U.S. patent application number 10/744625 was filed with the patent office on 2004-08-12 for use of multi-layer thin films as stress sensors.
Invention is credited to Ganapathi, Srinivasan K., Prakash, Shiva.
Application Number | 20040154405 10/744625 |
Document ID | / |
Family ID | 31497959 |
Filed Date | 2004-08-12 |
United States Patent
Application |
20040154405 |
Kind Code |
A1 |
Ganapathi, Srinivasan K. ;
et al. |
August 12, 2004 |
Use of multi-layer thin films as stress sensors
Abstract
The present invention provides a pressure sensing device that
includes at least one GMR sensor, and preferably an array of GMR
sensors, with each GMR sensor having a conducting spacer layer
interposed between two ferromagnetic layers. In an unbiased state,
the magnetization vector of each of the ferromagnetic layers is
preferably parallel to each other. Upon application of a current,
however, the magnetization vector of each ferromagnetic layer is
changed, preferably to an antiparallel position, in which state the
sensor is used to then sense stress applied thereto. Upon
application of stress, the magnetization vectors of both free
magnetic layers will rotate, thus causing a corresponding and
proportionally related change in the resistance of the sensor. This
change in resistance can be sensed and used to calculate the stress
applied thereto.
Inventors: |
Ganapathi, Srinivasan K.;
(Fremont, CA) ; Prakash, Shiva; (Santa Barbara,
CA) |
Correspondence
Address: |
PILLSBURY WINTHROP LLP
2475 HANOVER STREET
PALO ALTO
CA
94304-1114
US
|
Family ID: |
31497959 |
Appl. No.: |
10/744625 |
Filed: |
December 22, 2003 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10744625 |
Dec 22, 2003 |
|
|
|
09502406 |
Feb 10, 2000 |
|
|
|
6694822 |
|
|
|
|
60144843 |
Jul 20, 1999 |
|
|
|
Current U.S.
Class: |
73/763 |
Current CPC
Class: |
B82Y 25/00 20130101;
G01L 1/125 20130101 |
Class at
Publication: |
073/763 |
International
Class: |
G01L 001/00 |
Claims
I claim:
1. An apparatus for sensing pressure comprising: a sensor
including: first and second free ferromagnetic layers; and a non
magnetic conducting layer disposed between the first and second
free ferromagnetic layers, the non magnetic conducting layer
providing ferromagnetic coupling of the first and second free
ferromagnetic layers in an initial state such that magnetization
vectors of the first and second free ferromagnetic layers are
substantially parallel to each other, wherein applied pressure
causes a change in the magnetization vectors of the first and
second ferromagnetic layers through the property of non-zero
magnetostriction and a corresponding change in resistance of the
first and second ferromagnetic layers.
2. An apparatus according to claim 1 further including a plurality
of sensors, each that are formed in a two dimensional array and
operate as the sensor such that each sensor detects the pressure of
an area associated with that sensor.
3. An apparatus according to claim 1 wherein the first and second
ferromagnetic layers are comprised substantially of NiFe or CoFe,
either alone or in combination.
4. An apparatus according to claim 1 wherein the thickness of each
ferromagnetic layer is within the range of 0.5-15 nm.
5. An apparatus according to claim 4 wherein the thickness of the
non-magnetic conducting layer is within the range of 0.1 to 10
nm.
6. An apparatus according to claim 1 wherein the sensor further
includes a buffer layer disposed below the first ferromagnetic
layer to assist in uniform epitaxial growth of the first
ferromagnetic layer, and increased GMR response
7. An apparatus according to claim 6 wherein the buffer layer is
NiFeCr.
8. An apparatus according to claim 6 further including a capping
layer disposed above the second ferromagnetic layer.
9. An apparatus according to claim 1 further including a capping
layer disposed above the second ferromagnetic layer.
10. An apparatus according to claim 1 further including an
interlayer disposed between at least one of the first and second
ferromagnetic layers and the nonmagnetic conducting spacer, the
interlayer provided to increase GMR response and reduce
interdiffusion.
11. An apparatus according to claim 10 wherein the interlayer is
comprised of one of Co and CoFe.
12. An apparatus according to claim 1 further including an
interlayer disposed between each of the first and second
ferromagnetic layers and the nonmagnetic conducting spacer, each
interlayer provided to increase GMR response and reduce
interdiffusion.
13. An apparatus according to claim 12 wherein each interlayer is
comprised of one of Co and CoFe.
14. An apparatus according to claim 1 further including a specular
reflection underlayer disposed below the first ferromagnetic layer
to assist in increased GMR response.
15. An apparatus according to claim 14 wherein the underlayer is
one of NiO and gold.
16. An apparatus according to claim 1 further including a specular
reflection layer disposed above the second ferromagnetic layer to
assist in increased GMR response.
17. An apparatus according to claim 16 wherein the specular
reflection layer is one of NiO and gold.
18. An apparatus according to claim 1 further including a specular
reflection layer disposed above the second ferromagnetic layer to
assist in increased GMR response.
19. A apparatus according to claim 18 wherein the specular
reflection layer is one of NiO and gold.
20. An apparatus according to claim 1 wherein either one or both of
the first and second ferromagnetic layers is a laminate of a
plurality of layers that are antiferromagnetically coupled to each
other.
21. An apparatus according to claim 1 further including a
protective coating layer disposed above the second ferromagnetic
layer, said protective coating layer having a surface energy that
reduces deposits from adhering thereto.
22. An apparatus according to claim 21 wherein the protective
coating layer is a carbon based material.
23. An apparatus according to claim 22 wherein the carbon based
material is silicon carbide.
24. An apparatus according to claim 1 further including an
insulating layer disposed over the second ferromagnetic layer and a
conductive layer disposed over the insulating layer such that the
conductive layer provides for protection from electrostatic
discharge.
Description
RELATED APPLICATIONS
[0001] The present application is a divisional of U.S. application
Ser. No. 09/502,406 filed Feb. 10, 2000, which claims priority to
U.S. Provisional Application Serial No. 60/144,843 filed Jul. 20,
1999 and also claims priority to U.S. application Ser. No.
09/500,706 filed Feb. 9, 2000.
FIELD OF THE INVENTION
[0002] The present invention relates to Giant MagnetoResistive
(GMR) devices in conjunction with micromachined beams to measure
stresses with high sensitivity, and methods of making and using the
same.
BACKGROUND OF THE INVENTION
[0003] Stress and Pressure Sensors
[0004] The GMR effect has been widely reported in multi-layer thin
film sensors, where there are alternating ferromagnetic layers 12,
made from materials such as Cobalt, Iron or Nickel, separated by
nonmagnetic conductor layers 14, such as chromium or copper to form
a sensor 10 such as illustrated in FIG. 1. When a current is passed
along the length direction of the sensor, the electrical resistance
of the multi-layer stack of films varies as the relative angle
between the magnetizations of the individual ferromagnetic layers,
as shown in FIG. 1. The resistance is minimum when the
magnetization vectors between the neighboring ferromagnetic layers
are parallel to each other, and is maximum when the two vectors are
antiparallel to each other (at 180.degree.), as shown in FIG. 2. It
is to be noted that this is in contrast to the conventional AMR
effect, where the resistance is maximum when the magnetization
vector within a single magnetoresistive film is parallel to the
direction of the current, and minimum when it makes an angle of
90.degree. to the direction of the current. The AMR effect is shown
in FIGS. 3 and 4.
