Note: Descriptions are shown in the official language in which they were submitted.
~313~03
G-3410 C-4077
Copending Patent Applications
This patent application is related to the
following Canadian patent applications, which are
assigned to the same a~signee to which this patent
application is assigned:
Canadian patent application Serial No.
604,137, which is based on United States Patent NoO
4,926,154 entitled, "Indium Arsenide Magnetoresistor,"
filed in the names of Jo&eph P. Heremans and Dale L.
Partin; and
Canadian patent application Serial No.
604,.133, which i~ based on United States Patent No.
4,978,938 entitled, "Improved Magnetoreaistor," filed
ln the names of Dale L. Partin, Joseph P. Heremans and
Donald T. Morelli.
This patent application iE also related to
the following earlier filed United States patent
application, which also i6 assigned to th~ assignee of
this invention;
Canadian patent application Serial No.
604,131, which is based on United States Patent No.
4,926,122 entitled, "Position Sensor," filed in the
names of Thaddeus Schroeder and ~runo P. B. Lequesne.
Field of the Invsntion /-
This invention relates to a position sensor
and more particularly to an improved magnetic field
sensing system having an improved magnetoresistive
sensor for detectiny changes in magnetic flux passing
1 31 3403
sensor for detecting changes in magnetic 1ux passing
through a magnetic flux sensitive element.
sackground of the Invention
This invention is a further improvement on
the improved magnetic field 6ensing system already
being de~cribed and claimed in the a~ove-identified
Canadian patent application Serial No. 604,131, filed
in the names of Thaddeus Schroeder and Bruno P. ~.
Le~uesne and entitled, "Position Sensor. ~t
The need for accurately and easily sensing
position, speed or acceleration is growing,
particularly in the automotive fiald. Anti lock
braking ~ystems, traction control systems, electric
power steering, four-wheel steering and throttle
control are examples of functions that can use such
sensing. Such applications not only reguire accuracy
and precision, but frequently involve se~ere
environments. Cost of such systems is an important
factor, too.
~ or such applications, it is desira~le to
have a position sen~or (speed and acceleration can he
derived from a position signal) that is rugged and
reliable, small and inexpensive, capable of low
~including zero) speed sensing and relatively immune to
electromagnetic field interference from the other
systems used in an automobile.
A well-known form of position sensor is a
semiconductor magnetoresistive sensor. Such a sensor
comprises a magnetic circuit that includes two basic
parts. One of these parts, typically kept stationary,
includes a semiconductive sensing element that is
sensitive to the magnetic ~lux density passing through
'.~
j,
1 3 1 3~03
its surface, and further includes a permanent ~agnet
for creating a reference flux~ The other of the two
parts, termed the exciter, includes a high magnetic
permeability ele~ent with a series of ~eeth that ~oves
with relation to the stationary ele~ent ~or changing
the reluctan~e of the magnetic circuit and for cau~ing
the ~agnetic flux through the ~ensing element to Yary
in a f~shion corresponding to the position of the
teeth.
Such a sensor is sensitive to the magnetic
flux den~ity rather than to the rate of flux density
change and so it does not have a lower speed limit.
Thi~ also makes it less sensitive to E.M.I. Moreover,
its response is predictahly related to the distribution
of flux density over the surface of the sensing
element.
Typically, the stationary part includes a
magnetoresistive element including a semiconductive
element whose resistance varies with the magnetic flux
density passing through it in controllable fashion so
that an electrical output signal can be derived.
Moreover, when this magnetoresistor is produced ~rom a
high electron mobility se~iconductor, such as compound
6emiconductors like indium antimonide or indium
arsenide, a large electrical output signal can ~e
available. If the output signal is sufficiently large,
there is the possibility of providing an output signal
that requires little or no further a~plification, B
factor of considerable advantage.
It is desirable to have a position sensor of
high sensitivity so that a large electrical output
signal can be produced efficiently and of easy
1313~03
~anufacture so that it can be made reliably and at low
c06t.
The magnitude of the flux variation~ in the
~en~ing ~l~ment for a given chang~ in position of the
exciter i6 ~n i~portant factor ~n d~t~rmining the
~nsitivity of the ~en~or. Accordingly, a variety of
d~ign6 have been atte~pted h~th~rto to m~xim~ze the
change in the 1ux density through the ~ensor in
respon~e to a given change in exciter po6ition.
Typically, th~6e ~ttemptc involved including a ~lux
guid~ for the p~rmanent m~gnet included in the
~tatlonary part of the magnetic circuit to provide a
r~turn path or the magnetlc ficld of the magn~t.
Additionally, sometime~ a field conc~ntrator of
commen~urate ~ize has been provided contiguous to the
magnetore~istive element to concentrate flux throu~h
the magnetore~istive element.
~owever, for example, ~uch technique~ have
typi~ally produced m~gnetic circuit 6ensitivitie~ no
higher than about five percent for a typi~al exciter
de~ign having a three milli~eter tooth pitch and one
millimeter gap, where the ~en6itivity i~ defined ~ the
differsnee between the maxi~um and mini~um flux
den6ities 6ensed divided by the ~an flux ~ensity
6en~ed (half the ~um of the maximum and minimu~ ~lux
den6ities ~ensed).
Two CoDIpanion Canadian patent
application6 were concurrently filed her~lith, CSN 604,137
and CSN 604,133, which are more fully identified above.
CSN 604,137 and C~N 604,133 de~cribe the ~brication
and prop~rtie~ of a new type of magnetore~i~tor thin
film el~ment. CSN 604,137 detail~ the proce66 of
growing a thin film of indium ar6~nide (InA~), a narrow-gap
"~
,!,'
1313403
semiconductor, on a semi-insulating indium phosphide
~InP) substrate, and shows that this device has a
rathsr large sensitivity of electrical re~istance to
magnetic field CSN 604,133 outlines various methods of
enhancing the sensitivity of the device on the basi6 of
the existence of a thin surface layer (known as an
accumulation or inversion layer) of high density, high
mobility electrons. Such electron accumulation or
strong inversion layers can be induced in a variety of
semiconductor thin films mater~als. While the devices
described therein could be used in a wlde variety of
magnetic field sensing applications without significant
further development, the application of these
magnetoresistors as position sensors in more ~tringent
operating condition6 (such as those which exist in an
automobile) requires interfacing the magnetoresistor
with a suitable sensing system.
We have recognized that the Schroeder and
Lequesne (CSN 604,131) type of mag~etic circuit i~ so
effective in concentrating the mag~etic ~ield that
lesser sensitive magnetoresistors may still work well
enough to be useful at ~ome applications. In addition,
we have recognized that some of the less sensitive
magnetoresistor materials are magnetically #ensitive at
higher temperatures. We have also recognized that the
improved magnetoresistor concepts of CSN 604,137 and
CSN 604,133 provide enhancement to lesser magnetically
sensitive materials. We have thus recognized that the
combination of all these concepts could provide
e6pecially striking benefits. This patent application
specifically describes and claims that combination.
There are several reasons why the improved ~ag-
netoresistors described in CSN 604,137 and CSN 604,133
- 1313403
would be especially desirable for use in such a sensing
systemO The reasons will not be mention~d in order of
importance. ~irst, extreme compactness of these
sensors make their use ideal in any sensing location,
regardless of the space limitations. Secondly~ their
improved sensitivity to magnetic field affords the
designer a large amount of freedom in the placement of
the sensor with respect to the exciter wheel. This
means that the air ~ap between exciter and sensor can
be larger than for a less sensitive de~ice without any
diminution in magnitude of the electrical signal. This
could prove to be important in applications where
vibration and thermal expansion problems limit the
degree of proximity o~ the sensor to the exciter wheel.
