. The measured voltage drop on the MTJ (proportional to the resistance) is the indication of the information of the cell. The information on a cell can be changed during a write operation by sending write currents through digit lines DL1-3 and bit lines BL1-3, which are patterned at the bottom and on top of the memory cells. The currents will create magnetic fields (easy axis field and hard axis field) in the memory cell. The fields are programmed so that they are large enough to switch the magnetization of the free layer of the selected cell to a new direction. The bit lines are parallel with the hard axis of the cells, which creates a field in the easy axis, while the digit lines (sometimes called word lines) otherwise create a field in the hard axis. In some designs the relations can be reversed, i.e. the bit lines create hard axis field and the digit lines create easy axis field. Bit lines and digit lines are shown as being perpendicular, with magnetic tunnel junctions are placed at the intersections. The inset shows hard axis field (created by digit line) and easy axis field (created by bit line). The resultant field is directed at 45° with respect to the easy axis, which is able to rotate the magnetization of the free layer of the selected cell, while all unselected cells are not affected. The bottom electrodes of the cells are connected to the selection transistors with vias, which are used when reading. As the resultant field makes an angle of 45° with respect to the easy axis of the free layer of the cell, the switching field of the free layer is the smallest, thus writing can be done with the least current. The magnitude of resultant magnetic field at the crossing point is (|H_HA|+|H_EA|)/^2, in which H_HA and H_EA are the fields created in the hard axis and easy axis, respectively. They must generally have the same magnitude. For more information on such MRAMs, the reader is referred to for example P. K. Naji, M. Durlam, S Tehrani, J. Calder and M. F. DeHerrera, "A 256kb 3.0V 1 TIMTJ nonvolatile magnetoresistive RAM", IEEE Int. Sol id-State Circuits Conference 2001, section 7.6. and R. Scheuerlein, W. Gallagher, S. Parkin, A. Lee, S. Ray, R. Robertazzi, W. Reohr, "A 1 Ons Read and Write nonvolatile memory array using a magnetic tunnel junction and FET switch in each cell", IEEE Int. Solid-State Circuits Conference 2000, section TA 7.2. In general, both GMR and TMR result in a low resistance if the magnetization directions in the multilayer are parallel, and in a high resistance when the orientations of the magnetization are antiparallel. In TMR devices the sense current has to be applied perpendicular to the layer planes (CPP, Current Perpendicular to Plane) because the electrons have to tunnel through the barrier layer. In GMR devices the sense current usually flows in the plane of the layers (CIP, Current In Plane). Nevertheless, supported by the rapidly continuing miniaturization, the possibility of commercially available MRAMs using CPP TMR seems more likely. It is also known to provide dummy cells in an array of magnetic memory cells. An example is shown in US patent 6466475 in which dummy cells are provided around the edge of the array to avoid edge effects. Dummy cells are also known from US patent application 2002/0080646 for increasing MRAM read speed. An issue for MRAM is that it needs rather large current pulses to switch a bit, thus writing on MRAM is a large power consuming activity. HHA and HHE are in the order of a few tens of Oe (a few kA/m). To create such a field, a current of around 5 - 10 A must be sent to both lines. Reducing power consumption in MRAM is one of the important keys to the success of MRAM, especially for the mobile applications. Another issue is that in commercial GMR sensor products now, the only way to create a set of two orthogonal sensors for 2-dimensional field detection is to combine two separate orthogonal chips on a circuit board, or two separate orthogonal dies in the same chip package. Another issue is that in GMR or TMR sensors nowadays such as in magnetic read heads, flux-guides are created separately from the sensor elements, thus costs extra processing steps. According to a first aspect, the invention provides a magnetic sensor for sensing a magnetic field and having a magneto resistive sensing element having layers of magnetic material and one or more flux guides for concentrating the field onto the sensing element, the flux guide comprising a part of one or more of the same layers used for the sensing element. An advantage of using the same layer or layers for the flux guide is that the number of processing steps can be kept to a minimum. The guide layer can be formed in the same step as the corresponding layer of the sensing element. Even if the guide and the element are formed in separate steps, there is an advantage in using the same layer for both. Another advantage is that such magnetic field sensors can be integrated in MRAM chips, using exactly the same technology for MRAM. This means they can be added at little or no extra cost. Some embodiments use two dummy devices, which are placed next to a sensor, as flux guides for the sensor. The flux guides can be included without any extra processing steps during manufacture. Simulations show that in the case of external field sensor, the flux density inside the sensing layer of the sensor with flux guides can be increased considerably, e.g. by 53%. In the case of current sensor, the improvement has been found to be about 10%. Another advantage of using flux guides having the same layer or layers as the sensing element is that the demagnetization field (i.e. a shape anisotropy) of the pinned layer can be reduced. As an additional feature of some embodiments, the sensing element comprises a tunneling magnetic junction. Another such feature is the guide having a larger area than the sensing element. Another such feature is having guides on opposing sides of the sensing element. Another such feature is the guides being sized to be not fully saturated in the range of magnetic fields to be measured. Another such feature is the sensor having an elongate shape. This enables exploitation of anisotropy to tailor magnetic characteristics. Additional features which can form dependent claims include the sensor element being planar and the tunnel current being detected perpendicular to the plane of the planar element. Another such feature is the element being arranged to have a relationship between resistance and field, which shows substantially no hysteresis. Another such feature is that the sensor element has a free magnetic layer which has an easy axis oriented to be substantially perpendicular to the field being measured. For a magnetic element such as a free layer of an MTJ the total effective anisotropy has two sources: one is the intrinsic anisotropy of the material caused by the crystal structure, often called crystalline anisotropy, another is the shape anisotropy caused by the elongate shape. The former is usually smaller than the latter. If these two anisotropies are aligned, the total anisotropy is the sum of the two. If they are orthogonal, the effective anisotropy will follow the stronger one and the strength is simply the subtraction of the two. Even if the two anisotopies are orthogonal but the shape anisotropy is stronger, the effective anisotropy (easy axis) will still be in the direction of elongation. This is used in the case sensors have to integrated with MRAM. The crystalline anisotropy is the same for the whole chip but by patterning the shape of the sensors orthogonal to that of the MRAM cells different anisotropy directions for the MRAM cells and the sensors can be obtained. Hence, exploitation of the anisotropic properties of some materials provides an improved resistance characteristic, such as maximizing the sensitivity without introducing too much hysteresis. Another such feature is that the junction