Wa0055

Electronic Properties of Graphene in a Strong Magnetic Field
M. O. Goerbig
1
Laboratoire de Physique des Solides, Univ. Paris-Sud, CNRS UMR 8502, F-91405 Orsay, France
(Dated: November 24, 2011)
We review the basic aspects of electrons in graphene (two-dimensional graphite) exposed to a

strong perpendicular magnetic field. One of its most salient features is the relativistic quan-
arXiv:1004.3396v4 [cond-mat.mes-hall] 22 Nov 2011
tum Hall effect the observation of which has been the experimental breakthrough in identifying
pseudo-relativistic massless charge carriers as the low-energy excitations in graphene. The effect
may be understood in terms of Landau quantisation for massless Dirac fermions, which is also the
theoretical basis for the understanding of more involved phenomena due to electronic interactions.
We present the role of electron-electron interactions both in the weak-coupling limit, where the
electron-hole excitations are determined by collective modes, and in the strong-coupling regime
of partially filled relativistic Landau levels. In the latter limit, exotic ferromagnetic phases and
incompressible quantum liquids are expected to be at the origin of recently observed (fractional)
quantum Hall states. Furthermore, we discuss briefly the electron-phonon coupling in a strong
magnetic field. Although the present review has a dominating theoretical character, a close con-
nection with available experimental observation is intended.
PACS numbers: 81.05.ue, 73.43.Lp, 73.22.Pr
Contents IV. Magneto-Phonon Resonance in Graphene 31

A. Electron-Phonon Coupling 31
I. Introduction to Graphene 1 1. Coupling Hamiltonian 32
A. The Carbon Atom and its Hybridizations 2 2. Hamiltonian in terms of magneto-exciton operators 32
B. Crystal Structure of Graphene 3 B. Phonon Renormalization and Raman Spectroscopy 33
C. Electronic Band Structure of Graphene 4 1. Non-resonant coupling and Kohn anomaly 33
1. Tight-binding model for electrons on the 2. Resonant coupling 34
honeycomb lattice 4
V. Electronic Correlations in Partially Filled Landau
2. Continuum limit 7
Levels 35
D. Deformed Graphene 10
A. Electrons in a Single Relativistic Landau Level 35
1. Dirac point motion 11
1. SU(4)-symmetric model 36
2. Tilted Dirac cones 12
2. Symmetry-breaking long-range terms 37
3. Qualitative expectations for correlated electron
II. Dirac Equation in a Magnetic Field and the
phases 37
Relativistic Quantum Hall Effect 13
4. External spin-valley symmetry breaking terms 39
A. Massless 2D Fermions in a Strong Magnetic Field 13 5. Hierarchy of relevant energy scales 40
1. Quantum-mechanical treatment 14 B. SU(4) Quantum Hall Ferromagnetism in Graphene 41
2. Relativistic Landau levels 14 1. Ferromagnetic ground state and Goldstone modes 41
B. Limits of the Dirac Equation in the Description of 2. Skyrmions and entanglement 42
Graphene Landau Levels 17 3. Comparison with magnetic catalysis 45
C. Landau Level Spectrum in the Presence of an Inplane 4. The quantum Hall effect at ν = ±1 and ν = 0 46
Electric Field 19 C. Fractional Quantum Hall Effect in Graphene 48
D. Landau Levels in Deformed Graphene 19 1. Generalized Halperin wave functions 48
1. The generalized Weyl Hamiltonian in a magnetic 2. The use of generalized Halperin wave functions in
field 19 graphene 49
2. Tilted Dirac cones in a crossed magnetic and 3. Experiments on the graphene FQHE 51
electric field 20
VI. Conclusions and Outlook 51
III. Electronic Interactions in Graphene – Integer
Quantum Hall Regime 20 Acknowledgments 52
A. Decomposition of the Coulomb interaction in the
Two-Spinor Basis 22 A. Matrix Elements of the Density Operators 52
1. SU(2) valley symmetry 23
2. SU(4) spin-valley symmetric Hamiltonian 24 References 53
B. Particle-Hole Excitation Spectrum 24
1. Graphene particle-hole excitation spectrum at
B=0 25 I. INTRODUCTION TO GRAPHENE
2. Polarizability for B 6= 0 26
3. Electron-electron interactions in the random-phase
approximation: upper-hybrid mode and linear The experimental and theoretical study of graphene,
magnetoplasmons 28 two-dimensional (2D) graphite, has become a major is-
4. Dielectric function and static screening 29 sue of modern condensed matter research. A milestone
2
was the experimental evidence of an unusual quantum

ground state excited state
Hall effect reported in September 2005 by two different ( ~ 4 eV)
groups, the Manchester group led by Andre Geim and
a Columbia-Princeton collaboration led by Philip Kim ~ 4 eV
2px 2py 2pz 2px 2py 2pz
Energy
and Horst Stormer (Novoselov et al., 2005a; Zhang et al.,
2s 2s
2005).
The reasons for this enormous scientific interest are
manyfold, but one may highlight some major motiva-
tions. First, one may underline its possible technological
potential. One of the first publications on graphene in
2004 by the Geim group reported indeed an electric field 1s 1s
effect in graphene, i.e. the possibility to control the car-
Figure 1 Electronic configurations for carbon in the ground state
rier density in the graphene sheet by simple application (left) and in the excited state (right).
of a gate voltage (Novoselov et al., 2004). This effect is
a fundamental element for the design of electronic de-
vices. In a contemporary publication Berger et al. re- research, which have themselves grown to a consider-
ported on the fabrication and the electrical contacting able size and that require reviews on their own. As
of monolayer graphene samples on epitaxially grown SiC an example one may cite the review by Peres (Peres,
crystals (Berger et al., 2004). Today’s silicon-based elec- 2010), which is concerned with transport properties of
tronics reaches its limits in miniaturization, which is on graphene, or that by Kotov and co-workers on interac-
the order of 50 nm for an electric channel, whereas it has tion effects (Kotov et al., 2010). The present theoretical
been shown that a narrow graphene strip with a width review deals with electronic properties of graphene in a
of only a few nanometers may be used as a transistor strong magnetic field, and its scope is delimited to mono-
(Ponomarenko et al., 2008), i.e. as the basic electronics layer graphene. The vast amount of knowledge on bilayer
component. graphene certainly merits a review on its own.
Apart from these promising technological applications,
two major motivations for fundamental research may be
emphasized. Graphene is the first truely 2D crystal ever A. The Carbon Atom and its Hybridizations
observed in nature and possess remarkable mechanical
properties. Furthermore, electrons in graphene show rel-
In order to understand the crystallographic structure
ativistic behavior, and the system is therefore an ideal
of graphene and carbon-based materials in general, it is
candidate for the test of quantum-field theoretical models
useful to review the basic chemical bonding properties of
that have been developed in high-energy physics. Most
carbon atoms. The carbon atom possesses 6 electrons,
promenently, electrons in graphene may be viewed as
which, in the atomic ground state, are in the configu-
massless charged fermions living in 2D space, particles
ration 1s2 2s2 2p2 , i.e. 2 electrons fill the inner shell 1s,
one usually does not encounter in our three-dimensional which is close to the nucleus and which is irrelevant for
world. Indeed, all massless elementary particles hap-
chemical reactions, whereas 4 electrons occupy the outer
pen to be electrically neutral, such as photons or neutri- shell of 2s and 2p orbitals. Because the 2p orbitals (2px ,
nos.1 Graphene is therefore an exciting bridge between
2py , and 2pz ) are roughly 4 eV higher in energy than the
condensed-matter and high-energy physics, and the re- 2s orbital, it is energetically favorable to put 2 electrons
search on its electronic properties unites scientists with
in the 2s orbital and only 2 of them in the 2p orbitals
various thematic backgrounds. (Fig 1). It turns out, however, that in the presence of
Several excellent reviews witness the enormous re- other atoms, such as e.g. H, O, or other C atoms, it is
search achievements in graphene. In a first step those by favorable to excite one electron from the 2s to the third
Geim and Novoselov (Geim and Novoselov, 2007) and by 2p orbital, in order to form covalent bonds with the other
de Heer (de Heer et al., 2007) aimed at a rather global atoms.
experimental review of exfoliated and epitaxial graphene, In the excited state, we therefore have four equivalent
respectively. Furthermore, the review by Castro Neto quantum-mechanical states, |2si, |2px i, |2py i, and |2pz i.
(Castro Neto et al., 2009) was concerned with general A quantum-mechanical superposition of the state |2si
theoretical issues of electrons in graphene. Apart from with n |2pj i states is called spn hybridization. The sp1
the review by Abergel et al. (Abergel et al., 2010), more hybridization plays, e.g., an important role in the context
recent reviews concentrate on the subfields of graphene of organic chemistry (such as the formation of acetylene)
and the sp3 hybridization gives rise to the formation of
diamonds, a particular 3D form of carbon. Here, how-
1
ever, we are interested in the planar sp2 hybridization,
The neutrino example is only partially correct. The
observed oscillation between different neutrino fla-
which is the basic ingredient for the graphitic allotropes.
vors (νµ ↔ ντ ) requires indeed a tiny non-zero mass As shown in Fig. 2, the three sp2 -hybridized orbitals
(Fukuda, Y. et al. (Super-Kamiokande Collaboration), 1998). are oriented in the xy-plane and have mutual 120◦ angles.
3
H (a) (b)
(a) (b)
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
H C H
1111111111111
0000000000000
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
a2 K M’’ K’
00000000000000000
11111111111111111
0000000000000
1111111111111
0000000000000
1111111111111
C C 11111111111111111
00000000000000000
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111 00000000000000000
11111111111111111
00000000000000000
11111111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
0000000000000
1111111111111
a1 00000000000000000
11111111111111111
M
00000000000000000
11111111111111111
a*2 M’
δ2
0000000000000
1111111111111
0000000000000
1111111111111
00000
11111
00000
11111
o 00000000000000000
11111111111111111
00000
11111
00000
11111 00000000000000000
11111111111111111
00000000000000000
11111111111111111
00000
11111
00000
11111
00000
11111 120 00000000000000000
11111111111111111
Γ
00000
11111
00000000000000000
11111111111111111
δ1
00000
11111
00000
11111
00000
11111 00000000000000000
11111111111111111
00000
11111
00000
11111 00000000000000000
11111111111111111
00000000000000000
11111111111111111
00000
11111
00000
11111
00000
11111
00000
11111
00000
11111
C C K’ a1*
00000000000000000
11111111111111111
00000000000000000
11111111111111111 K
δ3
00000
11111 00000000000000000
11111111111111111
00000
11111
00000
11111
00000
11111
00000
11111
H y a=0.142 nm 00000000000000000
11111111111111111
00000000000000000
11111111111111111
00000000000000000
11111111111111111
00000
11111
00000
11111
00000
11111
H C 00000000000000000
11111111111111111
00000
11111
M’ M
00000000000000000
11111111111111111
00000000000000000
11111111111111111
00000000000000000
11111111111111111
00000000000000000
11111111111111111
00000000000000000
11111111111111111
00000000000000000
11111111111111111
x
H K M’’ K’
(c) (d) : A sublattice : B sublattice
Figure 3 (a) Honeycomb lattice. The vectors δ 1 , δ 2 , and δ 3

connect nn carbon atoms, separated by a distance a = 0.142 nm.
The vectors a1 and a2 are basis vectors of the triangular Bravais
lattice. (b) Reciprocal lattice of the triangular lattice. Its primitive
lattice vectors are a∗1 and a∗2 . The shaded region represents the first
Figure 2 (a) Schematic view of the sp2 hybridization. The orbitals Brillouin zone (BZ), with its center Γ and the two inequivalent
form angles of 120o . (b) Benzene molecule (C6 H6 ). The 6 carbon corners K (black squares) and K ′ (white squares). The thick part
atoms are situated at the corners of a hexagon and form covalent of the border of the first BZ represents those points which are
bonds with the H atoms. (c) The quantum-mechanical ground counted in its definition such that no points are doubly counted.
state of the benzene ring is a superposition of the two configurations The first BZ, defined in a strict manner, is, thus, the shaded region
which differ by the position of the π bonds. (d) Graphene may be plus the thick part of the border. For completeness, we have also
viewed as a tiling of benzene hexagons, where the H atoms are shown the three inequivalent cristallographic points M , M ′ , and
replaced by C atoms of neighboring hexagons and where the π M ′′ (white triangles).
electrons are delocalized over the whole structure.
and accessible to electronic-transport measurements only

The remaining unhybridized 2pz orbital is perpendicular since 2004.
to the plane.
For a detailed discussion of the different fabrication
A prominent chemical example for such hybridization techniques, the most popular of which are the exfoliation
is the benzene molecule the chemical structure of which technique (Novoselov et al., 2005b) and thermal graphi-
has been analyzed by August Kekulé in 1865 (Kekulé, tization of epitaxially-grown SiC crystals (Berger et al.,
1865, 1866). The molecule consists of a hexagon with 2004), we refer the reader to existing experimental re-
carbon atoms at the corners linked by σ bonds [Fig. 2 views (Geim and Novoselov, 2007; de Heer et al., 2007).
(b)]. Each carbon atom has, furthermore, a covalent Notice that, more recently, large-scale graphene has been
bond with one of the hydrogen atoms which stick out fabricated by chemical vapor deposition (Reina et al.,
from the hexagon in a star-like manner. In addition to 2009) that seems a promising technique not only for fun-
the six σ bonds, the remaining 2pz orbitals form three damental research but also for technological applications.
π bonds, and the resulting double bonds alternate with
single σ bonds around the hexagon. Because a double
bond is stronger than a single σ bond, one may ex-
B. Crystal Structure of Graphene
pect that the hexagon is not perfect. A double bond
(C=C) yields indeed a carbon-carbon distance of 0.135
nm, whereas it is 0.147 nm for a single σ bond (C–C). As already mentioned in the last section, the carbon
However, the measured carbon-carbon distance in ben- atoms in graphene condense in a honeycomb lattice due
zene is 0.142 nm for all bonds, which is roughly the av- to their sp2 hybridization. The honeycomb lattice is not
erage length of a single and a double bond. This equiv- a Bravais lattice because two neighboring sites are in-
alence of all bonds in benzene was explained by Linus equivalent from a crystallographic point of view.2 Fig.
Pauling in 1931 within a quantum-mechanical treatment 3 (a) illustrates indeed that a site on the A sublattice
of the benzene ring (Pauling, 1960). The ground state has nearest neighbors (nn) in the directions north-east,
is indeed a quantum-mechanical superposition of the two north-west, and south, whereas a site on the B sublattice
possible configurations for the double bonds, as shown has nns in the directions north, south-west, and south-
schematically in Fig. 2 (c). east. Both A and B sublattices, however, are triangular
Bravais lattices, and one may view the honeycomb lat-
These chemical considerations indicate the way to- tice as a triangular Bravais lattice with a two-atom basis
wards carbon-based condensed matter physics – any (A and B). The distance between nn carbon atoms is
graphitic compound has indeed a sheet of graphene as its a = 0.142 nm, which is the average of the single (C–C)
basic constituent. Such a graphene sheet may be viewed and double (C=C) covalent σ bonds, as in the case of
simply as a tiling of benzene hexagons, where the hydro-
gen are replaced by carbon atoms to form a neighboring
carbon hexagon [Fig. 2 (d)]. However, graphene has re-
mained the basic constituent of graphitic systems during 2 This needs to be clearly distinguished from a chemical point of
a long time only on the theoretical level. From an exper- view according to which they may be equivalent as in the case
imental point of view, graphene is the youngest allotrope of graphene where both types of sites consist of carbon atoms.
4
benzene. C. Electronic Band Structure of Graphene

The three vectors which connect a site on the A sub-
lattice with a nn on the B sublattice are given by As we have discussed in the previous section, three
a √ a √ electrons per carbon atom in graphene are involved in the
δ1 = 3ex + ey , δ 2 = − 3ex + ey , δ 3 = −aey , formation of strong covalent σ bonds, and one electron
2 2
(1) per atom yields the π bonds. The π electrons happen to
and the triangular Bravais lattice is spanned by the basis be those responsible for the electronic properties at low
vectors energies, whereas the σ electrons form energy bands far
√ away from the Fermi energy (Saito et al., 1998). This
√ 3a √
a1 = 3aex and a2 = ex + 3ey . (2) section of the introduction is, thus, devoted to a brief
2 discussion of the energy bands of π electrons within the
The modulus tight-binding approximation, which was originally calcu-
√ of the basis vectors yields the lattice spac-
ing, ã = √3a = 0.24 nm, and the area of the unit cell lated for the honeycomb lattice by P. R. Wallace in 1947
is Auc = 3ã2 /2 = 0.051 nm2 . The density of carbon (Wallace, 1947).
atoms is, therefore, nC = 2/Auc = 39 nm−2 = 3.9 × 1015
cm−2 . Because there is one π electron per carbon atom
that is not involved in a covalent σ bond, there are as 1. Tight-binding model for electrons on the honeycomb lattice
many valence electrons as carbon atoms, and their den-
sity is, thus, nπ = nC = 3.9 × 1015 cm−2 . As discussed in In the case of two atoms per unit cell, we may write
detail below, this density is not equal to the carrier den- down a trial wave function
sity in graphene, which one measures in electric transport (A) (B)
ψk (r) = ak ψk (r) + bk ψk (r), (5)
measurements.
The reciprocal lattice, which is defined with respect to where ak and bk are complex functions of the quasi-
the triangular Bravais lattice, is depicted in Fig. 3 (b). (A) (B)
momentum k. Both ψk (r) and ψk (r) are Bloch func-
It spanned by the vectors tions with

2π ey 4π (j)
X
a∗1 = √ ex − √ and a∗2 = ey . (3) ψk (r) = eik·Rl φ(j) (r + δ j − Rl ), (6)
3a 3 3a Rl
Physically, all sites of the reciprocal lattice represent where j = A/B labels the atoms on the two sublattices
equivalent wave vectors. The first Brillouin zone [BZ, A and B, and δ j is the vector which connects the sites
shaded region and thick part of the border of the hexagon of the underlying Bravais lattice with the site of the j
in Fig. 3 (b)] is defined as the set of inequivalent points atom within the unit cell. The φ(j) (r + δ j − Rl ) are
in reciprocal space, i.e. of points which may not be con- atomic orbital wave functions for electrons that are in the
nected to one another by a reciprocal lattice vector. The vicinity of the j atom situated at the position Rl − δ j
long wavelength excitations are situated in the vicinity at the (Bravais) lattice site Rl . Typically one chooses
of the Γ point, in the center of the first BZ. Furthermore, the sites of one of the sublattices, e.g. the A sublattice,
one distinguishes the six corners of the first BZ, which to coincide with the sites of the Bravais lattice. Notice
consist of the inequivalent points K and K ′ represented furthermore that there is some arbitrariness in the choice
by the vectors of the phase in Eq. (6) – instead of choosing exp(ik · Rl ),
4π one may also have chosen exp[ik·(Rl −δ j )], for the atomic
± K = ± √ ex . (4) wave functions. The choice, however, does not affect the
3 3a
physical properties of the system because it simply leads
The four remaining corners [shown in gray in Fig. 3 (b)] to a redefinition of the weights ak and bk which aquire a
may indeed be connected to one of these points via a different relative phase (Bena and Montambaux, 2009).
translation by a reciprocal lattice vector. These cristal- With the help of these wave functions, we may now
lographic points play an essential role in the electronic search the solutions of the Schrödinger equation Hψk =
properties of graphene because their low-energy excita- ǫk ψk . Multiplication of the Schrödinger equation by ψk∗
tions are centered around the two points K and K ′ , as is from the left yields the equation ψk∗ Hψk = ǫk ψk∗ ψk ,
discussed in detail in the following section. We empha- which may be rewritten in matrix form with the help
sise, because of some confusion in the literature on this of Eq. (5)
point, that the inequivalence of the two BZ corners, K
and K ′ , has nothing to do with the presence of two sub-

ak ak
(a∗k , b∗k ) Hk = ǫk (a∗k , b∗k ) Sk . (7)
lattices, A and B, in the honeycomb lattice. The form of bk bk
the BZ is an intrinsic property of the Bravais lattice, in-
dependent of the possible presence of more than one atom Here, the Hamiltonian matrix is defined as
in the unit cell. For completeness, we have also shown, (A)∗ (A) (A)∗ (B)
!
in Fig. 3 (b), the three crystallographically inequivalent ψk Hψk ψk Hψk
Hk ≡ (B)∗ (A) (B)∗ (B) = Hk† , (8)
M points in the middle of the BZ edges. ψk Hψk ψk Hψk
5
and the overlap matrix

! a3 a2
(A)∗ (A) (A)∗ (B)
ψk ψk ψk ψk B2 B1
Sk ≡ (B)∗ (A) (B)∗ (B) = Sk† (9)
ψk ψk ψk ψk A
a1
δ3
accounts for the non-orthogonality of the trial wave func-
tions. The eigenvalues ǫk of the Schrödinger equation B3
yield the electronic bands, and they may be obtained

from the secular equation Figure 4 Tight-binding model for the honeycomb lattice.
det Hk − ǫλk Sk = 0,

(10)
b. Solution for graphene with nearest-neighbor and next-
which needs to be satisfied for a non-zero solution of the nearest-neighour hopping. After these formal considera-
wave functions, i.e. for ak 6= 0 and bk 6= 0. The label λ tions, we now study the particular case of the tight-
denotes the energy bands, and it is clear that there are as binding model on the honeycomb lattice, which yields,
many energy bands as solutions of the secular equation to great accuracy, the π energy bands of graphene. Be-
(10), i.e. two bands for the case of two atoms per unit cause all atomic orbitals are pz orbitals of carbon atoms,
cell. we may omit the onsite energy ǫ0 , as discussed in the last
paragraph. We choose the Bravais lattice vectors to be
those of the A sublattice, i.e. δ A = 0, and the equivalent
site on the B sublattice is obtained by the displacement
a. Formal solution. Before turning to the specific case of δ B = δ AB = δ 3 (see Fig. 4). The nn hopping amplitude
graphene and its energy bands, we solve formally the sec- is given by the expression
ular equation for an arbitrary lattice with several atoms
per unit cell. The Hamiltonian matrix (8) may be writ- Z
ten, with the help of Eq. (6), as t ≡ d2 r φA∗ (r)∆V φB (r + δ 3 ), (14)

Hkij = N ǫ(j) sij ij
k + tk (11) and we also take into account next-nearest neighbor
(nnn) hopping which connects neighboring sites on the
same sublattice
where (δ ij ≡ δ j − δ i ),
Z
Z
Skij tnnn ≡ d2 r φA∗ (r)∆V φA (r + a1 ) (15)
sij
X
ik·Rl
k ≡ e d2 r φ(i)∗ (r)φ(j) (r + δ ij − Rl ) =
N
Rl
Notice that one may have chosen any other vector δ j or
(12)
a2 , respectively, in the calculation of the hopping ampli-
and we have defined the hopping matrix
tudes. Because of the R normalization of the atomic wave
Z functions, we have d2 rφ(j)∗ (r)φ(j) (r) = 1, and we con-
tij
X
k ≡ eik·Rl d2 r φ(i)∗ (r)∆V φ(j) (r+δ ij −Rl ) . (13) sider furthermore the overlap correction between orbitals
Rl on nn sites,
Z
Here, N is the number of unit cells, and we have sep-
s ≡ d2 r φA∗ (r)φB (r + δ 3 ). (16)
arated the Hamiltonian H into an atomic orbital part
H a = −(~2 /2m)∆ + V (r − Rl + δ j ), which satisfies the
eigenvalue equation H a φ(j) (r+δ j −Rl ) = ǫ(j) φ(j) (r+δ j − We neglect overlap corrections between all other orbitals
Rl ) and a “perturbative part” ∆V which takes into ac- which are not nn, as well as hopping amplitudes for larger
count the potential term that arises from all other atoms distances than nnn.
different from that in the atomic orbital Hamiltonian. If we now consider an arbitrary site A on the A sub-
The last line in Eq. (11) has been obtained from the fact lattice (Fig. 4), we may see that the off-diagonal terms
that the atomic wave functions φ(i) (r) are eigenstates of of the hopping matrix (13) consist of three terms corre-
the atomic Hamiltonian H a with the atomic energy ǫ(i) sponding to the nn B1 , B2 , and B3 , all of which have the
for an orbital of type i. This atomic energy plays the same hopping amplitude t. However, only the site B3 is
role of an onsite energy. The secular equation now reads described by the same lattice vector (shifted by δ 3 ) as the
det[tij λ
k − (ǫk − ǫ
(j) ij
)sk ] = 0. Notice that, if the the atoms site A and thus yields a zero phase to the hopping matrix.
on the different sublattices are all of the same electronic The sites B1 and B2 correspond to lattice vectors shifted
configuration, one has ǫ(i) = ǫ0 for all i, and one may by a2 and a3 ≡ a2 − a1 , respectively. Therefore, they
omit this on-site energy, which yields only a constant contribute a phase factor exp(ik · a2 ) and exp(ik · a3 ),
and physically irrelevant shift of the energy bands. respectively. The off-diagonal elements of the hopping
6
matrix may then be written as3 tAB k = tγk∗ = (tBA ∗

k ) , as
AB ∗ BA ∗
well as those of the overlap matrix sk = sγk = (sk ) ,
(sAA
k = sBB
k = 1, due to the above-mentioned normaliza-
tion of the atomic wave functions), where we have defined
the sum of the nn phase factors π∗
Energy
γk ≡ 1 + eik·a2 + eik·a3 . (17) K
K’ K’
The nnn hopping amplitudes yield the diagonal elements K
K
K’
of the hopping matrix, π
3
X
tAA = tBB cos(k · ai ) = tnnn |γk |2 − 3 ,

