Imaging and manipulating electrons in a 1D quantum dot with Coulomb blockade microscopy

Jiang Qian1,2 , Bertrand I. Halperin1 and Eric J. Heller1

1
Physics Department, Harvard University, Cambridge, MA 02138, USA
2 Arnold Sommerfeld Center for Theoretical Physics and Center for NanoScience, Ludwig-Maximilians-Universität München, Germany
(Dated: February 22, 2010)
Motivated by recent experiments by the Westervelt group, which used a mobile tip to probe the electronic
state of a segmented nanowire, we calculate shifts in Coulomb blockade peak positions, as a function of tip
location, which we term “Coulomb blockade microscopy”. We show that if the tip can be brought sufficiently
close to the nanowire, one can distinguish a high density electronic liquid state from a Wigner crystal state by
microscopy with a weak tip potential. In the opposite limit of a strongly negative tip potential, the potential
depletes the electronic density under it and divides the quantum wire into two partitions. There the tip can push
individual electrons from one partition to the other, and the Coulomb blockade micrograph can clearly track such
transitions. We show that this phenomenon can be used to qualitatively estimate the relative importance of the
electron interaction compared to one particle potential and kinetic energies. Finally, we propose that a weak tip
Coulomb blockade micrograph focusing on the transition between electron number N = 0 and N = 1 states may
be used to experimentally map the one-particle potential landscape produced by impurities and inhomogeneities.

I. INTRODUCTION

Studies of nanoscale electronic structures hold important

promise both as laboratories for few-body, interacting quan-
tum mechanical systems and as technological testbeds for
future classical or quantum computing technologies. Novel
probe technologies 1,2 are very important for studying elec-
tronic properties in nanoscale systems because they are often
beyond the resolution of conventional imaging techniques like
optical microscopy, and traditional transport measurements FIG. 1: Schematic geometry of Coulomb blockade microscopy of
can only measure spatially averaged physical properties such a quantum wire containing four electrons. When calculating the
as the conductance or the current. One scanning probe mi- electron-electron interaction and the electron-tip interaction, we as-
croscopy (SPM)2 technique utilizes a charged metallic tip to sume that the InP barriers have zero thickness, the InAs wire is in-
finitely long, and the substrate layers extend to infinity in x and y
perturb the local electronic density in a nanoelectronic struc-
directions.
ture while monitoring the resulting change in transport prop-
erties. Using this technique one can obtain spatially resolved
measurement of the electronic properties, including the local
electron density and, in principle, the wavefunction itself in II. MODEL
the case of a one-electron system3 (see discussions below).
This imaging technique has been fruitfully applied to study We consider a uniform InAs (dielectric constant ε = 15.4,
the flow of ballistic electrons across a range of two dimen- Bohr radius aB ≈ 34nm) nanowire of radius R = 10nm, which
sional heterostructures. lies in vacuum atop a SiOx (ε = 3.9) layer 100nm thick, sep-
arating it from conducting doped bulk silicon (see Fig. 1).
Recently, a series of experiments4 applied SPM techniques Electrons are modelled as point charges traveling along the
to study quantum wires. In these experiments a segment of center axis of the wire, confined to interval − L2 < x < L2
an InAs nanowire lying on top of a two-dimensional SiOx by hard walls, representing the InP layers. We consider
layer was isolated from the rest of the wire by two short length L from 110nm to 500nm. The electron-electron in-
InP segments, forming a one-dimensional quantum dot with teraction U(x1 − x2 ) was calculated using the commercial
lithographically defined boundaries. A negatively charged finite-element program ComsolTM to solve the classical Pois-
probe scanned controllably the two dimensional area around son equation for a point charge on the axis of an infinite wire
the wire and the conductance across the 1D quantum dot was above in a substrate with the geometry described in Fig. 1. At
measured as a function of the probe location. Both the voltage short distance ∆x, the potential was softened to account for
of the probe and its height above the surface can also be inde- the finite thickness of the electron wavefunction, by replac-
pendently varied. Motivated by these new experimental pos- ing ∆x−1 with [(∆x)2 + R2]−1/2 . Following the approximation
sibilities, we turn to exact diagonalization techniques to study used by Topinka5, we model the negatively charged probe as a
the conductance response of a few-electron quantum dot as a fixed point charge of strength q at a location ~r0 relative to the
function of a spatially varied probe potential, in order to illus- center point of the wire. This gives rise to a one-body poten-
trate the kind of information that can be extracted in the case tial V (x;~r0 , q) for an electron on the wire axis at point x, which
of a system of several electrons. we again obtain by solving the Poisson equation (results are
 2

