Dipolar superfluidity in electron-hole bilayer systems
Extracted source text ยท arXiv cond-mat/0404033
% This is a draft for a PRL paper for the eh superfluid %
% FINAL REVISIONS TO THIS DRAFT: 12.01.04 (editorial acceptance from PRL).
% This version reduces the size of the paper.
\documentclass[prl,showpacs,twocolumn,floatfix,amsfonts]{revtex4}
\usepackage{graphicx,graphics,color,epsfig}% Include figure files
\usepackage{bm}
\usepackage{amsmath}
\usepackage{amssymb}
%\usepackage[showrefs]{refcheck}
\newcommand{\bS}{{\bf S}}
\newcommand{\bd}{{\bf d}}
\newcommand{\bB}{{\bf B}}
\newcommand{\bQ}{{\bf Q}}
\newcommand{\bk}{{\bf k}}
\newcommand{\bv}{{\bf v}}
\newcommand{\bl}{{\bf l}}
\newcommand{\br}{{\bf r}}
\newcommand{\bp}{{\bf p}}
\newcommand{\s}{{\sigma}}
\newcommand{\bA}{{\bf A}}
\newcommand{\ba}{{\bf a}}
\newcommand{\bJ}{{\bf J}}
\newcommand{\bE}{{\bf E}}
\newcommand{\bb}{{\bf b}}
\newcommand{\bsig}{\mbox{\boldmath{$\sigma$}}}
\newcommand{\Det}{\,{\rm{Det}\,}}
\newcommand{\Tr}{\,{\rm{Tr}\,}}
\newcommand{\tr}{\,{\rm{tr}\,}}
\newcommand{\bn}{{\bf \nabla}}
\newcommand{\sign}{\,{\rm{sign}\,}}
\newcommand{\bomega}{\bar{\Omega}}
\newcommand{\wL}{\omega_{L}}
\newcommand{\COMMENT}{\large\bf}
\newcommand{\FigWidth}{\columnwidth}
\newcommand{\beqa}{\begin{eqnarray}}
\newcommand{\eeqa}{\end{eqnarray}}
\newcommand{\bI}{\bf{I}_s}
\renewcommand{\Re}{{\rm Re}}
\renewcommand{\Im}{{\rm Im}}
%-----------------------------------------------------------------------------%
\begin{document}
\preprint{}
\title{Dipolar superfluidity in electron-hole bilayer systems}
\author{Alexander V. Balatsky,$^1$ Yogesh N. Joglekar,$^1$ and Peter B.
Littlewood$^{1,2}$} \affiliation{$^1$ Theoretical Division, Los
Alamos National Laboratory, Los
Alamos, New Mexico 87544, USA\\
$^2$ Cavendish Laboratory (TCM), University of Cambridge,
Cambridge CB3 0HE, United Kingdom}
\date{\today}
\begin{abstract}
Bilayer electron-hole systems, where the electrons and holes are created via
doping and confined to
separate layers, undergo excitonic condensation when distance between the
layers is smaller than the typical distance between particles within a layer.
We argue that the excitonic condensate is a novel dipolar superfluid in
which the phase of the condensate couples to the {\it gradient} of the vector
potential. We predict the existence of dipolar supercurrent which can be
tuned by an in-plane magnetic field. Thus the dipolar superfluid offers an
example of excitonic condensate in which the {\it composite} nature of its
constituent excitons is manifest in the macroscopic superfluid state. We also
discuss various properties of this superfluid including the role of vortices.