k k = 2tnnn ky
i=1
(18) kx
and one obtains, thus, the secular equation

tk − ǫk (t − sǫk )γk∗
AA
Figure 5 Energy dispersion as a function of the wave-vector com-
det =0 (19)
(t − sǫk )γk tAA
k − ǫk
ponents kx and ky , obtained within the tight-binding approxima-
tion, for tnnn /t = 0.1. One distinguishes the valence (π) band from
with the two solutions (λ = ±) the conduction (π ∗ ) band. The Fermi level is situated at the points
where the π band touches the π ∗ band. The energy is measured in
tAA
k + λt|γk |
units of t and the wave vector in units of 1/a.
ǫλk = . (20)
1 + λs|γk |
This expression may be expanded under the reasonable
c. Energy dispersion of π electrons in graphene. The en-
assumptions s ≪ 1 and tnnn ≪ t, which we further jus-
ergy dispersion (21) is plotted in Fig. 5 for tnnn /t = 0.1.
tify at the end of the paragraph,
It consists of two bands, labeled by the index λ = ±,
′
ǫλk ≃ tAA 2 2
k + λt|γk | − st|γk | = tnnn |γk | + λt|γk |
each of which contains the same number of states. Be-
" 3
# cause each carbon atom contributes one π electron and
each electron may occupy either a spin-up or a spin-
X
= t′nnn 3 + 2 cos(k · ai )
i=1
down state, the lower band with λ = − (the π or valence
v band) is completely filled and that with λ = + (the π ∗
u 3
u X or conduction band) completely empty. The Fermi level
+λtt3 + 2 cos(k · ai ), (21) is, therefore, situated at the points, called Dirac points,
i=1
where the π band touches the π ∗ band. Notice that only
where we have defined the effective nnn hopping ampli- if tnnn = 0 the energy dispersion (21) is electron-hole
tude t′nnn ≡ tnnn − st, and we have omitted the unim- symmetric, i.e. ǫλk = −ǫ−λk . This means that nnn hop-
portant constant −3tnnn in the second step. Therefore, ping and nn overlap corrections break the electron-hole
the overlap corrections simply yield a renormalization symmetry. The Dirac points are situated at the points
of the nnn hopping amplitudes. The hopping ampli- kD where the energy dispersion (21) is zero,
tudes may be determined by fitting the energy dispersion
(21) obtained within the tight-binding approximation to ǫλkD = 0. (22)
those calculated numerically in more sophisticated band-
structure calculations (Partoens and Peeters, 2006) or to Eq. (22) is satisfied when γkD = 0, i.e. when
spectroscopic measurements (Mucha-Kruczyński et al.,
2008). These yield a value of t ≃ −3 eV for the nn "√ #
hopping amplitude and t′nnn ≃ 0.1t, which justifies the 3a D √ D
ReγkD = 1 + cos (kx + 3ky )
above-mentioned expansion for t′nnn /t ≪ 1. Notice that 2
this fitting procedure does not allow for a distinction be- "√ #
3a √ D
tween the “true” nnn hopping amplitude tnnn and the + cos D
(−kx + 3ky ) = 0 (23)
contribution from the overlap correction −st. We, there- 2
fore, omit this distinction in the following discussion and
drop the prime at the effective nnn hopping amplitude, and, equally,
but one should keep in mind that it is an effective pa-
rameter with a contribution from nn overlap corrections. "√ #
3a D √ D
ImγkD = sin (kx + 3ky )
2
"√ #
3a D
√ D
3 The hopping matrix element tAB
k corresponds to a hopping from + sin (−kx + 3ky ) = 0. (24)
the B to the A sublattice. 2
7
The last equation may be satisfied by the choice kyD = 0, the eigenvalue equation Hk (tnnn = 0)Ψλk = λt|γk |Ψλk ,
and Eq. (23), thus, when which does not take into account the nnn hopping cor-
√ ! rection. Indeed, these eigenstates are also those of the
3a D 4π Hamiltonian with tnnn 6= 0 because the nnn term is pro-
1 + 2 cos k =0 ⇒ kxD = ± √ . portional to the one-matrix 1. The solution of the eigen-
2 x 3 3a
(25) value equation (29) yields
Comparison with Eq. (4) shows that there are, thus, two
inequivalent Dirac points D and D′ , which are situated γk∗ λ
at the points K and K ′ , respectively, aλk = λ b = λe−iϕk bλk (30)
|γk | k
4π
kD = ±K = ± √ ex . (26)
3 3a and, thus, the eigenstates
Although situated at the same position in the first BZ, it
is useful to make a clear conceptual distinction between 1 1
the Dirac points D and D′ , which are defined as the Ψλk = √ , (31)
2 λeiϕk
contact points between the two bands π and π ∗ , and the
crystallographic points K and K ′ , which are defined as
the corners of the first BZ. There are, indeed, situations where ϕk = arctan(Imγk /Reγk ).
where the Dirac points move away from the points K and As one may have expected, the spinor represents an
K ′ , as we will discuss in Sec. I.D. equal probability to find an electron in the state Ψλk on
Notice that the band Hamiltonian (8) respects time- the A as on the B sublattice because both sublattices are
∗
reversal symmetry, Hk = H−k , which implies ǫ−k = ǫk built from carbon atoms with the same onsite energy ǫ(i) .
for the dispersion relation. Therefore, if kD is a solution
of ǫk = 0, so is −kD , and Dirac points thus necessarily
occur in pairs. In graphene, there is one pair of Dirac
points, and the zero-energy states are, therefore, doubly
degenerate. One speaks of a twofold valley degeneracy, 2. Continuum limit
which survives when we consider low-energy electronic
excitations that are restricted to the vicinity of the Dirac In order to describe the low-energy excitations, i.e.
points, as is discussed in Sec. I.C.2. electronic excitations with an energy that is much smaller
than the band width ∼ |t|, one may restrict the exci-
tations to quantum states in the vicinity of the Dirac
d. Effective tight-binding Hamiltonian. Before considering points and expand the energy dispersion around ±K.
the low-energy excitations and the continuum limit, it is The wave vector is, thus, decomposed as k = ±K + q,
useful to define an effective tight-binding Hamiltonian, where |q| ≪ |K| ∼ 1/a. The small parameter, which gov-

0 γk∗
erns the expansion of the energy dispersion, is therefore
Hk ≡ tnnn |γk |2 1 + t . (27) |q|a ≪ 1.
γk 0
It is evident from the form of the energy dispersion
Here, 1 represents the 2 × 2 one-matrix (21) and the effective Hamiltonian that the basic entity
to be expanded is the sum of the phase factors γk . As
1 0
1= 0 1
. (28) we have already mentioned, there is some arbitrariness
in the definition of γk , as a consequence of the arbitrary
This Hamiltonian effectively omits the problem of non- choice of the relative phase between the two sublattice
orthogonality of the wave functions by a simple renor- components – indeed, a change γk → γk exp(ifk ) in Eq.
malization of the nnn hopping amplitude, as alluded to (17) for a real and non-singular function fk does not af-
above. It is therefore simpler to treat than the origi- fect the dispersion relation (21), which only depends on
nal one (8) the eigenvalue equation of which involves the the modulus of the phase-factor sum. For the series ex-
overlap matrix Sk , while it yields the same dispersion re- pansion, it turns out to be more convenient not to use the
lation (21). The eigenstates of the effective Hamiltonian expression (17), but one with fk = k · δ 3 , which renders
(27) are the spinors the expression more symmetric (Bena and Montambaux,
λ 2009),
λ ak
Ψk = , (29)
bλk
eik·δ 3 γk = eik·δ 1 + eik·δ 2 + eik·δ 3 (32)
the components of which are the probability amplitudes
of the Bloch wave function (5) on the two different sublat-
tices A and B. They may be determined by considering In the series expansion, we need to distinguish further-
8
more the sum at the K point from that at the K ′ point, the framework of the discussion of the zero-energy states
at the BZ corners. From Eq. (38) it is apparent that
3
ik·δ 3 the continuum limit |q|a ≪ 1 coincides with the limit
e±iK·δ j eiq·δ j
X
γq± ≡ e γk=±K+q = |ǫ| ≪ |t|, as described above, because |ǫq | = 3ta|q|/2 ≪
j=1
|t| then.
±i2π/3 1 2 It is convenient to swap the spinor components at the
≃ e 1 + iq · δ 1 − (q · δ 1 )
2 K ′ point (for ξ = −),

∓i2π/3 1 2 A B
+e 1 + iq · δ 2 − (q · δ 2 ) ψk,+ ψk,−
2 Ψk,ξ=+ = B , Ψ k,ξ=− = A , (39)
ψk,+ ψk,−
1 2
+ 1 + iq · δ 3 − (q · δ 3 )
2 i.e. to invert the role of the two sublattices. In this case,
= γq±(0) + γq±(1) + γq±(2) (33) the effective low-energy Hamiltonian may be represented
as
By definition of the Dirac points and their position at the
±(0) Hqeff,ξ = ξ~vF (qx σ x + qy σ y ) = ~vF τ z ⊗ q · σ, (40)
BZ corners K and K ′ , we have γq = γ±K = 0. We
limit the expansion to second order in |q|a. i.e. as two copies of the 2D Dirac Hamiltonian HD =
vF p · σ (with the momentum p = ~q), where we have
introduced the four-spinor representation
The first-order term is given by
a. First order in |q|a.
 A 
a √
h √ i ψq,+
γq±(1) = i ( 3qx + qy )e±i2π/3 − ( 3qx − qy )e∓i2π/3 B 
 ψq,+
2 Ψq =   (41)
B
 ψq,− 
3a
−iqy a = ∓ (qx ± iqy ), (34) A
ψq,−
2
√
which is obtained with the help of sin(±2π/3) = ± 3/2 in the last line via the 4 × 4 matrices
and cos(±2π/3) = −1/2. This yields the effective low-
energy Hamiltonian σ 0
τz ⊗ σ = , (42)
0 −σ
Hqeff,ξ = ξ~vF (qx σ x + ξqy σ y ), (35)
and σ ≡ (σ x , σ y ). In this four-spinor representation, the
4 first two components represent the lattice components at
where we have defined the Fermi velocity
the K point and the last two components those at the
3ta 3|t|a K ′ point. We emphasise that one must clearly distin-
vF ≡ − = (36)
2~ 2~ guish both types of pseudospin: (a) the sublattice pseu-
and used the Pauli matrices dospin is represented by the Pauli matrices σ j , where
“spin up” corresponds to the component on one sublat-
x 0 1 y 0 −i tice and “spin down” to that on the other one. A rotation
σ = and σ = . (37) within the SU(2) sublattice-pseudospin space yields the
1 0 i 0
band indices λ = ±, and the band index is, thus, in-
Furthermore, we have introduced the valley pseudospin timitely related to the sublattice pseudospin. (b) The
ξ = ±, where ξ = + denotes the K point at +K and valley pseudospin, which is described by a second set of
ξ = − the K ′ point at −K modulo a reciprocal lattice Pauli matrices τ j , the z-component of which appears in
vector. The low-energy Hamiltonian (35) does not take the Hamiltonian (40), is due to the twofold valley degen-
into account nnn-hopping corrections, which are propor- eracy and is only indirectly related to the presence of two
tional to |γk |2 and, thus, occur only in the second-order sublattices.
expansion of the energy dispersion [at order O(|q|a)2 ]. The eigenstates of the Hamiltonian (40) are the four-
The energy dispersion (21) therefore reads spinors
ǫλq,ξ=± = λ~vF |q|, (38) 
1
 
0

1  λe iϕ q 1  0
independent of the valley pseudospin ξ. We have already Ψξ=+
q,λ = √ 
ξ=−
 , Ψq,λ = √  ,
 
2 0 2 1
alluded to this twofold valley degeneracy in Sec. I.C.1, in 0 −λe iϕq
(43)
where we have, now,
4 The minus sign in the definition is added to render the Fermi

qy
velocity positive because the hopping parameter t ≃ −3 eV hap- ϕq = arctan . (44)
pens to be negative, as mentioned in the last section. qx
9
ηq clearly commutes with the massless 2D Dirac Hamil-

η=+ η=− tonian (40), and one may even express the latter as
conduction
band
(λ = +) Hqeff,ξ = ξ~vF |q|ηq , (49)
energy
0
valence which takes into account the two-fold valley degeneracy,

band in terms of the valley pseudospin ξ = ±. The band index
(λ = −) η=− η=+
λ, which describes the valence and the conduction band,
is therefore entirely determined by the chirality and the
K (ξ = +) K’ (ξ = −)
valley pseudospin, and one finds
momentum
λ = ξη , (50)
Figure 6 Relation between band index λ, valley pseudospin ξ, and
chirality η in graphene.
which is depicted in Fig. 6.
We notice finally that the chirality is a preserved quan-
tum number in elastic scattering processes induced by
b. Chirality. In high-energy physics, one defines the he- impurity potentials Vimp = V (r)1 that vary smoothly on
licity of a particle as the projection of its spin onto the the lattice scale. In this case, inter-valley scattering is
direction of propagation (Weinberg, 1995), suppressed, and the chirality thus conserved, as a conse-
q·σ quence of Eq. (50). This effect gives rise to the absence of
ηq = , (45) backscattering in graphene (Shon and Ando, 1998) and
|q|
is at the origin of Klein tunneling according to which a
which is a Hermitian and unitary operator with the eigen- massless Dirac particle is fully transmitted, under nor-
values η = ±, ηq |η = ±i = ±|η = ±i. Notice that mal incidence, through a high electrostatic barrier with-
σ describes, in this case, the true physical spin of the out being reflected (Katsnelson et al., 2006). This rather
particle. In the absence of a mass term, the helicity op- counter-intuitive result was first considered as a paradox
erator commutes with the Dirac Hamiltonian, and the and led to the formulation of a charged vacuum in the
helicity is, therefore, a good quantum number, e.g. in potential barrier (Klein, 1929), which may be indentified
the description of neutrinos, which have approximately in the framework of band theory with a Fermi level in
zero mass. One finds indeed, in nature, that all neutri- the valence band.
nos are “left-handed” (η = −), i.e. their spin is antipar-
allel to their momentum, whereas all anti-neutrinos are
“right-handed” (η = +). c. Higher orders in |q|a. Although most of the fundamen-
For massive Dirac particles, the helicity operator (45) tal properties of graphene are captured within the effec-
no longer commutes with the Hamiltonian. One may, tive model obtained at first order in the expansion of the
however, decompose a quantum state |Ψi describing a energy dispersion, it is useful to take into account second-
massive Dirac particle into its chiral components, with order terms. These corrections include nnn hopping cor-
the help of the projectors rections and off-diagonal second-order contributions from
the expansion of γk . The latter yield the so-called trigo-
1 − ηq 1 + ηq nal warping, which consists of an anisotropy in the energy
|ΨL i = |Ψi and |ΨR i = |Ψi. (46) dispersion around the Dirac points.
2 2
The diagonal second-order term, which stems from the
In the case of massless Dirac particles, with a well-defined nnn hopping, is readily obtained from Eq. (34),
helicity |Ψi = |η = ±i, one simply finds
9a2
1 − ηq 1 + ηq
ξ
Hnnn = tnnn |γqξ |2 1 ≃ tnnn |γqξ(1) |2 1 = tnnn |q|2 1,
|Ψ+ |Ψ+ 4
Li = |+i = 0, Ri = |+i = |+i
(51)
2 2
(47) independent of the valley index ξ.
and The off-diagonal second-order terms are tγq
ξ(2)
=
1 − ηq 1 + ηq −~vF a(qx − iξqy )2 /4. Notice that there is a natural
|Ψ−
Li = |−i = |−i, |Ψ−
Ri =
|−i = 0, energy hierarchy between the diagonal and off-diagonal
2 2
(48) second-order terms when compared to the leading lin-
such that one may then identify helicity and chirality. ear term; whereas the off-diagonal terms are on the or-
Because we are concerned with massless particles in the der O(|q|a) as compared to the energy scale ~vF |q|, the
context of graphene, we make this identification in the diagonal term is on the order O((tnnn /t)|q|a) and thus
remainder of this review and use the term chirality. roughly an order of magnitude smaller. We therefore
For the case of graphene, one may use the same def- take into account also the off-diagonal third order term
ξ(3)
inition (45), but the Pauli matrices define now the sub- tγq = −ξ~vF a2 (qx +iξqy )|q|2 /8, which also needs to be
lattice pseudospin instead of the true spin. The operator considered when calculating the high-energy corrections
10
ky
of the energy levels in a magnetic field (see Sec. II.B). (a)
3 (b) 2eV
Up to third order, the off-diagonal terms therefore read 0.4
qy 1.5eV
2
1eV
h a
tγqξ = ξ~vF (qx + iξqy ) − ξ (qx − iξqy )2
0.2
1
4 K’ Γ K K’
a2 2 -3 -2 -1 1 2 3 -0.4 -0.2 0.2 0.4
− |q| (qx + iξqy ) , (52) kx qx
8 -1
-0.2
-2
where one may omit the valley-dependent sign before the
-0.4
y-components of the wave vector by sweeping the sublat- -3
tice components in the spinors when changing the valley.

In order to appreciate the influence of the second-order Figure 7 Contours of constant (positive) energy in reciprocal
off-diagonal terms on the energy bands, we need to cal- space. (a) Contours obtained from the full dispersion relation (21).
culate the modulus of γqξ , The dashed line corresponds to the energy t + tnnn , which sepa-
rates closed orbits around the K and K ′ points (black lines, with
energy ǫ < t + tnnn ) from those around the Γ point (gray line, with
|q|a

3a
|γqξ | ≃ |q| 1 − ξ cos(3ϕq ) , (53) energy ǫ > t + tnnn ). (b) Comparison of the contours at energy
2 4 ǫ = 1 eV, 1.5 eV, and 2 eV around the K ′ point. The black lines
correspond to the energies calculated from the full dispersion re-
where we have used the parametrization qx = |q| cos ϕq lation (21) and the gray ones to those calculated to second order
and qy = |q| sin ϕq , and where we have restricted the within the continuum limit (54).
expansion to second order. Finally, the energy dispersion
(21) expanded to second order in |q|a reads 11111111111111111111
00000000000000000000
00000000000000000000
11111111111111111111
00000000000000000000
11111111111111111111
00000000000000000000
11111111111111111111
00000000000000000000
11111111111111111111
00000000000000000000
11111111111111111111
00000000000000000000
11111111111111111111
9a2

|q|a
t’nnn
00000000000000000000
11111111111111111111
00000000000000000000
11111111111111111111 t’nnn
00000000000000000000
11111111111111111111
ǫλq,ξ = tnnn |q|2 + λ~vF |q| 1 − ξ cos(3ϕq ) . a2
00000000000000000000
11111111111111111111
00000000000000000000
11111111111111111111
00000000000000000000
11111111111111111111
4 4 00000000000000000000
11111111111111111111
a1 t nnn
00000000000000000000
11111111111111111111
00000000000000000000
11111111111111111111 t nnn
00000000000000000000
11111111111111111111
(54) 00000000000000000000
11111111111111111111
t t
As mentioned in Sec. I.C.1, it is apparent from Eq. a a
(54) that the nnn correction breaks the electron-hole a’
symmetry ǫ−λ λ
q,ξ = −ǫq,ξ . This is, however, a rather axis of t’ t’nnn t’nnn
small correction, of order |q|atnnn /t, to the first-order deformation
effective Hamiltonian (40). The second-order expansion
of the phase factor sum γq yields a more relevant cor-
rection – the third term in Eq. (54), that is of order Figure 8 Quinoid-type deformation of the honeycomb lattice –
the bonds parallel to the deformation axis (double arrow) are mod-
|q|a ≫ |q|atnnn /t – to the linear theory. It depends ex- ified. The shaded region indicates the unit cell of the oblique
plicitly on the valley pseudospin ξ and renders the energy lattice, spanned by the lattice vectors a1 and a2 . Dashed and
dispersion anisotropic in q around the K and K ′ point. dashed-dotted lines indicate next-nearest neighbors, with charac-
The tripling of the period, due to the term cos(3ϕq ), is teristic hopping integrals tnnn and t′nnn , respectively, which are
a consequence of the symmetry of the underlying lattice different due to the lattice deformation.
and is precisely the origin of trigonal warping.
The trigonal warping of the dispersion relation is visu-
alized in Fig. 7, where we have plotted the contours of down, and Fig. 7 (a) indicates that the continuum limit
constant (positive) energy in Fourier space. The closed (54) yields extremely accurate results at these energies.
energy contours around the K and K ′ points at low en- We finally mention that, when higher-order terms in
ergy are separated by the high-energy contours around |q|a are taken into account, the chirality operator (45)
the Γ point by the dashed lines in Fig. 7 (a) at en- no longer commutes with the Hamiltonian. Chirality is
ergy |t + tnnn | the crossing points of which correspond therefore only a good quantum number in the vicinity of
to the M points. As mentioned above, the dispersion the Dirac points.
relation has saddle points at these points at the border
of the first BZ, which yield van Hove singularities in the
density of states. In Fig. 7 (b), we compare constant-
energy contours of the full dispersion relation to those D. Deformed Graphene
obtained from Eq. (54) calculated within a second-order
expansion. The contours are indistinguishable for an en- In the previous section, we have considered a per-
ergy of ǫ = |t|/3 ≃ 1 eV, and the continuum limit yields fect honeycomb lattice which is invariant under a 2π/3
rather accurate results up to energies as large as 2 eV. rotation. As a consequence, all hopping parameters
Notice that, in today’s exfoliated graphene samples on along the nn bonds δ j were equal. An interesting
SiO2 substrates, one may probe, by field-effect doping situation arises when the graphene sheet is deformed,
of the graphene sheet, energies which are on the order such that rotation symmetry is broken. In order to
of 100 meV. Above these energies the capacitor breaks illustrate the consequences, we may apply a uniax-
11
whereas in quinoid-type graphene b′ = b(1 + ε/2), which

gives
t′nnn = tnnn (1 − 2ε + bε/2d). (58)
The electronic properties of quinoid-type graphene

may then be described in terms of an effective Hamil-
tonian of the type (27),
0 γ̃k∗

Hk = tnnn hk 1 + t , (59)
γ̃k 0
with (Goerbig et al., 2008)

( "√
Figure 9 Band dispersion of the quinoid-type deformed the hon- √ t′nnn 3kx a

3
#
eycomb lattice, for a lattice distortion of δa/a = −0.4, with t = 3 hk = 2 cos 3kx a + 2 cos + ky a +ǫ
eV, tnnn /t = 0.1, ∂t/∂a = −5 eV/Å, and ∂tnnn /∂a = −0.7 eV/Å. tnnn 2 2
The inset shows a zoom on one of the Dirac points, D ′ . " √ #)
3kx a 3
+ cos − + ky a +ǫ , (60)
2 2
ial strain in the y-direction,5 a → a′ = a + δa,
in which case one obtains a quinoid-type deformation and the off-diagonal elements
(Fig. 8). The hopping t′ along δ 3 is then differ- √ !
ent from that t along δ 1 and δ 2 (Dietl et al., 2008; iky a(3/2+ǫ) 3
Farjam and Rafii-Tabar, 2009; Goerbig et al., 2008; γ̃k = 2e cos kx a + (1 − 2ǫ). (61)
2
Hasegawa et al., 2006; Wunsch et al., 2008; Zhu et al.,
2007), The resulting energy dispersion
∂t ǫλk = tnnn hk + λt|γ̃k | (62)
t → t′ = t + δa. (55)
∂a
is plotted in Fig. 9 for an unphysically large defor-
Furthermore, also four of six nnn hopping integrals are mation, ǫ = 0.4, for illustration reasons. Notice that
affected by the strain (see Fig. 8), the reversible deformations are limited by a value of
ǫ ∼ 0.1...0.2 beyond which the graphene sheet cracks
∂tnnn
tnnn → t′nnn = tnnn + δa. (56) (Lee et al., 2008). One notices, in Fig. 9, two effects
∂a of the deformation: i) the Dirac points no longer coin-
If one considers a moderate deformation ǫ ≡ δa/a ≪ 1, cide with the corners of the first BZ, the form of which
the effect on the hopping amplitudes may be estimated is naturally also modified by the deformation; and ii) the
with the help of Harrison’s law (Harrison, 1981), accord- cones in the vicinity of the Dirac points are tilted, i.e. the
ing to which t = C~2 /ma2 , where C is a numerical pref- nnn hopping term (60) breaks the electron-hole symme-
actor of order 1. One therefore finds a value try already at linear order in |q|a. These two points are
discussed in more detail in the following two subsections.
∂t 2t
= − ∼ −4.3 eV/Å and t′ = t(1 − 2ǫ) (57)
∂a a
1. Dirac point motion
which coincides well with the value ∂t/∂a ≃ 5 eV/Å,
which may be found in the literature (Dillon et al., 1977; In order to evaluate quantitatively the position of the
Saito et al., 1998). The estimation of the modified nnn Dirac points, which are defined as the contact points be-
hopping integral t′nnn is slightly more involved. One may tween the valence (λ = −) and the conduction (λ = +)
use a law tnnn (b, a) ≈ t(a) exp[−(b − a)/d(a)] familiar in bands, one needs to solve the equation γ̃kD = 0, in anal-
the context of the extended Hückel model (Salem, 1966), ogy with the case of undeformed graphene discussed in
where b is the nnn distance, and d ≈ a/3.5 ≈ 0.4 Å is Sec. I.C.1. One then finds
a caracteristic distance related to the overlap of atomic
√ ′
D D 2 t
orbitals. In undeformed graphene one has b = a 3, ky = 0 and kx a = ξ √ arccos − , (63)
3 2t
where the valley index ξ = ± denotes again the two in-
5
equivalent Dirac points D and D′ , respectively. As al-
In our simplified model, we only consider one bond length
changed by the strain. The more general case has been con- ready mentioned, the Dirac points D and D′ coincide, for
sidered by Peirera et al. (Pereira et al., 2009). However, the undistorted graphene, with the crystallographic points K
main effects are fully visible in the simplified model. and K ′ , respectively, at the corners of the first BZ. The
12
(a) (b) qy -direction, as one would expect for Dirac points, but it
is quadratic in the qx -direction (Dietl et al., 2008). This
is a general feature of merging points, which may only
occur at the Γ point or else at half a reciprocal lattice
vector G/2, i.e. in the center of a BZ border line (such as
the M points) (Montambaux et al., 2009a). Indeed, one
may show that in the case of a time-reversal symmetric
(c) (d) Hamiltonian, the Fermi velocity in the x-direction then
vanishes such that one must take into account the
quadratic order in qx in the energy band. Notice that
such hybrid semi-Dirac points, with a linear-parabolic
dispersion relation, are unaccessible in graphene be-
cause unphysically large strains would be required
(Lee et al., 2008; Pereira et al., 2009). However, such
points may exist in other physical systems such as cold
atoms in optical lattices (Hou et al., 2009; Lee et al.,
Figure 10 Topological semi-metal insulator transition in the model
(64) driven by the gap parameter ∆. (a) Two well-separated Dirac 2009; Wunsch et al., 2008; Zhao and Paramekanti,
cones for ∆ ≪ 0, as for graphene. (b) When lowering the modulus 2006; Zhu et al., 2007), the quasi-2D organic ma-
of the (negative) gap parameter, the Dirac points move towards a terial α−(BEDT-TTF)2 I3 (Katayama et al., 2006;
single point. (c) The two Dirac points merge into a single point Kobayashi et al., 2007) or VO2 /TiO2 heterostructures
at the transition (∆ = 0). The band dispersion remains linear in
the qy -direction while it becomes parabolic in the qx -direction. (d)
(Banerjee et al., 2009).
Beyond the transition (∆ > 0), the (parabolic) bands are separated
by a band gap ∆ (insulating phase). From Montambaux et al.,
2009a. 2. Tilted Dirac cones
distortion makes both pairs of points move in the same Another aspect of quinoid-type deformed graphene and
direction due to the negative value of ∂t/∂a. However, a consequence of the fact that the Dirac points no longer
unless the parameters are fine-tuned, this motion is dif- coincide with the BZ corners K and K ′ of high crystallo-
ferent, and the two pairs of points no longer coincide. graphic symmetry is the tilt of the Dirac cones. This
One further notices that Eq. (63) has (two) solutions may be appreciated when expanding the Hamiltonian
only for t′ ≤ 2t. Indeed, the two Dirac points merge (59) to linear order around the Dirac points ξkD , in-
at the characteristic point M ′′ at the border of the first stead of an expansion around the point M ′′ as in the
BZ (see Fig. 3). The point t′ = 2t is special insofar last subsection. In contrast to the undeformed case (51),
as it characterizes a topological phase transition between the diagonal components hk now yield a linear contribu-
a semi-metallic phase (for t′ < 2t) with a pair of Dirac tion (Goerbig et al., 2008), tnnn hξkD +q 1 ≃ ξ~w0 · q 1, in
cones and a band insulator (for t′ > 2t) (Dietl et al., terms of the tilt velocity
2008; Esaki et al., 2009; Montambaux et al., 2009a,b; √
Pereira et al., 2009; Wunsch et al., 2008). In the vicin- 2 3
w0x = (tnnn a sin 2θ + t′nnn a sin θ) and w0y = 0,
ity of the transition, one may expand the Hamiltonian ~
(59) around the merging point M ′′ (Montambaux et al., (65)
2009a,b), and one finds6 where we have defined θ ≡ arccos(−t′ /2t). The linear
model is therefore described by the Hamiltonian,7
~2 q 2
!
0 ∆ + 2m∗x − i~cqy
M
Hq = ~2 q 2
, (64) Hqξ = ξ~(w0 · q1 + wx qx σ x + wy qy σ y ), (66)
∆ + 2m∗x + i~cqy 0
with the renormalized anisotropic velocities
in terms of the mass m∗ = 2~2 /3ta2 and the velocity
c = 3ta/~ (Montambaux et al., 2009b). The gap param- √
3 t′ a