comprises a pinned magnetic layer having a magnetization oriented perpendicular to the easy axis of the free magnetic layer. Clearly the output can be anything from a logical signal indicating field detected or not, to an analog or digital signal indicating a measurement to a given level of precision. Suitable post processing of the detected output can be carried out to suit the precision or noise immunity of the application for example. Another aspect of the invention provides a current sensor for sensing a current in a conductor, comprising the above mentioned field sensor located to detect a field caused by the current. This is a particularly useful application of a field sensor. Another aspect of the invention provides a sensor for sensing a magnetic field, having a sensing element and a flux guide for concentrating the magnetic field on the sensing element, the sensing element being a magneto resistive sensing element such that an axis of elongation of the sensing element is substantially orthogonal to the magnetic field being sensed. This can provide a particularly sensitive sensor. Another aspect of the invention provides an integrated circuit having a built in current sensor comprising the above current sensor. As an additional feature, the current sensor is arranged to sense quiescent current (IDDQ). As another additional feature, the integrated circuit has multiple current sensors linked in a scan chain. A second aspect of the invention provides a magnetic memory having at least one memory cell, at least one of the cells having at least one flux guide for concentrating a magnetic field used for writing the cell. The use of a-flux guides can help reduce the writing current needed to write the memory cell. This is particularly important for battery powered or other mobile devices. The free layer of dummy MTJs that are placed next to a memory element can serve as flux-guides. Thanks to such flux-guides, in some embodiments, the write current can be reduced by about 10% and the total write power consumption can be reduced by 10-20%, depending on the implementation and whether only one or both fields (easy and hard axis) are equipped with the flux-guides. A feature of some embodiments is that the flux guides are co-planar with a free magnetic layer of the cell. This can provide particularly efficient concentration of the field. A feature of some embodiments is that the flux-guides are in the form of dummy MTJs, having all the layers of their associated memory cell. This can ensure the guides are co planar without requiring any extra process steps, and thus can be implemented at little or no extra cost. As an additional feature of some embodiments, the guide comprises a part of the same layer or layers used by the memory cell. This enables the flux guides to be formed without an extra processing step, which can help reduce costs and maintain reliability. As another such feature, the cell is a tunneling junction. As another such feature, the flux guide overlaps a via between layers. This helps to keep the arrangement compact, to maximize the number of cells that can be integrated. As another such feature, flux guides are formed on four sides of the cell. As another such feature, the flux guide comprises a dummy cell. Another aspect of the invention provides a magnetic sensor having a first sensing element for measuring a magnetic field in a first direction, and a second sensing element for measuring a magnetic field in a different direction, and a flux guide for at least one of the sensing elements, the flux guide being arranged to rotate the direction of the field through that sensing element. The rotation helps enable the sensor to determine a vector or direction of the field. It can make it simpler to manufacture, if it avoids the need for the sensors to have different orientations. It can enable an integrated 2D magnetic field sensor within an MRAM chip to be achieved. In some embodiments, the sensor has different flux guides having different geometries. The flux guides act as diverters to rotate the sensing direction of the sensing elements in two different directions. The combined signals from the sensors can give information about the field vector being measured. An advantage of this is that it is no longer necessary to have orthogonal exchange bias directions for the two sensing elements. If the flux guides are dummy devices, this can avoid the need for separate steps for patterning the flux guides, thus the sensor can be integrated more easily. In particular it can be integrated within an MRAM chip at little or no extra cost. The guides can be parallelogram-shaped. As an additional feature, flux guides are provided on both the first and the second sensing elements. As another such additional feature, the first and second sensors are arranged to detect orthogonal components of the field. This can make it easier to determine the direction of the field. As another such feature, the sensing elements are oriented so that they have the same exchange bias directions. This can be easier to manufacture, and helps enable them to be integrated. As another such feature, the sensing elements are integrated on the same chip. This helps enable a reduction in space and in manufacturing costs. As another such additional feature, flux guides are provided on both sides of the sensing elements, the flux guides for either one of the sensing elements being parallel. This enables a more efficient rotation of the flux. Another such additional feature is circuitry for determining a direction and magnitude of the field, from outputs of the first and second sensing elements. As another such feature, the circuitry is arranged to take into account an amount of background field not rotated or amplified by the flux guides. This can enable a more accurate calculation. As another additional such feature, the flux guides are arranged to couple the sensing elements predominantly by magnetostatic interaction rather than exchange interaction. This can increase the efficiency of the rotation of the field. The magnetic tunnel junction has been developed previously for memory applications, and the inventors have appreciated that it could be adapted for use as a sensor, despite the fact that memory cells and current sensors must have different characteristics. In a memory cell, a magnetoresistance loop (MR loop) of a free layer should be square with a relatively large coercivity (in the order of a few tens of Oe) and having two distinct remanence states. (Remanence refers to remanent magnetization after the applied field is reduced to 0 after the sample has reached saturation. Moreover, the center of the loop must be at zero field. In contrast, a current sensor must have on the one hand as large susceptibility to magnetic field as possible (for high sensitivity) and one the other hand must have small or no hysteresis. Another aspect provides corresponding methods of sensing magnetic field or of sensing current, or of writing an MRAM. Another aspect provides a method of manufacturing a sensor comprising the step of forming a layer of a sensing element and a layer of a flux guide for the sensing element, in the same step. Another aspect provides a method of manufacturing an MRAM having features mentioned above. The features of any of the dependent claims can be combined with each other or with any of the independent claims. Further advantages will be apparent to those skilled in the art, especially over other prior art not known to the inventors. How the present invention may be put into effect will now be described with reference to the appended schematic drawings. Obviously, numerous variations and modifications can be made without departing from the claims of the present invention. Therefore, it should be clearly understood that the form of the present invention is illustrative only and is not intended to limit the scope of the present invention.