Also, the outstanding temperature stability o~ the
~ensitivity o~ the improved magnetore6i6tors will allow
their application in extreme t~mperature environments,
such as automotive anti-lock braking systems, in which
temperatures can range from -50~C to ~200~C. Other
applications may require operation at temperatures as
high as +300C. We believe that the enhancement to
system sensitivity afforded by the CSN 604,131 concepts
and the enhancement to magnetoresistor sensitivity
afforded by the C~N 604,137 and CSN 604,133 concepts,
in combination, makes a wider group of semiconductor
materials now available for use in magnetic field
sensing. Materials that were previously considered as
unacceptably now can be used, and will provide
acceptable performance at much higher temperature.
This expands the range of applications where such
sensing is practical, and provides other bene~its as
well.
- J `~
;
1313~03
Accordingly, we think that the combination
proposed in this patent application is especially
attractive for automotive applications as part of
linear or rotary position measurement systems. The
sensitivity to magnetic field and high thermal
stability of these sensors would be especia,lly
beneficial.
Summary of the Invention
The present invention is a novel magnetic
circuit for use in a position senBor. It features a
novel type of magnetoresistor that significantly
improves the circuit. The combination is simple and
planar in geometry, which makes it amenable for batch
processing with a consequent saving in manufacturillg
cost. Moreover, it makes possible attainment of
sensitivities and/or sensing at higher temperatures
appreciably higher than prior art structures.
In particular, the novel magnetic circui~
employs a stationary part that comprise6 a permanent
magnet whose width is several times wider than that of
the magnetic sensing element an,d, advantageously, at
least about one and one half times the pitch of the
exciter teeth. The sensing element is a magnetoresistor
having an accumulation layer on its sensing area
surface. Moreoverj in the preferred embodiment for
~urther improvement in the sensitivity, the surace o~
the magnet adjacent to which the teeth pass is provided
with a thin layer of a magnetic material of high
permeability. The magnetic sensing element
advantageously is centered on this magnetic layer and
is as described in CSN 604,137 or CSN 604,133.
Additionally, the width of the magnetic sensing element
. .
1313~03
is desirably narrow for ~aximum sensitivity, but is
wide enouyh to have a suitable resistance for good
impedance matching with the electrical circuit used to
det~ct the change in properties resulting from the
magnetic flux being sensed. Preferably any flux guide
or field concentrator is avoided by using a ~agnet of
adequate ~trength.
It is characteristic of this magnetic csrcuit
that the passing teeth of the exciter essentially vary
only ~he ~patial distribution of the magnetic flux
density along the width of the magnet for creating
sharp local flux density variations that can been
readily sensed by the sensing element, while the total
flux density passing through the thin ferro~agnetic
lS layer remains essentially constant. By way o
contrast, in prior art magnetic circuits, the passing
teeth of the exciter vary the circuit reluctance and
consequently vary the total magnetic flux in the
circuit.
The invention will be better understood from
the followin~ more detailed description taken with the
accompanying drawings.
Brief Description of the Drawings
Figure lA is a schematic view of a
magnetoresistor, showing its electrical current flow
lines when no magnetic field is applied to it.
Figure lB is a sche~atic view of a
magnetoresistor, showing how the electrical current
flow lines are redirected in the plane of its major
surface when a magnetic field is applied perpendicular
to that surface.
- 1313403
Figure 2 is an isometric view showing a
magnetoresistor having two integral sensing areas
electrically in parallel.
Figure 3 is a three-dimensional or contour
S plot showing the change of electrical resi~tance in a
inqle elç~ent larger band gap ~emicollductor
magnetoresi~tor with ehanges in temper~ture and
magnetic field strength.
Figure 4 is a two-dimen6ional plot of the
~ fractional magnetoresistance over a wider temperature
range than shown in Figure 3.
Figure 5 i6 a two-dimensional plot showing
change in resistance with no magnetic field applied
over a wider temperature range than shown in Figure 3.
Fi~ure 6 i~ an elevational view showing a
semiconductor film in a pattern for providing a series
connected plurality of sensing areas integrated in a
~ingle magnetoresistor.
Figure 7A is an elevational view showing a
metallization pattern for superpositivn on the Figure 6
pattern~
Figure 7B is an elevational ~iew showing the
Figure 7A metallization pattern superimposed on the
Figure 6 semiconductor pattern to delineate the
plurality of sensing areas.
Figure 8 is a three-dimensional or contour
plot showing the change of electrical resistance of a
multiple sensing area magnetoresistor such as show in
Figure 7B.
Figures 9 and 10 are two-dimensional electron
~nergy to depth plots showing how electrons could be
confined in an accumulation layer under special layers
on surface of the sensing area of the magnetoresistor.
1 31 3403
Figures llA, llB, and llC are schematic views
showing a magnetoresistor having a gate electrode over
each of a plurality of sensing area~ to electrically
induce an accumulation layer in each sensing area. In
Figures 7~ and 7C, the gate electrodes are electrically
biased internally, by two diferent tischni~ues.
Figure 12 is a schematic vi~w showing a
~gnetoresistor having accumulativn layers not only in
the sensing area , but also as conductors making
electriGal contact to the edges of the sensing areas;
Figure 13 shows a typical magnetic circuit of
a prior art position sensor o~ the type using a flux
guide return path;
Figure 14 shows the magnetic circuit of a
position sensor in accordance with a preferred
embodiment of the present invention;
Figure 15 shows in more detail the stationary
sensing portion of the magnetic circuit shown in Fi~ure
2;
Figures 16A and 16B show the magnetic circuit
of Figure 2 for two different positions of its
permanent magnet relative to the exciter; and
Figures 17 and 18 are plots useful in
discussing design considerations of the invention.
Description of the Preferred Embodiments
As indicated above, a new approach to making
magnetoresistors is described and claimed in CSN
604,137 and CSN 604,133. It was found that if an
accumulation layer is induced in the surface of an
extremely thin film of semiconductive material, the
properties of the accumulation layer relevant to
magnetic sensitivity can dominate over those of the
remainder of the film.
1313403
Such accumulation layers can make higher band
gap semiconductor materials useful in magneto~ensors.
Such materials can be used at higher operating
temperatures than lower band gap se~iconduetiYe
S material, such as indium anti~onide. ~owever, it may
ev~n enhance the sensitiYity of indiuan antimonide
enough to allcw it to be used at higher temperatures.
In this discussion the term accu~ulat:ion layer is uaed.
In this patent applica~ion, the term accumulation layer
lQ is intended to also include an inversion layer, unless
otherwise noted.
The accumulation layer i~ especially directed
to use in magnetoresistors ~ade of higher band gap
semiconductive materials. However, it is expected to
be beneficial in magnetoresistors made of still other
semiconductive material6.
A typical magnetoresistor element consists of
a slab of semiconductor, typically rectangular in
shape, through which a current is passed. Such a
ma~neto resistor is described by S. ~ataoka in "Recent
development of Magnetoresistive Devices and
~pplications~" Circulars of Electrotechnical La~oratory
No. 182, Agency of Industrial Science and Technology,
Tokyo (December 1974).
In the absence of magnetic field, the current
lines go from one injecting electrode to the other in
parallel lines ~see Figure lA). This flow is between
electrodes along the top and bottom edges of the
rectangle in Figure lA. The geometry (a rectangle in
our example) is chosen so that an applied ~agnetic
field, perpendi`cular to the slab, increases the current
line trajectory (see Figure lA). The magnetic field
perpendicular to the plane of the paper thus lengthens
~ 11
-
1 31 3403
the current flow lines~ The longer length leads to
higher electrical resistance, so long as the resulting
lateral voltage difference is electrically shorted, as
ahown, by the top and bottom edge electrode~.