\end{abstract}
\pacs{73.20.Mf, 71.35.Lk}
\maketitle
%-----------------------------------------------------------------------------%
%\section{Introduction}
%\label{sec:intro}
{\it Introduction:} Superfluidity was discovered by Kapitsa in 1938 as the
ability of liquid Helium (He$^4$) to carry momentum without
dissipation~\cite{kapitsa}. Its phenomenological theory was developed by
Landau and collaborators, who introduced the notion of the condensate wave
function as an order parameter that describes the superfluid component of the
liquid~\cite{landau}. This superfluid component can carry momentum with no
dissipation. The cornerstone of the Landau theory of superfluidity is the
notion that phase of the superfluid condensate couples to the gauge potential
and that the condensate current is given by $\bJ=\rho_s(\bn \phi-e\bA)$ where
$\rho_s$ is a measure of the superfluid density, $\phi$ is the
phase of the condensate wave function and $\bA$ is the gauge potential (we use
units such that $c=1=\hbar$). He$^4$, being neutral, does not carry
electrical current although it can produce nontrivial gauge potentials by
rotation. Another example of a neutral superfluid is an excitonic condensate.
Electrons and holes in semiconductors can form (metastable) bound states
called excitons which are expected to behave as neutral bosons at low
densities, and therefore can undergo Bose-Einstein
condensation~\cite{moskalenko,keldysh,bilayer,rice,shev}. The condensate,
being neutral, will posses properties similar to superfluidity. During the
past few years electron-hole plasmas have been created by optically exciting
the electrons from a valence band to the conduction band and then spatially
confining the resulting electrons and holes to different quantum wells using
a static electric field~\cite{butov}. The investigation of their properties
has been limited to photoluminescence measurements: processes which can probe
phase-coherence but not superfluidity. In particular, they have not been
investigated via transport measurements which can provide a {\it direct}
signature of their superfluid properties.
It has been argued that quantum Hall {\it electron-electron} bilayers at total
filling factor $\nu=1$ provide a realization of excitonic superfluid. These
systems have been studied via transport; they show, among spectacular
transport properties, a small but finite dissipation in the counterflow
channel~\cite{em}. Here we focus on {\it electron-hole} systems in zero
magnetic field perpendicular to the layers, discussing effects due to the
presence of a physical dipole.
Recent developments in heterojunction fabrications open up an exciting
possibility of electron-hole bilayer systems where the electrons reside in
one layer and holes in the other layer separated by a distance $d$
($\sim$200\AA), and where the density of electrons or holes in individual
layers can be adjusted using independent gates~\cite{sivan}. These
systems have very weak but {\it nonzero} interlayer tunneling which allows
the electrons and holes to couple with an in-plane magnetic field applied
between the two layers. As a result of Coulomb attraction between electrons
and holes, the excitonic condensate is expected to occur when the typical
distance $r_s$ between electrons (holes) within a layer exceeds the distance
$d$ between the two layers~\cite{loz,pbl,palo} (In contrast, bilayer quantum
Hall systems can be mapped on to excitonic superfluids only near specific
filling factors such as $\nu=1$). These systems offer an alternate view of
the electron-hole excitonic condensate where the condensate is
neutral, yet has a well defined {\it dipole moment} associated with each
exciton. In this Letter, we argue that this excitonic condensate will
represent a qualitatively new kind of superfluid where the condensate is
neutral and carries no momentum density. We call this nominally neutral
superfluid a {\it dipolar superfluid}. The dipole moment associated with each
exciton in the condensate allows this liquid to couple to electromagnetic
fields in a nontrivial fashion. We find that the phase of the dipolar
superfluid couples to the {\it gradient of the gauge potential}. As a result,
we predict that it
will exhibit a neutral persistent dipolar current, consisting of equal and
oppositely directed currents in the two layers, upon application of an in-plane
magnetic field $\bB_{||}$. Thus, the composite structure of excitons is
manifest in the macroscopic superfluid state. In the
following paragraphs, we present the hydrodynamics of such a superfluid
based on the Ginzburg-Landau (GL) energy functional, state various predictions
which follow from it (including the existence of critical field
$\bB^{c}_{||}$ above which the dipolar superfluidity is destroyed), and then
briefly outline the derivation of the GL functional from a microscopic
Hamiltonian. Because there is no long-range order in two dimensions,
photoluminescence from a coherent exciton droplet is suppressed when the
thermal length for phase fluctuations reaches the size of the
droplet~\cite{pl}. On the other hand, the phase stiffness, which determines
superfluidity, persists up to the Kosterlitz-Thouless (KT) transition
temperature which is much higher. Therefore, a direct probe of phase
stiffness can provide a better signature of excitonic condensation. Although
the true nature of the superfluid phase transition (being KT-type) cannot be
described by the standard mean-field theory, we will concentrate on
temperatures much smaller than the KT-transition temperature, $T\ll T_{KT}$,
where the Hartree-Fock mean-field description is applicable~\cite{vig}.
