ARTIFICIAL ATOMS

The charge and energy of a sufficiently small particle of metal

or semiconductor are quantized just like those of an atom.
The current through such a quantum dot or one-electron
transistor reveals atom-like features in a spectacular way.

Marc A. Kasfner

The wizardry of modern semiconductor technology makes Figures la and lb show two kinds of what is
it possible to fabricate particles of metal or "pools" of sometimes called, for reasons that will soon become clear,
electrons in a semiconductor that are only a few hundred a single-electron transistor. In the first type (figure la),
angstroms in size. Electrons in these structures can which I call the all-metal artificial atom,1 electrons are
display astounding behavior. Such structures, coupled to confined to a metal particle with typical dimensions of a
electrical leads through tunnel junctions, have been given few thousand angstroms or less. The particle is separated
various names: single-electron transistors, quantum dots, from the leads by thin insulators, through which electrons
zero-dimensional electron gases and Coulomb islands. In must tunnel to get from one side to the other. The leads
my own mind, however, I regard all of these as artificial are labeled "source" and "drain" because the electrons
atoms—atoms whose effective nuclear charge is controlled enter through the former and leave through the latter—
by metallic electrodes. Like natural atoms, these small the same way the leads are labeled for conventional field-
electronic sytems contain a discrete number of electrons effect transistors, such as those in the memory of your
and have a discrete spectrum of energy levels. Artificial personal computer. The entire structure sits near a large,
atoms, however, have a unique and spectacular property: well-insulated metal electrode, called the gate.
The current through such an atom or the capacitance Figure lb shows a structure2 that is conceptually
between its leads can vary by many orders of magnitude similar to the all-metal atom but in which the confinement
when its charge is changed by a single electron. Why this is accomplished with electric fields in gallium arsenide.
is so, and how we can use this property to measure the lev- Like the all-metal atom, it has a metal gate on the bottom
el spectrum of an artificial atom, is the subject of with an insulator above it; in this type of atom the
this article. insulator is AlGaAs. When a positive voltage Vg is
To understand artificial atoms it is helpful to know applied to the gate, electrons accumulate in the layer of
how to make them. One way to confine electrons in a GaAs above the AlGaAs. Because of the strong electric
small region is by employing material boundaries—by field at the AlGaAs-GaAs interface, the electrons' energy
surrounding a metal particle with insulator, for example. for motion perpendicular to the interface is quantized, and
Alternatively, one can use electric fields to confine at low temperatures the electrons move only in the two
electrons to a small region within a semiconductor. Either dimensions parallel to the interface. The special feature
method requires fabricating very small structures. This is that makes this an artificial atom is the pair of electrodes
accomplished by the techniques of electron and x-ray on the top surface of the GaAs. When a negative voltage is
lithography. Instead of explaining in detail how artificial applied between these and the source or drain, the
atoms are actually fabricated, I will describe the various electrons are repelled and cannot accumulate underneath
types of atoms schematically. them. Consequently the electrons are confined in a
narrow channel between the two electrodes. Constrictions
sticking out into the channel repel the electrons and
Marc Kastner is the Donner Professor of Science in the create potential barriers at either end of the channel. A
department of physics at the Massachusetts Institute of
plot of a potential similar to the one seen by the electrons
Technology, in Cambridge.
is shown in the inset in figure 1. For an electron to travel

24 PHYSICS TODAY JANUARY 1993 © 1993 American Insrirure of Physics

 The many forms of artificial
atoms include the all-metal
atom (a), the controlled-barrier
atom (b) and the two-probe
atom, or "quantum d o t " (c).
Areas shown in blue are
metallic, white areas are
Gate insulating, and red areas are
semiconducting. The
dimensions indicated are
Drain approximate. The inset shows
a potential similar to the one in
the controlled-barrier atom,
plotted as a function of
position at the semiconductor-
insulator interface. The
electrons must tunnel through
potential barriers caused by
the two constrictions. For
capacitance measurements
with a two-probe atom, only
the source barrier is made thin
enough for tunneling, but for
current measurements both
source and drain barriers
'Gate are thin. Figure 1

