Magnetoelectric Sensors

FIELD OF THE INVENTION

This invention relates to magnetoelectric sensors and sensing methods.

BACKGROUND TO THE INVENTION

Physical phenomena may be commercialised if they deliver a disruptive technology that offers extraordinary savings in cost or improvements in performance, e.g. lipstick dye as the active medium in tuneable lasers. There is currently great scientific interest in magnetoelectric effects that may be either direct (change of electrical polarization P with applied magnetic field H) or converse (change of magnetization M with applied electric field E), but suggested applications in e.g. transducers and data storage have not been forthcoming. This is because single-phase materials perform poorly and the exploitation of large strain-mediated magnetoelectric coupling between two phases (e.g. magnetostrictive and piezoelectric materials) requires development work and the identification of end users.

Background prior art can be found in W. Eerenstein, N.D. Mathur, J.F. Scott, Nature, 442 (2006) 759; M. Fiebig, Journal of Physics D-Applied Physics, 38 (2005) R123; S.- W Cheong & M. Mostovoy, Nature Materials 6, 13 (2007); R. Ramesh & N. A. Spaldin, Nature Materials, 6, 21 (2007); A.D. Milliken, AJ. Bell, J.F. Scott, Applied Physics Letters, 90, 112910 (2007); W. Eerenstein, M. Wiora, J.L. Prieto, J.F. Scott, N.D. Mathur, Nature Materials, 6, 348, (2007); J. Ryu, S. Priya, K. Uchino, H.E. Kim, Journal of Electroceramics 8, 107 (Aug, 2002); Fetisov, Kamentsev and Ostashchenko, "Magnetoelectric effect in multilayer fenϊte-piezoelectric structures", Journal of Magnetism and Magnetic Materials 272-276 (2004); US5,675,252; GB709528A; US2004/0126620; and WO00/60369. Further background prior art can be found in: WO 2006/116041 (and equivalently US 2006/238281), US 3,713,056 and US

4,366,463; and also in J. Zhai, Z. Xing, S. Dong, J. Li, and D. Viehland, Appl. Phys. Lett. 93, 072906 (2008).

SUMMARY OF THE INVENTION

According to a first aspect of the invention there is therefore provided use of a multilayer capacitor (MLC) as a magnetoelectric sensor, said capacitor having a laminar multilayer structure comprising a plurality of layers of barium titanate (BaTiO ₃ )-based dielectric and first and second interdigitated nickel electrode planes between said layers.

The inventors have recognised that certain types of multilayer capacitor have a structure that in the present context is novel and makes them advantageous for use as a magnetoelectric sensor. More particularly the multilayer structure in which a plurality of layers are electrically-connected in parallel provides an arrangement which is useful for practical applications, even when what is detected is an output voltage. This is because in practical applications the magneto-electric sensor drives a load and thus the ability to deliver more current to the load provides an enhanced sensing capability. In embodiments this enables, for example, a magnetically-powered magnetic sensing system/device.

Furthermore, the combination of nickel and barium titanate is surprisingly effective - these would not normally be considered the materials of first choice because, for example, nickel is not particularly strongly magnetostrictive and because there is a range of materials which have a substantially greater piezoelectric response than barium titanate. (The skilled person will understand that neither the nickel nor, in particular, the barium titanate are pure in a multilayer capacitor; typically the barium titanate will include between 5-10% additives, for example plasticizer to facilitate processing during manufacture and/or materials to reduce a temperature coefficient of capacitance of the capacitor, and so forth).

In some preferred embodiments the piezoelectric material (barium titanate) is surrounded by material in which substantially no electric field is generated (because, in embodiments, this bounding material substantially lacks magnetostrictive (nickel)

electrodes to either side). It might be thought that this would be undesirable since, potentially, the presence of material which was not under the influence of a magnetostrictive effect could reduce the overall response of the sensor. However in practice it is useful in providing mechanical stability for the sensor over time, broadly speaking helping to hold the sensor together.