The features of the invention will be better understood by reference to the accompanying drawings, which illustrate preferred embodiments of the invention. In the drawings: Fig. 1 shows a prior art MRAM design, Fig. 2 shows a prior art sensor with flux guides, Figs. 3 and 4 shows views of a first embodiment of the present invention, Fig. 5 shows a flux density profile of an embodiment, Figs. 6 and 7 show images of magnetic field lines around sensors, Figs. 8 and 9 show field lines around conductors and sensors, Fig. 10 shows a flux density graph of another embodiment, Figs. 1 1 and 12 show views of a prior art MRAM, Figs. 13 and 14 show views of an MRAM according to another embodiment of the present invention, Fig. 15 shows a view of another embodiment of the present, Figs. 16 to 18 show views of a TMJ sensor, Figs. 19 and 20a and b show views of sensors according to another embodiment of the present invention, for determining a field direction, and Fig. 21 shows flux density from first and second sensing elements, according to another embodiment of the present invention.

The present invention will be described with respect to particular embodiments and with reference to certain drawings but the invention is not limited thereto but only by the claims. The drawings described are only schematic and are non-limiting. In the drawings, the size of some of the elements may be exaggerated and not drawn on scale for illustrative purposes. Where the term "comprising" is used in the present description and claims, it does not exclude other elements or steps. Where an indefinite or definite article is used when referring to a singular noun e.g. "a" or "an", "the", this includes a plural of that noun unless something else is specifically stated. Furthermore, the terms first, second, third and the like in the description and in the claims, are used for distinguishing between similar elements and not necessarily for describing a sequential or chronological order. It is to be understood that the terms so used are interchangeable under appropriate circumstances and that the embodiments of the invention described herein are capable of operation in other sequences than described or illustrated herein. Moreover, the terms top, bottom, over, under and the like in the description and the claims are used for descriptive purposes and not necessarily for describing relative positions. It is to be understood that the terms so used are interchangeable under appropriate circumstances and that the embodiments of the invention described herein are capable of operation in other orientations than described or illustrated herein. Figs. 3 and 4, a first embodiment of the invention: The inventors have appreciated that MTJ devices can be adapted for use as magnetic field sensors. These have many applications, for example in contact-less measurements of currents, e.g. power current or IDD current. This can be in any kind of CMOS chip or other chip. Of course it can be implemented in MRAM chips. The same manufacturing technology used for MRAM cells can be used with little change, to build integrated current sensors. They are particularly suitable for power pin testing and IDDx testing in MRAM chips or chips containing embedded MRAM, since implementation of these sensors does not need to cost any extra mask or extra process steps. If the sensors are to reside inside the MRAM chip, they must use exactly the same technology as MRAM without any extra processing steps. The MRAM/sensor stack will be deposited and then patterned at once in between two metallization levels in the back- end of the process. This can make the sensor design more stringent and less flexible, though it will often be preferred because of low cost. For instance, making flux guides in the conventional way would cost several extra processing steps such as isolation, deposition, lithography and etching therefore it is not desired and in many cases not allowed. Embodiments of the invention can provide a simple way to build flux-guides for the magnetic field sensors, which are integrated in the MRAM chip and can share exactly the same MRAM process. As shown in Figs. 3 and 4, the free layer of two dummy MTJ elements, arranged in the neighborhood of the real sensor element, can be used as flux -guides to concentrate more flux into the sensor, thus making the detection limit lower. Briefly speaking, it has been found by simulations that using the dummy devices as flux-guides, flux density in the case of the field sensor can be increased by 53%. For the current sensor, which uses the same way to generate field as in MRAM, the improvement can be about 10%. Fig. 3 shows a side view of the sensing element 210, and dummy devices for use as flux guides 120. The layers correspond to those of the memory cell shown in Fig. 1. There is a sensing layer (free magnetic layer 200), an insulating layer (tunnel barrier 202), a pinnedjnagnetic layer 204, and an antiferromagnetic AF layer 206 which is used to "pin" the magnetization of the pinned layer to a fixed direction (exchange-bias direction). In the example shown in this Figure, there is also an underlayer 208 acting as the bottom contact. For simplicity, only 4 active layers are shown in the magnetic tunnel junction (MTJ) stack shown in Fig. 3. In practice there may be more layers, which are not relevant to the principle of operation. The sensing layer of the two neighboring dummy MTJ devices can act as flux guides for the working (real) sensor, which is placed in between them. This design allows the flux guides for the field sensor to be created without any extra processing steps, which suits applications such as integration into MRAM chips. The free layers of the two dummy devices now act as flux guides because their magnetization is free to rotate. The pinned layers of the dummy devices will not influence the flux change during the measurement because their magnetization is fixed. Fig. 4 shows a top view. It is preferred that the two dummy devices (acting as flux guides) are larger than the sensor for better efficiency and homogeneous field produced in between them. The permeability of the flux guides can be tuned by varying their geometry. The aspect ratio and size of the dummy devices should be chosen such that they are not fully saturated in the field range to be measured but still have sufficiently large permeability for concentrating the flux. The flux guides should be patterned as close as possible to the sensor, but still isolated from it. The closer they are, the more efficient are the flux guides. Of course the minimum spacing depends on the lithography resolution and the etching technique. In spite of the small spacing between the flux guides and the sensor, the location of a top contact on top of the sensor is not critical. In fact, it can have some tolerance. Even if the top contact of the sensor makes contact with one of the flux guides, this would not normally cause any problem because the flux guides are not connected at the bottom. If used as a built in or integrated current sensor using MRAM technology, the current conductor can be patterned in the same lower metallization layer as is used for e.g. word or bit lines of MRAM. The conductor is galvanically isolated from the MTJ device. When