The best geometry for this effect to occur is
one where the current injecting electrode6 are along
the longest side of the reotangle, and the ratio of
this dimension ~"width") to the shorte6t di~ension
("length"] is a~ large a~ possible. Such an optimal
device geo~etry hence leads to a very low resi~tance.
K~taoka teaches that the magnetic field sensitivity of
such devices is best when the devices are made out o~
semiconductors with as large a carrier mobility as
possible. The resistivity of such devices is made Less
temperature-dependent when the semiconductor material
contains a large donor concentration, giving a large
carrier density. These last two constraints i~ply that
semiconductors with high electrical conductivity are
best suited for practical applications.
Combined with the geometrical restriction~
described earlier, one can deduce that the final
magnetoresistor element will have a low resistance.
This has a practical drawback. Under a constant
; voltage, the power dissipated by the device scales as
the inverse of the resistance. To limit ohmic heating
(which would limit the operational temperature range of
the sensor, if not destroy the sensor itself~ while
maintaining a large voltage output during sensor
interrogation, it is desirable that a magnetoresistive
element have a resistance around lkW~ We con~ider this
to typically be equivalent to a resistance of about
300W-3kW. A number of ways have been proposed to
achieve such resistances. For example, as Kataoka has
12
1313~03
13
pointed out, one can put a number of elementary devices
in ~eries. Making a plurality of sensing areas as
integral parts of a single element i5 shown in Figure
2. While only two ~ensing areas (i.e., devices) ~re
ghown, on could make an element with tens or hundred6
of integral sensing areas i.e., devices).
If the ~etal-semiconductor (~agnetic-field
independent) interfacial contact resistance of one ~uch
elementary device is an appreciable fraction of the
semiconductor resistance of this elementary device~ it
will lower the sensitivity to a magnetic field. Thus,
metals must be deposited which have a very low
metal-semiconductor interfacial contact resistance to
avoid this sensitivity degradation. In most cases we
would prefer that the interfacial contact resistance
between the sensing area and its electrodes be 10-100
times less than the resistance of the sensing area
between those electrodes. Another option which
alleviates the problem of low magnetoresistor de~ice
resistance has been to use active layers that are as
thin as possible. This has been done by thinning
waers of indium antimonide ~InSb), which were sliced
from bulk ingots, down to thicknes~es as small as 10
microns. The wafer thinning process is a very
difficult process, since any residual damage from the
thinning process will lower the electron mobility.
Reducing electron mobility will decrease the
~ensitivity to a magnetic field of devices made from
this material.
Another approach has been to deposit fil~ of
InSb onto an insulating substrate. On the other hand,
in this latter case, the electron mobility of the
resulting films is reduced to a fraction of that of
,~ ~
' 13
1 31 3403
14
bulk InSb. This reduction occurs because of defects in
the film. With typical mobilities of 20,000
cm2V lsec 1, these films produce devices with greatly
reduced ~ensitivity to a ~agnetic fie].d compared to
devices made from bulk In5b. The usuzll device
6tructure for the prior magnetoresi~tor~ made ~rom a
i1m i~ schematically ~hown in Figure 2.
The great ma~ority of the prior work until now ha6
focused on InSb. This can be understood from th~ data
in the following Table I.
TABLE I.
Potential Magnetoresistor Materials at 300~
15 Semiconductive Maximum Crystal Energy
Material Electron L~ttice ~and
MQb2li~y -1 Constant Gap
(cm V sec ) (A) ~ eV)
InSb 7B,000 6.478 0.17
Bil-xSbx (xC0.2) 32,000 6.429(Bi) 0-0.02
InAs 32,000 6.058 0.35
; InO 53GaO.47 14,000 5.869 0.75
(on InP~
GaAs 8,000 5.654 1.4
GaSb 5,000 6O095 0.68
InP 4,500 5.869 1.27
Since the magnetoresistance effect is
proportional to electron ~obility squared for small
magnetic fields, InSb is highly preferable. However,
1 31 3403
the difficulty of growing co~pound semiconduc~ors in
general, and the fact tha~ there is no suitable,
lattice-matched, insula~ing sub~teate upon which it may
be grown led us to try growing Bi fil~ns. Such work has
5 been previously reported by Partin e~ al. in Phy_ical
Reviews B, 38, 3818-3824 tl988~ and by Hereman6 ~t al.
in Phy~ical Reviews B, 38, 10280-10284 5198~ u~ce~
wa~ obtained in growing epitaxial ~i thin fil~s~ with
mobilities as high as 25,000 cm2V~1 sec~1 at 300 K ~and
27,000 cm2V lsec 1 for Bi1 XSbx ~t 300~).
Magnetoresistors made from these films had very low
~ensitivities. Modeling studies which we have just
completed indicate that this is to our knowledge an
unrecognized ~ffect of the fact that the energy band
structure of Bi has several degenerate conduction band
minima. Other high mobility materials shown in Table I
have a single, non-degenerate conduction band minimum.
InSb thin films (on semi-insulating GaAs substrates)
were then grown using the metal organic chemical vapor
deposition (M~CVD) growth techniques. After many
m~nths of effort, films with electron mobilities of
only 5,000 em2V~1sec 1 were produced.
Growth of Indium Arsenide (InAs) on
semi-insulating GaAs, and also on semi-insulating InP
substrates, was tried. By semi-insulating we mean such
high resistivity that they can be considered as
substantially insulating. These latter substrate~ were
made ~emi-insulating by doping the~ with Fe. They were
tried in addition to Ga~s because there is less lattice
mis~atch with InAs (see Table I ) . After some time, we
were able to produce InAs films with a room temperature
~obility of 13,000 cm2V lsec 1 on InP substrates, and
1 31 3403
16
of lower mobility on GaAs substrates. The better InAs
films were formed by the following process.
An MOCVD reactor manufactured by Emcore
Corporation was used. InP substrates were heated to
the growth temperature in an atmosphere of 40 torr of
- high purity (palladium diffused3 hydrogen to which a
~oderate quantity of arsine was addedl (B0 9CCM, or
~tandard cubic centi~eters per minute ~ . This produced
about 0.02 mole fraction of arsine. The arsine was
used to retard thermal decomposition of the InP surface
caused by loss of the more volatile phosphorus. The
way in which arsine reduces the surface roughening
during this process is not well understood. Phosphine
would have been preferred, but was not available at the
time in our reactor. After reaching a temperature o
600C., the arsine flow was reduced to 7 SCCM, and
ethyl-dimethyl indium (EDMIn) was introduced to the
growth chamber by bubbling high purity hydrogen (100
SCCM) through ~DMIn which was held at 40C. Higher or
lower arsine flows during growth gave lower mobilities
and worse surface morphologies. After 2.5 hours of
InAs growth time, the EDMIn flow to the growth chamber
was stopped and the samples were cooled to room
temperature in an arsine-rich atmosphere ~as during
heat up).
The thickness of the resulting InAs film was
2.3 mm. From conventional Hall effect measurements at
300 ~, the electron density was 1.4x1016 cm 3 and the
electron mobility was 13,000 cm2V~1sec~1. These are
effectively averages since the electron density and
mobility may vary within a ilm. ~he fil~ was not
intentionally doped. Even thou~h this is a very
disappointing mobility, ~ crude magnetoresistor was
16
-- 1313~03
17
made, since this required very little effort. A
rectangular ~ample was cleaved from the growth and In
~etal was hand soldered along two opposing sdges of the
~ample, and leads were connected to the In. The length,
which is the vertical dimension in Figures lA and 1~,
was 2 mm and the width, which was the horizontal
dimension in Figures lA and lB, wafi 5 mm.