In some preferred methods the capacitor is of the type which has one or more sets of floating electrodes partially overlapping electrodes connected to connections to the multilayer capacitor. This provides an advantage for a magnetoelectric sensor in that magnetostrictive strain within the device tends to be concentrated in the regions of overlap of connected electrodes with floating electrodes and/or between sets of floating electrodes. This therefore has the effect of reducing strain propagation within the device since active regions are separated from one another by passive regions of the device, again overall contributing to mechanical stability and robustness, and hence to longevity.

The invention further provides a magnetic field sensor configured to use a multilayer capacitor, as described above, as a magnetoelectric sensor. In embodiments such a magnetic field sensor is configured to use the output from a multilayer capacitor of the type described above to provide a signal for sensing a magnetic field. The signal may be used in a range of different ways including, for example, to provide an indication of a sensed field.

In another aspect the invention provides a magnetoelectric sensor, the sensor comprising: first and second electrical connections; a plurality of layers of magnetostrictive metal, a first set of said layers of said magnetostrictive metal being connected to said first electrode connection and a second set of said layers of said magnetostrictive metal being connected to said second electrode connection, said first and second sets of layers of magnetostrictive metal being interdigitated; and a plurality of layers of piezoelectric material between said layers of magnetostrictive material; wherein a said layer of piezoelectric material and layers of said magnetostrictive metal to either side of the said layer of piezoelectric material define a magnetoelectric sensing element; and wherein a plurality of said magnetoelectric sensing elements are

connected electrically in parallel such that a change in magnetic field is able to cause a current to flow generated by said plurality of said layers of piezoelectric material.

In some preferred embodiments there are at least 5, 10, 20, 50 or more layers of piezoelectric material between the layers of magnetostrictive material. In embodiments layers of magnetostrictive material alternate with layers of piezoelectric material. In general, a magnetic field causes strain in the magnetostrictive material which is transferred, by mechanical coupling, to the piezoelectric material, which in turn generates an electric field causing a current to flow through any load connected to the sensor.

In embodiments although the first and second sets of layers of magnetostrictive metal are interdigitated, depending upon the configuration of the electrodes, electrodes making a connection to different (opposite) external electrodes of the sensor need not themselves overlap since there may be one or more sets of intermediate, floating electrodes within the interdigitated metal layers. Thus, as mentioned above, some preferred embodiments of the sensor include one or more floating electrodes, these preferably only partially overlapping the electrodes connected to electrical contacts to the sensor from which said current is drawn. Again, as previously mentioned, in some preferred embodiments a region in which the current is generated is bounded on one or more faces by piezoelectric material in which substantially no current is generated.

Magnetostrictive materials which may be employed include cobalt, iron, nickel, Permalloy, or a related nickel-iron magnetic alloy. Some particularly preferred magnetostrictive materials comprise rare earth alloys; of these Terfenol-D is particularly advantageous having very high magnetostriction (Terfenol-D comprises an alloy of Terbium, Dysprosium and Iron, the name being a contraction of the names of these constituents). Some particularly piezoelectric materials include the following and doped versions thereof: lead titanate, lead zirconate titanate (PZT), barium titanate, barium strontium titanate (BST) and lead magnesium niobate-lead titanate (PMN-PT).

A magnetoelectric sensor as described above may be incorporated into a magnetic field sensing device powered by current from the sensor. In embodiments the device may

comprise an indicator to provide an indication of a change in a sensed magnetic field, powered by the sensor. The indictor may comprise, for example, a very low voltage lamp (for example a lamp powered by a voltage of less than 1 volt) and/or an audible alert and/or a wireless transmitter, or the like. Such a device may need no permanent internal power source such as a battery.

The invention also provides a method of sensing a magnetic field using a magnetoelectric sensor as described above.

The skilled person will appreciate that although a sensor as described above is particularly useful for generating a current, nonetheless it may also be used to generate a voltage in situations in which substantially no current flows, for example when driving the gate of a field effect transistor to sense a magnetic field. Thus in use the sensor may be employed in either a "voltage-output" mode or a "current-output" mode.

