Coherent Exciton Transport

arXiv:1301.1726v1 [cond-mat.
mes-hall] 9 Jan 2013
Coherent exciton transport in

semiconductors
Massimo Rontani
CNR-NANO Research Center on nanoStructures and bioSystems at Surfaces (S3),
Via Campi 213a, 41125 Modena, Italy
L. J. Sham
Department of Physics, University of California San Diego,
Gilman Drive 9500, La Jolla, CA 92093-0319
Acknowledgements
We thank Leonid Butov for critically reading the manuscript. This work is supported
by EU-FP7 Marie Curie Initial Training Network “Indirect Excitons: Fundamental
Physics and Applications (INDEX)”.
Contents
1 Coherent exciton transport in semiconductors 1

1.1 Introduction 1
1.2 Physical systems 5
1.3 Two-band versus BCS model 10
1.4 Andreev reflection at the interface between excitonic insulator
and semimetal 18
1.5 A perfect insulator 25
1.6 Josephson oscillations between exciton condensates in electro-
static traps 30
1.7 Conclusions 37
References 38
1
Coherent exciton transport in
semiconductors
1.1 Introduction
An exciton is a particle-like neutral excitation of solids and molecules composed of one
electron and one hole bound together by the mutual electrical attraction [134,206,221,
228, 286]. Its creation through internal charge separation is most frequently caused by
the absorption of light and its demise is occasioned by electron-hole recombination,
mostly with emission of light and less frequently non-radiatively. The many-electron
ground state of the system, being an insulator, is immune to excitation until the
excitation energy reaches a threshold G known as the energy gap. When external
influences such as the electromagnetic field and lattice vibrations are ignored, the
exciton may be viewed as a robust state of an excited electron plus the hole which
has been left behind in the valence electron states [230]. The hole acquires its positive
charge from the loss of an electronic charge from the ground state whose total charge
is neutralized by that of the ions in the molecule or solid.
The photon-exciton interaction is responsible for the optical excitation (though
not necessarily in the visible frequency range) of the exciton and for its spontaneous
recombination emitting a photon (a quantum unit of light). The dipole matrix element
responsible for the transition between the energy states is strong when the electron
and hole wave functions overlap in space or match in wave vector. From Planck’s law,
the frequency of the emitting light EX /h is proportional to the energy loss EX in
returning the exciton state back to the ground state, with h being Planck’s constant.
If the constituent electron and hole of the exciton are mostly localized at an ion,
the exciton is localized, but with some probability to hop from site to site. Such a
Frenkel exciton is common in molecules and molecular solids. At the other extreme, if
the electron and hole wave functions are widespread as extended orbitals in a molecule
or Bloch waves in a crystal, their bound state as the exciton can have their center of
mass moving through the system with ease. Such Wannier (or Wannier-Mott) excitons
are most common in broad-band and small-gap semiconductors (a semiconductor is
distinguished from an insulator qualitatively by a smaller energy gap, with the fre-
quency of the emitting light from the exciton spanning the range from visible light to
very far infrared). Wannier excitons resemble the hydrogen atom or, more closely, the
positronium system composed of an electron and a positron. Because of the dielectric
screening of the electrical force in small-gap materials and sometimes the small effec-
tive mass of the electron, the Wannier exciton radius is 10 to 100 times larger than
the positronium radius, which is approximately 0.1 nm.
2 Coherent exciton transport in semiconductors
Excitons, being made of two fermions, behave as bosons on the scale larger than the
exciton radius and therefore may macroscopically occupy a single quantum state [20,
175,40,125,3,124,77]. If the exciton lifetime is long enough to allow for reaching quasi-
equilibrium, the dilute and cold gas of optically generated excitons may undergo Bose-
Einstein condensation (BEC) [50, 87, 198]. The critical temperature for exciton BEC,
of the order of 1 K for typical densities in semiconductors, is basically the temperature
at which the thermal De Broglie wavelength becomes comparable to the average inter-
exciton separation. The possibility of achieving BEC of excitons by shining light on
solids has been thoroughly investigated in the last fifty years (see the reviews [99, 178,
87, 157, 176, 143, 237, 30, 31, 32, 112, 240]). Semiconductors are particularly appealing
for this goal as they may provide excitons with a a lifetime (hundreds of ns in bilayer
structures [30]) longer than the time required for cooling.
In an indirect-gap semiconductor such as silicon, where the momentum of the
exciton does not match that of the photon, the excitons are generally formed after
relaxation of optical excitations with initial energy much higher than the gap. The
indirect exciton has a long lifetime because its recombination with the emission of a
photon requires the conservation of momentum to be satisfied by the assistance of a
lattice vibration or trapping by a defect. Consequently, the excitons have time to form
a large pool known as an electron-hole drop (see the reviews [209, 122]). Alternatively,
the delay in optical recombination may be due to the symmetry of the crystal, as in the
direct-gap oxide Cu2 O which has conduction and valence bands of like parity hence
the optical dipolar transition is forbidden [110]. The chapter by Kuwata-Gonokami in
Volume 1 focuses on the aspects of Bose-Einstein condensation of optically generated
excitons in semiconductors.
In a direct-gap material, the spatial separation of the electron and hole can be en-
forced by housing them in two layers sufficiently close to maintain their electric attrac-
tion [232,160,71]. The recombination of such indirect excitons may then be controlled
by changing the electron and hole wave function overlap with an electric field [5, 300].
An interesting phenomenon is the laser spot excitation of these indirect excitons, lead-
ing to the formation of two concentric luminous circles centered at the laser spot plus
other localized bright spots randomly placed between the circles. Whereas the forma-
tion of the inner ring [34] is due to the migration of indirect excitons away from the
laser spot as optically inactive excitons, the localized bright spots [34] as well as the
outer ring [34,238,37,204] form on the boundaries between electron-rich and hole-rich
regions. At low temperature the outer ring is a necklace of evenly spaced bright spots,
whose origin is not fully understood [103]. This system of excitons in a double quantum
well is considered a good candidate for condensation. These experiments and related
work in double quantum wells are reviewed in [237, 30, 31, 32, 256, 112].
Another possible—and elusive—mechanism of condensation of excitons as bosons is
that excitons form spontaneously at thermodynamic equilibrium even in the absence
of an optical excitation. Such process signals the transition to a permanent phase
known as excitonic insulator (EI) [177, 134, 123, 55, 114], which is originated by the
instability of the normal ground state of either a semiconductor or a semimetal against
the spontaneous formation of bound electron-hole pairs. The wave function of the
strongly correlated EI ground state is formally similar to that proposed by Bardeen,
Introduction 3
Cooper, and Schrieffer for superconductors [16]. As a matter of fact, both excitons
and Cooper pairs are absent except as fluctuations in the normal high-temperature
phase and form only in the ordered, low-temperature phase—respectively the EI and
the superconductor. Besides, both condensation of excitons and that of Cooper pairs
are best described in the reciprocal space of the crystal solid. The EI phase is reviewed
in [96, 135, 99, 50, 186, 87, 157, 176, 237, 156, 112, 240].
It is intriguing to observe that condensation of other types of bosons composed of
two fermions leads to spectacular manifestations of quantum mechanical coherence,
such as the superfluidity ensuing from the pairing of 3 He atoms [149], Fermi alkali
atoms confined in optical traps [21, 80], nucleons in neutron stars [79, 196], the super-
conductivity induced by Cooper pairs in metals [54], and the non classical momenta
of inertia in nuclei [169, 23]. The above phenomena may regarded as distinct realiza-
tions of superfluidity, associated to the coherent, dissipationless flow of charge and /
or mass. However, excitons are neutral and stay dark unless recombine radiatively, as
shown in Table 1.1, which compares the distinct features of the condensates made of
composite bosons. The signature of the macroscopic order of the exciton condensate
is, at present, controversial for the superfluid transport but its other manifestations
will be discussed next.
The aim of this Chapter is to illustrate some recent theoretical proposals con-
cerning the detection of coherent exciton flow [212, 213, 214]. The reader may refer to
the literature reviewed in Sec. 1.2 for a discussion of the conceptual and experimen-
tal difficulties inherent in the realization of exciton condensates. Here we set aside
such difficulties and adopt in a pedagogical way the simplest mean-field description
of the condensate, on which we lay our theoretical development in order to detect the
transport properties of the exciton condensate.
In particular, we focus on the exciton analogues of two phenomena, i.e., Andreev
reflection and Josephson effect, which are hallmarks of superconducting behavior, and
stress the crucial differences between excitons and Cooper pairs. Our first main conclu-
sion is that the excitonic insulator is the perfect insulator in terms of both charge and
heat transport, with an unusually high resistance at the interface with a semimetal—
the normal phase of the condensed state. Such behavior, which should be contrasted
with the high electrical conductance of the junction between superconductor and nor-
mal metal, may be explained in terms of the coherence induced into the semimetal by
the proximity of the exciton condensate. Then we show that the exciton superflow may
be directly probed in the case that excitons are optically pumped in a double-layer
semiconductor heterostructure: we propose a correlated photon counting experiment
for coupled electrostatic exciton traps which is a variation of Young’s double-slit ex-
periment.
We last mention that, due to the interaction between electrons and light, not only
can an exciton decay irreversibily into a photon or vice versa, but it can also exchange
roles with the photon in a quantum-mechanically coherent fashion. Thus, the exciton
may exist in the solid in the superposition state of an exciton and a photon, known as
polariton. Whereas the photon energy varies linearly with its momentum at the speed
of light in the vacuum, the exciton energy depends on the square of its center-of-mass
momentum. For small momenta, the exciton and the photon can approximately match
Table 1.1 Excitonic insulator (EI) versus superconductor à la Bardeen-Cooper-Schrieffer (BCS).

The interface referred to in the Table is the junction between normal and condensed phase. For a
general discussion of the condensates made of composite bosons see [136]. For specific EI features
see [114] (Meissner effect), [297] (superconductivity), [298] (superthermal conductivity), [212, 213]
(Andreev reflection), and [214] (Josephson oscillations).
Physical property Excitonic insulator BCS-like superconductor
Nature of the composite boson Exciton Cooper pair

Boson charge Neutral 2e
Boson momentum Crystal momentum Crystal momentum
(commonly ignored in the free
electron gas approximation)
Boson mass Effective mass Effective mass of the electron
quasiparticle in the Fermi level
region (of thickness provided by
phonon Debye frequency)
Type of long-range order Diagonal Off-diagonal
Superfluidity ? Superconductivity
Meissner effect No Yes
Superthermal conductivity No No
Nature of the quasiparticle Electron (hole) Bogoliubon
Andreev reflection Yes Yes
Interface electric conductance Decreased Increased
Interface thermal conductance Decreased Decreased
Proximity effect Yes Yes
Josephson oscillations Yes Yes
both their momentum and energy values, the coupling mixing the two states into two
superpositions of photon and exciton with an energy splitting. Thus, the massless
photon is slowed down by the massive exciton by virtue of the quantum-mechanical
superposition. The Chapter by Yamamoto in Volume 1 deals with aspects of polariton
condensation.
The structure of this Chapter is the following: After a review of previous work
(Sec. 1.2), in Sec. 1.3 we illustrate the mean-field theory of the EI emphasizing its
relation with the BCS theory of superconductors. We then introduce the phenomenon
of Andreev reflection in Sec. 1.4 and analyze its observable consequences in Sec. 1.5.
Section 1.6 on the Josephson effect ends the Chapter.
Physical systems 5
1.2 Physical systems

This section briefly reviews recent theoretical and experimental works on exciton con-
densation, focusing on diverse physical systems. Without attempting an exhaustive
review, we refer the reader to more comprehensive essays whenever available.
1.2.1 Bose-Einstein condensation of optically generated excitons

The pursuit of Bose-Einstein condensation of optically generated excitons in semi-
conductors, which dates back to the sixties, presently focuses on both classic systems
such as Cu2 O and novel low dimensional structures (for reviews see [99, 178, 87, 157,
176, 143, 237, 120, 30, 31, 156, 32, 256, 112, 240, 141, 224]). A very active field concerns
“indirect” excitons. Such excitons are made of spatially separated electrons and holes,
hosted in two quantum wells that are sufficiently close to maintain electrical attrac-
tion between the carriers of opposite charge. This setup has several advantages: (i)
The overlap of electron and hole wave functions is controlled by applying an elec-
tric field along the growth direction of the bilayer heterostructure, thus increasing
the exciton recombination time by orders of magnitude with respect to the single-
well value [5, 300]. (ii) The confinement effect along the growth direction increases
the exciton-phonon scattering rate, improving exciton thermalization [292]. (iii) The
dipolar repulsion among indirect excitons disfavors the formation of biexcitons and
electron-hole droplets [38,156,246,222,267,148,269,47] as well as effectively screens the
in-plane disorder potential [111,220,102,107,208,7]. (iv) As the electric field parallel to
the growth direction may be laterally varied using suitably located electrodes, one may
tailor the in-plane effective potentials for excitons, thus realizing artificially controlled
traps [109,97,44,86,75,102,107,223,104,7,8], ramps [94,74], lattices [295,296,208,207],
“exciton circuits” [101, 106, 88], and “exciton conveyers” [279].
Exciton traps may also be created by means of the uncontrolled in-plane disor-
der of the double quantum well [299, 36, 34, 102, 107], the strain experienced by the
heterostructure [257, 119, 183, 179, 268, 285], the laser-induced confinement [98, 9], the
magnetic field [46]. The realization and control of exciton traps is a key capability to
reach exciton BEC: As the long range order in two dimensions is smeared by quan-
tum fluctuations, a weaker requirement for the macroscopic occupation of the lowest
exciton level is that the exciton coherence length exceeds the trap size [31].
The present evidence of exciton BEC is based on distinct features of the emit-
ted light (photoluminescence, PL) that appear at low temperature: (i) The PL dy-
namics exhibits bosonic stimulation of the scattering of hot optically dark excitons
into optically active low-energy states [35]. (ii) The PL signal becomes noisy in a
broad range of frequencies, as it occurs in the presence of coherence [38, 139, 142].
(iii) The degree of polarization of the emitted light increases with decreasing tem-
perature [144, 145, 103], consistently with gauge symmetry breaking. (iv) The exciton
mobility is enhanced, which may be attributed to superfluid behavior [33]. (v) The
radiative decay rate increases, which may be explained in terms of “superradiance” of
a macroscopic dipole [33] or collective behavior at the onset of condensation [144,145].
(vi) The PL lineshape narrows and departs from the Maxwell-Boltzmann distribu-
tion [139, 144, 145, 146], as it may be expected for the macroscopic population of a
single exciton state.
However, some of the signatures [110, 239, 155, 72] listed above, taken separately,
may have different explanations than exciton condensation [187, 188, 118] (for a dis-
cussion see Ref. [31]). The most compelling evidence of BEC is probably the direct
measure of coherence through interferometric techniques [282,256,103,104,227,105,6],
which accesses the macroscopic exciton wave function in real space. On the theory
side, the light emitted by excitons just after the onset of condensation is predicted to
be coherent [190, 68, 189], with a sharply focused peak of radiation in the direction
normal to the quantum-well plane [121, 294]. Besides, the instability leading to the
external ring of evenly placed bright spots discussed in the introduction [34, 37] is
possibly linked to exciton quantum degeneracy [153].
An intriguing issue is the role played by spin [165] in exciton condensation. In bilay-
ers, the exciton spin is the component Jz of the angular momentum along the direction
perpendicular to the planes, discriminating between optically active (Jz = ±h̄) and
inactive states (Jz = ±2h̄). The most urgent questions concern the multi-component
nature of the condensate [67,283], the possibility of dark-exciton condensation [48,49],
the role of spin-orbit coupling [95, 39, 281, 233]. This research is fueled by the recent
experimental evidence that the spin-relaxation time of indirect exciton is long and con-
sequently exciton spin transport is long-ranged [152,138], as well as that spin textures
and polarization vortices appear together with the onset of long-range coherence [103].
1.2.2 Excitonic insulator in mixed-valence semiconductors