3ta 2
eter ∆ = t′ − 2t changes its sign at the transition – it is wx = sin θ and wy = 1+ ǫ .
~ 2 ~ 3
negative in the semi-metallic and positive in the insulat-
ing phase, where it describes a true gap (Fig. 10).
The Hamiltonian (64) has quite a particular form in
the vicinity of the merging points: it is linear in the 7 This model may be viewed as the minimal form of the generalized
Weyl Hamiltonian (with σ0 ≡ 1)
X
HW = ~vµ · q σµ ,
6 µ=0,...,3
We do not consider the diagonal part of the Hamiltonian, here,
i.e. we choose tnnn = 0, because it does not affect the position which is the most general 2 × 2 matrix Hamiltonian that yields
of the Dirac points. a linear dispersion relation.
13
Diagonalizing the Hamiltonian (66) yields the disper- observe the effect, e.g. in angle-resolved photoemis-
sion relation sion spectroscopy (ARPES) measurements (Damascelli,
q 2004) that have been successfully applied to epitaxial
ǫξλ (q) = ~w0 · q + λ~ wx2 qx2 + wy2 qy2 , (67) graphene (Bostwick et al., 2007) and graphitic samples
(Zhou et al., 2006). Notice that the Dirac cones are
and one notices that the first term (~w0 ·q) breaks indeed naturally tilted in α−(BEDT-TTF)2 I3 (Katayama et al.,
the symmetry ǫξλ (q) = ǫξλ (−q) in each valley, i.e. it tilts 2006; Kobayashi et al., 2007), where the Dirac points oc-
the Dirac cones in the direction opposite to w0 , as well as cur at positions of low crystallographic symmetry within
the electron-hole symmetry ǫλ (q) = −ǫ−λ (q) at the same the first BZ.
wave vector.8 Indeed, the linearity in q of the generalized
Weyl Hamiltonian (66) satisfies only the symmetry Hqξ =
−H−qξ
inside each valley. II. DIRAC EQUATION IN A MAGNETIC FIELD AND
THE RELATIVISTIC QUANTUM HALL EFFECT
Furthermore, one notices that the chiral symme-
try is preserved even in the presence of the tilt term
if one redefines the q chirality operator (45) as ηq = As already mentioned in the introduction, a key ex-
periment in graphene research was the discovery of a
(wx qx σ x + wy qy σ y )/ wx2 qx2 + wy2 qy2 , which naturally particular quantum Hall effect (Novoselov et al., 2005a;
commutes with the Hamiltonian (66). The eigenstates Zhang et al., 2005), which unveiled the relativistic nature
of the chirality operator are still given by of low-energy electrons in graphene. For a deeper under-
standing of this effect and as a basis for the following
1 1 parts, we discuss here relativistic massless 2D fermions
ψη = √ −iϕ , (68)
2 ηe q in a strong quantizing magnetic field (Sec. II.A). The
limits of the Dirac equation in the treatment of the high-
with tan ϕk ≡ wy qy /wx qx , and one notices that these field properties of graphene are discussed in Sec. II.B,
states are also the natural eigenstates of the Hamiltonian and we terminate this section with a discussion of the
(66). relativistic Landau level spectrum in the presence of an
One finally notices that not all values of the tilt pa- in-plane electric field (Sec. II.C) and that of deformed
rameter w0 are indeed physical. In order to be able to graphene (Sec. II.D).
associate λ = + to a positive and λ = − to a negative
energy state, one must fulfill the condition
A. Massless 2D Fermions in a Strong Magnetic Field
w̃0 < 1, (69)
in terms of the tilt parameter In order to describe free electrons in a magnetic field,
one needs to replace the canonical momentum p by the
gauge-invariant kinetic momentum (Jackson, 1999)
s 2 2
w0x w0y
w̃0 ≡ + . (70)
wx wy p → Π = p + eA(r), (72)
In the particular case of the deformation in the y-axis, where A(r) is the vector potential that generates the
which is discussed here and in which case w0y = 0 [see magnetic field B = ∇ × A(r). The kinetic momentum is
Eq. (65)], the general form of the tilt parameter reduces proportional to the electron velocity v, which must natu-
to w̃0 = w0x /wx . Unless this condition is fulfilled, the rally be gauge-invariant because it is a physical quantity.
iso-energetic lines are no longer ellipses but hyperbolas. In the case of electrons on a lattice, the substitution
In quinoid-type deformed graphene, the tilt parameter (72), which is then called Peierls substitution, remains
may be evaluated as (Goerbig et al., 2008) correct as long as the lattice spacing ã is much smaller
than the magnetic length
tnnn sin 2θ t′nnn

2
w̃0 = 2 + ≃ 2 (tt′nnn −t′ tnnn ) ≃ 0.6ǫ,
t sin θ t t r
~
(71) lB = , (73)
where we have used Eqs. (57) and (58). Even at mod- eB
erate deformations (ǫ < 0.1), the tilt of the Dirac cones which is the fundamental length scale in the presence
is on the order of 5%, and one may therefore hope to of a magnetic field. Because ã = 0.24 nm and lB ≃
p
26 nm/ B[T], this condition is fulfilled in graphene for
the magnetic fields, which may be achieved in today’s
high-field laboratories (∼ 45 T in the continuous regime
8 In the absence of the tilt term ~w0 · q1, this is a consequence
of the symmetry σz Hσz = −H, which is satisfied both by the
and ∼ 80 T in the pulsed regime).
effective Hamiltonian (27) for tnnn = 0 and the linearised version With the help of the (Peierls) substitution (72), one
(40) in each valley for undeformed graphene. may thus immediately write down the Hamiltonian for
14
charged particles in a magnetic field if one knows the of the one-dimensional harmonic oscillator. These lad-
Hamiltonian in the absence of the field, der operators play the role of a complex gauge-invariant
momentum (or velocity), and they read
H(p) → H(Π) = H(p + eA) = H B (p, r). (74)
lB lB
Notice that because of the spatial dependence of the â = √ (Πx − iΠy ) and â† = √ (Πx + iΠy ) , (78)
2~ 2~
vector potential, the resulting Hamiltonian is no longer
translation invariant, and the (canonical) momentum where we have chosen the appropriate normalization such
p = ~q is no longer a conserved quantity. For the Dirac as to obtain the usual commutation relation
Hamiltonian (40), which we have derived in the preced-
ing section to lowest order in |q|a, the Peierls substitution [â, â† ] = 1. (79)
yields
It turns out to be helpful for practical calculations to
ξ eff,ξ
HB = ξ~vF (qx σ x + qy σ y ) → HB = ξvF (Πx σ x + Πy σ y ). invert the expression for the ladder operators (78),
(75)
We further notice that, because electrons do not only ~ ~
â† + â and Πy = √ â† − â . (80)

Πx = √
possess a charge but also a spin, each energy level re- 2lB i 2lB
sulting from the diagonalization of the Hamiltonian (75)
is split into two spin branches separated by the Zeeman
effect ∆Z = gµB B, where g is the g-factor of the host 2. Relativistic Landau levels
material [g ∼ 2 for graphene (Zhang et al., 2006)] and
µB = e~/2m0 is the Bohr magneton, in terms of the In terms of the ladder operators (78), the Hamiltonian
bare electron mass m0 . In the remainder of this sec- (75) becomes
tion, we concentrate on the orbital degrees of freedom
√ ~vF

which yield the characteristic level structure of electrons ξ 0 â
HB = ξ 2 . (81)
in a magnetic field and therefore neglect the spin degree lB â† 0
of freedom, i.e. we consider spinless fermions. Effects
related to the internal degrees of freedom are discussed One remarks
√ the occurence of a characteristic frequency
in a separate section (Sec. V) in the framework of the ω ′ = 2vF /lB , which plays the role of the cyclotron
quantum-Hall ferromagnet. frequency in the relativistic case. Notice, however, that
this frequency may not be written in the form eB/mb
because the band mass is strictly zero in graphene, such
1. Quantum-mechanical treatment that the frequency would diverge.9
The eigenvalues and the eigenstates of the Hamilto-
One may easily treat the Hamiltonian (75) quantum- nian (81) are readily obtained by solving the eigenvalue
mechanically with the help of the standard canonical ξ
equation HB ψn = ǫn ψn , in terms of the 2-spinors,
quantization (Cohen-Tannoudji et al., 1973), according
to which the components of the position r = (x, y) and

un
the associated canonical momentum p = (px , py ) sat- ψn = . (82)
vn
isfy the commutation relations [x, px ] = [y, py ] = i~ and
[x, y] = [px , py ] = [x, py ] = [y, px ] = 0. As a conse- We thus need to solve the system of equations
quence of these relations, the components of the kinetic
momentum no longer commute, and, with the help of the ξ~ω ′ â vn = ǫn un and ξ~ω ′ â† un = ǫn vn , (83)
commutator relation (Cohen-Tannoudji et al., 1973)
which yields the equation
df
[O1 , f (O2 )] = [O1 , O2 ] (76) ǫ 2
dO2 â† â vn =
n
vn (84)
~ω ′
between two arbitrary operators, the commutator of
which is an operator that commutes itself with both O1 for the second spinor component. One may therefore
and O2 , one finds identify, up to a numerical factor, the second spinor com-
ponent vn with the eigenstate |ni of the usual number
~2

∂Ay ∂Ax operator â† â, with â† â|ni = n|ni in terms of the inte-
[Πx , Πy ] = −ie~ − = −i 2 , (77)
∂x ∂y lB ger n ≥ 0. Furthermore, one observes that the square
of the energy is proportional to this quantum number,
in terms of the magnetic length (73).
For the quantum-mechanical solution of the Hamilto-
nian (75), it is convenient to use the pair of conjugate op-
erators Πx and Πy to introduce ladder operators in the 9 Sometimes, a density-dependent cyclotron mass mC is formally
same manner as in the quantum-mechanical treatment introduced via the equality ω ′ ≡ eB/mC .
15
(a) (b)
The particular form of the n = 0 spinor (88) associated
+,n=4
4 +,n=3 with zero-energy states merits a more detailed comment.
+,n=2 B One notices that only the second spinor component is
energy
2 +,n=1 non-zero. Remember that this component corresponds

to the B sublattice in the K-valley (ξ = +) and to the
n=0 η
A sublattice in the K ′ -valley (ξ = −) – the valley pseu-
0 R
1 2 3
magnetic field 4 B 5
dospin therefore coincides with the sublattice pseudospin,
and the two sublattices are decoupled at zero energy. No-
-2
−,n=1 r tice that this is also the case in the absence of a magnetic
−,n=2 field, where the relation (50) between the chirality, the
−,n=3
-4
−,n=4 band index and the valley pseudospin is only valid at non-
zero values of the wave vector, i.e. not exactly at zero
Figure 11 (a) Relativistic Landau levels as a function of the mag- energy. Indeed, the chirality can no longer be defined as
netic field. (b) Semi-classical picture of cyclotron motion described the projection of the sublattice pseudospin on the direc-
by the cyclotron coordinate η, where the charged particle turns tion of propagation q/|q|, which is singular at q = 0. At
around the guiding center R. The gray region depicts the uncer-
tainty on the guiding center, as indicated by Eq. (98).
zero energy, it is therefore useful to identify the chiral-
ity with the valley pseudospin. Notice, however, that this
particularity concerns, in the absence of a magnetic field,
ǫ2n = (~ω ′ )2 n. This equation has two solutions, a posi- only a non-extensive number of states (only two) because
tive and a negative one, and one needs to introduce an- of the vanishing density of states at zero energy, whereas
other quantum number λ = ±, which labels the states of the zero-energy LL n = 0 is macroscopically degenerate,
positive and negative energy, respectively. This quantum as discussed in the following paragraphs.
number plays the same role as the band index (λ = + for
the conduction and λ = − for the valence band) in the
zero-B-field case discussed in the preceding section. One a. LL degeneracy. A particular feature of LLs, both rel-
thus obtains the spectrum (McClure, 1956) ativistic and non-relativistic ones, consists of their large
degeneracy, which equals the number of flux quanta
~vF √ NB = A × B/(h/e) threading the 2D surface A occupied
ǫλ,n = λ 2n (85) by the electron gas. From the classical point of view,
lB
this degeneracy is related to the existence of a constant
√
of relativistic Landau levels (LLs) that disperse as λ Bn of motion, namely the position of the guiding center, i.e.
as a function of the magnetic field [see Fig. 11(a)]. Notice the center of the classical cyclotron motion. Indeed, due
that, as in the B = 0 case, the level spectrum is two-fold to translation invariance in a uniform magnetic field, the
valley-degenerate. energy of an electron does not depend on the position of
Once we know the second spinor component, the first this guiding center. Translated to quantum mechanics,
component is obtained from Eq. (83), which reads un ∝ this means that the operator corresponding to this guid-
â vn ∼ â|ni ∼ |n − 1i because of the usual equations ing center R = (X, Y ) commutes with the Hamiltonian
H(p + eA).
√ √
â† |ni = n + 1|n + 1i and â|ni = n|n − 1i (86) In order to understand how the LL degeneracy is re-
lated to the guiding-center operator, we formally decom-
for the ladder operators, where the last equation is valid pose the position operator
for n > 0. One then needs to distinguish the zero-energy r=R+η (90)
LL (n = 0) from all other levels. Indeed, for n = 0, the
first component is zero because into its guiding center R and the cyclotron variable
η = (ηx , ηy ), as depicted in Fig. 11(b). Whereas the
â|n = 0i = 0 (87) guiding center is a constant of motion, as mentioned
above, the cyclotron variable describes the dynamics of
In this case one obtains the spinor the electron in a magnetic field and is, classically, the
time-dependent component of the position. Indeed, the
0 cyclotron variable is perpendicular to the electron’s ve-
ψn=0 = . (88)
|n = 0i locity and thus related to the kinetic momentum Π by
In all other cases (n 6= 0), one has positive and nega- Πy Πx
ηx = and ηy = − , (91)
tive energy solutions, which differ among each other by eB eB
a relative sign in one of the components. A convenient which, as a consequence of the commutation relations
representation of the associated spinors is given by (77), satisfy

ξ 1 |n − 1i [Πx , Πy ] 2
ψλ,n6=0 = √ . (89) [ηx , ηy ] = = −ilB , (92)
2 ξλ|ni (eB)2
16
whereas they commute naturally with the guiding-center where we have introduced the flux density
components X and Y . Equation (92) thus induces the
commutation relation 1 B
nB = 2 = h/e , (100)
2πlB
2
[X, Y ] = −[ηx , ηy ] = ilB , (93)
which is nothing other than the magnetic field measured
in order to satisfy [x, y] = 0. in units of the flux quantum h/e, as already mentioned
These commutation relations indicate that the compo- above. The ratio between the electronic density nel and
nents of the guiding-center operator form a pair of conju- this flux density then defines the filling factor
gate variables, and one may introduce, in the same man-
ner as for the kinetic momentum operator Π, the ladder nel hnel
ν= = , (101)
operators nB eB
1 1 which characterizes the filling of the different LLs.

b̂ = √ (X + iY ) and b̂† = √ (X − iY ), (94)
2lB 2lB
which again satisfy the usual commutation relations b. The relativistic quantum Hall effect. The integer
[b̂, b̂† ] = 1 and which naturally commute with the Hamil- quantum Hall effect (IQHE) in 2D electron systems
tonian. One may then introduce a number operator b̂† b̂ (v. Klitzing et al., 1980) is a manifestation of the LL
associated with these ladder operators, the eigenstates of quantization and the macroscopic degeneracy (100) of
which satisfy the eigenvalue equation each level, as well as of semi-classical electron localiza-
tion due to the sample impurities.10 In a nutshell, this
energy quantization yields a quantization of the Hall re-
b̂† b̂|mi = m|mi. (95)
sistance
One thus obtains a second quantum number, an integer h
m ≥ 0, which is necessary to describe the full quantum RH = , (102)
e2 N
states in addition to the LL quantum number n, and the
completed quantum states (88) and (89) then read where N = [ν] is the integer part of the filling factor
(101), while the longitudinal resistance vanishes.11 The
ξ ξ 0 resistance quantization reflects the presence of an incom-
ψn=0,m = ψn=0 ⊗ |mi = (96)
|n = 0, mi pressible quantum liquid with gapped single-particle and
density excitations. In the case of the IQHE, at integer
and filling factors, the gap is simply given by the energy dif-
ference between adjacent LLs which must be overcome
ξ ξ 1 |n − 1, mi by an electron that one adds to the system. Notice that
ψλn,m = ψλn ⊗ |mi = √ , (97)
2 ξλ|n, mi if one takes into account the electron spin and a vanish-
ing Zeeman effect, the condition for the occurence of the
respectively. IQHE is satisfied when both spin branches of the last LL
One may furthermore use the commutation relation n are completely filled, and one thus obtains the Hall-
(93) for counting the number of states, i.e. the degener- resistance quantization at the filling factors
acy, in each LL. Indeed, this relation indicates that one
may not measure both components of the guiding cen- ν IQHE = 2n, (103)
ter simultaneously, which is therefore smeared out over a
surface i.e. for even integers. Odd integers may principally be
observed at higher magnetic fields when the Zeeman ef-
2
∆X∆Y = 2πlB , (98) fect becomes prominent, and the energy gap is then no
longer given by the inter-LL spacing but by the Zeeman
as it is depicted in Fig. 11(b). The result (98) for the gap. This picture is naturally simplistic and needs to be
surface occupied by a quantum state may be calculated modified if one takes into account electronic interactions
rather simply if one chooses a particular gauge, such as
the Landau or the symmetric gauge for the vector poten-
tial, but its general derivation is rather involved (Imry,
1997). This minimal surface plays the same role as the 10 Strictly speaking, the IQHE requires only the breaking of trans-
surface (action) h in phase space and therefore allows lation invariance, which in a diffusive sample is due to impuri-
us to count the number of possible quantum states of a ties. In a ballistic sample, translation invariance is broken via
the sample edges (Büttiker, 1992).
given (macroscopic) surface A, 11 A simultaneous measurement of the Hall and the longitudinal re-
sistance requires a particular geometry with at least four electric
A A contacts [for a recent review on the quantum Hall effect, see Ref.
NB =
∆X∆Y
= 2 = nB × A,
2πlB
(99)
(Goerbig, 2009)].
17
– their consequences, such as the fractional quantum Hall

effect or ferromagnetic states are discussed, in the con-
text of graphene, in Sec. V.
The phenomenology of the relativistic quantum Hall
effect (RQHE) in graphene is quite similar to that of
the IQHE. Notice, however, that one is confronted not
only with the two-fold spin degeneracy of electrons in
graphene (in the absence of a strong Zeeman effect),
but also with the two-fold valley degeneracy due to the
presence of the K and K ′ points in the first BZ, which
govern the low-energy electronic properties. The filling
factor therefore changes by steps of 4 between adjacent
plateaus in the Hall resistance. Furthermore, the fill-
ing factor (101) is defined in terms of the carrier density
which vanishes at the Dirac point. This particle-hole Figure 12 Transmission spectroscopy on epitaxial multilayer
symmetric situation naturally corresponds to a half-filled graphene (Plochocka et al., 2008). The inset shows a representative
zero-energy LL n = 0, whereas all levels with λ = − are transmission spectrum. The main figure represents the positions of
the absorption lines as a function of the square-root of the mag-
completely filled and all λ = + levels are unoccupied. netic field. The dashed lines correspond to transitions calculated
In the absence of a Zeeman effect and electronic inter- at linear order, in agreement with the Dirac equation, whereas one
actions, there is thus no quantum Hall effect at ν = 0, notices downward deviations in the high-energy limit.
and the condition of a completely filled (or empty) n = 0
LL is found for ν = 2 (ν = −2). As a consequence, the
signature of the RQHE is a Hall-resistance quantization the conduction and the valence band, and the transitions
at the filling factors (Gusynin and Sharapov, 2005, 2006; have the energies
Peres et al., 2006) ~vF hp √ i
∆n,λ = 2(n + 1) − ξ 2n , (105)
ν RQHE = 2(2n + 1), (104) lB
where λ = + denotes an intraband and λ =
which needs to be contrasted to the series (103) of the − an interband transition (Abergel and Fal’ko, 2007;
IQHE in non-relativistic 2D electron systems. The se- Iyengar et al., 2007; Sadowski et al., 2006). One there-
ries (104) has indeed been observed in 2005 within the fore obtains families √of resonances the energy of which
quantum Hall measurements (Novoselov et al., 2005a; disperses as ∆n,λ ∝ B, as it has been observed in the
Zhang et al., 2005), which thus revealed the relativistic experiments [see Fig. 12, where we show the results from
character of electrons in exfoliated graphene. More re- Ref. (Plochocka et al., 2008)]. Notice that the dashed
cently, the RQHE has been observed also in epitaxial lines in Fig. 12 are fits with a single fitting parameter
graphene with moderate mobilities (Jobst et al., 2010; (the Fermi velocity vF ), which matches well all experi-
Shen et al., 2009; Wu et al., 2009). mental points for different values of n in the low-energy
regime.
Moreover, the relativistic LLs have later been directly
c. Experimental
√ observation of relativistic Landau levels.
observed in scanning-tunneling spectroscopy in graphene
The Bn dispersion of relativistic LLs has been observed
on a graphite substrate12 (Li et al., 2009a) as well as on
experimentally in transmission spectroscopy, where one
epitaxial graphene (Song et al., 2010).
shines monochromatic light on the sample and measures
the intensity of the transmitted light. Such experiments
have been performed both on epitaxial (Sadowski et al., B. Limits of the Dirac Equation in the Description of
2006) and exfoliated graphene (Jiang et al., 2007a). Graphene Landau Levels
When the monochromatic light is in resonance with a
dipole-allowed transition from the (partially) filled (λ, n) Transmission spectroscopy is an ideal tool for the study
to the (partially) unoccupied LL (λ′ , n ± 1), it is ab- of the high-energy part of the LL spectrum when consid-
sorbed due to an electronic excitation between the two ering the transitions (λ = −, n) → (λ = +, n ± 1), for
levels. Notice that, in a non-relativistic 2D electron gas, n ≫ 1. As discussed in Sec. I.C.2, one expects devi-
the only allowed dipolar transition is that from the last ations [of order O(|q|2 a2 )] from the linear dispersion in
occupied LL n to the first unoccupied one n + 1. The this limit. These deviations renormalize the energy of
transition energy is ~ωC , independently of n, and one the LLs and thus the transition energies.
therefore observes a single absorption line (cyclotron res-
onance) that is robust to electron-electron interactions,
as a consequence of Kohn’s theorem (Kohn, 1961).
In graphene, however, there are many more allowed 12 With the help of the same technique, relativistic LLs had before
transitions due to the presence of two electronic bands, been identified even in graphite (Li and Andrei, 2007).
18
In order to quantify the effect (Plochocka et al., 2008), which does not commute with â† â and which needs to be
we may use the Peierls substitution (72) and the expres- treated apart. If we neglect this trigonal-warping term
sions (80) in the terms (51) and (52) corresponding to for a moment, the LL energies are obtained from the
the higher-order diagonal and off-diagonal band terms, quadratic equation
respectively. This yields the Hamiltonian 2
4w22 − w12 a2 2

′ 2 ′ 3tnnn a
(~ω ) n − 2 n ≃ ǫn − ~ω √ n .
′ ∗
ξ h hξ
HB = , (106) 8 lB 2tlB
hξ h′ (112)
where the diagonal elements read In order to account for the trigonal-warping term in
the eigenvalue equation (109), we may use a perturbative
3tnnn a † treatment, which is justified because of the small param-
h′ = ~ω ′ √ â â, (107)
2tlB eter a/lB . There is no contribution at first order since
hn|â(†)3 |ni = 0 due to the orthogonality of the eigenstates
and the off-diagonal ones are hn|n′ i = δn,n′ . At second order, one obtains
aw1 2 a2 w22 †2