[0005] Typically, the change in electrical resistance of a GMR
multi-layer stack for a full rotation of the magnetization vector
from a parallel to an antiparallel state can be anywhere from 2% to
greater than 50%, and for the AMR effect, the change in resistance
for a 90 rotation is 1.5-3.5%. Therefore, if the magnetizations of
some layers in a GMR multi-layer stack can be made to rotate under
the application of a magnetic field, the GMR stack theoretically
will provide a greater sensitivity magnetic field sensor than a
conventional AMR film. However, one challenge in doing this is that
the exchange coupling magnetic field between the alternating
ferromagnetic layers in a GMR stack is very large, on the order of
2000 Oersted. As a result, to make an individual ferromagnetic
layer rotate in relation to a neighboring ferromagnetic layer
requires enormous magnetic fields. If the interlayer distance
between the neighboring ferromagnetic layers is increased to reduce
the exchange coupling field between the layers, then the GMR ratio
(percentage resistance change) decreases correspondingly. As a
result, it has not been possible to exploit the full extent of the
classical GMR effect.
[0006] One approach to overcome the problem described above is the
spin valve magnetic field sensor, a device that utilizes a version
of the GMR effect. The spin valve essentially consists of two
ferromagnetic layers 52 and 54 separated by a nonmagnetic
conducting spacer layer 56 as shown in FIG. 5. Of these two layers
52 and 54, layer 54 is a "pinned layer", in which the magnetization
vector is pinned in one direction. The other layer 52 is a soft
ferromagnetic layer, called the "free layer", whose magnetization
vector is free to rotate in the plane of the film. The separation
between the pinned layer 54 and the free layer 52 is chosen such
that the coupling field between the two layers is not too large. In
the quiescent state, the magnetization of the free layer 52 is
oriented at 90.degree. to that of the pinned layer 54, as shown by
the bold arrow in FIG. 5. Under the application of a relatively
weak magnetic field, the rotation of the magnetic moment of the
free layer 52 (as shown by the dashed arrows in FIG. 5) leads to a
change in the relative angle of the magnetization between the
pinned layer 54 and free layer 52, and results in a change in
resistance of the device.
[0007] FIG. 6 depicts a typical resistance change of the device as
a function of the applied magnetic field to the device. In the
quiescent state, the resistance of the device is represented by the
point X on the graph, and the change in resistance is linear for
changes in magnetic field almost up to the point of saturation of
the device as shown in FIG. 6.
[0008] The magnetization vector of the pinned layer in a spin valve
device is usually held in place through antiferromagnetic exchange
coupling between the pinned layer and a hard magnetic material (the
"pinning layer" 58 shown in FIG. 5 for example) such as CrMnPd,
PtMn, FeMn, NiMn, etc. Other methods to fix the magnetization of
the pinned layer include permanent magnet biasing, current induced
biasing, etc.
[0009] The classical Giant MagnetoResistive (GMR) effect, as it is
described above, has also been contemplated as being used to
measure mechanical strain induced by stress. This principle
involves generating a rotation of the magnetization vector of the
ferromagnetic film under the application of mechanical stress even
in the absence of a magnetic field, which results in a resistance
change of the film, which in turn can be used to infer the degree
of stress. However, one still needs to overcome the large exchange
coupling field between the alternating ferromagnetic layers, and in
order to do this, it has been suggested to use an externally
applied magnetic field to aid the rotation of one of the layers
under an applied stress, and to measure the resulting change in
resistance. However, in practice, this is very difficult to
implement, since it is not possible to apply such large magnetic
fields in sensors that are widely deployed in the field.
Additionally, this method causes serious accuracy problems, since
the effects of the externally applied magnetic field and the stress
on the sensor need to be decoupled.
[0010] It has also been suggested, such as in U.S. Pat. No.
5,856,617, to use a spin valve device of the type described above
to measure strains in the cantilever tip of an atomic force
microscope. In such a suggested strain gauge device, an example of
which is illustrated in FIG. 5, the free layer 52 is made to be of
non-zero magnetostriction, so that under zero magnetic field
conditions, the free layer magnetization vector rotates under the
application of stress to the cantilever beam, and the resulting
change in relative magnetization vector angle between the free
layer 52 and the pinned layer 54 leads to a resistance change in
the device. The strain gage device is thus a conventional top spin
valve, with the free layer 52 comprising an alloy of NiFe, Ni and
Co and being deposited directly onto the substrate, a non-magnetic
conducting layer 56 disposed between the free layer 52 and the
pinned layer 54, and with the antiferromagnetic (AFM) layer 58 that
is used for pinning the pinned layer 54 being on top of the stack.
Although this device may find some use in measuring strains on
atomic force microscope cantilever tips, there are several
disadvantages to the use of this device as a general purpose strain
gauge. The device's drawbacks relate mostly to the performance,
reliability and processing limitations that are inherent with this
type of design, and are listed below.
[0011] First, since an antiferromagnetic (AFM) layer 58 is used to
pin the pinned layer 54 through exchange coupling, the device is
subject to reliability concerns, since extended exposure of the AFM
material to elevated temperatures around 150-200 C can cause
"depinning" of the pinned layer, which destroys the effectiveness
of the sensor. This is especially true if the antiferromagnetic
material that is chosen has a low "Blocking temperature" (the
temperature at which the antiferromagnet starts to lose its
exchange anisotropy). Furthermore, if the AFM material chosen is
one that needs high temperature annealing, this introduces other
processing problems such as the compatibility of the film with the
substrate on which the multi-layer stack is being deposited, due to
thermal mismatch concerns and delamination of the stack. Moreover,
most manganese based AFMs have poor corrosion resistance.
[0012] Second, it is very difficult to maintain the magnetization
of the free layer 52 to be pointing in a direction that is at
90.degree. to the pinned layer 54 in the quiescent state. There are
several competing magnetic torques that affect the net quiescent
state magnetic moment of the free layer, including the intrinsic
stresses, the shape demagnetizing fields, current induced fields,
interlayer coupling fields with the pinned layer 54, and finally
the intrinsic anisotropy of the free layer 52. It is very difficult
to balance all these moments to arrive at a final moment vector
that is pointed at 90.degree. to the pinned layer 54. This is
especially true at the edges of the sensor. As a result, there is a
corresponding reduction in the sensitivity, and error in the
linearity and offset of the device.
[0013] Third, using the configuration such as described above,
there are several problems associated with stability of the device,
since the free layer 52 is magnetically very soft. During the
measurement of stress and strain, even very small magnetic fields
in the vicinity of the device can affect the rotation of the free
layer magnetization. This is further compounded by problems at the
edges of the sensor, where stress effects and other demagnetization
effects apply torques on the magnetization vector of the free layer
52. Therefore, it is desirable to have a free layer 52 that is
relatively insensitive to small extraneous fields.
[0014] Fourth, it is difficult to process the device and deposit
the multi-layer thin films on specialty substrates, such as Teflon
or other flexible substrates like kapton on which the strain is to
be measured. This is due to the fact that most AFMs require
elevated temperature annealing steps to get the required magnetic
properties. Elevated temperature annealing on such specialty
substrates is likely to introduce severe stresses due to thermal
mismatch, and peeling due to lack of adhesion.
