You are on page 1of 7

Solid-State Electronics Vol. 37, Nos 4-6, pp.

793-799, 1994

~ ) Pergamon

Copyright 1994ElsevierScienceLtd
Printed in Great Britain.All rights reserved
0038-1101/94 $6.00+0.00

SINGLE-ELECTRON T U N N E L I N G A N D COULOMB
CHARGING EFFECTS IN ULTRASMALL
DOUBLE-BARRIER HETEROSTRUCTURES
M. TEWORDTt,V. J. LAW,J. T. NICHOLLS,L. MARTiN-MORENO~,D. A. RITCHIE,M. J. KELLY,
M. PEPPER,J. E. F. FROST,R. NEWBURYand G. A. C. JONES
Cavendish Laboratory, University of Cambridge, Cambridge CB3 0HE, England
Abstract--An extensivestudy of charge transport through submicron-diameterdouble barrier heterostructure diodes is reported. The occupation of the quantum well with single electrons, starting from zero, is
observed in the form of sharp steps in the current-voltage characteristics. The magnitude of the current
steps can be controlled by changing the barrier thicknesses and thus their transparency for tunneling
electrons. The plateau width of the current steps is related to the energies of the electron states in the
quantum well that are affected by the lateral quantum confinement, and by Coulomb charging effects.
Diameter dependent studies of the tunneling current suggest that the lateral quantum confinement can
result from the surface depletion potential, potential fluctuations, or single impurities. High magnetic field
studies confirm this conclusion. The contribution of the Coulomb charging energy is investigated by using
an asymmetric double barrier profile. It is shown that tunneling through submicron-diameter double
barrier heterostructures provides valuable insight into the electronic properties of quantum boxes
containing a few electrons.

1. INTRODUCTION
Advanced technology has now made it possible to
fabricate semiconductor nanostructures in which
electrons are confined in small boxes that are a few
hundred angstroms in size[l]. These tiny artificially
crafted electron systems are often referred to as
"quantum dots" or "Coulomb islands". The interest
in semiconductor quantum dots lies in their characteristic electronic properties. Firstly, when the size of
the quantum dot is smaller than the Fermi wavelength and the electron mean free path, three-dimensional quantum confinement of the electronic states
can be observed in vertical[2] and lateral[3] tunneling
spectroscopy, and in optical spectral4]. Secondly, the
small size of the quantum dots implies that their
electrostatic charging energy is quite significant. In
analogy to classical electrostatics of a capacitor C, the
Coulomb charging energy of a quantum dot is
Ec(N) ~ N 2 e 2 / 2 C , where N is the number of electrons in the dot, and e is the electron charge.
Coulomb charging effects were first observed in
small metal grains[5], but the quantum confinement
energies are too small to affect the classical charging
tPresent address: Max-Planck-lnstitut for Festk6rperforschung, Heisenbergstr. 1. 70569 Stuttgart 80,
Germany.
:l:Present address: lnstituto de Ciencia de Materiales, Consejo Superior de lnvestigacionesCientificas, Universidad
Autonoma de Madrid, 28049 Madrid, Spain.
Present address: Department of Physics, University of
Surrey, Guildford. Surrey GU2 5XH, England.
fAlso at the Toshiba Cambridge Research Centre, 260
Cambridge Science Park, Cambridge CIM4WE, England.
ssr 37:4-6--a