from the source to the drain, it must tunnel through the this kind of structure, most of the experiments have been
barriers. The "pool" of electrons that accumulates be- done without one, so I call this the two-probe atom.
tween the two constrictions plays the same role that the
small particle plays in the all-metal atom, and the Charge quantization
potential barriers from the constrictions play the role of One way to learn about natural atoms is to measure the
the thin insulators. Because one can control the height of energy required to add or remove electrons. This is
these barriers by varying the voltage on the electrodes, I usually done by photoelectron spectroscopy. For example,
call this type of artificial atom the controlled-barrier the minimum photon energy needed to remove an electron
atom. Controlled-barrier atoms in which the heights of is the ionization potential, and the maximum energy of
the two potential barriers can be varied independently photons emitted when an atom captures an electron is the
have also been fabricated.3 (The constrictions in these electron affinity. To learn about artificial atoms we also
devices are similar to those used for measurements of measure the energy needed to add or subtract electrons.
quantized conductance in narrow channels as reported in However, we do it by measuring the current through the
PHYSICS TODAY, November 1988, page 21.) In addition, artificial atom.
there are structures that behave like controlled-barrier Figure 2 shows the current through a controlled-
atoms but in which the barriers are caused by charged barrier atom7 as a function of the voltage VK between the
impurities or grain boundaries.24 gate and the atom. One obtains this plot by applying a
Figure lc shows another, much simpler type of very small voltage between the source and drain, just large
artificial atom. The electrons in a layer of GaAs are enough to measure the tunneling conductance between
sandwiched between two layers of insulating AlGa As. One them. The results are astounding. The conductance
or both of these insulators acts as a tunnel barrier. If both displays sharp resonances that are almost periodic in Vg.
barriers are thin, electrons can tunnel through them, and By calculating the capacitance between the artificial atom
the structure is analogous to the single-electron transistor and the gate we can show2-8 that the period is the voltage
without the gate. Such structures, usually called quantum necessary to add one electron to the confined pool of
dots, have been studied extensively.56 To create the electrons. That is why we sometimes call the controlled-
structure, one starts with two-dimensional layers like barrier atom a single-electron transistor: Whereas the
those in figure lb. The cylinder can be made by etching transistors in your personal computer turn on only once
away unwanted regions of the layer structure, or a metal when many electrons are added to them, the artificial
electrode on the surface, like those in figure lb, can be used atom turns on and off again every time a single electron is
to repel electrons everywhere except in a small circular added to it.
section of GaAs. Although a gate electrode can be added to A simple theory, the Coulomb blockade model, ex-

PHYSICS TODAY JANUARY 1993 25

 10

o
Q
o
o

285 290 295 300

GATE VOLTAGE Vg (millivolts)

II
10" 10-'

10" •b 10- 2 -
LU
o O
5
O
3
a Q
10"3" 10" 3 -
o
Conductance of a controlled-barrier o 8 1 \
atom as a function of the voltage VK on 71
/ 1 1 \
the gate at a temperature of 60 mK. At f 1 1 \
low I/, (solid blue curve) the shape of I 1
1
the resonance is given by the thermal 10" 10- AL/« .
distribution of electrons in the source
that are tunneling onto the atom, but at . 1
high VK a thermally broadened
1
\ ^
Lorentzian (red curve) is a better 1
10- 10" 1 1
description than the thermal distribution 282.5 282.9 291.0 292.2
alone (dashed blue curve). (Adapted
Vg (millivolts) Vg (millivolts)
from ref. 7.) Figure 2

plains the periodic conductance resonances.9 (See PHYSICS The gap in the tunneling spectrum is the difference
TODAY, May 1988, page 19.) This model is quantitatively between the "ionization potential" and the "electron
correct for the all-metal atom and qualitatively correct for affinity" of the artificial atom. For a hydrogen atom the
the controlled-barrier atom.10 To understand the model, ionization potential is 13.6 eV, but the electron affinity,
think about how an electron in the all-metal atom tunnels the binding energy of H", is only 0.75 eV. This large
from one lead onto the metal particle and then onto the difference arises from the strong repulsive interaction
other lead. Suppose the particle is neutral to begin with. between the two electrons bound to the same proton. Just
To add a charge Q to the particle requires energy Q2/2C, as for natural atoms like hydrogen, the difference between
where C is the total capacitance between the particle and the ionization potential and electron affinity for artificial
the rest of the system; since you cannot add less than one atoms arises from the electron-electron interactions; the
electron the flow of current requires a Coulomb energy difference, however, is much smaller for artificial atoms
ei/2C. This energy barrier is called the Coulomb blockade. because they are much bigger than natural ones.
A fancier way to say this is that charge quantization leads By changing the gate voltage Ve one can alter the
to an energy gap in the spectrum of states for tunneling: energy required to add charge to the particle. Vg is
For an electron to tunnel onto the particle, its energy must applied between the gate and the source, but if the drain-
exceed the Fermi energy of the contact by e'2/2C, and for a source voltage is very small, the source, drain and particle
hole to tunnel, its energy must be below the Fermi energy will all be at almost the same potential. With Vg applied,
by the same amount. Consequently the energy gap has the electrostatic energy of a charge Q on the particle is
width e'2/C. If the temperature is low enough that
kT<e'~/2C, neither electrons nor holes can flow from one E=QVg +Q2/2C (1)
lead to the other. For negative charge Q, the first term is the attractive