′ † 2
hξ = ξ~ω â − ξ √ â − (~ω ′ )2

2 â â . (108) a
2 2lB 4lB δn = − × 3n [1 + O(1/n)] , (113)
8 lB
Naturally, to lowest order in a/lB , one obtains again the
Hamiltonian (81). The dimensionless parameters w1 and which needs to be added to the right-hand side in Eq.
w2 are artificially added to the expressions and play the (112). Interestingly, trigonal warping thus yields the
role of fitting parameters in the comparison with experi- same correction to the energies of the relativistic LLs as
mental measurements, as will be discussed below. They the third-order term in the expansion of the band disper-
measure the deviation from the tight-binding-model ex- sion, although trigonal warping occurs at second order in
pectation, w1 = w2 = 1. Notice furthermore that, since the absence of a magnetic field, as we have discussed in
we are interested in the n ≫ 1 limit, we do not care Sec. I.C.2. This effect is due to the anisotropy of the
about corrections related to the ordering of the ladder band correction; in the presence of a magnetic field, the
operators, such that we identify a†2 a2 ≃ a2 a†2 ≃ (a† a)2 cos(3ϕq ) term in Eq. (54) is averaged over the angle ϕq ,
in the following parts. and therefore only contributes at second order in the per-
In the calculation of the LL spectrum, one may pro- turbation theory described above. This eventually yields
ceed in the same manner as in Sec. II.A.2 – the eigen- a correction of order (a/lB )2 to the LL energy, as does the
value equation (84) for the second spinor component now third-order term in the correction of the band dispersion.
becomes One finally obtains, in the large-n limit, where these
corrections become relevant, the energies of the relativis-
hξ h†ξ vn ≃ (ǫn − h′ )2 vn , (109) tic LLs (Plochocka et al., 2008)
which is asymptotically correct in the large-n limit, where vF 3tnnn a

ǫλn = ~ n (114)
we may neglect the commutator [hξ , h′ ] on the right-hand lB t lB
side of the equation.13 The combination Ĥξ ≡ hξ h†ξ is
( 2 )
vF √ 3w2 a
now interpreted as some fake Hamiltonian which needs +λ~ 2n 1 − n [1 + O(1/n)] ,
lB 8 lB
to be diagonalized in order to obtain the modified LLs.
Notice that n remains a good quantum number if one
considers h′ on the right-hand side of the eigenvalue equa- independent of the valley index ξ, where O(1/n) stands
tion. The left-hand side consists of a term for corrections of order 1/n. Notice that the fitting pa-
" # rameters w1 and w2 cannot be determined independently
2 2
2
4w2 − w1 a 2 from a fit to the spectroscopic measurement, but only the
Ĥ0 ≃ (~ω ′ )2 â† â − â† â

(110) combination w2 ≡ (w12 + 2w22 )/3. Equation (114) gener-
8 lB
alizes a calculation for the relativistic LLs when only nnn
that contains powers of â† â and thus respects the quan- hopping is taken into account (Peres et al., 2006).
tum number n, but in addition it contains the trigonal- In Fig. (12), we show experimental results obtained
warping term from high-field transmission spectroscopy on multi-layer
epitaxial graphene (Plochocka et al., 2008). Qualita-
w1 (~ω ′ )2 a †3 tively, one notices a downward renormalization of the
Ĥt.w. = −ξ √ (â + â3 ), (111) transition energies
2 2lB
∆n = ǫλ=+,n − ǫλ=−,n (115)
13 The commutator would yield relative corrections that are on in the interband regime for large values of n, in agree-
the order of 1/n and a/lB as compared to the energy scale ment with Eq. 114. Notice that because transmission
(tnnn /t)(a/lB )n that dominates h′ . spectroscopy is sensitive to energy-level differences, the
19
√ √
nnn correction in Eq. (114) yields only a correction on such that the LLs (85), which scale as B ′ = B[1 −
the order of (tnnn /t)(a/lB )/n ∼ < 1%/n at B ∼ 25 T as (E/vF B)2 ]1/4 with the magnetic field, read
compared to the energy scale t(a/lB )n of the transition,
whereas the other term yields a correction on the order ~vF √
ǫ′λ,n = λ [1 − (E/vF B)2 ]1/4 2n, (117)
of (a/lB )2 ×n ∼ 0.5%×n. The latter corrections thus be- lB
come more relevant in the large-n limit than the nnn cor-
rection. Indeed, the experiment (Plochocka et al., 2008) where the primes indicate the physical quantities in the
was not capable of probing the electron-hole symmetry moving frame of reference. When measuring the energy
breaking associated with the nnn term, whereas a quan- in the original (lab) frame of reference, the above energy
titative study of the high-energy transitions revealed a spectrum also needs to be transformed into this frame
good semi-quantitative agreement with the calculated LL of reference, which amounts to being multiplied by an-
spectrum (114). However, it has been shown that the other factor (116), such that the spectrum of relativistic
simple-minded tight-binding approach (with w = 1) un- LLs in the presence of an in-plane electric field becomes
derestimates the higher-order band corrections and that (Lukose et al., 2007)
the best fit to Eq. (114) is obtained for a value of w = 2.8. ~vF √ E
The origin of this discrepancy is yet unexplained, and ǫλ,n;k = λ [1 − (E/vF B)2 ]3/4 2n + ~ k. (118)
it may be interesting to perform high-field transmission lB B
spectroscopy measurement also on single-layer exfoliated The quantum-mechanical derivation of this result will be
graphene in order to understand whether the stronger discussed in more detail in Sec. II.D.1 in the context of
downward renormalization of the LLs is due to interlayer the generalized Weyl Hamiltonian in a magnetic field.
couplings in the epitaxial multi-layer sample.
D. Landau Levels in Deformed Graphene

C. Landau Level Spectrum in the Presence of an Inplane
Electric Field As we have discussed in Sec. I.D, a uniaxial strain
deforms the graphene sheet and modifies the electronic
A remarkable consequence of the relativistic character structure. The induced anisotropy of the Fermi velocity
of electrons in graphene and the Lorentz invariance of wx 6= wy is essentially washed out by the magnetic field,
the Dirac equation is their behavior in crossed magnetic which yields an effective averaging over the Fermi surface,
and electric fields, where the magnetic field remains per- √
vF → vF′ = wx wy . More spectacular are the two fur-
pendicular to the graphene sheet and the electric field is ther consequences of the deformation; (a) the tilt of the
applied in the plane. Remember that in a non-relativistic Dirac cones accounted for in the generalized Weyl Hamil-
2D electron systems, the electric field E = Eey (in the tonian (66) and (b) the topological phase transition due
y-direction) simply lifts the LL degeneracy and adds a to the Dirac point motion. The implication for the LL
term ~(E/B)k to the LL energies, where k is the wave spectrum are briefly reviewed in the following sections.
vector in the x-direction. At a fixed wave vector k, the
LL spacing is unaffected by the in-plane field.
The situation is different for relativistic electrons in 1. The generalized Weyl Hamiltonian in a magnetic field
graphene, as a consequence of the Lorentz invariance of
the Dirac equation. One may choose a reference frame With the help of the Peierls substitution (72) and the
in which the electric field vanishes as long as the drift expression of the kinetic momentum in terms of ladder
velocity vD = E/B is smaller than the Fermi velocity, operators (80), the generalized Weyl Hamiltonian (66)
which plays the role of an upper bound for the physically may be cast into the form
significant velocities in the same manner as the speed
of light in relativity (magnetic regime).14 In addition
p w̃
~ 2wx wy (âeiϕ + H.c.)

ξ 2
0
â
to the lifted LL degeneracy, the LL spacing is reduced HB = ξ w̃0 .
lB â† 2 (âe
iϕ
+ H.c.)
(Lukose et al., 2007; Peres and Castro, 2007), as may be (119)
seen from a Lorentz boost into the reference frame which where
moves at the drift velocity and in which the electric field
vanishes. In this reference frame, the magnetic field is w0x w0y
w̃0 eiϕ ≡ +i , (120)
reduced by the factor wx wy
p
1 − (E/vF B)2 , (116) in terms of the effective tilt parameter (70) and the angle
ϕ between the x-axis and the direction of the effective tilt
(w0x /wx , w0y /wy ), renormalized by the Fermi velocities
wx and wy in the x- and y-direction, respectively.
14 In the opposite case, vD > vF , one may choose a reference system The Hamiltonian (119) may be solved quantum-
in which the magnetic field vanishes (electric regime) (Jackson, mechanically in a straight-forward, but lengthy manner
1999). (Morinari et al., 2009; Peres and Castro, 2007). Instead,
20
one may also obtain the result in a simpler semi-classical 2. Tilted Dirac cones in a crossed magnetic and electric field
treatment (Goerbig et al., 2008), with the help of the
Onsager relation (Lifshitz and Kosevich, 1956; Onsager, One notices that the form (126) of LLs for tilted Dirac
1952) according to which the surface S(ǫ) enclosed by cones is the same as that of the LL spectrum (118) if
a trajectory of constant energy ǫ in reciprocal space is one interprets the drift velocity vD = E/vF B as an ef-
quantized as fective electric-field induced tilt. The magnetic regime
E/B < vF corresponds then to the regime of closed or-
Z ǫ
bits (w̃0 < 1) and the open hyperbolic orbits may be
2
S(ǫ)lB = (2π)2 dǫ′ ρ(ǫ′ ) = 2π(n + γ), (121)
0 identified with the electric regime E/B > vF . Mathe-
matically, the generalized Weyl Hamiltonian with an in-
where n is an integer denoting the energy level which co- plane electric field may still be cast into the form (119)
incides with the Landau level in the full quantum treat-
E
ment. The additional contribution γ is related to a Berry ξ
HB ξ
→ HE/B ξ′
= HB +~ k 1, (127)
phase acquired by an electron during its cyclotron orbit. B
Usually, one has γ = 1/2 except if there is an extra Berry where HB ξ′
is the same as that of Eq. (119) if one replaces
phase of π, which in our case yields γ = 0, as for graphene the tilt parameter w̃0 exp(iϕ) by (Goerbig et al., 2009)
with no tilt (Mikitik and Sharlai, 1999). If one considers
wξx wξy
a density of states which scales as ρ(ǫ) ∝ ǫα , the energy w̃ξ (E)eiϕξ (E) ≡ +i . (128)
levels thus scale as wx wy
Here, the renormalized tilt velocity is given by
ǫn ∼ [B(n + γ)]1/(1+α) , (122)
E×B
wξ = (wξx , wξy ) ≡ w0 − ξ , (129)
in the large-n limit. B2
Because the density of states vanishes linearly at the and the angle ϕξ is the angle between this velocity and
Dirac point, as in the case of no tilt, i.e. α = 1, the the x-axis.
scaling argument (122) yields the energy levels, The resulting energy spectrum is given by
√
√ ~v ∗ √ ~ wx wy 3/4 √ E
ǫλ,n ≃ λ 2 F n, (123) ǫξλ,n;k (E) = λ 1 − w̃ξ (E)2 2n + ~ k .
lB lB B
(130)
as for unconstrained graphene, apart from a renormaliza- Naturally one obtains the result (118) for undeformed
tion of the Fermi velocity. The latter is readily obtained graphene in an in-plane electric field, for wx = wy = vF
from the calculation of the total number of states below and w0 = 0, as well as the LL spectrum (126) for the
the energy ǫ within the positive energy cone, generalized Weyl Hamiltonian with tilted Dirac cones for
zero in-plane field (E = 0). However, the most interest-
1 1 ǫ2 ing situation arises when both the tilt and an in-plane
Z
N+ (ǫ) = dq̃x dq̃y = , field are present, in which case one observes a lifting of
(2π)2 ~2 wx wy ǫ+ (q̃)≤ǫ 2π~2 vF∗2 2
(124) the valley degeneracy that is maximal when the electric
where we have defined q̃x/y ≡ wx/y qx/y , and the renor- field is applied perpendicular to the tilt velocity, E ⊥ w0
malized Fermi velocity is (Goerbig et al., 2009).
Notice that, in order to obtain an effect on the or-
h i der of ∼ 1%, extremely large electric fields would be
vF∗2 = wx wy (1 − w̃02 )3/2 , (125) required (on the order of 106 V/m) for a 10% deforma-
tion of the lattice (Goerbig et al., 2009). It seems there-
in terms of the effective tilt parameter (70). This yields fore difficult to observe the effect in graphene, e.g. in
the result high-field transmission spectroscopy or transport mea-
√ surments, whereas the effect may be more visible in
~ wx wy √
ǫλ,n = λ (1 − w̃02 )3/4 2n (126) α−(BEDT-TTF)2 I3 , where the Dirac cones are naturally
lB tilted (Katayama et al., 2006; Kobayashi et al., 2007)
and where lower electric fields would be required for a
for the LL spectrum in the presence of a tilt, which comparable effect due to a roughly ten times smaller ef-
coincides with the one obtained from the full quan- fective Fermi velocity.
tum treatment (Morinari et al., 2009; Peres and Castro,
2007). One finally notices that the LL spacing becomes
zero for w̃0 = 1, which corresponds to the condition (69) III. ELECTRONIC INTERACTIONS IN GRAPHENE –
of maximal tilt for the Dirac cones, as discussed in Sec. INTEGER QUANTUM HALL REGIME
I.D – indeed for values of w̃0 larger than 1, the isoener-
getic lines are no longer closed elliptic orbits but open In the preceding sections, we have discussed the elec-
hyperbolas, for which the energy is not quantised. tronic properties of graphene within a one-particle model,
21
i.e. we have neglected the Coulomb interaction between expression (133) is the same as that of the fine structure
the electrons. In many materials, the one-particle picture constant α = e2 /~εc = 1/137 in quantum electrodynam-
yields the correct qualitative description of the electronic ics (Weinberg, 1995) if one replaces the Fermi velocity by
properties and is modified only quantitatively if one in- the velocity of light, which is roughly 300 times larger.
cludes the electron-electron interactions within perturba- One therefore calls αG alternatively the graphene fine
tion theory (Giuliani and Vignale, 2005; Mahan, 1993). structure constant.
Notice, however, that there exists a class of materials
– strongly correlated electron systems – the electronic
properties of which may not be described correctly, not Long-range versus short-range interactions. Another im-
even on the qualitative level, within a one-particle pic- portant aspect of interacting electrons is the range of the
ture. interaction potential. Whereas the underlying Coulomb
In order to quantify the role of the electronic inter- potential e2 /εr is long-range, short-range interaction
actions, i.e. the correlations, in graphene, one needs to models such as the Hubbard model are often – and suc-
compare the characteristic Coulomb energy Eint = e2 /εℓ cessfully – used in the description of correlated met-
at the average inter-electronic distance (ε is the dielectric als. The use of such a short-range interaction may be
constant describing the environment the 2D electron gas justified by the screening properties of interacting elec-
is embedded in) to the kinetic one Ekin (kF ) at the same trons, which are correctly captured in a Thomas-Fermi
length scale, given in terms of the Fermi wave vector kF , approach (Giuliani and Vignale, 2005; Mahan, 1993) ac-
ℓ ∼ kF−1 , cording to which the Coulomb interaction potential is
screened above a characteristic screening length λT F ∼
Eint 1/kT F .16 In 2D, the Thomas-Fermi wave vector
rs = . (131)
Ekin
kT F ≃ rs kF (134)
If this dimensionless interaction parameter becomes very is given in terms of the dimensionless interaction param-
large, rs ≫ 1, the electrons are strongly correlated. In eter (131) and the Fermi wave vector kF .17
non-relativistic 2D metals with a parabolic band dis- One notices that, for metals with a parabolic dis-
persion, Ekin ∼ ~2 kF2 /mb , the dimensionless parameter persion relation, the Thomas-Fermi wave vector is sim-
reads ply given in terms of the inverse effective Bohr radius,
mb e 2 1 kT F ∼ 1/a∗0 , independent of the electronic density. Un-
rs = ℓ∼ ∗ , (132) less the band mass is very small as compared to the bare
~2 ε a0 kF
electron mass or the dielectric constant of the host ma-
in terms of the effective Bohr radius a∗0 = a0 εm0 /mb , terial very large, the Coulomb interaction is therefore
where a0 = 0.5 Å is the Bohr radius in vacuum and screened on the atomic scale. A description of such sys-
mb /m0 the ratio between the band and the bare electron tems in the framework of short-range interaction mod-
mass. The relevance of electronic correlations therefore els, such as the Hubbard model, then becomes better
increases in the dilute limit when ℓ ≫ a∗0 . Notice that the justified than in systems with a small band mass or a
parameter rs , which is also called Wigner-Seitz radius, prominent dielectric constant (such as in 2D electron
plays the role of a length measured in units of the effective systems in GaAs heterostructures). Typical examples,
Bohr radius a∗0 . where short-range interaction model yields valuable phys-
The same argument applied to graphene yields a com- ical insight, are heavy-fermion compounds [for a review
pletely different result. Whereas the scaling of the see Ref. (Coleman, 2003)].
Coulomb energy remains the same, that of the kinetic The situation is again drastically different in graphene
energy is changed due to the linearity of the band dis- where the Thomas-Fermi wave vector (134) becomes
persion. As a consequence the dimensionless interaction 2.2 √
parameter in graphene reads kTGF ≃ αG kF ≃ kF ∼ nel , (135)
ε
Eint e2 2.2 i.e. it vanishes at the Dirac points where the carrier
αG = = ≃ , (133) density goes to zero, and the screening length then di-
Ekin ~εvF ε
verges.18 Notice that even for doped graphene, where
independent of the carrier density.15 The correlations
are therefore in an intermediate regime but may be de-
creased if the graphene sheet is embedded in an environ-
16 Notice that the Thomas-Fermi approach is restricted to static
ment with a large dielectric constant. Notice that the
screening effects, whereas dynamic screening require a more com-
plex treatment, e.g. in the framework of the random-phase ap-
proximation.
17 2
In three space dimensions, the relation reads kT 2
F ≃ rs k F .
15 In contrast to an electron system with a parabolic band disper- 18 Due to this divergence of the screening length, one principally
sion, this parameter can no longer be interpreted as a dimension- needs to describe screening beyond the level of linear-response
less radius, and we therefore use the notation αG rather than rs . theory (Katsnelson, 2006).
22
one may typically induce carrier densities on the order ρ(q) = ρ↑ (q) + ρ↓ (q). For notational convenience, we
of 1012 cm−2 , the screening length is λT F ∼> 10 nm, i.e. neglect the spin index in the following discussion keeping
much larger than the lattice scale. in mind that the spin SU(2) symmetry is respected. The
One thus comes to the conclusion that the rele- density operators may be decomposed in the basis of
vant electronic interactions in graphene are long-range the spinor wave functions (96) and (97) for relativistic
Coulomb interactions that may not be captured, in con- electrons in graphene,
trast to other materials with a parabolic band dispersion,
within models such as the Hubbard model (Herbut, 2006, †
e−iq·r ψλ′ n′ ,m′ ;ξ′ c†λn,m;ξ cλ′ n′ ,m′ ;ξ′ ,
X
ρ(q) = ψλn,m;ξ
2007a). We therefore investigate, in this section, the fate
λn,m;ξ
of the long-range Coulomb interaction in a strong mag- λ′ n′ ,m′ ;ξ ′
netic field. In Sec. III.A, we decompose the Coulomb in- (137)
teraction Hamiltonian in the two-spinor basis of the low- where
(†)
cλn,m;ξ
are fermion operators in second quantiza-
energy electronic wave functions in graphene and com-
tion that annihilate (create) an electron in the quantum
ment on its symmetry with respect to the valley pseu-
state
dospin. The role of these interactions in the particle-hole
excitation spectrum is studied in Sec. III.B, where we
1∗n |n − 1, mi

discuss the resulting collective excitations in the IQHE ψλn,m;ξ=+ = eiK·r
regime, which allows for a perturbative treatment. The λ2∗n |n, mi
strong-correlation regime of partially filled LLs (regime −λ2∗n |n, mi

and ψλn,m;ξ=− = e−iK·r . (138)
of the fractional quantum Hall effect) is presented sepa- 1∗n |n − 1, mi
rately in Sec. V. In order to avoid confusion in the case of inter-valley
coupling, we use now a representation in which the first
spinor component represents the amplitude on the A
A. Decomposition of the Coulomb interaction in the sublattice and the second on the B sublattice for both
Two-Spinor Basis valleys. Contrary to the expressions (96) and (97), the
state (138) is valid for both n = 0 p and n 6= 0 by us-
Generally, the Coulomb interaction for 2D electrons ing the short-hand notation 1∗n ≡ (1 − δn,0 )/2 and
may be accounted for by the Hamiltonian
2∗n ≡
p
(1 + δn,0 )/2. Furthermore, we have explicitly
1X taken into account the rapidly oscillating part exp(iξK·r)
Hint = v(q)ρ(−q)ρ(q), (136) due to the two different valleys, whereas the expressions
2 q
(96) and (97) are only concerned with the slowly varying
envelope function. Explicitly, the Fourier components of
in
R 2 terms of the † Fourier components ρ(q) =
the density operator (137) then read
d r exp(−iq · r)ψ (r)ψ(r) of the electronic den-
sity ψ † (r)ψ(r) and the 2D Fourier transformed 1/r ′ ′
ξ,ξ ξ,ξ
X
Coulomb potential, v(q) = 2πe2 /ε|q|. If one takes ρ(q) = Fλn,λ′ n′ (q)ρ̄λn,λ′ n′ (q), (139)
into account the electronic spin σ =↑, ↓, the Coulomb λn,λ′ n′
interaction respects the associated SU(2) symmetry, ξ,ξ ′
and the Fourier components are then simply the

sum of the densities ρσ (q) in both spin orientations, in terms of the reduced density operators
′ XD ′
E
ρ̄ξ,ξ
λn,λ′ n′ (q) = m e−i[q+(ξ−ξ )K]·R m′ c†λn,m;ξ cλ′ n′ ,m′ ;ξ′ , (140)
m,m′
which may also be interpreted as magneto-exciton operators associated with a particular inter-LL transition (see Sec.
IV.A), and the form factors
−iq·η
ξ,ξ
Fλn,λ′ n′ (q) ≡ Fλn,λ ∗ ∗
′ n′ (q) = 1n 1n′ n − 1 e n′ − 1 + λλ′ 2∗n 2∗n′ n e−iq·η n′ (141)
for intra-valley and