a current to be measured is sent through the line, it creates a magnetic field around itself, which will be sensed by the MTJ device. The resistance change of the MTJ device is the indication of the field and thus of the current. Figs. 5 to 10, simulation results: Simulation results using the finite- element technique are shown in these Figures. The simulations were performed for two cases: magnetic field sensor and current sensor, both cases are calculated with and without the flux guides. Fig. 5 shows the X component of flux density (magnetic induction B) across the free layer of a magnetic sensor for two cases: with and without the flux guides. Here the X direction is referred to the direction parallel with the sensor plane. It can be clearly seen that with the flux guides, the flux density in the sensor is drastically increased by 53%. In the simulation, only the free layers of the sensor and the flux guides are taken into account for the sake of simplicity. The permeability of the sensing layer (free layer) of the sensor and the flux guides are supposed to be about 200 and 1000 respectively. These are practical values and depend on the geometry of the flux guides and the sensor. The sensor width is lμm, the flux guide width is 5 μm and the spacings between them are lOOnm. The free layer thickness is 5nm, which is normally used in MRAM stack. A constant and homogeneous field of 5 Oe is applied in the direction of the free layer plane. Fig. 5 (revised): Fig. 6 shows the calculated images of the field lines for the prior art case of a sensing element 210 without flux guides. Fig. 7 shows a corresponding image for the case according to an embodiment of the invention with flux guides 120. The same density scale is used in both images for comparison. In the case of sensor with flux guides, it is obvious that due to the flux guides, the field lines are more concentrated on the sensing layer of the sensor. In Fig. 7, only a part of the flux guides are shown for the sake of conciseness, they actually extend further in both directions. Figs. 8 and 9 show corresponding images of field lines for a current sensor, with and without flux guides respectively. The same geometry and parameters are applied to the current sensor case, in which, instead of applying a homogeneous field to the sensor, a conductor 130 having a cross section of 300nm (height) x 1500nm (lateral) is placed beneath the sensor at a distance of 150nm. A current of 1.5mA is sent through the conductor (in the direction perpendicular to the drawing), which creates around itself a circular field. At the sensor location, the created magnetic field is about 50e, which is compatible to the cases in Fig. 6 and Fig. 7. Again, in Fig. 9 only a part of the flux guides are shown for the sake of conciseness, they actually extend further in both directions. From the images, it can be seen that the improvement is not as clear as that shown in Figs. 6 and 7. This is because only relatively small parts of the flux guides near the sensor, where the flux lines are closer to the horizontal direction, are actually contributing to the flux guide effect. Detailed simulation data show that the flux density inside the sensor with flux guides is increased by around 10% compared to the case without flux guides, as shown in the graph of Fig. 10 of flux density across the sensor width at different distances from the sensor center. Furthermore, the flux guides can be used in combination with the cladding of the current line (as shown in WO 02/41367 A2), which is often used for MRAM bit/digit lines. The effect of the flux guides in this case is some 10% increase over the improved field created by the conductor line with a cladding layer. Other effects: In a sensor without flux guides, the elongated shape of the sensor induces a shape anisotropy of: K = 1/2 μo Ms² (Nχ-Nγ), in which Ms is the saturation magnetization of the layer, and Nx and Ny are demagnetization factors of the transverse and longitudinal directions, respectively. This shape anisotropy tends to align the magnetization of any magnetic layer of the sensor to the longitudinal direction. The larger the aspect ratio of the element, the stronger the shape anisotropy becomes. For sensors having a width of around 1 micron, an elongated shape with aspect ratio (AR) of 5 to 7 is desired for stabilizing the free layer and thus reducing hysteresis. However, for the pinned layer, whose magnetization is by purpose pinned in the transverse direction, the shape anisotropy will tends to force the magnetization away from the pinned direction, which may result in bad behavior of the sensor. By placing two larger dummy devices close to the two edges of the sensor, the demagnetizing field of the pinned layer of the sensor is reduced thus reducing the shape anisotropy. This is because the pinned layers of the two dummy devices are also pinned in the same direction as that of the sensor, thus makes its pinned direction less unfavorable. Similarly to the effect mentioned in the previous section, the free layer (sensing layer) would see the same reduction in the shape anisotropy. The effect, however, is smaller because the saturation magnetization of the free layer (usually made of permalloy) is smaller than that of the pinned layer (usually made of CoFe). The free layer in this case behaves as if it has a larger effective width, thus smaller AR. Nevertheless, some shape anisotropy of the free layer is useful to suppress hysteresis but on the other hand, the shape anisotropy tends to degrade the sensitivity of the sensor. Therefore a compromise value should be found for each particular application. During design of the geometry of the sensor with flux guides, the reduction in the effective shape anisotropy of the free layer must be taken into account to obtain the desired value of the shape anisotropy. Although the embodiments described refer to magnetic sensors using TMR effect, the idea is however not restricted thereto, but can also be applied to GMR sensors for example. Figs. 11 to 15, flux guides applied to enhance writing of MRAM cells: The same flux guide principles discussed above can be applied also to enhance MRAM, by increasing the flux density on cell created by the write current. This can save up to 20% of power consumption during writing. The implementation of the flux-guides need not cost any extra process steps, only the design of the mask has to be changed. Furthermore, if the flux-guides can be fitted into unused space in between the MTJ memory cells 110, then there is no need to enlarge the cell size. Figs. 11 and 12 show top and side views of a prior art MRAM without flux guides, corresponding to the design of Fig. 1. In the word or digit line direction, the cells can be arranged as close as possible, as long as each cell is large enough to house the isolation transistor on the silicon substrate and there is still no significant magnetic coupling between the cells. Placing flux-guides along this direction would normally enlarge the cell size. However, along the bit line direction, the spacings between the cells are significantly wider since some space is needed for the bottom electrode vias 150. This unused space can be used for placing dummy MTJs as flux-guides. Figs. 13 and 