26 PHYSICS TODAY JANUARY 1993

 Q o = - Ne

V4)e

Total energy (top) and

tunneling energies (bottom) for
an artificial atom. As the gate
voltage is increased the charge
Qo for which the energy is
minimized changes from
- / V e t o - ( / V + V4)e. Only
the points corresponding to
discrete numbers of electrons
on the atom are allowed
(dots on upper curves). Lines
in the lower diagram
EF
indicate energies needed for
electrons or holes to tunnel
onto the atom. When
Qo = - (/V + '/, )e the gap in
tunneling energies vanishes
• Increasing gate voltage Vg and current can
flow. Figure 3

interaction between Q and the positively charged gate therefore analogous to moving through the periodic table
electrode, and the second term is the repulsive interaction for natural atoms by increasing the nuclear charge.
among the bits of charge on the particle. Equation 1 shows The quantization of charge on a natural atom is
that the energy as a function of Q is a parabola with its something we take for granted. However, if atoms were
minimum at Qo = — CVg. For simplicity I have assumed larger, the energy needed to add or remove electrons
that the gate is the only electrode that contributes to C; in would be smaller, and the number of electrons on them
reality, there are other contributions.7 would fluctuate except at very low temperature. The
By varying Vg we can choose any value of Qo, the quantization of charge is just one of the properties that
charge that would minimize the energy in equation 1 if artificial atoms have in common with natural ones.
charge were not quantized. However, because the real
charge is quantized, only discrete values of the energy E Energy quantization
are possible. (See figure 3.) When Qo = — Ne, an integral The Coulomb blockade model accounts for charge quanti-
number N of electrons minimizes E, and the Coulomb zation but ignores the quantization of energy resulting
interaction results in the same energy difference e2/2C for from the small size of the artificial atom. This confine-
increasing or decreasing TV by 1. For all other values of Qo ment of the electrons makes the energy spacing of levels in
except Qo = — (N + V2 )e there is a smaller, but nonzero, the atom relatively large at low energies. If one thinks of
energy for either adding or subtracting an electron. the atom as a box, at the lowest energies the level spacings
Under such circumstances no current can flow at low are of the order f?/ma2, where a is the size of the box. At
temperature. However, if Qo = — (N+ \)e the state with higher energies the level spacings decrease for a three-
Q= — Ne and that with Q= —(N+ l)e are degenerate, dimensional atom because of the large number of standing
and the charge fluctuates between the two values even at electron waves possible for a given energy. If there are
zero temperature. Consequently the energy gap in the many electrons in the atom, they fill up many levels, and
tunneling spectrum disappears, and current can flow. The the level spacing at the Fermi energy becomes small. The
peaks in conductance are therefore periodic, occurring all-metal atom has so many electrons (about 107) that the
whenever CVg =Q0= —(N+ V2)e, spaced in gate voltage level spectrum is effectively continuous. Because of this,
by e/C. many experts do not regard such devices as "atoms," but I
As shown in figure 3, there is a gap in the tunneling think it is helpful to think of them as being atoms in the
spectrum for all values of Vg except the charge-degener- limit in which the number of electrons is large. In the con-
acy points. The more closely spaced discrete levels shown trolled-barrier atom, however, there are only about 30-60
outside this gap are due to excited states of the electrons electrons, similar to the number in natural atoms like
present on the artificial atom and will be discussed more in krypton through xenon. Two-probe atoms sometimes
the next section. As Ve is increased continuously, the gap have only one or two electrons. (There are actually many
is pulled down relative to the Fermi energy until a charge- more electrons that are tightly bound to the ion cores of
degeneracy point is reached. On moving through this the semiconductor, but those are unimportant because
point there is a discontinuous change in the tunneling they cannot move.) For most cases, therefore, the spec-
spectrum: The gap collapses and then reappears shifted trum of energies for adding an extra electron to the atom is
up by e 2 /C Simultaneously the charge on the artificial discrete, just as it is for natural atoms. That is why a dis-
atom increases by 1 and the process starts over again. A crete set of levels is shown in figure 3.
charge-degeneracy point and a conductance peak are One can measure the energy level spectrum directly
reached every time the voltage is increased by e/C, the by observing the tunneling current at fixed Vg as
amount necessary to add one electron to the artificial a function of the voltage Vds between drain and
atom. Increasing the gate voltage of an artificial atom is source. Suppose we adjust Vg so that, for example,