D E D E h i∗
+,−
Fλn,λ ∗ ∗
′ n′ (q) = λ1n′ 2n n e−i(q+2K)·η n′ − 1 − λ′ 1∗n 2∗n′ n − 1 e−i(q+2K)·η n′ = Fλ−,+
′ n′ ,λn (−q) (142)
for inter-valley processes. Here, we have used the decomposition r = R + η of the position operator into its guiding
center and cyclotron coordinate (see Sec. II.A.1) and the fact that f1 (η)f2 (R)|n, mi = f1 (η)|ni ⊗ f2 (R)|mi, for two
23
arbitrary functions f1 and f2 . The full expressions for the matrix elements in Eqs. (140), (141), and (142) may be
found in Appendix A.
In terms of the reduced density operators (140), the interaction Hamiltonian (136) reads
1X
vλξ11...ξ ξ1 ,ξ3 ξ2 ,ξ4
X
Hint = n1 ...λ4 n4 (q)ρ̄λ1 n1 ,λ3 n3 (−q)ρ̄λ2 n2 ,λ4 n4 (q),
4
(143)
2 q
λ1 n1 ...λ4 n4
ξ1 ...ξ4
(a) ν3 , ξ 1 ν4 , −ξ 2 (b) ν3 , −ξ ν4 , −ξ
right changes the valley. Such a process would re-
quire a momentum transfer of ±K, i.e. of the wave
vector connecting the two valleys, and therefore
does not respect momentum conservation, in the
ν1 , ξ 1 ν1 , ξ absence of a magnetic field. Naturally, momentum
ν2 , ξ 2 ν2 , ξ
is not a good quantum number here due to to the
(c) ν3 , −ξ ν4 , ξ (d) ν3 , ξ 1 ν4 , ξ 2 magnetic field, but momentum conservation man-
ifests itself by an exponential suppression of such
processes. In order to appreciate this point, we
need to consider the Gaussian in the form factors
ν1 , ξ
ν2 , −ξ
ν1 , ξ 1
ν2 , ξ 2
(141) and (142),
′ ′ 2 2
Figure 13 Diagrammatic representation of the interaction ver- ξ,ξ −|q+(ξ−ξ )K| lB /4
Fλn,λ′ n′ (q) ∝ e , (146)
tex (we use the short-hand notation νi = (λi ni , mi ) for the
quantum numbers); (a) vertex associated with terms of the form
′
ξ,ξ,ξ ,−ξ ′
ξ,−ξ,ξ ,ξ ′ ′ as discussed in Appendix A [see Eq. (A2)]. One
vλ n ...λ n (q) or vλ n ...λ n (q), (b) vertex of Umklapp type,
1 1
ξ,ξ,−ξ,−ξ
4 4 1 1 4 4
ξ,−ξ,−ξ,ξ
therefore sees that the interaction vertex contains
vλ (q),
(c) vertex of backscattering type, vλ (q)
1 n1 ...λ4 n4 1 n1 ...λ4 n4 a Gaussian term
and (d) vertex respecting the SU(2) valley-pseudospin symmetry
′ ′
ξ,−ξ,ξ ,−ξ ′ ′ 2
vλ (q). +|q±K|2 )l2B /4
1 n1 ...λ4 n4 vλξ,ξ,ξ ,−ξ
1 n1 ...λ4 n4
(q) ∝ e−(q (147)
′ 2 2
/4)l2B /2
∼ e−(|q | +|K|
2 2 2 2
where the interaction vertex is defined as ∼ e−|K| lB /8 ∼ e−#lB /a ,
2πe2 ξ1 ,ξ3 where # represents an unimportant numerical fac-
vλξ11...ξ4
n1 ...λ4 n4 (q) = F (−q)Fλξ22 ,ξ4
n2 ,λ4 n4 (q).
ε|q| λ1 n1 ,λ3 n3 tor and where we have shifted the momentum
(144) q′ = q ± K/2 in the second step. The processes
associated with the diagram in Fig. 13(a) are thus
2
exponentially suppressed in lB /a2 ≃ 104 /B[T] and
1. SU(2) valley symmetry may safely be neglected in the range of physically
accessible magnetic fields.
One notices that, in contrast to the SU(2) symmetry
associated with the physical spin, the Hamiltonian (143) • The same fate is reserved for the diagram in Fig.
does not respect a similar valley-pseudospin symmetry 13(b), which represents a process of Umklapp type.
due to possible inter-valley couplings. An SU(2) valley- In this case, the vertex reads
pseudospin symmetry would be respected for the case
2
+|q−K|2 )l2B /4
ξ1 = ξ3 and ξ2 = ξ4 , i.e. if the interaction vertex (144) vλξ,ξ,−ξ,−ξ
1 n1 ...λ4 n4
(q) ∝ e−(|q+K| (148)
−|K|2 l2B /2 −#l2B /a2
vλξ11...ξ ∼e ∼e ,
n1 ...λ4 n4 (q) ∝ δξ1 ,ξ3 δξ2 ,ξ4 .
4
(145)
2
One may show, however, that the SU(2) valley- which is again exponentially small in lB /a2 .
pseudospin symmetry is approximately respected when
• The situation is different for backscattering-type di-
considering the different classes of interaction vertices de-
agrams [Fig. 13(c)], in which case the interaction
picted in Fig. 13.
vertex is
• Consider the diagram in Fig. 13(a), which rep- 2
+|q±K|2 )l2B /4
′
,−ξ ′ vλξ,−ξ,−ξ,ξ
1 n1 ...λ4 n4
(q) ∝ e−(|q±K| . (149)
resents a vertex of the type vλξ,ξ,ξ
1 n1 ...λ4 n4
(q) or
′ ′
vλξ,−ξ,ξ ,ξ
1 n1 ...λ4 n4
(q). In this case, the particle on the One may then redefine the wave vector q′ = q ± K,
left remains in the same valley whereas that on the which is eventually an integration variable in the
24
interaction Hamiltonian (143), and the interaction The above argument, which generalizes symmetry con-
vertex becomes siderations for the interactions in a single relativistic
LL (Alicea and Fisher, 2006; Doretto and Morais Smith,
2πe2 ′2 2
2007; Goerbig et al., 2006; Herbut, 2007b), shows that
vλξ,−ξ,−ξ,ξ
1 n1 ...λ4 n4
(q′ ) ∝ ′
e−q lB /2 (150)
ε|q ∓ K| although the valley SU(2) symmetry is not an exact sym-
2πe2 −q′2 l2B /2 metry, such as the SU(2) symmetry associated with the
∼ e . physical spin, it is approximately respected by the long-
ε|K|
range Coulomb interaction. Valley-symmetry breaking
As an order of magnitude, with |K| ∼ 1/a, one then terms are due to lattice effects beyond the continuum
notices that the backscattering interactionpvertex is limit and therefore suppressed by the small factor a/lB ,
suppressed by a factor of a/lB ∼ 0.005 × B[T] as which quantifies precisely corrections due to effects on the
compared to the leading energy scale e2 /εlB . lattice scale. If one takes into account the additional spin
degree of freedom, the resulting four-fold spin-valley de-
• The leading interaction vertex is therefore the generacy may then be described within the larger SU(4)
SU(2) valley-pseudospin symmetric one depicted in symmetry group, which turns out to be relevant in the
Fig. 13(d), for which the rapidly oscillating contri- description of strong-correlation effects in partially filled
bution at K vanishes, as may be seen directly from LLs (Sec. V).
the form factors (142).
2. SU(4) spin-valley symmetric Hamiltonian
The SU(4)-symmetric part of the interaction Hamiltonian (143) finally reads
sym 1X
vλsym
X
Hint = 1 n1 ...λ4 n4
(q)ρ̄λ1 n1 ,λ3 n3 (−q)ρ̄λ2 n1 ,λ4 n4 (q), (151)
2 q
λ1 n1 ...λ4 n4
where the symmetric interaction vertex is
2πe2
vλsym
1 n1 ...λ4 n4
(q) = Fλ1 n1 ,λ3 n3 (−q)Fλ2 n2 ,λ4 n4 (q), (152)
ε|q|
in terms of the reduced density operators
m e−iq·R m′ c†λn,m;ξ,σ cλ′ n′ ,m′ ;ξ,σ ,

X ξ,ξ X X X
ρ̄λn,λ′ n′ (q) ≡ ρλn,λ′ n′ (q) = (153)
ξ=± ξ=± σ=↑,↓ m,m′
where we have explicitly taken into account the spin in- In the remainder of this section, we neglect the
dex σ =↑, ↓ in the last line. symmetry-breaking part of the Hamiltonian and consider
We finally notice that the graphene form factors (141) the Coulomb interaction to respect the SU(2) valley sym-
may also be rewritten in terms of the LL form factors metry.
Fn,n′ (q) = n e−iq·η n′ , (154)
which arise in a similar decomposition of the Coulomb B. Particle-Hole Excitation Spectrum
interaction in Landau states in the non-relativistic 2D
electron gas, as The considerations of the previous subsection allow us
to discuss the role of the Coulomb interaction within
Fλn,λ′ n′ (q) = 1∗n 1∗n′ Fn−1,n′ −1 (q) + λλ′ 2∗n 2∗n′ Fn,n′ (q). a perturbative approach in the IQHE regime for ν =
(155) ±2(2n + 1), where the (non-interacting) ground state
To summarize the differences and the similarities between is
√ non-degenerate
√ and√separated by the cyclotron gap
the interaction Hamiltonians in graphene and the non- 2(~vF /lB )( n + 1 − n) from its excited states. Quite
relativistic 2D electron system, one first realizes that its generally, the inter-LL transitions evolve into coher-
structure is the same if one replaces the LL form factor ent collective excitations, as a consequence of these
(154) by the graphene form factors (141) and if one takes Coulomb interactions. Prominent examples in the
into account the larger (approximate) internal symmetry non-relativistic 2D electron gas are the upper-hybrid
SU(4), due to the spin-valley degeneracy, instead of the mode (sometimes also called magneto-plasmon), which
spin SU(2) symmetry. is the magnetic-field counterpart of the usual 2D plas-
25
(a) (b) close to the Fermi energy, the allowed excitations imply
a wave-vector transfer that lies in between q = 0 (Ia) and
CB q = 2qF (Ib). At non-zero values ǫ of the transfered en-
Ib Ia ergy, one needs to search for available quantum states at
EF
larger wave vectors, and the particle-hole pair wave vec-
IIa
tor is then restricted to ǫ/~vF < q < 2qF + ǫ/~vF , as a
−qF qF q
IIb
consequence of the linear dispersion relation in graphene.
This gives rise to the region I, which describes the intra-
VB
band particle-hole continuum, and its linear boundaries
in the PHES described by the spectral function in Fig.
14(b).
Figure 14 Zero-field particle-hole exctiation spectrum for doped In addition to intraband excitations, one notices that
graphene. (a) Possible intraband (I) and interband (II) single-pair interband excitations become possible above a threshold
excitations in doped graphene. The exctiations close to the Fermi energy of ǫF , where an electron at the top of the VB (at
energy may have a wave-vector transfer comprised between q = 0 q = 0) may be promoted to an empty state slightly above
(Ia) and q = 2qF (Ib), in terms of the Fermi wave vector qF .
(b) Spectral function Im Π0 (q, ω) in the wave-vector/energy plane. the Fermi energy. The associated wave-vector transfer is
The regions corresponding to intra- and interband excitations are naturally q = qF . The point (qF , ǫF ) marks the bottom
denoted by (I) and (II), respectively. of the region II in Fig. 14(b), which determines the region
of allowed interband excitations (interband particle-hole
continuum). Direct interband excitations with zero wave-
mon (Giuliani and Vignale, 2005), and magneto-excitons vector transfer are possible above an energy of 2ǫF .
(Kallin and Halperin, 1984). In the present subsection, Another aspect of the PHES in Fig. 14 is the strong
we discuss how these modes manifest themselves in concentration of spectral weight around the central diag-
graphene in comparison with the non-relativistic 2D elec- onal ω = ~vF |q|. This concentration is a particularity
tron gas. of graphene due to the electrons’ chirality (Polini et al.,
2008). Indeed, if one considers a 2qF backscattering
process in the vicinity of the Fermi energy in the CB,
1. Graphene particle-hole excitation spectrum at B = 0 Eq. (50) indicates that the chirality, i.e. the projec-
tion of the sublattice-pseudospin on the direction of the
Before discussing the particle-hole excitation spectrum wave vector, is preserved. The inversion of the direc-
(PHES) for graphene in the IQHE regime, we briefly re- tion of propagation in the 2qF process would therefore
view the one for B = 0 as well as its associated collec- require an inversion of the A and B sublattices that is
tive modes (Ando, 2006a; Hwang and Das Sarma, 2007; not supported by most of the scattering or interaction
Shung, 1986; Wunsch et al., 2006). Quite generally, the processes. This effect is reflected by a strong suppres-
PHES is determined by the spectral function sion of the spectral weight when approaching the right
1 boundary of the region I in the PHES associated with
S(q, ω) = − Im Π(q, ω), (156) processes of the type Ib in Fig. 14(a). Similarly, the
π
conservation of the electrons’ chirality (50) favors 2qF
which may be viewed as the spectral weight of the al- processes in the interband region (II) and the suppres-
lowed particle-hole excitations, in terms of the polariz- sion of direct q = 0 interband excitations of the type IIa
ability Π(q, ω), which plays the role of a density-density in Fig. 14(a). Notice that, although the direction of the
response function (Giuliani and Vignale, 2005; Mahan, wave vector is inverted in a 2qF process, this indicates
1993). still the absence of backscattering because the group ve-
The particle-hole excitations for non-interacting elec- locity v = ∇q ǫλq /~ = λvF q/|q| remains unchanged – the
trons in doped graphene are depicted in Fig. 14.19 In change in the sign due to the inversion of the wave vector
contrast to the PHES of electrons in a single parabolic is indeed canceled by the one associated with the change
band (the non-relativistic 2D electron gas), there are two of the band index.
different types of excitations: intraband excitations [la-
beled by I in Fig. 14(a)], where both the electron and the
hole reside in the conduction band (CB), and interband a. Formal calculation of the spectral function. In order
excitations [labeled by II in Fig. 14(a)], where an electron to obtain the spectral function, it is apparent from
is promoted from the valence band (VB) to the CB. In un- Eq. (156) that one needs to calculate the polarizabil-
doped graphene, there exist naturally only interband ex- ity Π(q, ω) of the 2D system, which may be found with
citations (II). If the electron and the hole have an energy the help of the Green’s functions G(q, ω),
dω ′ X
Z
Π(q, ω) = −iTr G(q′ , ω ′ )G(q + q′ , ω + ω ′ ),
19
2π ′
We consider here only the case of a Fermi energy ǫF in the con- q
duction band, for simplicity. (157)
26
(a) (b) level of the random-phase approximation (RPA), which

amounts to calculating a geometric series of bubble
q+q’,ω+ω’
II diagrams and which has shown to yield reliable re-
sults for doped graphene (Hwang and Das Sarma, 2007;
Sabio et al., 2008; Wunsch et al., 2006). The RPA has
I also been applied to undoped graphene (González et al.,
q’, ω’ 1994, 1999), but its validity has been questioned
(Gangadharaiah et al., 2008; Kotov et al., 2007) because
of the vanishing density of states, which would require to
take into account diagrams beyond the RPA (Katsnelson,
2006). The RPA polarizability then becomes
Figure 15 (a) Particle-hole bubble diagram (polarizability), in
terms of Green’s functions G(q, ω) (lines). (b) Spectral func- Π0 (q, ω)
tion Im ΠRP A (q, ω) for doped graphene in the wave-vector/energy ΠRP A (q, ω) = , (160)
plane. The electron-electron interactions are taken into account εRP A (q, ω)
within the RPA. We have chosen αG = 1 here.
in terms of the polarizability (158) for non-interacting
electrons and the dielectric function
where Tr means the trace since the Greens functions
are 2 × 2 matrices as a consequence of the matrix char- 2πe2 0
εRP A (q, ω) = 1 − Π (q, ω). (161)
acter of the kinetic Hamiltonian. Diagrammatically, ε|q|
the polarizability may be represented by the so-called
bubble diagram shown in Fig. 15(a), and one finds The spectral function associated with the RPA polar-
for non-interacting electrons in graphene (Ando, 2006a; izability (160), which is shown in Fig. 15(b), reveals
Hwang and Das Sarma, 2007; Polini et al., 2008; Shung, the characteristic coherent 2D plasmon mode, which
1986; Wunsch et al., 2006) corresponds to the solution of the implicit equation
εRP A (q, ωpl ) = 0 and the dispersion relation of which
reads
′
λ
ǫ̃λq′ +q

g X n ǫ̃ q ′ − n
Π0 (q, ω) = ′ Cλλ′ (q′ , q′ +q), r
A ′ ′ ǫ̃λq′ − ǫ̃λq′ +q + ~ω + iδ 2e2 ǫF
q ,λ,λ ωpl (q) ≃ q (162)
(158) ~2 ε
where ǫ̃λq = λ~vF |q| − ǫF is the energy of the quantum in the small-q limit (Shung, 1986; Wunsch et al., 2006).
state ψλ (q) measured from the Fermi energy ǫF , g = 4 Interestingly, this equation is valid also for non-
takes into account the four-fold spin-valley degeneracy, relativistic electrons in conventional 2D electron sys-
and n(ǫ̃λq ) is the Fermi-Dirac distribution function that tems (Stern, 1967) if one takes into account the dif-
reduces to a Heavyside step function n(ǫ̃λq ) = Θ(−ǫ̃λq ) ference in the density dependence of the Fermi energy
at zero temperature. Equation (158) is nothing other (ǫF = πnel /mb for non-relativistic 2D electrons and
than the Lindhard function (Giuliani and Vignale, 2005; √
ǫF = ~vF πnel p in graphene) as well as that in the Fermi
Mahan, 1993), apart from the factor velocity (vF = 2ǫF /mb for non-relativistic electrons,
1 + λλ′ cos θq′ ,q′ +q as compared to a constant vF in graphene). Notice that
Cλλ′ (q′ , q′ + q) ≡ , (159) the dispersion relation is restricted to small values of q
2
(as compared to the Fermi wave vector kF ), whereas the
in terms of the angle θq′ ,q′ +q between q′ and q′ + q, numerical solution presented in Fig. 15 indicates that
which takes into account the particular chirality prop- the asymptotic dependence of the plasmon mode is in-
erties of graphene – as already mentioned above, this deed given by the central diagonal ωuh (q) ∼> vF q (Shung,
chirality factor vanishes for backscattering processes, i.e. 1986; Wunsch et al., 2006). Therefore, contrary to the
for intraband (λ = λ′ ) 2qF processes with θq′ ,q′ +q = π plasmon in 2D metals with a parabolic dispersion rela-
as well as for interband (λ = −λ′ ) q = 0 processes with tion, the plasmon in graphene does not enter region I,
θq′ ,q′ +q = 0 or 2π. but only the interband particle-hole continuum (region
Notice finally that the quantity δ in Eq. (158) is an II). In this region, the Landau damping is less efficient,
infinitesimal energy in the case of pure graphene and may and the coherence of the mode thus survives to a certain
be used (for finite values) as a phenomenological measure extent without decaying into incoherent particle-hole ex-
of the impurity broadening δ ≃ ~/τ , in terms of a life time citations.
τ of the excitations.
2. Polarizability for B 6= 0
The
b. Polarizability in the random-phase approximation.
diagrammatic approach is particularly adapted for tak- In the case of a strong magnetic field applied perpen-
ing into account the electronic interactions on the dicular to the graphene sheet, one needs to take into ac-
27
count the quantization of the kinetic energy into rela- larizability. One finds a similar expression for the zero-
tivistic LLs described in Sec. II.A.1, as well as the spino- temperature polarizability of non-interacting electrons as
ξ
rial eigenfunctions ψnλ,m in the calculation of the po- in Eq. (158),
√
X Θ(ǫF − λ~ω ′ √n) − Θ(ǫF − λ′ ~ω ′ n′ )
Π0B (q, ω) =g √ √ |Fλn,λ′ n′ (q)|2 , (163)
λ~ω ′ n − λ′ ~ω ′ n′ + ~ω + iδ
′
λn,λ n′
in terms of the graphene form (a) (b)

√ factors (141) and the char-
acteristic frequency ω ′ = 2vF /lB introduced in Sec. II
II.A.1. One notices that the first part is nothing other
than a Lindhard function (Giuliani and Vignale, 2005;
Mahan, 1993) for relativistic
√ LLs filled up to the Fermi
energy ǫF = ~(vF /lB ) 2NF , which is chosen to be situ- I
ated between a completely filled (NF ) and a completely
empty (NF +1) LL in the CB (IQHE regime). The second
factor is the modulus square of the graphene form factors
which plays the role of the chirality factor Cλ,λ′ (q′ , q′ +q)
in the absence of a magnetic field (Roldán et al., 2009, Figure 16 Particle-hole excitation spectrum for graphene in a per-
2010; Shizuya, 2007).20 pendicular magnetic field. We have chosen NF = 3 in the CB and
a LL broadening of δ = 0.05~vF /lB (a) and δ = 0.2~vF /lB (b).
As for the zero-field case, one may distinguish two con-
The ultaviolet cutoff is chosen such that Nc = 70.
tributions to the polarizability, one that may be viewed as
a vacuum polarizability Πvac (q, ω) and that stems from
interband excitations when the Fermi level is at the Dirac which verify Πλn (q, ω) = −Π−λn (q, ω). The vacuum po-
point, and a second one that comes from intraband exci- larization may then be defined as
tations in the case of doped graphene. Because undoped
graphene with zero carrier density does not correspond
Nc
to an IQHE situation – as we have already discussed in X
Sec. II.A, the zero-energy LL n = 0 is only half-filled Πvac (q, ω) = − Π+n (q, ω) (166)
n=1
then –, we define, here, the vacuum polarizability with
respect to the completely filled zero-energy level. where Nc is a cutoff that delimits the validity of the
In order to describe more explicitly the different con- continuum approximation. Notice that, already in the
tributions to the polarizability, we define the auxiliary absence of a magnetic field, the validity of the contin-
quantities (Roldán et al., 2009) uum approximation is delimited by a maximal energy
2 ∼ t. One may then introduce √ an upper level cutoff with
|Fλn,λn′ (q)| the help of ǫNc = ~(vF /lB ) 2Nc ∼ t, that leads to
Πλn,λ′ n′ (q, ω) = √ √
λ~vF n − λ′ ~vF n′ + ~ω + iδ Nc ∼ 104 /B[T ], which is a rather high value even for
+ (ω + → −ω − ) (164) strong magnetic fields. However, due to the fact that the
separation between LLs in graphene decreases with n,
where ω + → ω − indicates the replacement ~ω + iδ → it is always possible to obtain reliable semi-quantitative
−~ω − iδ and results from smaller values of Nc .
The spectral function S(q, ω) = −Im Π0B (q, ω)/π is
shown in Fig. 16 for NF = 3 and for two different values
X n−1
X of the phenomenologically introduced LL broadening δ.
Πλn (q, ω) = Πλn,λ′ n′ (q, ω) (165)
One notices first that the spectral weight is restricted to
λ′ n′ =0
Nc
the two regions I and II corresponding to the intraband
X X and interband particle-hole continuum, respectively, in
+ Πλn,λ′ n′ (q, ω) + Πλn,−λn (q, ω)
the zero-field limit. This is not astonishing because the
λ′ n′ =n+1
electron-hole pair wave vector remains a good quantum
number also in the presence of a magnetic field and be-
cause the overlap between the electron and hole wave
20 A similar expression for the polarizability has also been obtained functions is largest in these regions; if one considers the
in Refs. (Berman et al., 2008; Tahir and Sabeeh, 2008) though pair with its overall charge neutrality, its motion is unaf-
with approximate form factors. fected by the magnetic field. Indeed, the pair momentum
28
may be viewed as the sum of the pseudomomenta asso- (a) (b)

ciated with the guiding-center variable for the electron,
2
R × ez /lB , and the hole −R′ × ez /lB
2
, respectively. Each II II
of the pseudomomenta is naturally a constant of motion

because so is the guiding center, as we have discussed in I I
Sec. II.A.1. One therefore obtains the relation
2 2
q = ∆R × ez /lB or ∆R = |q|lB , (167)
where ∆R = R − R′ is the distance between the guiding

center of the electron and that of the hole. The bound- Figure 17 Particle-hole excitation spectrum for graphene in a per-
aries of the PHES in Fig. 16 may then be obtained from pendicular magnetic field. The Coulomb interaction is taken into
the decomposition (90), which yields η ′ − η ≤ ∆R ≤ account within the RPA. We have chosen NF = 3 in the CB and a
LL broadening of δ = 0.05~vF /lB (a) and δ = 0.2~vF /lB (b). The
η′ √
+ η, with the help of√the average values η ≡ h|η|i = ultaviolett cutoff is chosen such that Nc = 70.
lB 2n + 1 and η ′ = lB 2n′ + 1,
√ √ √ √
2n′ + 1− 2n + 1 ≤ qlB ≤ 2n′ + 1+ 2n + 1. (168)
Interestingly, it is this structure of diagonal lines that
√ survives in more disordered samples in which the hori-
Because the energy scales also with n, one obtains the
zontal lines associated with inter-LL transitions start to
linear boundaries of the particle-hole continua as in the
overlap, i.e. once the LL spacing is smaller than the level
zero-field case mentioned above.
broadening δ.22
In contrast to these similarities with the zero-field
PHES, one notices a structure in the spectral weight
that is due to the strong magnetic field. As a con-
sequence of the relativistic LL quantization, the spec- 3. Electron-electron interactions in the random-phase
tral weight approximation: upper-hybrid mode and linear magnetoplasmons
√ corresponds√ to inter-LL
√ transitions at ener-
gies ω = 2~(vF /lB )( n − λ n′ ), where n > NF and
n′ ≤ NF (for λ = +) or n′ > 0 (for λ = −). For larger The PHES of non-interacting electrons in graphene
values of NF , or quite generally when increasing the en- gives already insight into the collective modes which
√
ergy, the level density increases due to the n scaling of one may expect once electron-electron interactions are
the LLs and the transitions. The LL structure is there- taken into account. Indeed the regions of large spec-
fore only visible in the lower part of PHES, in the clean tral weight evolve into coherent collective excitations as
limit δ = 0.05~vF /lB [Fig. 16(a)], whereas the inter-LL a consequence of these interactions. Because the re-
transitions are blurred at larger energies or even for the gions of large spectral weight are organised in lines par-
lower transitions in the case of less clean samples [Fig. allel to the central diagonal ω = ~vF |q|, as mentioned
16(b), for δ = 0.2~vF /lB ].21 above, one may expect that the dominant collective ex-
In addition to the (blurred) LL structure in the PHES, citations are roughly linearly dispersing modes instead
one notices another structure of the spectral weight, of the more conventional weakly dispersing magneto-
which is organised in lines parallel to the central diago- excitons, which emerge from the inter-LL transitions
nal ω = ~vF |q|. This weight is again decreased when ap- (Kallin and Halperin, 1984). It has though been argued
proaching the right boundary of the intraband continuum that such magneto-excitons may play a role at low en-
(region I) and the left one of the interband continuum ergies in clean samples with low doping (Iyengar et al.,
(region II), due to the above-mentioned chirality proper- 2007) and that they may renormalize the cyclotron en-
ties of electrons in graphene. The emergence of diagonal ergy at zero wave vector (Bychkov and Martinez, 2008).
lines is a consequence of the graphene form factors (141) As in the zero-field case, we take into account the
the modulus square of which intervenes in the polariza- Coulomb interaction within the RPA [see Eq. (160)].
tion function. Indeed, these form factors Fλ(n+m),λ′ n (q) The resulting spectral function is shown in Fig. 17 for the
are (associated) Laguerre polynomials with n + 1 zeros same choice of parameters as in the non-interacting case
(Gradshteyn and Ryzhik, 2000) due to the overlap be- (Fig. 16). Furthermore, we have chosen a dimensionless
tween the wave function of the hole in the level λ′ n and interaction parameter αG = 1, here, which corresponds
that of the electron in the LL λ(n + m) (Roldán et al.,
2010). These zeros in the inter-LL transitions are organ-
ised in lines that disperse parallel to the central diagonal
22 This behavior is in stark contrast to that of non-relativistic 2D
and thus give rise to zones of vanishing spectral weight.
electrons, where the LL spacing is constant and given by the
cyclotron energy ~eB/mb . If this quantity is larger than the
level broadening δ, there is no qualitative difference between low
and high energies, and the horizontal lines associated with the
21 The value δ = 0.2~vF /lB is a reasonable estimate for today’s inter-LL excitations (multiples of the cyclotron energy) remain
exfoliated graphene samples on an SiO2 substrate (Ando, 2007a). well separated.
29
to a dielectric constant of ε ≃ 2. (a) (b)

One notices the prominent mode that evolves in the
originally forbidden region for particle-hole excitations.
This mode, which is called upper-hybrid mode, is the
magnetic-field descendent of the 2D plasmon mode (162)
and acquires a density-dependent cyclotron gap ωC =
√
eBvF2 /ǫF = eBvF /~vF πnel . Its dispersion relation in
the small-q limit is then given by
q Figure 18 Static polarization function Π0 (q, ω = 0) for non-
ωuh (q) ≃ 2 (q) + ω 2
ωpl (169) interacting electrons in graphene without (a) and with (b) a mag-
C netic field. To
√ compare both √ cases, we have chosen a Fermi wave
√ vector qF = 2NF + 1/lB = 7/lB that corresponds to NF = 3.
s 2
2~e2 vF πnel eBvF2