picture. In comparison, in semiconductors, where


both the electron effective mass and the Fermi energy
are small, the confinement energies and the Coulomb
charging energies[6] are comparable and both can be
observed simultaneously[7]. This was first demonstrated in conductance studies of quantum dots
formed in a high mobility two-dimensional electron
gas (2DEG)18].
A particularly interesting case occurs when the
quantum dot is occupied by only a few electrons. This
is difficult to achieve in systems in which the 2DEG
is laterally confined, where isolated Coulomb islands
contain typically 100 electrons[9]. Recently, several
groups have successfully observed the incremental
charging of a quantum dot electron by electron,
starting from zero [2,10-16]. They studied transport
through ultrasmall diameter resonant tunneling
diodes (RTDs), where a GaAs quantum dot is formed
between the two AIGaAs barriers. If the vertical
confinement is strong enough, the lowest energy state
in the dot lies above the Fermi energy in the emitter
at zero bias, and no electrons occupy the dot. At low
temperatures, incremental charging of the dot can be
observed in form of sharp current steps in the tunneling current[10,12]. The electrons are then stored
dynamically in the quantum dot, i.e. the system is not
in equilibrium. Recent theoretical models for
transport through submicron scale RTDs consider
lateral quantum confinement[l7], Coulomb charging
effects[6], and both contributions at the same
time[7,181.
Ashoori et al.[19] measured the capacitance signal
of a quantum dot as it is charged up with single
electrons, again starting from zero. in their device, no

793

794

M. TEWORDT et al.
Table 1. Summary of the double-barrier heterostructures

Structure
(l)
(2)
(3)
(4)

Barrier width
(nm)=
10 (7)
7.1 (7.1)

Well width
(nm)
14
5.5

Spacer
(nm)
7
5

Dopingb
(10~scm 3)
0.02
~ 1.5

5.0 (5.25)
4.3 (4.3)

5.5
6.1

5
5

~ 1.5
~ 1.5

'Top (bottom) barrier.


bDoping of contacts close to the spacer layers.

electrons tunnel through the dot, and therefore the


discrete electron states can be probed under conditions closer to equilibrium.
In this paper, we present an extensive study of
tunneling through nanometer scale A I G a A s - G a A s
double-barrier quantum dots. The observation of
single-electron charging is described under various
experimental conditions that affect the electron transmission and the energetics of the quantum dot.
Parameters studied include diameter, barrier thickness and asymmetry, magnetic field, and temperature.
It will be shown that this system is ideally suited to
study quantum dots in the limit of very few electrons.

2. SAMPLES AND EXPERIMENT

The diodes studied here were processed from four


A I G a A s - G a A s double barrier heterostructures
grown by molecular beam epitaxy on an n +-type
(100) G a A s substrate. The undoped active layers of
structure 1 comprise a G a A s quantum well (thickness
w = 14nm) sandwiched between two A10.33Ga067As
barriers (top barrier thickness b I = l0 nm, substrate
side barrier thickness b2 = 7 nm), and G a A s spacer
layers (b0 = 7 nm). The top and bottom G a A s contacts, starting from the spacers, consist of a 350 nm
layer Si-doped to 2 x 10 ~6cm -3, a 28 nm thick layer
with the doping graded
from 2 X 1016 tO
1.4 X 1018 cm -3, and a 350 nm thick layer, Si-doped
to 1.4 x 1018 cm 3. The other three heterostructures
2-4 are symmetric and have no graded doping profile
in the contacts. The main design parameters of
heterostructures 1-4 are summarized in Table 1. Free
standing single R T D s with diameters between 0. i and
10/am were fabricated from the heterostructures by
employing electron-beam lithography and CH4-H2
metalorganic reactive ion etching[10].
All samples were measured at T = 4.2 K or in a
dilution refrigerator operating at base temperature of
the mixing chamber (T = 20 mK). The diodes have to
be measured at low temperatures, because the discrete
electronic states can only be observed when the
energy spacing of the quantum dot states is smaller
than the thermal broadening k 8 T of the Fermi energy
in the contacts. This condition is typically fulfilled at
temperatures around T ~ I K.
Figure 1 shows the current-voltage ( l - V ) characteristics of an asymmetric double barrier diode (structure 1) with large diameter a = 10pro. Resonance
peaks from tunneling through the first 2D subband

are observed in both bias polarities. The asymmetry


of the I - V results from considerable charging of the
quantum well only in forward bias. In forward bias,
the rate of tunneling into the well is higher than that
of tunneling out, leading to much higher electron
charging of the quantum well than in reverse bias[20].
Therefore, the current peak in forward bias is larger
than in reverse bias. Charging also leads to more
inefficient biasing of the quantum well, resulting in a
higher peak voltage in forward bias as compared to
reverse bias[20].
3. TUNNELING IN SMALL DIAMETER DIODES