PHYSICS TODAY JANUARY 1990 27

Qo= —(N+ V4 )e and then begin to increase Vds. The fields, so electrons added to the atom reside on its surface.
Fermi level in the source rises in proportion to Vds relative Because of this, the electron-electron interaction is always
to the drain, so it also rises relative to the energy levels of e2/C (where C is the classical geometrical capacitance),
the artificial atom. (See the inset to figure 4a.) Current independent of the number of electrons added. This is
begins to flow when the Fermi energy of the source is exactly the case for which the Coulomb blockade model
raised just above the first quantized energy level of the was invented, and it works well: The conductance peaks
atom. As the Fermi energy is raised further, higher are perfectly periodic in the gate voltage. The difference
energy levels in the atom fall below it, and more current between the "ionization potential" and the "electron
flows because there are additional channels for electrons affinity" is e'2/C, independent of the number of electrons
to use for tunneling onto the artificial atom. We measure on the atom.
an energy level by measuring the voltage at which the In the controlled-barrier atom, as you can see from
current increases or, equivalently, the voltage at which figure 4, the level spacing is one or two tenths of the energy
there is a peak in the derivative of the current, d//dVds. gap. The conductance peaks are not perfectly periodic in
(We need to correct for the increase in the energy of the gate voltage, and the difference between ionization
atom with Vds, but this is a small effect.) Many beautiful potential and electron affinity has a quantum mechanical
tunneling spectra of this kind have been measured5 for contribution. I will discuss this contribution a little later
two-terminal atoms. Figure 4a shows one for a controlled- in more detail.
barrier atom.7 In the two-probe atom the electron-electron interac-
Increasing the gate voltage lowers all the energy tion can be made very small, so that one can in principle
levels in the atom by eVg, so that the entire tunneling reach the limit opposite to that of the all-metal atom. One
spectrum shifts with Vg, as sketched in figure 3. One can can find the energy levels of a two-probe atom by
observe this effect by plotting the values of Vds at which measuring the capacitance between its two leads as a
peaks appear in d//dVds. (See figure 4b.) As Vg increases function of the voltage between them.6 When no tunnel-
you can see the gap in the tunneling spectrum shift lower ing occurs, this capacitance is the series combination of
and then disappear at the charge-degeneracy point, just as the source-atom and atom-drain capacitances. For ca-
the Coulomb blockade model predicts. You can also see pacitance measurements, two-probe atoms are made with
the discrete energy levels of the artificial atom. For the the insulating layer between the drain and atom so thick
range of Vds shown in figure 4 the voltage is only large that current cannot flow under any circumstances.
enough to add or remove one electron from the atom; the Whenever the Fermi level in the source lines up with one
discrete levels above the gap are the excited states of the of the energy levels of the atom, however, electrons can
atom with one extra electron, and those below the gap are tunnel freely back and forth between the atom and the
the excited states of the atom with one electron missing source. This causes the total capacitance to increase,
(one hole). At still higher voltages (not shown in figure 4) because the source-atom capacitor is effectively shorted
one observes levels for two extra electrons or holes and so by the tunneling current. The amazing thing about this
forth. The charge-degeneracy points are the values of Vg experiment is that a peak occurs in the capacitance every
for which one of the energy levels of the artificial atom is
degenerate with the Fermi energy in the leads when
Vds = 0, because only then can the charge of the atom
fluctuate.
In a natural atom one has little control over the
spectrum of energies for adding or removing electrons.
There the electrons interact with the fixed potential of the
nucleus and with each other, and these two kinds of
interaction determine the spectrum. In an artificial atom,
however, one can change this spectrum completely by
altering the atom's geometry and composition. For the all- >
metal atom, which has a high density of electrons, the
energy spacing between the discrete levels is so small that
it can be ignored. The high density of electrons also
results in a short screening length for external electric
-2 -1 0 1
DRAIN-SOURCE VOLTAGE Vds (millivolts)