The full black line represents the total polarizability, whereas the
≃ q + √ , red dotted and the blue dashed lines show the intraband and the
~2 ε ~vF πnel
interband contributions, respectively. From (Roldán et al., 2010).
as may be shown easily within a hydrodynamic ap-
proach that has been successfully applied to the upper- One notices that up to 2qF , the zero-field static po-
hybrid mode in non-relativistic 2D electron systems larizability [Fig. 18(a)] remains constant. Indeed, the
(Chiu and Quinn, 1974). It is apparent from Fig. 17 interband contribution (blue dashed line) increases lin-
that this mode remains coherent also within the region early with the wave vector (Ando, 2006a; González et al.,
II, which corresponds to the B = 0 interband particle- 1999)
hole continuum.
In addition to the upper-hybrid mode, one notices lin- q
− Πvac (q) = (172)
early dispersing coherent modes in the regions I and II 4~vF
that emerge from the lines of large spectral weight in and thus cancels the linear decrease of the intraband con-
the non-interacting PHES, as expected from the quali- tribution (red dotted line),
tative discussion above. We may call these modes lin-
ear magneto-plasmons (Roldán et al., 2009, 2010) in or- q
− Πdop (|q| ∼
< 2qF ) ≃ ρ(ǫF ) 1 − , (173)
der to distinguish them clearly from the upper-hybrid 2qF
mode (169) and the weakly dispersing magneto-excitons
at low doping (Iyengar et al., 2007). These modes are where
more prominent in the interband than in the intraband ǫF
ρ(ǫF ) = (174)
region although they are also visible there.√They are gen- 2π~2 vF2
uine to graphene with its characteristic Bn LLs that
is the density of states per unit area at the Fermi energy.
inevitably overlap in energy, above a critical LL n, if
At wave vectors larger than 2qF , the intraband contri-
level broadening is taken into account, and they may not
bution dies out, and the polarizability is dominated by
be captured in the usual magneto-exciton approximation
interband excitations.
where the collective modes are adiabatically connected
With the help of these two contributions, we may
to the inter-LL excitations (Bychkov and Martinez, 2008;
rewrite the static dielectric function (170) as
Iyengar et al., 2007; Kallin and Halperin, 1984).
παG Πdop (q)

εRP A (q) = ε∞ 1 + , (175)
2ε∞ Πvac (q)
4. Dielectric function and static screening
where we have defined
We terminate this section with a brief discussion of the π
ε∞ ≡ εRP A (|q| → ∞) = 1 + αG , (176)
dielectric function (161) in the static limit 2
i.e. the value that the static dielectric function ap-
2πe2 0
εRP A (q) ≡ εRP A (q, ω = 0) = 1 − Π (q, ω = 0), proaches at large wave vectors. Notice that this is pre-
ε|q| cisely the RPA result for the intrinsic dielectric constant
(170) for undoped graphene (González et al., 1999). The above
by comparing the B 6= 0 to the zero-field case, as shown form of the static dielectric function may be cast into a
in Fig. 18. As mentioned above, one may distinguish Thomas-Fermi form,
two separate contributions to the static polarizability,
the vacuum polarizability Πvac (q) ≡ Πvac (q, ω = 0) due qF
εT F (q) ≃ ε∞ 1 + α∗G , (177)
to interband excitations and the intraband contribution q
Πdop (q) ≡ Πdop (q, ω = 0), which is only present in doped
graphene, in terms of the effective coupling constant
αG αG
α∗G = = , (178)
Π0 (q, ω = 0) = Πvac (q) + Πdop (q). (171) ε∞ 1 + παG /2
30
1.0 αG∗
0.8
2/π 0.6
Out[22]=
0.4
0.2
αG
2 4 6 8 10
Figure 19 Effective coupling constant α∗G as a function of the bare

coupling αG . The dashed line indicates the asymptotic value 2/π
obtained for large values of the bare coupling (rs ≫ 1). Figure 20 Static dielectric function for graphene in the IQHE
regime for NF = 1, 2 and 3 (blue, red and green curves, in increas-
ing order). From (Roldán et al., 2010).
which is plotted in Fig. 19.
One notices that interband excitations yield a contri-
bution to the dielectric constant that originally takes into
account the dielectric environment in which the graphene negligible role [see Fig. 18(a)], are still governed by the
sheet is embedded, ε → ε∗ = εε∞ . This is a direct conse- bare coupling constant αG ∼ vscr (q ≪ qF )/~vF q. How-
quence of the linear behavior of the vacuum polarization ever, processes at or above the Fermi wave vector, such
(172) as a function of the wave vector and thus specific as those that are relevant in the electronic transport, are
to graphene. Furthermore, one may also define an effec- characterized by the effective coupling constant α∗G ∼
tive Thomas-Fermi wave vector qT∗ F = qT F /ε∞ = α∗G qF , > qF )/~vF q, which saturates at a value of 2/π as
vscr (q ∼
which describes the screening length in the presence of mentioned above. If we consider the value (133), αG ≃
the vacuum polarization. As a consequence of the sat- 2.2, for the bare coupling constant of graphene in vac-
uration of the effective coupling constant (178) at large uum, the effective coupling is roughly four times smaller
values of αG , the effective Thomas-Fermi vector is thus α∗G ≃ 0.5, such that the electrons in doped graphene ap-
always on the order of the Fermi wave vector unless proach the weak-coupling limit. The situation is differ-
αG ≪ 1, where α∗G ∼ αG . The relevant effective param- ent in undoped graphene, where recent renormalization-
eters are summarized in the table below for freestanding group (Herbut et al., 2009a,b; Juričić et al., 2009) and
graphene and graphene on mainly used substrates.23 lattice gauge theoretical calculations (Drut and Lähde,
2009a,b) indicate a flow towards strong coupling at mod-
erate values of αG .
graphene ε αG ε∞ α∗G In Fig. 18(b), we have plotted the static polarizability
in vacuum 1 2.2 4.5 0.5 for graphene in the IQHE regime. Qualitatively, the re-
on SiO2 2.5 0.9 2.4 0.38 sult agrees with the zero-field behavior, with a (roughly)
on h-BN 2.3 1 2.4 0.39 linearly increasing vacuum polarizability and a decreas-
on SiC 5.5 0.4 1.6 0.25 ing intraband contribution, apart from some superim-
posed oscillations due to the overlap functions that are
Table I Dielectric constant ε, ε∞ , bare coupling αG , and effective reflected by the form factors (141). An important differ-
coupling α∗G for graphene in vacuum and popular substrates. ence is manifest in the small wave-vector limit of the po-
larizability. In contrast to the zero-field case, where the
polarizability saturates at a non-zero density of states,
the system is gapped in the IQHE regime with a result-
Finally, the screened Coulomb interaction potential ing vanishing density of states at the Fermi energy. This
may be approximated as gives rise to a q 2 behavior of the polarizability at small
wave vectors. Furthermore, the static dielectric function,
2πe2 which is shown in Fig. 20 (Roldán et al., 2010; Shizuya,
vscr (q) ≃ . (179)
εε∞ (1 + α∗G qF /q)q 2007), does no longer diverge in this limit, contrary to
the zero-field Thomas-Fermi result (177). Indeed, the
One notices from this expression that processes at wave small-q behavior may be approximated as
vectors q ≪ qF , where the interband excitations play a
3/2
εRP A (q) − 1 ∝ αG NF qlB , (180)
23 The dielectric constant ε is then the average of the dielectric
constant of the substrate material and that of the vacuum. which is the same as for non-relativistic 2D elec-
31
(a) (b) (c)

trons (Aleiner and Glazman, 1995).24 The maximum −dt/da
electron
K’
of the static√dielectric function is obtained at qlB ∼ q
on A
k+q
1/qF√lB ∼ 1/ 2NF + 1, at the value εmax ≃ εRP A (q ∼ TO
phonon
q
g ~ −dt/da Γ K
(q~0)
1/lB 2NF + 1) ∼ αG√NF . It therefore scales as εmax ∝ LO
u(r) phonon k
NF in contrast to a NF scaling in non-relativistic 2D phonon
electron
on B
systems (Aleiner and Glazman, 1995). At large wave
vectors, the static dielectric function saturates at the electron−phonon Brillouin zone
optical phonons
coupling
same value ε∞ as in zero magnetic field.
Figure 21 Electron-phonon coupling in graphene. (a) Optical
phonons in graphene, with a wave vector q in the vicinity of the Γ
point at the center of the first BZ [see part (c)]. The amplitude of
IV. MAGNETO-PHONON RESONANCE IN GRAPHENE the LO phonon is in the direction of propagation (black arrows),
that of the TO phonon perpendicular (red arrows). The optical
In the previous section, we have discussed the role of phonons modify the bond lengths of the honeycomb lattice. (b) As
a consequence of the modified bond lengths, the electronic hopping
electron-electron interactions in the IQHE regime, where is varied, and the electron-phonon coupling is off-diagonal in the
a perturbative (diagrammatic) approach may be applied. sublattice index.
Similarly, one may treat the electron-phonon interaction
in a perturbative manner in this regime. This is the
subject of the present section, before discussing again renormalization of the optical phonons at the Γ point
electron-electron interactions in the strong coupling limit in Sec. IV.B. More specifically, we consider both
of partially filled LLs (Sec. V). non-resonant (Sec. IV.B.1) and resonant coupling (Sec.
As a consequence of the honeycomb-lattice structure IV.B.2), the latter being specific to graphene in a mag-
of graphene, with two inequivalent sublattices, there are netic field when the phonon is in resonance with an al-
four in-plane phonons, two acoustic and two optical ones. lowed inter-LL transition (magneto-phonon resonance)
Each phonon type occurs in a longitudinal (longitudinal (Goerbig et al., 2007).
acoustic, LA, and longitudinal optical, LO) and a trans-
verse (transverse acoustic, TA, and transverse optical,
TO) mode. In addition to these four phonons, one finds A. Electron-Phonon Coupling
two out-of-plane phonons, one acoustic (ZA) and one op-
tical (ZO) [for a review of phonons in graphene, see Refs. The LO and TO phonons in graphene are schematically
(Saito et al., 1998) and (Wirtz and Rubio, 2004)]. Here, represented in Fig. 21(a). As already mentioned above,
we concentrate on the in-plane optical phonons LO and we concentrate on phonons at small wave vectors, in the
TO, which couple to the electronic degrees of freedom. vicinity of the Γ point. The origin of the electron-phonon
More specifically, we discuss these phonons at the Γ point coupling may easily be understood from the variation of
(E2g modes) in the center of the first BZ, which are at- the bond length caused by the phonon, which affects the
tributed to the G-band at ~ωph ≃ 0.2 eV in the Raman electronic hopping amplitude between nn carbon atoms.
spectra [see e.g. Refs. (Ferrari et al., 2006; Graf et al., As we have discussed in Sec. I.D, the effect may be quan-
2007; Gupta et al., 2006; Pisana et al., 2007; Yan et al., tified with the help of Harrison’s law (Harrison, 1981),
2007)]. which yields ∂t/∂a ≃ −4.3 eV/Å [see Eq. (57)]. The
One of the most prominent effects of electron-phonon order of magnitude for the bare electron-phonon energy
coupling in metals and semiconductors is the so-called is then obtained by p multiplying this variation with the
Kohn anomaly (Kohn, 1959), which consists of a sin- typical length scale ~/M ωph , which characterizes the
gularity in the phonon dispersion due to a singular- amplitude of a lattice vibration of frequency ωph within
ity in the electronic density-density response function. the harmonic approximation. The intervening mass M is
The analog of the Kohn anomaly in graphene yields a that of the carbon atom. Indeed, a tight-binding calcu-
logarithmic divergence of the phonon frequency when lation (Ando, 2006b; Ishikawa and Ando, 2006) corrob-
the bare phonon frequency coincides with twice the orates this argument, apart from a numerical prefactor
Fermi energy (Ando, 2006b; Castro Neto and Guinea, 3/2, and yields a bare electron-phonon coupling
2007; Lazzeri and Mauri, 2006). We investigate how this s
renormalization manifests itself in graphene in a strong 3 ∂t ~
g=− ≃ 0.26 eV. (181)
magnetic field (Ando, 2007b; Goerbig et al., 2007). In 2 ∂a M ωph
Sec. IV.A, we consider the specific form of the electron-
phonon coupling and discuss its consequences for the This value agrees well with ab-initio calculations
(Piscanec et al., 2004), although it is though slightly
lower than the value determined experimentally, which
ranges from g ≃ 0.3 eV (Faugeras et al., 2009;
24 Notice, however, that the expression becomes exact only in the Pisana et al., 2007) to g ≃ 0.36 eV (Yan et al., 2007).
large-NF limit and that in non-relativistic 2D electron systems, Furthermore, one notices that, because the electron-
−1/2
the coupling constant rs also depends on NF , rs ∼ NF . phonon coupling is mediated by a bond variation between
32
sites that belong to two different sublattices, the coupling where σ ∧ u(r) = [σ × u(r)]z = σ x uy (r) − σ y ux (r) is
constant is off-diagonal in the sublattice basis. This is the 2D cross product between the Pauli matrices and the
diagrammatically depicted in Fig. 21(b). displacement field.
1. Coupling Hamiltonian 2. Hamiltonian in terms of magneto-exciton operators
The above considerations help us to understand the As a consequence of the off-diagonal character of the
different terms in the Hamiltonian, electron-phonon coupling (185), one notices that the in-
tervening matrix elements are proportional to δn,n±1 , and
H = Hel + Hph + Hcoupl
one thus obtains the selection rules
which serves as the basis for the analysis of the electron-
phonon coupling. The Hamiltonian for 2D electrons in a λn → λ′ (n ± 1), (186)
magnetic field,
in analogy with the magneto-optical selection rules dis-
XZ
Hel = d2 r ψξ† (r)HB
eff,ξ
ψξ (r) cussed in Sec. II.A.1. Furthermore, if we fix the energy
25
ξ √ the dipole√transition√(105) to be ∆n ≡ ∆n,λ=− =
of
2~(vF /lB )( n + 1 + n), there are two possible tran-
ǫλ,n c†λn,m;ξ cλn,m;ξ
X
= (182) sitions, which may be distinguished by the circular po-
λn,m;ξ larization of the phonon they are coupled to. Indeed,
may be written, in second quantization, in terms of the the form of the electron-phonon coupling (185) indicates
one-particle Hamiltonian (75) and the fermionic fields that the -polarized phonon is coupled to the transi-
X tion −(n + 1) → +n, whereas the -polarized phonon
ψξ (r) = ψλn,m;ξ (r)cλn,m;ξ , couples to the −n → +(n + 1) interband transition
λn,m (Goerbig et al., 2007).
It is then convenient to introduce magneto-exciton op-
where ψλn,m;ξ (r) is the wave function in position space
erators, associated with the above-mentioned inter-LL
associated with the spinor (138).
transitions
The lattice vibration is characterized by the relative
displacement u(r) between the two sublattices, which
p
† i 1 + δn,0 X †
may be decomposed in terms of the bosonic operators φ (n, ξ) = c+n,m;ξ c−(n+1),m;ξ ,
Nn
bµ,q and b†µ,q , p
m
s † i 1 + δn,0 X †
~ φ (n, ξ) = c+(n+1),m;ξ c−n,m;ξ , (187)
Nn

bµ,q + b†µ,−q eµ,q e−iq·r ,
X
u(r) = m
µ,q
2Nuc M ωµ (q)
(183) where the index A = , characterizes the angular mo-
where eµ,q denotes the two possible linear polariza- mentum of the excitation and where the normalization
tions √(µ = LO,TO) at the wave vector q and Nuc = factors
A/(3 3a2 /2) is the number of unit cells in the system. q
The phonon Hamiltonian then reads, on the level of the Nn = 2(1 + δn,0 )NB ν̄−(n+1) − ν̄+n
harmonic approximation, q
Nn =

X and 2(1 + δn,0 )NB ν̄−n − ν̄+(n+1)
Hph = ~ωµ (q)b†µ,q bµ,q , (184)
µ,q are used to ensure the bosonic commutation rela-
in terms of the phonon dispersion ωµ (q). Notice that, at tions of the exciton operators, [φA (n, ξ), φ†A′ (n′ , ξ ′ )] =
the Γ point, the frequencies of the LO and TO phonons δA,A′ δξ,ξ′ δn,n′ , including the two-fold spin degeneracy.
coincide, and one has ωph ≡ ωµ (q = 0). It is then con- These commutation relations are obtained within the
venient to describe the phonon modes in terms of mean-field approximation with hc†λn,m;ξ cλ′ n′ ,m′ ;ξ′ i =
√ circu- δξ,ξ′ δλ,λ′ δn,n′ δm,m′ (δλ,− + δλ,+ ν̄λn ), where 0 ≤ ν̄λn ≤ 1
larly polarized modes, u (r) = [ux (r) + iuy (r)]/ 2 and
u (r) = u∗ (r). is the partial filling factor of the n-th LL, normalized to
Finally, taking into account the above considerations 1.
on the electron-phonon coupling, the coupling Hamilto- One notices that the magneto-exciton operators are,
nian reads (Ando, 2006b; Castro Neto and Guinea, 2007; apart from a normalization constant, nothing other
Ishikawa and Ando, 2006)
r
2M ωph X
Z
Hcoupl = g d2 r ψξ† (r) [σ ∧ u(r)] ψξ (r), 25 We consider mainly interband transitions here, which may have a
~
ξ chance to be in resonance with the phonon of energy ~ωph ∼ 0.2
(185) eV.
33
than the reduced density operators (140), φ† (n, ξ) ∝ is equivalent to a particle-hole propagation associated
†
ρ̄ξ,ξ ξ,ξ
+n,−(n+1) (q = 0) and φ (n, ξ) ∝ ρ̄+(n+1),−n (q =
with a polarization bubble [see Fig. 15(a)], but the ex-
0), respectively, at zero wave vector. Notice fur- pression (193) also takes into account the square of the
thermore that, because of the relative sign ξ between effective coupling constant which is due to the double
the kinetic part (182) and the electron-phonon cou- occurence of the electron-phonon coupling – first when
pling Hamiltonian (185), the optical phonons couple the phonon is converted into a magneto-exciton and the
to the valley–anti-symmetric magneto-exciton combina- second time when the magneto-exciton is destroyed by
√ creating a phonon. The last term in Eq. (193) takes
tion φA,as (n) = [φA (n, ξ = +) − φA (n, ξ = −)]/ 2.
This needs to be contrasted to the magneto-optical into account the (only) possible intra-band magneto-
coupling (Abergel and Fal’ko, 2007; Iyengar et al., 2007; exciton from
√ the last filled
√ LL NF √to NF + 1 with energy
∆˜ NF = 2(~vF /lB )( NF + 1 − NF ), which we have
Sadowski et al., 2006), where the photon couples to the
valley-symmetric omitted in the Hamiltonian (188) because it is irrelevant
√ mode φA,s (n) = [φA (n, ξ = +) + for resonant coupling. The parameter Nc is√the same
φA (n, ξ = −)]/ 2.
high-energy cutoff, defined by ǫNc = ~(vF /lB ) 2Nc ∼ t,
The magneto-exciton operators (187) allow one to
as in Sec. III.B of the preceding chapter.
rewrite the electron-phonon Hamiltonian at the Γ point
The renormalized phonon frequencies ω̃A may be ob-
(q = 0) in a bosonic form as (Goerbig et al., 2007)
tained from the Dyson equation (191), by searching the
poles of the dressed phonon propagator
∆n φ†A,τ (n)φA,τ (n) + ~ωph b†A bA
X X X
H =
τ =s,as A,n
h
A
i DA (ω)−1 = 0 = DA (ω̃A )−1 − χA (ω̃A ), (194)
gA (n) b†A φA;as (n) + bA φ†A;as (n) , (188)
X
+
and one finds (Ando, 2007b; Goerbig et al., 2007)
A,n
" N #
in terms of the effective coupling constants 4ωph X c
∆n gA 2
(n) ˜ NF g 2 (NF )
∆
2 2 A
ω̃A −ωph = 2 − ∆2 + 2 2 .
q ~ ~2 ω̃A n ~ ω̃A − ∆ ˜2
p n=NF +1 N F
g (n) = g (1 + δn,0 )γ ν̄−(n+1) − ν̄+n , (195)
q p
and g (n) = g (1 + δn,0 )γ ν̄−n − ν̄+(n+1) , (189)
√ 1. Non-resonant coupling and Kohn anomaly
with the constant γ ≡ 3 3a2 /2πlB 2
. One therefore re-
marks that, although the bare coupling constant g is Before discussing resonant coupling, i.e. when the
rather large [see Eq. (181)], the effective coupling is re- phonon frequency is in resonance with a possible inter-
duced by a factor of a/lB , LL excitation, in a strong magnetic field, we comment on
a p the relation between Eq. (195) and the (non-resonant)
gA (n) ∼ g ∼ 1 . . . 2 meV B[T]. (190) renormalization of the phonon frequency in zero mag-
lB
netic field. The zero-field limit may indeedPbe obtained
from Eq.R (195) if one replaces the sum n by an in-
B. Phonon Renormalization and Raman Spectroscopy tegral dn, i.e. if the spacing between the LLs van-
ishes, ∆˜ NF → 0. Linearizing Eq. (195), if one replaces
The Hamiltonian (188) shows that a phonon may be ω̃A → ωph in √ the denominators, and using the approxi-
√ √
destroyed by exciting a magneto-exciton, and the associ- mation n + n + 1 ≈ 2 n yields
ated Dyson equation for the dressed phonon propagator p √
D(ω) reads vF ωph ωph + 2 2NF vF /lB
ω̃ ≃ ω̃0 +λ̃ 2NF − ln √ ,
lB 4 ωph − 2 2NF vF /lB
DA (ω) = D0 (ω) + D0 (ω)χA (ω)DA (ω), (191) (196)
√ 2 −3
in terms of the bare bosonic phonon propagator where λ̃ = (2/ 3π)(g/t) ≃ 3.3 × 10 is the dimen-
sionless electron-phonon coupling constant introduced in
1 2ω Refs. (Ando, 2006b, 2007b), and
D0 (ω) = 2 (192)
~ ω 2 − ωph
Nc 2
2 ∆n gA (n)
Z
and ω̃0 ≃ ωph + dn 2 (197)
~ 0 ~ ωph − ∆2n
2
Nc 2 ˜ NF g 2 (n)
X 2∆n gA (n) 2∆ A is the physical phonon frequency at zero doping. In-
χA (ω) = + . (193)
~2 ω 2 − ∆2n ~ 2ω2 − ∆ ˜2 deed, the frequency ωph is not relevant in a physi-
n=NF +1 NF
cal measurement in graphene even if it occurs in the
The form of the last expression is transparent; the Hamiltonian, but one measures ω̃0 at zero doping and
magneto-exciton is a boson, and its propagator is there- B = 0. Equation (196) coincides precisely with the
fore of the same form as that of the bare phonon. It zero-field result (Ando, 2006b; Castro Neto and Guinea,
34
(a) 22 boson modes read (Ando, 2007b; Goerbig et al., 2007)

(a)
1.75
Energy in units of ω ph
n=2
1.5 s 2
n=1
1.25 n=0 ± 1∆n 1 ∆n 2 (n),
11
ω̃A (n) = + ω̃0 ∓ − ω̃0 + gA
2 ~ 4 ~
0.75
0.5
(198)
0.25 and the resulting phonon–magneto-exciton anticrossing
B0 is depicted in Fig. 22(a).
0 5 1010 1520 20 25
30 30 4035
Magnetic Field [T] The above-mentioned anticrossing of the coupled
phonon–magneto-exciton modes has been observed in
(b) recent Raman experiments on epitaxial graphene. Re-
member that Raman spectroscopy is sensitive to the
phonon component of the mixed modes (Faugeras et al.,
2009). The results are shown in Fig. 22(b) and corrob-
orate the theoretically expected behavior (Ando, 2007b;
Goerbig et al., 2007). Indeed, one may obtain the oscil-
lating behavior from a numerical solution of Eq. (195) if
one expresses the equation in terms of ω̃0 instead of ωph
and if one takes into account a finite broadening of the
levels. If the phonon is out of resonance with an inter-LL
transition, its frequency is essentially field-independent
and coincides with the energy of the E2g line at 1586
cm−1 ≃ 0.2 eV. When it approaches the resonance (by
increasing the magnetic field), its energy is shifted up-
wards as a consequence of the anticrossing but rapidly
dies out in intensity once the magneto-exciton component
+
becomes dominant in the ω̃A mode. Upon further in-
−
crease of the magnetic field, the ω̃A mode becomes more
Figure 22 (a) From Ref. (Goerbig et al., 2007). Anticrossing of
coupled phonon-magneto-exciton modes as a function of the mag- phonon-like and therefore visible in the Raman spectra.
netic field. (b) From Ref. (Faugeras et al., 2009). Colour plot of The fine structure of the high-field resonant electron-
the Raman spectra as a function of the magnetic field. The contin- phonon coupling may furthermore be investigated by
uous white lines indicate the magnetic field for which the phonon is sweeping the chemical potential when the magnetic field
in resonance with an inter-LL excitation of energy ∆n . Top: data
of the full B-field range. Bottom: zoom on the range from 0 to 10
is held fixed at resonance. The effect is most pronounced
T. for the resonance ~ωph ≃ ∆n=0 , which is expected at
B ≃ 30 T [see Fig. 22(a)]. In this case, the mode
consists of an equal-weight superposition of the √ phonon
and the magneto-exciton (cos θ = sin θ = 1/ 2), and
2007; Lazzeri and Mauri, 2006) if one identifies the chem- the E2g band would appear as two lines, at the energies
ical potential with the energy of the last filled LL, µ = ~ω̃ ± = ~ω̃0 ± gA , for the case of undoped graphene.26
√
2NF ~vF /lB (Goerbig et al., 2007). With the above estimation (181) for the bare electron-
phonon coupling constant, one obtains for the line split-
ting 2gA ∼ 16 meV (∼ 130 cm−1 ), which largely exceeds
the G-band width observed in Refs. (Ferrari et al., 2006;
Graf et al., 2007; Gupta et al., 2006; Pisana et al., 2007;
Yan et al., 2007).
2. Resonant coupling It is apparent from the expressions (190) for the ef-
fective coupling constants g and g that the splitting
may be controled by the LL filling factor. Exactly at
Apart from the non-resonant coupling discussed in the zero doping, both coupling constants coincide, g = g ,
preceding section, the high-field electron-phonon cou- but upon electron doping the transition −1 → 0 associ-
pling reveals a linear effect when the phonon is in res- ated with the -polarization becomes weaker due to the
onance with a particular magneto-exciton, ~ωph ≃ ∆n . reduced number of final states in n = 0, whereas the
In this case, the sum on the right-hand side in Eq. 0 → +1 transition (with polarization ) is strengthened.
(195) is dominated by a single term and may be ap-
2
proximated by 2(ωph /~)gA (n)/(~ω̃A − ∆n ). This re-
sults in a fine structure of mixed phonon–magneto-
exciton modes, φA,as (n) cos θ + bA sin θ with frequency 26 Notice, however, that only an oscillation of the phonon mode,
+ −
ω̃A and φA,as (n) sin θ−bA cos θ with frequency ω̃A [where and not a splitting, has been observed in the experiment by
cot 2θ = (∆n −~ω̃0 )/2gA ]. The frequencies of these mixed Faugeras et al. (Faugeras et al., 2009).
35
As a consequence, the associated coupling constants are in a partially filled relativistic LL (Sec. V.A). This
increased and decreased, respectively, until the coupling model yields a qualitative understanding of the above-
constant g vanishes at ν = 2. mentioned correlated electronic phases in the context of
The above-mentioned filling-factor dependence has a graphene, as compared to non-relativistic electrons. In
direct impact on the Raman lines (Goerbig et al., 2007). Sec. V.B, we apply this model to the quantum Hall ferro-
Whereas at ν = 0, one expects two lines separated by magnetism with an internal SU(4) symmetry that is the
the energy 2g = 2g , the degeneracy in the circular po- relevant symmetry in graphene LLs and discuss its rela-
larization is lifted between 0 < ν < 2.27 One therefore tion with the experimentally observed degeneracy lifting
expects to observe four lines instead of two, where the in- of the zero-energy LL n = 0 (Zhang et al., 2006). We ter-
ner ones are associated with the polarization , whereas minate this section with a review of the specific FQHE in
the outer ones with increased splitting correspond to the graphene (Sec. V.C), which has recently been observed in
opposite polarization . The separation between the in- the two-terminal (Bolotin et al., 2009; Du et al., 2009) as
ner lines vanishes then at ν = 2, where the splitting of the well as in the four-terminal geometry (Dean et al., 2011;
outer lines is maximal and where one expects to observe Ghahari et al., 2011).
three lines.
A. Electrons in a Single Relativistic Landau Level