14 show the top and side views of a first embodiment of an MRAM according to the invention. The design corresponds to that of Fig. 1, and corresponding reference numerals have been used as appropriate. Every MTJ memory element (denoted as MTJ in the Figure) is sandwiched between two dummy MTJ s which serve as flux-guides (FG) 120. The dummies are arranged in the bit line direction. The dummies and the memory elements can be patterned at the same time, in exactly the same way. The two flux-guides lying partly on top of the same via can share the same bottom electrode as the memory element. However the connection in the bottom electrodes must always be broken between the cells. The bit line is connected to top electrode of the memory element through the top electrode via 160, which is placed exactly on top of the memory element. Therefore the dummies are open-circuit and do not influence the memory element electrically. The spacings between the dummies and the memory element should be as small as possible, e.g. around lOOnm or smaller depending on the technology capability. The dummies should only occupy the unused space in between the elements. Using the dummies as flux-guides, the write current sent to a digit line (which generates hard-axis field in this example) can be reduced by about 10% to obtain the same field as in the case without flux-guides. In the other words, the power consumption for the hard axis field, being proportional to the current squared, can be reduced by almost 20%. If total power consumption for both fields (hard and easy axis fields) is considered, the total write power consumption can be reduced by about 10% (assuming that the bit line current is equal to the digit line current). The flux-guides (FG), in the form of dummy MTJs, are patterned in between the memory elements in the bit line directions. If somehow a larger space between the memory elements in the digit lines direction is required, because for instance, a larger area is needed for the isolation transistor, this unused space can be used by putting flux-guides in both directions. In Fig. 15, a second embodiment is shown where the dummy cells are placed in both directions i.e. hard axis and easy axis field directions. Otherwise the design corresponds to that of Fig. 13 and 14, and corresponding reference numerals have been used as appropriate. In this case, the currents in both the bit line and digit line can be reduced by 10%, and the write power consumption for both fields can be reduced by 20%. Figs. 16 to 21, flux guides for 2-D sensors: By way of introduction to such sensors, development of TMR sensors will be described briefly. Integration of MRAM into CMOS technology has created a new opportunity for including magnetic field sensors within MRAM chips because MRAM and magnetic sensor are based on the same effect, the Tunnel Magnetoresistance (TMR) effect. The magnetic sensors will use exactly the same technology as MRAM, only with different design of geometry; therefore the integration of magnetic sensors within the MRAM chip can add extra value to the MRAM chip at no extra cost. Applications of integrated magnetic sensors may include: highly sensitive current sensors for power-pin testing or IDDx testing, external field sensors for write control of MRAM, active shielding for MRAM, integrated compasses, and so on. As mentioned previously, magnetoresistive sensors can be based on one of the magnetoresistance effects, such as the Giant Magnetoresistance (GMR) or Tunnel Magnetoresistance (TMR). A TMR sensor contains a sensor element which is a Magnetic Tunnel Junction (MTJ, as shown in Figs. 4 and 5 described above). Figs. 16 to 19 show schematic drawings of a magnetic field sensor based on the TMR effect. Corresponding reference numerals have been used where appropriate. The magnetization of the free layer 200 is relatively free to rotate depending on the applied field whereas that of the pinned layer always points to a fixed direction, in this example, the x direction. The free layer has an easy axis in the y direction, commonly by patterning the element elongated in the y direction. In the absence of an applied field, the magnetization of the free layer (Mfree) rests in the direction of the easy axis. When a magnetic field is present in the x direction, Mfree rotates out of the y direction, towards the direction of the field (x direction). The stronger the field, the smaller is the angle θ (i.e. the angle between M and x direction). The resistance of the MTJ (measured by sending a sensing current perpendicular to the stack from the top to the bottom electrodes of the MTJ) is changed according to the following transfer function: R(θ)/R(0) = 1 + MR*(1 -cosθ)/2 in which R(θ) and R(0) are the resistances of the MTJ when θ is not zero and is zero, respectively; MR is the magnetoresistance ratio, typically about 30-40%. A TMR sensor is normally designed to operate properly when the field to be measured is directed perpendicular to the easy axis of the free layer, i.e. in the direction of the pinned magnetization. This direction is called the sensing direction of the sensor. Except for current sensors, many other applications need 2D field detection capability, i.e. the sensor system must provide information about the projection of the magnetic field vector onto the plane of the chip. This field projection, hereinafter called planar field H, can be decomposed into Hx and Hy, each being detected by a separate sensor. For detecting a 2D field vector, two orthogonal sensors are required. Each sensor is sensitive to only one component (Hx or Hy) of the field and the combined signals from both of them provide information about the direction and magnitude of the planar field. In GMR and TMR sensors, the sensing direction is decisively determined by the direction of the magnetization of the pinned layer. As mentioned above, this layer is pinned by the antiferromagnetic layer underneath. This pinning effect is called the exchange bias. The exchange bias direction is set during processing by annealing the stack to a few hundreds of degrees C then cooling down in the presence of a magnetic field. After cooling down and removing the field, the exchange bias direction remains in the previously applied field direction. This step can be done before or after patterning the sensor elements. Because the sensors are patterned from the same multilayer stack, they normally have the same exchange bias direction. In commercial GMR sensor products nowadays, the only way to create a set of two orthogonal sensors for 2D field detection is to combine two separate orthogonal chips on a PCB or two separate orthogonal dies in the same chip package. In order to integrate 2 orthogonal sensors onto the same substrate, local annealing of separate sensors is required. This is very difficult or impossible in practice. Especially if the sensors are integrated with MRAM, this process step is currently not available because for MRAM, only one exchange bias direction is possible. For a GMR sensor, one way to locally change the exchange bias direction, is by sending high current pulses through some selected sensor strips in the presence of a field oriented in a desired direction, reference K-M. Lenssen, "Magnetoresistive sensors