Discrete energy levels of an artificial atom

can be detected by varying the drain-source 0.5
voltage. When a large enough Vdi is applied, o
electrons overcome the energy gap and tunnel
from the source to the artificial atom. (See
inset of a.) a: Every time a new discrete state
is accessible the tunneling current increases,
giving a peak in d//d l/ ds . The Coulomb
blockade gap is the region between about
— 0.5 mV and + 0.3 mV where there are no \ .Vv •
peaks, b: Plotting the positions of these peaks
at various gate voltages gives the level w -0.5
spectrum. Note how the levels and the gap \ \ ' . . • . ' , . . ; -

lJil_
move downward as VR increases, just as
1 ••. .

305 306 307 308 309 310 311

sketched in the lower part of figure 3.
(Adapted from ref. 7.) Figure 4 GATE VOLTAGE Vg (millivolts)

28 PHYSICS TODAY JANUARY 1993

 Capacitance of a
-650 - 600 - 550 -500 two-probe atom that has
DRAIN-SOURCE VOLTAGE Vds (millivolts) only one barrier thin
enough to allow
tunneling, a: The
time a single electron is added to the atom. (See figure 5a.) capacitance has a peak
The voltages at which the peaks occur give the energies for every time a single
adding electrons to the atom, just as the voltages for peaks electron is added to the
in dI/dVds do for the controlled-barrier atom or for a two- atom. The positions of
probe atom in which both the source-atom barrier and the the peaks give the
atom-drain barrier are thin enough for tunneling. The energy spectrum of the
first peak in figure 5a corresponds to the one-electron atom, b: Peaks in
artificial atom. capacitance plotted
Figure 5b shows how the energies for adding elec- versus applied magnetic
trons to a two-probe atom vary with a magnetic field field. The green line
perpendicular to the GaAs layer. In an all-metal atom indicates the rate of
the levels would be equally spaced, by e2/C, and would be change of the energy
independent of magnetic field because the electron- expected when the
electron interaction completely determines the energy. magnetic field
By contrast, the levels of the two-probe atom are irregu- dominates. (Adapted
larly spaced and depend on the magnetic field in a from ref. 6.) Figure 5
systematic way. For the two-probe atom the fixed
potential determines the energies at zero field. The level
spacings are irregular because the potential is not highly constant-interaction model.
symmetric and varies at random inside the atom because Now think about what happens when one adds
of charged impurities in the GaAs and AlGaAs. It is electrons to a controlled-barrier atom by increasing the
clear that the electron-electron interactions that are the gate voltage while keeping Vds just large enough so one
source of the Coulomb blockade are not always so can measure the conductance. When there are N — 1
important in the two-probe atom as in the all-metal and electrons on the atom the N— 1 lowest energy levels are
controlled-barrier atoms. Their5 relative importance de- filled. The next conductance peak occurs when the gate
pends in detail on the geometry. voltage pulls the energy of the atom down enough that the
Fermi level in the source and drain becomes degenerate
Artificial atoms in a magnetic field with the Ath level. Only when an energy level is
Level spectra for natural atoms can be calculated theoreti- degenerate with the Fermi energy can current flow; this
cally with great accuracy, and it would be nice to be able to is the condition for a conductance peak. When Vg is
do the same for artificial atoms. No one has yet calculated increased further and the next conductance peak is
an entire spectrum, like that in figure 4a. However, for a reached, there are TV electrons on the atom, and the Fermi
simple geometry we can now predict the charge-degener- level is degenerate with the (N + l)-th level. Therefore to
acy points, the values of Vg corresponding to conductance get from one peak to the next the Fermi energy must be
peaks like those in figure 2. From the earlier discussion it raised by e2/C + (EN ( [ — EN), where EN is the energy of
should be clear that in such a calculation one must take the A^th level of the atom. If the energy levels are closely
into account the electron's interactions with both the fixed spaced the Coulomb blockade result is recovered, but in
potential and the other electrons. general the level spacing contributes to the energy
The simplest way to do this is with an extension of the between successive conductance peaks.
Coulomb blockade model.11"13 It is assumed, as before, It turns out that we can test the results of this kind of
that the contribution to the gap in the tunneling spectrum calculation best if a magnetic field is applied perpendicu-
from the Coulomb interaction is e2/C no matter how many lar to the GaAs layer. For free electrons in two
electrons are added to the atom. To account for the dimensions, applying the magnetic field results in the
discrete levels one pretends that once on the atom, each spectrum of Landau levels with energies (n + V2 )fuoc,
electron interacts independently with the fixed potential. where the cyclotron frequency is ac = eB/m*c, and m* is
All one has to do is solve for the energy levels of a single the effective mass of the electrons. In the controlled-
electron in the fixed potential that creates the artificial barrier atom and the two-probe atom, we expect levels
atom and then fill those levels in accordance with the that behave like Landau levels at high fields, with
Pauli exclusion principle. Because the electron-electron energies that increase linearly in B. This behavior occurs
interaction is assumed always to be e2/C, this is called the because when the field is large enough the cyclotron