V. ELECTRONIC CORRELATIONS IN PARTIALLY
FILLED LANDAU LEVELS Quite generally, the origin of strongly-correlated elec-
tron phases is a quenched kinetic energy, where the par-
This last section is devoted to the physics of in- tially filled LL is separated by the cyclotron gap from
teracting electrons in the strong-correlation limit of a the neighboring ones such that inter-LL excitations con-
partially filled LL. The motivation stems from non- stitute high-energy degrees of freedom. The Coulomb in-
relativistic quantum Hall systems in GaAs heterostruc- teraction, which may though be small with respect to the
tures, where these correlations lead to the formation cyclotron gap, remains then as the only relevant energy
of incompressible quantum-liquid phases, which display scale which dominates the low-energy degrees of free-
the fractional quantum Hall effect (FQHE) (Tsui et al., dom if we can neglect disorder effects. This leads to the
1982), as well as of exotic electron-solid phases, seemingly counter-intuitive finding of strongly-correlated
such as the high-field Wigner crystal (Andrei et al., phases in weakly-correlated matter.
1988; Williams et al., 1991) or the theoretically pre- In order to quantify the degree of separation between
dicted bubble and stripe phases (Fogler et al., 1996; the energy scales, one may use a similar argument as
Koulakov et al., 1996; Moessner and Chalker, 1996). the one that led us to the definition of the dimension-
The latter are likely to be at the origin of highly less interaction parameter (131), introduced at the be-
anisotropic transport properties at half-filled higher ginning of Sec. III. Indeed, one needs to compare the
2
LLs (Du et al., 1999; Lilly et al., 1999), particular elec- Coulomb interaction energy E√ int = e /εRC at the char-
tron transport under microwave irradiation (Lewis et al., acteristic length scale RC = lB 2n + 1 to the LL spacing
2005, 2004, 2002), and an intriguing reentrance of the ~ωC = ~eB/mb , where we concentrate on non-relativistic
IQHE in n = 1 and n = 2 (Cooper et al., 1999; electrons first,
Eisenstein et al., 2002).
It is therefore natural to ask whether such strongly- e2
rsB = , with vF (n, B) ≡ RC ωC .
correlated phases exist also in graphene and if so what ~εvF (n, B)
the differences are with respect to non-relativistic 2D √ (199)
electrons. Moreover, the fact that the electrons reside If one identifies the Fermi wave vector kF ≃ 2n/lB , one
at the surface opens up the possibility to probe these obtains the same expression as for the zero-field coupling
phases by spectroscopic means, such as scanning tunnel- constant (132),
ing spectroscopy, which has already been applied success-
fully in the analysis of the electron density distribution mb e2 −1 1
rsB = rs = k ∼ ∗ . (200)
of exfoliated (Martin et al., 2008) and epitaxial graphene ~2 ε F a0 kF
(Mallet et al., 2007), as well as graphene on graphite sub-
This means that the degree of LL mixing is still governed
strates (Li et al., 2009b).
by rs , and the inter-LL excitations are well separated
After a short discussion of the Coulomb interaction in
from the low-energy intra-LL degrees of freedom unless
graphene as compared to non-relativistic 2D electrons,
rs becomes very large. Notice, however, that rs ∼ 1 in
we introduce the basic model of interacting electrons
most 2D electron systems.
In the case of partially filled relativistic LLs in
graphene, one is tempted to apply the same argument
27 – if the Coulomb interaction e2 /εRC is sufficiently small
We present the argument for a Fermi energy in the CB, i.e. ν > 0,
but the situation is generic, and the argument also applies in the as compared to the LL spacing ∆ ˜ n , the relevant degrees
VB if one interchanges the polarizations. of freedom are those which couple quantum states in the
36
(a) (b) describe the density fluctuations inside the LL λn that

interest us here. Notice that we have dropped the index
LL λn in the definition of the projected density operators;
separation
X they satisfy the quantum-mechanical commutation rela-
X x x x x
tions (Girvin et al., 1986)
q′ ∧ q lB
2

Figure 23 (a) Completely filled topmost LL. Due to the Pauli ′
principle, the only possible excitations are inter-LL transitions. (b) [ρ̄(q), ρ̄(q )] = 2i sin ρ̄(q + q′ ), (204)
2
Partially filled LL. For sufficiently small values of rs (or αG ), the
inter-LL excitations constitute high-energy degrees of freedom that
may be omitted at low energies, where the relevant degrees couple where q′ ∧ q = (q′ × q)z = qx′ qy − qy′ qx is the 2D vec-
states within the same LL. tor product between q′ and q, and these commutation
relations are independent of the LL index λn. The in-
formation about the LL is indeed waved to the effective
same LL, whereas inter-LL excitations may be considered interaction potential
as frozen out (see Fig. 23). Although this seems a rea-
sonable assumption for the lowest LLs, one is confronted 2πe2 2
with the apparent problem that the LL spacing rapidly vn (q) = [Fn (q)] , (205)
εq
decreases once the Fermi level resides in higher LLs,
√ ~vF √ in terms of the LL form factors [see Eq. (141) and their
√ ~vF
˜n =
∆ 2 ( n + 1 − n) ≃ √ . (201) explicit form (A2), discussed in Appendix A]
lB lB 2n
22
1 q lB
Notice,
√ however, that this decrease is balanced by the Fn (q) = (1 − δn,0 )Ln−1
2 2
1/ 2n scaling of the characteristic Coulomb interaction, 2 2
such that even in higher LLs the separation between low- q lB 2 2
+(1 + δn,0 )Ln e−q lB /4 , (206)
and high-energy degrees of freedom is governed by the 2
dimensionless coupling constant
√ independent of the band index λ (Goerbig et al., 2006;
B e2 /εlB 2n e2 Nomura and MacDonald, 2006). The Hamiltonian re-
αG = √ = = αG , (202) sulting from Eq. (151)
~vF /lB 2n ~εvF
which coincides with the scale-invariant zero-field cou- 1X

Hn = vn (q)ρ̄(−q)ρ̄(q) (207)
pling constant (133). From the interaction point of view, 2 q
the restriction of the electron dynamics to a single par-
tially filled LL in the large-n limit is therefore as justified therefore defines, together with the commutation rela-
as for the lowest relativistic LLs. Naturally, this state- tion (204) the model of strongly-correlated electrons re-
ment only holds true in the absence of disorder that in- stricted to a single relativistic LL. The model respects the
duces stronger LL mixing for n ≫ 1 than in n = 0 or SU(4) spin-valley symmetry, and naturally, there is no
±1. kinetic energy scale because all processes involve states
within the same LL.
1. SU(4)-symmetric model
a. Algebraic properties. The SU(4) spin-valley symmetry
Formally, the above-mentioned separation into high- is formally described with the help of the spin and valley-
and low-energy degrees of freedom may be realized with pseudospin operators
the help of the reduced density operators (140). For the
moment, we only consider the case where ξ = ξ ′ , i.e. S̄ µ (q) = (S µ ⊗ 1) ⊗ ρ̄(q) and I¯µ (q) = (1 ⊗ I µ ) ⊗ ρ̄(q),
we concentrate on the valley-symmetric model, in which (208)
case the reduced (intra-valley) density operators (153) respectively, that are tensor products between the pro-
fall into two distinct classes: for n 6= n′ or λ 6= λ′ , the jected density operators (203) and the operators S µ
operators ρ̄λn,λ′ n′ (q) describe density fluctuations corre- and I µ , which are (up to a factor 1/2) Pauli matri-
sponding to inter-LL transitions of an energy equal to or ces and that describe the spin and valley degrees of
larger than the LL separation ∆ ˜ n , whereas the projected
freedom, respectively. The operators (S µ ⊗ 1) and
density operators (1 ⊗ I µ ) are the generators of the SU(2)×SU(2) sub-
group of SU(4). However, once combined in a ten-
ρ̄(q) ≡ ρ̄λn,λn (q) (203) sor product with the projected density operators ρ̄(q),
†
X X X
−iq·R ′ the SU(2)×SU(2)-extended magnetic translation group
= m e m cλn,m;ξ,σ cλn,m′ ;ξ,σ
ξ=± σ=↑,↓ m,m′ is no longer closed due to the non-commutativity of
37
the Fourier components of the projected density opera- 2. Symmetry-breaking long-range terms
tors. The commutators [S̄ µ (q), I¯ν (q′ )] yield the remain-
ing generators of the SU(4)-extended magnetic trans- When decomposing the Coulomb interaction in the
lation group (Douçot et al., 2008; Ezawa and Hasebe, two-spinor basis (Sec. III.A), we have seen that the
2002; Ezawa et al., 2003). SU(4)-symmetric model yields the leading energy scale,
Physically, the operators introduced in Eq. (208) play whereas the only relevant symmetry-breaking term is as-
the role of projected spin and valley-pseudospin densi- sociated with backscattering processes at an energy scale
ties, where the LL projection is induced by the projected roughly a/lB times smaller than the leading one. When
charge-density operator ρ̄(q). Their non-commutativity restricted to a single relativistic LL λn, these backscat-
with the projected charge density, [S̄ µ (q), ρ̄(q′ )] 6= 0 and tering terms yield a contribution
[I¯µ (q), ρ̄(q′ )] 6= 0, which are due to the commutation re-
1 X X sb
lation (204), is at the origin of the (pseudo-)spin-charge Hnsb = vn (q)ρ̄ξ,−ξ (−q)ρ̄−ξ,ξ (q), (212)
entanglement in quantum Hall systems: as we discuss 2 q
ξ=±
in more detail in Sec. V.B.2, this entanglement yields
(pseudo-)spin-texture states that carry an electric in ad- in terms of the effective backscattering potential
dition to their topological charge.
2πe2 +,− 2
vnsb (q) = Fλn,λn (q)
εq
b. Validity of the model. With the help of the Hamilto- 2πe2 (1 − δ0,n )
= (qy − Ky )2 lB
2
(213)
nian (207), we may render more transparent the model εq 2n
assumption of electrons restricted to a single relativistic
|q − K|2 lB2
2
2 2
LL. We need to show that the energy scale that governs × L1n−1 e−|q−K| lB /2 ,
the model (207) and its resulting phases is indeed given 2
by e2 /εRC and not e2 /εlB . As an upper bound for the where we have made use of Eq. (142) and of the explicit
energy scale, one may use the energy of a completely filled expressions for the intervening matrix elements (A4).
LL described by hc†λn,m;ξ,σ cλn,m′ ;ξ,σ i = δm,m′ , the mean- The effect of this symmetry-breaking term will be dis-
field energy hHn i/N of which is simply the exchange en- cussed in more detail in Sec. V.B in the context of the
ergy,28 SU(4) quantum Hall ferromagnetism. The term (212) is
∞
only relevant in relativistic LLs n 6= 0 as a consequence
1X e2
Z
n
EX =− vn (q) = −
2
dq [Fn (q)] . (209) of the factor (1 − δn,0 ) in the expression (213) for the
2 q 2ε 0 backscattering potential (Goerbig et al., 2006). This is a
consequence of the chiral symmetry of the zero-energy LL
In order to estimate the integral in the large-n limit, one (Arikawa et al., 2008) where the sublattice index is con-
may use the scaling form (Abramowitz and Stegun, 1970; funded with the valley pseudospin, as may be seen from
Gradshteyn and Ryzhik, 2000) of the Laguerre polyno- the expression (88) for the associated wave functions.
mials Notice, however, that there may occur other symmetry-
breaking terms in n = 0 as a consequence of short-range
√
22
q lB 2 2 interactions on the lattice scale (Alicea and Fisher, 2006;
Ln e−q lB /4 ≃ J0 (qlB 2n + 1), (210) Doretto and Morais Smith, 2007; Herbut, 2007b).
2
in terms of the Bessel function J0 (x), such that one

obtains by a simple √ change Rof the integration variable 3. Qualitative expectations for correlated electron phases
R∞ ∞ √
0
dq[F n (q)] 2
≃ (l B 2n)−1 0 dx[J0 (x)]2 = c/lB 2n,
where c is a numerical factor of order one. The exchange The model of interacting electrons in a single rela-
energy of a completely filled LL n therefore scales with tivistic LL has the same structure as the one for non-
n ≫ 1 as relativistic LLs – in both cases, one has an interaction
Hamiltonian that is quadratic in the projected density
n e2 e2 operators (203) which satisfy the commutation relations
EX ≃ −c √ ≃ −c , (211) (204). This is a noticeable result in the sense that,
εlB 2n εRC
whereas non-relativistic 2D electron systems are gov-
in agreement with the model assumption of a separation erned by Galileian invariance, the electrons in graphene
between high- and low-energy degrees of freedom and the are embedded in a Lorentz-invariant “space-time”. How-
definition (202) of the coupling constant αB ever, once restricted to a single LL, the electrons forget
G.
about their original spatial symmetry properties and are
governed by the magnetic translation algebra, which is at
the origin of the commutation relations (204). As a con-
28 The direct energy is compensated by the positively charged back- sequence and in contrast to the IQHE, the differences
ground (“jellium model”) (Mahan, 1993) between strongly-correlated electrons in graphene and
38
(a)
√
J(qlB 2n) + O(1/n). This result agrees indeed to lead-
effective interaction potential

12
n=0 (relativistic and non−relativistic)
ing order in 1/n with the scaling expression of the form
10
10 factors (154) for usual non-relativistic 2D electrons.
n=1 (relativistic)
8 The strongest difference in the interaction potentials is
n=1 (non−relativistic)
6
n=5 (relativistic)
thus found for n = 1, which in graphene is quite reminis-
5
4 cent to that in n = 0, apart from a reduced repulsion at
2
very short distances, whereas for non-relativistic 2D elec-
n=5 (non−relativistic)
trons it has an additional structure [see Fig. 24(a)]. The
2 44 66 88 10
10 behavior of the effective interaction potential may also
r/l B
(b)
be analyzed with the help of Haldane’s pseudopotentials
n=0 (Haldane, 1983)
pseudopotentials V m
n
0.8 0.8
n=1 (rel.) 1 X 2 2
0.6
n=1 (non−rel.) Vℓn = vn (q)Lℓ (q 2 lB
2
)e−q lB /2 , (215)
0.6 2π q
0.4 0.4 which represent the interaction between pairs of elec-
trons in a magnetic field, in a relative angular momentum
1
0 2
1 3
2
4
3 5
4 6
5 state with quantum number ℓ. This√quantum number is
m related to the average distance lB 2ℓ + 1 between the
Figure 24 (a) From Ref. (Goerbig et al., 2006); comparaison two particles constituting the pair and is a good quan-
between the relativistic (black curves) and non-relativistic (grey tum number for any two-particle interaction potential
curves) potentials for the LLs n = 0, 1, and 5 in real space. The v(ri − rj ). The pseudopotentials for graphene are shown
dashed line shows the potential in n = 0, which is the same in in Fig. 24(b) for n = 0 and 1.
both the relativistic and the non-relativistic case. (b) Pseudopo-
Haldane’s pseudopotentials are an extremely helpful
tentials for n = 0 (black circles), n = 1 relativistic (blue) and
n = 1 non-relativistic (green). The lines are a guide to the eye. quantity in the understanding of the possible FQHE
The open circles represent pseudopotentials with even relative pair states which one may expect in 2D electron systems. One
angular momentum that are irrelevant in the case of completely notices first that as a consequence of the anti-symmetry
spin-valley polarized electronic states. The energies are given in of a two-particle wave function under fermion exchange,
units of e2 /ǫlB .
the relative angular-momentum quantum number ℓ must
be an odd integer, i.e. only the pseudopotentials with
odd values of ℓ play a physical role in the description
non-relativistic 2D electrons do not stem from their re- of two interacting electrons of the same type (spin or
spective space-time properties, as one would expect from valley). Even-ℓ pseudopotentials become relevant if the
a mean-field Chern-Simons approach (Khveshchenko, SU(4) spin-valley pseudospin is not completely polarized,
2007; Peres et al., 2006). in the treatment of two electrons with different internal
The differences between graphene and non-relativistic quantum number σ or ξ. One then notices that the n = 1
2D electrons are rather to be seeked in the larger inter- pseudopotentials, apart from a the difference in Vℓ=0 , are
nal symmetry – instead of an SU(2) spin symmetry, one much more reminiscent of those in n = 0 than of those
has an SU(4) spin-valley symmetry if one neglects the for non-relativistic 2D electrons in the same LL n = 1
small symmetry-breaking term (212) in the interactions. [see Fig. 24(b)]. If one considers polarized electrons, one
Another difference arises from the different effective in- therefore expects essentially the same strongly-correlated
teraction potential (205), instead of electronic phases in graphene for n = 1 as for n = 0
2 2 2 (Goerbig et al., 2006), as also corroborated by numerical
2πe2 q lB 2 2 studies for FQHE states (Apalkov and Chakraborty,
vnnon−rel (q) = Ln e−q lB /2 (214)
εq 2 2006; Goerbig and Regnault, 2007; Papić et al.,
2009; Töke and Jain, 2007; Töke et al., 2006)
for the usual 2D electron gas. As one may see from the and electron-solid phases (Poplavskyy et al., 2009;
graphene form factors (206), the effective interaction po- Zhang and Joglekar, 2007, 2008). Because the pseu-
tential in graphene for n 6= 0 is the average of the non- dopotentials (215) are systematically larger in n = 1
relativistic ones in the adjacent LLs n and n − 1, whereas than in n = 0 (apart from the short-range component for
for n = 0 there is no difference between the relativistic ℓ = 0), the gaps of the FQHE states in n = 1 are larger
and the non-relativistic case (see Fig. 24), as a conse- than the corresponding ones in n = 0, as one may also see
quence of the above-mentioned chiral properties. from numerical calculations (Apalkov and Chakraborty,
One notices, furthermore, that the difference between 2006; Töke et al., 2006).
the relativistic and non-relativistic effective interaction As much as we have emphasized the similarity between
potentials become less prominent in the large-n limit [see the n = 0 and n = 1 LLs in graphene, we need to stress
n = 5 in Fig. 24(a)]. This may be understood from the the difference between the n = 1 LL in graphene as
approximate expression√(210) of the form√ factors, which compared to n = 1 in non-relativistic 2D electron sys-
yields Fn (q) ≃ [J0 (qlB 2n + 1) + J0 (qlB 2n − 1)]/2 ≃ tems. Remember that the physical phase diagram in the
39
non-relativistic n = 1 LL is extremely rich; an intrigu- is more involved because there is no physical field that
ing even-denominator FQHE has been observed at ν = couples directly to the valley pseudospin, as suggested by
5/2 (Willett et al., 1987) and probably possesses non- the otherwise intuitive form (217). There have however
Abelian quasiparticle excitations (Greiter et al., 1991; been proposals that such an effect may be achieved with
Moore and Read, 1991). Furthermore, a particular com- the help of strain-induced disordered gauge fields that
petition between FQHE states and electron-solid phases, mimic large-scale ripples (Meyer et al., 2007) and that
which is characteristic of the non-relativistic n = 1 LL yield an easy-plane anisotropy in n = 0 (Abanin et al.,
(Goerbig et al., 2003, 2004), is at the origin of the reen- 2007a), similarly to the backscattering term (213) in
trance phenomena observed in transport measurements higher LLs. Quite generally, a valley-degeneracy lifting
(Eisenstein et al., 2002; Lewis et al., 2005). These phe- may be achieved indirectly in the zero-energy LL n = 0
nomena are absent in the n = 0 LL, and the similarity via fields that couple to the sublattice index. This is due
A B
between the n = 0 and the relativistic n = 1 LL thus to the fact that the components ψK,σ and ψK ′ ,σ vanish
leads to the expectation that FQHE states corresponding as a consequence of the chiral properties, which identify
to the 5/2 state in non-relativistic quantum Hall systems the sublattice and the valley pseudospins in n = 0, as we
and the above-mentioned reentrance phenomena are ab- have discussed in Sec. II.A.2.
sent in the n = 1 LL in graphene. This expectation In order to illustrate this indirect lifting of the valley
has recently been corroborated in exact-diagonalization degeneracy, we consider the term (Haldane, 1988)
studies on the non-Abelian 5/2 state (Wojs et al., 2010).
MH = M Ψ† (1valley ⊗ τAB
z
⊗ 1spin ) Ψ, (218)
which breaks the lattice inversion symmetry and opens a

4. External spin-valley symmetry breaking terms mass gap at the Dirac point in the absence of a magnetic
field. In the presence of a B-field, the LL spectrum (85)
Before we consider the different phases due to electron- is modified by the term (218) and reads
electron interactions, we start with an analysis of the
different external effects,29 which are capable of lifting
s
~2 v 2
the four-fold spin-valley degeneracy. ǫλ,n;ξ = λ M 2 + 2 2 F n, (219)
The probably most familiar external symmetry- lB
breaking term is the Zeeman effect, which lifts the two-
for n 6= 0, independent of the valley index ξ, whereas the
fold spin degeneracy while maintaining the SU(2) symme-
fate of the n = 0 LL depends explicitly on ξ,
try associated with the valley pseudospin. The size of the
Zeeman splitting is given by the energy ∆Z ∼ 1.2B[T] K, ǫn=0;ξ = ξM, (220)
for a g-factor that has been experimentally determined as
g ∼ 2 (Zhang et al., 2006). If we adopt a compact eight- such that the valley degeneracy is effectively lifted. No-
spinor notation to take into account the four different tice, however, that due to the vanishing components ψK,σ A
spin-valley components, in addition to the two sublattice and ψK B
′ ,σ , the mass term (218) is now indistinguishable
components, the Zeeman term has the form
(in n = 0) from the above-mentioned valley Zeeman term
(217),
∆spin ∼ Ψ† 1valley ⊗ 1AB ⊗ τspin
z

Z Ψ (216)
∼ A†
ψK,↑ A
ψK,↑ − A†
ψK,↓ A
ψK,↓ + A†
ψK A
′ ,↑ ψK ′ ,↑ − A†
ψK A
′ ,↓ ψK ′ ,↓
Mn=0
H ∼ Ψ† (1valley ⊗ τAB
z
⊗ 1spin ) Ψ
∼ Ψ τvalley ⊗ 1AB ⊗ 1spin Ψ,
† z

+ (A ↔ B), (221)
where the tensor product consists of the valley pseu- whereas this is not the case for the LLs n 6= 0, where
dospin (represented by the Pauli matrices τvalley µ
and the valley degeneracy is only lifted by an explicit val-
ley Zeeman effect. A mass term of the form (218)
1valley ), the sublattice pseudospin (τAB and 1AB ), and
µ
typically arises in the presence of a frozen out-of-plane
µ
the true spin (τspin and 1spin ). For a better understand-
phonon that yields a crumbling of the graphene sheet
ing, we have given the explicit expression in terms of
(Fuchs and Lederer, 2007).
spinor components in the second line of Eq. (216).
More recent studies have concentrated on a sponta-
A possible valley-degeneracy lifting, that one could de- neous deformation of the graphene sheet due to frozen
scribe with the help of a “valley Zeeman effect” similarly in-plane phonons that yield a Kekulé-type distortion
to Eq. (216), (Hou et al., 2010; Nomura et al., 2009). This distor-
tion, which is associated with a characteristic wave vec-
∆valley ∼ Ψ† τvalley
z
⊗ 1AB ⊗ 1spin Ψ,

Z (217) tor 2K and which therefore couples the two valleys, di-
rectly breaks the valley degeneracy, via a term MK =
Mx + My , with
29 By external effects we mean those that are not caused by the ∆x,y † x,y
mutual Coulomb repulsion between the electrons. Mx,y = Ψ τvalley ⊗ 1AB ⊗ 1spin Ψ. (222)
2
40
Such a term yields the same energy spectrum (219) and ishingly small in graphene, whereas an extrinsic Rashba-
(220) as the mass term (218) if one replaces M
q by ∆kek /2, type spin-orbit coupling in graphene can be on the order
with the characteristic energy scale ∆kek = ∆2x + ∆2y ≃ of 1 K (Min et al., 2006).
2B[T] K, (Ajiki and Ando, 1995; Hou et al., 2010). No-
tice that this energy scale is slightly larger than, but
5. Hierarchy of relevant energy scales
roughly on the same order as, the Zeeman energy scale.
Finally, we mention another class of terms that break
These energy scales associated with external fields need
the spin-valley degeneracy and that have received recent
to be compared to the p characteristic (bare) interaction
interest in the framework of research on topological in- 2
sulators [for recent reviews see (Hasan and Kane, 2010; energy e /ǫlB ≃ 625( B[T]/ε) K, which is, for experi-
Qi and Zhang, 2011)]. In an original work, Haldane ar- mentally accessible magnetic fields, much larger than ∆Z
gued that a time-reversal-symmetry breaking term with or ∆kek . As we have discussed in Sec. III.B.4, inter-band
an inhomogeneous flux distribution inside each hexagon LL excitations screen the bare Coulomb interaction and
opens a gap in a honeycomb lattice with zero magnetic yield a contribution to the dielectric constant. In the ab-
field (Haldane, 1988). Most saliently, he showed that sence of a quantizing magnetic field, we have seen that
one may thus achieve a quantum Hall effect without an this dielectric constant is given by [see Eq. (176)]
external magnetic field, a system that is now often re-
π π e2
ferred to as the “quantum anomalous Hall insulator” ε∞ = 1 + αG = 1 + , (224)
(Hasan and Kane, 2010; Qi and Zhang, 2011). A simi- 2 2 ~εvF
lar situation arises when spin-orbit interactions are taken
where we remember that ε is the extrinsic dielectric con-
into account, which are of the form
stant of the surrounding medium. As one may notice
∆SO † z from Fig. 18, the vacuum contribution Πvac (q) (thick
z z