[0015] Fifth, it is very difficult with this sensor to separate out
the intrinsic effects of the strains developed within the films due
to thermal mismatches between the substrate and the device during
elevated temperature processing or operation of the device. Since
the free layer is extremely sensitive to stresses, and it cannot
separate between an externally applied stress and an intrinsic
stress introduced during processing, linearity may be lost due to
processing stress, and random offsets may be encountered in the
field due to varying environmental conditions.
[0016] Sixth, this strain gauge is limited to realizing a total GMR
response of about 3-5%. Noting that this response corresponds to a
complete 180.degree. relative change in angle between pinned and
free layer magnetization vectors, the design described allows only
half of the total GMR effect to be realized, since the maximum
difference in angle between the pinned and free layer
magnetizations between the quiescent state and the fully stressed
state at zero magnetic field is only 90.degree., and not
180.degree.. Therefore, it is unlikely that one can obtain a
resistance change greater than 3-4% with this design.
[0017] Seventh, if the device is used to measure both tensile and
compressive stresses simultaneously, the maximum signal output is
reduced to one fourth of the total GMR response for either sign of
stress, and linearity of the response is compromised.
[0018] Eighth, it is difficult to separate the temperature induced
drift in the sensor due to an inherent material characteristic
called the "temperature coefficient of resistance (TCR)". The TCR
denotes the change of quiescent resistance of the device at zero
field, zero stress conditions as a function of the temperature.
Typically, this number is about 0.15-0.2%/.degree. C. for GMR
sensors. When the sensor is being used as a strain gauge, however,
it is difficult to separate out the resistance change of the sensor
into ambient temperature induced effects and stress induced
effects. As a result, strain gage measurement circuitry usually
involves elaborate circuitry to compensate for the temperature
induced changes in the sensor. A sensor such as the one described
is subject to the same drawbacks as other typical piezoresistive
sensors available today.
[0019] For all the reasons described above, it is necessary to make
substantial improvements in order to realize a stress sensor that
has high sensitivity (large response), good stability, and good
reliability.
[0020] Use of Pressure to Image Fingerprints
[0021] The fingerprint sensing industry uses several different
technologies to image fingerprints. The two most prominent
technologies are optical based sensors and capacitance based
sensors. Optical sensors use a light source, lenses and a prism to
image the "ridges" and valleys on a fingerprint, based on
differences in the reflected light from the features. The
conventional capacitance sensor uses semiconductor type processing
to fabricate a two-dimensional array of capacitors. The individual
sensors form one plate of the parallel plate capacitor, while the
finger itself, when placed on the array, acts as the second plate.
Upon contact with the array of sensors, the individual distance
from each sensor to the skin is measured using capacitive
techniques. The difference in distance to skin at the ridges and
valleys of a fingerprint provide the means to replicate the
fingerprint. An example of the use of capacitive sensors to measure
the spacing is shown in FIGS. 7A and 7B.
[0022] Both the above techniques fundamentally measure the spacing
between the fingerprint features, and the sensor. The measurement
of spacing is inherently subject to several distortion effects:
since the height difference between the ridges and valleys is only
of the order of 50 microns, any parameter which affects the spacing
between the finger and the sensor will affect the measurement. For
example, both types of sensors are very sensitive to the thickness
of the protective coating. They are also sensitive to oils or
grease on the finger, and the presence or absence of moisture on
the finger. In addition, most of these sensors are affected by the
ambient temperature at the time of sensing. Under very hot or very
cold conditions, the capacitive sensor can provide erroneous
readings. The combined effect of all these variables results in a
very distorted image of the fingerprint, as shown in FIG. 7C.
[0023] As a result of the above drawbacks to spacing based
reproduction of fingerprints, it would be very useful to be able to
use the difference in pressure exerted by the ridges and valleys of
a fingerprint on a sensor to replicate the fingerprint image. In
principle, a pressure based fingerprint sensor would be impervious
to the drawbacks listed above, such as wet or dry conditions on the
fingertip, presence of oil or grease on the fingertip, thickness of
protective coatings, etc, and would produce a "digital" response,
depending on whether the sensor experiences a ridge or not. This
situation is illustrated in FIGS. 7D and 7E, where the pressure
sensor can highlight only the ridges, which are the lines of
interest in a fingerprint. However, due to a variety of factors,
including the very low sensitivity and inability to provide the
required resolution, pressure based sensors have not been deployed
for the replication of fingerprints.
[0024] Accordingly, there remains a need for a device suitable for
use as a stress and/or pressure sensor that has high sensitivity
yet can provide high lateral resolution. Moreover, there further
remains a need for a sensor that is suitable for use in fingerprint
identification and verification that is less sensitive to adverse
conditions such as extreme temperatures and skin oils and
grease.
SUMMARY OF THE INVENTION
[0025] It is an object of the present invention to provide an
improved GMR sensor.
[0026] It is another object of the present invention to provide a
GMR sensor that does not require an additional layer to pin any
ferromagnetic layer, but instead uses an external current source to
properly bias the ferromagnetic layers.
[0027] It is another object of the present invention to provide a
GMR sensor that allows both ferromagnetic layers to freely rotate,
thus increasing the dynamic sensing range of the GMR sensor and
allowing the entire GMR response to be sensed.
[0028] It is a further object of the present invention to provide a
GMR sensor capable of sensing both compressive stress and
tension.
[0029] It is a further object of the present invention to provide a
GMR sensor that can be adapted to have substantial independence
from temperature shifts.
[0030] Another object of the invention is to provide a GMR sensor
that is suitable for use in fingerprint identification and
verification.
[0031] Another object of the invention is to provide a GMR sensor
that is suitable for use in fingerprint identification and
verification and that is less sensitive to adverse conditions such
as extreme temperatures and skin oils and grease. Another object of
the invention is to provide a GMR sensor that is suitable for use
in fingerprint identification and verification and that is less
sensitive to transient ESD voltages, and also mechanical
abrasion.
[0032] The present invention fulfills these and other objects of
the present invention by providing a pressure sensing device that
includes at least one GMR sensor, and preferably an array of GMR
sensors, with each GMR sensor having a conducting spacer layer
interposed between two ferromagnetic layers. In an unbiased state,
the magnetization vector of each of the ferromagnetic layers is
preferably parallel to each other. Upon application of a current,
however, the magnetization vector of each ferromagnetic layer is
changed, preferably to an antiparallel position, in which state the
sensor is used to then sense stress applied thereto. Upon
application of stress, the magnetization vectors of both free
magnetic layers will rotate, thus causing a corresponding and
proportionally related change in the resistance of the magnetic
material of the sensor. This change in resistance can be sensed and
used to calculate the stress applied thereto.
[0033] While the above provides an overview of the invention, there
exist numerous other significant aspects and advantages that will
become apparent in the discussion provided hereinafter.
BRIEF DESCRIPTION OF THE DRAWINGS
[0034] The above and other objectives, features, and advantages of
the present invention are further described in the detailed
description which follows, with reference to the drawings by way of
non-limiting exemplary embodiments of the present invention,
wherein like reference numerals represent similar parts of the
present invention throughout several views and wherein:
[0035] FIG. 1 is a perspective view of a prior art GMR multi-layer
stack, using alternating ferromagnetic layers separated by
non-magnetic spacers.
[0036] FIG. 2 is a graph illustrating the resistance response of a
GMR multi-layer stack to changes in relative angle between the
alternating ferromagnetic layers.