The number of electrons charging up the quantum


well increases with bias and is proportional to the
diode area. If the diode diameter is made very small,
the incremental charging with single electrons can be
resolved. Every electron that occupies the well contributes to the current by a discrete amount AI = e/z,
where z is the transit time. The widths of the resulting
current plateaus in bias depend on the spacing in
energy of the electronic states in the well. The two
main contributions to the energy spacing are the
lateral confinement energy and the electrostatic
charging energy.
In Fig. 2(a), a typical small diameter R T D is

4 |

<

2
o

T = 50mK

dia = lOp.m

-4
-160

-80

80

160

240

Bias (mV)
Fig. I. Current-voltage (I-V) characteristics of a large
diameter asymmetric double-barrier diode from strucutre 1.
The band structure graphs in the inset illustrate the charging
only under forward bias.

795

Single-electron tunneling and Coulomb charging

100-1000nm

box
Eo
-eV

GaAs
AIGaASGaAs

~
O~
="

( C):)

~ bo]

b2 w

electron
accumulation
~

b 1 b0

Fig. 2. (a) Schematic sketch of the submicron-diameter double-barrier diode. (b) Band diagram of
structure 1 and reverse and (c) forward bias.

depicted schematically. Electrons tunnel from the


contact regions through the AIGaAs barrier into the
quantum well. The electron motion is laterally
confined between the diode sidewalls in both contacts
and in the quantum well. Thus, one-dimensional
subbands are formed in the contacts and discrete
zero-dimensional (0D) states are obtained in the well
("quantum dot"). The 0D states can be probed by
measuring the tunnel current upon applying a bias
between the contacts.
The discrete confinement energy in the well
is Ez+E,.,,, where E z is the energy from the
vertical quantum confinement between the bartiers. We tentatively assume that discrete states
E,.m result from the lateral surface depletion
potential which can be modeled using a two-dimensional, cylindrically symmetric parabolic potential
V (x, y) = (l ~2)m'co o(x 2 + y 2), with quantum energies E,.,,=(2n + l m t + 1)hco0 with the radial
n = 0, 1. . . . and the azimuthal m = 0,-I-1,_2 . . . .
quantum numbers. The lateral confinement energies
can also be modeled using the hard-wall potential
E , j . = ( n 2 + m ' - ) n 2 h 2 / 2 m * a 2, where a is the wall
separation.
The lateral confinement potential is not necessarily
from the surface depletion potential. Particularly
in systems containing only a few electrons and in
the presence of impurities, the confinement can
result from potential fluctuations[16] or from single
impurities[ 13].
The Coulomb charging energy can be modeled
using a single effective capacitance C. This is the usual
approach in studies of resonant tunneling in quantum
dots[6,7], although recently, it has been noted that
electron-electron interactions within the dot can be
important[23]. The resonant energies of the quantum
dot states can be written as:
E ( n , m , N ) h o ~ = ( e 2 / C ) ( N - I/2)+ Ez + E,,.,,,. (!)

The Coulomb charging energy of the quantum dot


containing N electrons gives the t e r m (e2/C)
( N - 1/2)[7]. It can be estimated from the effective
dot capacitance C by considering the barriers with
thicknesses bj and b2 as dielectrics:
C ~ (Eeod ~o.d/4n) (b ~ i + b ~ l ).

(2)

The current for a single electron tunneling through


the quantum well is given by[10,12]:
AI = e (F I F2)/(F m+ F2)

(3)

where F, and F 2 are the tunnel rates through the first


and second barrier, respectively. Every time a discrete
energy state in the dot falls below Fermi energy in the
emitter, a new current step A! will occur, leading to
a current-voltage staircase. The tunnel rates F, and
F 2 can be estimated in the WKB approximation by:

= ( E J h ) e x p { - ( 2 b / h )x/(2m*[V0 - E,])},

(4)

where b is the barrier thickness, V0 is the barrier


height, and E r is the electron energy. Other models
are more appropriate when considering the influence
of temperature and dimensionality of the emitter
contacts on the shape of the I ( V ) steps[l,6,7,8, I0,17].
Figure 2(b) and (c) show schematic band diagrams
that illustrate tunneling in a laterally confined asymmetric RTD fabricated from structure 1. Under reverse bias [Fig. 2(b)], there are 0D states in the
accumulation layer at the interface of the thick
barrier (box 1) and in the well (box 2). When in
resonance, the energies of the 0D states in box ! and
box 2 line up and electrons can tunnel between the 0D
states, leading to spikes in the I - V characteristics[21,22]. The tunneling rate F~ through the thick
(emitter) barrier is much lower than the tunneling

796

M. TEWORDTet al.

rate F2 through the thin (collector) barrier,


and therefore, at low biases, there is at most one
electron at a time in box 2. The spacing of adjacent
single particle states A E , = hco0 is related to the
measured spike spacing in bias A Vs through the
relation AV s = ~revAE,/e, where r/~v= (b~ + 0.5w)/
(b~ + w + b 2 + be) is the fraction of the bias voltage
dropped between the emitter and the well. Thus the
resonance spacings in bias are related to the 0D single
electron spectrum E, in box 2.
The forward bias situation for tunneling is depicted
in Fig. 2(c). When the 0D states in the well fall below
the electrochemical potential in the emitter contact,
electrons can tunnel from the I D subbands in the
emitter through the thin barrier (with tunneling rate
F2), which is too transparent to form an accumulation layer. Every dot state below the Fermi energy
contributes to the current by a discrete AI given in
equation (3). Increasing the bias will lead to an
incremental charging and thus to a current-voltage
staircase. If the quantum dot contains N electrons, a
current step will occur when an additional electron
tunnels into the quantum dot, and the energy of the
dot will increase by A E ( N , N + I ) = A E , + A E
c.
Thus,
the plateau
widths in bias, A V p =
A E ( N , N + 1)/e, fo,, may be larger than the spikespacing A Vs in reverse bias, where only the single
electron spectrum E, is probed.
4. R E S U L T S

AND

DISCUSSION

4.1. Fine structure at the current threshold


Figure 3(a) shows the I ( V ) of a diode fabricated
from structure
1, with conducting diameter
a = 150 nm. The conducting diameter was estimated
by extrapolating the 2D resonance peak current from
a larger diameter diode. Since the asymmetry of the
I ( V ) curve is retained compared to Fig. 1, we can
assume that the quantum well is charged considerably
in forward bias. The interesting feature of this data
is the appearance of fine structure at low biases.
In Fig. 3(b), the I - V curve of a diode from
structure 1 with diameter a = 300 nm is shown at low
biases. At T = 4.2 K (see inset), a smooth bump is
observed at the current threshold in each current
direction. At T = 20 inK, fine-structure in forward
bias has developed into a sharp current-voltage
staircase, whereas in reverse bias, a complicated
sequence of current spikes has appeared. The steps in
forward bias have a magnitude between 5 and 14 pA,
and the plateau widths range in bias between 1 and
5 mV. Spacings of the spikes in reverse bias are
AV s -~ 0.5-1 mV, are an order of magnitude smaller
than the plateau widths. The current steps in forward
bias are related to the incremental charging of the
well with single electrons. The first current step
corresponds to one electron tunneling through the
well, the second step to two electrons, and so on. The
magnitude of the current steps can be calculated
using the W K B approximation (Section 4.3).

30T=370mK
20~.~

dia = 1 5 O h m

1Or-

(~

xlO0

-10 -200

I
200

i
400

(a)

i
600

/ /

T=2OmK
~l~

100

dia=3OOnm

O[

, ~L

, I /

t-"

ll~II tit l

0
-100

_,--

/.
A

I
-40

I tt t

111

t
(b)

1
30

t...=

[
40

J
50

1
60

Bias ( m V )

Fig. 3. (a) I - V characteristics of a 150 nm diameter asymmetric double-barrier diode at T -~ 370 inK. (b) Expanded
I - V at T ~ 20 mK of a 300 nm diameter diode fabricated
from structure 1 at low current levels, showing in forward
bias the incremental charging of the quantum d o t electron
by electron, starting from zero. The inset shows that the
current steps are w a s h e d o u t at T = 4.2 K. In reverse bias,
tunneling between two quantum dots in series is observed.

