Microscopic verification of topological electron-vortex binding in the lowest-Landau-level crystal state
Extracted source text ยท arXiv cond-mat/0406744
%\documentclass[aps,science,preprint,groupedaddress]{sci}
%\documentclass[aps,prl,preprint,psfig,superscriptaddress]{revtex4}
\documentclass[aps,prl,twocolumn,groupedaddress,showpacs,floatfix]{revtex4}
\def\bc{\begin{center}}
\def\ec{\end{center}}
\def\be{\begin{equation}}
\def\ee{\end{equation}}
\renewcommand{\vec}[1]{\mbox{\boldmath$#1$}}
\usepackage{longtable}
\usepackage{epsfig}
\begin{document}
\title{A Fermion Crystal with Quantum Coherence}
\author{Chia-Chen Chang, Gun Sang Jeon, Jainendra K. Jain}
\address{Department of Physics, 104 Davey Laboratory, The Pennsylvania State University,
Pennsylvania 16802}
\date{\today}
\begin{abstract}
When two-dimensional electrons are subjected to a very strong
magnetic field, they are believed to form a triangular Wigner crystal.
We demonstrate that, in the entire crystal phase, this crystal is very well represented
by a composite-fermion-crystal wave function, revealing that it is not a simple
Hartree-Fock crystal of electrons but an inherently quantum
mechanical crystal characterized by a non-perturbative binding of
quantized vortices to electrons, which establishes a long range quantum coherence in
it. It is suggested that this has qualitative consequences for experiment.
\end{abstract}
\pacs{73.43.-f,71.10.Pm.}
\maketitle
Wigner argued that electrons form a crystal when the interaction energy
of electrons dominates over their kinetic energy~\cite{Wigner}.
One such situation occurs when a two-dimensional electron system
is exposed to a strong magnetic field ($B$). The kinetic energy is
quantized into Landau levels under a
magnetic field, and the number of occupied Landau levels is called the
filling factor, $\nu=\rho hc/eB$, $\rho$ being the two-dimensional
electron density. At sufficiently strong magnetic fields,
when all electrons fall into the lowest Landau level ($\nu<1$),
the kinetic energy is no longer relevant, and the nature of the
state is determined solely by the Coulomb interaction.
One might expect a crystal for any $\nu<1$, but
for a range of filling factors the system condenses into a quantum Hall
liquid, characterized by dissipationless transport and precisely quantized plateaus of
Hall resistance~\cite{Tsui}. There are strong indications, however, that a crystal occurs
at sufficiently low $\nu$, and its properties have been studied in many
experimental~\cite{Jiang,Goldman,Li,Buhmann,Santos,Kukushkin,Ye,Pan,Shayegan}
and theoretical~\cite{Fukuyama,Maki,Levesque,Lam,Chui,Zhu,Price,Zheng,Fertig,Murthy,Haldane,Peterson}
works.
Our understanding of the physics of the FQHE liquid is
secure in part because the microscopic wave functions~\cite{Laughlin,Jain}
incorporating this physics have been shown to be extremely
accurate through comparisons against {\em exact} wave functions known for
systems containing a finite number of electrons.
The liquid wave functions become progressively worse with decreasing
$\nu$, consistent with the expectation that a crystal
becomes energetically favorable at some point. Indeed, a Hartree-Fock
wave function describing an ``electron crystal" (EC) provides a better approximation
for the ground state below certain filling factor. However, the electron crystal
is not particularly good quantitatively. The aim of this paper is to show
that the low $\nu$ phase is strikingly accurately described as a
quantum crystal of composite fermions, where composite fermions are bound
states of electrons and quantized vortices.
The composite-fermion (CF) crystal wave function offers a quantitative description
of the crystal state at a similar level of accuracy as the
CF liquid wave functions do for the FQHE liquid. One might have expected
residual CF correlations to survive in the crystal phase in the proximity of
the phase boundary separating the CF liquid and the crystal, but we
find that the CF crystal is realized even deep inside the crystal phase,
down to arbitrarily low fillings. We estimate the temperature
below which thermal fluctuations should not destroy the quantum effects due
to the formation of composite fermions in the crystal.
The gaussianly localized wave packet for an electron in the lowest Landau level
localized at $\vec{R}=(X,Y)$ is given by~\cite{Maki}
\be
\phi_{\vec{R}}(\vec{r})=\frac{1}{\sqrt{2\pi}} \exp\left(-\frac{1}{4}(\vec{r}-\vec{R})^2+\frac{i}{2}
(xY-yX)\right)
\ee
where the magnetic length, $l_0=\sqrt{\hbar c/eB}$, has been taken as the unit of length.