[0037] FIG. 3 is a top view of a prior art MR layer, illustrating
the angle between the magnetization direction and the current
direction.
[0038] FIG. 4 is a graph illustrating the resistance response of
the prior art MR layer stack to changes in relative angle between
the magnetization vector and the current direction shown in FIG.
3.
[0039] FIG. 5 is a perspective view of a prior art spin valve,
using a pinned ferromagnetic layer and a free ferromagnetic layer,
separated by a non-magnetic spacer.
[0040] FIG. 6 is a graph illustrating the resistance response of
the prior art spin valve to changes in relative angle between the
pinned and free ferromagnetic layers shown in FIG. 5.
[0041] FIGS. 7A-7E are side views of a fingertip on a prior art
capacitive sensor, and on a potential pressure sensor according to
the present invention which illustrate the use of spacing versus
pressure for replicating the fingerprint, and the distortion
associated with the prior art.
[0042] FIGS. 8A(1)-(3) are perspective views, not to scale, of a
GMR strain gauge sensor with current biased free layers, according
to a preferred embodiment of the present invention.
[0043] FIG. 8B is a graph illustrating the resistance response of
the strain gauge sensor of the preferred embodiment as a function
of the stress induced field.
[0044] FIG. 8C is a graph illustrating the resistance response of
the strain gauge sensor of the preferred embodiment as a function
of the applied stress for varying sense currents.
[0045] FIGS. 8D(1) and (2) are perspective views of other
embodiments of a GMR strain gauge sensor with current biased free
layers, according to a preferred embodiment of the present
invention.
[0046] FIGS. 8E(1)-(4) are perspective views, not to scale, of a
GMR strain gauge sensor with current biased free layers biased so
as to be capable of sensing tension and compression, according to a
preferred embodiment of the present invention.
[0047] FIG. 9 is a side view, not to scale, of a GMR strain gauge
sensor with current biased free layers, with one layer with
non-zero magnetostriction, according to the present invention.
[0048] FIG. 10 illustrates a beam or diaphragm substrate on which a
giant magnetoresistive sensor is deposited and patterned in
accordance with the invention.
[0049] FIG. 11 illustrates a two dimensional array of beams or
diaphragms employing giant magnetoresistive devices such as those
illustrated in FIG. 8A or 8D or 8E for use as pressure sensors in
accordance with the invention.
[0050] FIG. 12 illustrates the use of a conductive layer separated
by an insulating spacer from the sensor on the beam, which
conductive layer's function is to dissipate ESD charges to ground
and protect the sensor from ESD damage.
[0051] FIG. 13 illustrates an example of circuitry for performing
electronic measurement in a two dimensional array of pressure
sensors employing giant magnetoresistive sensor such as that
illustrated in FIG. 11 in accordance with the invention.
[0052] FIG. 14 illustrates a method for fabricating a beam or
diaphragm which would employ a giant magnetoresistive device such
as that illustrated in FIG. 8 for the purpose of pressure sensing
in accordance with the invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0053] Before describing the initial preferred embodiment in
detail, an overview of the concepts that are used by the present
invention will be first provided. Subsequently, the preferred
embodiments and alternative embodiments will be discussed. The
rotation of the magnetization vector of a soft ferromagnetic layer
that comes about from the magnetoelastic driving force is
proportional to the product of the stress and the magnetostriction.
The sensors of the present invention based upon this concept are
essentially multi-layer thin film stacks, which are deposited onto
the substrate to be monitored, and photolithographically patterned
to have a high aspect ratio, where the length is greater than the
width of the sensor. A uniaxial compressive or tensile stress in
bending, acting upon the length of the sensor, produces a rotation
of the magnetization vector of the soft ferromagnetic layer, even
under the absence of an external magnetic field. A requirement for
this magnetization rotation is the sensor's property of
magnetostriction, which must be non-zero (>+10.sup.-7 or
<-10.sup.-7), preferably +/-10.sup.-5 and whose sign must be
appropriate for the sign of the stress. The rotation of the
magnetization in turn produces a change in the resistivity of the
magnetic material, and, in the configuration wherein the current is
flowing through the device, preferably along the long axis
direction, causes a corresponding change in the output voltage.
[0054] This invention describes a preferred embodiment that uses an
externally applied current as the pinning source to fix the
magnetizations of the ferromagnetic layers. Specifically, the
preferred embodiment includes two layers of ferromagnetic material
(each of the layers typically a combination of NiFe, CoFe),
separated by a thin non-magnetic conducting spacer material such as
copper. The copper provides for a ferromagnetic coupling between
the two ferromagnetic layers, such that their magnetizations are
pointed in the same direction, which in the case of the preferred
embodiment is along the length of the sensor. Each of the
ferromagnetic layers is further made of a material of non-zero
magnetostriction.
[0055] Under the application of a predetermined current, the field
induced by the current is sufficient to induce a magnetic field in
each of the ferromagnetic layers which tends to rotate the
magnetization of each layer towards the width direction of the
sensor. However, the magnetizations of the two layers tend to
rotate in opposite directions, as dictated by the right hand rule.
If the current is of sufficient magnitude, the field induced by the
current is large enough to overcome the ferromagnetic coupling
between the two layers and their anisotropy, and the magnetizations
of the two layers orient themselves at 180.degree. to each other,
each pointing along the width direction of the sensor. This is the
quiescent, zero-field, zero stress state of the device. The
resistance of the sensor is maximum under this condition. If the
two ferromagnetic layers are of substantially the same magnetic
moments, the flux closure is complete and edge effects are
eliminated, allowing for uniform magnetization over the entire
film.
[0056] Under application of stress, for an appropriate combination
of sign of magnetostriction and sign of stress along the length of
the sensor, the magnetizations of the two ferromagnetic layers will
now rotate back towards the length direction. As they rotate, the
angle between their moment vectors decreases, and the resistance of
the device decreases. One can thus use this resistance change as a
measure of the stress. In the fully saturated state, the moments
are parallel to each other along the length direction, and the
resistance of the sensor is at a minimum.
[0057] With these principles having been set forth, the first
preferred embodiment will be described with reference to FIGS.
8A(1)-8A(3), which illustrate a strain gauge sensor 80 that
utilizes the GMR effect, but eliminates the AFM layer and the
necessity to have the free layer oriented perpendicular to the
pinned layer in the quiescent state. It also provides for realizing
the entire GMR response, rather than just half the response from
the sensor 80.
[0058] The sensor 80 preferably has a rectangular shape from a top
view, with current flowing along the length axis of the sensor. The
sensor multi-layer stack includes at least two ferromagnetic layers
82 as shown in FIG. 8A(1), with a nonmagnetic conductor spacer
layer 84 such as copper between them. Both of these ferromagnetic
layers 82 are free layers, with their magnetizations free to rotate
under the presence of applied magnetic fields or stress. Both of
the free layers have non-zero magnetostriction of the same sign and
substantially the same magnitude. The thickness of the free layers
82 can range from 0.5-15 nm (typical value is around 5-8 nm) and
spacer layer 84 thickness is chosen such that there is a weak
ferromagnetic coupling between the two ferromagnetic layers,
without unduly compromising the GMR response. The thickness of this
spacer layer 84, when copper is used, is typically in the range of
0.1-10 nm in order to provide appropriate coupling. Leads 85 are
connected to each length end of the device to provide for passing a
known current through the device and measuring the resulting
voltage of the device.
[0059] In the as deposited state, the magnetization in each
individual ferromagnetic layer 82 is pointed in a direction along
the length of the sensor 80, as shown in FIG. 8A(1). The
ferromagnetic coupling field between the two layers 82 (shown in
detail as 82A and 82B, which are referred to specifically when
needed) ensures that the magnetizations are therefore parallel to
each other in the absence of a bias current. However, when a bias
current of sufficient magnitude is applied, a current induced field
is generated in each of the ferromagnetic layers 82 as dictated by
the right hand rule. This current induced field tends to align the
magnetizations in each of the layers along the width direction, but
to be antiparallel (180.degree.) to each other, as shown in FIG.
8A(2).
[0060] In a typical design for such a sensor 80, one should
consider the bias current requirements to align the magnetizations
of the two ferromagnetic layers 82 to be antiparallel to each other
in the zero stress, zero field state. The ferromagnetic coupling
between the two ferromagnetic layers 82 is in the range of about
2-5 Oe for a copper thickness of 2-4 nm. Since there is an added
intrinsic anisotropy of 5 Oe for both of the layers 82 in the
length direction, a current induced field of about 7-10 Oe is
needed in order to completely rotate the magnetizations of the two
layers 82 towards the width direction, as shown in FIG. 8A(2). The
field induced by the current in each of the layers is given by
H.sub.current=H.sub.I=(4.pi./3)I/w Equation 1
[0061] where I is the current in milliamps, and "w" is the width of
the sensor in microns. Assuming a sensor width of 2 microns, a
current between 3-5 mA would completely rotate the magnetizations
of the ferromagnetic layers 82 to the width direction.
[0062] In a zero stress, zero field state, under the application of
sufficient bias current as dictated by the design criteria above,
the moments, or magnetization vectors, of the two free layers 82
become antiparallel to each other, aligned along the width
direction of the sensor 80, resulting in a maximum resistance. In
the case where the as deposited anisotropy of the free layers
(H.sub.k) is in a direction along the length axis, for a full
rotation of the magnetization vectors into an antiparallel
configuration under the application of a bias current, we define
the minimum required current, I.sub.min, as occurring when
H.sub.I>(H.sub.ILC+H.sub.k+H.sub.ch) Equation 2a
[0063] where H.sub.ILC is the interlayer coupling field and
H.sub.ch is the hard axis coercivity which also needs to be
overcome by the bias current for rotation of the magnetization
vector. It is to be noted that this treatment is for one embodiment
and does not take into account other embodiments where the
intrinsic anisotropy H.sub.k, is pointed in a direction along the
width of the sensor; nor does it consider the effect of the
relative magnitudes of H.sub.k and H.sub.ILC. These other
embodiments are just as viable as stress sensors, except that the
current requirements will change depending on the specific
embodiment used.
[0064] Equation 2a can be rewritten as
I.sub.min=(3w/4.pi.)*(H.sub.ILC+H.sub.k+H.sub.ch) Equation 2b
[0065] Now we consider the effect of an externally applied stress
on the magnetization vector of the ferromagnetic films, and thus
the resistance of the multi-layer stack. The stress induced
anisotropy field on ferromagnetic materials has been well
documented and understood; see, for example, A. C. Tam and H.
Schroeder, "A New High-Precision Optical Technique to Measure
Magnetostriction of a Thin Magnetic Film Deposited on a Substrate",
IEEE Transactions on Magnetics, Vol. 25, No. 3, May 1989, pp.
2629-2637). This stress induced field is given by
H.sub..sigma.=3.lambda.(.sigma.))/M.sub.s Equation 3
[0066] where H.sub..sigma. is the stress induced anisotropy tending
to rotate the magnetic moment of the free layers, .lambda. is the
magnetostriction of the free layer, .sigma. is the incremental
uniaxial applied stress on the film, and M.sub.s is the saturation
moment of the alloy used for the free layer. If .lambda.>0, the
stress induced anisotropy tends to rotate the magnetization in the
direction of tension (.sigma.>0); whereas if .lambda.<0, it
tends to rotate the magnetization in the direction of compression
(.sigma.<0).
[0067] The sign of the magnetostriction of the free layers 82 is
chosen such that, under the application of stress to the beam that
serves as the substrate for the multi-layer stack, the
magnetizations of both the free layers will rotate towards the
length direction of the beam. As the magnetizations rotate, the
angle between their moment vectors decreases, and the resistance of
the multi-layer stack decreases. One can thus use this resistance
change as a measure of the stress. The orientation of the intrinsic
anisotropies of the ferromagnetic layers 82, and the ferromagnetic
coupling between them will ensure that their magnetic moments will
align themselves to be parallel, rather than antiparallel, under
the application of stress, as shown in FIG. 8A(3). With the two
moment vectors now parallel, the resistance of the device is at a
minimum.
[0068] As an example, if the two ferromagnetic free layers are made
of negative magnetostriction, and if a compressive stress is
applied along the length direction of the sensor 80, the
magnetizations of the two layers 82 tend to rotate back towards the
length direction of the sensor. Due to the weak ferromagnetic
coupling, they rotate towards each other to be in a parallel state,
resulting in a decrease in resistance until it reaches a
minimum.
[0069] Quantitatively, under the application of stress, the minimum
condition for a full rotation of the magnetization vectors back to
a parallel configuration to counteract the effect of the bias
current can be written as
H.sub..sigma.=H.sub.I Equation 4
[0070] This equation suggests that the two externally induced
magnetic torques, the current induced anisotropy and the stress
induced anisotropy fields have to be considered together in
determining the final magnetization vector direction of the
ferromagnetic layers. When Equation 4 is satisfied, the
magnetization vectors of the two ferromagnetic layers are parallel
to each other and pointed along the length direction of the sensor,
and the resistance of the sensor is at a minimum. When
H.sub..sigma.<H.sub.I, the magnetization vector points at an
angle .theta. to the width direction dictated by the equation
H.sub..sigma./H.sub.I=sin .theta.. When there is no stress, and the
layer magnetizations are biased by current only, the resistance of
the device is at a maximum.
[0071] These principles are illustrated in FIG. 8B, which plots the
variation of device resistance as a function of the stress induced
anisotropy. Point A represents the resistance under zero applied
stress, and Point B represents the state when the stress induced
anisotropy exactly cancels the current induced anisotropy.
[0072] In order to quantitatively calculate the gauge factor or
sensitivity of this strain gauge, one can analyze the transfer
curve of FIG. 8B. We first calculate the slope of this transfer
curve by noting that under the application of stress, the maximum
resistance change .DELTA.R.sub.max in FIG. 8B occurs as soon as the
stress induced anisotropy field (H.sub..sigma.) equals the current
induced anisotropy field (H.sub.I), as long as I>I.sub.min.
[0073] Consequently,
For H.sub..sigma.=H.sub.I; .DELTA.R=.DELTA.R.sub.max for
I>Imin
or
(H.sub..sigma./H.sub.I)=(.DELTA.R/.DELTA.R.sub.max) for
I>I.sub.min
[0074] where .DELTA.R is defined as -(R.sub.max-R).