PHYSICS TODAY JANUARY 1990 29

radius is much smaller than the size of the electrostatic po- artificial atoms, absorption and emission of electrons plays
tential well that confines the electrons, and the electrons this role, so we had better understand how this process
act as if they were free. Levels shifting proportionally to works. Think about what happens when the gate voltage
B, as expected, are seen experimentally. (Seefigure5b.) in the controlled-barrier atom is set at a conductance peak,
To calculate the level spectrum we need to model the and an electron is tunneling back and forth between the
fixed potential, the analog of the potential from the atom and the leads. Since the electron spends only a finite
nucleus of a natural atom. The simplest choice is a time r on the atom, the uncertainty principle tells us that
harmonic oscillator potential, and this turns out to be a the energy level of the electron has a width fi/r.
good approximation for the controlled-barrier atom. Fig- Furthermore, since the probability of finding the electron
ure 6a shows the calculated level spectrum as a function of on the atom decays as e~' / r , the level will have a
magnetic field for noninteracting electrons in a two- Lorentzian line shape.
dimensional harmonic oscillator potential. At low fields This line shape can be measured from the transmis-
the energy levels dance around wildly with magnetic field. sion probability spectrum T(E) of electrons with energy E
This occurs because some states have large angular incident on the artificial atom from the source. The
momentum and the resulting magnetic moment causes spectrum is given by
their energies to shift up or down strongly with magnetic
field. As the field is increased, however, things settle (2)
down. For most of the field range shown there are four F2 +(E-ENf
families of levels, two moving up, the other two down. At
the highest fields there are only two families, correspond- where F is approximately fi/r and EN is the energy of the
ing to the two possible spin states of the electron. iVth level. The probability that electrons are transmitted
Suppose we measure, in an experiment like the one from the source to the drain is approximately proportion-
whose results are shown in figure 2, the gate voltage at al15 to the conductance G. In fact, G~(e2/h)T, where e2M
which a specific peak occurs as a function of magnetic is the quantum of conductance. It is easy to show that one
field. This value of Vg is the voltage at which the 7Vth en- must have G<e2/h for each of the barriers separately to
ergy level is degenerate with the Fermi energy in the observe conductance resonances. (An equivalent argu-
source and drain. A shift in the energy of the level will ment is used to show that electrons in a disordered
cause a shift in the peak position. The blue line in figure conductor are localized for G < e2/h. See, for example, the
6a is the calculated energy of the 39th level (chosen fairly article by Boris L. Al'tshuler and Patrick A. Lee in PHYSICS
TODAY, December 1988, page 36.) This condition is
arbitrarily for illustration purposes), so it gives the
prediction of the constant-interaction model for the equivalent to requiring that the separation of the levels is
position of the 39th conductance peak. As the magnetic greater than their width F.
field increases, levels moving up in energy cross those Like any spectroscopy, our electron spectroscopy of
moving down, but the number of electrons is fixed, so artificial atoms has a finite resolution. The resolution is
electrons jump from upward-moving filled levels to determined by the energy spread of the electrons in the
downward-moving empty ones. The peak always follows source, which are trying to tunnel into the artificial atom.
the 39th level, so it moves up and down in gate voltage. These electrons are distributed according to the Fermi-
Figure 6b shows a measurement14 of Vg for one Dirac function,
conductance maximum, like one of those in figure 2, as a
function of B. The behavior is qualitatively similar to that /IE) = - (3)
predicted by the constant-interaction model: The peak E-E¥ +1
expl
moves up and down with increasing B, and the frequency kT
of level crossings changes at the field where only the last where EF is the Fermi energy. The tunneling current is
two families of levels remain. However, at high B the given by
frequency is predicted to be much lower than what is
observed experimentally. While the constant-interaction
model is in qualitative agreement with experiment, it is =jj-T(E) [f(E) - f(E - eVds)} dE (4)
not quantitatively correct.
To anyone who has studied atomic physics, the Equation 4 says that the net current is proportional to the
constant-interaction model seems quite crude. Even the probability f(E)T{E) that there is an electron in the source
simplest models used to calculate energies of many- with energy E and that the electron can tunnel between
electron atoms determine the charge density and potential the source and drain minus the equivalent probability
self-consistently. One begins by calculating the charge for electrons going from drain to source. The best
density that would result from noninteracting electrons in resolution is achieved by making Vds4kT. Then
the fixed potential, and then one calculates the effective [/IE) - f(E - eV^^eV^idf/dE), and / i s proportional to
potential an electron sees because of the fixed potential Vds, so the conductance is // V4s.
and the potential resulting from this charge density. Then Figure 2 shows that equations 2-4 describe the
one calculates the charge density again. One does this experiments well: At low Vg, where F is much less than
repeatedly until the charge density and potential are self- kT, the shape of the conductance resonance is given by the
consistent. The constant-interaction model fails because it resolution function df/dE. But at higher Vg one sees the
is not self-consistent. Figure 6c shows the results of a self- Lorentzian tails of the natural line shape quite clearly.
consistent calculation for the controlled-barrier atom.14 It The width F depends exponentially on the height and
is in good agreement with experiment—much better width of the potential barrier, as is usual for tunneling.
agreement than the constant-interaction model gives. The height of the tunnel barrier decreases with Vg, which
is why the peaks become broader with increasing Vg. Just
Conductance line shapes as we have control over the level spacing in artificial
In atomic physics, the next step after predicting energy atoms, we also can control the coupling to the leads and
levels is to explore how an atom interacts with the therefore the level widths. It is clear why the present
electromagnetic field, because the absorption and emis- generation of artificial atoms show unusual behavior only
sion of photons teaches us the most about atoms. For at low temperatures: When kT becomes comparable to

30 PHYSICS TODAY JANUARY 1993

 7.5
Effect of magnetic field on energy level
spectrum and conductance peaks.
a: Calculated level spectrum for noninteracting
electrons in a harmonic oscillator electrostatic
potential as a function of magnetic field. The
blue line is the prediction that the constant-
interaction model gives for the gate voltage for
the 39th conductance peak, b: Measured
position of a conductance peak in a
controlled-barrier atom as a function of field.
c: Position of the 39th conductance peak
versus field, calculated self-consistently. The
scale in c does not match that in b because
parameters in the calculation were not
precisely matched to the experimental
conditions. (Adapted from
ref. 14.) Figure 6

2. U. Meirav, M. A. Kastner, S. J. Wind, Phys. Rev. Lett. 65, 771

(1990). M. A. Kastner, Rev. Mod. Phys. 64, 849 (1992).
3. L. P. Kouwenhoven, N. C. van der Vaart, A. T. Johnson, W.
Kool, C. J. P. M. Harmans, J. G. Williamson, A. A. M. Staring,
C. T. Foxon, Z. Phys. B 85, 367 (1991), and refs. therein.
4. V. Chandrasekhar, Z. Ovadyahu, R. A. Webb, Phys. Rev. Lett.
67, 2862 (1991). R. J. Brown, M. Pepper, H. Ahmed, D. G.
Hasko, D. A. Ritchie, J. E. F. Frost, D. C. Peacock, G. A. C.
Jones, J. Phys.: Condensed Matter 2, 2105 (1990).
5. B. Su, V. J. Goldman, J. E. Cunningham, Science 255, 313
(1992). M. A. Reed, J. N. Randall, R. J. Aggarwal, R. J. Matyi,
T. M. Moore, A. E. Wetsel, Phys. Rev. Lett. 60, 535 (1988). M.
Tewordt, L. Martin-Moreno, J. T. Nicholls, M. Pepper, M. J.
Kelly, V. J. Law, D. A. Ritchie, J. E. F. Frost, G. A. C. Jones,
Phys. Rev. B 45, 14407 (1992).
6. R. C. Ashoori, H. L. Stormer, J. S. Weiner, L. N. Pfeiffer, S. J.
1.0 1.5 2.0 2.5 3.0 Pearton, K. Baldwin, K. W. West, Phys. Rev. Lett. 68, 3088
(1992).
MAGNETIC FIELD (tesla)
7. E. B. Foxman, P. L. McEuen, U. Meirav, N. S. Wingreen, Y.
Meir, P. A. Belk, N. R. Belk, M. A. Kastner, S. J. Wind, "The
the energy separation between resonances, the peaks Effects of Quantum Levels on Transport Through a Coulomb
Island," MIT preprint (July 1992). See also A. T. Johnson,
overlap and the features disappear. L. P. Kouwenhoven, W. de Jong, N. C. van der Vaart,
C. J. P. M. Harmans, C. T. Foxon, Phys. Rev. Lett. 69, 1592
Applications (1992).
The behavior of artificial atoms is so unusual that it is nat- 8. A. Kumar, Surf. Sci. 263, 335 (1992). A. Kumar, S. E. Laux, F.
