Twodimensional electron gas magnetic field sensors

J. Heremans, D. L. Partin, D. T. Morelli, B. K. Fuller, and C. M. Thrush

Citation: Applied Physics Letters 57, 291 (1990); doi: 10.1063/1.103717

View online: http://dx.doi.org/10.1063/1.103717
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/57/3?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Density Response of an Interacting TwoDimensional Electron Gas under a Uniform Static Magnetic Field and an
Intense Laser Field
AIP Conf. Proc. 832, 526 (2006); 10.1063/1.2204554

Field electron emission from two-dimensional electron gas

J. Vac. Sci. Technol. B 23, 657 (2005); 10.1116/1.1886820

Characteristic energy losses of electrons in a two-dimensional electron gas in a magnetic field

Low Temp. Phys. 25, 986 (1999); 10.1063/1.593852

The high-frequency conductivity tensor of a two-dimensional electron gas with electron impurity states in a
magnetic field
Low Temp. Phys. 23, 55 (1997); 10.1063/1.593336

Two-dimensional electron gas in a magnetic field and point potentials

Low Temp. Phys. 23, 15 (1997); 10.1063/1.593330

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.248.155.225 On: Sat, 22 Nov 2014 08:08:06
Two-dimensional electron gas magnetic field sensors
J. Heremans, D. L. Partin, D. T. Morelli, B. K. Fuller, and C. M. Thrush
General Motors Research Laboratories, 30500 l'Wound Road, Warren, Michigan 48090

(Received 26 February 1990; accepted for publication 7 May 1990)

We describe the use of accumulation layers of electron charge in applications as
magnetoresistive devices. We consider two such systems: an InGaAs/InP heterostructure in
which we identify a two-dimensional electron gas from the observation of the quantum Hall
effect, and InAs films, in which a strong surface accumulation of charge is inferred from depth
profiling studies of the galvanomagnetic coefficients. Magnetoresistive devices fabricated from
these materials exhibit outstanding field sensitivity and temperature stability due to the
existence of electrons of relatively high density and mobility in the accumulation regions. We
also model the magnetosensitivity of our devices.

Electronic devices capable of sensing magnetic field
strength are becoming increasingly popular and commer-
cially important. The simplest such device is the magnetore-
sistor, When used in conjunction with a permanent magnet,
for instance, a magneto resistor can be used as a potentio-
meter and as a very accurate position sensor. Hall effect de-
vices give a voltage output that is more linear with magnetic
field, but the signal level is much lower than that which can
be obtained with magnetoresistors. Therefore, Hall sensors
typically require much amplification. Magnetoresistors are
not new; engineering applications were considered even 20
years ago. !.2 In this early work it was quickly realized that
the most sensitive magnetic field sensors are those fabricated
from high mobility semiconductors, because the relative
change in resistance is proportional to the square of the mo-
bility. The semiconductor with by far the highest mobility
(at room temperature) is lnSb, and this was naturally the
most popular material for commercial applications over not
too broad a temperature range.
If, on the other hand, one envisions using these sensors
at more extreme temperatures, such as those experienced
near an automobile engine, one requires a material whose
resistivity and sensitivity to magnetic field do not depend
greatly on temperature. This in principle can be achieved by
using semiconductors with large extrinsic carrier densities
or large energy gaps. Unfortunately, in bulk semiconduc-
tors, the general trend is that electron mobility decreases
with increasing electron density; indeed, magnetoresistors
made from InSb are useful only over a rather limited tem-
perature range around and below room temperature, be-
cause their resistance quickly becomes impractically low
above 125°C. The carrier mobility-carrier density trend ob-
served in most bulk semiconductors would seem to preclude
the existence of a semiconductor magnetoresistor with both
high field sensitivity and low temperature sensitivity.
In this letter we describe the use of a two-dimensional
electron gas (2DEG) in magnetoresistive devices. The chief
advantage of this approach is that the 2DEG contains elec-
trons of both high mobility and high density. This leads to
devices which are capable of maintaining a high level of sen-
sitivity over wide temperature ranges.
We have grown two types of III-V semiconductor de-
vices: an InGaAs/InP heterostruture, which contains a high
mobility 2DEG at the InP interface,3 and an InAs film also
grown on an insulating InP substrate. The InAs film con-
tains a space-charge accumulation layer at the air interface.
The growth of these structures is by low-pressure metalor-
ganic chemical vapor deposition (MOCVD). Details ofthe
growth of lattice-matched In O. 53 Ga0 .47 As/InP heterostruc-
tures have been given elsewhere. 4 For the case of InAs,
growth was performed in the same reactor using arsine and
ethyldimethylindium as reactive source compounds.
We have made two types of samples: Hall bars and mag-
netoresistor devices. The Hall bars are mesa-etched six-point
bridge patterns on which one can measure the galvanomag-
netic coefficients Pxx and Pxy' For each material, p xx and p xy
were measured from 0.3 to 350 K in fields from 0 to 2 T (7 T
at 0.3 K). InAs films have room-temperature mobilities
ranging from 15000 to 21 000 cm 2 /V s. Upon cooling, the
mobility of the InAs samples increases to 50 000 cm 2 IV s at
100 K, then it decreases again. The InO. 53 G3t1.47As films have
a mobility of 11 000 cm 2 /V s at room temperature, which
increases to 110 000 below 20 K.
Magnetoresistive field sensing elements are typically
rectangles defined in the film by a mesa etch. The current is
passed along the short dimension of the rectangle to give rise
to the "geometrical magnetoresistance effect". I Multiele-
ment devices were fabricated using photolithography. They
consisted of 12 single-element rectangular devices in series.
The length to width ratio of each rectangle was 1:2. These
were produced by first etching a dog-bone shaped mesa
through the whole film and into the substrate. Gold ( ~ 1000
A thick) was then evaporated to provide both contact pads
and contact areas between the individual InAs or
1110 . 53 Gllo,47As elements. Active InAs or InC. 53 Ga0,47As rec-
tangles are defined by the mesa on their short sides and by
the gold stripes on their long sides. The resistance of each
magnetoresistor was measured from - 60 to 280°C (213 to
553 K) in fields from 0 to 1.2 T.
Evidence of accumulation of electrical charge in a thin
layer and the existence of a 2DEG in our InGaAs/InP het-
erostructure is provided by the observation of the quantum
Hall effect (QHE), as shown in Fig. 1. We observe plateaus
in the transverse resistance R~y and zeros in the longitudinal
resistance R xx (the current is in the x direction and the mag-
netic field in the z direction). At 0.3 K, the areal carrier
density, as determined by QHE, is 3.MX 1011 cm- 2 ; from
the ordinary Hall effect at low fields we measure

291 AppL Phys. Lett. 57 (3), 16 July 1990 0003-6951/90/290291-03$02.00 @ 1990 American Institute of Physics 291

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.248.155.225 On: Sat, 22 Nov 2014 08:08:06

5
4 em -3, even though the density averaged over the entire film
II
6 is on the order of 10 16 cm - 3. The mobility of the electrons in
oL------''-------l'-------''----'.L.l---'------l=----
o :1
Field EI
FIG. 1. Longitudinal and transverse resistance measured at 0.3 K on an
InGaAs/lnP heterostructure. Numbers indicate quantum Hall indices.
Nlotal 3.9X lOll cm- 2 . Thus, over 90% of the conduction
= devices as a function of applied field and temperature. We
electrons are confined to the 2DEG. The room-temperature
mobility and density of this sample are 11 000 cm 2 V-I S-1
and 4.6X 1011 cm- 2 , respectively. If one assumes that the
density of the 2DEG is temperature independent, then it can
be concluded that the galvanomagnetic properties of this
structure are determined primarily by the electrons in the
2DEG.
Turning now to the InAs films, evidence for the exis-
tence of accumulation of charge at the air interface is pro-
vided by a depth profiling study of the sheet resistance and
Hall voltage using the Van der Pauw technique. One of the
samples was etched away in steps using O. 5 vol % bromine in
methanol, with van der Pauw data taken at room tempera-
ture between each step. From these data the carrier density
and mobility were computed, and the results are shown in
Fig. 2. We see that while the mobility decreases with increas-
ing etched depth into the film, the areal density remains
nearly constant. These results are consistent with a thin elec-
tron accumulation layer of fairly high mobility at the sur-
face. Such a surface accumulation of charge at the InAs sur-
2 face has been the subject of some study over the years, 5 and it
has been shown that its thickness is -100 A. This means
that the local electron density at the surface is -4.5X 10 18
the InAs film studied in Fig. 2 is about 15 500 cm 2 ;V s; the
films used to make the device studied in Figs. 3 and 4 have a
room-temperature mobility on the order of 21 000 cm 2 ;V s.
This can be compared with the mobility of bulk samples
doped from 1 to 5 X 10 18 cm- 3 ,6 which is 10 000 cm 2 ;V s.
Thus there exists a rather high density electron layer offairly
high mobility at the InAs surface. The measurements of Pxv
4 6
at 20 K made on an unetched film give an areal electron
density of3.5 X 10 12 cm 2; at 300 K this density is 4.5 X 10 12
cm -3. These mobility and density data bode well for tem-
perature-stable magneto resistor operation.
Magnetoresistors were then made from both materials.
In Fig. 3 we show contour plots of the resistance of these
see that both types of magnetoresistors are capable of main-
taining a relatively flat response with respect to temperature.
In order to model the magnetoresistance of a rectangle
made from a homogeneous semiconductor, we must solve
the two-dimensional boundary value problem involving the
flow of current in a solid when a Hall effect is present. This
has been done by Wick. 7 In the case of our InAs and InGaAs
films, in which the electron density depends on depth into
the film, the two-dimensional modef no longer strictly ap-
plies, and must be generalized. Wick's mode1 7 involves a
conformal mapping of the rectangular geometrical device on

"--'10.00 25000
N
I
E 9.00 22500 ,--..
U U
13.00 Q)
20000
fIl
N
7.00 17500 >
0 c ...........
'-'
~ 6.00 .:'_-s--w 15000 N E

5.00 #< -0 >I< 12500 U

~ 'L
* /'b*
,. o * * '-"'

Vi 4.00 0
/
10000
~
Z
W / :J
0 J.oo I 7500
ill
I
0
-' 2.00 I 5000
«
w I
:::2'
0::: 1.00 ~ 2500
«
0.00 a
0 2 J 4 5
THICKNESS (~m)

FIG. 2. Depth profiling study of an InAs film: (left, *) areal density vs FIG. 3. Resistance of magnctoresistors made from (a) an InGaAs/lnP het-
depth etched (right, D) mobility vs depth etched. The initial film thickness erostructurc' and (b) an InAs film, as functions of magnetic field and tem-
was 4.5 p,m. perature.

292 Appl. Phys. Lett., Vol. 57, No.3, 16 July 1990 Heremans et al. 292

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.248.155.225 On: Sat, 22 Nov 2014 08:08:06
 2000 (here, L: W = 1:2) of a rectangular magnetoresistor element,
,
0..., __ T=-43 C we can calculate its magnetosensitivity. Figure 4 shows a
,.-.,
• ...r;;i? T=59 C comparison of the measured resistance of a 12 element mag-
E 1500 o
r.
--;;;;;::T=76 C netoresistor with that calculated using the above theory
S from the geometry and the values ofPxx andpxy, in the tem-
"-J
~ 1000 perature range where the data overlap. There are no adjusta-
~ ble parameters in this model. The 7% difference between the
til
[f) calculation and the experimental data can be attributed to
"-J
Ct: 500 spatial variations of the mobility over the wafer, and uncer-
tainties in the processing of the devices.
In conclusion, we have shown that it is possible to make
O+-~~~'-T-~~-'~~~-'
0.0 0.5 1.0 1.5 sensitive and stable magnetoresistive sensors out ofmateriuls
FIELD (T) possessing a 2DEG offairly high mobility and density, and
that one can accurately calculate the sensitivity of such de-
FIG. 4. Comparison of calculated (full line) and measured (data points)
resistance of an InAs device at variOlls temperatures from ... 43 to 76°C. VIces.
We acknowledge the help ofK. Baucom, and many use-
ful discussions with D. Baker and T. Schroeder.

the real axis of a {; plane using a Schwarz-Christoffel trans-

formation. This step is unaffected by our thickness-depen-
dent electron density. A second step, however, involves an
additional mapping from this (; plane onto a parallelogram in
current-voltage space. In Wick's theory, the angle ofthe par-
allelogram is the Hall angle, f) = arctan [R lJB / P ], where
RH is the Hall coefficient and p is the zero-field resistivity. In
our case, at each field value B, the average Hall angle inte-
grated over the film thickness is given by p = arctan
[Pxy/Pxx ]. Therefore, by measuring the Hall resistivity Pxy
and longitudinal resi.stivity Pxx on a long, thin bar of the
semiconductor film, and knowing the length-to-width ratio
IH. Weiss, Structure and Applications of Galvanomagnetic Devices (Perga-
mon, New York, 1969).
25. Kataoka. Circulars of the Electrotechnical Laboratory, No. 182, Tokyo
(1974).
'L D. Zhu. P. E. Sulewski, K. T. Chan, K. Muro, J. M. Ballantyne, and A.
J. Sievers, J. Appl. Phys. 58, 3145 (1985).
4J. Heremans, D. T. Morelli, B. K. Fuller, C. M. Thrush, D. L. Partin, and
K. C. Baucom (unpublished).
5D. C. Tsui, Phys. Rev. Lett. 24, 303 (1970); Phys. Rev. 4, 4438 (1971);
ibid. 12, 5739 (1975).
"T. C. Harman, H. L Goering, and A. C Beer, Phys. Rev. 104, 1562
(1956).
7R. F. Wick, J. App\. Phys. 25, 741 (1954).

293 Appl. Phys. Lett., Vol. 57, No.3, 16 July 1990 Heremans et al. 293
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.248.155.225 On: Sat, 22 Nov 2014 08:08:06