The wave function for the electron crystal (EC) is constructed by
placing electrons on a triangular lattice $\vec{R}_j$, the lowest
energy solution for the classical problem, and then antisymmetrizing
the product:~\cite{Maki}
\be
\Psi^{EC}=\frac{1}{\sqrt{N!}}\sum_P \epsilon_P \prod_{j=1}^N \phi_{\vec{R}_j}(\vec{r}_{Pj})
\ee
where the sum is over all permutations $P$ and $\epsilon_P$ is $+1$ for even
permutations and $-1$ for odd permutations.
With the lattice constant $a=(4\pi/\sqrt{3}\nu)^{1/2}l_0$,
the overlap integral between nearest neighbor electron
wave functions~\cite{Maki}, $\exp(-a^2/2l_0^2)=\exp(-3.627/\nu)$,
decays rapidly with decreasing $\nu$.
We will work with the symmetric gauge, $\vec{A}=(B/2)(-y,x,0)$, for which
the total angular momentum $L$ is a good quantum number.
Because the wave function $\Psi^{EC}$ is not an eigenstate of angular momentum,
we follow the method of Yannouleas and Landman~\cite{Landman} to project
it onto a definite $L$, denoting the resulting wave function $\Psi^{EC}_{L}$.
Such projection amounts to creating a rotating crystal, implying that
the crystalline structure is not apparent in the density but
in the pair correlation function. The explicit expression
for $\Psi^{EC}_{L}$ is given in Ref.~\onlinecite{Landman}.
Following the standard procedure of
the composite fermion (CF) theory~\cite{Jain,Kamilla}, we construct the following
wave function:
\be
\Psi^{^{2p}CFC}_L=\prod_{j<k}(z_j-z_k)^{2p}\Psi^{EC}_{L^*}
\label{CFC},
\ee
\be
L^*=L-pN(N-1).
\ee
It is interpreted as a CF crystal (CFC),
because the Jastrow factor $\prod_{j<k}(z_j-z_k)^{2p}$ binds $2p$
quantized vortices to each electron in $\Psi^{EC}$ to convert it into a composite fermion.
The composite fermions of different flavors are denoted by $^{2p}$CF, and
their crystals by $^{2p}$CFC. A variational wave function analogous to
that in Eq.~(\ref{CFC}) was considered earlier by Yi and Fertig~\cite{Fertig}
and was shown to have lower energy than the Lam-Girvin~\cite{Lam}
wave function in the filling factor range $0.1<\nu<0.2$.
Narevich, Murthy, and Fertig~\cite{Murthy} used a
Hamiltonian formulation of composite fermions to estimate gaps
and shear modulus on either side of the $\nu=1/5$ quantum Hall state.
However, it is not known how close $\Psi^{^{2p}CFC}_L$ is to
the actual ground state and what is the filling factor
range of its applicability. It is noted that Jastrow-type wave
functions were earlier used to estimate the effect of zero
point motion in the $^4$He crystal~\cite{Nosanow}.
\begin{figure}
\centerline{\epsfig{file=Jainfig1.eps,width=8cm}}
\caption{
\label{fig1}
The correlation energy of the optimal CF crystal, i.e., the
\% deviation of its Coulomb energy from the Coulomb energy of the
uncorrelated electron crystal, for $N=6$ particles.
The superscript $2p$ on $^{2p}$CFC indicates the
vortex quantum number of composite fermions. The energy of the electron crystal
for $L>400$ is taken from Yannouleas and Landman~\protect\cite{Landman}.
The deviation of the exact energy from the electron crystal energy is also
shown for $L\leq 145$; for larger angular momenta, where the exact energy is
not available, we show an accurate approximation, $V_{CF}^{(2)}$
(explained in the text), as an independent reference. For $2p>6$, the
number of vortices carried by composite fermions
is shown in brackets near the diamond. The energy difference
per particle between the electron and the CF crystals is given in the inset,
quoted in units of $e^2/\epsilon l_0$,
where $l_0$ is the magnetic length and $\epsilon$ the dielectric constant of
the host semiconductor.
}
\end{figure}
Comparisons with exact solutions
have served as a rigorous and unbiased means for establishing the physics of various
strongly correlated states in the lowest Landau level. We now subject
$\Psi^{^{2p}CFC}$ to such tests. Exact wave functions
can be obtained (using the Lanczos method) for up to $N=7$ particles
in the low-$\nu$ region of interest.