[0075] From equations 1 and 3, this can be written as
(3.lambda.(.DELTA..sigma.)*3*w/4.pi.*I*M.sub.s)=(.DELTA..sup.R/.DELTA.R.su-
b.max) for I>I.sub.min
or
(.DELTA.R/.sigma.)=(9.lambda.w.DELTA.R.sub.max/4.pi.*I*M.sub.s) for
I>I.sub.min Equation 5
[0076] This equation is plotted in FIG. 8C, illustrating the
dependence of resistance on applied stress, and the effect of
changing the bias current.
[0077] In order to calculate the sensitivity of the strain gauge,
we rewrite the variation of resistance with strain rather than
stress:
(.DELTA.R/.epsilon.)=E(9.lambda.w.DELTA.R.sub.max/4.pi.*I*M.sub.s)
for I>I.sub.min Equation 6
[0078] where E represents the Young's modulus of the film.
[0079] It is instructive to insert some typical values into
Equation 6 and obtain an estimate for the gauge factor for the
device proposed. With
[0080] W=2 .mu.m
[0081] .lambda.=1-2 e-5
[0082] E=2.4e11 dyne/sq.cm.
[0083] .DELTA.R=0.08-0.12
[0084] M.sub.s=1000 emu/cc
[0085] I=5 mA
[0086] GF=550-1500
[0087] Equation 6 illustrates quantitatively the dependence of the
strain gage performance on different parameters. Therefore, in
addition to affecting the gauge factor of the strain gauge through
choice of materials (primarily by varying the magnetostriction),
the sensitivity of the strain gauge sensor depends on the magnitude
of the current used to bias the sensor, the width of the sensor and
the magnitude of the GMR response .DELTA.R.sub.max. By increasing
the current beyond the minimum bias current, 1 mm, the sensitivity
of the strain gauge can be reduced, but the dynamic range of strain
increases correspondingly.
[0088] The performance of the sensor 80 illustrated in FIG. 8A can
be enhanced by adding layers in addition to the layers previously
discussed to obtain enhanced GMR response and maximum reliability.
For example, as shown in FIG. 8D(1) a buffer underlayer 86 of high
resistance material such as Tantalum is typically deposited below
the ferromagnetic layer 82A to ensure epitaxial growth of the
ferromagnetic layers. Similarly, the stack is "capped" off by
another similar buffer layer 88 such as tantalum or ruthenium over
the ferromagnetic layer 82B to prevent oxidation of the materials
of the stack, and to aid further processing of the material.
Improvements can be made in the GMR response by adding other or
additional underlayers, such as, for example, an underlayer 86A of
Ni.sub.48Fe.sub.12 Cr.sub.40 alloy just below the lower
ferromagnetic layer 82A. This layer of Ni--Fe--Cr can either
replace the Tantalum underlayer 86, or be deposited between the
Tantalum buffer underlayer 86 and the ferromagnetic free layer 82A,
as illustrated. This layer serves to improve the GMR response by
affecting the epitaxial growth morphology of the ferromagnetic
layers.
[0089] In addition, FIG. 8D(2) illustrates other layers that can be
added that can improve performance. As shown, although the free
layers 82A and 82B are a soft magnetic material such as Ni--Fe, it
is typical to include a thin interlayer, shown as 83A and 83B, of a
material, such as Cobalt, or an alloy of Co--Fe with a thickness of
0.2-2 nm at each interface of the free layers 82A and 82B with the
spacer layer 84 to enhance the GMR effect, and also to prevent
interdiffusion between the Ni--Fe of the free layers 82 and the
copper of the spacer layer 84. Additionally, the use of a metal
oxide layer 89 such as NiO, or a metal layer such as gold, with a
thickness of about 5 nm for specular reflection can be used to
increase the GMR response. The NiO layer 89 is placed between the
upper ferromagnetic free layer 82B and the tantalum capping layer
88, and improves the GMR response by eliminating the diffuse
scattering of electrons at the interface between the upper
ferromagnetic layer and the tantalum capping layer of the stack.
The specular reflecting metal oxide layer can also be used as a
substitute to the Ni--Fe--Cr underlayer 86A for the lower
ferromagnetic layer 82A.
[0090] The entire stack of thin films as described above is
deposited onto a substrate that is shaped in the form of a membrane
or a beam, as shown in FIGS. 8D1 and 8D2. After deposition onto the
membrane or beam, the multi-layer stack as described above is
patterned into a rectangular shape using photolithography
techniques. The sensor is positioned at a predetermined location on
the beam, such that it experiences a maximum state of stress. The
prior art for the locations of maximum state of stress in a beam or
membrane under load are well documented. If the length and width
dimensions of the sensor are made small enough relative to the
length and width dimensions of the beam, then a uniform state of
stress is introduced over the surface area of the sensor, and one
can expect a linear response from the sensor. The membrane or beam
is designed to flex under the application of a stress in the
vertical direction as shown in FIGS. 8D(1) and 8D(2), such that
there is either a tensile or compressive stress.
[0091] As mentioned previously, this embodiment corrects several of
the problems associated with the conventional GMR sensor used for
stress sensing. First, it eliminates the use of the AFM layer, by
using current to pin the sensor in the quiescent state. In addition
the ease of processing the wafer (the entire wafer can now be
processed using a low temperature process without any high
temperature annealing) also provides for better reliability by
reducing the susceptibility of the sensor to delamination from the
substrate due to differences in coefficient of thermal expansion.
It also reduces susceptibility of the sensor due to the absence of
manganese based AFM alloys.
[0092] Second, this embodiment obviates the need for the free layer
to be aligned at 90.degree. to the pinned layers in the as
deposited state through delicate balancing of the different
magnetic torques on the free layer. The use of a bias current is
very effective in setting the magnetizations of both the free
layers in the antiparallel state, resulting in a known condition
prior to stress measurement. The use of a bias current, combined
with efficient flux closure between the ferromagnetic layers also
eliminates the role of edge instabilities in the magnetization of
the free layer.
[0093] A third advantage of this embodiment is that one can realize
the entire GMR response because, in the quiescent state (with
sufficient current and in a zero stress state), the resistance is
maximum when the magnetizations are at 180.degree. to each other,
whereas under the application of stress in the saturated condition,
the magnetizations are parallel, resulting in a minimum
resistance.
[0094] A fourth advantage of this preferred embodiment is the
ability to detect both signs of stresses simultaneously using the
same sensor, which will now be described with reference to FIGS.
8E(1)-8E(4). The sensor 80 is physically the same as in the
embodiments of FIG. 8A or 8D, as illustrated in FIG. 8E(1). In
order to sense both signs of stresses simultaneously, however, the
current through the sensor is adjusted such that, in the quiescent
state under zero stress conditions, the current induced field in
each of the two neighboring ferromagnetic layers 82 is oriented at
an angle of 45.degree. to the direction of current, as shown in
FIG. 8E(2). As a result, the two magnetizations are now at
90.degree. to each other, and the resistance is at the midpoint of
the curve in FIG. 6, denoted by X. If the two ferromagnetic layers
82 have negative magnetostriction, for example, then the
application of a uniaxial compressive stress along the length of
the sensor causes the magnetizations of each of the films to rotate
towards each other, and towards the length direction of the sensor,
as shown in FIG. 8E(3). The resistance of the device therefore
decreases under these conditions. However, if the sensor
experiences uniaxial tension along the length of the sensor, the
magnetizations now rotate away from each other, towards the width
direction of the sensor as shown in FIG. 8E(4), resulting in an
increase in resistance. One can thus operate the sensor 80 to
detect both tension as well as compression in the sensor. The
linear behavior of the GMR response curve ensures a linear response
for a stress of either sign. The advantage of this embodiment over
the prior art is that it allows one to realize twice the signal
output for a stress of either sign.