%-----------------------------------------------------------------------------%
%\section{hydrodynamics}
%\label{sec: hydro}
{\it GL Energy Functional and Effective Action:} Let us consider a bilayer
system with electrons in the top layer and holes in the bottom layer. We
introduce a notation where $\pm$ subscript corresponds to the top and bottom
layer respectively. With the coordinate system shown in
Fig.~\ref{fig: ehbilayer} we get $\br_\pm=\br\pm\bd/2$ where $\bd=d\hat{z}$
is a vector normal to the layers and $\br$ is a two-dimensional vector within
the layer. We define the excitonic condensate order parameter as
\beqa
\Delta(\br) = \langle c^{\dag}_+(\br) c_-(\br)\rangle =
|\Delta(\br)| \exp\left[i\Phi(\br)\right]
\label{eq: op}
\eeqa
where $c^{\dag}_{\pm}(\br)$ creates an electron in the top (bottom) layer at
position $\br$, and we have used the fact that $c_{-}(\br)=c^{\dag}_{h}(\br)$
where $c^{\dag}_{h}(\br)$ creates a {\it hole} at position $\br$ in the
bottom layer. Upon a gauge transformation
$c_{\pm}(\br)\rightarrow\exp\left[ie\varphi_{\pm}(\br)\right]c_{\pm}(\br)$,
the order parameter $\Delta(\br)$ will transform as
$\Delta(\br)\rightarrow\exp\left[i\Phi(\br)\right]\Delta(\br)$
with
\beqa
\Phi(\br)\rightarrow\Phi(\br)-e\varphi_{+}(\br)+e\varphi_{-}(\br)
\label{eq: gauge}
\eeqa
where $-e<0$ is the electron charge. We will call the phase of the order
parameter, $\Phi(\br)$, the {\it dipolar phase}. It is ``approximately''
charge-neutral since the condensate is an electron-hole condensate. However,
the electron and the hole operators are always spatially separated: electrons
are in the top layer and the holes are in the bottom layer. Therefore as one
winds the phases of $c_{+}(\br)$ and $c_{-}(\br)$ in the same direction,
since $\varphi_{\pm}(\br)=\varphi(\br\pm\bd/2)$, the phases that enter into
the shift of the dipolar phase $\Phi(\br)$ {\it are not fully compensated}.
This phase shift that enters in the gauge transformation of a nominally
charge-neutral dipolar phase, Eq.(\ref{eq: gauge}), is crucial for the
hydrodynamics of the dipolar superfluid. Now we determine the coupling of the
dipolar phase $\Phi(\br)$ to external gauge potentials in the top and bottom
layers, $\bA_{\pm}(\br)=\bA(\br\pm\bd/2)$. To be consistent with $U(1)$ gauge
transformation
$\bA_{\pm}(\br)\rightarrow\bA_{\pm}(\br)+\nabla\varphi_{\pm}(\br)$ the
dipolar phase must transform as
\beqa
\nabla\Phi(\br)\rightarrow \nabla\Phi(\br)-e\bA_+(\br)+e\bA_{-}(\br).