ural to ask whether they will be useful for applications to Stern, Appl. Phys. Lett. 54, 1270 (1989).
electronics. Some clever things can be done: Because of 9. D. V. Averin, K. K. Likharev, in Mesoscopic Phenomena in
the electron-electron interaction, only one electron at a Solids, B. L. Al'tshuler, P. A. Lee, R. A. Webb, eds., Elsevier,
time can pass through the atom. With devices like the Amsterdam (1991), p. 173.
"turnstile" device1617 shown on the cover of this issue the 10. H. van Houton, C. W. J. Beenakker, Phys. Rev. Lett. 63, 1893
two tunnel barriers can be raised and lowered indepen- (1989).
dently. Suppose the two barriers are raised and lowered 11. D. V. Averin, A. N. Korotkov, Zh. Eksp. Teor. Fiz. 97, 1661
sequentially at a radio or microwave frequency v. Then, (1990) [Sov. Phys. JETP 70, 937 (1990)].
with a small source-drain voltage applied, an electron will 12. Y. Meir, N. S. Wingreen, P. A. Lee, Phys. Rev. Lett. 66, 3048
tunnel onto the atom when the source-atom barrier is low (1991).
and off it when the atom-drain barrier is low. One 13. C. W. J. Beenakker, Phys. Rev. B 44, 1646 (1991).
electron will pass in each time interval v~\ producing a 14. P. L. McEuen, E. B. Foxman, J. Kinaret, U. Meirav, M. A.
current ev. Other applications, such as sensitive elec- Kastner, N. S. Wingreen, S. J. Wind, Phys. Rev. B 45, 11419
trometers, can be imagined.918 However, the most inter- (1992).
esting applications may involve devices in which several 15. R. Landauer, IBM J. Res. Dev. 1, 223 (1957).
artificial atoms are coupled together to form artificial 16. L. P. Kouwenhoven, A. T. Johnson, N. C. van der Vaart, W.
molecules161719 or in which many are coupled to form Kool, C. J. P. M. Harmans, C. T. Foxon, Phys. Rev. Lett. 67,
artificial solids. Because the coupling between the artifi- 1626 (1991).
cial atoms can be controlled, new physics as well as new 17. L. J. Geerligs, V. F. Anderegg, P. A. M. Holweg, J. E. Mooij, H.
applications may emerge. The age of artificial atoms has Pothier, D. Esteve, C. Urbina, M. H. Devoret, Phys. Rev. Lett.
only just begun. 64, 2691 (1990).
18. H. Grabert, M. H. Devoret, eds., Single Charge Tunneling,
References Plenum, New York (1992).
1. T. A. Fulton, G. J. Dolan, Phys. Rev. Lett. 59, 109 (1987). 19. R.J.Haug.J.M. Hong, K.Y.Lee, Surf. Sci. 263, 415(1991). •

PHYSICS TODAY JANUARY 1993 31