[0095] The preferred embodiment of this invention also allows for
automatic temperature compensation of the temperature induced drift
of the sensor under varying ambient temperature conditions during
the long term, continuous monitoring of stress. In order to
separate out the temperature induced resistance change of the
sensor from the stress induced resistance change, a current pulse
of very small duration is passed through the sensor, typically on
the order of milliseconds, but of sufficient magnitude to
completely re-orient the magnetizations to their initial
antiparallel state even under the presence of stress. The resulting
resistance is noted, and represents the "quiescent state"
resistance of the device, even though the sensor is experiencing
stress, since the ferromagnetic layers have been instantaneously
forced into an antiparallel state by the current pulse. However,
this resistance also represents the resistance of the sensor at its
ambient operating temperature, which resistance varies according to
the temperature coefficient of resistance (TCR) of the device. It
is possible to match this resistance using simple electronic
circuitry against a previously determined, well calibrated curve of
the resistance variation of the sensor as a function of current and
temperature under zero stress conditions. The resistance change due
to temperature and pressure can thus be separated.
[0096] The magnitude of the current pulse for temperature
compensation is chosen according to Equations 1-4 above such that
it is sufficient to rotate the magnetizations antiparallel to each
other even under the presence of the largest stress that the sensor
is expected to experience. This instantaneous current pulse ensures
that the sensor goes back to its "quiescent state" of antiparallel
alignment even under the application of stress.
[0097] Another significant advantage of the preferred embodiment
lies in the ability of the user to adjust the response properties
of the strain gage sensor in a dynamic fashion. FIG. 8C shows an
example of this application. According to Equation 6, the
sensitivity and dynamic range of the strain gauge sensor is
inversely proportional to the applied current for any value of
current greater than a predetermined minimum.
[0098] While the preferred embodiment with the operating conditions
has been described above to obtain optimum performance and
reliability, an alternative embodiment, as shown in FIG. 9, has a
similar multi-layer stack structure to that of the preferred
embodiment, and includes ferromagnetic layers 82A and 82B and a
conducting spacer layer 84, as shown, and can also include, for
example an underlayer 86 and a capping layer 88 as shown. However,
in this case, one of the ferromagnetic layers has a non-zero
magnetostriction of a certain sign to allow its rotation under the
application of stress; while the other ferromagnetic layer has
non-zero magnetostriction of a sign opposite to that of the first
layer. FIG. 9 illustrates this as ferromagnetic layer 82A having
the non-zero magnetostriction, and ferromagnetic layer 82B having a
magnetostriction of opposite sign. Under the application of a
sufficient sense current, in a manner similar to that described
above, the magnetization vectors of the two layers 82 align,
preferably in a manner that is antiparallel to each other,
resulting in maximum resistance. Under the application of stress,
either tensile or compressive, however, in a zero field condition,
only the magnetization vector of one of the free layers rotates,
and in the fully saturated state, it aligns itself at 90.degree. to
the magnetization of the other ferromagnetic layer. One can thus
realize half of the total GMR response from such a structure. In
this configuration, if the two free ferromagnetic layers have
magnetostrictions of opposite sign, one can sense stresses of
either sign, either tensile or compressive. In this case, the
sensor will provide the same type of output signal regardless of
whether the applied stress is compressive or tensile in nature.
This type of configuration is useful when it is desirable to detect
just the presence of stress, regardless of whether it is tensile or
compressive.
[0099] Pressure Sensor Application of the GMR Sensor
[0100] In accordance with an aspect of the invention, a multi-layer
stack 1010 is the basis for a novel stress or pressure sensor,
which can be used, for example, for fingerprint sensing. In order
to use it, the underlying base structure for the GMR sensor 1010 is
fabricated as a suspended bridge, a cantilevered beam 1004 or a
similar kind of membrane that is supported over a cavity 1008 that
is formed in a bulk substrate and allows for deformation of the
cantilevered beam 1004, as shown in FIG. 10. An example of a method
of fabricating this is described in more detail below. A GMR stack
1010 comprising the two ferromagnetic layers, the non-magnetic
conducting spacer layer and the underlayer and capping layers are
deposited on deformable beam 1004, and leads 1012 are connected
thereto to send current I through it and to sense the resulting
voltage. It should be apparent that the support structure for MR
layer 1010 can be a variety of different structures other than the
beam 1004 of this example, such as a sealed membrane over an
enclosure, or any other form that may be suitable for detecting
deflection and stress.
[0101] Under the application of a force on the beam 1004, it
deflects. Note that in this configuration of the beam, the stress
direction needs to be either unidirectional or if bidirectional,
needs to be different along the two principal axes. If the length
of the beam is "l", the width "b", and the thickness "t", for a
load per unit length "W", the maximum compressive stress in the
center of the top surface of the beam is
.sigma..sub.max=Mt/2I=(1/4)(W/b)(l.sup.2/t.sup.2) (Equation 7)
[0102] where M is the moment at the center of the beam and I is the
moment of inertia about the vertical axis. The stress is tensile at
the edges of the top surface of the beam, and the sensor can also
be positioned at these locations, with the appropriate sign of
magnetostriction.
[0103] The maximum stress and strain occur on the surface of the
beam. Since the multi-layer stack 1010 is located on the surface of
the beam 1004, and is a very thin set of films, for calculation and
illustration purposes, one can assume that the GMR element is
subjected to the maximum stress and strain.
[0104] Under these conditions, the multi-layer stack is subjected
to the maximum compressive stress on the beam surface, as the beam
bends in response to applied pressure. The magnetostriction of the
device causes the resistance of the element to change depending on
the sign of the applied stress, as described earlier. In this way,
by measuring the resistance prior to and during the application of
the stress, the difference in resistance gives a good indication of
the magnitude of the stress. If the system is calibrated, this can
give an idea of the absolute stress as well as just the presence of
a pressure point.
[0105] In accordance with an aspect of the invention, the
sensitivity of the device as described above exceeds the
sensitivity of a capacitance based sensor or a piezoresistive
sensor. As a result, considerable miniaturization can be realized
with GMR sensor 1010. With conventional fabrication methods, the
GMR element can be made as small as one to two microns long while
maintaining considerable sensitivity (e.g., >1%). From equation
7, this sensitivity varies directly as the square of the length of
the beam, and inversely as the square of the thickness. As a
result, if the thickness is appropriately tailored for the length
of the beam, good sensitivity can be obtained even for beam lengths
of 1-2 microns and a thickness of 0.2 micron. It is desirable to
make the beam as long as possible within the constraints of the
product, because sensitivity only improves with the length of the
beam. In contrast, a capacitance based sensor or piezoresistance
based sensor would require a beam thickness of the order of a few
nanometers for a beam length of 2 micron in order to get any valid
signal for small values of applied pressure. This would make the
beam very weak from a mechanical standpoint, and make it
susceptible to fracture.
[0106] Preferably, in order to achieve the maximum sensitivity of
the sensor to an applied stress, in addition to designing the beam
for maximum deflection under a given load, it is appropriate to
choose an alloy and deposition conditions that would ensure a
maximum .DELTA.R/R response from the material, as well as a maximum
magnetostriction coefficient.
