Magnetic field angle sensor with GMR or MTJ elements

The invention relates to the general field of measurement of the direction of a magnetic field with particular reference to coverage of a full 360 degree range at minimum cost.

The present invention discloses a single chip solution for a GMR (giant magneto-resistance) or MTJ (magnetic tunneling junction)-based sensor with a full 360-degree range capability for magnetic field measurement. This design greatly enhances sensitivity relative to prior art devices. It does not require an additional Hall sensor in order to provide a full 360 degree range of measurement. It may be implemented using either a GMR (giant magneto-resistance) or an MTJ (magnetic tunnel junction) device, which terms we will use inter-changeably in the description that follows.

As illustrated in FIG. 1 , a GMR structure is deposited as a multi-layer structure starting with a seed-layer(11)/AFM(12)/AP2(13)/Ru(14)/AP1(15)/Cu(16)/free layer(17)/capping layer(18), where a ferromagnetic sub-layer AP2, non-magnetic spacer Ru and a ferromagnetic reference sub-layer AP1 form an anti-parallel synthetic pinned layer. In an MTJ structure, the Cu layer just below the free layer is replaced by an insulating tunnel barrier layer, (typically AlO x ). The synthetic layer is further pinned by anti-ferromagnetic layer (AFM). The pinning field, or exchange anisotropy, is related to the exchange coupling between an antiferromagnetic (AFM) layer and a ferromagnetic sub-layer (AP2) [2].

In a conventional angle sensor, the sensing elements are four long AMR (anisotropic magneto-resistance) [1] stripes oriented in a diamond shape with the ends connected together by metallization to form a Wheatstone bridge, as shown in FIG. 2 . The top and bottom connections of the four identical elements are given a direct current stimulus in the form of a supply voltage (Vs), with the remaining side connections to be measured as ΔV. With no magnetic field present, the side contacts should be at the same voltage. To have the elements' magnetization directions align with an externally applied magnetic field, the latter must be large enough to saturate the permalloy material.

With the AMR elements connected in this fashion to form the Wheatstone bridge, the side contacts will produce a different voltage (ΔV) as a function of the supply voltage, MR ratio, and the angle; which is the angle between the element current flow and element magnetization (M). One set of this bridge only provides a measurement of angles ranging from −45 degree to +45 degree. Combined with a second bridge which is oriented 45-degree in rotation from the first set, a wide range of angles, from −90 degree and +90 degrees, can be measured.

In this prior art design, due to the characteristic of the AMR effect, one of the AMR Wheatstone bridges only detects within a 90-degree angle range while two AMR Wheatstone bridges with 45-degrees orientation difference only allow measurement over a 180-degree angle range. In order to measure a full 360-degree angle, an additional Hall sensor must be used in combination with the two Wheatstone bridges.

1. Honeywell application note “Applications of Magnetic Position Sensors”

2. Taras Pokhil, et., “Exchange Anisotropy and Micromagnetic Properties of PtMn/NiFe bilayers,” J. Appl. Phys. 89, 6588 (2001)

A routine search of the prior art was performed with the following additional references of interest being found:

U.S. Pat. No. 7,095,596 (Schmollngruber et al.) discloses a 360 degree angle sensor comprised of two Wheatstone bridge circuits. U.S. Pat. No. 6,927,566 (Apel et al.) shows four GMR cells arranged at angle of 90 degrees to one another to measure 0 to 360 degrees.

In U.S. Pat. No. 6,640,652, Kikuchi et al. show a device for detecting a change in direction of a magnetic field, rather than the absolute direction of a static field. It uses two Wheatstone bridge arrangements. Each bridge has a pair of GMR elements connected in parallel on each of two sides. The GMR elements on each side are connected in series. The pairs of GMR elements are all parallel, antiparallel, or orthogonal relative to one another.

None of the prior art discussed above provides a method to simultaneously set pinned magnetizations in reference layers of four (or two) GMR elements along various directions during wafer level processes. Consequently the GMR elements have to be cut out of the same wafer and rearranged at 90 degrees to one another to be able to measure 0 to 360 degrees. This adds higher cost for manufacturing and greater error during angle detection.