The fine structure under low reverse bias indicates


that the well is not occupied with electrons between
the spikes. The most reasonable explanation for this
observation is that the first electron tunnels from
discrete states in the accumulation layer at the emitter-barrier interface [box 1 in Fig. 2(b)], in accordance with experimental data obtained from double
quantum dot structures[21]. The current step at
- 35 mV may result from a second electron tunneling
through box 2. It is not clear, however, why the
magnitudes of current spikes in reverse bias are larger
than in forward bias, because the transit time should
be smaller.
The smaller spacing of spikes as compared to the
plateau widths is in qualitative agreement with the
assumption that in reverse bias charging effects do
not contribute, because only one electron occupies
the well at a time. The fine-structure in reverse bias
contains information on the single-electron spectrum
in box 2.
4.2. Diameter dependence
It was discussed in Section 3 that both confinement
energy and charging energy decrease with increasing
quantum box size. If the lateral confinement is due to
the surface depletion potential of the diode sidewalls[2], we wish to control the confinement by
changing the diode diameter.
Figure 4 shows the I - V characteristics of four
diodes from structure I with different conducting

Single-electron tunneling and Coulomb charging


I

'

'

~ 120
SO

~30

/ //0
d=101~m

c-

'

Onm

F]
,

20

J
I

40

(a)

60

,,~Q.200150 30Ohm/i

~ ~

100

,..,v

80

.....

797

capacitance of the order 10-~6 F, and thus according


to eqn (2) a diameter a.~,160nm. Similar considerations hold when the plateau width is interpreted
as being related to lateral quantization. This result
clearly favours an explanation that the current steps
are caused by localized states within the well (maybe
due to potential fluctuations) and will have to be
subject to further investigations. It can also be seen
in Fig. 4 that the current threshold decreases with
diode diameter, and at a = l0/~m it is just above zero
bias. This observation will lead us below to argue in
favour of a model of lateral confinement from potential fluctuations.
In reverse bias [Fig. 4(b)], the spike-like I - V fine
structure is retained for all diode diameters. This
observation supports our model, that electrons tunnel
from fully quantized emitter-states in box l into box
2. The fine-structure is difficult to quantify, but we
cannot see any clear diameter dependence of the
magnitude or spacing of the spikes. Again we observe
a decrease of the current threshold with increasing
diode diameter.
4.3. Dependence on barrier thickness and symmetry

-50

-100

-150

Bias (mV)
Fig. 4. l - V characteristics of four diodes from structure 1
with conducting diameters i0 #m, 300 nm, 300 nm, and
150nm, in (a) forward bias, and (b), reverse bias. The
single-electron charging can be observed in all diodes, but
only in the forward bias direction.

diameters 150nm, 300nm, and 10/~m. In forward


bias [Fig. 4(a)], the current displays sharp steps above
the threshold voltage for all four diodes. The magnitudes of the current steps are about 10 pA for the 150
and 300 nm diameter diodes, and about 5 pA for the
10/~m device, i.e. there is only a very weak dependence on diode diameter. It is remarkable that nearly
the same current steps are observed for diodes that
differ in area by more than three orders of magnitude.
This result shows that the magnitudes of the current
steps are dependent on the heterostructure profile and
clearly supports our model of single electron tunneling [see eqn (3)] in which the magnitude of current
steps is independent of the area of the diode. The
value of the first current-step calculated using the
WKB approximation (Section 3) is AI = 15 pA. The
values of F 1 and F2 were calculated using equations
(3, 4), with values for bl, b2, w, b0, from Table I,
V0 = 256 meV (for x = 0.3), and assuming a linear
voltage drop over the double-barrier profile. The
calculation is in good agreement with the experimental current steps A1.
The most remarkable result of the data in Fig. 4 is
the observation of current steps even in a diode with
diameter a = 10 pm. In the Coulomb blockade model
a plateau width of a few mV [see Fig. 4(a)] implies a

The current for a single electron tunneling through


the quantum well is given by eqn (3). We can
therefore use the experimental dependency of current
step magnitudes on barrier thickness to test our
model of single electron tunneling.
We have studied the dependence of current step
magnitudes from four diodes with barrier thicknesses
b =4.3, 5.2, and 7.1 nm (nearly symmetric) and
b = l0 nm (thicker barrier of the asymmetric structure l). All diodes have different diameters a, because
the lateral confinement is difficult to control by
processing. In Fig. 5, the first current step is plotted
for four diodes with different barrier thicknesses b.
The offset in bias has been changed for clarity. The
temperature is T ~ 20 mK.
The magnitude of current steps decreases exponentially with barrier thickness. This is in agreement with
eqn (3) for AI, and eqn (4) for the tunnel rates. The
experimental current steps are 5-10pA, 0.5nA,
12nA, and 50nA for barrier thicknesses 10,

20mY b: L,.3nm 5.2nm 7.1nm 10.Onn


"' ~ d:60nm 80nm 300nm 300nm

,<r=

.<c:

-4"

Positive BiQs
Fig. 5. I - V characteristics of four diodes fabricated from
structures 1-4, with increasing barrier thickness b (and
various electronic diameters). The current steps A/decrease
with increasing barrier thickness. See Table I for details.

798

M.TEWORDT et al.
(b)

(a)

O000000 O0

120

oooOOOoo0oo

i I ~ l n+contact

30mK
-

Double barrier

20nA

J ~ [ ~ n +contact / /

,j

~100
w

0
--

B=O

0- 8O

1.5T

~ 4 . 5 T
8OT

ooo0

";i!i i i i i i i i i i l
ooooooooooooooooooo

::::::::::::::::::::::::::

g~ 6o

...

oooooooooooO0ooO o

60

40

20

20

40

60

Voltage (mY)

L,I,,,I,,,I,,,I,,,I,,,I,,,

40
0

10 12 14

Magnetic Field

(T)

Fig. 6. (a) I-V characteristics of a symmetric RTD from structure 3 with a conducting diameter of 83 nm,
as a function of magnetic field applied parallel to the current direction. (b) Positions of the current steps
in bias vs magnetic field according to (a).
7.1, 5, and 4.3nm, respectively (compare with
Table 1). The corresponding values calculated using
the W K B approximation using structure parameters
from Table 1 are 15pA, 2 n A , 18 nA, and 50hA.
While there is good agreement between experiment
and theory, we can say that the dependency on
barrier thickness supports the model of singleelectron tunneling [eqn (3)]. Other numerical
calculations[10] show
that
in most
cases,
quantitative agreement is obtained between theory
and model.

4.4. Tunneling in high magnetic fields


In this section we discuss the q u a n t u m dot spectra
in high magnetic fields. The data was obtained from
an 83 n m diameter RTD from structure 3116]. The
I - V characteristics in a magnetic field parallel to the
current direction are shown in Fig. 6(a). Figure 6(b)
shows resonance positions in forward bias; with
increasing magnetic field B all the current steps shift
to higher bias, following a parabolic shape. In
high magnetic field the traces run parallel to
V=(l/e~l)(ht~c/2 ), where ~oc=eB/m, is the cyclotron frequency. The cross-over to the linear increase occurs at around B = 6T, the lateral quantum
confinement can be estimated to be of the order of the
magnetic length l0 = ~
~ 10 nm at this field.
This strong confinement implies quantum energies of
the order of ~ 30 meV, which is much greater than
the energies AE ~, er/AV = e(0.3855)5-12mV =
2-5 meV obtained from the plateau widths. We therefore propose that electrons tunnel into laterally separated minima of a disorder potential in the q u a n t u m
dot[16,19].
In the circular symmetric, two-dimensional harmonic oscillator model, the eigenenergies are[24]:

E(B),.m = (2n + I m [ + 1){(h~oc/2)2 + (ho90)2}1j2


+ m (ho~J2).

(5)

We can fit the data in Fig. 6(b), using the formulas


for the energy equations (1), (2) and (5), and for the
conversion to resonance bias (see Section 3). Assuming that electrons occupy the lowest states with
n = m = 0 in the minima, we can fit several curves
and obtain confinement energies between 9 and
35 meV (for ~ = 0.3855~).5). The minima might arise
from potential fluctuations in the central region
which result from single impurities sitting in the well
or from randomly distributed donors in the contacts
(i.e. beyond the spacer layers, in a similar way as has
been described for high mobility heterojunctions[25]).
The parallel shift of the resonances with magnetic
field suggests that the lateral extension of the wavefunctions of d ~ 10 nm in all the states are about the
same.
The model corroborates with the data obtained
from the diameter dependence of the I - V fine-structure in Fig. 4. With increased diode diameter, the
probability for lower energy minima to occur in the
disorder potential increases, resulting in lower bias
resonances. For tunneling through such a low lying
state the effective barrier height is increased and
current steps are decreased according to eqn (3).
We note that other groups have claimed the observation of lateral q u a n t u m confinement from the
surface depletion potential[2]. These experiments are
very encouraging, but potential fluctuations and
single impurities have to be ruled out for such claims
to be substantiated.

5.CONCLUSIONS
We have observed the incremental charging of
the double-barrier diode with single electrons, start-

Single-electron tunneling and Coulomb charging


ing from zero, in the form of current steps with
magnitude A l = e / r .
Increasing the transmission
by decreasing the barrier thicknesses leads to
correspondingly larger current steps, in agreement
with simple model calculations. The step-like I - V is
therefore a clear demonstration of single-electron
tunneling.
Much more problematic is our understanding of
the energetics involved. We have shown that the
plateau-widths in bias of the current steps can explain
the lateral confinement energies and electrostatic
charging energies for submicron diameter quantum
dot diodes. However, fine-structure was also observed in much larger diameter diodes. The plateau
widths in that data were an order of magnitude too
large to explain confinement or charging effects. High
magnetic field data suggested that electrons in the
lowest states are laterally confined to only d ~ 10 nm.
F r o m this, we have concluded that lateral confinement is due to a disorder potential. The drawback is
that presently we cannot control this disorder potential. Therefore, our target is to find a way to control
lateral confinement as well as we can control vertical
confinement. Possible improvements could be obtained by decreasing the impurity concentration, by
growing smoother barrier interfaces, or by reducing
the surface damage during the reactive ion etch.
In conclusion, we have demonstrated that submicron diameter R T D s are very promising devices to
study the electronic properties of quantum boxes
containing only a few electrons. However, we have to
improve the experiment with respect to the purity of
the heterostructures in order to eliminate potential
fluctuations and impurities from the dot.
Acknowledgements--This work has been supported by the
SERC. M.T. acknowledges support from the Commission
of the European Communities and a Charles and Katherine
Darwin Fellowship of Darwin College, Cambridge. L.M.M.
acknowledges the Spanish Ministerio de Educacion y Ciencia for support. J.T.N. is supported by the Leverhulme Trust
and the I. Newton Trust.

REFERENCES

1. Marc A. Kastner, Physics Today, Jan., p. 25 (1993).


2. M. A. Reed et al., Phys. Rev. Lett. 60, 535 (1988);
Festk6rperprobleme: Adv. Solid St. Phys. 29, 267
(1989).
3. C. G. Smith et al., J. Phys. C: Solid State Phys. 21, L893
(1988).