As expected, the resistance of the device was
low (abou~ 50 W) since we did not have many ele~ent~ in
serie60 However, the magnetoresistance effect was
large. It is shown in Figure 3. Furthermore, the
device resistance and magnetoresistanee were
surprisingly stable with te~perature in the range shown
in Figure 3, which is -50C. to ~lOO~C. A second,
similar device was tested less thoroughly at
temperatures a~ high as ~230C. The results of thi~
latter testing are shown in Figures 4 and 5. In Figure
4, the applied magnetic field was 0.4 Tesla. The
fractional ~agnetoresistance is plotted as a function
of temperature between E = O . 4 Tesla and ~ ~ 0. ~espite
the fact that the indium metal used for contacts has a
melting point of 156C., the magnetoresistor still
functioned very surprisingly well at 230C., with the
fractional increase in resistance for a given magnetic
field (0.4 Tesla) reduced by less than one half
compared to the response near room temperature (a~
~hown in Figure 4).
The device resistance in zero magnetic field,
R(0), decreased over the same temperature range by a
factor of 5 (as shown in Figure 5). Ne also fQund this
to be surprisingly good, even ta~ing into account the
relatively large energy gap of InAs.
~: 17
1 31 3~03
Our own detailed analysis of transport data
from these films suggests that there are current
carriers with two differen~ mobilities present. In
retrospect, it looks like our result~ are related to an
accumulation layer of electrons at the surface of the
~- ~en6ing layer. We have now recognized that Wieder has
reported in Appl. Phys. ~etters, 25, 20Ç ~1974) that
6uch an aceu~ulation layer exi~ts just inside the InA~
near the air/In~s interface. $here appear o us to be
some errors in the Weider report. However, we think
that the basic conclusion that an electron accumulation
layer exists is correct. These electrons are spatially
separated from the positive charge at the air/InA6
interface. Thus, they are scattered relatively little
by this charge, resulting in a higher mobility than
would normally be the case. They al50 exist in a very
high density in such an accumulation layer, so that as
the temperature increases, the density of thermally
generated carriers is a relatively small fraction of
the density in the accumulation layer. This helps
stabilize the resistance (at zero magnetic field) with
te~perature. Thus, it appears that the relatively low
measured electron mobility of 13,000 cm2V lsec~1 is an
average for electrons in the accumulation layer and for
those in the remainder of the thickness of the film.
Thus, normally one would want to grow a
relatively thick layer of InAs to make a good
~agnetoresistor, since crystal quality ~and mobility)
generally i~prove with thickness when growing on a
lattice-mismatched substrate. However, the thicker the
layer becomes, the greater its conductivity becomes and
the less apparent the benefits or presence of a surface
accumulation layer would be. Thus, our current
lB
1313403
19
understanding of our devices suggests that relatively
thinner layers are preferable, even if the average film
mobility decreases somewhat, since this will make the
conductivity of the surface accumulation layer a greater
fraction of the total film conductivity. The exact
relationships between film thickness, crystal quality and
properties of the surface accumulation layer are currently
under study. We currently prefer to use a nominally undoped
layer of a thickness of approximately 1-3 micrometers.
Multi-element magnetoresistors were ~ubseguently
made from this material using Au (or Sn) metallization.
First, conventional photolithography techniques were used
to etch away unwanted areas of an Indium Arsenide (InAs)
film from the surface of the Indium Phosphide (InP)
substrate to delineate the pattern shown in Figure 6. A
dilute solution (0.5%) of bromine in methanol was used to
etch the InAs. Then, a blanket layer of ~u metallization
1000 Angstroms thick was deposited using conventional
vacuum evaporation techniques over the entire surface o
the sample, after removing the photoresist. Conventional
photolithography was then used to etch away unwanted
areas of the Au film to delineate the gold pattern shown
in Figure 7A. A dilute aqueous solution of KCN was used
for this step. (~e think dissolved oxygen is helpful,
which can diffuse into the solution from ambient air or
be supplied in the form of a very small addition of
hydrogen peroxide.) The resultant composite of the two
patterns, with the gold pattern overlying the InAs film
pattern, is shown in Figure 7~.
Leads were then attached by silver epoxy to
the large AU end bonding pads. Leads could also be
attached by normal and accepted filamentary wire
bonding techniques. If so, and especially if a modern
19
., .:,,
1313~03
wire bonding apparatus were used, the bonding pads
could easily be made much smaller. ~lso, many devices
~uch as ~hown in Figures 6, 7 and 7A could be made
~imultaneousl~ using conventional integrated circuit
teohnology. The resulting devices typically have a
resi6tance near 1 KW (typically ~ or - 20~) at room
temperature in zero magnetic ~ield. Surprisingly, the
; magnetoresi~tance effect on the multisensing area
device was much larger than the effect on a single
sensing area device. For comparison, of these effects
at a given magnetic field, see Figures 8 and 3. In the
multi-element device (i.e., plural sensing area
element), the sensing areas had a length to width ratio
o~ 2/5. We do not understand why the multi-element
device works b~tter since the length to width ratio of
each element is 2/5, the same as for the single element
device characterized in Figure 3, which was fabrlcated
using part of the same InAs grown layer. Another
multi-element magnetoresistor was made similarly to the
one just described, but with a length to width ratio of
4/5. It had nearly as large a magnetoresistance as the
one made according to the patterns in Figs. 4 and 5.
Again, we do not yet understand this, but the resulting
devices work very well. Even a device with a length to
width ratio of 6/5 works well.
The relative stability of these
magnetoresistors with temperature also now appears to
be increasingly important, since some automotive
applications require operation from _50C. to as high
as +170C. to ~200C., and there are known applications
reyuiring even higher temperatures (to 300C.). There
is reason to believe that our invention will provide
. . ~: ~, ...
~ 3 1 3L~ 03
~agnetoresistors operating at temperature as high as
300~C., and even higher.
A potential problem with InAs
magnetoresistors made in accordance with thi~ invention
ls the potential importance of the air/InAs interace,
which ~ight cause the device characteristics to be
sen~itive ko changes in ~he composition of ambient air,
or cause the characteristics to slowly change with time
or thermal history because of continued o~idation of
the surface. Coating the surface~ of two d~vices with
a particular epoxy made by Emerson and Cuming, a
division of Grace Co. has been tried. The epoxy used
was "Stycast," number 1267. Parts A and B were ~ixed,
applied to the devices, and cured at 70C. ~or two
hours. We did not observe any significant change~ in
the device characteristics at room temperature as a
result of this encapsulation process. We have not yet
sy~tematically tested these devices at other
temperatures, but we are encouraged by this preliminary
result. We think other forms of encapsulants need to
be explored, such as other epoxies and thin film
dielectrics, such as SiO2 or Si3N4. Since exactly what
occurs at the air/InAs interface which causes the
accumulation layer is not yet known, intended for
exploration is the depositing of a thin film of
dielectric or high energy gap semiconductor (such as
GaAs, In1 xGa~As, In1 xAlxAs, or AlSb) right after
growth of the InAs is co~plete, and before expo~ure to
air. We hope that this will still result in an
accu~ulation layer at the interface between InAs and
the dielectric or high energy gap semiconductor.