x (nm) 12
L=110nm
-200 -100 0 100 200 L=500nm
60 20 10
V(x) z0= 30nm
z0= 50nm 8
50 z0=100nm 16

ρ(x)*L
U(∆x) (meV)

V(x) (meV)
40 U(∆x)
12
4
30
8 2
20
0
10 4 0 0.1 0.2 0.3 0.4 0.5
x/L
0 0
0 100 200 300 400 500
∆x (nm)
FIG. 3: (Color Online) Electronic densities for a L = 110nm wire and
for a L = 500nm wire in absence of tip potential. Only the right half
is shown, as the plot is symmetric about x = 0. We rescale the x and
FIG. 2: (Color Online) (Dotted)Interaction potential U(∆z).(Solid, ρ (x) with wire length L to ease the comparison.
long and short dashed) Tip V (x) potentials with a tip charge q = e
and locations ~r0 = (0, 0, z0 ) where z0 = 30nm, 50nm, 100nm.

shown in Fig. 2).

In this paper, we diagonalize the exact 1D many-body state for our wire is ∆E ≈ 148meV . The most confined ge-
Hamiltonian with the Lanczos method6 for up to electron ometry we discuss is shown in Fig. 6: a L = 110nm wire with
number N = 4: four electrons under an extremely strong tips, squeezing them
into the both ends of the wire. Even in that case, the total
h̄2 2 N N i−1
∑ ∑ ∑ U(xi , x j )Ψ = E Ψ,
energy per electron is less than 140meV, not enough to cause
− ∇ Ψ + V (x i ;~
r0 , q)Ψ +
2m∗ i=1 i=1 j=1
an excited transverse mode, with full interaction effects taken
(1) into account. Thus we expect the quantum wire to be well ap-
where Ψ is the full many-body wavefunction, depending proximated by a 1D model for all the parameters we explored
on the position xi and spin σi of the electrons. To con- in this paper.
nect to the experimentally observable variables, we consider
the Coulomb blockade peak positions of the transition from The existence of higher transverse modes will lead to a
(N − 1) to N electron ground states. The conductance through renormalization of the effective electron-electron interaction
the quantum wire is maximum when the chemical potential (generally a softening of the potential at short distances) even
difference between the lead and the wire, controlled by the if the energies of the modes are well above the Fermi energy.
voltage Vg on a back gate, is equal to the ground state en- This is because two electrons that come close together in the
ergy difference between the two states in question. We may lowest transverse mode will mix, virtually, with states where
write this condition as ∆E ≡ EN − EN−1 = α Vg + β , where the electrons are in excited modes. As our calculations use,
β is a constant and α is the proportionality constant between already, a crude phenomenological cut-off at short distances,
changes in the back gate voltage and the chemical potential in we do not include explicitly effects of this renormalization.
the quantum dot. We probe the electronic states in the quan- Technically, the existence of higher transverse modes can also
tum wire through the dependence of ∆E on the tip position lead to three-body and four-body effective interactions, aris-
~r0 and potential strength q. An interesting set of spatially re- ing from three-body and four-body collisions, but we do not
solved information about the electrons in the wire can be ex- expect such higher body terms to be important in the wires
tracted from this function, and we call this method “Coulomb under consideration.
blockade microscopy”. It is a special application of the “scan-
ning probe microscopy” developed by the Westervelt group2. We want to emphasize that although we choose a specific
In calculations in this paper we focus on the transition from wire geometry and a specific form of interaction that model
N = 3 to N = 4 electrons, but most of our conclusions are some characteristics of the setup of the Westervelt group’s
easily generalizable to other ground state transitions. on-going experiments, most features of the Coulomb block-
Finally, we note that for four non-interacting electrons with ade micrographs we discuss below are applicable to any 1D
spin in a wire of radius R = 10nm, when the dot length quantum dot system under a mobile potential. Indeed, our
L > Lm = 18.2 nm, the lowest four single particle energy levels discussions of the qualitative features of Coulomb blockade
are all longitudinal modes. The shortest wire length we con- micrographs under both the weak-tip and the strong-tip lim-
sider in this paper L = 110nm ≫ Lm , we therefore expect the its rely only on the general properties of 1D electronic sys-
wires under consideration can be well approximated as strictly tems, independent of the specific geometry and interaction we
1D under the assumption of weak interaction effects. Indeed, adopt. Our numerical results mostly serve for illustrative pur-
the gap between the transverse ground state and first excited poses.
 3