In principle, any intrinsic semiconductor that may be turned into a semimetal, either
by applying stress or by suitable alloying, may undergo a transition to the permanent
EI phase. Favorable conditions are the presence of an indirect gap, which weakens the
detrimental effect of dielectric screening on the exciton binding, as well as the nesting of
electron and hole Fermi surfaces, which maximizes electron-hole pairing. Nevertheless,
early experiments focusing on simple materials, such as divalent fcc metals (Ca, Sr,
Yb) and group V semimentals (As, Sb, Bi) were unable to confirm the existence of the
EI. References [96, 135, 99, 50, 186, 87, 157, 176, 156, 112] review the work on the EI.
Recently, a few experiments have pointed to the realization of the EI phase in
mixed-valent semiconductors. The first class of candidate materials consists in rare-
earth chalcogenides, such as TmSex Te1−x [28,272], Sm1−x Lax S [272,270], Sm1−x Tmx S,
YbO and YbS [270]. These compounds all crystallize in the NaCl structure and undergo
a semiconductor-semimetal transition as the band gap G is changed from positive to
negative values by applying high hydrostatic pressure to the sample.
When the direct gap of TmSe0.45 Te0.55 , formed between the localized 4f 13 levels
and the 5d conduction-band states, is closing with external pressure, an indirect band
gap develops between the highest valence Tm 4f 13 level Γ15 at the Γ point and the
mimimum of the ∆2′ conduction band 5d states at the X point of the Brillouin zone.
As the otherwise localized 4f band is broadened and shows a maximum at Γ due to
p(Se,Te)-f (Tm) covalent hybridization [113], it is tempting to use a simple two-band
model for interpretation, similar to the one ilustrated in Sec. 1.3. On the basis of low-
temperature resistivity and Hall mobility measurements, the authors of [28] attribute
the resistivity increase with the vanishing gap to a condensation of free carriers into
excitons, placing the EI phase between semimetal and semiconductor, close to G ≈ 0.
Physical systems 7
Later, the same group has reported a linear increase of thermal conductivity and
diffusivity with decreasing temperature and attributed it to exciton superfluidity [271].
Fehske and coworkers [26,27,289] have suggested theoretically that the EI phase in the
pressure-temperature phase diagram is narrower that the experimental claim, being
surrounded by a “halo” regions made of preformed excitons coexisting with the normal
semiconductor phase. The presence of this halo, precursor of the EI, explains the
experimental findings and rules out the idea of a heat supercurrent, which conflicts
with the general argument [114, 298] that a flowing condensate carries no entropy and
thus no heat.
Other candidate systems for the EI phase are the transition-metal chalcogenides
TiSe2 [217, 278, 258, 166, 247, 10, 195, 131, 43, 205, 174, 264, 173, 167, 42], Ta2 NiSe5 [274],
TaSe2 [252], and the possibly ferromagnetic EI GdI2 [18, 253] (see Ref. [216] for a
review). The main evidence relies on the hole quasiparticle band structure, as extracted
from angular-resolved photoemission [258, 166, 247, 10, 195, 131, 43, 205, 174, 167, 274].
The much studied TiSe2 , at a critical temperature of around 200 K, develops a charge
density wave [217] which does not fit the standard model based on Fermi surface
nesting [90, 216] but it is consistent with the presence of an EI (see Sec. 1.3). In fact,
the spanning wave vector of the charge density wave is the distance in reciprocal space
between Ti 3d-electrons and Se 4p-holes, which are bound by Coulomb attraction.
Therefore, the excitonic instability drives the charge density wave and may possibly
couple with a periodic lattice distortion [247,195,43,174,173,167,42], though alternate
scenarios [131,264,216] have been suggested. Recent time-resolved photoemission data
link the artificially induced collapse of the charge-ordered TiSe2 state to screening due
to transient generation of free charge carriers, supporting the excitonic origin of the
phase transition [210, 100].
A third class of candidate systems consists in Kondo insulators [2] and heavy-
fermion materials [147], which are mixed-valence semiconductors characterized by a
flat f -type valence band plus a dispersive—say d-type—conduction band, typically
exhibiting strongly correlated behavior. Such systems (e.g. SmB6 ) are often modeled
by the Falicov-Kimball Hamiltonian, which takes into account the strong inter-band
Coulomb interaction [66,200,211,215]. Sham and coworkers have shown [201,200] that
the exciton condensate made of f holes and d electrons may spontaneously break the
lattice inversion symmetry and lead to a ferroelectric phase transition of electronic
origin, whereas conventional ferroelectricity is associated to lattice distortion [133].
The predicted experimental signatures, supported by some evidence [273, 83], include
the divergence of the static dielectric constant, a ferroelectric resonance in the mi-
crowave absorption spectrum, and a nonvanishing susceptibility for second-harmonic
generation.
If intraband hybridization dominates over Coulomb interaction, then the exciton
condensate wave function acquires a different type of symmetry—p-wave—which ex-
cludes the ferroelectric scenario but allows the coupling with the lattice [60]. In this
latter case the excitonic instability manifests itself as a spontaneous lattice deformation
which may explain some of the phase transitions known as ferroelastic [29].
1.2.3 Permanent exciton condensation in bilayers
In order to investigate permanent exciton condensation in semiconductor bilayers, one

strategy is to host electrons in the first layer and holes in the second layer [232, 160].
This task is nowadays accomplished by means of suitable electric gates which allow
to separately contact the layers [235, 116, 199, 127, 202, 251]. The spacer beetween the
two quantum wells suppresses the inter-layer tunneling which induces exciton recom-
bination, but it is sufficiently thin to provide strong inter-layer Coulomb interaction
(see Ref. [92] for a recent review). This setup allows for measuring the Coulomb drag
resistance, which is the inverse ratio of the electric current measured in one layer to
the open-circuit voltage developed in the other layer in turn. Such drag resistance is
predicted to diverge in the presence of exciton condensation, as the exciton binding
correlates the motion of carriers in the two layers [266, 108, 115]. Recent measure-
ments [52, 225, 51] point to low-temperature anomalies in the Coulomb drag which
may originate from an excitonic instability, though other strongly correlated phases
are possible [92].
An alternate strategy is to place electrons in both layers in the presence of the
magnetic field (see [63, 64, 259] for reviews). The field bends classical electron trajec-
tories into circular cyclotron orbits. As such orbits may be placed all across the plane,
overall their quantized energies consist in highly degenerate “Landau levels”. Since
the level degeneracy is the number of quanta of magnetic flux that cross the plane,
for sufficiently high fields and identical layers the lowest Landau level in each layer
will be half filled by electrons (single layer filling factor ν = 1/2, total filling factor
νT = 1). Note that in this quantum Hall effect regime, routinely detected through the
quantization of the Hall resistance, Landau levels may be considered either half filled
or half empty. Therefore, one may switch to the excitonic parlance [140,284], regarding
one layer as filled by electrons and the other one by holes: In this picture the exciton
“vacuum” has the lowest Landau level totally filled in one layer (ν = 1) and empty in
the other layer (ν = 0), thus excitons are created by moving electrons from one layer,
which leaves a hole behind, to the other one [164].
There is significant evidence, based on low-temperature transport experiments,
that the bilayer ground state is a condensate of excitons. The first hint is a huge
enhancement of inter-layer tunneling solely due to many-body effects, clearly pointing
to strong inter-layer coherence [244]. The most compelling observations are based on
counterflow measurements [232, 160, 248], where the electric currents of opposite sign
and like magnitude that flow in the two layers provide zero total electric current and
a net exciton flow. For filling factors other than νT = 1 the Hall voltages separately
measured in the two layers are equal and opposite in sign, whereas for νT = 1 they
both drop to zero, consistently with the flow of an uncharged object such an exciton
[126, 260, 277]. To prevent edge states—always present at the boundary of quantum
Hall systems and unrelated to excitons—from playing a role in transport, the Coulomb
drag has been recently measured in the “Corbino” annular geometry, confirming the
excitonic nature of transport [255, 254, 69, 180], whereas the superfluid character of
the exciton flow is unclear. The above scenario is supported by the measurement of
quasiparticle and collective excitations by means of tunneling [245, 81] and inelastic
light scattering [161, 117] spectroscopies.
Physical systems 9
Interesting theoretical predictions concern the response of the bilayer exciton con-
densate to external electromagnetic fields [15, 115, 243, 62] and impurities [61], as well
as the transport properties of hybrid circuits including exciton condensates and super-
conductors [57, 192, 242]. For weak inter-layer interaction or filling factors other than
νT = 1, bilayers are predicted to undergo phase transitions to other strongly correlated
phases, such as paired two-dimensional Laughlin liquids and Wigner solids [284], or
peculiar excitonic charge density waves [45].
1.2.4 Graphene-based systems

Graphene—a recently discovered allotrope of sp2 bonded carbon—is a one-atom thick
two-dimensional honeycomb lattice [76, 185, 219, 137]. Its peculiar electrical and me-
chanical properties—chemical stability, high mobility, easiness of making electric contacts—
have stimulated observations by means of different electron spectroscopies and scan-
ning probes. Intensive investigations have uncovered new physics (e.g. Klein tunnelling,
anomalous types of quantum Hall effect), rooted in the unusual character of quasipar-
ticle excitations, that, in the neighborhood of the Fermi energy, are massless chiral
Dirac fermions. In fact, conduction and valence bands form specular cones whose
apexes touch in the two inequivalent points K and K′ , located at the corners of the
hexagonal two-dimensional Brillouin zone. These two points, which map into each
other by a rotation of 2π/6 [89], are the Fermi surface of the undoped system, hence
graphene is a zero-overlap semimetal.
In principle, graphene is a good candidate system for EI, since: (i) the density of
states vanishes at the charge neutrality point, hence the long-range Coulomb interac-
tion is unscreened (ii) the perfect electron-hole symmetry of Dirac cones favors the
nesting of electron and hole isoenergetic surfaces. Khveshchenko [130] was the first to
suggest that graphene hides a latent excitonic insulator instability. The EI phase is a
charge density wave alternating between the two inequivalent triangular sublattices,
its spamming wave vector connecting K and K′ in reciprocal space. A stack of graphite
layers in a staggered (ABAB...) configuration, with the atoms located in the centers
and corners of the hexagons in two adjacent layers, respectively, could stabilize the EI
by enforcing interlayer Coulomb interaction.
After this seminal prediction, many theoretical works have tried to estimate the size
of the EI transport gap as well as the stability of the EI phase (see for example [262,
58,73,84,275] and references therein as well as the reviews [184,137,226]). The absence
of consensus is not surprising, as the many-body problem in graphene is presently an
open issue [137, 261]. Experiments show that electrons in graphene allegedly behave
as non-interacting particles [76, 185, 219], except for small effects related to velocity
renormalization [65], coupling with phonons [76,185,219] / plasmons [25] (here we are
not concerned with the fractional quantum Hall effect [59,24], induced by the magnetic
field). Therefore, if the EI energy gap ever exists, it must be smaller than the present
spectroscopic resolution.
A related theoretical proposal concerns permanent exciton condensation in double-
layer graphene [159, 56]. The idea is to separately contact the two layers, which are
spaced by a dielectric medium, in order to induce the same quantity of charge with
opposite sign in the two layers. With respect to the double-layer made of usual semi-
conductors mentioned in Sec. 1.2.3, here the advantage is the smaller value of the
transverse electric field required to polarize the bilayer, due to the zero energy gap of
graphene. The estimate of the Kosterlitz-Thouless temperature required to undergo
the EI phase is debated theoretically [170, 290, 128, 19, 78, 129, 172, 158, 241, 250, 194];
recent Coulomb drag measurements [132, 85] point to the importance of inter-layer
interactions.
In the absence of a dielectric spacer, undoped bilayer graphene is predicted—among
other proposals—to undergo an excitonic ferroelectric phase that spontaneously breaks
which-layer symmetry and polarizes the layers in charge [171, 181, 291]. The excitonic
instability, which opens an energy gap, appears to sensitively depend on the interaction
range [91]: in the case of finite range, the expected electronic phase is nematic and
gapless [151, 263] (see reviews [163, 182]). The experimental observation of a transport
gap at the charge neutrality point is controversial [276, 168, 265, 70].
1.3 Two-band versus BCS model

In this section we contrast the mean-field theory of the EI to the BCS theory of
superconductors. We compare the equations of two model junctions, (i) one between EI
and semimetal (SM-EI), (ii) the other one between superconductor and normal metal
(N-S). In both cases the phase boundary is due to the variation of the order parameter
that changes along the direction perpendicular to the interface, tending respectively to
zero in the bulk normal phase and to a constant value inside the condensed phase. The
quasiparticle amplitudes for both SM-EI and N-S junctions are formally identical and
are used in Section 1.4 to compute the flow of charge and heat through the junction.
1.3.1 The SM-EI junction

We start studying the junction between semimetal (SM) and EI on the basis of a
spinless two-band model. The SM has overlapping isotropic conduction and valence
bands (b and a, respectively) of opposite curvature and one electron per unit cell,
hence the Fermi surface is a sphere in momentum space, located at the nesting of the
two bands; since there are N electrons and 2N states available, the nesting occurs at
zero energy in Fig. 1.1. One may turn the SM into an EI by either changing the SM
stoichiometric composition through suitable alloying or applying stress, which opens
a gap of size 2∆ in the bulk EI in virtue of the strong inter-band Coulomb interaction
[cf. right panel of Fig. 1.1(a)]. The variation of the EI order parameter ∆(r) (defined
below) along the coordinate z normal to the interface determines the effective interface
potential, as shown in Fig. 1.2(a).
The Hamiltonian of the SM-EI junction is
HSM-EI = H0 + H1 + H2 . (1.1)
Here H1 is the kinetic term which embodies the effect of the ideal and frozen crystal
lattice on electrons with the envelope function in the effective mass approximation
[162]:
XZ
H1 = d r ψi† (r) εi (r) ψi (r) . (1.2)
i=a,b
Two-band versus BCS model 11
B A I D C
Energy (a.u) ω
0 + +
2∆
- - +
-q q q -k k
SM EI
a b
-kF 0 kF -kF 0 kF
Crystal wave vector kz (a.u.)
(a)
B I 1
0
C
1
0
1
0 1
0
1
0 1
0
1
0 0
1 1
0
1
0
0
1
A D
kz
a kx
b
(b)
Fig. 1.1 Junction between semimetal (SM, left) and excitonic insulator (EI, right). (a) Quasi-
particle energy ω vs wave vector kz . The labels mark the allowed elastic scattering channels for
an incoming electron (labeled I) with kz = q + . In particular, A is the interband (Andreev)
reflection, B the intraband reflection, C the intraband transmission, and D the interband
transmission. The size of the EI gap is 2∆. (b) Isoenergetic contour lines in the (kx , kz ) space
for the energy ω shown in panel a. The arrows point to the group velocities of electrons in
the different scattering channels.
The field operator ψa (r) [ψb (r)] annihilates an electron in the valence (conduction)
energy band at the position r in space. The real-space band operators εi (r) appearing
in Eq. (1.2) take the form:
εa (r) = G/2 + (2m)−1 ∇2 ; (1.3a)

εb (r) = −G/2 − (2m)−1 ∇2 . (1.3b)
Here m is the (positive) effective mass, G is the (positive) band overlap, and the
energies are measured from the Fermi surface [114]. Throughout this work we put
h̄ = 1 and assume that the system has unit volume. The valence- and conduction-
band energy levels of the non-interacting bulk crystal, eigenvalues of H1 , are
εa (ka ) = G/2 − (2m)−1 ka2 , (1.4a)

εb (kb ) = −G/2 + (2m)−1 kb2 , (1.4b)
where ka and kb refer to the respective band extrema. We assume the valence-band
has a single maximum at k = 0 whereas the conduction-band a single minimum at
k = w, and ignore complications due to the presence of equivalent extrema. The Fermi
wave vector is given by kF2 = mG. The two-body term H2 consists of the inter-band
Coulomb interaction,
Z
H2 = d r d r ′ ψa† (r) ψb† (r ′ ) V2 (r − r ′ ) ψb (r ′ ) ψa (r) , (1.5)
with V2 (r) being the dielectrically screened Coulomb potential [123]. Renormalization
effects due to intra-band Coulomb interaction and temperature dependence are taken
into account into the energy band structure (1.4). The one-body term H0 is the sum
of two parts,
H0 = V + Vhyb . (1.6)
V is the intra-band term,
XZ
V= d r ψi† (r) V (r) ψi (r) , (1.7)
i=a,b
which includes the effects of the band offset as well as those of possible impurities and
defects at the interface, such as a thin insulating layer, via the single-particle potential
V (r). The potential V (r) can also include the effect of a voltage bias applied to the
junction in a steady-state regime. The inter-band term,
Z
Vhyb = d r ψb† (r) Vhyb (r) ψa (r) + H.c., (1.8)
hybridizes b and a bands by means of the potential Vhyb (r). This term may be orig-
inated e.g. by the change of an element in the SM compound. The influence of the
potential Vhyb (r) on exciton condensation, which depends on the symmetry of the
bands involved, is discussed in Refs. [201, 200, 60].
Both the hybridization potential and the EI order parameter contribute additively
to the small band gap formed between the two overlapping bands [cf. right panel in
Fig. 1.1(a)] hence their separate contributions cannot be distinguished by spectroscopy,
as it was early recognized [114, 136, 93]. However, in this Chapter we show that the
transport properties across the SM-EI junction are distinctive features of the EI. The
key point is in the length scale of the variation of the effective interface potential
which reflects or transmits the electron. A component of the effective potential is the
position dependent order parameter ∆(r), which decreases from the bulk value in the
EI region to zero in the SM region [see Fig. 1.2(a)]. The length scale of the change
is the coherence length in the EI, much longer than the lattice constant. Since this
scenario is most likely if the lattices of the two components are as similar as possible,
we classify as homogeneous the junction with ∆(r) being the only contribution to the
interface potential.
On the other hand, the one-electron interface potential Vhyb (r) due to the change
in hybridization has an abrupt variation on a length scale of the order of a few atomic
layers, thus considered in our context as a heterogeneous junction (this includes the
case of a Schottky barrier). We have studied the common physical features of the
heterogeneous junction, including the abrupt band edge discontinuity, the short-ranged
interface potential, and the impurities at the interface. Whereas the charge carriers in
the heterogeneous junction experience uninteresting intraband reflection, electrons in
the homogeneous junction change valley when backscattered as a feature of the EI band
mixing. In the rest of the Chapter we focus on the homogeneous junction and refer
the reader elsewhere [212] for the study of the heterogeneous junction (corresponding
to the case Vhyb (r) 6= 0).
Following Sham and Rice [230], we introduce the electron quasi-particle amplitudes,
f (r, t) = hΨ0 |ψ̃b (r, t)|Ψek i, (1.9a)

g(r, t) = hΨ0 |ψ̃a (r, t)|Ψek i. (1.9b)
Here |Ψ0 i and |Ψek i are the exact interacting ground states with N and N +1 electrons,
respectively; the quantum index k labeling the electron quasiparticle means the crystal
momentum only in the bulk phase as the overall translational symmetry is destroyed
by the presence of the junction. States and operators are written in the Heisenberg
representation [1] (flagged by the tilde symbol on operators):
ψ̃i (r, t) = exp(i [H − µ N ] t) ψi (r) exp(−i [H − µ N ] t) , (1.10)
where µ is the chemical potential (here µ = 0 due to electron-hole symmetry) and the
number operator N is defined by
X Z
N = d r ψi† (r) ψi (r) . (1.11)
i=a,b
Writing down the Heisenberg equations of motion for the operators ψ̃i (r, t), exploit-
ing the mean-field approximation to express them in a closed form, and neglecting
unessential intra-band Hartree terms, we derive a set of two coupled integro-differential
equations for the amplitudes f (r, t) and g(r, t):
∇2 k2