\label{eq: gradphi}
\eeqa
\begin{figure}[thbp]
\begin{center}
\vspace{1cm}
\includegraphics[width=2.8in]{ehbilayer.eps}
\caption{Schematic bilayer electron-hole system. The electrons and holes
form bound excitons which condense at low densities. We predict that an
in-plane field $\bB_{||}$ will produce equal but opposite currents
$\bJ_{+}$ and $\bJ_{-}$ in the two layers.}
\label{fig: ehbilayer}
\end{center}
\vspace{-8mm}
\end{figure}
The gauge potentials in top and bottom layers enter with opposite signs since
they couple to oppositely charged electrons and holes respectively.
Eq.(\ref{eq: gradphi}) shows that, in contrast to an ordinary neutral
superfluid, {\it the phase of the dipolar superfluid couples to the
difference of the gauge potentials between the two layers.}
The GL energy functional for the dipolar superfluid will depend on the order
parameter $\Delta(\br)$, and for inhomogeneous state should depend only on
the gauge-invariant combinations (\ref{eq: gradphi}) involving gradient of
the dipolar phase $\Phi(\br)$. In particular, the gradient part of free
energy will be given by
\beqa
F=\frac{1}{2}\rho_{d}\int_{\br}\left[\nabla\Phi(\br)-e\bA_{+}(\br)+
e\bA_{-}(\br)\right]^2
\label{eq: glenergy}
\eeqa
where $\rho_d$ is the dipolar superfluid density~\cite{cav}. It follows
that the currents in the top and bottom layers are
\beqa
\bJ_{\pm}(\br)=-\frac{\delta F}{\delta\bA_{\pm}(\br)}=\pm e\rho_d\left[
\nabla\Phi(\br)-e\ba(\br)\right].
\label{eq: sfcurrent}
\eeqa
Here we have introduced the antisymmetric and symmetric combinations of gauge
potentials, $\ba\equiv(\bA_{+}-\bA_{-})$ and
${\cal A}\equiv(\bA_{+}+\bA_{-})$, which couple to the dipolar phase
$\Phi(\br)$ and the total phase
$\varphi_T=\varphi_{+}+\varphi_{-}$ respectively. Eq.(\ref{eq: sfcurrent})
implies that, as far as the dipolar condensate is concerned, there is no net
electric current, $\bJ=(\bJ_{+}+\bJ_{-})=0$. This result is
naturally expected since excitons do not carry any electric or mass current
unless one breaks them apart. On the other hand, the excitonic condensate
{\it does} carry a net dipolar current
\beqa
\bJ_d(\br)=2e\rho_d\left[\nabla\Phi(\br)-e\ba(\br)\right].
\label{eq: dipolarcurrent}
\eeqa
Eq.(\ref{eq: dipolarcurrent}) is the central result of this paper. It has
exactly the same form as a supercurrent in a superconductor,
$\bJ=e\rho_{s}(\nabla\phi-2e\bA)$. Therefore, in analogy with a
superconductor, we expect persistent dipolar currents produced by the
external gauge potential $\ba$. For a smoothly varying gauge potential, the
antisymmetric combination
$\ba(\br)=\bA(\br+\bd/2)-\bA(\br-\bd/2)\approx d\partial_{z}\bA(\br)$ can be
tuned by varying a uniform in-plane magnetic field. To be specific, let us
consider magnetic field $\bB_{||}=-B_{||}\hat{y}$ between the two layers,
generated by gauge potential $\bA(\br,z)=(-B_{||}z,0,0)$. In this case,
assuming the dipolar condensate phase is uniform, we get
\beqa
\bJ_{d} = 2{e^2}\rho_d dB_{||}\hat{x}
\label{eq: pcurrent}
\eeqa
Thus, we predict that {\it a uniform in-plane magnetic field will induce
persistent and opposite currents in the top and the bottom layers in the
direction perpendicular to magnetic field}. This is a direct consequence of
the non-compensation of gauge potential acting on the electron and hole part
of the dipolar condensate.