In order to still have a very low
metal-semiconductor contact resistance between the InAs
21
1 31 3403
and the contact and shorting bar metallization, it ~ay
be necessary to modify the processing sequence
previously described in connection with Figures 6, 7A
and 7~. For example, with an inverse of ~he ma~k
conte~plated in the previous discussilDn, the
photoresist on the surface could then be used as a ~ask
for wet etching (e.g~, by wet ch~mical~ or reactive
ions, or ion beams) of the dielectric or high energy
gap semiconductor layer to expose the InAs. Au or
1~ other ~etals could then be deposited by vacuu~
evaporation ~or by other conventional processes, such
as sputtering, electroplating, etc.) and then the
photoresist could be removed, resulting in lift-off of
the undesired regions of metal. Alternatively, after
etching through to the InAs, the photore~ist could be
removed, Au or other metal could be deposited uniformly
across the surface, and then after deposition of
photoresist the mask pattern in Figure 7A could be
aligned with the pattern etched into the dielectric and
~ the Au could be patterned as before.
; As an additional alternative, if a
sufficiently thin layer (e.g., 200 Angstroms) of high
; energy gap semiconductor is present, the original
processing sequence described could be modified by
deposition of a low ~elting temperature eutectic alloy,
such as Au-Ge, Au-Ge-Ni, Ag-Sn, etc., in place of Au.
After patterning similarly to the way Au was (or using
the inverse of the mask in Figure 7 and lift-off), the
sample is heated to a moderate temperature, typically
to somewhere in the range of 360C. to 500C. for Au-Ge
based alloys, thus allowing the liquid ~tal to locally
dissolve the thin layer of high energy gap
semiconductor, effectively contacting the InAs.
1313~03
23
In most recent work, the InAs growth
procedures are changed somewhat. The procedures are
the same as before, but the InP wafer iB heated to
460C in a larger arsine mole fraction (0.1). After
0.5 minute at 460~C, during which the native oxide on
InP is believed to desorb, the ~emperature is lowered
to 400C and 200 Angstroms of In~s thickness is grown.
The temperature is then raised to ths growth
temperature of 625C (with the arsine mole fraction
still 0.1), and then EDMIn is introduced while the
arsine flow is abruptly reduced to 5 SCCM ~about 0.001
mole ~raction). The EDMIn is kept at 50C, and the
high purity hydrogen is bubbling through it at a rate
of 75 SCCM. Again, the arslne ~low of 5 SCCM seems
near-optimal for these growth conditions. The
re~ulting films have somewhat enhanced sensitivity to a
magnetic field relative to those grown earlier.
While all of our recent work has concentrated
on magnetoresistors fabricated from InAs films on
semi-insulating (i.e., substantially electrically
insulating) InP substrates, we think that a more mature
growth capability will permit films of InAs with nearly
comparable quality to be grown on semi-insulating GaAs
substrates as well. In either case, other growth
techniques such as molecular beam epitaxy liquid phase
epitaxy or chloride-transport vapor phase epitaxy may
also prove useful.
We are describing and claiming the above-
mentioned Indium Arsenide (InAs) thin film devices,fabrication processes, and operating characteristics
in a separate Canadian patent application Serial
No. 604,137 entitled, "Indium ~rsenide Magneto-
resistor," that is being si~ultaneously filed
23
1 31 3403
24
with this patent application in the names of J. P.
Hereman~ and D. L. Partin.
On the other hand, we think that the presence
cf what may be a naturally occurring accumulation layer
in the above-mentioned thin film In~s ~agnetore6istors
i~ what ~akes th~m work so well, and which enabled
production of a practical device. We believe that this
fundamental concept is new to ~agnetore~istors, and
that this thought can be expanded in a aultiplicity of
ways, not only to Indium Arsenide but to other
semiconductive materials as well. In this patent
application we further describe and claim a variety of
techniques by which an accumulatlon layer can be
induced in the semiconductor layer, by other than a
naturally occurrence or inherent occurrence as a result
of the fabrication process.
The following discussion describes some of
the other ways of inducing or enhancing an electron
accumulation or inversion layer in InAs thin films and
in other semioonductive materials in thin film form, to
attain effective high mobilities. There are three
basic advantages to the use of strong electron
accumulation layers in magnetoresistor active regions.
It is repeated here that the term electron accumulation
layer, as used in this patent application is also
intended to include electron inversion layers.
First, electron accumulation layers or strong
ele~tron inversion layers can contain a density of
electrons significantly larger than the intrinsic
den~ity at any given temperature. This must improve
- the temperature stability, since the thermally excited
carriers are a small fraction of the accumulated or
strongly inverted ones.
1 31 3403
Sesond, accumulation layers enhance the
mobility of the carriers in the semiconductor. This
effect has been experimentally observed in thin indium
arsenide tInAs) fil~s, ~speci~lly at higher
temperatures. They will enhance the ~;ensitlvity of ths
~agnetoresistor. One po6sible cause of thi6 ~ffect may
be that in such accumulated or stzongly inverted layers
large electron densities can be achieved without the
presence of a large density of ionized impurities in
the same spatial region, which would limit th~ carrier
mobility. This efect is similar to the "modulation
doping" of layers described by ~. Burns in Solid State
Physics, pp. 726 747, ~cademic Press (1985). Such an
effect is us~d in the fabrication of
~5 ~igh--Electron-Mobility-Transistors (HEMTs).
Third, accumulation or strong inversion
layers are inherently close to the surface or interface
of a semi~onductor. This makes it relatively easy to
induce, enhance, or control these accumulation or
strong inversion layers through the use of thin film
structures deposited on top of the semiconductor,
po~sibly in combination with voltage bia~es.
Accumulation layers have been used in silicon
MOSFET Hall plates, and is described by H. P. Baltes et
al. in Proc. IEEE, 74, pp. 1107-1132, especially pp.
1116-7, (1986). In the MOSFET Hall effect devices, a
biased gate electrode in a Metal-Oxide-Semiconductor
was used to generate a suitably thin electron layer
close to the Se~iconductor-Oxide interface. Four
electrodes were then used to eontact that layer: a
~ource and a drain through which current is passed, and
two intermediate electrodes across which the Hall
voltage is generated. Further, saltes et al. ibid.
1 3 1 3~03
26
also describe a split-drain MQSFET using an
accumulation-layer based sensor with only four
electrodes (one ~ource, two drains, and one gate)O One
of the virtues of a magnetoresistor over a ~all efect
device i~ that the magnetoresistor has ~nly two
@lectrod~s. In order to preserve this in our improved
magnetoresi~tor concept, we propose to use, in
conjunction with a magnetoresistor layout such as
described in Figure 2, a number of new ways to generate
accumulation or inversion layers without using
externally biased gate electrodes.
In a first embodiment, we make use of the
fact that the natural interface between InAs and air is
known to g2nerate an electron accumulation layer in
InAs. A similar effect may exist ln InSb, and the
technique may therefore be applicable to thin film
magnetoresistors made with this semiconductor material~
We would, however, not expect such devices to work as
well as InAs at very high temperatures. The very small
energy gap of InSb (see Table I ) would cause thermal
generation of carriers that would cause increased
conductivity in the InSb film adjacent to the
accumulation layer, making the conductivity of the
accumulation layer a relatively small fraction of the
total device conductivity. Thus, the benefits of the
accumulation layer would be lost at a lower temperature
in InSb than in the higher energy band gap InAs.~ We
experimentally grew a 2.3 mm thick epitaxial layer of
InAs on an insulating InP substrate using Metal Organic
Chemical Vapor Deposition ~MOCVD). Hall and
magnetoresistance measurements on the layer in the
temperature range of 350K to 0.5K, and in magnetic
fields up to 7 Tesla re~eal the presence of at least
- 1 31 3403
two "types" of carriers, in roughly equal
concentrations, but with very di~ferent mobilities ~by
a factor of 2 to 3). In retrospective view of the
afore-mentioned Weider publication, it is reasonahle to
5 assume that one of them is the accumulation layer
located near the air interface. We bui~t ~wo 2 ~
long, 5 mm wide magnetoresistor~ out of thi6 fil~ which
develop a very u~able magnetic field sensitiv~ty, while
~aintaining good temperature ~tability (~ee Figure~ 3,
1~ 4, and 5~. We believe it is possible to preserve this
sensitivity after covering the InAs surface with a
suitable encapsulating coating (e.g., an epoxy or other
dielectric).