∂f (r, t)
Z
i = − − F + V (r) f (r, t) + dr ′ ∆(r, r ′ ) g(r ′ , t) , (1.12a)
∂t 2m 2m
2
∇ k2

∂g(r, t)
Z
i = + F + V (r) g(r, t) + dr ′ ∆∗ (r ′ , r) f (r ′ , t) . (1.12b)
∂t 2m 2m
The built-in coherence of the exciton condensate, ∆(r, r ′ ), appearing in Eqs. (1.12)
for k > kF is defined as
∆(r, r ′ ) = V2 (r − r ′ ) hΨ0 |ψ̃b (r) ψ̃a† (r ′ ) |Ψ0 i. (1.13)
Except for the factor V2 (r − r ′ ), ∆(r, r ′ ) is the exciton macroscopic wave function.
In fact, hΨ0 |ψ̃b (r) ψ̃a† (r ′ ) |Ψ0 i is the average on the many-electron ground-state of the
operator destroying an electron-hole pair, i.e., one b-band electron at r and one a-band
hole at r ′ [the electron creation operator ψ̃a† (r ′ ) may be regarded as a hole destruction
operator]. Such average is zero in the SM phase, since for k > kF b-band levels are
empty and a-band levels filled, but it acquires a finite value in the EI phase. Besides,
the finiteness of ∆(r, r ′ ) reflects the new periodicity in real space of the EI phase, as
the electron density shows an additional charge-density-wave-like order characterized
by the wave vector w displacing the extrema of a and b bands [135]. For k < kF the
roles of electrons and holes are exchanged hence the definition (1.13) of ∆(r, r ′) is
modified accordingly.
The built-in coherence ∆(r, r ′ ) generically depends on both center-of-mass and
relative-motion coordinates, but inside the EI bulk the center-of-mass part of the
condensate wave function is a plane wave with zero momentum, hence ∆ depends
only on the relative coordinate r − r ′ . We expect ∆(r − r ′ ) to smoothly vanish when
|r − r ′ | becomes larger than the characteristic length, the exciton radius. This allows
us to easily find the bulk solution of the system of eqns (1.12a-1.12b) [V (r) = 0] in
terms of the two-component plane wave

fk (r, t) uk i(k·r−ωt)
= e , (1.14)
gk (r, t) vk
with energy
q
2
ω(k) = ξk2 + |∆k | , (1.15)
where ξk = k 2 − kF2 /(2m) and ∆k is the Fourier component of ∆(r). The amplitudes

are normalized as

2 1 ξk 2 2
|uk | = 1+ , |uk | + |vk | = 1, (1.16)
2 Ek
and the relative phase between uk and vk is given by
uk ∆k
= . (1.17)
vk ω(k) − ξk
When ∆k = 0, the amplitude (1.14) is the trivial solution with uk = 1 and vk = 0,

i.e., a conduction-band plane wave. When excitons form a condensate, solution (1.14)
is admissible only if the self-consistency condition derived by the definition of ∆(r)
is satisfied. This condition, which can be easily obtained from Eq. (1.13), is formally
analogous to the BCS gap equation:
X V2, k−p ∆p
∆k = , (1.18)
p
2ω(p)
with V2, k being the Fourier component of V2 (r).

In general, the amplitudes f (r, t) and g(r, t) are the position space representation
of the stationary electron-like elementary excitation across the whole junction. Taken
together, they signify the wave function of the quasiparticle: f (g) is the probability
amplitude for an electron of belonging to the conduction (valence) band. They satisfy
the normalization condition
Z h i
2 2
d r |f (r, t)| + |g(r, t)| = 1, (1.19)
and have always positive excitation energy ω due to the definitions (1.9-1.10). The
probability current density J(r, t) can be found starting from the definition of the
probability density ρ(r, t) for finding either a conduction- or a valence-band electron
2 2
at a particular time and place, ρ(r, t) = |f | + |g| . After some manipulation of the
equations of motion (1.12), one derives the continuity equation
∂ρ
+ ∇ · J = 0, (1.20)
∂t
where
∗∇ ∗∇

J = Im f f −g g . (1.21)
m m
Note that the two terms appearing in the rhs of Eq. (1.21), referring respectively to
conduction and valence band electrons, have opposite sign since the curvature of the
two bands is opposite. One can verify that the semiclassical group velocity of the
quasiparticle, vg = ∇k ω, coincides with the velocity v given by the full quantum
mechanical expression (1.21), with J = ρ v.
1.3.2 The N-S junction

In this section we highlight the suggestive parallelism between the formalism intro-
duced in section (1.3.1) and the treatment of quasiparticle excitations in conventional
superconductors, as modeled by the BCS theory.
The Hamiltonian of the N-S junction is
HN-S = H0′ + H1′ + H2′ . (1.22)
Electrons in the metal experience the crystal lattice potential through H1′ ,
1 X
Z
H1′ = − d r ψσ† (r) ∇2 ψσ (r) . (1.23)
2m
σ=↑,↓
The space-dependent field operator ψσ (r) annihilates an electron with spin σ in the
two-fold degenerate conduction energy band, whose energy dispersion is taken to be
parabolic for simplicity,
ε(k) = k 2 /(2m), (1.24)
ε(k) being the eigenvalue of H1′ . With respect to the SM, the role of a and b bands
is replaced by the two spin flavors ↑ and ↓. The Fermi wave vector kF is fixed by the
condition that there are N electrons in the system, with µ = kF2 /2m.
e
e = he (a)
6 1 5 2 4 3
∆(z)
Energy (a.u.)
e e e
ω
e e e
SM EI
∆(z)
z
e e
h=e (b)
6 1 5 2 4 3
∆(z)
Energy (a.u.)
e e e
ω
h h h
N S
∆(z)
z
Fig. 1.2 Andreev reflection. (a) SM-EI junction. An incoming b-band electron is backscat-
tered into the a band. The thicker (thinner) curves are the renormalized (bare) bands at
different values of the z coordinate, with ∆(z) being the corresponding order parameter.
Only the relevant low-energy portion of the spectrum is shown here, with numbers from 1 to
6 pointing to the time sequence of the reflection process (the arrows represent group veloci-
ties). The inset illustrates that the reflection process may alternatively be seen as a coherent
flow of excitons from the SM into the EI. (b) N-S junction. In contrast to panel a, an electron
is backscattered as a hole, hence the whole process may be thought of as the dissipationless
flow of Cooper pairs through the interface.
The relevant two-body interaction H2′ is attractive and short-ranged,

Z
H2′ = −g d r ψ↑† (r) ψ↓† (r) ψ↓ (r) ψ↑ (r) , (1.25)
with g being a positive constant parametrizing the combined effect of Coulomb and
electron-phonon interaction in the vicinity of the Fermi surface [54]. The short-range
interaction (1.25) does not affect electrons with parallel spin as a consequence of Pauli
exclusion principle. The effective potential (1.25) results from the competition between
Coulomb repulsion and the screening effect of the positive ions in the lattice. Close to
the resonance frequency of the ion motion, the ions give a very large response to the
perturbation induced by an electron charge. The resulting cloud of the moving electron
plus the polarized ions has a net positive charge, then inducing a weak electron-electron
attraction, whose characteristic energy is a tiny fraction of the Fermi energy. As in
the case of the SM-EI junction, the boundary between N and S phases is determined
by the variation along z of the pair potential associated to H2′ , defined below [see
Fig. 1.2(b)]. A residual effect of Coulomb interaction is to shift the energy levels, that
are already renormalized in the dispersion relation (1.24). The one-body term
X Z
H0′ = d r ψσ† (r) V ′ (r) ψσ (r) , (1.26)
σ=↑,↓
arises from the possible impurities and defects at the interface, as well as the applied
bias voltage.
In order to find out the quasiparticles of the N-S junction, we follow the same
approach as for the SM-EI junction, with one important difference [12] that derives
from the following definition of the amplitudes:
f ′ (r, t) = hΨ′0 |ψ̃↑ (r, t)|Ψbk i, (1.27a)

′
g (r, t) = hΨ′0 |ψ̃↓† (r, t)|Ψbk i. (1.27b)
Here |Ψbk i is the state with one quasiparticle added to the many-electron ground state
|Ψ′0 i. According to eqns (1.27), the number of particles is not a constant of motion,
as we add both an electron [eqn (1.27a)] and a hole [eqn (1.27b)] to |Ψ′0 i. This is
allowed within the gran canonical framework, where the chemical potential µ is the
independent thermodynamic variable instead of N .
The resulting ‘Bogoliubov-de Gennes’ equations of motion are:
∂f ′ (r, t) ∇2 k2

i = − − F + V ′ (r) f ′ (r, t) + ∆′ (r) g ′ (r, t) , (1.28a)
∂t 2m 2m
∂g ′ (r, t)
2
∇ kF2

i = + − V (r) g ′ (r, t) + ∆′∗ (r) f ′ (r, t) ,
′
(1.28b)
∂t 2m 2m
with the local pair potential ∆′ (r) being defined as
∆′ (r) = −ghΨ′0 |ψ̃↓ (r) ψ̃↑ (r) |Ψ′0 i. (1.29)
The space-dependent parameter ∆′ (r) may be regarded as the center-of-mass wave

function of the condensate made of Cooper pairs. The latter are bound pairs of two
electrons with opposite spins, as it is evident from the definition (1.29). The product
of the two operators that destroys a Cooper pair, appearing in the so-called anomalous
average (1.29), does not commute with the number operator N as it breaks the U (1)
gauge symmetry of total Hamiltonian HN-S . Nevertheless,
√ since the number of electrons
is macroscopic, the number fluctuations, of order N , are small and may be neglected
with respect to the average
√ value of N . Moreover, since a value of one or two is still
small with respect to N , the ground states with either N or N − 2 electrons must be
regarded as identical, so the anomalous character (i.e., breaking the gauge symmetry
of HN-S ) of the definitions (1.29) and (1.27) is physically irrelevant.
It is remarkable to observe that the systems of equations (1.12) and (1.28), re-
spectively describing the SM-EI and the N-S junctions, are formally identical in the
homogeneous case. This corresponds to put respectively V (r) = 0 in eqns (1.12) and
V ′ (r) = 0 in eqns (1.28), as well as to take the built-in exciton coherence in (1.12) as a
local operator, ∆(r, r ′ ) = δ(r − r ′ ) ∆(r) (then ∆k does not depend on k). However,
the quasiparticle amplitudes for the two model junctions signify profoundly different
types of single-particle excitations.
In the EI, to obtain a free electron in the b-band—for k > kF —one has to break an
exciton among those forming the condensate, that is a bound pair of a b-band electron
and a-band hole. The way to do this is to either create an electron in the b band,
whose amplitude component is f , or destroy a hole in the a band, whose amplitude
component is g.
In the S, to obtain an unbounded single particle with spin ↑, one has to break
a Cooper pair in the condensate. This is accomplished by means of either creating
an electron with spin ↑ or destroying an electron with spin ↓. The latter option is
equivalent to creating a hole with spin ↑, as a consequence of time-reversal symmetry.
The components f ′ and g ′ are the amplitudes for the propagation of the electron and
the hole, respectively. Table 1.1 compares the key features of the EI with those of the
S, with regards to both the ground state and the quasiparticle excitations.
In the following we are interested in comparing the SM-EI and N-S systems. In order
to stress their formal analogy, hereafter we drop the prime symbol to label quantities
referring to the N-S junction (as f , g, ∆, etc) and use the same notation in both cases.
With this convention, formulae (1.14–1.21) obtained for the SM-EI junction hold for
the N-S junction, too.
1.4 Andreev reflection at the interface between excitonic insulator

and semimetal
In this section we introduce the phenomenon of Andreev reflection at the N-S bound-
ary as a paradigm to discuss the transport through the SM-EI junction. There are
three qualitatively important results that are common to both systems: (i) all three
cartesian components of the velocity change sign when the quasiparticle is reflected
(Sec. 1.4.1) (ii) the ratio of incident quasiparticles C(ω) which are transmitted through
the interface depends on the coherence factors of the condensate, being strongly sup-
pressed close to the gap (Sec. 1.4.2) (iii) the condensate induces pairing on the normal
side of the junction (proximity effect, Sec. 1.4.3).
1.4.1 Velocity inversion at the interface
The electrical transport across the N-S junction exhibits high conductance behavior
at vanishing applied voltage bias. This evidence seems to contradict the fact that
quasiparticle excitations are gapped in the S (see Fig. 1.3): quasiparticles in the bulk
N approaching the junction with energy smaller than the gap, 0 < ω < ∆, cannot
penetrate into the bulk S. This paradox is solved by Andreev reflection, which is
illustrated below.
Andreev reflection at the interface between excitonic insulator and semimetal 19
B A I D C
ω
Energy (a.u)
N S
∆
0 + - + - +
-q q q -k k
-kF 0 kF -kF 0 kF
Crystal wave vector kz (a.u.)
Fig. 1.3 Junction between normal metal (N, left) and superconductor (S, right). The plot
shows the quasiparticle energy ω vs wave vector kz in the two bulk phases. The labels mark
the allowed elastic scattering channels for an incoming particle (labeled I) with kz = q + . A
is the Andreev reflection, B the normal reflection, C the normal transmission, and D the
cross-branch transmission. Note that the energy of the particle is positive: on the N side the
hole branch for |kz | < kF is obtained by inverting the energy of the portion of band filled
with electrons (showed as a dashed curve) with respect to the Fermi surface.
Consider ∆(z) to be a smooth complex increasing function of z, tending respectively

to the asymptotic values zero when z → −∞, inside the bulk N, and ∆0 when z → +∞,
inside the bulk S [Fig. 1.2(b)]. Following Andreev [12], we note that the medium under
consideration is completely homogeneous with an accuracy 2m |∆0 | /kF2 —a very small
quantity in typical superconductors. Therefore, we seek a solution of eqns (1.28) in
the form
f (r) = eikF n·r η(r) , g(r) = eikF n·r χ(r) , (1.30)
where n is some unit vector and η(r) and χ(r) are functions that vary slowly compared
to eikF n·r . Substituting (1.30) in (1.28) and neglecting higher derivatives of η and χ,
we obtain
(ivF n · ∇ + ω) η(r) − ∆(z) χ(r) = 0, (1.31a)

(ivF n · ∇ − ω) χ(r) + ∆∗ (z) η(r) = 0, (1.31b)
where vF = kF /m. It is easy to find for z → ±∞ the asymptotic form of the solutions
of eqns (1.31) describing the reflection of the quasiparticle falling on the junction.
When z → −∞ we put ∆(z) = 0. Then

η 1 ik1 ·r 0 ik2 ·r
= C1 e + C2 e , (1.32)
χ 0 1
SM EI SM EI N S N S
e
000
111
e
e e
000
111
e
111
000 111
000
000
111
h
000
111
000
111 000
111
000
111 000
111
000
111
000
111
000
111
000
111
000
111
111
000 000
111
000
111
=
000
111
=
000
111 000
111
000
111
000
111
000
111 111
000
000
111
000
111 000
111
000
111 000
111
000
111
e
000
111
000
111
h
e e
00
11
11
00
00
11
00
11
00
11
h (a) 00
11
11
00
00
11
00
11
00
11
00 11
11 00
00
11
00
11
00
11
00
11 e (b)
Fig. 1.4 Sketch of Andreev reflection for quasiparticles approaching the junction from the
normal-phase side. (a) SM-EI junction. The reflected left-going electron is equivalent to a
right-going hole. (b) N-S junction. The reflected left-going hole is equivalent to a right-going
electron.
where n · k1 = ω/vF , n · k2 = −ω/vF ; C1 and C2 are arbitrary constants. The first

term on the rhs of eqn (1.32) corresponds to an electron whose velocity v (or J) lies
along n, and the second term to a hole whose velocity lies in the opposite direction to
n (in fact ω/vF ≪ kF since 2m |∆0 | /kF2 ≪ 1). If nz > 0, then the wave function (1.32)
describes an electron incident on the boundary and reflected as a hole on the N side;
if nz < 0, it describes an incident hole reflected as an electron. When z → +∞ we put
∆(z) = ∆0 in eqn (1.31). The solution describing the transmitted wave (Jz > 0) has
for ω > |∆0 | the form
1 + vF n · k3 /ω eiϕ/2
p
η C3
=√ p eik3 ·r , (1.33)
χ 2 1 − vF n · k3 /ω e−iϕ/2
where C3 is a constant, ϕ is the phase of ∆0 ,

q
n · k3 = vF−1 ω 2 − |∆0 |2 for nz > 0, (1.34a)
q
n · k3 = −vF−1 ω 2 − |∆0 |2 for nz < 0. (1.34b)
As expected, for ω < |∆0 | the functions η and χ decay exponentially as z → +∞,
hence the quasiparticle is prevented from entering the bulk S. However, all three carte-
sian components of the velocity of the reflected particle change sign [see Fig. 1.4(b)].
This remarkable behavior, which is not due to interface roughness since we take the
interface to be completely flat, is the key to explain the electric transport through
the N-S junction. An electron with velocity v is Andreev-reflected into a hole with
velocity −v which carries exactly the same current as the incident electron. In fact, in
virtue of time-reversal invariance, the hole moving with velocity −v may be regarded
as an electron moving with velocity v [Fig. 1.4(b)]. Therefore, we may understand the
process of Andreev reflection as an electron above the Fermi surface forming a Cooper
pair with another electron below the Fermi surface on the N side: such pair moves
to the S side merging into the condensate, whereas the second electron leaves a hole
behind in the N Fermi sea [287].
The results obtained in this section hold also for the SM-EI junction, provided one
links η and χ components to the probability amplitudes of an electron of being in
either b or a band, respectively, as illustrated in Fig. 1.4(a). Apparently, the reflection
process at the SM-EI junction seems the usual reflection of an electron from the gap
barrier. However, the complete reversal of the velocity vector suggests that the reflected
electron may be regarded as an incoming hole with the same velocity as the incoming
electron. The idea is that the overall reflection process may be thought of as the flow
of electron-hole pairs—excitons—from the SM to the EI side, where they merge into
the exciton condensate. Below we substantiate this alternate interpretation.
1.4.2 Coherence factors in the transmission coefficients