9.68 L=500nm x0(nm)

z0= 30nm
z0= 50nm 0 50 100 150 200 250 300
z0=100nm 73.68 9.68
9.64 L=110nm

L=110nm E4-E3(meV)

L=500nm E4-E3(meV)
L=500nm
9.64
E4-E3(meV)

9.6 73.66
9.6
9.56
9.56
73.64
9.52 9.52

9.48 73.62 9.48

0 50 100 150 200 250 0 10 20 30 40 50
x0 (nm) L=110nm x0(nm)

FIG. 4: (Color Online) Coulomb blockade micrographs for a 1D FIG. 5: (Color Online) q = 0.02e weak tip limit for the N = 3 to
dot with L = 500nm and tip charge q = 0.02e for three tip poten- N = 4 Coulomb blockade transition for densities shown in Fig. 3.
tial shown in Fig.2. Again, the right half is shown. The tip distance to the wire is z0 = 30nm. Only the right half is
shown.

III. ELECTRONIC DENSITY

L=500nm x0(µm)
0 50 100 150 200 250 300
In the absence of a probe potential V , both the N = 3 and 190
L=110nm
40

N = 4 wires the electronic density profile ρ (x) undergoes a L=500nm

L=110nm E4-E3(meV)

L=500nm E4-E3(meV)
35
180
crossover as a function of L from a liquid state characterized
30
by a 2kF Friedel oscillations to a quasi-Wigner crystal state 170
characterized by a 4kF density oscillation. Such a crossover 25

160
from a liquid state to quasi-Wigner states with a decrease in 20

density a very generic phenomenon for 1D interacting fermion 150 15

system7,8 . Indeed, for any interacting decaying no faster than 0 10 20 30
L=110nm x0(µm)
40 50

x−2 at long distance a quasi-Wigner crystal state is known

to emerge at low density8 . This requirement for interaction
will hold for a system with long screening length as compared
to mean inter-particle distance, as is the case for our geom- FIG. 6: (Color Online) q = 8e strong tip limit for the N = 3 to N = 4
etry when the screening doped silicon layer is relatively far Coulomb blockade transition for densities shown in Fig. 3 The tip to
(100µ m) away. For our specific geometry and interaction, the wire distance is 30nm. Only the right half is scanned.
crossover happens at around density ρ ∗ ≈ 35µ m−1. A Wigner
crystallized density variation is shown in the dashed curve of
Fig. 3 for a quantum dot of L = 500nm, whereas for L = 110 quasi-Wigner crystal state on the right can be detected in the
the four electron density exhibits Friedel oscillations. Coulomb blockade micrograph, whereas when z0 = 100nm
away, the tip potential becomes much too broad to resolve the
fine features of the density oscillations. We note that although
IV. WEAK TIP LIMIT the resolution of the tip is largely determined by the distance
z0 , the contrast of a Coulomb blockade micrograph, i.e. the
Now we introduce a weak tip potential, corresponding to a magnitude of the 4kF variations in the micrographs, can be
negatively charged tip of strength q = 0.02e, scanning above improved by modestly increasing the tip potential.
the center axis of the quantum wire along its direction (1, 0, 0), By contrast, in Fig. 5 the L = 110nm micrograph at z0 =
with the tip location vector ~r0 = (x0 , 0, z0 ). For a 1D quan- 30nm does not show features of Wigner crystal oscillations.
tum dot of length L = 500nm, which as shown in Fig. 3 has However, this micrograph does not by itself give a clearcut
4kF Wigner-crystal density variation, let us consider the three indication of the absence of Wigner crystal order for L =
tip heights above the quantum wire, z0 = 30nm, 50nm, 100nm, 110nm. With the current interaction and tip parameters, one
corresponding to the three tip potential shown in Fig. 2. The cannot observe the crossover from the Wigner crystal to the
resulting Coulomb blockade peak position ∆E as a function of Friedel oscillations because it happens at a inter-particle spac-
the tip coordinate x0 along the wire, i.e. the Coulomb block- ing ∆x ≈ 30nm, below the resolution of the micrograph even
ade micrograph, is shown in Fig. 4. Clearly, in Fig. 2 the at z = 30nm. To be more specific, we may define the onset
closer the tip approaches the wire, the more localized is the tip of Wigner-crystal order for our four-electron system as the
potential and a sharper tip potential make it easier to resolve length L at which there first appears a local minimum of the
the density variations, this is reflected in Coulomb blockade mean density ρ (x) in the vicinity of x/L = 0.25. According
micrograph scans in Fig. 4. At z0 = 30nm, 50nm from the tip to our calculations, this should occur at L=135 nm. However,
to the center of the wire, the 4kF density oscillation of the with the tip at height 30nm, in the weak charge limit, the reso-
 4

(2,2) (4,0)
L R L R
density
300 0.04

(a) (b)

200
(2,2) (3,1) (4,0)
x0 (nm)

0.02
L R L R L R

100

(c) (d) (e)