Magnetoelectric sensors as described above can exhibit a strong temperature dependence which can mask the signal generated by a changing magnetic field. It would therefore be useful to provide temperature compensation for such a sensor.

Thus according to a further aspect of the invention there is provided a temperature- compensated magnetoelectric sensor system, the sensor system comprising: first and second sensor system electrical connections; first and second magnetoelectric sensors each having respective first and second electrodes, each of said first and second magnetoelectric sensors having a temperature response in which, with an increase in temperature, a voltage on said first electrode becomes more positive with respect to said second electrode, each of said first and second magnetoelectric sensors further having an anisotropic magnetic response in which a change in voltage across said first and second electrodes with a change in sensed magnetic field is greater when the sensor is in a first orientation with respect to said magnetic field than when the sensor is in a second orientation; and wherein, in said temperature-compensated magnetoelectric sensor system, one of said first and second magnetoelectric sensors is in said first orientation and the other of said first and second magnetoelectric sensors is in said second orientation, and wherein said electrodes of said first and second magnetoelectric sensors

are electrically connected to one another between said sensor system electrical connections such that with an increase in said temperature an increase in voltage across said first and second electrodes of one of said magnetoelectric sensors is compensated for, at said sensor system electrical connections, by an increase in voltage across said first and second electrodes of the other of said magnetoelectric sensors.

The skilled person will recognise that it is arbitrary which one of the sensors is in the first and which is in the second (different) orientation (for example, it could be the first sensor in the first orientation and vice versa); likewise it is arbitrary as to which sensor is considered as compensating which (for example the sensor in the first orientation could be considered to be sensor which is compensated). In some preferred implementations a magnetoelectric sensor used in the sensor system is as described above.

The first and second magnetoelectric sensors may be electrically connected to one another either in series or in parallel. In both cases the sensors are connected to one another so that the effects of temperature change on one opposes the effect of the same temperature change on the other. Preferably the temperature responses of the two magnetoelectric sensors are substantially matched to one another. In embodiments one of the orientations corresponds to a magnetic so-called easy axis of the sensor, more particularly of a magnetic material of the sensor, and the other orientation corresponds to a hard magnet axis of the sensor. In embodiments the first and second sensor orientations are substantially perpendicular to one another. In embodiments the first and second magnetoelectric sensors have oppositely directed electrical polarisation or at least oppositely directed components of electrical polarisation. The skilled person will appreciate that in embodiments of a sensor system and plurality of series and/or parallel coupled first and second magnetoelectric sensors may be employed.

In a further aspect the invention provides a keyboard, keypad, button or switch incorporating a magnetoelectric sensor or sensor system as described above.

Thus in a further related aspect the invention provides a keyboard or keypad comprising a plurality of pressure-activated buttons or keys, each having a magnet and an

associated passive magnetoelectric effect sensor, and wherein when a said button or key is pressed the respective magnet approaches the associated passive magnetoelectric effect sensor such that the sensor produces a voltage by magnetostriction of the sensor.

In preferred embodiments the sensors employed comprise temperature-compensated sensors, more particularly each sensor comprising a pair of anisotropic magnetoelectric sensing devices electrically connected such that they provide opposing (voltage or current) signals with a change in temperature. In some implementations each magnet is attached to a button or key or to a membrane supporting a button or key and the magnetoelectric effect sensors are mounted on a substrate below the buttons or keys.

BRIEF DESCRIPTION OF THE DRAWINGS

These and other aspects of the invention will now be further described, by way of example only, with reference to the accompanying figures in which:

Figure 1 shows magnetoelectric coupling in a multilayer capacitor (MLC) based on BaTiO ₃ with interdigitating Ni-electrodes showing (A) a schematic MLC cross-section (a magnetic easy axis lies in the plane of the electrodes, and the out-of-plane direction is hard); (B) in direct magnetoelectric measurements at 300 K after poling in E=I 02 kV/cm, the magnitude of the magnetoelectric voltage V(H) (symbols and dashed lines) across the MLC approximately tracks the MLC magnetization M(H) (solid lines), for both hard and easy axes; (C) F(0.5 T, T) with H || easy after poling in E = ± 102kV/cm, and depolarizing - the remnant MLC polarization P _x (T) approximately tracks the values of I F(0.5 T, 7)| measured after poling, and both quantities tend to zero near the Curie temperature Tc=393 K of pure BaTiO ₃ ; and (D) the converse magnetoelectric effect M(E) at 300 K in H= 0 is established by plotting M(f) and E(t) after saturating mμ ₀ H= 0.5 T before time t=0. H and M were aligned along a magnetically easy direction, and E = + 100 V7 9.8 μm (all dashed lines are guides to the eye);

Figures 2a and 2b show, respectively, an example of a 1206 surface mount multilayer capacitor and a scanning electron micrograph of a cross-section through a multilayer capacitor;

Figures 3 a and 3 b show, respectively, a schematic diagram of a magneto-electric sensor according to an embodiment of the invention, and an equivalent electrical circuit for the sensor of Figure 3 a;

Figure 4 shows a schematic diagram of a magnetoelectric sensor system according to an embodiment of the invention, with two MLCs as the MR elements - the MLCs, whose orientations differ by 90°, should be poled in opposite senses prior to electrical connection;

Figure 5 shows the magnetization M of a single MLC, as a function of applied magnetic field H, for easy in-plane (solid) and hard normal (dashed) directions with respect to the Ni-based planar electrodes; data were taken using a vibrating sample magnetometer, and M was calculated using the total electrode volume;

Figure 6 shows open-circuit ME voltage V generated across the terminals of a single MLC as a function of H applied along easy in-plane (solid) and hard normal (dashed) directions with respect to the Ni-based planar electrodes; H was swept over a period of ~10 ¹⁰ ω x lμF = 10 ⁴ s, such that there is negligible leakage of charge from the MLC during the measurement; data were corrected for linear thermal drift by equalizing the end points of multiple V(H) sweeps;

Figure 7 shows the ME response V of a sensor comprising two MLCs (solid, middle) to an optimally oriented H (see description later) swept over a period of ~100 s; the corresponding simulation (dashed, middle) is based on the easy and hard axis ME data for a single MLC, reproduced here (solid and dashed, lower & upper) from Fig. 6; data were corrected for linear thermal drift by equalizing the end points of multiple V(H) sweep;

Figure 8 shows temperature dependence of the voltage V across a single MLC (lower curve), and the sensor comprising two MLCs (upper line); the sweeps were performed on warming by δT~ 17 K over ~1000 s; thermal lag reduces dV/dT at the start of the sweeps, so to avoid underestimating the room-temperature value we work with δV/δT; the data were data at H= 0, but are substantially independent of H; and

Figure 9 shows use of a magnetoelectric sensor according to an embodiment of the invention in a keyboard or keypad.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

The use of nickel electrodes is reducing the cost of industrially manufactured multilayer capacitors (MLCs), but, by serendipity, nickel is magnetic. We demonstrate magnetoelectric effects in these one-cent devices by recording a change of voltage in response to an applied magnetic field. These MLCs are therefore cheap, room- temperature magnetic-field sensors that do not require electrical power. In multilayer capacitors (MLCs) based on BaTiO ₃ (doped with a few percent of proprietary additives), Ni is replacing Ag/Pd as the electrode material in order to cut costs. Here we show that these MLCs exhibit magnetoelectric coupling, even though they were designed for another purpose.

In a vibrating sample magnetometer with electrical access, all measurements were performed on a MLC (0.6 μF, AVX Corp., Coleraine, Northern Ireland) with 81 parallel capacitors (each of active area 4.5 mm ) formed from interdigitated Ni electrodes (1.5 μm thick) separated by BaTiO ₃ -based dielectric layers (9.8 μm thick) - Figure Ia. AU magnetoelectric measurements were quasi-static.