To proceed we specify the functional form of the interface potential, assuming that
the excitonic coherence is a step function at the SM-EI interface, ∆(r) = ∆ θ(z), with
∆ > 0. Moreover, we introduce a simple-minded model for the effect of disorder at the
origin (e.g. an insulating layer) through the δ-potential V (r) = H δ(z). In the following
we abandon the slowly varying amplitude approximation and look for solutions of the
full eqns (1.28), requiring a larger number of scattering channels than those used in
Sec. 1.4.1.
Carriers coming from the bulk SM with energies slightly outside the EI gap have,
say for the incident electron at I, two reflection channels, A and B, and two trans-
mission channels at C and D (see Fig. 1.1). If the energy lies within the gap, only
the two reflection channels are possible. Whereas the interface—by breaking the lat-
tice translational symmetry—can in principle connect different parts of the Brillouin
zone [229], here the relevant regions of the wave vector space are the two valleys near
the gaps in the bulk EI for those states with the same component of the wave vector
parallel to the interface.
We consider the elastic scattering at equilibrium, matching wave functions of the
incident (I), transmitted (C and D), and reflected (A and B) states at the boundary,
following the approach of Ref. [22]. In the bulk EI, there are a pair of magnitudes of
k associated with ω, namely
√ q
k = 2m kF2 /2m ± (ω 2 − ∆2 )1/2 .
±
(1.35)
The total degeneracy of relevant states for each ω is fourfold: ±k ± . The two states
±k + have a dominant conduction-band character, whereas the two states ±k − are
mainly valence-band states. Using the notation

f (z)
Ψ(z) = (1.36)
g(z)
the wave functions degenerate in ω are

u0 ±ik+ z v0 ±ik− z
Ψ±k+ = e , Ψ±k− = e , (1.37)
v0 u0
with the amplitudes u0 , v0 defined as

s s
(ω 2 − ∆2 )1/2 (ω 2 − ∆2 )1/2

1 1
u0 = 1+ , v0 = 1− , (1.38)
2 ω 2 ω
possibly extended in the complex manifold. With regards to the SM bulk, ∆ = 0 and
the two possible magnitudes of the momentum q reduce to q ± = [2m(kF2 /2m ± ω)]1/2 ,
with wave functions

1 ±iq+ z 0 ±iq− z
Ψ±q+ = e , Ψ±q− = e , (1.39)
0 1
for conduction and valence bands, respectively. The appropriate boundary conditions
′ ′
are: (i) Continuity of Ψ at z = 0, so ΨEI (0) = ΨSM (0) ≡ Ψ(0). (ii) [fEI (0) − fSM (0)] /(2m) =
′ ′
Hf (0) and [gEI (0) − gSM (0)] /(2m) = −Hg(0), the derivative boundary conditions
appropriate for δ-functions. (iii) Incoming (incident), reflected and transmitted wave
directions are defined by their group velocities, i.e., an electron incident from the left
is transmitted with v > 0 and reflected with v < 0.
If an electron incident on the interface from the SM with energy ω > ∆ has wave
vector q + , the four outgoing channels, with probabilities A, B, C, D, have respectively
wave vectors q − , −q + , k + , −k − , as shown in Fig. 1.1. C is the probability of trans-
mission through the interface with a wave vector on the same (i.e., forward) side of
the Fermi surface as q + (i.e., q + → k + , not −k − ), whereas D gives the probability of
transmission on the back side of the Fermi surface (i.e., q + → −k − ). B is the probabil-
ity of intra-branch reflection, whereas A is the probability of Andreev (cross-branch)
reflection. The steady state solution of system (1.12) is
ΨSM (z) = Ψinc (z) + Ψrefl (z), ΨEI (z) = Ψtrans (z),
where

1 iq+ z 0 iq− z 1 −iq+ z
Ψinc (z) = e , Ψrefl(z) = a e +b e ,
0 1 0

u0 ik+ z v0 −ik− z
Ψtrans (z) = c e +d e . (1.40)
v0 u0
Applying the boundary conditions, we obtain a system of four linear equations in the
four unknowns a, b, c, and d, which we solve at a fixed value for ω. We introduce the
dimensionless barrier strength Z = H/vF , and approximate k + = k − = q + = q − ≈ kF ,
on the basis that the ratio 2m∆/kF2 is small. The quantities A, B, C, D, are the ratios
of the probability current densities of the specific transmission or reflection channels
to the current of the incident particle, e.g. A = |JA /Jinc |, and so on. The conservation
of probability requires that
A + B + C + D = 1. (1.41)
This result is useful in simplifying expressions for energies below the gap, ω < ∆,
where there can be no transmitted electrons, so that C = D = 0. Then, Eq. (1.41)
reduces simply to A = 1 − B.
Table 1.2 Transmission and reflection coefficients for the SM-EI junction. A
gives the probability of Andreev reflection (cross-branch), B of ordinary reflec-
tion, C of transmission without branch crossing, and D of cross-branch trans-
1/2
mission. Here θ = ω 2 + 4Z 2 ω 2 + (1 + 4Z 4 )(∆2 − ω 2 ) − 8Z 3 ω ∆2 − ω 2 ,
2 2 2 2 2 2 2 2 1/2

γ = Z v0 − u0 + (iZ + 1/2) 2u0 , and u0 = 1 − v0 = 1/2 [1 + (ω − ∆ ) /ω].
A B C D
Z2 1
No condensate 0 1+Z 2 1+Z 2 0
General form
∆2
ω<∆ θ 1−A 0 0
u20 v02 (u20 −v02 )2 Z 4 +Z 2 u20 (u20 −v02 )(1+Z 2 ) v02 (u20 −v02 )Z 2
ω>∆ |γ|2 |γ|2 |γ|2 |γ|2
No barrier
ω<∆ 1 0 0 0
v02 u20 −v02
ω>∆ u20
0 u20
0
Strong barrier
∆2
ω<∆ 4Z 4 (∆2 −ω 2 ) 1−A 0 0
u20 v02 1 u20 v02
ω>∆ Z 4 (u20 −v02 )2
1− Z 2 (u20 −v02 ) Z 2 (u20 −v02 ) Z 2 (u20 −v02 )
We find
u0 v0
a= ,
γ
u20 − v02 Z 2 − iZ

b= ,
γ
u0 (1 + iZ)
c= ,
γ
iv0 Z
d=− , (1.42)
γ
γ = Z 2 v02 − u20 + (iZ + 1/2) 2u20 .

The probability coefficients are actually the currents, measured in units of vF . For
2 2
example, A = |JA | /vF = |a| , and D = |d| / v02 − u20 . The expression for the energy
dependences of A, B, C, and D can be conveniently written in terms of the so-called
coherence factors u0 and v0 . The results are given in Table 1.2. For convenience, in
addition to the general results we also list the limiting forms of the results for zero
barrier (Z = 0) and for a strong barrier [Z 2 (u20 − v02 ) ≫ 1], as well as for ∆ = 0 (the
semimetal case).
In the absence of disorder (Z = 0), the dependence of the transmission coefficient
C(ω) on energies close to the gap is
p
C(ω) = 2 2 (ω − ∆) /∆ ω ≈ ∆, (1.43)
whereas below the gap the electron is totally Andreev-reflected and the transmission is
zero. The ordinary reflection channel is completely suppressed (B = 0) as well as cross-
branch transmission (D = 0). The situation is depicted in Fig. 1.5(a). Even above the
Transmission / reflection coefficient

1
A
Z=0 B Z=-1
0.8
C
0.6
B
0.4 C
0.2 A A
(a) (b)
B D D A
0
0 ∆ 0 ∆
Energy ω
Fig. 1.5 Transmission and reflection coefficients at the SM-EI boundary vs quasiparticle
energy ω. (a) Z = 0. (b) Z = −1. A gives the probability of Andreev reflection, B gives
the probability of ordinary reflection, C gives the transmission probability without branch
crossing, and D gives the probability of transmission with branch crossing. The parameter Z
measures the interface transparency.
gap, ω > ∆, there is a high probability for Andreev reflection, which strongly depends
on ω. For energies close to the gap, ω ≈ ∆, Andreev reflection is almost certain, A ≈ 1.
This is the cause for the low value of interface conductance. The effect is washed out
by the opacity of the interface: as |Z| increases [Z = −1 in Fig. 1.5(b)], the total
reflection probability A + B loses its dependence on ω, and the dominant reflection
channel changes from the Andreev one (A) into the ordinary one (B).
The interface opacity Z is the handle to tune the effect of excitonic coherence
on transport, as discussed in the next sections. Remarkably, the overall set of results
of Table 1.2 is formally identical to the analogous quantities obtained for the N-S
junction (e.g. compare with Table II of Ref. [22]), the only slight difference being the
behavior for Z 6= 0. In fact, due to the different boundary conditions, the coefficients
of the N-S junction are even functions of Z, whereas those of Table 1.2 do not have a
definite parity with respect to the sign of Z for ω < ∆ (there is a mistake in the entry
corresponding to the sub-gap strong-barrier case appearing in Table II of Ref. [22]).
Nevertheless, the expressions for coefficients in the strong-barrier case are the same
for the SM-EI and the N-S junction.
The appearance of coherence factors u and v in the coefficients of the SM-EI
junction demonstrates that the electron-hole condensate strongly affects the transport
and in general the wave function of carriers, by means of both inducing coherence on
the SM side and altering transmission features.
1.4.3 Proximity effect

From the results of the previous two subsections it follows that the condensate on the
right hand side of the junction induces pairing correlations in the normal phase on the
left hand side, even if there the order parameter ∆ is zero as interactions are switched
A perfect insulator 25
off. In the N-S junction the pairing induced by the proximity effect correlates electrons
with opposite spins, whereas in the SM-EI junction it correlates electron-hole pairs.
To show that the exciton condensate induces exciton order on the SM side it is
sufficient to compute the built-in coherence hΨ0 |ψ̃b (r) ψ̃a† (r)|Ψ0 i. This would be zero
in an isolated SM but by the proximity with the condensate acquires the value
X
hΨ0 |ψ̃b (r) ψ̃a† (r)|Ψ0 i = fk (r) gk∗ (r)
k

∆ ω
Z
≈ 2 d ω N (ω) cos arctan +2 z , (1.44)
ω vF
with N (ω) being the density of states in the SM. Inside the gap (ω ≈ 0) each quasi-
particle contributes to the sum (1.44) with a term ∼ exp [i arctan (∆/ω) + 2iωz/vF].
The only coordinate dependence enters this expression via the phase factor, 2ωz/vF,
which represents the relative phase shift of conduction- and valence-band components
of the wave function. If ω = 0, then these components keep constant relative phase
arctan (∆/ω) all the way to z = −∞, where no pairing interactions exist. Therefore,
the reflected electron—equivalent to an incoming hole—has exactly the same veloc-
ity as the incident particle, and will thus retrace exactly the same path all the way to
z = −∞. At finite energy, the z dependent oscillations provide destructive interference
on the pair coherence. Hence, the paths of incident and reflected electrons part ways
away from the interface. Analogous considerations apply to the incident electron and
to the Andreev-reflected hole in a sub-gap scattering event at the N-S interface [287].
1.5 A perfect insulator

From the results for transmission and reflection probabilities obtained in Sec. 1.4.2, we
derive in the linear response regime the values of the electrical and thermal conduc-
tances of the SM-EI junction, G and GT , respectively. The derivation is standard and
it proceeds along the lines explained e.g. in Refs. [22,53,215]. The Seebeck coefficient is
zero due to the symmetry of the model [293,215]. Then, except for an additive phonon
contribution to the thermal conductance, the interface thermoelectric properties are
completely determined by G and GT .
Both G and GT have an activation threshold at low temperature, T , proportional
to the gap ∆, as shown in Fig. 1.6 (cf. the curves for the transparent barrier with
Z = 0). At first sight, the transport properties of the SM-EI junction seem dramat-
ically different from those of the N-S junction, as the latter may sustain an electric
supercurrent at vanishing bias voltage whereas the former exhibits insulating behavior.
However, a closer examination shows that the two junctions share essential features.
In fact, the functional dependence of GT on T and ∆ is the same for both the
SM-EI and N-S junctions. Remarkably, the time rate of entropy production, Ṡ, is
the same very low value in both cases, pointing to the disipationless character of the
flow of charge and heat. This is seen by the relation between Ṡ and the transport
coefficients [293],
Ṡ = G(δV )2 /T + AGT (δT )2 /T 2 , (1.45)
with δT and δV being respectively the temperature and voltage drops at the interface
and A being the cross-sectional area. In the N-S junction the superfluid component
[ ∆ N(εF) vF / 4 units]
[e A N(εF) vF / 4 units] 1
6
0.8 5
Z=9
4 Z=7
0.6
3 Z=0
2
0.4 Z=0
2
GT (Z +1)
G (Z +1)
0.2
1
2
2
(a) (b)
0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
T (∆ / KB units) T (∆ / KB units)
Fig. 1.6 (a) Electrical conductance G of the SM-EI junction vs temperature T . The curves
correspond to different values of the dimensionless barrier opacity, Z = 0, 1, 3, 5, 7, 9. For
each value of Z, G is divided by the corresponding transmission coefficient when ∆ = 0, i.e.,
C∆=0 = (Z 2 + 1)−1 . KB is the Boltzmann constant, A is the interface cross-sectional area,
N (εF ) is the density of states evaluated at the Fermi energy εF . (b) Thermal conductance
GT of the SM-EI junction vs T . The curves correspond to Z = 0, 1, 3, 5, 7.
does not carry any entropy. Therefore, the terms proportional to G and GT only
include the contribution of quasiparticles which, when they cross the N-S interface,
experience the same electric and thermal resistance as electrons do across the SM-EI
boundary.
To shed light on the dissipationless motion of electrons in the linear transport
regime, we vary the opacity of the SM-EI junction. The coherence between the two
sides of the interface is diminished as the dimensionless barrier strength Z increases
from zero (clean junction) to finite values (tunneling regime). Figure 1.6 displays G
and GT for increasing values of Z. Since the transmission coefficient C(ω) decreases
uniformly in the absence of any electron-hole pairing (cf. Table 1.2), C∆=0 = (Z 2 +
1)−1 , we rescale conductances dividing them by C∆=0 . Naively, we would expect that
the insertion of an insulating layer would reduce the conductances. On the contrary, the
effect is just the opposite: as Z increases, G/C∆=0 and GT /C∆=0 increase, eventually
reaching saturation in the tunneling regime. This shows that the exciton order induced
in the SM side by EI makes the junction less conductive for charge and heat transport.
The plot of the differential conductance (dI/dV )/C∆=0 vs the bias voltage V at low T
(Fig. 1.7) allows clear monitoring of the transition from the transparent to the opaque
limit, where transport is recovered. The effect is maximum for eV ≈ ∆ and as T → 0,
when the differential conductance becomes proportional to C(eV ) + D(eV ).
dI/dV (Z +1) [e A N(εF)vF / 4 units]

3
Z = 11
KBT / ∆ = 0.1
2
2
Z=0
1
2
0.5 1 1.5
Bias voltage V (∆ /e units)
Fig. 1.7 Differential conductance (dI/dV )/C∆=0 of the SM-EI junction vs bias voltage V ,
computed at KB T /∆ = 0.1. The curves correspond to different values of the dimensionless
barrier opacity, Z = 0, 1, 3, 5, 7, 9, 11. For each value of Z, dI/dV is divided by the corre-
sponding transmission coefficient when ∆ = 0, i.e., C∆=0 = (Z 2 + 1)−1 .
1.5.1 Charge versus exciton current
As anticipated at the end of Sec. 1.4.1, the transport features discussed above which
distinguish the excitonic insulator from the normal insulating state may be explained
by two alternate physical pictures. The conventional view is that electrons below the
energy gap cannot contribute to transport as they are back scattered by the gap barrier,
∆, formed by the proximity effect of the EI. The less conventional view is to make use
of the analogy with the N-S junction. Instead of counting the electrons in the valence
band as negatively charged carriers of the current, we may start with the state with
the valence band filled to the top as carrying zero current even under an electrical or
thermal current and regard each unoccupied state in the valence band as a positively
charged carrier — a hole — moving in the direction opposite to the electron. Then the
reflected electrons are replaced by incoming holes towards the barrier. Therefore, the
incident conduction electron and the valence hole may be viewed as a correlated pair
moving towards the interface [Fig. 1.4(a)]. The novelty is that a constant electron-
hole current moves from the SM to the EI below the gap, where electric transport
is blocked. As the electron-hole pair approaches the interface from the SM side, the
exciton current is converted into the condensate supercurrent: the global effect is that
in the steady state an exciton current exists flowing constantly and reversibly all the
way from the SM to the EI without any form of dissipation.
The above scenario follows from the continuity equation for the electron-hole cur-
rent. The probability density ρe-h (r, t) for finding either a conduction-band electron
2 2
or a valence-band hole at a particular time and place is ρe-h (r, t) = |f | + 1 − |g| .
Thus, the associated continuity equation is
∂ρe-h
+ ∇ · Je-h = 0, Je-h = Jpair + Jcond . (1.46)
∂t
One component of the electron-hole current,
1
Jpair = Im {f ∗ ∇f + g ∗ ∇g} , (1.47)
m
is the density current of the electron-hole pair, similar to the standard particle carrier
J = m−1 Im{f ∗ ∇f −g ∗∇g} with an important difference in sign. The other component,
∇ · Jcond = −4 Im {f ∗ g ∆} , (1.48)
depends explicitly on the built-in coherence of the electron-hole condensate ∆, and

may be described as the exciton supercurrent of the EI state.
Let us go back to our picture of ∆(z) smoothly varying in space (Fig. 1.2), with
the junction being divided into small neighborhoods at position r and each being a
homogeneous system. If ω < |∆0 |, each electron wave function, solution of eqn (1.28),
carries zero total electric current eJ, which is the sum of the equal and opposite
incident and reflected fluxes, and finite and constant electron-hole current Je-h = 2vF n,
with n a unit vector. When z → −∞, far from the interface on the SM side, the
supercurrent contribution Jcond is zero. As z increases and ∆(z) gradually rises, both
J and Je-h conserve their constant value, independent of z, since quasiparticle states
are stationary. However, their analysis in terms of incident and reflected quasi-particles
is qualitatively different.
From the electron point of view, we see in Fig. 1.2(a) that the incoming conduction-
band particle approaching the EI boundary sees its group velocity progressively re-
duced (from time step 1 to time step 3), up to the classical turning point (time step 4)
where it changes direction and branch of the spectrum: there is no net electric current.
From the exciton point of view, as the contribution to the electron-hole current
Jpair vanishes approaching the boundary, since the group velocity goes to zero at the
classical turning point where the wave function becomes evanescent, Jpair is converted
into the supercurrent Jcond . Excitons therefore can flow into the EI side without any
resistance, and the sum Je-h of the two contributions, Jpair and Jcond , is constant
through all the space [Fig. 1.4(a)].
As ω exceeds |∆0 |, J acquires a finite value and Je-h monotonously decreases.
However, close to the gap, electron transmission to the EI side is still inhibited [cf. eqn
(1.43)] by the pairing between electrons and holes of the condensate: an electron can
stand alone and carry current only after its parent exciton has been “ionized” by
injecting — say — a conduction-band electron or by filling a valence-band hole in the
EI. The ionization costs an amount of energy of the order of the binding energy of
the exciton, |∆0 |. Therefore, as long as ω ≈ |∆0 |, the competition between exciton
and electron flow favors Andreev reflection, which is the source of both electric and
thermal resistances.
In equilibrium, there is no net charge or heat flow, since quasi-particles with v and
−v compensate each other. However, if a heat current flows, the net drift velocity of
electrons and holes locally “drags” the exciton supercurrent, which otherwise would
be pinned by various scattering sources [93].
1.5.2 A concrete example