Alternatively, one can view the dipolar persistent current $\bJ_d$ as arising
from ``perfect diamagnetism'' of electrons and holes. Indeed, turning on the
magnetic field $\bB_{||}$ produces in-plane electric fields $\bE_{\pm}$ which
are equal and opposite. These electric fields accelerate electrons and holes
{\it in the same direction} in respective layers. We emphasize that these
time-dependent electric fields are physical and they cannot be gauged away in
a simply-connected geometry. The condensate phase
stiffness does not allow the resulting current to decay, thus giving rise to
persistent dipolar supercurrent. Since the dipolar phase couples to
$\ba\sim\bB_{||}$ it follows that the curl of the gauge potential
$\nabla\times\ba$, {\it which will introduce vortices in the dipolar phase},
is determined by gradients in $\bB_{||}$ over the length-scale $d$ and is
necessarily small for externally applied fields. In addition, the constraint
$\nabla\cdot\bB_{||}=0$ necessitates other gradients in the magnetic field to
compensate for the gradients which induce vorticity in the gauge potential
$\ba$. Thus creation of vortices in the dipolar phase requires nontrivial
texture in the external magnetic field over very short length-scales. In this
sense, the dipolar superfluid is robust against creation of vortices. This
feature might make the particle-hole condensate more robust against
vortex-induced dissipation in the counterflow channel.
It is straightforward obtain the effective action for dipolar superfluid from
the GL energy functional,
\beqa
{\cal S}=\int_{\br,t}\left[n(\partial_t\Phi-ea_{0})-\frac{\rho_d}{2}
(\nabla\Phi-e\ba)^2-\frac{n^2}{2C}\right],
\label{eq: action1}
\eeqa
where $a_0=(A_{0+}-A_{0-})$ is the difference between electrical potentials
in the two layers, we have approximated the potential energy by a
quadratic term with mass $1/C$ for the number fluctuations, and confined
ourselves to a long-wavelength description~\cite{cav2}. The first
term in the action is standard and results from the commutation relation
between the dipolar condensate number and the phase, $[n,\Phi]=i$.
Integrating out the massive fluctuations leads to the standard effective
action for the phase,
\beqa
{\cal S}_\Phi=\int_{\br,t}\left[\frac{C}{2}(\partial_t \Phi)^2-
\frac{\rho_d}{2}(\bn\Phi)^2\right].
\label{eq: action2}
\eeqa
Eq.(\ref{eq: action2}) implies that the phase fluctuation mode (superfluid
sound mode) is a gapless collective mode with dispersion $\omega_{c}=v_ck$
where $v_c=\sqrt{\rho_d/C}$ is the collective mode velocity. It follows
from the analogy with a neutral superfluid that a state with dipolar
supercurrent $\bJ_d$ will be stable only if the superfluid velocity,
defined by $\bJ_d=2en_d\bv_s=2e\bv_s/(\pi r_s^2)$, is smaller than the
collective mode velocity $v_c$. Since the dipolar supercurrent is linearly
proportional to $B_{||}$, we predict that for magnetic fields greater than
a critical field
$B^{c}_{||}=n_d/(ed\sqrt{\rho_d C})$, the dipolar superfluid state will be
destroyed by collective phase fluctuations. For typical bilayer parameters
($n_d\sim 10^{11}/$cm$^2$) we get the
critical field $B^c_{||}\sim 100$T much larger than typical experimental
field values. We can also define critical current as the dipolar current
which leads to pair-breaking effects in which an exciton gives rise to an
electron and a hole. This current is given by
$J^{c}_{d}=e\rho_d|\Delta|/v_F$ where $v_F$ is the Fermi velocity. Since the
dipolar current is proportional to $B_{||}$, this criterion provides another
bound on the maximum value of the in-plane field
(for typical parameters we get $B_{||}^c\sim 10$T)~\cite{cav2}.