The MLC studied here has an industry standard footprint: "1206" (0.120 x 0.060) inch ² , 3.05 x 1.52 mm ² ). A Princeton Measurements Corporation vibrating sample magnetometer was used for all magnetic, ferroelectric, and magnetoelectric measurements, with a twisted pair of wires soldered to the sample providing electrical access. (The first report of the measurement probe with wires appeared in Nature Materials 6, 348 (2007)). In this magnetometer, the measurement axis and Hare

parallel. V was measured (as a dc signal) using a digital multimeter with an internal resistance of 10Mω. V was found to be symmetric about H = O (not shown). Mwas calculated over the electrode volume. P _r values were extracted from a series of temperature dependent P(E) loops taken at 50 Hz out to E = 4.9kV/cm. During all experiments involving an applied E, the current was monitored. The current always dropped to or below 1 x 10 ^" A several seconds after changing the applied E.

Equivalent results were obtained for this MLC across several temperature cycles, and similar results were obtained in other Ni / BaTiO ₃ MLCs with different specifications.

Figure Ib shows that at room temperature, the MLC develops a voltage V across its terminals (+ and -, Figure Ia) in an applied field H. The magnitude of this direct magnetoelectric response V(H) approximately tracks M(H), reflecting the role of the magnetostriction of Ni in the strain-mediated coupling. The input per unit output sensitivity dV/dH = 7.OxIO ^"6 V/Oe is largest for μoH< 0.1 T along an easy axis, such that 0.1 T generates 7mV. The MLC is in effect less sensitive than a SQUID, but operates at room temperature and only costs one cent. Compared to an expensive and larger sandwich using optimal materials (Terfenon-D and single-crystal Pb(Mn ₅ Nb)O ₃ - PbTiO ₃ ) (J. Ryu, S. Priya, K. Uchino, H.E. Kim, Journal of Electroceramics 8, 107 (Aug, 2002)), the MLC sensitivity is 10 ⁵ times smaller, and the magnetoelectric coupling constant dE/dH = 7.1 xlO ^"3 V/cm.Oe is 10 ³ smaller.

Figure Ic shows V(O.5 T) with H [[ easy after different electrical poling histories, at selected temperatures. The weak temperature dependence around room temperature is attractive for applications. The observed dependence of V(0.5 T) on both poling history and remnant polarization P ₁ (Figure Ic) reflects the fact that the dielectric of the MLC is in fact ferroelectric. Here, ferroelectricity breaks device symmetry and guarantees the piezoelectricity required for the strain-mediated direct coupling. Compared with pure BaTiO ₃ , Pr is suppressed but the Curie temperature is similar (Figure Ic).

Figure Id shows converse magnetoelectric effects (M(E)) in the MLC. Sudden changes in E applied across the terminals produce sharp changes in M. At 10 s, for example, the application of E = 100 kV/cm produces a 16% drop in M, which compares favourably

with the 26% drop in M (due to the application of E = 4 kV/cm) for an epitaxial La _{0 7} Sr ₀₃ MnO ₃ film on a single crystal BaTiO ₃ substrate at room temperature. The converse magnetoelectric coupling constant μodλd/dE = 3.4 x 10 ^~10 s/m for this specific measurement of the MLC is however relatively small. Subsequent removals and reversals of E produce further switches in M.

Geometrically, MLCs are inadvertently well-designed magnetoelectric transducers for three reasons. First the laminar nature of MLCs simplifies strain fields and consequently enhances coupling. Second, the large capacitance of the MLC permits magnetically induced output currents to be large. Third, the rigidity of the multilayer structure and its inactive surroundings (the unaddressed BaTiO ₃ ) should inhibit device failure via cracking (albeit at the potential expense of performance).

Figure 2a shows a photograph of a 1206-style surface mount multilayer capacitor resting on a one-cent coin illustrating a comparison between the two in terms of both size and cost. Figure 2b shows a scanning electron micrograph of a vertical cross- section through an example multilayer capacitor with floating electrodes (in Figure 2b the nickel electrodes are the dark lines against the lighter dielectric material). The structure of Figure 2b has, effectively, seven magnetoelectric sensor layers but more typically there is a much larger number of layers, for example greater than 50 or greater than 100 layers (in two particular examples, 80 and 290 layers).