As a concrete example of the aforementioned conversion of free-exciton current into
condensate supercurrent, consider the quasiparticle steady state of Eq. (1.40). For
ω < ∆, k + and k − in the EI (z > 0) have small imaginary parts which lead to an
exponential decay on a length scale λ, where
−1/2
ω2

vF
λ= 1− 2 . (1.49)
2∆ ∆
The quasiparticles penetrate a depth λ before the electron-hole current Jpair is con-
verted to a supercurrent Jcond carried by the condensate; right at the gap edge the
length diverges. For clarity, we define C and D here as the transmission probabilities
at z ≫ λ, while for ω > ∆ plane-wave currents are spatially uniform and we need not
specify the position at which they are evaluated.
When the interface is transparent, Z = 0, the steady state (1.40) is specified by
b = d = 0, a = v0 /u0 , and c = 1/u0 . Below the gap coherence factors u0 and v0
2
are complex and equal in modulus. For ω < ∆, |a| = 1, which means the incident
conduction-band electron is totally reflected into the SM valence band. Thus, the
electron-hole current Jpair carried in the semimetal equals 2vF , but Jpair of the excitonic
insulator is exponentially small for z ≫ 0. Explicitly,
2
|c|

2 2 + ∂ +
Jpair = (|u0 | + |v0 | ) Im (eik z )∗ (eik z ) .
m ∂z
Letting k + ≈ kF + i/(2λ), we have
Jpair = 2vF e−z/λ . (1.50)
The “disappearing” electron-hole current reappears as exciton current carried by the

condensate. Recalling the definition of Jcond ,
∂Jcond /∂z = −4 Im{f ∗ g ∆} ,
by integration we obtain
Z z
2 ′
Jcond = −4∆ |c| d z ′ e−z /λ
Im[u∗0 v0 ] = 2vF 1 − e−z/λ . (1.51)
0
This is the desired result, explicitly showing the supercurrent Jcond increasing to an
asymptotic value as z → ∞, at the same rate as the free electron-hole current Jpair
dies away.
1.6 Josephson oscillations between exciton condensates in

electrostatic traps
The Josephson effect is a a macroscopic coherent phenomenon which has been ob-
served in systems as diverse as superconductors [17], superfluid Helium [14, 193, 249],
and Bose-Einstein condensates of trapped ultra-cold atomic gases [41,234,4,154]. Since
Josephson oscillations appear naturally when two spatially separated macroscopic wave
functions are weakly coupled, they have been predicted for bosonic excitations in solids
as well, like polaritons [280, 218] and excitons optically pumped in suitable semicon-
ductor heterostructures [231]. However, unlike the polaritons, which have a photonic
component allowing for easy detection, excitons stay dark unless they recombine ra-
diatively. So far, it is unclear how the exciton Josephson effect could be observed. One
proposal relied on the observation of quasiparticle excitations from the spectral prop-
erties of the emitted light [218]. A drawback of this idea is that spectral properties
are not unambiguously linked to the condensed phase. In this Section we propose the
interference of emitted photons as a direct probe of the exciton Josephson effect.
Condensed excitons are predicted to act as coherent light sources [190,68,189] (see
discussion below). If Josephson oscillations occur between two exciton traps, in princi-
ple they can be probed by measuring the interference of the beams separately emitted
from the traps. However, in the time interval before recombination and contrary to
the polariton case, there are too few photons emitted for an adequate signal to noise
ratio, so one has to average the signal over many replicas of the same experiment [103].
Here we show that such ensemble averaging blurs the signature of the Josephson effect
except in the relevant case of exciton “plasma” oscillations [198]. For the latter the
dipole energy difference between the traps modulates the visibility α of interference
fringes, providing a means for detection.
The subsections below are organized as follows: we first introduce the double quan-
tum well system illustrating a feasible scheme to manipulate electrically the exciton
phase (Sec. 1.6.1); we then set the theoretical framework (Sec. 1.6.2), and we eventually
propose a correlated photon counting experiment (Sec. 1.6.3).
1.6.1 Electrical control of the exciton phase
Consider a double quantum well where electrons and holes are separately confined
in the two layers. In experiments aiming at Bose-Einstein condensation of excitons,
electron-hole pairs are optically generated at energies higher than the band gap, left
to thermalize, form excitons, and, at sufficiently low temperature and high density,
condense before radiative decay [112, 103]. Let z be the growth axis of the two wells
separated by distance d. The electrons in the conduction band and holes in the valence
band move in the planes z = d, 0, respectively (Fig. 1.8). In the experiments [31,32], an
electric field Fz is applied along z to suppress inter-layer tunneling, thereby quenching
the exciton recombination.
Fabrications [97, 106, 44] of electrostatic traps with suitably located electrodes to
provide lateral confinement for the excitons have been implemented. The double quan-
tum well is sandwiched between two spacer layers, providing insulation from planar
electrodes lithographed on both sides of the coupled structure. Each electrode controls
a tunable gate voltage, Vg (x, y, z), which localizes in a region of the xy plane the field
Josephson oscillations between exciton condensates in electrostatic traps 31
Fz (a) (b)
Energy
UX(x)
V0
µ
d z ∆U
x
Fig. 1.8 (a) Double quantum well energy profile along the growth direction z. Electrons and
hole move freely in the xy planes at z = d and z = 0, respectively. Fz is the component of
the electric field parallel to the exciton electric dipole. (b) Effective exciton potential profile
vs x for the double-trap setup.
component along z, Fz (x, y, z) = −∂Vg (x, y, z)/∂z, while Fx and Fy are small and
can be neglected as well as the the dependence of Fz on z. The vertical field Fz (x, y)
makes the electrostatic potential energy of the exciton dipole depend on the lateral
position, UX (x, y) = −edFz (x, y) (e < 0) [cf. Fig. 1.8(b)]. In this way, potential traps
for excitons are designed with great flexibily, with in situ control of the height, width,
and shape of the potential barriers [97, 104].
First, we focus on the quasi-equilibrium situation before radiative recombination,
where excitons condense in two coupled electrostatic traps, both within the condensate
coherence length. Figure 1.8(b) depicts the exciton potential profile UX (x, y = 0)
along the x axis, with a link between two identical traps. The potential barrier allows
tunneling between the condensates in the two traps, whose macroscopic wave functions
are Ξ1 (x, y, t) and Ξ2 (x, y, t), respectively. The optical coherence in a single trap is of
the form
Ξ(x, y, t) = Ψ†a (x, y, 0, t)Ψb (x, y, d, t) ,

(1.52)
where h. . .i denotes quantum and thermal average in the grand canonical formalism.
In eqn (1.52) Ψ†a (x, y, 0, t) and Ψb (x, y, d, t) are respectively the hole and electron
destruction operators, with the vacuum being the semiconductor ground state with
no excitons. With respect to definition (1.13), here by putting (x, y) = (x′ , y ′ ) we
ignore the internal structure of the exciton relative-motion wave function. The reason
is that we focus on the ideal BEC limit of dilute weakly interacting excitons, n a2B ≪ 1,
with aB being the two-dimensional effective Bohr radius and n the exciton density.
Therefore, Ξ(x, y, t) is the macroscopic wave function for the center-of-mass motion of
excitons, which may be written in the form
√
Ξ(x, y, t) = ns eiϕ , (1.53)
with ns being the density of the exciton condensate and ϕ the phase. We recall that
only the relative phase between the two condensates has measurable effects [11, 150].
For a gauge transformation of the gate potential Vg → Vg − c−1 ∂χ(t)/∂t, which

leaves the field Fz unaltered, the field operators Ψ gain a phase,

ie
Ψa → Ψa exp χ(x, y, 0, t) ,
h̄c

ie
Ψb → Ψb exp χ(x, y, d, t) . (1.54)
h̄c
Throughout this section we indicate explicitly the reduced Planck’s constant h̄. The
macroscopic wave function, by eqn (1.52), also gains a phase,
e
ϕ→ϕ+ [χ(z = d, t) − χ(z = 0, t)] . (1.55)
h̄c
Hence, the frequency of time oscillation of the condensate is given by the electrostatic
energy of the exciton dipole in the external field [15], U = −edFz :
1
ϕ = ϕ(0) + edFz t , (1.56)
h̄
with ϕ(0) being the time-independent zero-field value. In the absence of the bilayer
separation of the electrons and the holes, their gauge phases gained in the electric
field would cancel each other resulting in no time dependence driven by U . Equation
(1.56) shows that the experimentally controllable dipole energy difference between the
two traps depicted in Fig. 1.8(b), ∆U = −ed(Fz1 − Fz2 ), drives the relative phase
between the two condensates, thereby creating Josephson oscillations as a means for
measuring the Josephson tunnel between the traps.
1.6.2 Exciton Josephson oscillations

We next introduce the usual two-mode description of inter-trap dynamics based on
the Gross-Pitaevskii (GP) equation [280, 218, 231, 236, 203, 288, 198]. Exciton-exciton
correlation [191] beyond the GP mean field may be neglected due to the repulsive
character of the dipolar interaction between excitons in coupled quantum wells. The
condensate total wave function solution is
Ξ(x, y, t) = Ξ1 (x, y, N1 ) eiϕ1 + Ξ2 (x, y, N2 ) eiϕ2 , (1.57)
where both the trap population Ni (t) and the condensate phase ϕi (t) possess the entire
time dependence for the ith trap (i = 1, 2), and Ξi (x, y, Ni ) is a real quantity, with
Z Z
dx dy Ξ2i (x, y, Ni ) = Ni (t). (1.58)
The dynamics of the GP macroscopic wave function Ξ(x, y, t) depends entirely on the
temporal evolution of two variables, the population imbalance k(t) = (N1 − N2 )/2 and
the relative phase φ(t) = ϕ1 − ϕ2 of the two condensates. Here we consider a time
interval much shorter than the exciton lifetime (10 — 100 ns) and ignore the spin
structure. Therefore, the total population is approximately constant, N1 (t) + N2 (t) =
N.
The equations of motion for the canonically coniugated variables h̄k and φ are
derived from the effective Hamiltonian
k2 δJ p 2
HJ = Ec + ∆U k − N − 4k 2 cos φ , (1.59)
2 2
under the condition k ≪ N (Ref. [198]). Ec = 2dµ1 /dN1 is the exciton “charging”
energy of one trap, where µ1 is the chemical potential of trap 1, whereas δJ is the
Bardeen single-particle tunnelling energy,
h̄2
Z
∂ξ2 ∂ξ1
δJ = dy ξ1 − ξ2 , (1.60)
m ∂x ∂x x=0
where m is√the exciton mass. The single-particle orbital ξi (x, y) is defined through
Ξi (x, y) = Ni ξi (x, y).
The various dynamical regimes associated to certain intitial conditions (k(0), φ(0)),
including π oscillations and macroscopic quantum self-trapping, are exhaustively dis-
cussed in Refs. [236, 203]. Two cases are specially relevant:
AC Josephson effect. Under the conditions ∆U ≫ N Ec /2, ∆U ≫ δJ , one easily

obtains
∆U δJ N
φ(t) = − t + φ(0) , k̇ = sin φ . (1.61)
h̄ 2h̄
Equation (1.61) shows that, analogous to the case of two superconductors separated
by a thin barrier, if the phase difference φ between the condensates is not a multiple of
π, an exciton supercurrent 2k̇ flows across the barrier. Remarkably, in the presence of
an electric field gradient along z, an exciton flux oscillates back and forth between the
two traps, with frequency ∆U/h̄. As an exciton goes through the barrier, it exchanges
with the field the dipole energy acquired or lost in the tunneling process. The analogy
with the AC Josephson effect for superconductors is clear: in that case a bias voltage
V is applied across the junction, and the energy 2eV is exchanged between field and
Cooper pairs, as the latter experience a potential difference of V when penetrating the
potential barrier.
Plasma oscillations. This case concerns small oscillations around the equilibrium po-
sition (k, φ)eq = (0, 0). The Hamiltonian (1.59) may then be linearized into the form
k2

δJ 1 δJ N
HJ = 2 + Ec + δJ N φ2 + ∆U k − . (1.62)
2 N 4 2
It follows that both k and φ oscillate in time with plasma frequency

1p
ωJ = δJ (N Ec /2 + δJ ). (1.63)
h̄
Note that ∆U displaces the equilibrium position from (k, φ)eq = (0, 0) to
(k, φ)eq = (−∆U N δJ /2(h̄ωJ )2 , 0). (1.64)

detector
optical delay
τ
trap 1 trap 2
Ξ1 , ϕ1 Ξ2 , ϕ2
Fig. 1.9 Proposed experimental setup to measure the time correlation of the photons emitted
from the two traps. The delay time τ of one optical path is externally controlled.
1.6.3 Correlated photon counting experiment

Figure 1.9 illustrates the correlated photon counting setup which we propose to probe
Josephson oscillations. The detector measures the intensity I(τ ) of the sum of the two
beams separately emitted from the traps. A delay time τ is induced in one of the two
beams, as in Ref. [282]. The fields are simply proportional to the order parameters Ξi
of the traps. In fact, Ξ(x, y, t) is associated with a macroscopic electric dipole moment,
P (t) = x̂Px (t) ± iŷPy (t), (1.65)
which couples to photons:

Z
Px (t) = dx dy x Ξ(x, y, t), (1.66)
and similarly for Py . The built-in dipole hP (t)i 6= 0 oscillates with frequency (µ +
EX )/h̄, where EX is the optical gap minus the exciton binding energy, and µ accounts
for exciton-exciton interaction [190, 68, 189]. This macroscopic oscillating dipole is
equivalent to a noiseless current, which radiates a coherent field [82].
Therefore, the measured intensity I(τ ) is
I(τ ) = 2I0 [1 + hcos φ(τ )i] , (1.67)
assuming that the fields emitted from the two traps have the same magnitude (and
intensity I0 ) but different relative phase φ, which is evaluated at the delayed time τ .
I(τ ) may be written as
I(τ ) = 2I0 [1 + α cos φ0 (τ )] , (1.68)
where φ0 (τ ) is the phase averaged over many measurements, defined by the condition
hsin [φ(τ ) − φ0 (τ )]i = 0, (1.69)
and
α = hcos [φ(τ ) − φ0 (τ )]i (1.70)

is the fringe visibility, i.e., the normalized peak-to-valley ratio of fringes,
Imax − Imin
α= , (1.71)
Imax + Imin
with Imax (Imin ) being the maximum (minimum) value of I(τ ), and 0 ≤ α ≤ 1.
Equation (1.68) has a few important caveats. Since I(τ ) is an average, the temporal
inhomogeneous effect will blur the interference fringes, i.e., α < 1. Other dephasing
mechanisms include exciton recombination and inelastic exciton-phonon scattering
[282], as well as inelastic [282] and elastic exciton-exciton scattering, which in first
instance may all be neglected for short τ , low T , and n a2B ≪ 1, respectively.
The most immediate caveat is that the exciton condensates in decoupled traps must
acquire a relative phase if initially they condense separately without a definite phase
relation. This scenario is analogous to the case of interference between independent
laser sources first discussed by Glauber [82] and later studied experimentally for mat-
ter waves [13]. Even though a one-shot measurement with sufficient resolution would
display fringes, the relative phase φ0 (τ ) is also subject to intrinsic dephasing effects
by quantum fluctuations [82]. The latter are significant noise sources which affect α,
when φ and k are quantized into canonically conjugated quantum variables whereas
in the GP theory used so far they were classical variables whose fluctuations where
neglected.
In the following, we quantize Hamiltonian (1.59) in order to properly evaluate
α = hcos [φ(τ ) − φ0 (τ )]i as a quantum statistical average in finite traps. Therefore, we
follow Ref. [197] and introduce the commutator
h i
φ̂, k̂ = i. (1.72)
The operator k̂ now appearing in the quantized version of Hamiltonian (1.59) takes
the form −i∂/∂φ, whereas the ground state wave function is defined in the space of
periodical functions of φ with period 2π. If condensate oscillations are mainly

coherent,
the variance of φ is small and the visibility is approximated by α = 1 − 21 (∆φ)2 .