We will present below detailed results for $N=6$;
the study of $N=5$ and $N=7$ particles is consistent with
our conclusions below. The filling factor of the finite system will
be defined by the expression $\nu=N(N-1)/2L$, which gives the
correct value of $\nu$ in the thermodynamic limit. For $N=6$, the lowest
energy classical configuration has one particle at the center, with
the remaining five forming a ring around it~\cite{Peeters}.
The wave functions in Eq.~(\ref{CFC}) for $2p\neq 0$ have rather complicated
correlations built into them, but the interaction energy per particle,
\be
V=\frac{1}{N}\frac{\langle\Psi^{^{2p}CFC}_L| \sum_{j<k}
\frac{e^2}{\epsilon r_{jk}}|\Psi^{^{2p}CFC}_L \rangle}
{\langle\Psi^{^{2p}CFC}_L|\Psi^{^{2p}CFC}_L \rangle} \;,
\ee
can be evaluated by the Metropolis Monte Carlo method at least for many large values
of $2p$ (the computation time increases rapidly as $2p$ is reduced).
The total energy also has contributions from
electron-background and background-background interactions, but these terms
are the same for different crystal wave functions for a given $L$, so are not
relevant for comparisons.
Fig.~\ref{fig1} shows the correlation energy of the optimal CF crystal,
defined as the deviation of its energy from that of the uncorrelated electron crystal.
(The computationally accessible range of $2p$ allows us to determine
the minimum CFC energy. The only exceptions are the largest two values of $L$,
where we show the energy at the smallest $2p$ studied; the minimum CF energy may
be still lower here.) The exact correlation energy is also shown for $L\leq 145$.
For $L>145$, the dimension of the Fock space ($D$) is too large for an exact treatment.
As an independent reference point, we
obtain an accurate approximation to the exact energy by the method of ``CF diagonalization,"
wherein the Coulomb Hamiltonian is diagonalized in a correlated CF basis, the dimension of
which is much smaller than the dimension of the full basis needed
for the exact state; gradually increasing the
basis size gives an increasingly better approximation~\cite{Jeon}.
Fig.~\ref{fig1} quotes $V^{(2)}_{CF}$ (using the notation in Ref.~\onlinecite{Jeon}),
obtained with a correlated CF basis of dimension 150. $V^{(2)}_{CF}$ has been
shown to be very precise~\cite{Jeon}: for the six particle system
it is within 0.02\% of the exact energy for $L\leq 145$ and we expect
similar level of accuracy for higher $L$ as well.
The minimum energy for all $L$ is obtained at a non-zero
value of $2p$, which establishes that the CF crystal
provides a better variational state than the electron crystal.
Most significantly, the CF crystal is essentially the
exact solution for $\nu \leq 1/7$ ($L\geq 105$).
For $100 < L < 145$, the energy of the optimal
CF crystal is approximately within $0.02$\% of the exact energy. Tables~\ref{Otable}
and \ref{Etable} show how Laughlin's liquid wave function, $\Psi^{EC}$, and $\Psi^{CFC}$
compare with the exact wave function for $\nu=1/5$, $1/7$ and $1/9$.
As indicated earlier, the liquid wave function worsens and $\Psi^{EC}$ improves with
decreasing $\nu$, but neither is very good at small fillings.
In contrast, $\Psi^{CFC}$ is surprisingly close to the exact state.
Its overlap with the exact wave function is $\sim$99\% for $\nu=1/7$ and $1/9$, while
its energy deviates from the exact energy
by 0.016\% and 0.006\%, respectively. To appreciate the significance
of these comparisons, we note: (i)
The exact state is a linear combination of a large number of Slater determinant
basis functions (see Table~\ref{Otable}), involving $D-1$ parameters,
and yet, a single CFC wave function captures its physics almost exactly.
(ii) The CFC wave function for
$\nu\leq 1/7$ is more accurate than accepted wave functions for the
FQHE liquid states. For example, for the system of six particles at
$\nu=1/3$ in the disk geometry, the energy of Laughlin's wave function~\cite{Laughlin}
is off by 0.15\% and its overlap with the exact state is 0.964, in spite of the fact
that the dimension of the Fock space is much smaller ($D=1206$).
For larger $L$, the energy of the CFC is
lower than $V^{(2)}_{CF}$, with the possible exception of the last two points,
where we may not have the optimal CFC.
We take these comparisons as a definitive verification of the formation
of a CF crystal at low $\nu$.
\begin{table}
\caption{The last three columns give the
overlaps of CF crystal (CFC), electron crystal (EC), and Laughlin's
wave function with the exact ground state wave function at several filling factors $\nu$.
The overlap is defined as $|\langle
\Psi^{trial}|\Psi^{exact}\rangle|^2 /
\langle \Psi^{trial}|\Psi^{trial}\rangle \langle \Psi^{exact}|\Psi^{exact}\rangle$.
The second column gives $D$, the dimension of the basis space for $N=6$ electrons,
and $L$ is the total angular momentum of the state.
\label{Otable}
}
\begin{tabular}{|c | c | c | c | c|}\hline
$ \nu\;\;\; (L) $ & $D$ & CFC & EC & Laughlin \\ \hline \hline
1/5 $\phantom{0}$(75) & $\phantom{0}$19858 & 0.891 & 0.645 & 0.701 \\ \hline
1/7 (105) & 117788 & 0.994 & 0.723 & 0.504 \\ \hline
1/9 (135) & 436140 & 0.988 & 0.740 & 0.442 \\ \hline
\end{tabular}
\end{table}
Because every particle sees quantized vortices on every other particle, the formation
of composite fermions implies a long range quantum coherence in the
crystal phase. To get a feel for how the binding of vortices to
electrons affects the inter-particle correlations, we show in Fig.~\ref{fig:pair}
the pair correlation function $g(x)$
for several candidate wave functions as well as the exact ground state
for $\nu=1/7$; $g(x)$ is the probability
of finding a pair of particles at an arc distance $x$ on a circle of radius $R$.
($R$ is chosen to match the distance of a particle in the parent classical
crystal from the center of the disk.) The result shows that the
crystalline correlations are slightly weakened by the formation of
composite fermions. It is perhaps counter-intuitive that such an effect
should lead to a lower energy even at very low fillings.
\begin{table}
\caption{\label{Etable}
Interaction energies per particle for the exact ground state, the CF crystal (CFC),
the electron crystal (EC), and Laughlin's wave function for six particles at
several filling factors. The uncertainty in the last digit from Monte Carlo sampling
is given in parentheses.
}
\vspace*{0.2in}
\begin{tabular}{| c| c| c | c | c|}\hline
$ \nu\;\;\; (L) $ & exact & CFC & EC & Laughlin \\ \hline \hline
1/5 $\phantom{0}$(75) & 2.2019 & 2.2042(5) & 2.2196 & 2.2093(2) \\ \hline
1/7 (105) & 1.8533 & 1.8536(2) & 1.8622 & 1.8617(2) \\ \hline
1/9 (135) &1.6305 & 1.6306(1) & 1.6361 & 1.6388(1) \\ \hline
\end{tabular}
\end{table}
Of interest is the nature of the thermodynamic state,
obtained in the limit $N\rightarrow \infty$ at a fixed filling factor.
Finite size studies do not necessarily provide a reliable insight into
the thermodynamic state. For example, at
$\nu=1/5$ the CFC gives the best description for $N=6$ but the thermodynamic
state here is known to be a liquid~\cite{Jiang,Goldman}.
However, when an extremely precise and unambiguous description of the finite $N$
state is obtained, as is the case at $\nu\leq 1/7$,
we consider that to be a strong indication for the
nature of the state in the thermodynamic limit.
In any case, even though our finite $N$ study cannot give the precise $\nu$ value
where a transition from liquid to crystal takes place, it does make a compelling
case that whenever the thermodynamic state is a crystal, it is a crystal of
composite fermions, even in regions of the phase diagram far from the CF liquid.
The quantum character of the crystal is not fragile, and ought to be observable at
presently attainable temperatures, even at very small $\nu$.
The energy difference per particle, $V^{CFC}-V^{EC}$, shown in the
inset of Fig.~\ref{fig1}, gives a crude estimate for the temperature below
which the quantum nature of the crystal should be robust to thermal
fluctuations. The relevant temperatures appear to be
well within the present experimental reach -- for example, for
parameters appropriate for GaAs, the quantum
crystal regime is estimated to be below $\approx$ 25 mK (at $B=25$T)
even at $\nu=1/33$. From the $N$ dependence,
we have estimated that the energy difference shown in the inset
underestimates the thermodynamic energy difference by approximately
a factor of two. It is interesting to note that even as the
energy difference between the CF and the electron
crystals decreases as $\nu\rightarrow 0$, $2p$ continues to rise.
Thus, CF flavors of up to very high $2p$ are predicted to occur in the
crystal state. In the liquid phase, $^2$CFs and $^4$CFs have
definitely been observed, and there is evidence also for $^6$CFs
and $^8$CFs at relatively high temperatures.~\cite{Pan}
\begin{figure}
\centerline{\epsfig{file=Jainfig2.eps,width=8cm}}
\caption{
\label{fig:pair}
The pair correlation functions for the CF crystal (solid circles), the electron crystal
(empty squares), and Laughlin's wave function (empty triangles)
on a circle of radius $R=6.445l_0$ for six particles at $\nu=1/7$.
The solid line shows the exact pair correlation function.
}
\end{figure}
Quantum crystals of neutral bosons have been studied previously, where quantum
mechanical exchange due to a significant overlap between the neighboring
atoms is responsible for the quantum behavior, including a possible
``supersolid" phase in which Bose-Einstein condensation can
coexist with crystallinity~\cite{Leggett,Chan}. In the lowest
Landau level, the overlap between (uncorrelated) electron wave packets at
neighboring sites is negligible in the filling factor region of interest
(the overlap integral is $10^{-15}$ for $\nu=1/9$). The quantum nature
of the CF crystal owes its origin to the long range Coulomb interaction.
Given that the CF liquid behaves qualitatively differently from
an electron liquid, one may ask in what ways the properties of the
CF crystal are distinct from those of an electron crystal.
We mention here a few examples where the CFC can provide natural
explanations for certain
experimental facts, although further work will be needed to make the
connection with experiment more direct and to clarify other possible
implications. The issue is obviously
relevant to experiments that exhibit transitions between
the liquid and crystal phases. Re-entrant transitions between
the FQHE liquid and an insulating state, thought to be a pinned
crystal, have been observed~\cite{Jiang,Goldman} in going from
$\nu=2/5$ to $\nu=1/5$. In the filling factor range $1/5>\nu>1/9$,
the low temperature insulating state melts into a CF liquid upon raising
temperature~\cite{Pan}, as indicated by the appearance of FQHE like
structure. These observations become less baffling knowing that the
crystal is itself made of composite fermions rather than electrons,
thus requiring a less drastic reorganization of the state at the transition.
Another result, perhaps puzzling for an electron crystal, is that the Hall
resistance of the pinned crystal is close to the value it would have for
a liquid~\cite{Goldman2}. If the current is carried by composite
fermions instead, then the Hall voltage
induced by the accompanying vortex current (the vortices effectively behave
as magnetic flux quanta~\cite{Jain,Lopez}), through an effective
Faraday effect, is roughly consistent with the observation.
(Zheng and Fertig~\cite{Zheng} considered a similar mechanism for transport
by correlated interstitial defects.) The unexpectedly
small activation energy in the crystalline state, compared to theoretical
predictions based on an electron crystal, as well as its non-monotonic
filling factor dependence has also been
rationalized in terms of a CF crystal~\cite{Fertig,Murthy}.
In summary, we have shown that the two-dimensional electron system
in high magnetic fields provides an example of an intrinsically quantum mechanical
crystal of fermions with non-trivial correlations.
Partial support of this research by the National Science Foundation
under grant no. DMR-0240458 is gratefully acknowledged.
\begin{thebibliography}{99}
\bibitem{Wigner}
E.P. Wigner, Phys. Rev. {\bf 46}, 1002 (1934).
\bibitem{Tsui}
D.C. Tsui, H.L. Stormer, and A.C. Gossard,
Phys. Rev. Lett. {\bf 48}, 1559 (1982).
\bibitem{Jiang}
H.W. Jiang {\it et al.},
Phys. Rev. Lett. {\bf 65}, 633 (1990).
\bibitem{Goldman}
V.J. Goldman {\it et al.},
{\it Phys. Rev. Lett.} {\bf 65}, 2189 (1990).
\bibitem{Li} Y. P. Li {\em et al.},
Phys. Rev. Lett. 67, 1630 (1991).
\bibitem{Buhmann} H. Buhmann {\em et al.}, Phys. Rev. Lett. {\bf 66}, 926 (1991).
\bibitem{Santos} M. B. Santos {\em et al.},
Phys. Rev. Lett. 68, 1188 (1992).
\bibitem{Kukushkin} I. V. Kukushkin {\em et al.},
Phys. Rev. Lett. {\bf 72}, 3594 (1994).
\bibitem{Ye} P. D. Ye {\em et al.},
Phys. Rev. Lett. {\bf 89}, 176802 (2002).
\bibitem{Pan}
W. Pan {\it et al.}, Phys. Rev. Lett. {\bf 88}, 176802 (2002).
\bibitem{Shayegan} M. Shayegan in {\em Perspectives in Quantum Hall Effects},
S. Das Sarma, A. Pinczuk, Eds. (Wiley, New York, 1997).
\bibitem{Fukuyama} H. Fukuyama, P.M. Platzman, and P.W. Anderson,
Phys. Rev. {\bf B 19}, 5211 (1979).
\bibitem{Maki}
K. Maki and X. Zotos, Phys. Rev. B {\bf 28}, 4349 (1983).
\bibitem{Levesque} D. Levesque, J.J. Weis, and A.H. MacDonald, Phys. Rev.
B {\bf 30}, 1056 (1984).
\bibitem{Lam}
P.K. Lam and S.M. Girvin, Phys. Rev. B {\bf 30}, 473 (1984).
\bibitem{Chui} K. Esfarjani and S.T. Chui, Phys, Rev. B {\bf{42}}, 10758
(1990).
\bibitem{Zhu}
X. Zhu and S. G. Louie, Phys. Rev. Lett. {\bf 70}, 335 (1993).
\bibitem{Price}
R. Price, P. M. Platzman, and S. He, Phys. Rev. Lett. {\bf 70}, 339 (1993);
R. Price {\em et al.}, Phys. Rev. B {\bf 51}, 2017 (1996).
\bibitem{Zheng}
L. Zheng and H.A. Fertig, Phys. Rev. Lett. {\bf 73}, 878 (1994).
\bibitem{Fertig}
H. Yi and H.A. Fertig, Phys. Rev. B {\bf 58}, 4019 (1998).
\bibitem{Murthy}
R. Narevich, G. Murthy, and H.A. Fertig,
Phys. Rev. B {\bf 64}, 245326 (2001).
\bibitem{Haldane} K. Yang, F.D.M. Haldane, and E.H. Rezayi, Phys. Rev. B
{\bf{64}}, 081301(2001).
\bibitem{Peterson}
S.S. Mandal, M.R. Peterson, and J.K. Jain,
Phys. Rev. Lett. {\bf 90}, 106403 (2003).
\bibitem{Laughlin} R.B. Laughlin, Phys. Rev. Lett. {\bf 50}, 1395 (1983).
\bibitem{Jain}
J.K. Jain, Phys. Rev. Lett. {\bf 63}, 199 (1989); Phys. Rev. B {\bf 40},
8079 (1989).
\bibitem{Landman}
C. Yannouleas and U. Landman, Phys. Rev. B {\bf 68}, 035326 (2003);
{\em ibid.} {\bf 69}, 113306 (2004).
\bibitem{Kamilla}
J.K. Jain and R.K. Kamilla, Int. J. Mod. Phys. B {\bf 11}, 2621 (1997);
Phys. Rev. B {\bf 55}, R4895 (1997).
\bibitem{Nosanow} L.H. Nosanow, Phys. Rev. {\bf 146}, 120 (1966).
\bibitem{Peeters}
V.M. Bedanov and F.M. Peeters, Phys. Rev. B {\bf 49}, 2667 (1994).
\bibitem{Jeon}
G.S. Jeon, C.-C. Chang, and J.K. Jain, Phys. Rev. B {\bf 69},
241304(R) (2004).
\bibitem{Leggett}
A.J. Leggett, Phys. Rev. Lett. {\bf 25}, 1543 (1970).
\bibitem{Chan}
E. Kim and M.H.W. Chan, Nature {\bf 427}, 225 (2004).
\bibitem{Goldman2}
V.J. Goldman {\em et al.}, Phys. Rev. Lett. {\bf 70}, 647 (1993).
\bibitem{Lopez} A. Lopez and E. Fradkin, Phys. Rev. B {\bf 44}, 5246 (1991);
B.I. Halperin, P.A. Lee, and N. Read, Phys. Rev. B {\bf 47}, 7312 (1993);
A.S. Goldhaber and J.K. Jain, Phys. Lett. A {\bf 199}, 267 (1995).
\end{thebibliography}
\end{document}