In a second embodiment, a capping layer of
large-gap semiconductor such as GaAs, InP, AlSb, or
Inl yAlyAS can be grown on top of the narrow-gap active
layer semiconductor (typically InAs or Inl xGaxAs with
O<x<0.5, although a similar structure using InSk can be
conceived). In this capping layer, we put donor-type
impurities, such as Si, Te, Se, or S. These will
release an electron, which will end up in the layer
where it has minimum energy, i.e., the narrow-gap
semiconductor. This leaves a layer of positively
ionized donor-impurities in the large-gap capping
layer; but they are spatially removed from the
electrons in the active layer, and hence do not
significantly scatter them.
In a third embodiment, we propose to deposit
a layer of ~etal on top of the device active region
3~ with the purpose of creating a Schottky barrier. A
plot of the electron energy levels adjacent the
metal-semiconductor interface in this third embodiment
i5 shown in Figure 5. In referring to Figure 9, it
27
1313~03
can be seen that there will be a depletion of the top
region of the active narrow-gap semiconductor~ If the
acti~e layer is thin enough (1000-2000 Angstro~s), thi~
will confine electrons in the active layer toward6 the
~ubstrate, resultinq in electrical properties ~i~ilar
to tho~e of an accumulation layer. ~Letals that
generally form Schott~y barriers to III-V compound~,
such as Au or Al ~ay be useful, although we have not
adequately studied this ~tructure experimentally yet.
In a fourth embodiment, we propose to deposit
on the active layer of a narrow-gap semiconductor a
layer of large gap semiconductor, or of a dielectric
such as SiO2 or Si3N4, and on top o that a gate
electrode. An electron energy plot going through the
layers at the relevant interfaces i6 shown in Figure
10. The metal of the gate electrode in Figure 10 can
be chosen such as that it induces an accumulation
region near the semiconductor-dielectric interface, by
effect of the difference between the electron af~inity
; 20 in the semiconductor, and the work function in the
metal. Conversely, a different metal with larger work
function can be used to deplete the
semiconductor-dielectric interface and
electrostatically confine the electrons near the
substrate, much as in the third embodiment mentioned
above.
In a fifth embodiment, it is suggested that
the gate electrodes described in the fourth embodiment
be biased so as to generate accumulation layers in the
semiconductor under them. Such a concept is
schematically shown in Figures llA, llB, and llC. In
Figure 11A, it can be seen that if desired, one could
use one or more added contacts to separately bias the
28
1313403
29
gate electrodes. This would not ordinarily be
preferred but could be done. It would not be preferred
because one of the advantages to a magnetoresistor
resides in that it only has two contacts. We are only
showing it here for complsteness. On the other hand,
additional contacts are not actually necessary. The
gate electrodes can be electrically biased by an
~nternal resistor circuit, examples of which are shown
in Figures llB and llC.
1~ Reference is now specifically made to the
fifth type of embodiments shown in Figures 11~ and llC.
Since the gate leakage currents are very minimal, a
very high resistance (>lMW) circuit can be used for
biasing. Afi a special case in Figure llB, reslstor R1
can be made extremely large (open circult), and the
other resistors can all be made to have zero resistance
(short circuit). Thus, the full positive bias applied
to one external electrode ~relative to the other
external electrode) is applied to all gates in this
special case. An alternative is to connect the gates
over each semiconductor region with the shorting bar
between two other semiconductor regions located such
that the potential difference between the gate (i.e.,
the shorting bar) and the active region induces an
accumulation layer in the latter. This latter version
of internal biasing of the gate electrodes is shown
Figure llC. A special case of this configuration is
one in which each gate is connected to ~he adjace~t
~horting bar. In this configuration, each element
might be considered to be a MISFET transistor with gate
and drain shorted.
In the five preceding embodiments, the
accumulation layers were used only to enhance the
29
1313403
desirable transport properties of the se~iconductor in
the sensing area. The geometry of the magnetoresistor,
i.e., the length over width ratlo of eaGh active
element, was still defined by the use of ~etallic
shorting bars. The structure o Figure llB can be
extended to defi~e the geome~ry of the magne~ore~i~tive
element~ the~selves, by modulating the carrier density
and hence the conductivity, inside the semiconductor
active layer. This forms a sixth embodiment of thi~
inYention. An example of such a structure is
schematically shown in Figure 12. ~gain, an external
(integrated into the chip) resistance network is u~ed
in this sixth embodlment to bias a ~ucce~sion of gate
electrode6 to create a series of ~trongly accumulated
regions. These can be used instead o~ metallic
shorting bars to create geometrical magnetoresistance.
Such a structure could potentially be superior to one
in which metallic shorting bars are used, because
ield-insensitive contact resistances between the metal
and the semiconductor would be eli~inated.
Again, a special case can be considered for
this sixth embodiment, as was considered in the fifth
embodiment. In this special case of the sixth
embodiment, the resistor R1 of Figure 12 i6
open-circuited and the other resistors (R2, R3 ...) are
short circuited, so that the entire positive bias
applied to one external electrode is also applied to
each gate. Thus, the natural accumulation layer
normally present on an InAs surface would exist between
the gates as in Figure llA, but have a lower electron
density. If desired, the gates could be biased
negatively to eliminate the electron accumulation
layers between the gates, or even to generate a strong
1 31 3403
inversion layer with carriers of the opposite type
(holes). While the emphasis of this record of
invention is on devices with only two external leads,
the gates could be connected through a resi~tor network
to a third external lead, making thi~ version of the
magnetic field sensor externally controllable throu~h a
voltage bias externally supplied to the gate lead. As
hereinbefore indicated, a similar three terminal d~vice
could be made with the device shown in Figure llA.
In a seventh embodiment, a lightly p-type
film is grown (typically doped with Zn, Cd, Mg, se, or
C). In the case of InAs, the surface would, we
believe, still have a strongly degenerate electron
layer, but it would be an inversion layer. Such an
inversion laycr would have a large electron density
near the surface, and then a relatively thick
~typically 0.1 mm to 1 mm or more, depending on dopant
density) region of very low carrier density, similar to
the ~pace charge region of an n+/p ~unction. This
2~ might be advantageously be used to reduce the
conductivity of the film ad~acent to the electron
strong inversion layer. At very high device operating
temperatures, the intrinsic carrier density of narrow
energy gap semiconductors like InAs would tend to
defeat this strategy somewhat, and other, higher energy
gap semiconductor6 such as In1_xGaxAs might be
preferred (see Table I). InO 53GaO.47
case, since it can be lattice-matched to
semi-in~ulating InP substrates. This makes it easier
; 30 to grow such films with high crystalline quality.
The acceptor dopants mentioned above (i,.e.,
Zn, Cd, Mg, Be, and C) have small activation energies
in the III-V compounds of interest (see Table 1).
~ !
1 31 3403
32 ~
However, ~here are other acceptor dopants with
r~latively large activation energies, such ~e, in
InO 53Gao 47AS. ThiS means that relatively large
thermal energy is required to make the iron ionize and
S contribute a hole to conduction. Hob~ever~ the iron
will co~pen~ate a concentration of donor impuritie6
frequently present in the material so that they do not
contribute electrons to the conduction bandO Thus,
doping this material with iron, will make it tend to
have a high resistivity, except in the electron rich
accumulation layer. It would in this case be
desirable to grow a thin undoped InO 53Gao 47As layer
~e.g., 0.1 micrometer thick, after correcting for iron
diffusion efects) on top of the iron doped layer in
order to obtain the highest possible electron mobility
and density in the accumulation layer. It is
recognized however, that findin~ suitable dopants with
large activation energies may not be practical for
~maller band gap semiconductive materials.
Furthermore, the other embodiments discussed above
could also be used in conjunction with this one
advantageously to reduce the conductivity oE the film
adjacent to the high electron density region.
The emphasis of the above discussion has been
on electeon accumulation or inversion layers. Hole
accumulation or inversion layers could also be used.
However, èlectrons are usually preferred as current
carriers in magnetoresistors since they have higher
mobilities in the materials shown in Table }.
With reference now to the drawings, Figure 13
shows a typical prior art form of position ~ensor 10 in
which the magnetic circuit comprises an exciter portion
12 of ferromagnetic material made up of a succession of
1 31 3~03
teeth 12A spaced by gaps 12B and a stationary sensing
portion comprising the permanent magnet 14 supporting
on one surface the sensing element 16 and a flux guide
18 for providing a return path for the~ magnetic field.
As shown, the width of each tooth i~ ~bout equal to the
width of the ~ag~et and of the sensing element.
Optionally, a field concentrator (not hown) ~ay be
localized over the sensing element 16 in the form of a
thin layer of a high per~eability erromasnetic
material.
The exciter 12 typically is a plate with
spaced teeth along one edge and is adapted to ~ove
horizontally so that its teeth pass under the permanent
magnet 14 and the sensing element 16 in accordance with
the movement of a position that i6 being sensed.
Alternatively, the exciter may be a circular plate,
with teeth around its circumference interspersed with
610ts, that rotates about a fixed center for varying
the position of the teeth relative to the sensing
~0 element. The exciter is typically of a high
permeability ferromagnetic ~aterial, such as iron.
The permanent magnet is polarized vertically
in the plane of the paper, as indicated. The ~ensing
element typically is a magnetoresistor, a two terminal
element whose resistance increases with increasing
magnetic flux passing vertically through its bulk and
typically had nearly the same width as the magnet. The
sensing element 16 is as hereinbefore de~cribed.
The flux guide 18 also is advantageously of a
high permeability material, such as iron, and its
presence can increase the flux density thrnugh the
sensor by providing an efficient return path for the
flux passing through the exciter. To this end, the
1313403
34
center-to-center spacing of adjacent teeth o~ the
exciter and the center-to center spacing of the
magnetic path formed by the permanent magnet and the
flux guide are made essentially equal, as shown. Such a
flux guide, however, in fact adds little to the
sensitivity and so is unnecessary if adequate flux
density is provided, either by a magnet of sufficient
thickness vr choice of magnet ~aterial.
Typical di~ensions might be about one
millimeter both for the vertical thickness and for the
horizontal width of the magnet 12, similarly about one
~illimeter for the height and width of each tooth 12A,
about two millimeters for the width of a gap 12B, and
about one millimeter for the separation between a tooth
and the magnet in the position shown. The flux guide
18 typically would be of the same scale and would add
about another millimeter to the height of the magnet
path. The lateral dimension of the magnet normal to
the plane of the drawing typically is wide enough to
keep low any edge effects in the sensing element.
With a magnetic circuit of this kind, the
maximum sensitivity that is obtained tends to be less
than about five percent. Moreover, sensors are known
in which the stationary part of the magnetic circuit
includes a pair of magnetic sensing elements for use as
~eparate legs of a differential sensor. In such cases,
the two sensing elements typically are so spaced that
when one of the sensing elements is positioned directly
opposite one tooth, the other 6ensing element is
positioned directly opposite the center of the gap
between adjacent teeth to maximize the difference of
the outputs fro~ the time sensing element. Such sen~ors
34
1 31 3403
provide higher sensitivities but at the expense of
greater complexity.
In Figure 14, there is ~hown a po5ition
sensor 20 in accordance with a preferred e~bodiment of
the present invention. Its ~agnetic circuit includes
the exciter 12 that may be si~ilar to the exciter
included in the position sensor 10 shown in Figure 13
and so the same reference number is used. ~he
~tationary portion of the magnetic oircuit is ~hown in
greater detail in Figure 15. It includes a permanent
magnet 22, magnetized vertically as shown, and on its
bottom surface there is provided the sensing element 16
that may be similar to sensing element 16 in the
po~ition sensor 10 of Figure 13. In accordance with a
feature of the invention, intermediate between the
~ensing element 16 and the permanent magnet 22 there is
included a layer 24 of high permeability magnetic
material, such as iron, that covers the entire bottom
surface of the permanent magnet 22. Additionally, to
ensure that this layer does not electrically short the
sensing element 16, there is included an insulating
layer 26 intermediate between the sensing element 16
and the layer 24. If the layer 24 were of a
non-conducting material, such as high permeability
ferrite, the insulating layer 26 would become
unnecessary and so might be omitted.
In sensor 20, in accordance with a ~eature of
the invention, for increased sensitivity the width W of
the permanent magnet 22 is considerably wider than the
3a typical width of the prior art sensor 10 shown in
Figure 13. Advantageously, the width of the permanent
magnet is made to be the sum of the width of one tooth
and two gaps of the exciter, as shown, as seen in
- 1 31 3403
36
Figure 14, and so about one and one half times the
pitch of the teeth of the exciter. By way of contra~t,
in the sensor shown in Figure 13, the width of the
permanent magnet 14 essentially matche~ that of a tooth
12~ of the exciter. Moreover, the improvement in
~ensitivity provided by thi~ increase in magnet width
i8 ~urther augmented by the presence of the magnetic
layer lB.
For ~aximum magnetic ~ensitivity, in our
design it is another feature that the width of the
~ensing element is desirably as narrow as is
convenient. However, for electrical cireuit
efficiency, it is desirable that the element have a
6ufficiently high resistance, for example, at least 100
lS ohms, which imposes practical limits on how narrow the
element may be. Also the sensing element needs to be
wide enough to have adequate power dissipation
capabilities. Nevertheless, the sensing element
typically would be significantly narrower than the
tooth element unless the exciter design involved
unusually narrow teeth. As shown, the sensing element
16 i~ provided at opposite ends with electrodes 16A and
16B by means of which it may be connected into an
appropriate electrical circuit. These are typically
metallic platings deposited on the insulating layer 26.
The ferromagnetic layer illustratively can be about 0.1
millimeters thick and of a material such as low carbon
steel 1008. rrhe result is a geometry made up of a
series of planar layers that is easy to manufacture.
The sensing element 16 typically is cho~en in
aceordance with the particular application intended. A
magnetoresistor is preferred, for reason apparent ~rom
the foregoing. We prefer that the magne~ic field be
1 31 3~03
applied perpendicularly to the major Eace of the
sensing area in the sensing element.
Figures 16A and 16B illustrate the conditions
for ~axi~um and mini~um ~lu~ through the s~nsing
~lement 16 respectively for the po~ition sensor 20
6hown in Figure 14. As seen in Figure 16~, when the
~ensing element 16 is directly opposite a tooth 12A of
the exciter, the flux density represented by lines 30
through sensing element 16 is comparatively high.