The most interesting case concerns plasma oscillations. For ∆U = 0, the ground
state of the quantized version of the harmonic oscillator Hamiltonian (1.62) is a Gaus-
sian, with φ0 = 0, independent from τ , and minimal spreading ∆φ2 ≈ (Ec /2δJ N )1/2 .

Therefore, the interferometer output is time-independent, I = 2I0 (1+α), showing con-

structive interference, I > 2I0 , with
r
Ec
α=1− . (1.73)
8δJ N
Not surprisingly, the visibility is controlled by the ratio Ec /δJ N , reaching the maxi-
mum α = 1 as Ec /δJ N → 0. In fact, α is given by the balance between the competing
effects of tunnelling (∝ δJ N ), which enforces a well-defined inter-trap phase, and in-
verse compressibility (∝ Ec ), which favors the formation of separate number states in
the two traps, thus separating the two macroscopic wave functions.
Beam intensity I(τ) / I0 3.9
3.6
(a)
3.9
3.6
(b)
3.3
0 1 2 3 4 5
Delay time τ ( 1 / ωJ )
Fig. 1.10 (Color online). Beam intensity I(τ )/I0 vs delay time τ , for ∆U/h̄ωJ = 0.2, 0.5
(dashed and solid lines, respectively). (a) T = 0 and α = 1, 0.8 (black and red [light gray]
lines, respectively). (b) α(T = 0) = 0.94 and kB T /h̄ωJ = 0, 1, 2 (black, red [light gray], and
blue [dark gray] lines, respectively).
A small finite value of ∆U in eqn (1.62) displaces the equilibrium position of the
harmonic oscillator. Noticeably, the ground state is a coherent state with harmonic
evolution of the average phase in time,
∆U
φ0 (τ ) = − sin (ωJ τ ), (1.74)
h̄ωJ
whereas α is unchanged. The Gaussian probability density characteristic of the ground

state for ∆U = 0 now is simply carried back and forth in φ space in the same motion
as the expectation value φ0 (τ ).
This key feature allows for directly monitoring τ -dependent plasma oscillations of
frequency ωJ through the photon correlation measurement (cf. Fig. 1.10). We evaluate
the effect of thermal fluctuations on α via the formula
P
n αn exp [−βEn ]
α(T ) = P , (1.75)
n exp [−βEn ]
where β = 1/kB T and 2(αn − 1) = (∆φ)2 n is the variance of φ in the nth excited

state whose energy is En . At low T , the excited states may be approximated as those
of the harmonic oscillator, giving
r
Ec 1 1
α(T ) = 1 − + βh̄ωJ . (1.76)
2δJ N 2 e −1
The above results are summarized by the formula

Conclusions 37

∆U
I(τ ) = 2I0 1 + α(T ) cos sin ωJ τ , (1.77)
h̄ωJ
which is valid for Ec /δJ N ≪ 1.

For small dipole energy variations, ∆U/h̄ωJ ≪ 1, the oscillating part within the
square brackets of eqn (1.77) may be written as −α(T )/2(∆U/h̄ωJ )2 sin2 ωJ τ . This
shows that the visibility α(T ) of fringes, which oscillate like sin2 ωJ τ , is modulated
by the experimentally tunable factor (∆U/h̄ωJ )2 /2. The dependence of I(τ ) on ∆U
is illustrated in Fig. 1.10 for two values of ∆U/h̄ωJ . As ∆U/h̄ωJ is increased [from
0.2 (dashed lines) to 0.5 (solid lines)], the amplitude of oscillations of I(τ ) shows a
strong non-linear enhancement, providing a clear signature of Josephson oscillations.
The oscillation amplitudes are larger for higher values of α [cf. Fig. 1.10(a)], and fairly
robust against thermal smearing [cf. Fig. 1.10(b)]. In fact, Fig. 1.10(b) shows that the
oscillation of I(τ ) is still clearly resolved for temperatures as high as T ≈ h̄ωJ /kB . At
even higher temperatures α(T ) displays anharmonic effects [197], with α(T ) → 0 as
T → ∞.
The AC Josephson effect cannot be observed within our scheme. In fact, for large
values of ∆U , the term proportional to cos φ appearing in the Hamiltonian (1.59) may
be neglected in first approximation, and the ground state wave function is a plane wave,
(2π)−1/2 exp [in̄φ], where n̄ is the integer closest to −∆U/Ec . Since the probability
density, (2π)−1 , is constant, the phase is distributed randomly and the visibility is
zero. Therefore, the correction to α coming from the inclusion in the calculation of the
term neglected in (1.59) will be small and fragile against fluctuations.
1.7 Conclusions
The venerated topic of exciton Bose Einstein condensation is facing a rebirth as its
investigation is fueled by advances in novel materials and technologies, as for double-
layer semiconductors and graphene. Indeed, growing evidence shows that the concept
of exciton condensation is a paradigm of many-body behavior. The observation of the
coherence properties of the condensate, including superfluidity, is an important long-
term goal of this field. In this Chapter we have explained how the exciton analogues of
Andreev reflection and Josephson oscillations may be linked to measurable quantities.
We hope that these ideas may stimulate further experiments along this path.
References
[1] Abrikosov, A. A., Gorkov, L. P., and Dzyaloshinski, I. E. (1975). Methods of

quantum field theory in statistical physics. Dover, New York.
[2] Aeppli, G. and Fisk, Z. (1992). Comments Condens. Matter Phys., 16, 155.
[3] Agranovich, V. M. and Toshich, B. S. (1967). Zh. Eksp. i Teor. Fiz., 53, 149.
[Sov. Phys.–JETP 26, 104 (1968)].
[4] Albiez, M., Gati, R., Fölling, J., Hunsmann, S., Cristiani, M., and Oberthaler,
M. K. (2005). Phys. Rev. Lett., 95, 010402.
[5] Alexandrou, A., Kash, J. A., Mendez, E. E., Zachau, M., Hong, J. M., Fukuzawa,
T., and Hase, Y. (1990). Phys. Rev. B , 42, R9225.
[6] Alloing, M., Fuster, D., Gonzalez, Y., Gonzalez, L., and Dubin, F. (2012).
arXiv:1210.3176.
[7] Alloing, M., Lemaı̂tre, A., and Dubin, F. (2011). Europhys. Lett., 93, 17007.
[8] Alloing, M., Lemaı̂tre, A., Galopin, E., and Dubin, F. (2012). Phys. Rev. B , 85,
245106.
[9] Alloing, M., Lemaı̂tre, A., Galopin, E., and Dubin, F. (2012). arXiv:1202.3301.
[10] Anderson, O., Manzke, R., and Skibowski, M. (1985). Phys. Rev. Lett., 55, 2188.
[11] Anderson, P. W. (1996). Rev. Mod. Phys., 38, 298.
[12] Andreev, A. F. (1964). Zh. Eksp. i Teor. Fiz., 46, 1823. [Sov. Phys.–JETP 19,
1228 (1964)].
[13] Andrews, M. R., Townsend, C. G., Miesner, H.-J., Durfee, D. S., Kurn, D. M.,
and Ketterle, W. (1997). Science, 275, 637.
[14] Avenel, O. and Varoquaux, E. (1985). Phys. Rev. Lett., 55, 2704.
[15] Balatsky, A. V., Joglekar, Y. N., and Littlewood, P. B. (2004). Phys. Rev.
Lett., 93, 266801.
[16] Bardeen, J., Cooper, L. N., and Schrieffer, J. R. (1957). Phys. Rev., 108, 1175.
[17] Barone, A. and Paterno, G. (1982). Physics and applications of the Josephson
effect. Wiley, New York.
[18] Bascones, E., Burkov, A. A., and MacDonald, A. H. (2002). Phys. Rev. Lett., 89,
086401.
[19] Bistritzer, R., Min, H., Su, J.-J., and MacDonald, A. H. (2008). arXiv:0810.0331.
[20] Blatt, J. M., Böer, K. W., and Brandt, W. (1962). Phys. Rev., 126, 1691.
[21] Bloch, I., Dalibard, J., and Zwerger, W. (2008). Rev. Mod. Phys., 80, 885.
[22] Blonder, G. E., Tinkham, M., and Klapwijk, T. M. (1982). Phys. Rev. B , 25,
4515.
[23] Bohr, A. and Mottelson, B. R. (1998). Nuclear structure — Vol. I and II. World
Scientific, Singapore.
[24] Bolotin, K. I., Ghahari, F., Shulman, M. D., Stormer, H. L., and Kim, P. (2009).
Nature, 462, 196.
References 39
[25] Bostwick, A., Speck, F., Seyller, T., Horn, K., Polini, M., Asgari, R., MacDonald,
A. H., and Rotenberg, E. (2010). Science, 328, 999.
[26] Bronold, F. X. and Fehske, H. (2006). Phys. Rev. B , 74, 165107.
[27] Bronold, F. X., Fehske, H., and Röpke, G. (2007). J. Phys. Soc. Jpn., 76, 27.
Supplement A.
[28] Bucher, B., Steiner, P., and Wachter, P. (1991). Phys. Rev. Lett., 67, 2717.
[29] Bulou, A., Rousseau, M., and Nouet, J. (1992). Key Eng. Mater., 68, 133.
[30] Butov, L. V. (2003). Solid State Commun., 127, 89.
[31] Butov, L. V. (2004). J. Phys.: Condens. Matter , 16, R1577.
[32] Butov, L. V. (2007). J. Phys.: Condens. Matter , 19, 295202.
[33] Butov, L. V. and Filin, A. I. (1998). Phys. Rev. B , 58, 1980.
[34] Butov, L. V., Gossard, A. C., and Chemla, D. S. (2002). Nature, 418, 751.
[35] Butov, L. V., Ivanov, A. L., İmamoğlu, A., Littlewood, P. B., Shashkin, A. A.,
Dolgopolov, V. T., Campman, K. L., and Gossard, A. C. (2001). Nature, 86,
5608.
[36] Butov, L. V., Lai, C. W., Ivanov, A. L., Gossard, A. C., and Chemla, D. S. (2002).
Nature, 417, 47.
[37] Butov, L. V., Levitov, L. S., Mintsev, A. V., Simons, B. D., Gossard, A. C., and
Chemla, D. S. (2004). Phys. Rev. Lett., 92, 117404.
[38] Butov, L. V., Zrenner, A., Abstreiter, G., Böhm, G., and Weimann, G. (1994).
Phys. Rev. Lett., 73, 304.
[39] Can, M. Ali and Hakioğlu, T. (2009). Phys. Rev. Lett., 103, 086404.
[40] Casella, R. C. (1963). J. Phys. Chem. Solids, 24, 19.
[41] Cataliotti, F. S., Burger, S., Fort, C., Maddaloni, P., Minardi, F., Trombettoni,
A., Smerzi, A., and Inguscio, M. (2001). Science, 293, 843.
[42] Cazzaniga, M., Cercellier, H., Holzmann, M., Monney, C., Aebi, P., Onida, G.,
and Olevano, V. (2012). Phys. Rev. B , 85, 195111.
[43] Cercellier, H., Monney, C., Clerc, F., Battaglia, C., Despont, L., Garnier, M. G.,
Beck, H., Aebi, P., Patthey, L., Berger, H., and Forró, L. (2007). Phys. Rev.
Lett., 99, 146403.
[44] Chen, G., Rapaport, R., Pfeiffer, L. N., West, K., Platzman, P. M., Simon, S.,
Vörös, Z., and Snoke, D. (2006). Phys. Rev. B , 74, 045309.
[45] Chen, X. M. and Quinn, J. J. (1991). Phys. Rev. Lett., 67, 895.
[46] Christianen, P. C. M., Piazza, F., Lok, J. G. S., Maan, J. C., and van der Vleuten,
W. (1998). Physica B , 251, 624.
[47] Cohen, K., Rapaport, R., and Santos, P. V. (2011). Phys. Rev. Lett., 106, 126402.
[48] Combescot, M., Betbeder-Matibet, O., and Combescot, R. (2007). Phys. Rev.
Lett., 99, 176403.
[49] Combescot, M., Moore, M. G., and Piermarocchi, C. (2011). Phys. Rev. Lett., 106,
206404.
[50] Comte, C. and Nozières, P. (1982). J. Physique, 43, 1069.
[51] Croxall, A. F., Das Gupta, K., Nicoll, C. A., Thangaraj, M., Beere, H. E., Farrer,
I., Ritchie, D. A., and Pepper, M. (2009). Phys. Rev. B , 80, 125323.
[52] Croxall, A. F., Gupta, K. Das, Nicoll, C. A., Thangaraj, M., Beere, H. E., Farrer,
I., Richtie, D. A., and Pepper, M. (2008). Phys. Rev. Lett., 101, 246801.
40 References
[53] Datta, S. (1995). Electronic transport in mesoscopic systems. Cambridge Univer-