and memory", NATO ASI article, December 2001. The heat generated by the current will reset the exchange bias direction of the selected sensors to the new direction, while the rest is still intact (remains in the previously set direction). The method is initially used to set opposite exchange bias directions for different sensors in a Wheatstone bridge. This method is only possible for non- integrated sensors and only for GMR sensors having no barrier layer. For TMR sensors, sending such a high current with consequently a high voltage is not feasible because the device can break down easily when exposed to a voltage of > about 1 V. A simple way to produce a pair (or pairs) of integrated TMR sensors for 2D field measurements such as 2D external field measurements or for an electronic compass will now be described. Flux guides using dummy devices, patterned under different angles, are used to decompose the field vector into 2 components while the exchange bias directions of the sensors are still parallel. That means the device does not require locally different exchange bias directions, and so no extra processing steps are needed beyond those used for manufacturing an MRAM. The method is therefore very suitable for integration of the sensors within an MRAM chip. The same flux guiding principle describe above in relation to Figs. 1 to 15 is used. However, in this case, the flux guides are used not (only) for increasing the flux density but more for decomposing the planar field into two rotated components, each being detected by a separate sensor. These sensors have the same exchange bias direction but they are coupled to flux guides having different orientations. The principle of the method is shown in Fig. 19 which shows a schematic top view of 2D magnetic sensors A and B and associated flux guides. Sensor A, 310 and sensor B, 320 have the same exchange bias direction in the x direction and the easy axis in the y direction, thus they are designed to sense best in the x direction. Each sensor has 2 parallelogram-shaped flux guides 300 on its left and right sides. The two sensor-flux guide groups are separated from each other at some sufficient distance to avoid mutual interaction. The flux guides are dummy MTJ devices so they can be patterned in the same way as the working sensors. Each flux guide of sensor A has 2 sides that are parallel with the longitudinal direction of the sensor, and the other 2 sides are in the y' direction, which forms an angle α with x. In this example, α is 45°, but it can be a different value. Each flux guide of sensor B has 2 sides that are parallel with the longitudinal direction of the sensor, and the other 2 sides are in the x' direction, which forms an angle β with x. Generally β= -α, i.e. the flux guides of the two sensors are mirrored over the x direction. Mirroring of the flux guides is necessary to obtain equivalent and symmetrical signals from the sensors. To prevent the flux-guides from being saturated at zero field, in which the magnetic moments are preferred to be oriented in one direction, the parallelogram-shaped flux guides must have small anisotropy. An important source of anisotropy in the free layer is the shape anisotropy, caused by the elongated shape of the layer. Therefore the shape of the flux guides must not be (clearly) elongated in any direction. Moreover, the moments should split into domains at zero field. This can be done by making the guides large enough (at least several microns in each side). When an applied planar field H is present, the flux guides concentrate flux along their longitudinal direction and create magnetic charges at two sides of the sensor (see example sensor A, Fig. 19). The charges consequently produce flux lines in the x direction, which coincides with the sensing direction of the sensor. Because the flux lines prefer the travel along the tilted sides of the flux guides, the flux is strongest when the field is parallel to this direction, in this example, this is the y' direction for sensor A. Apparently, the flux guides act as a flux diverter to rotate the sensing direction of the sensor, from x to y' direction for sensor A. Similarly, sensor B is most sensitive to a field directed in the x' direction. When both sensors are in use, a planar field H seems to be decomposed into two orthogonal directions, namely x' and y\ The signals from the sensors are unambiguously recorded by an electronic circuit, which then combines and gives information about the field vector, such as direction and magnitude. The behaviors of the sensors would look as if there were two orthogonal sensors. The phenomenon above is the ideal case, in which it has been assumed that the field created by the flux guides is much stronger than the background field. In practice, the situation is somewhat different. The sensor element in fact sees two fields: one created by the flux guides, which is in the x direction, another is the background field of the planar field H, which is not "amplified" by the flux guides. The latter is expected to be significantly smaller than the former. Consequently, the resultant flux inside the sensor is not directed in the x direction but (slightly) tilted depending on the direction of the applied planar field. However, later simulations will show that unambiguous signals can still be obtained from the sensors. In other applications, when the flux guides are used merely to increase the flux density, the gap between the flux guides and the sensor should be as small as possible. However, in the current invention, the flux guides act as a flux diverter rather than only acting as a flux concentrator. Consequently, it is preferable that the sensor couple to the flux guides by magnetostatic interaction rather than exchange interaction. This is because the exchange coupling is known to maintain strongly the direction of the magnetic moments with little change. If there were some exchange coupling between the sensor and the flux guides, the flux lines when entering the sensor would prefer to continue in the same direction as they travel inside the flux guides, i.e. in e.g. the y' direction for sensor A. Of course, this reduces the effectiveness of the sensor. Simulation results (not shown) have proved this argument. The spacing between the sensor and the flux guides should be tuned in such a way that the exchange coupling is broken while the magneto static interaction is still strong. In practice, the minimum spacing would be about 10-20nm. In this section, an example of a pair of sensors using parallelogram-shaped flux guides is simulated using the finite element method. In the simulation, only the free layer is considered for the sake of simplicity. The sensor size is 6 x 1 μm². The flux guides have 2 tilted sides oriented at 45° and have dimensions of 6 and 8μm in sides. The spacing between the flux guides and the sensor is 0.2μm. A homogeneous magnetic field of 2 Oe is applied to the system. Figs. 20a and 20b show examples of simulated magnetic flux line images for sensor A when the field is oriented at 0° and 45°. The images show that the flux guides indeed divert the flux lines to a direction close to the x direction at the sensor location. The flux images of sensor B are similar but mirrored over the