It has been an object of at least one embodiment of the present invention to provide a method for measuring the direction of a magnetic field over the full 360-degree range.

Another object of at least one embodiment of the present invention has been to provide a single chip solution that implements said method.

Still another object of at least one embodiment of the present invention has been to provide a process to simultaneously set directions of pinned magnetization in reference layers of non-parallel MR devices such as GMR or MTJ devices.

A further object of at least one embodiment of the present invention has been for said method, when used with said device, to measure said angle to an accuracy of about 0.5 degree.

These objects have been achieved by fabricating a pair of Wheatstone bridges, driven by a common voltage supply, that is built out of three basic building blocks. The latter are all GMR and/or MTJ resistive elements that have a large shape anisotropy. These three basic blocks are oriented, relative to one another as follows: −45-degree (type-A), 0-degree (type-B) and +45-degree (type-C).

In the presence of the field whose direction is to be measured, each device will have a different electrical resistance so different voltages will appear at the four nodes of the two Wheatstone bridges. The resistive elements in the two bridges are arranged so that the voltage difference between the nodes of the first bridge is proportional to the sine of the angle being measured while the voltage difference between the nodes of the second bridge is proportional to the cosine of the angle being measured. These two voltages are inputted to a microcontroller where their quotient is computed, thereby eliminating the proportionality constant accompanying each signal. The result is the tangent of the angle under measurement making the value of the angle itself is readily computable

FIG. 1 is a schematic cross-section of a typical GMR device.

FIG. 2 illustrates a structure of prior art in which an electric current flows across a Wheatstone bridge constructed of four identical AMR stripes.

FIG. 3( a ) shows the three basic building blocks that are used to form the angle measuring device of the present invention

FIG. 3( b ) illustrates the direction of magnetization M of each of the free layers in the three MR resistive elements as well as the magnetizations R of their respective reference layers.

FIG. 4 is a schematic illustration of the device of the present invention.

FIG. 5( a ) shows arrows representing the magnetization directions of both the AP2 and AP1 (reference) layers with an external magnetic field applied along −x direction.

FIG. 5( b ) shows arrows representing the magnetization directions of both the AP2 and AP1 (reference) layers which align along the direction of the shape anisotropy for each of the three types of MR resistive element after removing the external magnetic field.

One class of anti-ferromagnetic materials includes the ordered tetragonal (fct) alloys such as PtMn, PtPdMn, NiMn, etc. The as-deposited state of these materials is a non-magnetic fcc structure and the ferromagnetic sub-layer has no exchange bias (pinning) with similar coercivity to that of a pure ferromagnetic layer. For all types of GMR stripes, in order to set pinning directions of synthetic anti-parallel layers along their own longitudinal axes, the magnetic moment of sub-layer AP2 is designed to be higher than that of the reference sub-layer AP1, resulting in a non-zero net magnetic moment of the synthetic anti-parallel pinned layer.

Once deposited, the GMR film is patterned into rectangular stripes that have a very large aspect ratio, whereby each stripe has a large shape anisotropy generated by the net magnetic moment along its longitudinal axis. Before thermal annealing is initiated, a large magnetic field is applied along the −x direction yjat is sufficient to almost saturate (magnetically) both sub-layer AP1 and sub-layer AP2. This large magnetic field and is then gradually reduced to zero followed by high temperature thermal annealing in the absence of any external magnetic field.

A pair of Wheatstone bridges is now constructed using different orientations of these GMR stripes as the building blocks. All stripes are made with exactly the same geometry and process. We will refer to these stripe orientations as follows:

−45-degree (type-A), 0-degree (type-B) and +45-degree (type-C), as shown in FIG. 3 a . In FIG. 3 b , M A , M B and M C represent magnetizations of free layers in the three types of GMR stripes, respectively, while Ref A , Ref B and Ref C represent the magnetization of the reference (i.e. pinned) layers in the three types of GMR stripe.