sity Press, Cambridge (UK).
[54] de Gennes, P. G. (1999). Superconductivity of metals and alloys. Westview Press,
Boulder (Colorado).
[55] des Cloizeaux, J. (1965). J. Phys. Chem. Solids, 26, 259.
[56] Dillenschneider, R. and Han, J. H. (2008). Phys. Rev. B , 78, 045401.
[57] Dolcini, F., Rainis, D., Taddei, F., Polini, M., Fazio, R., and MacDonald, A. H.
(2010). Phys. Rev. Lett., 104, 027004.
[58] Drut, J. E. and Lände, T. A. (2009). Phys. Rev. Lett., 102, 026802.
[59] Du, X., Skachko, I., Duerr, F., Luican, A., and Andrei, E. Y. (2009). Nature, 462,
192.
[60] Duan, J.-M., Arovas, D. P., and Sham, L. J. (1997). Phys. Rev. Lett., 79, 2097.
[61] Dubi, Y. and Balatsky, A. V. (2010). Phys. Rev. Lett., 104, 166802.
[62] Eastham, P. R., Cooper, N. R., and Lee, D. K. K. (2012). Phys. Rev. B.
[63] Eisenstein, J. P. (2003). Solid State Commun., 127, 123.
[64] Eisenstein, J. P. and MacDonald, A. H. (2004). Nature, 432, 691.
[65] Elias, D. C., Gorbachev, R. V., Mayorov, A. S., Morozov, S. V., Zhukov, A. A.,
Blake, P., Ponomarenko, L. A., Grigorieva, I. V., Novoselov, K. S., Guinea, F.,
and Geim, A. K. (2011). Nature Phys., 7, 701.
[66] Falicov, L. M. and Kimball, J. C. (1969). Phys. Rev. Lett., 22, 997.
[67] Fernández-Rossier, J. and Tejedor, C. (1997). Phys. Rev. Lett., 78, 4809.
[68] Fernández-Rossier, J., Tejedor, C., and Merlin, R. (1998). Solid State Com-
mun., 108, 473.
[69] Finck, A. D. K., Eisenstein, J. P., Pfeiffer, L. N., and West, K. W. (2011). Phys.
Rev. Lett., 106, 236807.
[70] Freitag, F., Trbovic, J., Weiss, M., and Schönenberger, C. (2012). Phys. Rev.
Lett., 108, 076602.
[71] Fukuzawa, T., Kano, S. S., Gustafson, T. K., and Ogawa, T. (1990). Surf.
Sci., 228, 482.
[72] Fukuzawa, T., Mendez, E. E., and Hong, J. M. (1990). Phys. Rev. Lett., 64, 3066.
[73] Gamayun, O. V., Gorbar, E. V., and Gusynin, V. P. (2009). Phys. Rev. B , 80,
165429.
[74] Gärtner, A., Holleitner, A. W., Kotthaus, J. P., and Schuh, D. (2006). Appl.
Phys. Lett., 89, 052108.
[75] Gärtner, A., Prechtel, L., Schuh, D., Holleitner, A. W., and Kotthaus, J. P. (2007).
Phys. Rev. B , 76, 085304.
[76] Geim, A. K. and Novoselov, K. S. (2007). Nature Mat., 6, 183.
[77] Gergel’, V. A., Kazarinov, R. F., and Suris, R. A. (1968). Zh. Eksp. i Teor.
Fiz., 54, 298. [Sov. Phys.–JETP 27, 159 (1968)].
[78] Gilbert, M. J. and Shumway, J. (2009). J. Comput. Electron., 8, 51.
[79] Ginzburg, V. L. and Kirzhnits, D. A. (1964). Zh. Eksp. i Teor. Fiz. Pisma, 47,
2006. [Sov. Phys.–JETP 20, 1346 (1965)].
[80] Giorgini, S., Pitaevskii, L. P., and Stringari, S. (2008). Rev. Mod. Phys., 80, 1215.
[81] Giudici, P., Muraki, K., Kumada, N., and Fujisawa, T. (2010). Phys. Rev.
Lett., 104, 056802.
References 41
[82] Glauber, R. J. (1965). Quantum optics and electronics, p. 63. Gordon and Breach,
New York.
[83] Glushkov, V. V., Demishev, S. V., Ignatov, M. I., Paderno, Y. B., Shitsevalova,
N. Y., Kuznetsov, A. V., Churkin, O. A., Sluchanko, D. N., and Sluchanko, N. E.
(2006). J. Solid State Chem., 179, 2871.
[84] González, J. (2012). Phys. Rev. B , 85, 085420.
[85] Gorbachev, R. V., Geim, A. K., Katsnelson, M. I., Novoselov, K. S., Tudorovskiy,
T., Grigorieva, I. V., MacDonald, A. H., Morozov, S. V., Watanabe, K., Taniguchi,
T., and Ponomarenko, L. A. (2012). Nature Phys., 8, 896.
[86] Gorbunov, A. V. and Timofeev, V. B. (2006). JETP Lett., 84, 329.
[87] Griffin, A., Snoke, D. W., and Stringari, S. (ed.) (1995). Bose-Einstein conden-
sation. Cambridge University Press, Cambridge (UK).
[88] Grosso, G., Graves, J., Hammack, A. T., High, A. A., Butov, L. V., Hanson, M.,
and Gossard, A. C. (2009). Nature Photonics, 3, 577.
[89] Grosso, G. and Pastori Parravicini, G. (2000). Solid State Physics (1st edn).
Academic Press, San Diego.
[90] Grüner, G. (2000). Density waves in solids. Westview Press, Boulder, CO.
[91] Guinea, F. (2010). Physics, 3, 1.
[92] Gupta, K. Das, Croxall, A. F., Waldie, J., Nicoll, C. A., Beere, H. E., Farrer, I.,
Ritchie, D. A., and Pepper, M. (2011). Advances in Condens. Matter Phys., 2011,
727958.
[93] Guseı̆nov, R. R. and Keldysh, L. V. (1972). Zh. Eksp. i Teor. Fiz., 63, 2255. [Sov.
Phys.–JETP 36, 1193 (1973)].
[94] Hagn, M., Zrenner, A., Böhm, G., and Weimann, G. (1995). Appl. Phys. Lett., 67,
232.
[95] Hakioğlu, T. and Şahin, M. (2007). Phys. Rev. Lett., 98, 166405.
[96] Halperin, B. I. and Rice, T. M. (1968). Solid State Phys., 21, 115.
[97] Hammack, A. T., Gippius, N. A., Yang, S., Andreev, G. O., Butov, L. V., Hanson,
M., and Gossard, A. C. (2006). J. Appl. Phys., 99, 066104.
[98] Hammack, A. T., Griswold, M., Butov, L. V., Smallwood, L. E., Ivanov, A. L.,
and Gossard, A. C. (2006). Phys. Rev. Lett., 96, 227402.
[99] Hanamura, E. and Haug, H. (1977). Phys. Rep., 33, 209.
[100] Hellmann, S., Rohwer, T., Kalläne, M., Hanff, K., Sohrt, C., Stange, A., Carr,
A., Murname, M. M., Kapteyn, H. C., Kipp, L., Bauer, M., and Rossnagel, K.
(2012). Nature Commun., 3, 1069.
[101] High, A. A., Hammack, A. T., Butov, L. V., Hanson, M., and Gossard, A. C.
(2007). Optics Lett., 32, 2466.
[102] High, A. A., Hammack, A. T., Butov, L. V., Mouchliadis, L., Ivanov, A. L.,
Hanson, M., and Gossard, A. C. (2009). Nano Lett., 9, 2094.
[103] High, A. A., Leonard, J. R., Hammack, A. T., Fogler, M. M., Butov, L. V.,
Kavokin, A. V., Campman, K. L., and Gossard, A. C. (2012). Nature, 483, 584.
[104] High, A. A., Leonard, J. R., Remeika, M., Butov, L. V., Hanson, M., and Gos-
sard, A. C. (2012). Nano Lett., 12, 2605.
[105] High, A. A., Leonard, J. R., Remeika, M., Butov, L. V., Hanson, M., and Gos-
sard, A. C. (2012). Nano Lett., 12, 5422.
42 References
[106] High, A. A., Novitskaya, E. E., Butov, L. V., Hanson, M., and Gossard, A. C.
(2008). Science, 321, 229.
[107] High, A. A., Thomas, A. K., Grosso, G., Remeika, M., Hammack, A. T., Mey-
ertholen, A. D., Fogler, M. M., Butov, L. V., Hanson, M., and Gossard, A. C.
(2009). Phys. Rev. Lett., 103, 087403.
[108] Hu, B. Y.-K. (2000). Phys. Rev. Lett., 85, 820.
[109] Huber, T., Zrenner, A., Wegscheider, W., and Bichler, M. (1998). Phys. Status
Solidi A, 166, R5.
[110] Hulin, D., Mysyrowicz, A., and Benoı̂t à la Guillame, C. (1980). Phys. Rev.
Lett., 45, 1970.
[111] Ivanov, A. L. (2002). Europhys. Lett., 59, 586.
[112] Ivanov, A. L. and Tikhodeev, S. G. (ed.) (2008). Problems of condensed matter
physics: Quantum coherence phenomena in electron-hole and coupled matter-light
systems. Oxford University Press, Oxford.
[113] Jansen, H. J. F., Freeman, A. J., and Monnier, R. (1985). Phys. Rev. B , 31,
4092.
[114] Jèrome, D., Rice, T. M., and Kohn, W. (1967). Phys. Rev., 158, 462.
[115] Joglekar, Y. N., Balatsky, A. V., and Lilly, M. P. (2005). Phys. Rev. B , 72,
205313.
[116] Kane, B. E., Eisenstein, J. P., Wegscheider, W., Pfeiffer, L. N., and West, K. W.
(1994). Appl. Phys. Lett., 65, 3266.
[117] Karmakar, B., Pellegrini, V., Pinczuk, A., Pfeiffer, L. N., and West, K. W. (2009).
Phys. Rev. Lett., 102, 036802.
[118] Kash, J. A., Zachau, M., Mendez, E. E., Hong, J. M., and Fukuzawa, T. (1991).
Phys. Rev. Lett., 66, 2247.
[119] Kash, K., Worlock, J. M., Sturge, M. D., Grabbe, P., Harbison, J. P., Scherer,
A., and Lin, P. S. D. (1988). Appl. Phys. Lett., 53, 782.
[120] Kavoulakis, G. M. (2003). J. Low Temp. Phys., 132, 297.
[121] Keeling, J., Levitov, L. S., and Littlewood, P. B. (2004). Phys. Rev. Lett., 92,
176402.
[122] Keldysh, L. V. (1986). Contemp. Phys., 27, 395.
[123] Keldysh, L. V. and Kopaev, Yu. V. (1964). Fiz. Tverd. Tela., 6, 2791. [Sov.
Phys. Solid State 6, 2219 (1965)].
[124] Keldysh, L. V. and Kozlov, A. N. (1968). Zh. Eksp. i Teor. Fiz., 54, 978. [Sov.
Phys.–JETP 27, 521 (1968)].
[125] Keldysh, L. V. and Kozlov, Z. N. (1967). Zh. Eksp. i Teor. Fiz. Pisma, 5, 238.
[Sov. Phys.–JETP Lett. 5, 190 (1968)].
[126] Kellog, M., Eisenstein, J. P., Pfeiffer, L. N., and West, K. W. (2004). Phys. Rev.
Lett., 93, 036801.
[127] Keogh, J. A., Das Gupta, K., Beere, H. E., Ritchie, D. A., and Pepper, M. (2005).
Appl. Phys. Lett., 87, 202104.
[128] Kharitonov, M. Y. and Efetov, K. B. (2008). Phys. Rev. B , 78, 241401(R).
[129] Kharitonov, M. Y. and Efetov, K. B. (2010). Semicond. Sci. Tech., 25, 034004.
[130] Khveshchenko, D. V. (2001). Phys. Rev. Lett., 87, 246802.
[131] Kidd, T. E., Miller, T., Chou, M. Y., and Chiang, T.-C. (2002). Phys. Rev.
References 43
Lett., 88, 226402.

[132] Kim, S., Jo, I., Nah, J., Yao, Z., Banerjee, S. K., and Tutuc, E. (2011). Phys.
Rev. B , 83, 161401(R).
[133] Kittel, C. (1986). Introduction to solid state physics. John Wiley, New York.
[134] Knox, R. S. (1963). Theory of excitons. Volume Supplement 5, Solid State
Physics. Academic Press, New York.
[135] Kohn, W. (1968). Many-body physics, p. 351. Gordon and Breach, New York.
[136] Kohn, W. and Sherrington, D. (1970). Rev. Mod. Phys., 42, 1.
[137] Kotov, V. N., Uchoa, B., Pereira, V. M., Guinea, F., and Neto, A. H. Castro
(2012). Rev. Mod. Phys., 84, 1067.
[138] Kowalik-Seidl, K., Vögele, X. P., Rimpfl, B. N., Manus, S., Kotthaus, J. P.,
Schuh, D., Wegscheider, W., and Holleitner, A. W. (2010). Appl. Phys. Lett., 97,
011104.
[139] Krivolapchuk, V. V., Moskalenko, E. S., Zhmodikov, A. L., Cheng, T. S., and
Foxon, C. T. (1999). Solid State Commun., 111, 49.
[140] Kuramoto, Y. and Horie, C. (1978). Solid State Commun., 25, 713.
[141] Kuwata-Gonokami, M. (2011). Comprehensive semiconductor science and tech-
nology, Volume 2, p. 213. Elsevier, Amsterdam, The Netherlands.
[142] Kuz’min, R. V., Krivolapchuk, V. V., Moskalenko, E. S., and Mezdrogina, M. M.
(2010). Fiz. Tverd. Tela., 52, 1184. [Sov. Phys. Solid State 52, 1260 (2010)].
[143] Larionov, A. V. and Timofeev, V. B. (2001). JETP Lett., 73, 301.
[144] Larionov, A. V., Timofeev, V. B., Hvam, J., and Soerensen, C. (2000). JETP
Lett., 71, 117.
[145] Larionov, A. V., Timofeev, V. B., Hvam, J., and Soerensen, K. (2002). JETP
Lett., 75, 200.
[146] Larionov, A. V., Timofeev, V. B., Ni, P. A., Dubonos, S. V., Hvam, I., and
Soerensen, K. (2002). JETP Lett., 75, 570.
[147] Lee, P. A., Rice, T. M., Serene, J. W., Sham, L. J., and Wilkins, J. W. (1986).
Comments Condens. Matter Phys., 12, 99.
[148] Lee, R. M., Drummond, N. D., and Needs, R. J. (2009). Phys. Rev. B , 79,
125308.
[149] Leggett, A. J. (2006). Quantum liquids (1st edn). Oxford University Press,
Oxford.
[150] Leggett, A. J. and Sols, F. (1991). Found. Phys., 21, 353.
[151] Lemonik, Y., Aleiner, I. L., Toke, C., and Fal’ko, V. I. (2010). Phys. Rev. B , 82,
201408(R).
[152] Leonard, J. R., Kuznetsova, Y. Y., Yang, S., Butov, L. V., Ostatnický, T., Ka-
vokin, A., and Gossard, A. C. (2009). Nano Lett., 9, 4204.
[153] Levitov, L. S., Simons, B. D., and Butov, L. V. (2005). Phys. Rev. Lett., 94,
176404.
[154] Levy, S., Lahoud, E., Shomroni, I., and Steinhauer, J. (2007). Nature, 449, 579.
[155] Lin, J. L. and Wolfe, J. P. (1993). Phys. Rev. Lett., 71, 1222.
[156] Littlewood, P. B., Eastham, P. R., Keeling, J. M. J., Marchetti, F. M., Simons,
B. D., and Szymanksa, M. H. (2004). J. Phys.: Condens. Matter , 16, S3597.
[157] Littlewood, P. B. and Zhu, X. (1996). Physica Scripta, T68, 56.
44 References
[158] Lozovik, Y. E., Ogarkov, S. L., and Sokolik, A. A. (2012). Phys. Rev. B , 86,
045429.
[159] Lozovik, Y. E. and Sokolik, A. A. (2008). JETP Lett., 87, 55.
[160] Lozovik, Yu. E. and Yudson, V. I. (1976). Zh. Eksp. i Teor. Fiz., 71, 738. [Sov.
Phys.–JETP 44, 389 (1976)].
[161] Luin, S., Pellegrini, V., Pinczuk, A., Dennis, B. S., Pfeiffer, L. N., and West,
K. W. (2003). Phys. Rev. Lett., 90, 236802.
[162] Luttinger, J. M. and Kohn, W. (1955). Phys. Rev., 97, 869.
[163] MacDonald, A. H., Jung, J., and Zhang, F. (2012). Physica Scripta, T146,
014012.
[164] MacDonald, A. H. and Rezayi, E. H. (1990). Phys. Rev. B , 42, 3224.
[165] Maialle, M. Z., de Andrada e Silva, E. A., and Sham, L. J. (1993). Phys. Rev.
B , 47, 15776.
[166] Margaritondo, G., Bertoni, C. M., Weaver, J. H., Lévy, F., Stoffel, N. G., and
Katnani, A. D. (1981). Phys. Rev. B , 23, 3765.
[167] May, M. M., Brabetz, C., Janowitz, C., and Manzke, R. (2011). Phys. Rev.
Lett., 107, 176405.
[168] Mayorov, A. S., Elias, D. C., Mucha-Kruczynski, M., Gorbachev, R. V., Tu-
dorovskiy, T., Zhukov, A., Morozov, S. V., Katsnelson, M. I., Fal’ko, V. I., Geim,
A. K., and Novoselov, K. S. (2011). Science, 333, 800.
[169] Migdal, A. B. (1959). Zh. Eksp. i Teor. Fiz. Pisma, 37, 249. [Sov. Phys.–JETP
10, 176 (1960)].
[170] Min, H., Bistritzer, R., Su, J.-J., and MacDonald, A. H. (2008). Phys. Rev.
B , 78, 121401(R).
[171] Min, H., Borghi, G., Polini, M., and MacDonald, A. H. (2008). Phys. Rev. B , 77,
041407(R).
[172] Mink, M. P., Stoof, H. T. C., Duine, R. A., and MacDonald, A. H. (2011). Phys.
Rev. B , 84, 155409.
[173] Monney, C., Battaglia, C., Cercellier, H., Aebi, P., and Beck, H. (2011). Phys.
Rev. Lett., 106, 106404.
[174] Monney, C., Schwier, E. F., Garnier, M. G., Mariotti, N., Didiot, C., Beck, H.,
Aebi, P., Cercellier, H., Marcus, J., Battaglia, C., Berger, H., and Titov, A. N.
(2010). Phys. Rev. B , 81, 155104.
[175] Moskalenko, S. A. (1962). Fiz. Tverd. Tela., 4, 276. [Sov. Phys. Solid State 4,
199 (1962)].
[176] Moskalenko, S. A. and Snoke, D. W. (2000). Bose-Einstein condensation of
excitons and biexcitons. Cambridge University Press, Cambridge.
[177] Mott, N. F. (1961). Phil. Mag., 6(8), 287.
[178] Mysyrowicz, A. (1980). J. Physique Colloques, 41(C7), 281.
[179] Naka, N. and Nagasawa, N. (2005). J. Lumin., 112, 11.
[180] Nandi, A., Finck, A. D. K., Eisenstein, J. P., Pfeiffer, L. N., and West, K. W.
(2012). Nature, 488, 481.
[181] Nandkishore, R. and Levitov, L. (2010). Phys. Rev. Lett., 104, 156803.
[182] Nandkishore, R. and Levitov, L. (2012). Physica Scripta, T146, 014011.
[183] Negoita, V., Snoke, D. W., and Eberl, K. (1999). Phys. Rev. B , 60, 2661.
References 45
[184] Neto, A. H. Castro (2009). Physics, 2, 30.