X-axis. Flux density (induction B) in the x-direction at the centers of sensors A and B at different applied field directions has been calculated (Fig. 21). Because the sensors are sensitive to the field in the x-direction, the signal of the sensor is proportional to the field in this direction if the characteristic of the sensor within this field range is linear (this assumption is realistic). Therefore the curves of the sensor signals (magnetoresistance) would be similar to the B vs. angle curves shown in Fig. 21. Fig. 21 shows the Bx curves (similar to signal curves) are sine- wave-like and are shifted in phase from each other. This is exactly the purpose of the design. In the ideal case, the phase difference should be 90°, that means the planar field can be determined by an orthogonal coordinate system, x' and y', or the field is decomposed into 2 orthogonal components. However, as mentioned above, this is not always the case. In practice, the direction of the flux lines traveling across the sensor is not always parallel with the x direction, resulting in a smaller phase difference (smaller than 90°) between the signals. In the given example, the actual phase difference is about 65°. This means the coordinate system is non- orthogonal. Nevertheless, the signals from both sensors are still unambiguous and this still allows the direction of the applied field over the full angle range from 0 to 360° to be determined. The two signals shown in Fig. 21 can be converted into an angle signal using conventional combinational circuitry, or using conventional digital circuitry such as a microprocessor or a look up table for example. In conclusion a way to construct 2D magnetic field sensors by including flux guides of different geometries for two separate sensors has been shown. The flux guides act as flux diverters to rotate the sensing direction of the sensors to two different directions. The combined signals from the sensors give information about the field vector being measured. An advantage of this design is that it does not need orthogonal exchange bias directions for the two sensors, thus the sensors can be integrated in the MRAM chip without any extra process steps, or in other words, the sensors can be fully compatible with the MRAM process. The flux guides can in fact be dummy MTJ devices, which means they can be patterned in the same way, at same time as the sensors and MRAM elements. Therefore patterning the flux guides does not need to cost any extra steps either. Finite element simulations have proved that indeed two sensors structured in this way can provide unambiguous signals which can be used to determine the angle and magnitude of the field to be measured. Notably unlike other flux guide designs, the flux guides should be constructed in such as way that the exchange coupling between them and the sensor is suppressed. The angle of the guides need not be limited to 45° between the flux guides and the sensing direction (x direction), although 45° might be the optimum angle. The flux guides of two sensors do not need to be mirrored, although mirroring is an optimum configuration to obtain equivalent and symmetrical behaviors of the sensors. The flux guides with special shapes described above can also be patterned from a separated layer of soft magnetic material, rather than using the free layer of the dummy devices. In this case, the flux guides can be fabricated using conventional techniques involving some extra steps. There is still the advantage in this case that the exchange bias direction does not need to be altered to form a set of orthogonal sensors. In practice, the sharp corners of the parallelogram-shaped flux guides may pose a problem, that is they can act as pinning centers. This means that the magnetic moments near these corners may be pinned and get difficult to rotate during magnetic reversal process. A consequence is that it may increase the coercivity of the flux guides, which is not desired. An easy solution to this problem is to round off the sharp corners. Micromagnetic simulations (not shown) have proved that rounded corners would suppress well the pinning effect. The rounded shape will not significantly influence the guiding effect of the flux guides. Any suitable conventional circuitry can be used as the detection circuitry for measuring the resistance of the sensing element, to suit the application. Readout circuits for MRAMs are well known and can be used also for sensors. Typically an op-amp is used to amplify a voltage seen across a load resistor, which is coupled in series with the sensor, via a bias transistor. The bias voltage on the sensor is clamped to a relatively fixed value (about 200m V) and the change in resistance of the sensor causes a change in current, which results in a voltage change on the load resistor. This voltage change is then amplified. A disadvantage of the circuit is that it results in some variation of the clamping voltage when the resistance of the sensor is changed. An improved circuit is known from US 6,205,073 Bl, in the context of MTJ memory readout. In this design, a bias control op-amp output is also fed to an input of the bias transistor 340. A negative input of the bias control op-amp is fed by the voltage across the sensor. In this design, the negative feedback of the bias control op- amp allows an active way of clamping the voltage on the sensor, which can offer a more stable signal and faster readout time. Concluding remarks: The current sensors described above can be implemented in integrated circuits of many kinds, particularly CMOS circuits and MRAM circuits. Outputs of such sensors can be coupled in scan chains following established practice, to multiplex many sensor outputs onto one or more outputs of the integrated circuit. Such integrated circuits can be used in conventional consumer equipment, particularly mobile devices such as laptop computers, mobile phones and so on. As has been described above, a sensor for detecting magnetic field strength, has a sensor element using a magnetic tunnel junction, and detection circuitry, the sensor element having a resistance which varies with the magnetic field, the sensor element comprises a tunnel junction, and the detection circuitry is arranged to detect a tunnel current flowing across the tunnel junction. Shape anisotropy, such as elongation, is orthogonal to the magnetic field. The sensor can have a magneto resistive sensing element having layers of magnetic material and one or more flux guides for concentrating the field onto the sensing element, the flux guide comprising a part of one or more of the same layers used for the sensing element. By using the same layer or layers for the flux guide the guide layer can be formed in the same step as the corresponding layer of the sensing element. Such sensors can be integrated in MRAM chips. Flux guides can be used to rotate the field differently for each of 2 parallel sensing elements, to enable 2D sensors. As there is now no need for orthogonal exchange bias directions for the two sensors, they can be integrated more easily. Flux guides can also be used to concentrate the field for writing MRAM cells, and hence reduce write current. Other variations can be envisaged within the scope of the claims.