[0107] A magnetoresistive pressure sensor 1102 (that can be used
for fingerprint verification/identification, for example) employing
GMR sensors 1110 as described above will now be described in detail
with reference to FIG. 11. Several sensors 1110 as described above
are placed in an array 1106 of m rows by n columns on a substrate
1104, with a very fine pitch, and connected to electrodes 1108.
Because the sensors can be made as small as 1-2 microns in length,
the lateral resolution can be as good as 5 microns. For example,
when a finger is placed on the array, the ridges on the fingers
(which are spaced at about 400-500 microns, and are therefore
considerably wider than the sensor pitch in the array), apply a
force on the sensors that they come in contact with. The sensors
that fall between the ridges experience little to no stress. In
this way, one can generate a map of the contact points or the
ridges on the fingerprint, and get an accurate reproduction of the
fingerprint. However, in practice, the sensor length and pitch can
be made greater than 5 microns, since such a good resolution is
typically not required for a fingerprint image. Moreover, as the
sensor pitch decreases, the processing time for the image
increases, since the number of sensors in the image increases as
well. The length of the multi-layer stack is typically in the range
of 2-200 microns, and the width is in the range of 0.1-100 microns.
For a fingerprint image capture application, the sensor dimensions
are preferably 0.5-2 micron wide and 5-10 micron long.
[0108] In order to protect the GMR sensor during its use as a
fingerprint sensor, different schemes may be used. For ESD
protection, FIG. 12 shows a beam 1204 disposed over a cavity 1202,
with GMR stack 1210 formed on the beam 1204. A layer of insulating
material 1212 then coated on top of the GMR stack 1210, and then
another layer 1214 of a conductive metal (such as Titanium, copper,
etc.) is deposited, whose purpose is to bleed off transient charges
caused by ESD. This conductive film 1214 needs to be grounded, so
that the charges from the transient voltage spikes can be bled to
ground. This is illustrated in FIG. 12, with grounding leads 1216A
and 1216B shown from the conductive layer in order to dissipate ESD
charges to ground that is within the substrate.
[0109] Secondly, to protect the GMR stack from mechanical abrasion
or small impact, a hard coating can be deposited both above and
below the conductive ESD protection layer. The thicknesses of all
these layers would be in the range of 0.001 .mu.m-5 .mu.m. The
material used for the mechanically protective coatings could be a
material such as carbon based material, such as "diamond like
carbon" or silicon carbide, for example. These materials can also
be tailored to have surface energies such that undesired deposits,
such as debris or oils from the finger, are prevented from adhering
to the coating.
[0110] An example of electronic circuitry and a method that can be
used to probe an array of sensors either individually or in groups
is shown in FIG. 13. It includes a decoder 1302, a multiplexer
1304, and amplifier 1306, and A/D converter 1308. The techniques
for providing such electronic circuitry and method according to
this example are well known and detailed descriptions thereof are
not necessary for an understanding of the present invention.
[0111] In accordance with an aspect of a method of fingerprint
identification/verification in accordance with the invention,
however, to determine whether there is a stress on a particular
sensor, a baseline is first established with no finger on the
sensor, where the "quiescent" resistance of each element of the
array is measured. Then the readings are repeated with the finger
on the sensor, and the difference in voltages between the
"quiescent, unstressed" state and the "stressed" state is
calculated to determine the fingerprint pattern. The method of
scanning, providing power to each element, conversion of the values
from analog to digital, etc are well known. Note that the baseline
can desirably be established either immediately prior to or
immediately following the imaging of the fingerprint.
[0112] An advantage of the method of this invention is that it
always establishes a reference value, which eliminates the effect
of ambient temperature, humidity, stress, etc. The prior art, using
capacitive or optical means are unable to obtain such a reference
each time a measurement is taken, because they depend on the
presence of a finger to obtain a reading each time. Even though the
resistance of the sensor changes with temperature, this effect can
be automatically compensated for by establishing a reference value,
either immediately prior to or immediately following the
fingerprint imaging, as noted above.
[0113] It should be apparent to those skilled in the art that since
the MR sensor's output depends on a number of factors, it can be
used in a variety of alternative ways in this and other embodiments
other than for fingerprint identification/verification.
[0114] An example of a method for manufacturing a magnetoresistive
sensor in accordance with the invention will now be described with
reference to FIG. 14. As shown in FIG. 14A, a layer of silicon
nitride 1404 is first deposited on a silicon substrate 1402, then a
layer of polysilicon 1406, followed by another layer of silicon
nitride 1408. Then the pattern of the beam is etched using
photolithography by depositing a layer of photoresist 1410 as shown
in FIG. 14B, and etching through at least the top two layers of
silicon nitride and polysilicon as shown in FIG. 14C. This is
followed by preferentially etching out the second layer of
polysilicon underneath the silicon nitride in the beam portion
1412, thus forming a simply suspended beam 1412 of the upper layer
of silicon nitride as shown in FIG. 14D.
[0115] Alternatively, using similar techniques as are known in the
art, a thin silicon beam or membrane is made by starting from a
silicon wafer using conventional processing means as have been
described elsewhere in the literature. This is typically done
either by etching from the back side using anisotropic etchants, or
using single sided wafer processing, by first doping the wafer with
n-type doping elements (arsenic), covering with a layer of
epitaxial silicon, then etching using a chlorine gas plasma to
preferentially etch the N+ region. The preferential etching of the
N+ region under the top layer of epitaxial silicon leaves the upper
layer suspended as a beam, supported on either side.
[0116] It should be noted that even though the manufacturing
descriptions above are for the processing of silicon, one can use
other materials to achieve the same purpose as well. For example,
one could use a substrate of Aluminum, coat it with a layer of
insulating alumina (using thin film deposition techniques that are
well known), sputter another layer of metal (such as aluminum,
titanium, copper, etc.). One can now perform photolithography and
use dry etching techniques (such as ion milling) to etch down to
the underlying layer of aluminum, and then use etchants that are
selective to alumina to etch out the underlying alumina layer. In
this way, a "bridge" structure of aluminum or other metal can be
formed as well.
[0117] Once the beam is formed, a series of thin films is deposited
using thin film techniques (typically a cluster tool), representing
the magnetoresistive "stack". This series of materials comprises a
Tantalum layer (the underlayer), a layer of Ni--Fe--Cr (which is
the soft adjacent layer), another layer of Tantalum, followed by a
thin film of Nickel Iron (the MR material). This stack of thin
films is deposited on the entire substrate, and therefore will
cover the beam, and the exposed portions of the recess under the
beam. Only the portions of the stack on top of the beam are the
operative portions, hence a photolithographic process is used to
either wet etch or dry etch out the portions of the stack that
cover the rest of the substrate other than the portion 1414 on the
beam 1412, as shown in FIG. 14E. Typically, the beam can be made
anywhere from 0.2 micron to 20 microns thick; the length of the
beam can range from 2 micron to several hundred microns, and the
width of the beam can range from 2 micron to several microns. These
parameters depend on the mechanical properties of the substrate
material used to create the beam, and on the sensitivity required.
The beam can be designed to either have "standoff" from the
substrate, or can be a free standing structure, with the bottom of
the substrate completely etched out.
[0118] Although the present invention has been described in detail
with reference to the preferred embodiments thereof, those skilled
in the art will appreciate that various substitutions and
modifications can be made to the examples described herein while
remaining within the spirit and scope of the invention as defined
in the appended claims.
* * * * *