[185] Neto, A. H. Castro, Guinea, F., Peres, N. M. R., Novoselov, K. S., and Geim,
A. K. (2009). Rev. Mod. Phys., 81, 109.
[186] Nozières, P. and Comte, C. (1982). J. Physique, 43, 1083.
[187] O’Hara, K. E., Súilleabháin, L. Ó., and Wolfe, J. P. (1999). Phys. Rev. B , 60,
10565.
[188] O’Hara, K. E. and Wolfe, J. P. (2000). Phys. Rev. B , 62, 12909.
[189] Olaya-Castro, A., Rodrı́guez, F. J., Quiroga, L., and Tejedor, C. (2001). Phys.
Rev. Lett., 87, 246403.
[190] Östreich, T., Portengen, T., and Sham, L. J. (1996). Solid State Commun., 100,
325.
[191] Östreich, T. and Sham, L. J. (1999). Phys. Rev. Lett., 83, 3510.
[192] Peotta, S., Gibertini, M., Dolcini, F., Taddei, F., Polini, M., Ioffe, L. B., Fazio,
R., and MacDonald, A. H. (2011). Phys. Rev. B , 84, 184528.
[193] Pereverzev, S. V., Loshak, A., Backhaus, S., Davis, J. C., and Packard, R. E.
(1997). Nature, 338, 449.
[194] Phan, V.-N. and Fehske, H. (2012). New. J. Phys., 14, 075007.
[195] Pillo, T., Hayoz, J., Berger, H., Lévy, F., Schlapbach, L., and Aebi, P. (2000).
Phys. Rev. B , 61, 16213.
[196] Pines, D., Baym, G., and Pethick, C. (1969). Nature, 224, 673.
[197] Pitaevskii, L. and Stringari, S. (2001). Phys. Rev. Lett., 87, 180402.
[198] Pitaevskii, L. and Stringari, S. (2003). Bose-Einstein condensation. Oxford
University Press, Oxford.
[199] Pohlt, M., Lynass, M., Lok, J. G. S., Dietsche, W., von Klitzing, K., Eberl, K.,
and Mühle, R. (2002). Appl. Phys. Lett., 80, 2105.
[200] Portengen, T., Östreich, Th., and Sham, L. J. (1996). Phys. Rev. B , 54, 17452.
[201] Portengen, T., Östreich, Th., and Sham, L. J. (1996). Phys. Rev. Lett., 76, 3384.
[202] Prunnila, M., Laakso, S. J., Kivioja, J. M., and Ahopelto, J. (2008). Appl. Phys.
Lett., 93, 112113.
[203] Raghavan, S., Smerzi, A., Fantoni, S., and Shenoy, S. R. (1999). Phys. Rev.
A, 59, 620.
[204] Rapaport, R., Chen, G., Snoke, D., Simon, S. H., Pfeiffer, L., West, K., Liu, Y.,
and Denev, S. (2004). Phys. Rev. Lett., 92, 117405.
[205] Rasch, J. C. E., Stemmler, T., Müller, B., Dudy, L., and Manzke, R. (2008).
Phys. Rev. Lett., 101, 237602.
[206] Rashba, E. I. and Sturge, M. D. (1982). Excitons. North-Holland, Amsterdam.
[207] Remeika, M., Fogler, M. M., Butov, L. V., Hanson, M., and Gossard, A. C.
(2012). Appl. Phys. Lett., 100, 061103.
[208] Remeika, M., Graves, J. C., Hammack, A. T., Meyertholen, A. D., Fogler, M. M.,
Butov, L. V., Hanson, M., and Gossard, A. C. (2009). Phys. Rev. Lett., 102,
186803.
[209] Rice, T. M. (1977). Solid State Phys., 32, 1.
[210] Rohwer, T., Hellmann, S., Wiesenmayer, M., Sohrt, C., Stange, A., Slomski, B.,
Carr, A., Liu, Y., Avila, L. M., Kalläne, M., Mathias, S., Kipp, L., Rossnagel, K.,
and Bauer, M. (2011). Nature, 471, 490.
46 References
[211] Rontani, M. and Sham, L. J. (2000). Appl. Phys. Lett., 77, 3033.
[212] Rontani, M. and Sham, L. J. (2005). Phys. Rev. Lett., 94, 186404.
[213] Rontani, M. and Sham, L. J. (2005). Solid State Commun., 134, 89.
[214] Rontani, M. and Sham, L. J. (2009). Phys. Rev. B , 80, 075309.
[215] Rontani, M. and Sham, L. J. (2009). Properties and applications of thermoelectric
materials, p. 193. NATO Science for Peace and Security Series B: Physics and
Biophysics. Springer, Dordrecht, The Netherlands.
[216] Rossnagel, K. (2011). J. Phys.: Condens. Matter , 23, 213001.
[217] Salvo, F. J. Di, Moncton, D. E., and Waszczak, J. V. (1976). Phys. Rev. B , 14,
4321.
[218] Sarchi, D., Carusotto, I., Wouters, M., and Savona, V. (2008). Phys. Rev. B , 77,
125324.
[219] Sarma, S. Das, Adam, S., Hwang, E. H., and Rossi, E. (2011). Rev. Mod.
Phys., 83, 407.
[220] Savona, V. (2007). J. Phys.: Condens. Matter , 19, 295208.
[221] Schäfer, W. and Wegener, M. (2002). Semiconductor optics and transport phe-
nomena. Springer, Berlin.
[222] Schindler, C. and Zimmermann, R. (2008). Phys. Rev. B , 78, 045313.
[223] Schinner, G. J., Schubert, E., Stallhofer, M. P., Kotthaus, J. P., Schuh, D., Rai,
A. K., Reuter, D., Wieck, A. D., and Govorov, A. O. (2011). Phys. Rev. B , 83,
165308.
[224] Schwartz, R., Naka, N., Kieseling, F., and Stolz, H. (2012). New J. Phys., 14,
023054.
[225] Seamons, J. A., Morath, C. P., Reno, J. L., and Lilly, M. P. (2009). Phys. Rev.
Lett., 102, 026804.
[226] Semenoff, G. W. (2012). Physica Scripta, T146, 014016.
[227] Semkat, D., Sobkowiak, S., Manzke, G., and Stolz, H. (2012). Nano Lett., 12,
5055.
[228] Shah, J. (1999). Ultrafast spectroscopy of semiconductors and semiconductor
nanostructures (Second edn). Springer, Berlin.
[229] Sham, L. J. and Nakayama, M. (1979). Phys. Rev. B , 20, 734.
[230] Sham, L. J. and Rice, T. M. (1966). Phys. Rev., 144, 708.
[231] Shelykh, I. A., Solnyshkov, D. D., Pavlovic, G., and Malpuech, G. (2008). Phys.
Rev. B , 78, 041302(R).
[232] Shevchenko, S. I. (1976). Fiz. Nizk. Temp., 2, 505. [Sov. J. Low Temp. Phys. 2,
251 (1976)].
[233] Shim, Y.-P. and MacDonald, A. H. (2009). Phys. Rev. B , 79, 235329.
[234] Shin, Y., Saba, M., Pasquini, T. A., Ketterle, W., Pritchard, D. E., and Lean-
hardt, A. E. (2004). Phys. Rev. Lett., 92, 050405.
[235] Sivan, U., Solomon, P. M., and Shtrikman, H. (1992). Phys. Rev. Lett., 68, 1196.
[236] Smerzi, A., Fantoni, S., Giovanazzi, S., and Shenoy, S. R. (1997). Phys. Rev.
Lett., 79, 4950.
[237] Snoke, D. (2002). Science, 298, 1368.
[238] Snoke, D., Denev, S., Liu, Y., Pfeiffer, L., and West, K. (2002). Nature, 418,
754.
References 47
[239] Snoke, D., Wolfe, J. P., and Mysyrowicz, A. (1987). Phys. Rev. Lett., 59, 827.
[240] Snoke, D. W. (2011). Advances in Condens. Matter Phys., 2011, 938609.
[241] Sodemann, I., Pesin, D. A., and MacDonald, A. H. (2012). Phys. Rev. B , 85,
195136.
[242] Soller, H., Dolcini, F., and Komnik, A. (2012). Phys. Rev. Lett., 108, 156401.
[243] Sonin, E. B. (2009). Phys. Rev. Lett., 102, 106407.
[244] Spielman, I. B., Eisenstein, J. P., Pfeiffer, L. N., and West, K. W. (2000). Phys.
Rev. Lett., 84, 5808.
[245] Spielman, I. B., Eisenstein, J. P., Pfeiffer, L. N., and West, K. W. (2001). Phys.
Rev. Lett., 87, 036803.
[246] Stern, M., Gardimer, V., Umansky, V., and Bar-Joseph, I. (2008). Phys. Rev.
Lett., 100, 256402.
[247] Stoffel, N. G., Lévy, F., Bertoni, C. M., and Margaritondo, G. (1982). Solid
State Commun., 41, 53.
[248] Su, J.-J. and MacDonald, A. H. (2008). Nature Phys., 4, 799.
[249] Sukhatme, K., Mukharsky, Y., Chui, T., and Pearson, D. (2001). Nature, 411,
280.
[250] Suprunenko, Y. F., Cheianov, V., and Fal’ko, V. I. (2012). Phys. Rev. B , 86,
155405.
[251] Takashina, K., Nishiguchi, K., Ono, Y., Fujiwara, A., Fujisawa, T., Hirayama,
Y., and Muraki, K. (2009). Appl. Phys. Lett., 94, 142104.
[252] Taraphder, A., Koley, S., Vidhyadhiraja, N. S., and Laad, M. S. (2011). Phys.
Rev. Lett., 106, 236405.
[253] Taraphder, A., Laad, M. S., Craco, L., and Yaresko, A. N. (2008). Phys. Rev.
Lett., 101, 136410.
[254] Tiemann, L., Dietsche, W., Hauser, M., and von Klitzing, K. (2008). New J.
Phys., 10, 045018.
[255] Tiemann, L., Lok, J. G. S., Dietsche, W., von Klitzing, K., Muraki, K., Schuh,
D., and Wegscheider, W. (2008). Phys. Rev. B , 77, 033306.
[256] Timofeev, V. B. and Gorbunov, A. V. (2007). J. Appl. Phys., 101, 081708.
[257] Trauernicht, D. P., Mysyrowicz, A., and Wolfe, J. P. (1983). Phys. Rev. B , 28,
3590.
[258] Traum, M. M., Margaritondo, G., Smith, N. V., Rowe, J. E., and Salvo, F. J. Di
(1978). Phys. Rev. B , 17, 1836.
[259] Tutuc, E. and Shayegan, M. (2007). Solid State Commun., 144, 405.
[260] Tutuc, E., Shayegan, M., and Huse, D. A. (2004). Phys. Rev. Lett., 93, 036802.
[261] Uchoa, B., Reed, J. P., Gan, Y., Joe, Y. II, Fradkin, E., Abbamonte, P., and
Casa, D. (2012). Physica Scripta, T146, 014014.
[262] Vafek, O. and Case, M. J. (2008). Phys. Rev. B , 77, 033410.
[263] Vafek, O. and Yang, K. (2010). Phys. Rev. B , 81, 041401(R).
[264] van Wezel, J., Nahai-Williamson, P., and Saxena, S. S. (2010). Europhys.
Lett., 89, 47004.
[265] Velasco Jr, J., Jing, L., Bao, W., Lee, Y., Kratz, P., Aji, V., Bockrath, M., Lau,
C. N., Varma, C., Stillwell, R., Smirnov, D., Zhang, F., Jung, J., and MacDonald,
A. H. (2012). Nature Nanotech., 7, 156.
48 References
[266] Vignale, G. and MacDonald, A. H. (1996). Phys. Rev. Lett., 76, 2786.
[267] Vögele, X. P., Schuh, D., Wegscheider, W., Kotthaus, J. P., and Holleitner, A. W.
(2009). Phys. Rev. Lett., 103, 126402.
[268] Vörös, Z., Snoke, D. W., Pfeiffer, L., and West, K. (2006). Phys. Rev. Lett., 97,
016803.
[269] Vörös, Z., Snoke, D. W., Pfeiffer, L. N., and West, K. (2009). Phys. Rev.
Lett., 103, 016403.
[270] Wachter, P. (1995). J. Alloys and Compounds, 225, 133.
[271] Wachter, P., Bucher, B., and Malar, J. (2004). Phys. Rev. B , 69, 094502.
[272] Wachter, P., Jung, A., and Steiner, P. (1995). Phys. Rev. B , 51, 5542.
[273] Wachter, P. and Travaglini, G. (1985). J. Magn. Magn. Mater., 47-48, 423.
[274] Wakisaka, Y., Sudayama, T., Takubo, K., Mizokawa, T., Arita, M., Namatame,
H., Taniguchi, M., Katayama, N., Nohara, M., and Takagi, H. (2009). Phys. Rev.
Lett., 103, 026402.
[275] Wang, J.-R. and Liu, G.-Z. (2012). New J. Phys., 14, 043036.
[276] Weitz, R. T., Allen, M. T., Feldman, B. E., Martin, J., and Yacoby, A. (2010).
Science, 330, 812.
[277] Wiersma, R. D., Lok, J. G. S., Kraus, S., Dietsche, W., von Klitzing, K., Schuh,
D., Bichler, M., Tranitz, H.-P., and Wegscheider, W. (2004). Phys. Rev. Lett., 93,
266805.
[278] Wilson, J. A. (1977). Solid State Commun., 22, 551.
[279] Winbow, A. G., Leonard, J. R., Remeika, M., Kuznetsova, Y. Y., High, A. A.,
Hammack, A. T., Butov, L. V., Wilkes, J., Guenther, A. A., Ivanov, A. L., Hanson,
M., and Gossard, A. C. (2011). Phys. Rev. Lett., 106, 196806.
[280] Wouters, M. and Carusotto, I. (2007). Phys. Rev. Lett., 99, 140402.
[281] Wu, C., Mondragon-Shem, I., and Zhou, X.-F. (2008). arXiv:0809.3532.
[282] Yang, S., Hammack, A. T., Fogler, M. M., Butov, L. V., and Gossard, A. C.
(2006). Phys. Rev. Lett., 97, 187402.
[283] Yao, W. and Niu, Q. (2008). Phys. Rev. Lett., 101, 106401.
[284] Yoshioka, D. and MacDonald, A. H. (1990). J. Phys. Soc. Jpn., 59, 4211.
[285] Yoshioka, K., Chae, E., and Kuwata-Gonokami, M. (2011). Nature Commun., 2,
328.
[286] Yu, P. Y. and Cardona, M. (2004). Fundamentals of semiconductors (Third
edn). Springer, Berlin.
[287] Zagoskin, A. M. (1998). Quantum theory of many-body systems. Springer, New
York (NY).
[288] Zapata, I., Sols, F., and Leggett, A. J. (1998). Phys. Rev. A, 57, R28.
[289] Zenker, B., Ihle, D., Bronold, F. X., and Fehske, H. (2012). Phys. Rev. B , 85,
121102(R).
[290] Zhang, C.-H. and Joglekar, Y. N. (2008). Phys. Rev. B , 77, 233405.
[291] Zhang, F., Min, H., Polini, M., and MacDonald, A. H. (2010). Phys. Rev. B , 81,
041402(R).
[292] Zhao, W., Stenius, P., and İmamoğlu, A. (1997). Phys. Rev. B , 56, 5306.
[293] Ziman, J. M. (1960). Electrons and Phonons. Clarendon Press, Oxford (UK).
[294] Zimmermann, R. (2005). Solid State Commun., 134, 43.
References 49
[295] Zimmermann, S., Govorov, A. O., Hansen, W., Kotthaus, J. P., Bichler, M., and
Wegscheider, W. (1997). Phys. Rev. B , 56, 13414.
[296] Zimmermann, S., Schedelbeck, G., Govorov, A. O., Wixforth, A., Kotthaus, J. P.,
Bichler, M., Wegscheider, W., and Abstreiter, G. (1998). Appl. Phys. Lett., 73,
154.
[297] Zittartz, J. (1968). Phys. Rev., 165, 605.
[298] Zittartz, J. (1968). Phys. Rev., 165, 612.
[299] Zrenner, A., Butov, L. V., Hagn, M., Abstreiter, G., Böhm, G., and Weimann,
G. (1994). Phys. Rev. Lett., 72, 3382.
[300] Zrenner, A., Leeb, P., Schäfler, J., Böhm, G., Weimann, G., Worlock, J. M.,
Florez, L. T., and Harbison, J. P. (1992). Surf. Sci., 263, 496.

Coherent Exciton Transport

Uploaded by

Copyright:

Available Formats

Coherent Exciton Transport

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Coherent Exciton Transport

Uploaded by

Copyright:

Available Formats

arXiv:1301.1726v1 [cond-mat.

mes-hall] 9 Jan 2013

Coherent exciton transport in

1 Coherent exciton transport in semiconductors 1

Table 1.1 Excitonic insulator (EI) versus superconductor à la Bardeen-Cooper-Schrieffer (BCS).

Physical property Excitonic insulator BCS-like superconductor

Nature of the composite boson Exciton Cooper pair

1.2 Physical systems

1.2.1 Bose-Einstein condensation of optically generated excitons

1.2.2 Excitonic insulator in mixed-valence semiconductors

1.2.3 Permanent exciton condensation in bilayers

In order to investigate permanent exciton condensation in semiconductor bilayers, one

1.2.4 Graphene-based systems

1.3 Two-band versus BCS model

1.3.1 The SM-EI junction

εa (r) = G/2 + (2m)−1 ∇2 ; (1.3a)

εa (ka ) = G/2 − (2m)−1 ka2 , (1.4a)

f (r, t) = hΨ0 |ψ̃b (r, t)|Ψek i, (1.9a)

ψ̃i (r, t) = exp(i [H − µ N ] t) ψi (r) exp(−i [H − µ N ] t) , (1.10)

∆(r, r ′ ) = V2 (r − r ′ ) hΨ0 |ψ̃b (r) ψ̃a† (r ′ ) |Ψ0 i. (1.13)

and the relative phase between uk and vk is given by

When ∆k = 0, the amplitude (1.14) is the trivial solution with uk = 1 and vk = 0,

with V2, k being the Fourier component of V2 (r).

1.3.2 The N-S junction

HN-S = H0′ + H1′ + H2′ . (1.22)

The relevant two-body interaction H2′ is attractive and short-ranged,

f ′ (r, t) = hΨ′0 |ψ̃↑ (r, t)|Ψbk i, (1.27a)

with the local pair potential ∆′ (r) being defined as

∆′ (r) = −ghΨ′0 |ψ̃↓ (r) ψ̃↑ (r) |Ψ′0 i. (1.29)

The space-dependent parameter ∆′ (r) may be regarded as the center-of-mass wave

1.4 Andreev reflection at the interface between excitonic insulator

Consider ∆(z) to be a smooth complex increasing function of z, tending respectively

(ivF n · ∇ + ω) η(r) − ∆(z) χ(r) = 0, (1.31a)

where n · k1 = ω/vF , n · k2 = −ω/vF ; C1 and C2 are arbitrary constants. The first

where C3 is a constant, ϕ is the phase of ∆0 ,

1.4.2 Coherence factors in the transmission coefficients

the wave functions degenerate in ω are

with the amplitudes u0 , v0 defined as

Transmission / reflection coefficient

1.4.3 Proximity effect

1.5 A perfect insulator

dI/dV (Z +1) [e A N(εF)vF / 4 units]

1.5.1 Charge versus exciton current

depends explicitly on the built-in coherence of the electron-hole condensate ∆, and

1.5.2 A concrete example

Letting k + ≈ kF + i/(2λ), we have

Jpair = 2vF e−z/λ . (1.50)

The “disappearing” electron-hole current reappears as exciton current carried by the

∂Jcond /∂z = −4 Im{f ∗ g ∆} ,

1.6 Josephson oscillations between exciton condensates in

For a gauge transformation of the gate potential Vg → Vg − c−1 ∂χ(t)/∂t, which

1.6.2 Exciton Josephson oscillations

Ξ(x, y, t) = Ξ1 (x, y, N1 ) eiϕ1 + Ξ2 (x, y, N2 ) eiϕ2 , (1.57)

AC Josephson effect. Under the conditions ∆U ≫ N Ec /2, ∆U ≫ δJ , one easily

It follows that both k and φ oscillate in time with plasma frequency

(k, φ)eq = (−∆U N δJ /2(h̄ωJ )2 , 0). (1.64)

1.6.3 Correlated photon counting experiment

P (t) = x̂Px (t) ± iŷPy (t), (1.65)

which couples to photons: