Linear Chain of Coupled Quantum Dots
Extracted source text ยท arXiv cond-mat/9707065
\documentstyle[prl,multicol,aps]{revtex}
%\documentstyle[preprint,aps]{revtex}
\begin{document}
%\preprint{SNUTP 96-022}
\title{Linear Chain of Coupled Quantum Dots}
%\and
\author{Kicheon Kang}
\address{Department of Physics, Korea University, Seoul 136-701, Korea}
\author{Min-Chul Cha}
\address{Department of Physics, Hanyang University, Kyunggi-do,
Ansan 425-791, Korea}
\author{S.-R. Eric Yang}
\address{Department of Physics, Korea University, Seoul 136-701, Korea}
\maketitle
\draft
\begin{abstract}
A linearly coupled chain of spin-polarized quantum dots is
investigated under the condition that the number of
electrons is equal to or less than the number of the dots.
The chemical potential of the system, $\mu_{N}=E(N)-E(N-1)$,
satisfies, $(\mu_{N}+\mu_{N_{\ell}+2-N)}/2 \approx V+2t$
($N$, $N_{\ell}$, $V$, $E(N)$ and $t$ are the number of
electrons, the number of dots, and the strength of nearest
neighbor electron-electron interactions, the total groundstate
energy and the hopping integral between two adjacent dots).
This property will be reflected in the spacing between the
conductance peaks. The electron density structures are
determined using a quantum Monte Carlo method. As the number
of electrons is varied several correlated structures are found
that are commensurate/incommensurate with the periodic dot system.
Hartree-Fock theory fails to predict the correct electronic
structures of this system because several nearly degenerate
solutions exist.
\end{abstract}
\thispagestyle{empty}
\pacs{PACS numbers: 73.20.Dx, 73.20.Mf}
\begin{multicols}{2}
Recent advances in nano-fabrication techniques have made it possible
to make quantum dots \cite{K,Kas}.
These structures have several
similarities to atoms and are called artificial
atoms.
The physical parameters of these systems may be controlled easily:
the electron density can be varied significantly
by the substrate voltage $V_{g}$ and
the range of electron-electron interactions
can be controlled by changing the distance to the metallic layer.
When these
artificial atoms are coupled structures similar
to a molecule may arise.
Recently numerous groups have started to
investigate whether coupled two-dot systems really have molecular
properties. Properties of a two-dot
system have been studied both experimentally\cite{Kou,Wau,Bli,Vaa,Hof}
and theoretically\cite{Mat,Gol}.
When many dots are coupled an artificial one-dimensional crystal
may be created. It is possible
to couple up to ten or more dots together
since the size of the total system can be made smaller than the
phase coherence length.
Such a system has energy bands separated by energy gaps, and
transport properties of such a periodic crystal have been investigated
experimentally \cite{Kou2}.
When the length of such a periodic crystal is infinite the system
exhibts many intriguing properties. Depending of the values of the physical
parameters the system can have properties of a Luttinger liquid
or a generalized
Wigner crystal \cite{yang}.
It is unclear whether these properties remain in a finite system of coupled
dots.
The purpose of this paper is to examine general properties of
a linear chain of spin-polarized dots
under the condition $N\leq N_{\ell}$.
We find the following properties.
The chemical potential of the system satisfies,
$(\mu_{N}+\mu_{N_{\ell}+2-N})/2 \approx V+2t$.
%(Note that a ring of $N_{\ell}$ quantum dots satisfies
%$(\mu_{N}+\mu_{N_{\ell}+1-N})/2=V+2t$).
This property of the chemical potential
is reflected in the spacing between the conductance peaks \cite{SBA}.
The electron density structures,
determined by a quantum Monte Carlo
method, show several types of commensurate/incommensurate
structures
as the number of electrons is varied.
The groundstate electronic
properties of these structures
cannot be described by mean field theory (Hartree-Fock theory)
except for a few isolated cases. The reason
for this failure is the existence of many nearly degenerate solutions.
%The lowest energy state of an isolated dot can contain
%two electrons with opposite spins. When $N_{\ell}$ dots are coupled
%this energy state will
%develop into a band with $N_{\ell}$ states. Such a band can contain at
%most $2N_{\ell}$ electrons. The bands originating from
%the higher lying states of an isolated dot may be ignored when
%the characteristic electron-electron energy is smaller than the
%energy separations between the bands.
As shown by recent experiments when $N \leq N_{\ell}$ the
system can be easily spin-polarized
\cite{Kou2}.
%In such a case two electrons cannot be found in the same isolated dot
%because of Pauli exclusion principle.
%When the gate is close to the electrons the screening will be strong
%and electron-electron interactions will be short-ranged \cite{com2}.
The Hamiltonian of such a coupled dots
is given by \cite{com1}
\begin{eqnarray}
H &=& H_t +H_V \nonumber \\
&=& -t \sum_{i=1}^{N_{\ell}-1}(c^{+}_{i+1}c_{i}+c^{+}_{i}c_{i+1})+2tN \\
& & +V\sum_{i=1}^{N_{\ell}-1}n_{i}n_{i+1} \nonumber
\end{eqnarray}
The operator $c_{i}(c^{+}_{i})$
destroys (creates) an electron in the $i$-th dot.
The term $2tN$ allows to measure single particle energies form zero.
The presence of the end sites at $1$ and $N_{\ell}$ breaks
particle-hole symmetry.
However, the system can be mapped into a ring of
$(N_{\ell}+1)$-coupled dots by imposing the periodic boundary
condition $c_{N_{\ell}+2}=c_{1}$ and requiring the electron occupation
number at the site $N_{\ell}+1$ to be zero:
\begin{eqnarray}
H&=&-t \sum_{i=1}^{N_{\ell}+1}(c^{+}_{i+1}c_{i}+c^{+}_{i}c_{i+1})\nonumber \\
& &+V\sum_{i=1}^{N_{\ell}+1}n_{i}n_{i+1}+w_{N_{\ell}+1}n_{N_{\ell}+1}+2tN.
\end{eqnarray}
When the strength of the impurity potential energy, $w_{N_{\ell}+1}$, is large
and positive the occupation number at the site $N_{\ell}+1$ will be zero.
We rewrite the Hamiltonian using the hole creation and hole
occupation number operators $h_{i}^{+}=c_{i}$ and
$m_{i}=1-n_{i}$:
\begin{eqnarray}
H=H'+(V+2t)(2N-N_{\ell}-1),
\end{eqnarray}
where
\begin{eqnarray}
H'&=&t \sum_{i=1}^{N_{\ell}+1}(h^{+}_{i+1}h_{i}+h^{+}_{i}h_{i+1})\nonumber \\
& &+V\sum_{i=1}^{N_{\ell}+1}m_{i}m_{i+1}\nonumber \\
& &+w_{N_{\ell}+1}(1-m_{N_{\ell}+1})+2tN_{h}.
\end{eqnarray}
The quantity $N_{h}$
is the total number of holes, equal to
$N_{\ell}+1-N$. From this transformation we find the relation
\begin{eqnarray}
E(N)=E'(N_{\ell}+1-N)+(V+2t)(2N-N_{\ell}-1),
\end{eqnarray}
where $E(N)$ is the groundstate energy of $H$ with $N$ electrons, and
$E'(N)$ denotes the counterpart of $H'$. For sufficiently large
$N_{\ell}$ the impurity contributions to the groundstate energies are
negligible in both $H$ and $H'$, and we may set
\begin{eqnarray}
E'(N)\approx E(N)
\end{eqnarray}
>From this it follows
\begin{eqnarray}
(\mu_{N}+\mu_{N_{\ell}+2-N})/2\approx V+2t.
\end{eqnarray}
For comparison it should be noted that complete electron-hole
symmetry of a ring of $N_{\ell}$ dots leads to
\begin{eqnarray}
(\mu_{N}+\mu_{N_{\ell}+1-N})/2=V+2t. \label{eq:ring}
\end{eqnarray}
Note that the second subscript in Eq.~(\ref{eq:ring}) contains 1
instead of 2.
In a finite linear chain or in the presence of
disorder this relation
is not valid because electron-hole symmetry is broken.
We have solved Eq.~(1) by a Hartree-Fock method.
The ground state is determined by the following equations
\begin{equation}
H_{HF} |k\rangle = \epsilon_k |k\rangle
\end{equation}
Here the Hartree-Fock Hamiltonian is given by
\begin{eqnarray}
H_{HF}&=&-t\sum_{i=1}^{N_{\ell}-1} (c_i^{\dagger}c_{i+1}+ c_{i+1}^{\dagger}c_i)
+ \sum_{i=1}^{N_{\ell}} X_i c_i^{\dagger}c_i\nonumber\\
& &-V\sum_{i=1}^{N_{\ell}-1}\langle n_{i+1}\rangle \langle n_{i}\rangle.
\end{eqnarray}
where
\begin{displaymath}
X_i = V(\langle n_{i+1}\rangle + \langle n_{i-1}\rangle) .
\end{displaymath}
The electron density $\{ \langle n_i\rangle \}$ and the Hartree-Fock
eigenstates $\{ |k\rangle \}$ should be determined self-consistently.
Note that $ \langle n_0\rangle = \langle n_{N_{\ell}+1}\rangle=0$.
The total energy of $N$-electron system is given in terms of the
Hartree-Fock eigenstates $\{| k\rangle \}$ :
\begin{equation}
E(N) = \sum_{k=1}^N \epsilon_k - \frac{1}{2}\sum_{k,q=1}^N
\left( \langle kq |H_V| kq\rangle - \langle kq |H_V| qk\rangle \right),
\end{equation}
where
\begin{eqnarray*}
\langle kq |H_V| kq\rangle &=& 2V \sum_{i=1}^{N_{\ell}-1} |a_{i}(k)|^2
|a_{i+1}(q)|^2 , \\
\langle kq |H_V| qk\rangle &=& 2V \sum_{i=1}^{N_{\ell}-1}
a_{i}(k)^{*}a_{i}(q)a_{i+1}(k)^{*}a_{i+1}(q) .
\end{eqnarray*}
The amplitude of the $k$th eigenstate at site $i$ is $a_{i}(k)$.
Fig.~1 displays the lowest HF solutions
for an odd value of $N_{\ell}=7$.
Since they have nearly degenerate energies
we must include quantum fluctuations to find the true
groundstate.
Fig.~2 displays
the density profiles obtained using a quantum Monte Carlo
method \cite{Hir}.
We used typically 500 000 - 1 000 000 Monte Carlo steps to measure
the values and the statistical error is about the size of the symobols.
Quantum Monte Carlo and HF results differ significantly.
The HF theory overestimates the repulsive interactions, and consequently
favors structures with more oscillations. A commensurate
structure is found
at $N=4$. Generally, when $N_{\ell}$ is odd a commensurate state exists
for $N=(N_{\ell}+1)/2$. This type of states have interesting optical properties
\cite{yang}.
Quantum Monte Carlo results show
that the number of peaks is equal to $N$ when $N\leq N_{\ell}/2$.
The number of minima is equal to $N_{\ell}-N$ for $N > N_{\ell}/2$.
Fig.~3 displays quantum Monte Carlo density profiles for an even value
of $N_{\ell}=14$. In both cases the following structures are present:
%$(\bullet, \bullet, \circ, \circ, \bullet, \bullet)$,
%$(\bullet, \bullet, \circ, \circ)$,
$(\bullet, \bullet, \circ, \bullet, \bullet, \circ)$,
$(\circ, \circ, \bullet, \circ, \bullet, \circ, \circ, \bullet)$ where
the symbols $\bullet$ and $\circ$ denote relatively large and small occupation
numbers. HF or classical theory fails to predict these structures.
Note that when $N_{\ell}/N >1/2$ the occupation number
increases as the site index
moves away from the center and the values of $n_{i}$
are large at $i=1$ and $N_{\ell}$.
This is because the electrons feel strong mutual repulsion.
The opposite
is true for $N_{\ell}/N <1/2$. Note that, in contrast to infinite systems,
the groundstates do not
resemble a liquid state even for $V=1$. Generally when $N_{\ell}$ is even a
commensurate (periodic) state is absent, since two classically degenerate
state exists. Again we note that
the number of peaks is $N$ when $N\leq N_{\ell}/2$, and the number of
minima is $N_{\ell}-N$ when $N>N_{\ell}/2$.
The inset in Fig.~4 displays $N$ versus
$\mu_{N}$
for $N_{\ell}=15$ and
$V=4$. The distance between the $(N+1)$-th and $N$-th peaks is equal to
$\mu_{N+1}-\mu_{N}$ \cite{SBA}.
The constant charging model would fail to account for
these results since the separations between the peaks are not a constant.
We have tested the accuracy of Eq.~(7) numerically by plotting
$(\mu_{N}+\mu_{N_{\ell}+2-N})/2$ as a function of $N$. We see in Fig.~5
that even for small $N_{\ell}$ Eq.~(7) is well satisfied.
Kouwenhoven {\it et al.}\cite{Kou2} have shown that their experimental
data are better described by assuming presence of some disorder.
We have also carried out a similar calculation in the presence of an impurity,
and find that our result is in qualitative agreement with that of
Kouwenhoven {\it et al.}\cite{com2}
It would be interesting to test the validity of Eq.~(5) experimentally
in a clean coupled-dot system without impurities.
Strong correlation effects should be more visible in conductance positions
the larger the ratio $V/t$ is. A finite chain
of linearly coupled quantums dots
is well suited for observing both
commensurate/incommensurate electronic structures.
This work has been supported by the KOSEF
under grant 961-0207-040-2 and the Ministry of Education under grant
BSRI-96-2444.
M.C.C. has been supported in part by the Ministry of Education under
grant BSRI-96-2448 and by the KOSEF under grant 961-0202-008-2.
K.K. was partially supported by KOSEF-POST-DOC program.
S.R.E.Y. thanks L. P. Kouwenhoven for sending
preprints.
\begin{references}
\bibitem{K} For recent reviews see L. P. Kouwenhoven and P. L. McEuen,
in {\it Nano-Science
and Technology}, editor G. Timp, (to be published);
H. van Houten, C. W. Beenakker, and A. A. M. Staring, in {\it Single
Charge Tunneling}, edited by H. Grabert and M. H. Devoret
(Plenum, New York, 1992);
D. v. Averin and
K.K. Likharev, in {\it Mesoscopic Phenomena in Solids}, edited by B. L.
Altshuker, P.A. Lee, and R. Webb (Elsevier, Amsterdam, 1991);
U. Merkt, Adv. Solid state Phys. {\bf 30}, 77 (1990),
\bibitem{Kas}M.A.Kastner, Rev. Mod. Phys. {\bf 64}, 849 (1992);
Phys. Today {\bf 46}, No.1, 24 (1993).
\bibitem{Kou} L. Kouwenhoven, Science {\bf 268}, 1440 (1995).
\bibitem{Wau} F. R. Waugh, M. J. Berry, D. J. Mar, R. M. Westervelt,
K. L. Campman, and A. C. Gossard, Phys. Rev. Lett. {\bf 75}, 705 (1995).
\bibitem{Bli} R. H. Blick, R. J. Haug, J. Weis, D. Pfannkuche, K. v.
Klitzing, and K. Ebert, Phys. Rev.B {\bf 53}, 7899 (1996).
\bibitem{Vaa} N. C. van der Vaart, S. F. Godijn, Y. V. Nazarov,
C. J. P. M. Harmans, J. E. Mooji, L. W. Molenkamp, and C. T. Foxon,
Phys. Rev. Lett. {\bf 74}, 4702 (1995).
\bibitem{Hof} F. Hofmann, T. Heinzel, D. A. Wharam, J. P. Kotthaus, G. B\"{o}hm,W. Klein, G. Tr\"{a}nkle, and G. Weimann, Phys. Rev.B, {\bf 51}, 13872 (1995).
\bibitem{Mat} K. A. Matveev, L. I. Glazman, and H. U. Baranger,
Phys. Rev.B, {\bf 53}, 1034, (1996).
\bibitem{Gol} J. M. Golden and B. I. Halperin, Phys. Rev. B {\bf 54},
16757 (1996).
\bibitem{Kou2}L. P. Kouwenhoven, F. W. J. Hekking, B. J. van Wess,
and C. J. P. M. Harmans, Phys. Rev. Lett. {\bf 65}, 361, (1990).
R. J. Haug, J. M. Hong, and K. Y. Lee, Surf. Sci. {\bf 263}, 415 (1992).
\bibitem{yang}
S.-R. Eric Yang, Phys. Rev. B, to be published (1997), and references therein.
\bibitem{SBA}U. Sivan, R. Berkovits, Y. Aloni, O. Prus, A. Auerbach, and
G. Ben-Yoseph, Phys. Rev. Lett. {\bf 77}, 1123 (1996);
S.-R. Eric Yang, A. H. MacDonald, and M. D. Johnson,
Phys. Rev. Lett. {\bf 71}, 3194 (1993).
\bibitem{com1}When $N \leq N_l$ and in the presence of a strong magnetic field
electrons will be spin-polarized so that no dot can contain
more than one electron. In this case the system can be described our
tight binding Hamiltonian (Eq.(1)), and
the effect of the shape of the dot can be included
in the hopping parameter $t$.
\bibitem{Hir}J. E. Hirsch, R. L. Sugar, D. J. Scalapino, and R. Blankenbecler,
Phys. Rev. B {\bf26}, 5033 (1982).
\bibitem{com2}Kouwenhoven {\it et al.} interpretate
their data using a non-interacting
model, i.e. their system seems to be in the regime $V/t \leq 1$.
In such a case we find,
in agreement with them, that
the total energies come in pairs of two when an impurity is present.
\end{references}
\end{multicols}
\begin{figure}
\caption{Occupation numbers, $n_{i}$, as a function of site index, $i$.
The three lowest energy
states of a Hartree-Fock calculation are plotted. The strength of
nearest-neighbor electron-electron
interactions, the electron number, and the number of dots are
$V=5$, $N=3$, and $N_{\ell}=7$.
Here the energies are measured in units of $t$.
The graphs are shifted vertically for the sake of clarity of display.}
\end{figure}
\begin{figure}
\caption{Quantum Monte Carlo results of
occupation numbers for six different electron numbers $N$.
The parameters are $V=5$ and $N_{\ell}=7$.}
\end{figure}
\begin{figure}
\caption{Quantum Monte Carlo results of
occupation numbers for nine different electron numbers $N$.
The solid lines are for $V=4$ while
dotted lines are for
$V=1$. The total site number is $N_{\ell}=14$.}
\end{figure}
\begin{figure}
\caption{Chemical potential, $\mu_{N}$, measured in unit of $t$
is plotted as a function of $N$.
Results for $(N_{\ell},V)$ equal to (15,4), (14,4), and (14,1) are
represented by circles, inverted triangles, and triangles. The inset displays
$N$ vesus $\mu_{N}$ for $(N_{\ell},V)=(15,4)$. The energy
difference between two adjacent values of $\mu_{N}$
corresponds to the spacing between two nearby conductance
peaks.}
\end{figure}
\begin{figure}
\caption{The quantity $(\mu_{N}+\mu_{N_{\ell}+2-N})/2$, measured in unit of $t$
is plotted as a function of
$N$ to test whether it is approximately equal to a constant $V+2t$.
Results for $(N_{\ell},V)$ equal to (15,4), (14,4), (14,1), (7,4),
and (7,1) are
represented by circles, inverted triangles, and triangles, diamonds,
and squares.}
\end{figure}
\end{document}
>From kckang@silicon.korea.ac.kr Thu Jun 19 14:15 JST 1997
Received: from silicon.korea.ac.kr by venus.korea.ac.kr with SMTP
(1.38.193.4/16.2) id AA01911; Thu, 19 Jun 1997 14:15:55 +0900
Return-Path: <kckang@silicon.korea.ac.kr>
Received: by silicon.korea.ac.kr
(1.38.193.4/16.2) id AA02814; Thu, 19 Jun 1997 14:19:53 +0900
From: Kang Kicheon <kckang@silicon.korea.ac.kr>
Subject: revised manuscript
To: eyang@venus.korea.ac.kr
Date: Thu, 19 Jun 97 14:19:53 JST
Mailer: Elm [revision: 70.85]
Status: RO
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
% Manuscript (Revtex-file) and five figures (PS-file)
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% 1) Manuscript
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\documentstyle[preprint,aps]{revtex}
%\documentstyle[prb,multicol,aps]{revtex}
\begin{document}
%\preprint{SNUTP 96-022}
\title{Linear Chain of Coupled Quantum Dots}
%\and
\author{Kicheon Kang$^1$, Min-Chul Cha$^2$ and S.-R. Eric Yang$^1$}
\address{$^1$Department of Physics, Korea University, Seoul 136-701, Korea}
\address{$^2$Department of Physics, Hanyang University, Kyunggi-do,
Ansan 425-791, Korea}
\maketitle
\draft
\begin{abstract}
A linearly coupled chain of spin-polarized quantum dots is
investigated
under the condition that
the number of electrons is equal to or
less than the number of the dots.
The
chemical potential of the system, $\mu_{N}=E(N)-E(N-1)$, satisfies,
$(\mu_{N}+\mu_{N_{\ell}+2-N)}/2 \approx V+2t$
($N$, $N_{\ell}$, $V$, $E(N)$ and $t$ are the
number of electrons, the
number of dots, and the strength of
nearest neighbor electron-electron interactions, the total ground state
energy and the hopping
integral between two adjacent dots). This property will be reflected
in the spacing between the conductance peaks.
The electron density structures
are determined using a quantum Monte Carlo
method. As the number of electrons is varied
several correlated structures are found that are commensurate/incommensurate
with the periodic dot system.
Hartree-Fock theory fails to predict
the correct electronic structures of this system.
\end{abstract}
\thispagestyle{empty}
\pacs{PACS numbers: 73.20.Dx, 73.20.Mf}
%\begin{multicols}{2}
Recent advances in nano-fabrication techniques have made it possible
to make quantum dots \cite{K,Kas}.
These structures have several
similarities to atoms and are called artificial
atoms.
The physical parameters of these systems may be controlled easily:
the electron density can be varied significantly
by the substrate voltage $V_{g}$ and
the range of electron-electron interactions
can be controlled by changing the distance to the metallic layer.
When these
artificial atoms are coupled structures similar
to a molecule may arise.
Recently numerous groups have investigated
whether coupled two-dot systems really have molecular
properties. \cite{Kou,Wau,Bli,Vaa,Hof,Mat,Gol}.
When many dots are coupled an artificial one-dimensional crystal
may be created. It is possible
to couple up to ten or more dots together
since the size of the total system can be made smaller than the
phase coherence length.
Such a system has energy bands separated by energy gaps, and
transport properties of such a periodic crystal have been investigated
experimentally \cite{Kou2}.
When the length of such a periodic crystal is infinite the system
exhibts many intriguing properties. Depending of the values of the physical
parameters the system can have properties of a Luttinger liquid
or a generalized
Wigner crystal \cite{yang}.
It is unclear whether these properties remain in a finite system of coupled
dots.
The purpose of this paper is to examine general properties of
a linear chain of spin-polarized dots
under the condition $N\leq N_{\ell}$.
We find the following properties.
The chemical potential of the system satisfies,
$(\mu_{N}+\mu_{N_{\ell}+2-N})/2 \approx V+2t$.
%(Note that a ring of $N_{\ell}$ quantum dots satisfies
%$(\mu_{N}+\mu_{N_{\ell}+1-N})/2=V+2t$).
This property of the chemical potential
will be reflected in the spacing between the conductance peaks \cite{SBA}.
The electron density structures,
determined by a quantum Monte Carlo
method, show several types of commensurate/incommensurate
structures
as the number of electrons is varied.
The ground state electronic
properties of these structures
cannot be described by mean field theory (Hartree-Fock theory)
except for a few isolated cases. The reason
for this failure is the existence of many nearly degenerate solutions.
%The lowest energy state of an isolated dot can contain
%two electrons with opposite spins. When $N_{\ell}$ dots are coupled
%this energy state will
%develop into a band with $N_{\ell}$ states. Such a band can contain at
%most $2N_{\ell}$ electrons. The bands originating from
%the higher lying states of an isolated dot may be ignored when
%the characteristic electron-electron energy is smaller than the
%energy separations between the bands.
As shown by recent experiments when $N \leq N_{\ell}$ the
system can be easily spin-polarized
\cite{Kou2}.
%In such a case two electrons cannot be found in the same isolated dot
%because of Pauli exclusion principle.
%When the gate is close to the electrons the screening will be strong
%and electron-electron interactions will be short-ranged \cite{com2}.
