Pergamon
Vol. 92, No. 5, pp. 413-417, 1994 Elsevier Science Ltd Printed in Great Britain 0038-1098(94)$7.00+ .OO
Solid State Communications,
003X- 109X(94)005X2-6
ANGULAR
DISTRIBUTION
OF EMITTED
FOCUSING
EFFECT
S.Wakayama,
ELECTRONS
of Physics,
l-l
ELECTRON
AND LOW FIELD MAGNETORESISTANCE
I<.Tsukagoshi,
Department
FROM WIRE BY MAGNETIC
I<.Oto, S.Takaoka,
Faculty of Science,
Machikaneyama,
Toyonaka,
and K.Murase
Osaka
University
560, Japan
K.Gamo Department and
of Electrical
Research
Center
l-l
Engineering,
Faculty
for Extreme
Materials,
Machikaneyama,
( Received
Angular tron
distribution
focusing
dimensional with
distribution
In the two-dimensional
path
of electrons
ometry
PDEG
is comparable
has been
useful
properties(2-41. determined
6 is a tilted
vertical
direction
distribution
angle
has
magnetoresistance ter and 2DEG
The
a 2DEG
point
contacts
by varying
the width
which
is gradually
broadened
travel
adiabatically
forward,
the transversal formed
motion
to that
a collector
by using
the magnetic
or vertically
point
structure
connected
just
elec-
to a two
of the other
contact.
It is found
at the outlet
of the
device that
wire
the
rather
two
facing
at the outlet,
of longitudinal
f(B) of emitted
longitudinal
direction
(this
is called
Thus
collimation
effect)
The
confinement
tron
channels
represent
[9.
413
are 4.3
data
was formed
bridge
which
operated
Schottky
inset gates.
x
10”
cm-‘, patand metof elec-
potential
from
at 1.5 K by an AC
at 15 Hz with structure
of Fig. The
a mean
of Ti/Au
by electrostatic
3 PA. The schematic
in the upper
and
at the boundary
was measured
in-
dimen-
lithography,
by evaporation
potential
mobility Two
The device
electron-beam
were made
Resistance
current
high
mobility
illumination
by using
als.
tation
shift probes
ones by a clas-
and 7 pm, respectively.
gates
is shown
in the
would
experimental
on the
an electron
at 1.5 K under
was drawn
resistance
trans-
The
heterostructure.
density,
Schottky
the gates.
the angular
is collimated
electron
tern
of
positions
pair of oblique
the calculated
A is fabricated
of emit-
the electrons
wire is partly
device
6.4 x lo5 cm’/Vs
across
peak with
occurs.
with
of a GaAs/AlGaAs
free path
The
energy
effect
distribution
model.
low field
In a wire
the kinetic
motion. electrons
The
sional
other
the MEFE
collimation
billiard
from the
of the contact[5].
where
sical
terface
pairs
each
that
will also be compared
in
the angular
low field magnetoresistance.
field in a device
position
by the
we investigate and by using
It is expected
at 0 = O”,
boundary[5-7).
MEFE
transport
in a magnetic if the
trans-
the ballis-
direction
with
by using
the ge-
peak
D. c‘lrc1.ronic
10). In this letter,
free
(MEFE)
emitted
in a narrow
distribution
effect
determined
of a device
collector
)
low field magnetoresistance
size at low
regime,
MEFE
of the emission
been
mean
to understand
by electrons
against
f(0)
obliquely
A. nanostructures,
the electron
focusing
method
is apparently
electron
influences
electron
tic transport
where
is investigated
in a high
the
to a sample
a ballistic
strongly
The magnetic
on the
gas (2DEG)
heterostructure,
In such
of a sample
port(l].
depends
A. heterojunct.ions,
electron
GaAs/AlGaAs
temperatures.
wire and
University
560, Japan
1994 by A. Okiji
probes
and by using
of an emitter
strongly
with
Osaka
Science
the wire direct,ion.
Keywords:
mobility
gas,
20 April
electrons
in a device
electron
two pairs
than
of emitted
effect
Toyonaka,
of Engineering
of a sample
1. The shaded
separation
an exci-
AL
regions between
414
ANGULAR
DISTRIBUTION
OF EMITTED
ELECTRONS
Vol. 92, No. 5
4
.I
l.Opm
0’ -4’
I
1
I
I
-2
,
0
Gate Voltage ( V ) Fig.