Referring now to Figure 3a, this shows a schematic diagram of an embodiment of a magneto-electric sensor 300 according to the invention. The sensor has first and second electrical connections 302a, b to a plurality of layers of magnetostrictive metal 304, electrodes within layers 304a being connected to electrical connection 302a and electrodes within layers 304b being connected to electrical connection 302b. The layers of magnetostrictive metal include electrodes 306a, b which are directly connected to electrical connections 302a, b respectively, and one or more sets of floating electrodes 308. Layers of piezoelectric material 310 are disposed between the layers of magnetostrictive metal.

In the illustrated embodiment, which includes floating electrodes, current is generated substantially in regions 312 where the electrodes (including floating electrodes) coupled to the respective first and second electrical connections 302a, b overlap, thus breaking up active regions within the device. In the illustrated embodiment one or more additional layers of piezoelectric material 314 are included outside the outermost layers of magnetostrictive metal to help provide mechanical stability to the sensor.

Figure 3b illustrates, schematically, that the arrangement of Figure 3a effectively comprises a plurality of magnetoelectric sensors connected electrically in parallel. The skilled person will understand, however, that multiple sensors of the type illustrated in Figure 3a may additionally be connected in series, for increased voltage.

We have thus shown that in one-cent MLCs that can be mass-produced, magnetoelectric coupling is (1) easily measured, (2) a weakly varying function of temperature at room temperature, and (3) highly reproducible from sample to sample. MLC sensitivity dV/dH may be significantly improved via materials selection (J. Ryu, S. Priya, K. Uchino, H.E. Kim, Journal of Electroceramics 8, 107 (Aug, 2002)) and/or wiring the capacitor plates in series. The direct effect in MLCs may be exploited for energy harvesting, and for magnetic-field sensors that do not require electrical power, for example for underwater, space, health and safety, in-vivo, or toy applications.

We now describe a simple practical strategy to substantially eliminate unwanted temperature-dependent response seen in magnetoelectric (ME) magnetic-field sensors. The strategy involves electrically and thermally connecting two appropriately oriented magnetoelectric elements that are anisotropic and identical. Although this compromises the magnetoelectric performance, the thermal response is substantially nil in the absence of thermal gradients. For poled multilayer capacitors (MLCs) that are magnetoelectric, the magnitude of the temperature-induced voltage change is reduced from 59 mV/K for one MLC, to 1.0 mV/K for a pair of similar MLCs.

As we have described, an applied magnetic field H deforms that magnetostrictive material, and the transfer of strain to the piezoelectric material produces an electrical response that is paramaterized as a change of polarisation P, electric field E or voltage

V. Ferroelectrics are used in practice as these have the largest piezoelectric responses, but unfortunately the can also display a strong temperature dependence due to pyroelectricity.

In a ME heterostructure, primary pyroelectricity may be substantially eliminated by electroding two ferroelectric layers with suitably oriented polarizations. However, this does not eliminate secondary pyroelectric effects associated with the differential thermal expansion of the two materials in the heterostructure. We describe a technique which does not require permanent magnets, which represents a cost saving, and which does not require ME materials with specific orientations of magnetization and polarization, which means that the selection of materials and ME modes is unconstrained.

Preferred embodiments of the sensor use two matched, preferably substantially identical ME elements that display an anisotropic ME response. Without loss of generality, let us assume that the ME elements display piezoelectricity by virtue of displaying ferroelectricity. If the two ME elements were constrained to be at the same temperature, and were electrically connected in parallel with oppositely oriented polarizations, then the cancellation of pyroelectric effects would render the voltage across the elements independent of temperature. If the spatial orientation of the elements were identical, then ME effects would also cancel. But by selecting different orientations, the net ME response is reduced, rather than cancelled. A series arrangement of oppositely poled ME elements can offer ME voltage addition as well as temperature cancellation, but the charge generated is halved with respect to the parallel arrangement on which we focus here.

In more detail, the solid and dashed lines as shown in Fig. 6 (as described later) illustrate an anisotropic response of a single magnetoelectric sensor element, illustrating a relatively large response (with a sharp drop to zero at low field) when the magnetic field is applied in-plane, along an easy axis, and illustrating a small response of opposite polarity (positive) when the field is applied perpendicular to this plane (the plane in which the electrodes lie). Figure 7 shows (middle curve) the result when two magnetoelectric elements are connected together (in this example, in parallel), the two elements having different orientations, one perpendicular to the other: because the

response of a magnetoelectric element is anisotropic the outputs from the two magnetoelectric elements do not cancel each other out but instead there is, in effect, an average output response from the sensor. Thus the middle curves (in a vertical direction) in Figure 7 are in effect an average of the top and bottom curves (strictly speaking an average of the magnitudes of these curves); the middle curve with the dashed line represents an average of the lower, solid curve and the upper, dashed curve and, as can be seen, closely approximates the actual sensor output illustrated by the middle curve with the solid line.

The two connected magnetoelectric elements of Fig.4 are poled in opposite directions - that is the plates of one ME element which were connected to, say, a positive voltage terminal during the poling process are connected to the plates of the second ME element which were connected to a negative terminal during the poling process. In this way the effects of any temperature changes - which act on both the ME elements - cancel each other out. If the two ME elements had the same orientation, their magnetoelectric effects would also substantially cancel out. However the response of the ME elements is anisotropic and therefore by providing one of the ME elements in a different orientation to the other there is incomplete cancellation of the magnetoelectric response to a magnetic field whilst retaining substantially complete temperature cancellation. Thus returning again to Fig. 7, with the orientations shown in Fig. 4 a field applied along, say, the easy axis of one of the ME elements will be applied along the hard axis of the other ME element, resulting the average output response shown Fig. 7. A high degree of anisotropy is preferable and, as can be seen from Fig. 7, the response to a field in the hard direction is very small so that the effect of averaging is to approximately halve the response obtained from a field parallel to the easy (in-plane) direction.

Substantial cancellation of the effect of a temperature change can also be achieved by connecting the ME elements in series, when they are, in effect, connected in opposition to one another - in other words, the series connection between the two ME elements is formed by connecting together terminals poled using the same voltage polarity (either positive or negative).

In preferred embodiments, as described above both the ME elements are electrically poled using a relatively high electric field (for the case of a multilayer capacitor, greater than the manufacturers intended/specified maximum voltage). The ferroelectric material used in preferred ME sensing elements retains an remnant polarisation after this poling procedure. The skilled person will understand that although electrical poling of the material is convenient it is not essential - there are other mechanisms which can be employed to break the symmetry of an ME element.

By way of background, to assist understanding of the invention, broadly speaking heating reduces polarisation, that is the Ti ⁴⁺ cations in the BaTiO ₃ move back towards the centres of their unit cells. This causes the electrode plates that were poled positive (and pushed the Ti ⁴⁺ cations away) to develop a voltage that is positive with respect to the other plates.

For ME elements, we used MLCs with magnetostrictive Ni-based electrodes and piezoelectric BaTiO ₃ -based dielectric layers (Fig. 1). Such MLCs are inexpensive standard electronic components that function as magnetic-field sensors when poled to due strain-mediated ME coupling. The MLCs used here (1 μF, AVX Corp., model 12063C105KAT) differ slightly from those described above, and have 2 μm thick Ni- based planar electrodes spaced by 22 μm thick BaTiO ₃ -based dielectric layers, forming 108 capacitors in parallel, each of area 4.5 mm ² . All MLCs were poled in 250 V, and all electrical measurements were performed using a Keithley 2410 sourcemeter with internal resistance ~10 ¹⁰ ω. To form the sensor, two MLCs were placed just touching on a printed circuit board, electrically connected with silver paint, and overpainted with nail polish for thermal connectivity, mechanical stability, and electrical insulation. Soldering was avoided so that the ~130°C Curie temperature of the BaTiO ₃ -based dielectric was not exceeded.

We investigated the magnetic and ME behaviour of a single MLC: The MLC displays magnetic shape anisotropy (Fig. 5) and therefore an anisotropic ME response V(H) (Fig. 6). Although the ME response is relatively similar at different T (e.g. dF(0.5 T)ZdT ~ 0.06 mV/K over a wide range of T), V(T) is a strong function of T at any H (see later).

The ME response of the sensor comprising two MLCs (Fig. 4) is maximal when H lies parallel to the easy axis of one MLC and the hard axis of the other, i.e. along one of two directions. The measured output of this sensor configuration is presented in Fig. 7, and is seen to correspond closely to the expected response of [F(H _e asy)-F(Hh _a rd]/2, where V(H _easy ) and F(Hh _ard ) are the easy and hard axis responses of a single MLC (Fig. 6). As these two functions possess opposite signs, the sensor averages the magnitudes of the easy and hard axis ME responses of a single MLC.

The ME response of the sensor comprising two MLCs (Fig. 4) is maximal when H lies parallel to the easy axis of one MLC and the hard axis of the other, i.e. along one of two directions. The measured output of this sensor configuration is presented in Fig.7, and is seen to correspond closely to the expected response of [F(H _eaS y)-F(Hhaid]/2, where V(H _easy ) and F(Hh ₃ Td) ^ar e the easy and hard axis responses of a single MLC (Fig. 6). As these two functions possess opposite signs, the sensor averages the magnitudes of the easy and hard axis ME responses of a single MLC.

This reduction in ME performance (Fig. 7) is suffered in return for a significant reduction in temperature sensitivity. At any given H, whereas the single MLC shows AVfAT= -59 mV/K near room temperature, the corresponding figure for the sensor is AVIAT= 1.0 mV/K (Fig. 8). This reduction in temperature sensitivity means that a change of ~4 K would be required to mimic the 4 mV generated across the sensor terminals by a saturating η (Fig. 7), cf. ~0.1 K to mimic 7 mV across a single MLC (Fig. 6).

Thus, we have demonstrated a ME magnetic-field sensor based on two MLCs whose temperature dependencies cancel. Our strategy of temperature cancellation is relevant for applications where the temperature can change faster than the magnetic-field, i.e. low-frequency/quasi-static magnetic-field sensing. The MLCs were not designed as ME elements, and are therefore not optimized for this purpose. This explains why our low-field sensor output of 0.9 mV/cm Oe (Fig. 7) is three orders of magnetitude smaller than the corresponding figure of 1.2 V/cm Oe for magnetically biased back-to-back ME bilayers based on Terfenol-D and PbZro _.4 sTio _.52 0 ₃ . We have chosen parallel wiring to maximize the ME charge (4 mV x 2 μF = 8xlO ^"8 C, Fig. 7) for magnetic-field sensing,

whereas a series wiring would maximize the ME voltages presented here. The temperature dependence of the sensor is 59 smaller than it is for a single MLC, but his reduction may be increased substantially without limit as one tends towards substantially indentically poled MLCs of substantially identical capacitance.

Referring now to Figures 9a and 9b, these show an embodiment of a keyboard or keypad 900 employing a magnetoelectric sensor 902, preferably a temperature- compensated sensor, as described above. Each key 904 of the keypad comprises a small permanent magnet 906 mounted on a membrane 908 or the like which, when depressed as shown in Figure 9b, approaches the magnetoelectric sensor 902. The sensor then generates a voltage provided as an output from the keypad via an electrical connection (not shown). Such an arrangement can provide an inexpensive and reliable keypad, for example for a consumer electronic device such as a mobile phone or computer, or for, for example, an environment such an industrial environment where conditions are challenging for example because of dust or chemicals/contaminants or fluids, for example water, in particular where high reliability is desirable. One advantage of embodiments of a keyboard or keypad of the type shown in Figure 9 is that it can readily be made waterproof.

No doubt many other effective alternatives will occur to the skilled person. It will be understood that the invention is not limited to the described embodiments and encompasses modifications apparent to those skilled in the art lying within the spirit and scope of the claims appended hereto.