The Hamiltonian of such a coupled dots
is given by
\begin{eqnarray}
H &=& -t \sum_{i=1}^{N_{\ell}-1}(c^{+}_{i+1}c_{i}+c^{+}_{i}c_{i+1})+2tN
\nonumber \\
& & +V\sum_{i=1}^{N_{\ell}-1}n_{i}n_{i+1}
\end{eqnarray}
The operator $c_{i}(c^{+}_{i})$
destroys (creates) an electron in the $i$-th dot.
The term $2tN$ allows to measure single particle energies from zero.
Provided that $ N \leq N_{\ell}$, all electrons are easily
spin-polarized by the external magnetic field, and the system
can be described accurately by this Hamiltonian,
regardless of the shape of dot.
The presence of the end sites at $1$ and $N_{\ell}$ breaks
particle-hole symmetry.
However, the system can be mapped into a ring of
$(N_{\ell}+1)$-coupled dots by imposing the periodic boundary
condition $c_{N_{\ell}+2}=c_{1}$ and requiring the electron occupation
number at the site $N_{\ell}+1$ to be zero:
\begin{eqnarray}
H&=&-t \sum_{i=1}^{N_{\ell}+1}(c^{+}_{i+1}c_{i}+c^{+}_{i}c_{i+1})\nonumber \\
& &+V\sum_{i=1}^{N_{\ell}+1}n_{i}n_{i+1}+w_{N_{\ell}+1}n_{N_{\ell}+1}+2tN.
\end{eqnarray}
When the strength of the impurity potential energy, $w_{N_{\ell}+1}$, is large
and positive the occupation number at the site $N_{\ell}+1$ will be zero.
We rewrite the Hamiltonian using the hole creation and hole
occupation number operators $h_{i}^{+}=c_{i}$ and
$m_{i}=1-n_{i}$:
\begin{eqnarray}
H=H'+(V+2t)(2N-N_{\ell}-1),
\end{eqnarray}
where
\begin{eqnarray}
H'&=&t \sum_{i=1}^{N_{\ell}+1}(h^{+}_{i+1}h_{i}+h^{+}_{i}h_{i+1})\nonumber \\
& &+V\sum_{i=1}^{N_{\ell}+1}m_{i}m_{i+1}\nonumber \\
& &+w_{N_{\ell}+1}(1-m_{N_{\ell}+1})+2tN_{h}.
\end{eqnarray}
The quantity $N_{h}$
is the total number of holes, equal to
$N_{\ell}+1-N$. From this transformation we find the relation
\begin{eqnarray}
E(N)=E'(N_{\ell}+1-N)+(V+2t)(2N-N_{\ell}-1),
\end{eqnarray}
where $E(N)$ is the ground state energy of $H$ with $N$ electrons, and
$E'(N)$ denotes the counterpart of $H'$. For sufficiently large
$N_{\ell}$ the impurity contributions to the ground state energies are
negligible in both $H$ and $H'$, and we may set
\begin{eqnarray}
E'(N)\approx E(N) .
\end{eqnarray}
>From this it follows
\begin{eqnarray}
(\mu_{N}+\mu_{N_{\ell}+2-N})/2\approx V+2t.
\end{eqnarray}
For comparison it should be noted that the complete electron-hole
symmetry of a ring of $N_{\ell}$ dots leads to
\begin{eqnarray}
(\mu_{N}+\mu_{N_{\ell}+1-N})/2=V+2t. \label{eq:ring}
\end{eqnarray}
Note that the second subscript in Eq.~(\ref{eq:ring}) contains 1
instead of 2.
In a finite linear chain or in the presence of
disorder this relation
is not valid because electron-hole symmetry is broken.
We have first tried to solve Eq.~(1) by a Hartree-Fock method.
The ground state is determined by the following equations
\begin{equation}
H_{HF} |k\rangle = \epsilon_k |k\rangle .
\end{equation}
Here the Hartree-Fock Hamiltonian is given by
\begin{eqnarray}
H_{HF}&=&-t\sum_{i=1}^{N_{\ell}-1} (c_i^{\dagger}c_{i+1}+ c_{i+1}^{\dagger}c_i)
+ \sum_{i=1}^{N_{\ell}} X_i c_i^{\dagger}c_i\nonumber\\
& &-V\sum_{i=1}^{N_{\ell}-1}\langle n_{i+1}\rangle \langle n_{i}\rangle,
\end{eqnarray}
where
\begin{displaymath}
X_i = V(\langle n_{i+1}\rangle + \langle n_{i-1}\rangle) .
\end{displaymath}
The electron density $\{ \langle n_i\rangle \}$ and the Hartree-Fock
eigenstates $\{ |k\rangle \}$ should be determined self-consistently.
Note that $ \langle n_0\rangle = \langle n_{N_{\ell}+1}\rangle=0$.
The total energy of $N$-electron system is given in terms of the
Hartree-Fock eigenstates $\{| k\rangle \}$ :
\begin{equation}
E(N) = \sum_{k=1}^N \epsilon_k - \frac{1}{2}\sum_{k,q=1}^N
\left( \langle kq |U| kq\rangle - \langle kq |U| qk\rangle \right),
\end{equation}
where
\begin{eqnarray*}
\langle kq |U| kq\rangle &=& 2V \sum_{i=1}^{N_{\ell}-1} |a_{i}(k)|^2
|a_{i+1}(q)|^2 , \\
\langle kq |U| qk\rangle &=& 2V \sum_{i=1}^{N_{\ell}-1}
a_{i}(k)^{*}a_{i}(q)a_{i+1}(k)^{*}a_{i+1}(q) ,
\end{eqnarray*}
and $U$ represents the nearest-neighbor electron-electron interactions.
The amplitude of the $k$th eigenstate at site $i$ is $a_{i}(k)$.
Fig.~1 displays the lowest energy HF solutions
for an odd value of $N_{\ell}=7$.
Since they have nearly degenerate energies
we must include quantum fluctuations to find the true
ground state.
The HF mean field theory is a poor approximation since the electron-electron
interactions are short-ranged.
Fig.~2 displays
the density profiles obtained using a quantum Monte Carlo
method \cite{Hir}.
We used typically 500 000 - 1 000 000 Monte Carlo steps
and the statistical error is about the size of the symbols.
These calculations are performed at the temperature $0.1t$.
Quantum Monte Carlo and HF results differ significantly.
The HF theory overestimates the repulsive interactions, and consequently
favors structures with more oscillations. A commensurate
structure is found
at $N=4$. Generally, when $N_{\ell}$ is odd a commensurate state exists
for $N=(N_{\ell}+1)/2$. This type of states have interesting optical properties
\cite{yang}.
Quantum Monte Carlo results show
that the number of peaks is equal to $N$ when $N\leq N_{\ell}/2$.
The number of minima is equal to $N_{\ell}-N$ for $N > N_{\ell}/2$.
Fig.~3 displays quantum Monte Carlo density profiles for an even value
of $N_{\ell}=14$. For both $V/t=4$ and 1 the following structures are present:
%$(\bullet, \bullet, \circ, \circ, \bullet, \bullet)$,
%$(\bullet, \bullet, \circ, \circ)$,
$(\bullet, \bullet, \circ, \bullet, \bullet, \circ)$,
$(\circ, \circ, \bullet, \circ, \bullet, \circ, \circ, \bullet)$ where
the symbols $\bullet$ and $\circ$ denote relatively large and small occupation
numbers. HF or classical theory fails to predict these structures.
Note that when $N >N_l/2$ the occupation number
increases as the site index
moves away from the center and the values of $n_{i}$
are large at $i=1$ and $N_{\ell}$.
This is because the electrons feel strong mutual repulsion.
The opposite
is true for $N <N_l/2$. Note that, in contrast to infinite systems,
the ground states do not
resemble a liquid state even for $V=1$. Generally when $N_{\ell}$ is even a
commensurate (periodic) state is absent, since two classically degenerate
states exist. Again we note that
the number of peaks of conductance
is $N$ when $N\leq N_{\ell}/2$, and the number of
minima is $N_{\ell}-N$ when $N>N_{\ell}/2$.
The inset in Fig.~4 displays $N$ versus
$\mu_{N}$
for $N_{\ell}=15$ and
$V=4$. The distance between the $(N+1)$-th and $N$-th peaks is equal to
$\mu_{N+1}-\mu_{N}$ \cite{SBA}.
The constant charging model would fail to account for
these results since the separations between the peaks are not a constant.
We have tested the accuracy of Eq.~(7) numerically by plotting
$(\mu_{N}+\mu_{N_{\ell}+2-N})/2$ as a function of $N$. We see in Fig.~5
that even for small $N_{\ell}$ Eq.~(7) is well satisfied.
Kouwenhoven {\it et al.}\cite{Kou2} have shown that their experimental
data are better described by assuming presence of some disorder.
Our calculations for $V/t=1$ in the presence of an impurity
is in agreement with the result of Kouwenhoven {\it et al.}
It would be interesting to test the validity of Eq.~(5) experimentally
in a clean coupled-dot system without impurities.
Strong correlation effects should be more visible in conductance
peak positions
the larger the ratio $V/t$ is. A finite chain
of linearly coupled quantum dots
is well suited for observing both
commensurate/incommensurate electronic structures.
This work has been supported by the KOSEF
under grant 961-0207-040-2 and the Ministry of Education under grant
BSRI-96-2444.
M.C.C. has been supported in part by the Ministry of Education under
grant BSRI-96-2448 and by the KOSEF under grant 961-0202-008-2.
K.K. has been supported in part by KOSEF-Postdoctoral program.
S.R.E.Y. thanks L. P. Kouwenhoven for sending
preprints.
\begin{references}
\bibitem{K} For recent reviews see L. P. Kouwenhoven and P. L. McEuen,
in {\it Nano-Science
and Technology}, editor G. Timp, (to be published);
H. van Houten, C. W. Beenakker, and A. A. M. Staring, in {\it Single
Charge Tunneling}, edited by H. Grabert and M. H. Devoret
(Plenum, New York, 1992);
D. V. Averin and
K.K. Likharev, in {\it Mesoscopic Phenomena in Solids}, edited by B. L.
Altshuler, P.A. Lee, and R. Webb (Elsevier, Amsterdam, 1991);
U. Merkt, Adv. Solid state Phys. {\bf 30}, 77 (1990);
R. C. Ashoori, Nature {\bf 379}, 414 (1996).
\bibitem{Kas}M.A.Kastner, Rev. Mod. Phys. {\bf 64}, 849 (1992);
Phys. Today {\bf 46}, No.1, 24 (1993).
\bibitem{Kou} L. Kouwenhoven, Science {\bf 268}, 1440 (1995).
\bibitem{Wau} F. R. Waugh, M. J. Berry, D. J. Mar, R. M. Westervelt,
K. L. Campman, and A. C. Gossard, Phys. Rev. Lett. {\bf 75}, 705 (1995).
\bibitem{Bli} R. H. Blick, R. J. Haug, J. Weis, D. Pfannkuche, K. v.
Klitzing, and K. Ebert, Phys. Rev.B {\bf 53}, 7899 (1996).
\bibitem{Vaa} N. C. van der Vaart, S. F. Godijn, Y. V. Nazarov,
C. J. P. M. Harmans, J. E. Mooji, L. W. Molenkamp, and C. T. Foxon,
Phys. Rev. Lett. {\bf 74}, 4702 (1995).
\bibitem{Hof} F. Hofmann, T. Heinzel, D. A. Wharam, J. P. Kotthaus, G. B\"{o}hm,W. Klein, G. Tr\"{a}nkle, and G. Weimann, Phys. Rev.B, {\bf 51}, 13872 (1995).
\bibitem{Mat} K. A. Matveev, L. I. Glazman, and H. U. Baranger,
Phys. Rev.B, {\bf 53}, 1034, (1996).
\bibitem{Gol} J. M. Golden and B. I. Halperin, Phys. Rev. B {\bf 54},
16757 (1996).
\bibitem{Kou2}L. P. Kouwenhoven, F. W. J. Hekking, B. J. van Wess,
and C. J. P. M. Harmans, Phys. Rev. Lett. {\bf 65}, 361, (1990).
R. J. Haug, J. M. Hong, and K. Y. Lee, Surf. Sci. {\bf 263}, 415 (1992).
\bibitem{yang}S.-R. Eric Yang, Phys. Rev. B, to be published (1997).
\bibitem{SBA}U. Sivan, R. Berkovits, Y. Aloni, O. Prus, A. Auerbach, and
G. Ben-Yoseph, Phys. Rev. Lett. {\bf 77}, 1123 (1996);
S.-R. Eric Yang, A. H. MacDonald, and M. D. Johnson,
Phys. Rev. Lett. {\bf 71}, 3194 (1993).
\bibitem{Hir}J. E. Hirsch, R. L. Sugar, D. J. Scalapino, and R. Blankenbecler,
Phys. Rev. B {\bf26}, 5033 (1982).
%\bibitem{com2}Our results for long-range electron-electron interactions
%do not differ qualitatively from those of short range interactions.
%\bibitem{com3}Our calculations for $V=1$ in the presence of an impurity
%is in agreement with the result of Kouwenhoven {\it et al.}
\end{references}
%\end{multicols}
\begin{figure}
\caption{Occupation numbers, $n_{i}$, as a function of site index, $i$.
The three lowest energy
states of a Hartree-Fock calculation are plotted. The strength of
nearest-neighbor electron-electron
interactions, the electron number, and the number of dots are
$V=5$, $N=3$, and $N_{\ell}=7$.
Here the energies are measured in units of $t$.
}
\end{figure}
\begin{figure}
\caption{Quantum Monte Carlo results of
occupation numbers for six different electron numbers $N$.
The parameters are $V=5$ and $N_{\ell}=7$.}
\end{figure}
\begin{figure}
\caption{Quantum Monte Carlo results of
occupation numbers for nine different electron numbers $N$.
The solid lines are for $V=4$ while
dotted lines are for
$V=1$. The total site number is $N_{\ell}=14$.}
\end{figure}
\begin{figure}
\caption{Chemical potential, $\mu_{N}$, measured in unit of $t$
is plotted as a function of $N$.
Results for $(N_{\ell},V)$ equal to (15,4), (14,4), and (14,1) are
represented by circles, inverted triangles, and triangles. The inset displays
$N$ vs. $\mu_{N}$ for $(N_{\ell},V)=(15,4)$. The energy
difference between two adjacent values of $\mu_{N}$
corresponds to the spacing between two nearby conductance
peaks.}
\end{figure}
\begin{figure}
\caption{The quantity $(\mu_{N}+\mu_{N_{\ell}+2-N})/2$, measured in unit of $t$
is plotted as a function of
$N$ to test whether it is approximately equal to a constant $V+2t$.
Results for $(N_{\ell},V)$ equal to (15,4), (14,4), (14,1), (7,4),
and (7,1) are
represented by circles, inverted triangles, and triangles, diamonds,
and squares.}
\end{figure}
\end{document}