%-----------------------------------------------------------------------------%
%\section{Microscopic Theory}
%\label{sec: micro}
{\it Microscopic Theory:} We briefly describe the derivation of the dipolar
current, Eq.(\ref{eq: dipolarcurrent}), from a microscopic model. The
one-body Hamiltonian for an electron-hole bilayer system in the presence of
gauge potentials is
\beqa
H_{1} & = & \sum_{\bk\bk'}\left[c^{\dag}_{+\bk'}\frac{1}{2m_e}
\left(\bp-e\bA_{+}\right)^2_{\bk'\bk}c_{+\bk}\right.\\ \nonumber &
& +\left.
c_{-\bk}\frac{1}{2m_h}\left(\bp+e\bA_{-}\right)^2_{\bk'\bk}
c^{\dag}_{-\bk}\right],
\label{eq: micro}
\eeqa
where $m_e(m_h)$ is electron (hole) band-mass. The interaction term in the
microscopic Hamiltonian is a sum of the intralayer repulsive interaction,
$V_A(\bk)=2\pi e^2/k$, and interlayer attractive interaction,
$V_E(\bk)=-V_A(\bk)e^{-kd}$. We use Hubbard-Stratonovich transformation to
introduce the order-parameter fields and obtain mean-field equations~\cite{no}.
The attractive interaction leads to the pairing of electrons and holes near
the Fermi surface ($|\bk|\approx k_F$) and a nonzero dipolar condensate order
parameter $\Delta=\langle c^{\dagger}_{+}c_{-}\rangle=\langle c^{\dagger}_{e}
c^{\dagger}_{h}\rangle$. In the absence of external gauge potentials, the
mean-field Hamiltonian is given by
\beqa
H_0=\sum_{\bk\sigma\sigma'}c^{\dagger}_{\bk\sigma}\left[\epsilon_{3\bk}
\tau^{3}+\Re\Delta_{\bk}\tau^{1}+\Im\Delta_{\bk}\tau^{2}\right]
_{\sigma\sigma'}c_{\bk\sigma'}
\label{eq: mf}
\eeqa
where $\sigma,\sigma'=\pm$ is the layer index,
$\epsilon_{3}=(\epsilon_e+\epsilon_h)/2$ is the mean-band-energy and
$\epsilon_{e(h)\bk}=\hbar^2k^2/2m_{e(h)}$ is the electron (hole) dispersion.
In Eq.(\ref{eq: mf}) we have neglected a constant term proportional to
$\epsilon_0=(\epsilon_{e}-\epsilon_{h})/2$ since it vanishes for symmetric
electron-hole bilayers, $m_e=m_h=m^{*}$, and does not qualitatively change
the conclusions. The eigenvalues of the mean-field Hamiltonian are given by
$E_{\pm}(\bk)=\epsilon_{0\bk}\pm\sqrt{\epsilon_{3\bk}^2+|\Delta_{\bk}|^2}$
and the corresponding mean-field Green's function is
$G^{-1}_{0}(\bk,i\omega_n)=(i\omega_n-H_0)$. Now we will consider the
effect of small gauge potential perturbations on the condensate state. To
linear order, the change in the Hamiltonian is
$\delta\epsilon_{e(h)}=\mp e\left[\bp\cdot\bA_\pm+\bA_\pm\cdot\bp\right]
/(2m_{e(h)})$, which leads to the change in the Green's function,
$G^{-1}=G^{-1}_0-\delta G^{-1}$, where
\beqa
\delta G^{-1}=-\frac{e}{2}\left\{\frac{1}{m_e}[\bp,\bA_{+}]_{+}-\frac{1}{m_h}
[\bp,\bA_{-}]_{+}\right\}.
\label{eq: deltag1}
\eeqa
Note that the gauge potentials in the two layers have a relative minus sign
because the charges of the carriers in the two layers are opposite. This is
also consistent with the GL energy functional description in which the
dipolar phase is coupled to the
{\it difference} of gauge potentials in the two layers. The change in the
free energy due to a change in the gauge potential is given by
$\delta F=-\Tr\left(G_0\delta G^{-1}G_0\delta G^{-1}\right)/2$ where ``Tr''
denotes the trace over space-time and layer-index degrees of
freedom~\cite{no}. For symmetric electron-hole bilayers, in the
long-wavelength limit, the matrix $\delta G^{-1}$ in layer-index-space is
given by
$\delta G^{-1}=-e\bk_F\cdot\left[\ba{\bf 1}+{\cal A}\tau^3\right]/(2m^{*})$.
Evaluating the trace over space-time degrees leads to the dipolar superfluid
current $\bJ_d$
\beqa
\bJ_d=\frac{e^2v_F^2}{4}\tr \left[G_0\cdot{\bf 1}\cdot G_0
\left(\ba{\bf 1}+{\cal A}\tau^3\right) \right],
\label{eq: currentd}
\end{eqnarray}
with a similar expression for the total mass current $\bJ$.
Note that only the term proportional $\ba{\bf 1}$ survives in the
trace for the dipolar current $\bJ_d$. This reflects the fact that
the dipolar condensate phase $\Phi$ responds to the antisymmetric
combination of gauge potentials. Since we are expanding around $\bA=0$
and since the effect of $e\ba$ is the same as that of $\nabla\Phi$, only
the gauge-invariant combination enters the expression for the dipolar
current,
\beqa
\bJ_d=\frac{2ev_F^2}{8}\tr[G_0{\bf 1}G_0{\bf1}]\left(\nabla\Phi-e\ba\right)
=2e\rho_d\left[\nabla\Phi-e\ba\right].
\label{eq: dpcurrent}
\eeqa
Thus we recover Eq.(\ref{eq: dipolarcurrent}) with the appropriate definition
of dipolar superfluid density from a microscopic calculation. In general, if
we expand around a nonzero ${\cal A}$, which corresponds to nonuniform
fields, we will also get a nonzero total mass current $\bJ$.
%-----------------------------------------------------------------------------%
%\section{Summary and Conclusions}
%\label{sec: summary}
{\it Discussion:} The possibility of excitonic condensation in semiconductors
has been discussed in the literature for a long time. Recent claims of
observations of such condensates depend primarily on photoluminescence
measurements and do not directly probe their superfluid properties. We have
argued that excitonic condensates in electron-hole bilayer systems will
present a qualitatively different superfluid in which the phase of the
condensate is coupled to the {\it gradient} of the gauge potential. As a
result we predict that such systems will develop persistent dipolar
supercurrent with the application of an in-plane magnetic field. This
supercurrent can be detected via separate contacts to the electron and the
hole layers and will provide a {\it direct} signature of the superfluid
properties of excitonic condensates. As a possible realization of this
experiment in available samples, we propose the study of induced charges and
voltages in each layer in response to an {\it ac} in-plane magnetic field
with frequency $\omega$. In the absence of the excitonic condensate, Faraday
induction will lead to a dipolar current $J_{d}\sim\omega B$ which in turn
will induce equal and opposite charges $Q_{\pm}\sim\pm B$ in the two layers.
In contrast, in the presence of condensate, the dipolar current will be
$J_{d}\sim B$ and therefore the induced charges will be given by
$Q_{\pm}\sim\pm B/\omega$, leading to a very different frequency dependence.
In the current experimental setups, where the excitons are weakly confined in
a parallel trap and recombine by optical emission, our analysis implies that
the position of the spot, where recombinations take place, will oscillate
with an applied oscillating in-plane magnetic field.
We thank Nick Bonesteel, Jim Eisenstein, Mike Lilly, Allan MacDonald, and Kun
Yang for useful discussions. This work was supported by LDRD at Los Alamos.
%-----------------------------------------------------------------------------%
% notes on dimensions used in the paper:
% I have h=1=c. Thus [c]=[c^]=1/length. [\Delta]=1/Area.
% I get [\varphi]=1/q, [\rho_d]=Energy, [A]=1/length.q
% [J]=q/time.length (note the extra \delta-function in df/dA.
% [1/C]=Energy.Area, [v]=[\sqrt{\rho/C}]=Energy.length=length/time.
%-----------------------------------------------------------------------------%
\begin{thebibliography}{99}
\bibitem{kapitsa} P.L. Kapitsa, Nature {\bf 141}, 74 (1938).
\bibitem{landau} L.D. Landau, J. Phys. USSR {\bf 5}, 71 (1941).
\bibitem{moskalenko} S.A. Moskalenko, Fiz. Tverd. Tela {\bf 4} 276 (1962);
J. Blatt {\it et al.}, Phys. Rev. {\bf 126}, 1691 (1962).
\bibitem{keldysh} L.V. Keldysh, in {\it Bose-Einstein Condensation}, edited
by A. Griffin {\it et al.} (Cambridge University Press, Cambridge, 1995) and
reference therein.
\bibitem{bilayer} Yu.E. Lozovik and V.I. Yudson,
Pis'ma Zh. Eksp. Teor. Fiz.{\bf 22}, 556 (1975) [JETP Lett. {\bf 22}, 274
(1975)]; Solid State Commun. {\bf 19}, 391 (1976); Zh. Eksp. Teor. Fiz.
{\bf 71}, 738 (1976) [Sov. Phys. JETP {\bf 44}, 389 (1976)].
\bibitem{rice} T.M. Rice, Solid State Phys. {\bf 32}, 1 (1977);
J.C. Hensel {\it et al.}, {\it ibid.} {\bf 32}, 88 (1977).
\bibitem{shev} S.I. Shevchenko, Phys. Rev. Lett. {\bf 72}, 3242 (1992).
\bibitem{butov} L.V. Butov {\it et al.}, Nature {\bf 417}, 47 (2002),
{\it ibid.} {\bf 418}, 751 (2002); D. Snoke {\it et al.}, {\it ibid.}
{\bf 418}, 754 (2002); C.W. Lai {\it et al.}, Science {\bf 303}, 503 (2004).
\bibitem{em} J.P. Eisenstein, Science {\bf 305}, 950 (2004); J.P. Eisenstein and A.H.
MacDonald, Nature (London), {\bf 432}, 691 (2004).
\bibitem{sivan}U. Sivan {\it et al.}, Phys. Rev. Lett. {\bf 68}, 1196 (1992);
B.E. Kane {\it et al.}, App. Phys. Lett. {\bf 65}, 3266 (1994);
S. Vijendran {\it et al.}, J. Vac. Sci. Technol. B {\bf 17}, 3226 (1999);
M. Pohlt {\it et al.}, App. Phys. Lett. {\bf 80}, 2105 (2002); M. Lily,
{\it private communication.}
\bibitem{loz} Yu.E. Lozovik {\it et al.}, Phys. Scr. {\bf 55}, 491 (1997).
\bibitem{pbl} X.J. Zhu {\it et al.}, Phys. Rev. Lett. {\bf 74}, 1633 (1995);
P.B. Littlewood and X.J. Zhu, Phys. Scr. {\bf T68}, 56 (1996).
\bibitem{palo} S.De Palo {\it et al.}, Phys. Rev. Lett. {\bf 88}, 206401
(2002).
\bibitem{pl} J. Keeling {\it et al.}, Phys. Rev. Lett. {\bf 92}, 176402 (2004).
\bibitem{vig} G. Vignale {\it et al.}, Phys. Rev. Lett. {\bf 76}, 2786 (1996).
\bibitem{cav}Similar action for bilayer quantum Hall systems was derived in
K. Moon {\it et al.}, Phys. Rev. B {\bf 51}, 5138 (1995). However, in the
quantum Hall case, the predictions are different due the response of a
fully-filled Landau level.
\bibitem{cav2}We ignore instabilities associated with finite-momentum
order parameter fluctuations $\Delta_\bk$ because they will not
{\it qualitatively} change our predictions and conclusions.
\bibitem{no} J.W. Negele and H. Orland, {\it Quantum Many-Particle Systems}
(Addison Wesley, New York, 1988).
\end{thebibliography}
%-----------------------------------------------------------------------------%
\end{document}