Claims

CLAIMS:

1. A magnetic sensor for sensing a magnetic field and having a magneto resistive sensing element having layers of magnetic material and one or more flux guides for concentrating the field onto the sensing element, the flux guide comprising a part of one or more of the same layers used for the sensing element.

2. The sensor of claim 1 , the sensing element comprising a tunneling magnetic junction.

3. The sensor of claim 1 or 2, the flux guide having a larger area than the sensing element.

4. The sensor of any preceding claim having flux guides on opposing sides of the sensing element.

5. The sensor of any preceding claim, the flux guides being sized to be not fully saturated in the range of magnetic fields to be measured.

6. The sensor of any preceding claim, the sensing element having an elongate shape.

7. The sensor of any preceding claim the sensing element being planar and the tunnel current being detected perpendicular to the plane of the planar element.

8. The sensor of any preceding claim, the sensing element being arranged to have a relationship between resistance and field, which shows substantially no hysteresis.

9. The sensor of any preceding claim, the sensing element having a free magnetic layer which has an easy axis oriented to be substantially perpendicular to the field being measured.

10. The sensor of any preceding claim, the sensing element comprising a pinned magnetic layer having a magnetization oriented at an angle to the easy axis of the free magnetic layer.

11. The sensor of claim 10, wherein the angle is substantially perpendicular.

12. A magnetic sensor having a first sensing element for measuring a magnetic field in a first direction, and a second sensing element for measuring a magnetic field in a different direction, and a flux guide for at least one of the sensing elements, the flux guide being arranged to rotate the direction of the field through that sensing element.

13. The sensor of claim 12, the sensor having different flux guides having different geometries for each of the sensing elements.

14. The sensor of claim 12, the flux guides being arranged to rotate the magnetic field in different directions for each of the sensing elements.

15. The sensor of any of clams 12 to 14 having circuitry arranged to combine outputs from the sensing elements to give information about the magnetic field vector.

16. The sensor of any of claims 12 to 15, the first and second sensing elements and their flux guides being arranged to detect orthogonal components of the magnetic field.

17. The sensor of any of claims 12 to 16, the flux guides each comprising at least one side being oriented at an angle different from the sensitive direction of the sensing element.

18. The sensor of claim 17 wherein the flux guides have a parallelogram shape.

19. The sensor of any of claims 12 to 18, the sensing elements being integrated on the same chip.

20. The sensor of any of claims 12 to 19, the flux guides for either one of the sensing elements being oriented in parallel.

21. The sensor of claims 16 or any claim depending on claim 16, the circuitry being arranged to take into account an amount of background field not rotated or amplified by the flux guides.

22. A current sensor for sensing a current in a conductor, comprising a sensor as claimed in any preceding claim, located to detect a field caused by the current.

23. The current sensor of claim 22, a width of the sensor element in a direction parallel to a width of the conductor, being less than the width of the conductor.

24. An integrated circuit having a built in current sensor comprising the current sensor of claim 22 or 23.

25. The integrated circuit of claim 24, the current sensor being arranged to sense quiescent current (IDDQ).

26. The integrated circuit of claim 18 or 19, having multiple current sensors linked in a scan chain.

27. Consumer equipment having the integrated circuit of any of claims 24 to 26.

28. A method of sensing magnetic field using a sensor element having a magnetic tunneling junction having a resistance which varies with the magnetic field, the magnetic field being concentrated by a flux guide, characterized by the step of detecting a tunnel current flowing across the tunnel junction.

29. A method of manufacturing a sensor as claimed in any of claims 1 to 21, comprising the step of forming a layer of a sensing element and a layer of a flux guide for the sensing element, in the same step.

30. A magnetic memory having at least one memory cell, at least one of the cells having at least one flux guide for concentrating a magnetic field used for writing the cell.

31. The magnetic memory of claim 30, the guide comprises a part of the same layer or layers used by the memory cell.

32. The magnetic memory of claim 30 or 31 , the flux guides comprising dummy magnetic memory cells placed next to a corresponding memory element.

33. The magnetic memory of any of claims 30 to 32, the flux guides being co- planar with a free magnetic layer of the corresponding memory cell.

34. The magnetic memory of any of claims 30 to 33, the cell comprising a tunneling junction.

35. The magnetic memory of any of claims 30 to 34, flux guides being formed on two sides of the cell.

36. A method of manufacturing an MRAM comprising the step of forming a layer of a memory cell and a layer of a flux guide for the memory cell, in the same step.