0 0 FIG. 8: Schematic illustrations of partition of the four-electron state

-250 -200 -150 -100 -50 0 50 100 150 200 250
by the potential barrier under a strong tip potential. In (a) and (b),
x (nm)
single-electron energy levels are shown for non-interacting electrons,
for two positions of the tip. Because of spin degeneracy, we find ei-
ther two electrons in each well, or all four in the same well, when
FIG. 7: Electronic densities in L = 500nm, N = 4 quantum wire as a the energy of the single-electron ground state on the right becomes
tip q = 8e scan through 0 ≤ x0 ≤ 250nm. The density throughout the higher than the first excited level on the left. In (c)-(e), we “incor-
entire wire −250nm ≤ x ≤ 250nm is shown. porate” interaction energy into “single electron levels” schematically
by plotting the energy needed to add an additional electron. In draw-
ing energy levels this way, it is clear that the spin degeneracy in the
lution of the micrograph is of the order of 60nm, so we would non-interacting case is lifted by electron interaction, and an addi-
not see a secondary minimum in the micrograph signal until tional (3,1) partition will appear for the “energy level” arrangement
L ≥ 250nm. in (d). Filled circles show electrons in occupied levels.
To gain a more intuitive understanding of a weak tip
Coulomb blockade micrograph, we observe that a weak tip
only slightly disturbs the electron density as it scans across wire show two relatively sharp peaks for a large tip charge
the wire, thus, a simple first order perturbation theory should q = 8. This is in contrast with the case of a weak tip Fig. 5,
be a good approximation to compute the ground state energy where the Coulomb blockade micrographs show smooth spa-
in the presence of the tip potential: tial dependence as well as sensitivity to the electronic states in
Z
the absence of the tip potential.
E(~r, q) − E0(~r, q) = dx V (~r0 ; q, x)ρ (x), (2) To understand the physics of this strong tip limit we note
that the two sharp cusps in Fig. 6 represents discontinuous
slope changes in the N = 4 electron ground state energy as a
where ρ (x) is the non-interacting ground state density and function of tip position x0 . Similarly the deep valley in the fig-
E0 (~r0 , q) is its energy. We have checked that for tip charges ure corresponds to a cusp in N = 3 ground state energy. The
up to q = 0.1e the simple first order perturbation theory gives origin of these three discontinuities in slopes can be seen in
a decent fit to both the ground state energy and the Coulomb Fig. 7. In this limit, the negatively charged tip potential is so
blockade micrograph. Since both the width and the center strong that it depletes the electronic density under it. Thus the
location of the tip potential V (~r0 ; q, x) can be adjusted exper- tip creates an effective partition of the electrons in the wire
imentally, the Coulomb blockade microscopy with a weak tip into left and right sub-quantum dot. As shown in Fig. 7, as
potential provides a flexible way to map the electronic densi- the tip move from the center to right of the wire with four
ties in a quantum dot. Indeed, as an example of such flexibil- electrons, the partitions of the electrons undergoes two abrupt
ity, we find that one can improve the “contrast” of a Coulomb transitions (2, 2) → (3, 1) → (4, 0). These two transitions cor-
blockade micrograph; i.e., the prominence of the spatial vari- respond to the two upward cusps shown in the N = 4 curve
ations in micrographs like Fig. 4 as compared with the total in Fig. 6. Similarly, the discontinuous slope change shown on
energy shift E4 − E3 , can be improved by slightly increasing the N = 3 curve of the Fig. 6 corresponds to the transition be-
the tip potential while still staying within the weak tip pertur- tween the (2, 1) → (3, 0) partition of the ground state. Thus
bative approximation. the three discontinuities seen in the Coulomb blockade mi-
crographs in Fig. 6 correspond to, alternately, the transitions
between the integer partitioning of total electron numbers in
V. STRONG TIP LIMIT the N = 3 and N = 4 system. The upward slope of the curve
near x0 = 0 reflects an additional downward cusp at the origin,
In the opposite limit of strong tip, the Coulomb blockade due to the transition (1,2)→(2,1) in the N = 3 wire.
tip scans present a very different physical picture. In Fig. 6 To better understand the transitions between different parti-
we observe that irrespective of whether the electronic state is tions, let us consider the transitions in a model of N = 4 elec-
liquid or Wigner-crystal like as shown in Fig. 3, the Coulomb trons, with spin, which have no Coulomb repulsion between
blockade micrographs show similar behavior: in the case of them but interact with a repulsive tip potential. The scenario
N = 3 to N = 4 transition, both the L = 110nm and L = 500nm is illustrated in Fig. 8, panels (a) − (b). When tip is near the
 5

58 68 10
tip q=0.6 effective charge q=0
tip q=1.8 9 effective charge q=0.6
57 67 effective charge q=1.8
8
q=0.6 E4-E3(meV) 56 66

q=1.8 E4-E3(meV)
7
55 65
6

ρ(x)*L
54 64
5
53 63 4

52 62 3

51 61 2

1
50 60
0.09 0.1 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0
x0 (nm) 0 0.1 0.2 0.3 0.4 0.5
x/L

FIG. 9: (Color Online). q = 0.6e and q = 1.8e intermediate tip po- FIG. 10: (Color Online) Electronic density for a L = 180nm wire
tentials for N = 3 to N = 4 Coulomb blockade transition in a wire of under three tip potentials, with effective charges q = 0e, q = 0.6e
length L = 180nm. The tip distance to the wire is z0 = 30nm. Only and q = 1.8e. Only the right half is shown, as the plot is symmetric
the right half is shown. about x = 0. We rescale the x and ρ (x) with the wire length L.