1. Two terminal
resistance
sus applied
even voltage
Estimated
width
shown
at each
Schematic vertical
of an oblique
of both
voltage
and
collector show
peak
appears
Separation (V+,V-)
probe
with oblique
at oblique
and
probes
Fig.
2. Magnetic
gate
voltages.
dicate and
the
electron The
position
oblique
probes,
and
probes,
feet
respec-
with
and the collector with
oblique probes, while probes
. When
separation
negative
probes,
the
served.
of f45”
the upper
probes
are called
vertical
is comparable
to the
an emitter
electrons
the collector,
and
from
a peak
The electron
estimated
angles
diameter
emission
1 shows
the
oblique
wire
as a function
voltage
Vg=
-0.6
and
the wire
the
wire
gap
width
tance
-1.0,
The
and
Fig.
there
2.
resistance
The
voltage.
the gates
At gate
probes
magnetic
photograph.
1.0 pm) Since
the
fields.
is much
are dozens
larger of modes
for various magnetic
gate
electron
that
which
is
conducto its voltages
0.4 pm,
than
tiple
vertical
probes
peaks,
has
is observed
arrows
indicate
positions
in Fig.
re-
the Fermi
in the wire.
the
due
to mul-
The
The
of intensity
hand,
with
fields.
oblique
positions
oblique The right
probes.
against are
within
bound-
0.7 for vertical
magnetic
voltage
col-
of adjacent,
MEFE
with
and
specular
at the sample
are plotted
peak
by the gate
density
observed.
peak
the peaks
3.
and
to be about
in positive
=
emitter
by the ratio
the first
Jrst
of MEFE
are
of both
the
at B,,,,,
the
p of electrons
estimated
probes,
of
electron
reflections
been
region
is the
peaks
121. On the other
probes
age
n,
in the
is observed
between
is defined
probes[ll,
vertical
2, several
coefficient
which
At
where
boundary
reflection
The
gate
not
volt-
apparently
an experimental
er-
ror.
to the
to be proportional of the wire at gate
V are 1.0, 0.5 and
with
is observed
by the left arrow
In Fig.
affected
is depleted
V is equal
(about
be
of an
It is assumed -0.6
gates
-3.5
magnetoresistances in
terminal
widths
and
would
two
at Vg=
The wire width
wavelength,
rhown
-2.5
f(0) of MEFE.
under
into is ob-
position
is assumed
the estimated
spectively.
distribution
lector.
ary,
of vertical
are focused
of the gate
both
by a SEM wire
emitter
is constructed.
width
between
of the
width, Vg=
channel
a collector
peak
V, 2DEG
channel
determined
the
are called
of the magnetoresistance
by the shifted
Figure
and
peak
various
B > 0 in-
respectively.
AL is the separation
The lower probes
tilted
a cyclotron
between
is 7.5 pm.
of first
vertical
peak indicated
are connected
with
broken the first
-2fiJ2fm,/eAL,
which
effect
at B < 0 and
and
tively.
the emitter
focusing
arrows
(I+,I-)
for which
vertical
is
inset:
emitter Solid
trajectory
inset)
). Upper
between
is 7.5 pm.
the electron
wire ver-
of the wire.
(lower
( see text
view of the sample probes.
lines
side gates
of an oblique
gate
Magnetic Field ( kG )
To study tion
f(0)
electron which
boundary
motions
by
the
peak
position,
using
scattering
a classical within
141 are
conditions
and
between
MEFE
boundary[l3,
structure
emitter
relation
the
diffusive
sample
the
the
and
used
of device a collector
voltages
are
probes
is 1.0 pm,
the
focusing
ef-
( from
I+ to I-,
from
angular
billiard
the sample
taken
into
the
distance V+
between
to V-
in
and
at the The
separation
is 7.5 pm, the
model
account.
in the calculation A,
distribu-
we simulated
is based between
width
of wire
opposite
) is 20.0 pm,
on an of
probes and
an
ANGULAR
Vol. 92, No. 5
I ’
DISTRIBUTION
,
1
OF EMITTED ELECTRONS
I
1.5K
ns--4 -3x10” cme2
0 : oblique probes A : vertical probes 1 I
I
-3
-2
I
-1
Gate Voltage ( V )
Magnetic Field ( kG ) q--l-
Fig. 3. Gate voltage dependence of the position of the first peak. Both peak positions with oblique and vertical probes are not appreciably affected by the gate voltage.
density is 4.3 x 10” cm-*.
electron performed
(1) An electron angular
distribution
is ejected
was
from an emitter
with the
f(0).
(2) Each electron cyclotron
The calculation
as follows:
trajectory,
moves by the length
6 along the
where 6 is equal to 0.1 pm which is
small enough compared (3) After a cyclotron
motion by 6, the electron
elas-
tically scattered
in random directions
1 - exp(-6/l).
The mean free path (7 pm) which was
estimated
by the electron
is tentatively boundary,
density
and electron
mobility
a probability,
is either randomly
1 - p, or specularly
scattered
reflected
with
with a prob-
ability, p. The experimental
value 0.7 is adopted
specularity
enters any probes or moves
p. If the electron
for the
over ten times as long as the decay length, the calculation of the trajectory These
is stopped.
calculations
have been performed
5 x lo3 times for each magnetic the calculated ference
between
the number
4 shows
of electrons
are emitted
entering
for various function
The broken lines indicate
first peak Bfocus estimated electrons
Figure
0.5 Magnetic Field ( kG )
Fig. 4. Calculated MEFE with various f(0) at an emitter. Broken lines show the position of the first peak (B,,,,,) on the assumption that all electrons are emitted at a tilted angle 0 to vertical direction against the sample boundary. (a) If f(0) is cos(0) or cos(0 45”), the collimation effect is practically absent. The MEFE first peak position with f(6) = cos(0) is the almost the same with cos(0 - 45”). (b) If f(0) is COS’~(~) or COS’~(~- 45”), a strong collimation effect occurs. The ratio of the MEFE first peak position with f(0) = COS’~(~) to that with cosi”(B - 45”) is about 0.8.
result of the voltage difference from a dif-
probe V+ and the probe Vof f(0).
field.
more than
,cqS’O(y-457
0
with a probability,
used for the decay length I. At the sample
the electron
I-4”)
with a probe width 1 pm.
the positions
the forms
the same as those for f(0) = cos(0 - 45”) (Fig. 4(a)).
of the
a collimation
on the assumption
that all
in the vertical direction.
The po-
function shape f(0)
responding
contrast
distribution,
are almost
according
different
and
from those with
= COS’~(~- 45’) as shown in Fig. 4(b). to our observation
If
to a highly collimated
like f(0) = cosi”(0), the first peak position would be significantly
sition and shape of the first peak for f(B) = cos(Q), corto a usual classical
occurred
This is in
that the first peak positions
416
ANGULAR
DISTRZBU-I’ION
OF EMITTED
ELECTRONS
Or -90
( degree )
Angle of Emission
Fig. 5. The angular (a)The
angular
gate voltage the emitter
Vg=
-2.OV
which is determined
angular
distribution
of a point contact.
-2.0,
-3.OV
electrons
to measure
and vertical
resistance 0”).
are almost
the
10” from
-2.OV
of emitted
electrons,
magnetoresistance
with two pairs of an emitter other
across
2DEG
fields[8,15-171. electron
mobility
was measured
tation
cm-*,
Resistance
current
E2 is an oblique
probe
point
The
contacts.
1.4 x IO6 cm’/Vs
and 13 pm,
with a low exci-
probe,
between
the emitter
Cl and C2 are the emitter
and
( from El to Cl, or from E2 to C2 ) is 4 pm than a mean free path.
>f each channel
of the probes
minal resistances
and SEM
hgure captions m-&ted
an
shown in the inset of
is a vertical
distance
density,
at 0.4 K in the
and the collectors
which is much shorter
vestigated
electron
was measured
El
facing each
at very low magnetic
30 nA in a sample
The emitter
the collector
in the device B
and a mean free path
dark are 3.2 x IO”
distribution
and a collector
The two-dimensional
respectively.
Fig. 5.
the angular
electrons
of Fig. from
5.
was estimated photograph
The
probe
is fitted
of emitted
defining
the vertical
tges were independently
The width by two ter-
to measure
of
the electron
direction
relation
between
probe 0”).
electron
detected
Another
by Cl.
an emitter
5(a).
were fitted
becomes
grow around
collector
fluctuation,
was plot-
magnetoresisby cosN(8 + a) almost
slightly
6’ = 0”. appear
width,
the
are similar
the fine structures
wave interference
for instance,
sharper Since
effect by
due to uninten-
of gates and/or inhomogeneity
density[l8-201.
Similarly
the angular
of
distribu-
was in-
tion from theoblique
emitter
with a configuration
shown in the inset of Fig. 5(b).
corresponding
sym-
gate voltage of the emit-
the fine structures
shape
The
The volt-
applied to two pairs of side gates
The
and a collec-
The peaks appear
might be come from electron
of
(El)
of 6 in Fig.
to those at different
distribution
3”).
The electrons
magnetoresistance
the magnetoresistance
irregular
-
(E2).
of an angle 0 from the
L between
in 0. With increasing
the potential
probe
to the ground.
The observed
at which
are 0.4, 0.3,
and the collector(C1).
and the fine structures angles
-2.OV
would be directly
of the vertical
ter (El),
-1.5,
8 and B is 0 = arcsin[eB1/2fi,,!Z%J,
(12 < N 5 13,(~ metrically
estimated resistance
in the direction
ted as a function tances
direction.
from the oblique
where the separation tor is 4 pm.
The
by two terminal
-1.0,
were connected
vertical
tional
probe
Vg=
the emitter(E1)
from El
in
as shown in the inset of Fig. 5(a).
widths distribution
by cosN(6’ + p) {7 < N 5 9,p
electrons
emitted
as described
angular
The estimated The angular
the vertical
which is determined
two gates directly
probe(b).
Cl is 0.05 pm at the
The configuration
E2 at the gate voltage
distribution
same. In order to observe
of a point contact.
Inset:
about
Vg=
from the oblique
the angular
probes
and from the oblique
width of the collector
are 0.7, 0.6, 0.2 pm, respectively.
shifts
widths of the emitter
The magnetoresistance
The configuration
probe(a),
90
( degree )
(El).
of the emitted
The estimated
probes
from the vertical
{ 12 5 N < 13, cr -
probe
45
Angle of Emission
by two terminal
-1.0,
-1 .ov
I 0
-45
in 6’. The estimated
C2 is 0.05 pm at the gate voltage
0.2 pm, respectively.
with oblique
Vg=
from the vertical
widths of the collector
Inset:
emitted
symmetrically
probe is fitted by cosN(B +a)
distribution
(b)The
of electrons
appears
El at the gate voltage
of the vertical emission
distribution
distribution
Vol. 92, No. 5
probe (E2) was investigated
magnetoresistance
The
is shown in Fig. 5(b).
ANGULAR
Vol. 92, No. 5 The magnetoresistance
DISTRIBUTION
from the oblique probe was fitted
by cosN(O + p) {7 5 N 5 9,p N 3”}. ejected
from the emitter
enters the collector
If an electron
(E2) along the wire direction
(C2), the angular distribution
of the
by t,he replacement
low field magnetoresistance emitter
probe appeared
might shift by 45” from the vertical di-
). This suggests
rection.
Actually
the peak shifted by 6’ N 10” from the
electrons
vertical
direction.
lar distribution
The fact suggests
us that the angu-
is not decided only by the wire direction.
of the sharp distribution
COS’~(B- 45”) to COS’~(~). Moreover,
electrons
emitted
417
OF EMITTED ELECTRONS
outlet
measured
at 0 ‘v 10” (insted
strongly
depends
on the structure
wave character
of electrons
derstand
When the angles of emission
same. The angular distribution
are almost the
may depend on the struc-
ture just at the outlet of the wire. It should be noted that at the outlet of the oblique probes the escaping electron waves feels asymmetric In conclusion, tion of emitted MEFE
electrons
angular distribu-
in two kinds of devices by using
are almost
the same.
MEFE with a classical sive scattering.
The position
first peak for both oblique billiard
The calculated
is not shifted by the replacement
and vertical
We have calculated model considering position
also diffu-
of the first peak
of the broad distribution
f(0) from cos(8) to cos(t) - 45”), but it is shifted by 20%
just at the
of escaping electron
The to un-
at the outlet.
is lowered, the observation
of the peak shift becomes difficult with the oblique probe in MEFE experiments. The authors
would like to thank O.Matsuda
ful advice about computer
of the wire.
and low field magnetoresistance.
of the MEFE probes
termination
we have investigated
the behavior
of emitted
might be important
with oblique probes
probes
of 0 = 45”
than the wire direction.
This is the reason why the MEFE first peak positions and vertical
with the oblique
that the angular distribution
of the wire rather
f(0) from
the peak of the
NEC Co.,Ltd. try Co.,Ltd.
for providing
Research
supported
The authors
for supporting
for Young Scientists. edge the support
of
Electric Indus-
high mobility heterostructure. by a Grant-in-Aid
(B) from the Ministry
Science and Culture. knowledge
and K.Kasahara
and H.Okada of Sumitomo
The work is partially Scientific
simulation,
for use-
for
of Education,
(K.T. and K.O.) ac-
by JSPS Research
One of the authors
Fellowships
(S.T.) acknowl-
by Casio Science Promotion
and Shi-
mazu Science Foundation.
References [l] C. J. B. Ford, S. Washburn, M. Biittiker, C. M. Knoedler, a.nd J. M. Hong, Phys. Rev. Lett. 62,2724 (1989). [2] H. van Houten, C. W. J. Beenakker, J. G. Williamson, M. E. I. Broekaart, P. H. M. van Loosdrecht, B. J. van Wees, J. E. Mooij, C. T. Foxon, and J. J. Harris, Phys. Rev. B39, 8556 (1989). [3] H. van Houten, B. J. van Wees, J. E. Mooij, C. W. J. Beenakker, J. G. Williamson, and C. T. Foxon, Europhys. Lett. 5, 721 (1988). [4] C. W. J. Beena.kker, H. van Houten, and B. J. van Wees, Europhys. Lett. 7, 359 (1988). [5] T. Sakamoto, Y. Takagaki, S. Takaoka, K. Gamo, K. Murase, and S. Namba, Jpn. J. Appl. Phys. 30, L1186 (1991). [6] Y. Hirayama, S.Tarucha, T. Saku, and Y. Horikoshi, Phys. Rev. B44, 3440 (1991). [7] Y. Hirayama, T. Saku, S.Tarucha, and Y. Horikoshi, Appl. Phys. Lett. 58, 2672 (1991). [8] L. W. Molenkamp, A. A. M. Staring, C. W. J. Beenakker, R. Eppenga, C. E. Timmering, J. G. Williamson, C. J. P. M. Harmans, and C. T. Foxon, Phys. Rev. B41, 1274 (1990). (91 H. U. Baranger, D. P. DiVincenzo, R. A. Jalabert, and A. D. Stone, Phys. Rev. B44, 10637 (1991).
[lo] C. W. J. Beenakker,‘and H. van Houten, Phys. Rev. B39, 10445 (1989). [ll] K. Tsukagoshi, S. Takaoka, I<. Murase, K. Gamo, and S. Namba, Appl. Phys. Lett. 62, 1609 (1993) [12] F. Wakaya, Y. Takagaki, S. Takaoka, K. Murase, Y. Yuba, K. Gamo, and S. Namba, Superlatt. Microstruct. 11,273 (1992). [13] C. W. J. Beenakker, and H. van Houten, Phy. Rev. Lett. 63, 1857 (1989). [14] R. J. Blaikie, K. Nakazato, J. R. A. Cleaver, and H. Ahmed, Phys. Rev. B46, 9796 (1992). [15] K. L. Shepard, M. L. Roukes, and B. P. van der Gaag, Phys. Rev. Lett. 68, 2660 (1992). [IS] K. L. Shepard, M. L. Roukes, and B. P. van der Gaag, Phys. Rev. B46, 9648(1992). [17] M. Okada, M. Saito, M. Takatsu, K. Kosemura, T. Nagata, H. Ishiwari, and N. Yokoyama, Superlatt. Microstruct. 10,493 (1991). [18] J. A. Nixon, J. H. Davies, and H. U. Baranger, Phys. Rev. B43, 12638 (1991). [19] T. Ando, Phys. Rev. B44, 8017 (1991). [20] D. R. S. Cumming, H. Ahmed, and T. J. Thornton, Appl. Phys. Lett. 60, 2755 (1992).