799

4. Ch. Sikorski and U. Merkt, Phys. Re~,. Lett. 62, 2164


(1989); T. Demel et al., Phys. Re~'. Lett. 64, 788 (1990);
A. Lorke, J. P. Kotthaus and K. Ploog, Phys. Rer. Lett.
64, 2559 (1990).
5. T. A. Fulton and G. J. Dolan, Phys. Rer. Lett. 59, 109
(1987); J. B. Barner and S. T. Ruggiero, Phys. Rer. Left.
59, 807 (1987); I. Giaver and H. R. Zeller, Phys. Rev.
Lett. 20, 1504 (1968): J. Lambe and R. C. Jaklevic,
Phys. Rev. Lett. 22, 1371 (1969); reviewed by D. V.
Averin and K. K. Likharev, in Mesoscopic Phenomena
in Solids (Edited by B. Altshuler, P. A. Lee and R. B.
Webb), p. 169. Elsevier, Amsterdam (1991).
6. H. van Houten and C. W. J. Beenakker, Phys. Ret'. Lett.
63, 1893 (1989); L. I. Glazman and R. I. Shekhter, J.
Phys. Condens. Matter 1, 5811 (1989).
7. Atanas Groshev, Phys. Ret,. B 42, 5895 (1990); C. W. J.
Beenakker, Phys. Rev. B 44, 1646 (1991).
8. P. L. McEuen et al., Phys. Ret,. Lett. 66, 1926 (1991);
A. T. Johnson et al., Phys. Rer. Lett. 68, (1992); see also
in this context: U. Meirav, M. Kastner and S. J. Wind,
Phys. Ret'. Lett. 65, 771 (1990); L. P. Kouwenhoven
et al., Z. Phys. !]85, 367 (1991).
9. Smaller electron numbers have been reported in J. Weis,
R. J. Haug, K. von Klitzing and K. Ploog, Phys. Rev.
10. B 46, 12837 (1992).
M. Tewordt et al., J. Phys. Condens. Matter 2, 8969
(1990); Appl. Phys. Lett. 59, 1966 (1991): Phys. Rer. B
45, 14407 (1992).
11. S. Tarucha and Y. Hirayama, Phys. Ret.. B 43, 9373
12. (1991).
B. Su, V. J. Goldman, M. Santos and M. Shayegan,
Appl. Phys. Lett. 58, 747 (1991); B. Su, V. J. Goldman,
and J. E. Cunningham, Science, N.Y. 255, 313 (1992);
Phys. Rev. B 46, 7644 (1992).
13. M. W. Dellow et al. Phys. Rer'. Lett. 68, 1754 (1992).
14. P. Gu6ret, N. Blanc, R. Germann and H. Rothuizen,
Phys. Rev. Lett. 68, 1896 (1992).
15. A. Ramdane, G. Faini and H. Launois, Z. Phys. B85,
389 (1991); M. Van Hove et al., J. appl. Phys. 72, 158
(1992).
16. M Tewordt et al., Phys. Rev. B 46, 3948 (1992).
17. Garnett W. Bryant, Phys. Rev. B 39, 3145 (1989): B 44,
12837 (1991).
18. D. V. Averin, A. N. Korotkov and K. K. Likharev,
Phys. Rev. B 44, 6199 (1991).
19. R. C. Ashoori et al., Phys. Rev. Lett. 68, 3088 (1992);
see also W. Hansen et al., Phys. Rez'. Lett. 62, 2168
(1989); R. H. Silsbee and R. C. Ashoori, Phys. Rer.
Lett. 64, 1991 (1990).
20. A. Zaslavsky et al., Appl. Phys. Left. 53, 1408 (1988).
21. M. Tewordt et al., Appl. Phys. Lett. 60, 595 (1992).
22. Garnett W. Bryant, Phys. Rer. B 44, 3064 (1991).
23. Garnett W. Bryant, Phys. Rev. Lett. 59, 1140 (1987);
N. F. Johnson and M. C. Payne, Phys. Rev. Left. 67,
1157 (1991); U. Merkt, J. Huser and M. Wagner, Phys.
Ret,. B 43, 7320 (1991).
24. C. G. Darwin, Proc. Cambridge Phil. Soc. 27, 86 (1930);
V. Fock, Z. Phys. 47, 446 (1928).
25. J. A. Nixon and J. H. Davies, Phys. Re~'. B 41, 7929
(1990).

You might also like