VI. INTERMEDIATE TIP CHARGE

We have carried out calculations with various tip charges

center of the wire, the electrons are partitioned (2, 2) and both intermediate between the two limits discussed above. In gen-
electrons in each side reside in the single-particle ground state. eral, the larger the charge on the tip, the more readily one
As the tip moves rightward, the energy levels rise in the right sees the secondary minima in the micrograph signal, which
partition and fall in the left. When the first excited level on the are seen in Fig. 6 for a charge q = 8e, even at L = 110nm.
left partition crosses the ground state on the right, both elec- As one illustration, for a wire of length 180 nm, with a tip
trons will move to the left partition. Therefore, contrary to the height of 30 nm, we find that the micrograph signal shows a
interacting case, there is no energetically favorable state of secondary minimum at x/L ≈ 0.15 for q = 1.8e, but shows no
(3, 1) partitioning in the non-interacting system. In the N = 3 secondary minimum when q = 0.6e, as seen in Fig. 9 How-
case, the (2, 1) partition is not affected by this, and for non- ever, if we calculate the electron density in the wire when
interacting system the transition (2, 1) → (3, 0) will coincide the tip is over the center of the wire (x0 = 0), we find that
with the transition in N = 4, so the Coulomb blockade micro- a tip charge of q = 0.6e is enough to substantially modify the
graph will show only a single peak. This analysis can also density relative to the density in the absence of the tip. As
be generalized to a wire containing multiple non-interacting seen in Fig. 10, the electron density below the tip, at x = 0
electrons, such that all the (odd, odd) partitionings of elec- is reduced by a factor of three relative to the density with no
tron number will be missing. tip charge. Nevertheless, the oscillations seen in ρ (x) remain
qualitatively similar to structure seen in the absence of the tip.
As schematically illustrated in Fig. 8, panels (c) − (e), For example, the charged tip only pushes out the position of
when we take electron interaction into consideration, the one the secondary minimum in the density from x/L = 0.25 to
and two electron state would no longer be degenerate in either x/L = 0.3.
partition, so contrary to the non-interacting scenarios in panel In general, when we increase the tip potential to intermedi-
(a) − (b), here a (3, 1) partition can survive as an interme- ate values, the resolution of our micrographs improves com-
diate stage between the (2, 2) and (4, 0) partitions. With the pared with the weak tip limit. This is evident in the fact that at
non-interacting case in mind, we postulate that the distance q = 1.8e we can already see signs of Wigner crystallization at
between two peaks in N = 3 to N = 4 micrographs, corre- L = 180nm, in contrast with the weak tip case with q = 0.02e
sponding to the tip positions where (3, 1) partitioning in the where we can only detect quasi-Wigner crystal at L = 250nm.
N = 4 wire is stable, can serve as an indicator of the rela- On the other hand, at q = 1.8e there is no signature of Wigner
tive importance of the interaction energy versus the sum of crystallization when the electrons are in a liquid state in a
kinetic and single particle potential energies. The less impor- L = 110nm wire. Thus we do not have ”false positive” sig-
tant interaction is compared to single particle energies, the less nature of quasi-Wigner crystal, in contrast with the strong tip
splitting would the one and two particle energies be, and the limit described in SectionV, where the micrographs show N
smaller is the region of stable (3, 1) partition. This can be seen peaks for a N-electron wire regardless whether, in the wire in
in Fig. 6. The potential energy should have a larger share in the absence of the tip, the electrons are in a liquid or a quasi-
the total energy in the longer wire with lower electronic den- Wigner crystal state.
sity, and indeed we observe that the longer wire has a wider Because of these two characteristics, an intermediate tip po-
distance between the two peaks marking (2, 2) → (3, 1) and tential may help an experimentalist to reliably detect the pres-
(3, 1) → (4, 0) transitions. ence of a quasi-Wigner crystal state in a shorter, higher density
 6