The two Wheatstone bridges are energized by a common voltage supply (typically between about 0.5 and 5.0 volts).

During magnetic field angle sensing, the magnetic field is large enough to saturate and align all GMR free layer magnetizations in the same field direction. Respectively, resistances for these three types of GMR stripes are:

R A = R + dR · 1 - cos ⁡ ( π 4 - θ ) 2 R B = R + dR · 1 - cos ⁡ ( θ ) 2 R C = R + dR · 1 - cos ⁡ ( π 4 + θ ) 2

where R represents the resistance when the free layer magnetization and pinned reference layer magnetization are parallel, dR represents the resistance change when the free layer magnetization rotates to be anti-parallel to the pinned reference layer magnetization.

We further obtained:

For the Wheatstone bridge configurations shown in FIG. 4 , it can be shown that the voltage difference ΔV 1 is proportional to sin θ while the voltage difference ΔV 2 is proportional to cos θ.

Thus, ΔU 1 =A 1 sin θ, ΔU 2 =A 2 cos θ. where A 1 and A 2 are the respective output amplitudes of the two amplifiers. If we make A 1 =A 2 , then angle θ can be determined from θ=arctan (ΔU 1 /ΔU 2 ), since the amplifications now cancel out. Computation of the value of arctan (ΔU 1 /ΔU 2 ) can be accomplished in microcontroller 44 , seen in FIG. 4 , either by direct computation or from a lookup table. Also, the equalization of A 1 and A 2 can be achieved by simultaneously feeding the same signal to amplifiers 43 in FIG. 4 , connecting their outputs to a differential amplifier, and then adjusting amplification of either or both until the differential amplifier reads zero. Alternatively, the differential amplifier may be omitted and the ratio of the amplifier strengths can be stored in the microcontroller as a normalizing constant for use during the computation of θ.

The device illustrated in FIG. 4 is capable of measuring angle θ to an accuracy of ±0.5 degrees.

In addition to computing a value for θ, it is also necessary to determine in which quadrant θ lies. This is accomplished in the microcontroller 44 by a comparison of the signs of ΔU 1 and ΔU 2 (see FIG. 4 ). These are summarized in TABLE I as follows:

Formation of the GMR/MTJ Stripes:

For all types of GMR stripes, it is required to set pinning directions of synthetic anti-parallel layers along their own longitudinal axes. For this purpose, the magnetic moment of sub-layer AP2 is designed to be higher than that of the reference sub-layer AP1, giving a non-zero net magnetic moment of the synthetic anti-parallel pinned layer. This is accomplished by making AP1 thinner than AP2. Typically, AP1 would be between about 10 and 30 Angstroms thick while AP2 would be between about 20 and 50 Angstroms thick

After deposition, the GMR film is patterned into rectangular stripes with very large aspect ratio) generally 3:1 or greater, so that in each stripe a large shape anisotropy is generated by the net magnetic moment along its longitudinal axis.

Before performing thermal annealing, a large magnetic field (typically between about 100 and 10,000 Oe) is applied along the −x direction, at or near saturation of both sub-layer AP1 and sub-layer AP2, and is then gradually reduced. As a result, the thinner AP1 magnetization is first to rotate toward the +x direction due to the anti-parallel coupling with the thicker sub-layer AP2, making the net moment of the synthetic AFM structure point towards the external field (−x direction), as shown in FIG. 5( a ).

Finally as the magnetic field is reduced to zero, the AP1 magnetization (with the AP2 magnetization being in the opposite direction) settles down along its stripe's longitudinal direction due to its longitudinal shape anisotropy, as shown in FIG. 5( b ). Then a high temperature thermal anneal (at between about 250 and 350° C. for up to about 1000 minutes) is conducted without the application of an external magnetic field. As a result, the magnetizations of the reference layers are permanently pinned by their AFM layers to lie along each GMR's long axis direction. Therefore, the pinned directions for type-A, type-B and type-C stripes are set to be −45-degree, 0-degree and +45-degree, respectively, relative to the +x direction.