However, when the exciter has moved so that the sensing
element 16 is opposite the center of a gap 12B between
teeth, the flux density through the sensing element 16
i5 comparatively less. Typically, the maximum 1ux
density may be 0.2 ~esla and the mlnimum flux 0.15
Tesla for a 2 millimeter thick MQ2 magnet. MQ2 magnet
material is an NdFeB alloy that has an energy product
between 13 -15 MGOe, is isotropic and 100 percent dense
and i5 a trademarked product of General Motors
Corporation.
The role of the ferromagnetic layer 24 makes
it easier for the flux to travel towards or away from
the sensing element 16, thus increasing the ~aximum
flux and decreasing the minimum flux that passes
through the sensing element, and thereby increasing the
sensitivity, which is dependent on the difference
between the maximum and minimum fluxes sensed.
In particular, the movement of the exciter
teeth little affects the total flux density but does
vary the spatial distribution of the flux density along
the width of the magnet, crea~ing sharp local flux
den ity variations ~hat can be sensed by a localized
sensing element, such as a ~agnetoresistor. The
ferromagnetic layer permits the flux density to be
1 31 3403
38
distributed along the magnet width in a way that
reflects the profile of the air gap between the
stationary portion of the magnetic circuit and the
exciter. Where this air gap is narrow, the ~lux
densi~y is high, where this gap is wide, the flux
den~ity is low. Since this air "gap" i~ narrowest
alongside a tooth of the exciter, the ~lux density
there will be highest and this den~ity peak will follow
the tooth ~ovement along the width of the magnet. In
particular, our tests have shown that the addition of
the thin ferromagnetic layer 24 in the manner described
can essentially double the sensitivity of a sensor with
an already optimum width magnet. The optimum thickness
of the ferromagnetic layer is determined by the maximum
flux density it is desired to guide without saturation.
Layers even as thin as five microns have proven to be
useful for a sensed ~aximum flux density of about
O.12T. For this flux density improvement tends to
level off when the thickness reaches about 25 microns.
The magnetic layer 24 can be provided simply
as a thin metallic foil attached to the sur~ace of the
permanent magnet 22 using conventional adhesives.
Alternatively, magnets manufactured by compressing
and/or sinterinq magnetic powder, such as MQ2
2S previously described, can produce a ferromagnetic layer
as an integral part of the permanent magnet. To this
end there is introduced into the die cavity an
appropriate amount of iron powder, before or after the
magnetic powder is introduced, and then the powders are
compressed together. Moreover, the planar geometry
~akes feasible batch-processing whereby hundred~ of
magnetoresistors may be deposited simultaneously on a
relatively thin unmagnetized permanent magnet wafer
1 31 3403
having a ferromagnetic layer and an insulating layer.
The wafer would then be cut into separate sensors, the
sensors packaged, and the permanent magnets magnetizedO
It appears that the increase in sen6itivity
is ~chieved at the expense of a lowering of the ~ean
flux density. If this is of eoncern Eor effective
modulation of the par~icular magnetoresistor being
used, the mean flux density can be increased to the
desired level with little effeet on the sensitivity by
increasing the thickness o the magne~ and/or the
magnet type, thereby maintaining the desired planarity
of the sensor and avoiding the need for a flux guicle to
improve flux density. ~owever, in special instances
where neither of these expedients is adequate, a flux
guide may be induced to improve the flux density
involving teeth further along the exciter.
In order to translate optimally the high
magnetic sensitivity of the magnetic circuit described
into high electrical sensitivity, the sensing element
needs to be appropriately positioned on the ~agnet.
Figure 17 shows a typical envelope of maximum
attainable sensitivity plotted against the normalized
distance d/W of the sensing element where d is the
distance from the midpoint of the magnet of width W.
It can be seen that the peak attainable sensitivity is
at the midpoint of the magnet (d=0) and at a minimum at
each end of the magnet (d/W~0.5). Accordinglyt the
optimu~ location of the sensing element is at the
midpoint of the magnet.
It is also important to have a proper width
for the sensing element, particularly when the element
i~ a magnetoresistor that produces an electrical output
1 31 3403
signal eorresponding to the average of the flux density
across its surface.
The flux density distribution along the
length of the magnetoresistor, however, ~an be assumed
to be constant. Thus, one is required to consider the
flux density or ~ensitivity distribution~ only along
the magnetore~istor width. Because-of that, the
e~fective electrical sensitivity will be directly
related to the average magnetic sensitivity as
lQ det~rmined by integrating the magnetic 6ensitivity
distribution given in Figure 18 over the
magnetoresistor width WMR. Figure 18 shows how the
sensitivity varies along the magnet width for the
alignment shown in Figures 16A and 16B. Looking at the
sensitivity distribution, one would tend to maximi~e
the electrical sensitivity by attempting to make WMR as
small as possible. Small size, however, would lower
the resistance and power dissipation capability of the
~agnetoresistor, and in turn lead to a lower output
signal. The selection of WMR has to be a compromise
which takes into account several conflicting
requirements such as the practical limitations on the
magnetoresistor length, the best possible sensitivity,
sufficiently large resistance and power dissipation,
the lowest possible magnetoresistor cost (smaller
magnetoresistors are generally less expensive), etc.
Considering previously available magnetoresistor
technology, the minimum practical value of WMR for the
exciter design that has been discussed appear~ to be
about O.3mm which amounts to d/W - O.033 and yields an
effective magnetic sensitivity SM of about 28 percent.
We do not know at this time how this is affected by the
i~proved magnetoresistor contemplated in this
1313403
41
invention. A 0. 6mm width would stilL provide
sensitivity of about 26 perrent. The width WMR in any
case desirably should be less than the width of the
teeth in ths usual design. The height of the ~ensing
ele~ent may be small, typically tens of ~icrons,
whereby the planarity of the associated fiurface is
little disturbed by its presence.
It is also found in our design that the ratio
of tooth width T to tooth pitch P also af$ect
sensitivity. It has been found that the sensitivity
tends to be maximum for T/P ratios of about 0.25 but to
remain relatively flat over the range between 0.17 and
0.37.
It is also found in our design that the tooth
pitch affects sen~itivity and in particular that
increasing the tooth pitch can appreciably increase the
~ensitivity. For example, for the design discussed, a
change in pitch from 3mm to 5mm can increase the
maximum sensitivity to a~out 5~ percent when conditions
are optimized. Since sensitivity decreases with
increasing air gap size between the exciter and the
magnet, increasing the tooth pitch offers a way to
compensate for larger air gap sizes and offers a
designer an ability to trade off between air gap width
~5 and tooth pitch.
In addition, it is found that the ~tationary
portion of a sensor of the kind des~ri~ed can be used
effectively with a broad ran~e of exciter wheel tooth
pitch sizes. This feature offers a considerable cost
6aving potential, for example, for applications such a~
ABS designs that employ widely differing tooth pitch
sizes. If a sensor of a particular stationary design
is intended to operate with wheels having different
41
1313403
42
tooth pitch sizes, the maynet width preferably should
be cho~en to optimize the sensor for the smalle~t tooth
pitch 6ize so that the lowest sensitivity, encountered
- when using the exciter wheel of ~mallest tooth pitch
~ize, will be as high as possible~ As previou~ly
di6cu6sed, the optimum magnet width is about 1.5 times
the tooth pitch size.
It is to be understood that the specific
e~bodi~0nts de~sibed are merely illustrative o the
general principles of the invention and various
modifications may be devi6ed without departing from the
spirit and scope of the invention. For example, it i6
feasible to reverse the role~ of the stationAry portion
and the movable portion of the position sensor.
Additionally, the various dimensions and materials
mentioned are merely illustrative o~ a typical de~i~n
and other designs could necessitate other dimensions
and materials.
42