wire closer to the crossover from a liquid state. the other, and the accompanying Coulomb blockade micro-
graph can indicate the transitions between different partition-
ings. Furthermore, a feature of the resulting micrograph, the
VII. SINGLE ELECTRON distance between peaks marking the (odd, odd) partitioning,
can serve as an indicator of the relative strength of the inter-
Beyond the system of interacting electrons discussed above, action.
a possible further application of Coulomb blockade mi- In this paper we have chosen extreme values of the tip
croscopy is to experimentally “map” the rugged potential charge q to illustrate the physics in the two limits. However,
landscape produced by wire inhomogeneities and charged im- our calculations show that the discussions above hold true for
purities in the substrate. One would focus on the transition a wider range of moderately small and large values of q.
from N = 0 to N = 1 state, in which case the Coulomb block- To obtain a quantitative description of the energy shifts ex-
ade micrograph would reveal information about the single par- pected in Coulomb blockade microscopy, particularly in the
ticle density. By inverting the transformation in Eq. 2, one intermediate coupling regime, we see that it is necessary to
may be able to approximately obtain the single particle ground perform a realistic calculation, which takes into account both
state density ρ (x). In the absence of an external magnetic the electron-electron interaction and the non-linear effects of
field, the ground state wavefunction ψ (x) has no nodes and the charged tip on the electronic state of the wire. If one is
can be chosen to be ψ (x) = (ρ (x))1/2 It is then straightfor- prepared to carry out such a calculation, however, Coulomb
ward to invert the Schrödinger’s equation to extract the poten- blockade microscopy can be a powerful probe of interaction
tial landscape from the single particle wavefunction. effects in the wire.

VIII. SUMMARY
IX. ACKNOWLEDGEMENTS
In summary, in this paper we show that tracking the peak
position shift as a charged mobile tip moves above and across We would like to thank Erin Boyd, Halvar Trodahl and
a nanowire, a technique we term Coulomb blockade mi- Jesse Berezovsky and especially Bob Westervelt for helpful
croscopy, can reveal spatially-resolved information about the discussions. This work is supported in part by NSF grants
electronic density and states of a quantum 1D dot. A weak tip PHY-0646094 and DMR-0906475. Numerical work was per-
potential can serve as a probe with a tunable width, to reveal formed in part at the Center for Nanoscale Systems, a mem-
the spatial distribution of the electronic density in the wire. ber of the National Nanotechnology Infrastructure Network
A strong tip potential that depletes part of the wire can be (NNIN) supported by NSF award ECS-0335765. JQ is also
used to manipulate individual electrons from one partition to supported in part by NIM and DFG through SFB 631.

1 O. M. Auslaender, A. Yacoby, R. de Picciotto, K. W. Baldwin, 2008.

L. N. Pfeiffer and K. W. West Science, 295(5556):825–828, 2002. 5 M. A. Topinka. PhD thesis, Harvard University, 2002.
2 M. A. Topinka, R. M. Westervelt and E. J. Heller Physics Today, 6 J. K. Cullum and R. A. Willoughby. Lanczos Method for Large
56(47), 2003 and refs. therein. Symmetric Eigenvalue Computation, Vol. 1. Birkhauser Boston
3 P. Fallahi, A. C. Bleszynski-Jayich, R. M. Westervelt, J. Huang, Inc., Boston, MA, 1985.
J. D. Walls, E. J. Heller, M. Hanson and A. C. Gossard, Proc. 27th 7 G. A. Fiete, J. Qian, Y. Tserkovnyak, and B. I. Halperin Phys. Rev.
Int. Conf. on Phys. of Semiconductor, AIP, 2005, p. 779. B, 72, 045315, 2005.
4 A. C. Bleszynski-Jayich, L. E. Fröberg, M. T. Björk, H. J. Trodahl, 8 K. A. Matveev, Phys. Rev. B, 70, 245319, 2004.
Lars Samuelson and R. M. Westervelt, Phys. Rev. B, 77, 245327