HSO = Ψ τvalley ⊗ τAB ⊗ τspin Ψ (223) dashed lines) is only marginally modified by the magnetic
2
field, such that one may use e2 /εε∞ lB as an approxima-
and which provide again the same LL spectrum (219) tion for the interaction-energy scale for graphene taking
and (220) if one replaces M by ∆SO /2 (Kane and Mele, into account inter-band screening.
2005). In spite of the conceptually appealing prospect The relevant energy scales are summarized in the table
of the quantum spin Hall effect, which is revealed by below for different values of the magnetic field, in com-
this model because the spin orientation is locked to a parison with the interaction energy scales, taking into ac-
particular valley index via the term (223), the associ- count the effective dielectric constants for several widely
ated energy scale ∆SO ∼ 10 mK turns out to be van- used substrates from Tab. I.
energy value for arbitrary B for B = 6 T for B = 25 T

∆Z 1.2B[T] K 7K 30 K
∆kek 2B[T] K 12 K 50 K
p
e2 /εlB (bare) 625( B[T]/ε) K (1550/ε) K (3125/ε) K
2
p
e /εε∞ lB (vacuum) 139 B[T] K 344 K 694 K
p
e2 /εε∞ lB (on SiO2 ) 104 B[T] K 258 K 521 K
p
e2 /εε∞ lB (on h-BN) 109 B[T] K 270 K 543 K
p
e2 /εε∞ lB (on SiC) 71 B[T] K 176 K 355 K
∆sb < (e2 /εlB )(a/lB ) < 1B[T] K <6K < 25 K
Table II Energy scales at different magnetic fields. The first two lines show the energy scales associated with the major external symmetry-
breaking fields (Zeeman and √ Kekulé-type lattice distortion, ∆Z and ∆kek , respectively), which scale linearly in B. Below are shown the
interaction-energy scales (∝ B), the bare one with an unspecified dielectric constant and the ones for different substrates taking into
account inter-band screening via the term ε∞ [Eq. (224)]. The last line yields the interaction-energy scale associated with the intrinsic
symmetry breaking due to inter-valley coupling, discussed in Sec. V.A.2.
In view of the above discussion, one may conclude that yields the leading energy scale in the problem of elec-
the SU(4)-symmetric part of the Coulomb interaction trons in partially filled lower LLs, whereas external terms,
41
such as the Zeeman effect or spontaneous lattice distor- electrons in the spin-↑ states) is accompanied by a cost
tions, play a subordinate role. The remainder of this in kinetic energy – indeed, the Fermi energy for spin-
section is therefore concerned with a detailed discussion ↑ electrons is increased whereas that of spin-↓ electrons
of strongly-correlated electron phases that are formed to is lowered. The competition between the gain in inter-
minimize the Coulomb interaction. action and the cost in kinetic energy defines the degree
of polarization, i.e. how ferromagnetic the electrons ef-
fectively are. In the quantum Hall effect, however, we
B. SU(4) Quantum Hall Ferromagnetism in Graphene are confronted with a highly degenerate LL that may be
viewed as an infinitely flat band, such that the kinetic-
A prominent example of the above-mentioned strongly- energy cost for complete spin polarization is zero.
correlated phases is the generalized quantum Hall ferro- As an example of an SU(N ) quantum Hall ferromag-
magnet. It arises in systems with a discrete internal de- net, one may consider the state
gree of freedom described by an SU(N ) symmetry, such
B −1
k NY
that each single-particle quantum state ψn,m occurs in
c†m,i |vaci,
Y
N copies. Prominent examples are the non-relativistic |FMi = (225)
quantum Hall systems when taking into account the elec- i=1 m=0
tronic spin σ =↑, ↓ (N = 2) or bilayer quantum Hall sys-

which consists of k < N arbitrarily chosen completely
tems that consist of two parallel 2D electron gases, where
filled subbranches [i ∈ {(K, ↑), (K ′ , ↑), (K, ↓), (K ′ , ↓)},
the layer index may be viewed as a “spin” 1/2 [N = 2,
for the SU(N = 4) symmetry in graphene LLs], where we
or N = 4 if one also takes into account the physical spin
have omitted the LL index λn at the fermion operators
(Ezawa and Hasebe, 2002; Ezawa et al., 2003)].30 In this
to simplify the notation. The arbirariness in the choice
sense, graphene may be viewed as an SU(4) quantum
of the SU(N ) spin subbranches may be viewed as a spon-
Hall system as a consequence of its four-fold spin-valley
taneous symmetry breaking that accompanies the ferro-
degeneracy.
magnetism. Indeed, the state (225) is no longer invariant
under an SU(N ) rotation, but only under a rotation de-
scribed by the subgroup SU(k)×SU(N − k), where the
1. Ferromagnetic ground state and Goldstone modes
first factor indicates a symmetry transformation in the
fully occupied subbranches i = 1, ..., k and the second fac-
Quite generally, quantum Hall ferromagnetism arises tor one in the empty subbranches i = k +1, ..., N . There-
when the filling factor, defined from the bottom of fore the quantum Hall ferromagnet (225) is associated
the LL,31 is an integer that is not a multiple of N with an order parameter with a spontaneous symmetry
(Arovas et al., 1999). From the point of view of the ki- breaking described by the coset space SU(N )/SU(k) ×
netic Hamiltonian, one is thus confronted with a macro- SU(N − k) × U(1) ∼ U(N )/U(k) × U(N − k), where the
scopic ground-state degeneracy. Even if one has an inte- additional U(1) is due to the phase difference between the
ger filling factor, the situation is thus much more remi- occupied and the unoccupied subbranches (Arovas et al.,
niscent of the FQHE, i.e. the relevant energy scale is the 1999; Yang et al., 2006).
Coulomb interaction, and the system may be described in The coset space, with its N 2 − k 2 − (N − k)2 =
the framework of the model (207) of interacting electrons
2k(N − k) complex generators, defines also the Gold-
in a single (relativistic) LL. For the moment, we consider stone modes, which are nothing other than the k(N − k)
that there are no symmetry-breaking terms, such as the
spin-wave excitations of the ferromagnetic ground state
backscattering term (212) or Zeeman-type terms that are
(225).32 The number of the spin-wave modes may also
discussed below in Sec. V.A.4.
have been obtained from a simple inspection into the LL-
Qualitatively, one may understand the formation of a subbranch spectrum. Indeed, a spin wave can be de-
ferromagnetic ground state as a consequence of the repul- scribed with the help of the components of the projected
sive Coulomb interaction. In order to minimize this inter- density operators (203),
action, the electrons prefer to form a state described by
a maximally anti-symmetric orbital wave function that
m e−iq·R m′ c†m,i cm′ ,j ,
X
must then be accompanied by a fully symmetric SU(N ) ρ̄ij (q) = (226)
m,m′
spin wave function to satisfy an overall fermionic (anti-
symmetric) wave function. In a usual metal with a fi-
which represent coherent superpositions at wave vector
nite band dispersion, this ferromagnetic ordering (e.g. all
q of excitations from the occupied subbranch j to the
empty subbranch i. One has then k possibilities for the
choice of the initial subbranch j and N − k for the final
30 For a review on non-relativistic multi-component systems see
Ref. (Ezawa, 2000; Moon et al., 1995).
31 Remember that the filling factor in graphene is defined with re-
spect to the center of the n = 0 LL. There is thus a shift of 2 in 32 The complex generators come in by pairs of conjugate operators,
the filling factor as compared to the non-relativistic case. and each pair corresponds to one mode.
42
one, and one obtains therefore k(N − k) different spin- (a)

z
wave excitations, in agreement with the above group-

theoretical analysis. y
x
Notice that all spin-wave excitations have the
same dispersion, which may be calculated within
a mean-field approximation (Alicea and Fisher, 2006; (b)
Doretto and Morais Smith, 2007; Kallin and Halperin, z
1984; Yang et al., 2006),
Eq = hFM|ρ̄ij (−q)Hn ρ̄ij (q) − Hn |FMi y
x
2

X
2 q ∧ k lB
= 2 vn (k) sin , (227)
2
k
which saturates at large values of q = |q|, Figure 25 Excitations of the SU(2) ferromagnetic state. (a) Spin
n
X waves. Such an excitation can be continuously deformed into the
Eq→∞ = 2EX = vn (k), (228) ferromagnetic ground state [spin represented on the Bloch sphere
k (right) by fat arrow] – the grey curve can be shrinked into a sin-
gle point (b) Skyrmion with non-zero topological charge [from Ref.
i.e. at twice the value of the exchange energy (209). This (Girvin, 1999)]. The excitation consists of a a reversed spin at
result is not astonishing insofar as the large-q limit cor- the origin z = 0, and the ferromagnetic state is recovered at large
responds, as we have discussed in Sec. III.B.2 [see Eq. distances |z/ζ| → ∞. Contrary to spin-wave excitations, the spin
(167)], to an electron-hole pair where the electron is situ- explores the whole surface of the Bloch sphere and cannot be trans-
ated far away from the hole. The energy (228) is therefore formed by a continuous deformation into the majority spin (fat
arrow).
nothing other than the cost in exchange energy to create
a spin-flip excitation, i.e. an electron with reversed spin
and a hole in the ferromagnetic ground state. Because of
the large distance between the electron and the hole in terized by a particular elementary excitation that con-
such an excitation and the resulting decoupled dynamics, sists of a topological spin texture, the so-called skyrmion
one may be tempted to view this energy as the activa- (Sondhi et al., 1993). Similarly to a spin-wave in the
tion gap of the quantum-Hall state at ν = k, but we see limit qlB ≪ 1, the variation of the spin texture in a
below in Sec. V.B.2 that there exist elementary charged skyrmion excitation is small on the scale of the magnetic
excitations (skyrmions) that have, in some LLs, a lower length lB , such that its energy is determined by the spin
energy than these spin-flip excitations. stiffness (230) in the small-q limit. Indeed, one may show
In the opposite limit of small wave vectors (qlB ≪ 1), that its energy is given by (Ezawa, 2000; Moon et al.,
one may not understand the excitation in terms of decou- 1995; Sondhi et al., 1993)
pled holes and electrons, and the excitation can there-
fore not contribute to the charge transport. A Taylor
expansion of the sine in the spin-wave dispersion (227) Esk = 4πρns |Qtop |, (231)
yields the usual q 2 dispersion of the spin-wave Goldstone
modes, in terms of the topological charge Qtop , which may be
ρn 2 viewed as the number of times the Bloch sphere is covered
Eq→0 = s q 2 lB , (229)
2 by the spin texture [see Fig. 25(b)] and which we discuss
in terms of the spin stiffness in more detail below. Skyrmions are the relevant elemen-
tary excitations of the quantum Hall ferromagnetism if
1 X the energy (231) is lower than that of an added electron
ρns = vn (k)|k|2 lB
2
. (230)
4π (or hole) with reversed spin, which is nothing other than
k
n
the exchange energy (209), i.e. if Esk < EX . Whereas
One notices that the above results for the excitation en-
this condition is fulfilled in non-relativistic LLs only in
ergies do not depend on the size of the internal symmetry
the lowest one n = 0, skyrmions are the lowest-energy
group, but they can be derived within the SU(2) model
elementary excitations in the graphene LLs n = 0, 1 and
of the quantum Hall ferromagnetism (Moon et al., 1995;
2 (Töke et al., 2006; Yang et al., 2006), as a consequence
Sondhi et al., 1993) – the enhanced internal symmetry of
of the difference in the form factors.
graphene (or of a general N -component system) affects
only the degeneracies of the different modes. As in the case of the spin waves discussed above, the
skyrmion energy is independent of the size of the internal
symmetry group, and we will first illustrate the skyrmion
2. Skyrmions and entanglement texture in an effective SU(2) model, where the texture is
formed only from states within the last occupied (k) and
In addition to the above-mentioned spin-wave modes, the first unoccupied (k + 1) LL subbranch. The skyrmion
the SU(N ) ferromagnetic ground state (225) is charac- may then be described with the help of the wave function,
43
in terms of the complex coordinate z = (x − iy)/lB , given by Qtot = |ν−k|NB . The number of internal modes
is then Qtot (N − 1), in addition to one mode per charge
1 that corresponds to a simple translation z → z + a of
z ↑k (z) + ζ ↓k+1 (z)

|Sk,k+1 i = p
|ζ|2 + |z|2 the excitation (Douçot et al., 2008). As a consequence
(232) of the Coulomb repulsion, it is energetically favorable to
where | ↑k (z)i corresponds to states in the subbranch k form a state in which Qtot skyrmions of charge 1 are ho-
and | ↓k+1 (z)i to those in k + 1, at the position z. One mogeneously distributed over the 2D plane than a single
notices that at the origin z = 0 the “spin” associated with defect with charge Qtot (Moon et al., 1995). A natural
these two components is ↓ because the first component (semi-classical) candidate for the ground state of Qtot
of Eq. (232) vanishes, whereas the spins are ↑ at |z/ζ| → skyrmions is then a skyrmion crystal (Brey et al., 1995)
∞ (see Fig. 25), where the ferromagnetic ground state that has recently been revisited in the framework of the
is then recovered. The parameter ζ plays the role of SU(4) symmetry in graphene (Côté et al., 2007, 2008). In
the skyrmion size, measured in units of lB – indeed, for this case, the Qtot (N − 1) internal modes, which are dis-
|z| = |ζ|, both components are of the same weight and persionless zero-energy modes in the absence of electronic
the spin is therefore oriented in the xy-plane. interactions or Zeeman-type symmetry-breaking terms,
The skyrmion excitation (232) can also be illustrated are expected to yield N − 1 Bloch bands of Goldstone-
on the so-called Bloch sphere on the surface of which the type, in addition to the Qtot translation modes that form
(normalized) spin moves (see Fig. 25). The angles (θ a magnetic-field phonon mode of the skyrmion crystal
for the azimuthal and φ for the polar angle) of the spin with a characteristic ω ∝ q 3/2 dispersion (Fukuyama,
orientation on the Bloch sphere correspond to the SU(2) 1975).
parametrization |ψi = cos(θ/2)| ↑i + sin(θ/2) exp(iφ)| ↓i,
and the spin orientation at the circle |z| = |ζ| in the com-
a. Skyrmions and activation gaps in graphene. Quite gen-
plex plane describes the equator of the Bloch sphere. The
erally, the activation gap in quantum Hall states is the
topology of the skyrmion excitation becomes apparent
by the number of full circles the spin draws when going energy required to create a quasi-particle–quasi-hole pair,
in which the two partners are sufficiently well separated
around the origin of the xy-plane on the circle |z| = |ζ|.
More precisely, the topological charge Qtop is not defined to contribute independently to the charge transport. In
the framework of the quantum Hall ferromagnet, the
in terms of such closed paths, but it is the number of full
coverings of the Bloch sphere in a skyrmion excitation activation gap may be viewed as the energy to create
a skyrmion of topological charge Q = 1 and an anti-
[Qtop = 1 in the example (232)]. Notice that a spin-wave
excitation has a topological charge Qtop = 0 and corre- skyrmion of charge Q = −1 that are well-separated from
each other such that one may neglect their residual in-
sponds to an excursion of the spin on the Bloch sphere
teraction. The energy of such a skyrmion–anti-skyrmion
that is not fully covered and that can then be reduced
pair is then given, in the absence of symmetry-breaking
continuously to a single point describing the ferromag-
terms, by twice the energy in Eq. (231),
netic ground state [Fig. 25(a)].
The above considerations may be generalized to sys- ∆sym
a = 8πρns . (233)
tems with larger internal symmetries, i.e. to SU(N )
For graphene, the energies of the theoretical activation
quantum Hall ferromagnets. The state (232) is invariant
gaps for n = 0 and n = 1 are shown in the table below.
under the SU(N ) subgroup SU(k − 1) × SU(N − k − 1),
where the first factor describes a rotation of the occupied
subbranches that do not take part in the skyrmion exci- activation gap arbitrary value of B
p π e2
tation and the second factor is associated with a symme- 1
p
n=0 2 2 εε∞ lB
400( B[T]/εε∞ ) K
try transformation of the corresponding unoccupied sub- 7
p π e2 p
n = 1 32 175( B[T]/εε∞ ) K
branches k + 2, ..., N . A similar group-theoretical analy- 2 εε∞ lB
sis as the one presented in Sec. V.B.1 yields the number Table III Theoretical estimates for the activation gaps in the
of residual symmetry transformations (Yang et al., 2006) n = 0 and 1 graphene LLs due to well-separated skyrmion–anti-
2k(N − k) + 2(N − 1), where the first term describes the skyrmion pairs.
Goldstone modes of the ferromagnetic ground state, and
the second one corresponds to the N − 1 internal modes
of the skyrmion excitation.
For further illustration, we consider the scenario in
In addition to the topological charge, skyrmions in which the Zeeman effect is the only SU(4)-symmetry
quantum Hall systems carry an electric charge that co- breaking term.33 Due to the Zeeman effect, spin-↓ elec-
incides, for ν = k with the topological charge. Indeed,
the skyrmion state (232) describes an electron that is ex-
pelled from the origin z = 0 in the ↑-component, and its
net electric charge is therefore that of a hole. This means 33 The energetic argument remains valid in the case where the dom-
that skyrmions are excited when sweeping the filling fac- inant term is a valley-Zeeman effect if one interchanges the role
tor away from ν = k, and the net topological charge is of the spin and the valley pseudospin.
44
(a) z (b) z
trons are energetically favored. If only one spin-valley
branch of a particular LL is filled (k = 1), the spin mag- θS
θI
netization of the spin-valley ferromagnet is preferentially
y y
oriented in this direction whereas the valley polarization
φS x φI x
may point in any direction. The activation gap would
then be dominated by valley (anti-)skyrmions with no spin pseudospin
reversed physical spin such that one would not expect
any dependence of the gap on the in-plane component of
(c) z
the magnetic field, in agreement with the experimental
findings (Jiang et al., 2007b). α=0 α=0
The situation is different when both valley branches of
y
the spin-↓ branch are occupied; an excitation of the SU(4)
β x
ferromagnet with a full spin polarization would then nec-
essarily comprise reversed spins, and the corresponding entanglement
Zeeman energy must be taken into account in the energy
of the (spin) skyrmion–anti-skyrmion pair (233),
Figure 26 From Ref. (Douçot et al., 2008); Bloch spheres for
∆Za = 8πρns + 2Nrs ∆Z , (234) entangled spin-pseudospin systems. Bloch sphere for the spin
(a), pseudospin (b), and a third type of spin representing the
entanglement (c). In the case of spin-pseudospin entanglement
where Nrs ∼ |ζ|2 is the number of reversed spins in a sin- (| cos α| =
6 1), the (pseudo)spin-magnetizations explore the interior
gle (anti-)skyrmion. Notice that this number depends on of their spheres, respectively (black arrows).
the competition between the Zeeman effect itself, which
tries to reduce the skyrmion size ζ, and the cost in ex-
change energy due to the strong variation in small tex- four-spinor
tures (Moon et al., 1995; Sondhi et al., 1993).34 The en-
ergy of a skyrmion–anti-skyrmion pair in the spin channel α α
|Ψ(z)i = cos |ψS i|ψI i + sin eiβ |χS i|χI i , (235)
(with two completely filled valley sublevels) is therefore 2 2
larger than that (233) of a pair in the valley channel
when only one valley subbranch of the LL is completely where α and β are functions of the complex position z,
filled. Notice that this energy increase may even be sig- and the local two-component spinors |ψS i, |χS i, |ψI i,
nificant for large skyrmions because of the larger num- and |χI i are constructed according to
ber of reversed spins. As a thumbrule, the stability of a ! !
quantum Hall state is proportional to the activation gap, cos θ2 − sin θ2 e−iφ
which has in the present case been identified with the |ψi = and |χi = .
sin 2θ eiφ cos 2θ
skyrmion–anti-skyrmion energy and which is dominated
by the Coulomb interaction energy. Additional external (236)
symmetry-breaking terms, such as those discussed in Sec. The angles θ and φ define the usual unit vector
V.A.4, may enhance this stability although they provide
only a small correction to the activation energy. n(θ, φ) = (sin θ cos φ, sin θ sin φ, cos θ) , (237)
which explores the surface of the Bloch sphere depicted

in Fig. 25. Notice that one has two Bloch spheres, one for
b. Spin-valley entanglement in graphene.In an experimen- the unit vector n(θS , φS ) associated with the spin angles
tal measurement, one typically has not direct access to θS and φS and a second one for n(θI , φI ) for the valley-
the full SU(4) spin that describes the internal degrees pseudospin angles θI and φI (see Fig. 26). In addition,
of freedom in graphene LLs, but only to the SU(2) part one may associate a third Bloch sphere with the angles
associated with the physical spin, e.g. in a magnetiza- α and β that describe the degree of factorizability of the
tion measurement. It is therefore useful to parametrise wave functions and thus the degree of entanglement be-
the SU(4) spin in a manner such as to keep track of the tween the spin and the valley pseudospin (Douçot et al.,
two SU(2) copies associated with the physical spin and 2008).
with the valley pseudospin, respectively. This may be
With the help of the Schmidt decomposition (235), one
achieved with the so-called Schmidt decomposition of the
obtains immediately the reduced density matrices for the
spin and the valley-pseudospin sectors
α α
34 This energy cost may be evaluated from a gradient expansion of ρS = TrI (|ΨihΨ|) = cos2 |ψS ihψS | + sin2 |χS ihχS | ,
the energy in the magnetization fields. At leading order, one ob-
2 2
α α
tains, however, a non-linear sigma model that is scale-invariant, ρI = TrS (|ΨihΨ|) = cos2 |ψI ihψI | + sin2 |χI ihχI | ,
such that the energy cost must be calculated at higher orders 2 2
(Moon et al., 1995). (238)
45
respectively, and the local spin and valley-pseudospin state in graphene is not in the same universality class as
densities are simply that of the quantum-Hall bilayer – in the latter case the
symmetry of the (interaction) Hamiltonian is U(1) as a
maS = Tr(ρS S a ) = cos αhψS |S a |ψS i = cos α na (θS , φS ) consequence of the easy-plane anisotropy, and the sym-
(239) metry breaking is associated with a superfluid mode that
and disperses linearly with the wave vector, ω ∝ q. In con-
trast to this system, the interaction Hamiltonian (207)
mµI = Tr(ρI I µ ) = cos αhψI |I µ |ψI i = cos α nµ (θI , φI ), has the full SU(4) symmetry, and even for a sufficiently
(240) strong Zeeman effect, the symmetry is quite large with
where S a and I µ represent the components of the spin SU(2)↑ × SU(2)↓ , i.e. each spin projection ↑ and ↓ is gov-
and valley-pseudospin operators, respectively [see Eq. erned by the residual SU(2) valley symmetry and has the
(208)]. One notices from these expressions that, for the characteristic ω ∝ q 2 pseudospin-wave modes.
case α 6= 0 or π (i.e. cos2 α < 1), the local (pseudo)spin The connection between the two scenarios becomes
densities are no longer normalized, but they are of length transparent within a mean-field treatment of the
|mS/I |2 = cos2 α. Thus, in a semiclassical picture, the Coulomb interaction Hamiltonian. The quantum Hall
(pseudo)spin dynamics is no longer restricted to the sur- ferromagnetic states discussed in the previous subsec-
face of the Bloch sphere, but explores the entire volume tions may be described equivalently with the help of the
enclosed by the sphere (Fig. 26) (Douçot et al., 2008). mean-field order parameters
This result indicates that one may be confronted, in the
case of full entanglement (e.g. α = π/2), with an SU(4) D E
Ψ† τvalley
ν
⊗ 1AB ⊗ τspin
µ
Ψ , (241)
quantum Hall ferromagnet the (spin) magnetization of
which completely vanishes, as one would naively expect
for an unpolarized state. where Ψ denotes the same eight-spinor as in Sec. V.A.4.
Remember that a pure spin quantum Hall ferromagnet
ν
is obtained for τvalley = 1valley , whereas a pure valley-
3. Comparison with magnetic catalysis pseudospin ferromagnet is described by an order parame-
µ
ter (241) with τspin = 1spin . The remaining order param-
An alternative scenario proposed for the degeneracy eters describe states with a certain degree of spin-valley
lifting in n = 0 is that of the magnetic catalysis (Ezawa, entanglement, as discussed above.
2007; Gorbar et al., 2008; Gusynin et al., 2006; Herbut, Notice, however, that the choice of order parameters
2007b, 2008), which was discussed even before the dis- is not restricted to those in Eq. (241). Indeed, one may
covery of graphene (Gorbar et al., 2002; Khveshchenko, also opt for a mean-field calculation of the interaction
2001). According to this scheme, the Coulomb interac- Hamiltonian with the order parameters (Gorbar et al.,
tion spontaneously generates a mass term for the (origi- 2008; Gusynin et al., 2006)
nally massless) 2D electrons once the magnetic field in-
Ms = Ψ† τvalley
z z
⊗ 1spin Ψ

creases the density of states at zero energy by the for- ⊗ τAB (242)
mation of the highly degenerate n = 0 LL. As a con-
sequence of this mass generation, the particles condense and
in a state of coherent particle-hole pairs (excitonic con-
densation). The effect is at first sight reminiscent of the Mt = Ψ† (1valley ⊗ τAB
z
⊗ 1spin ) Ψ , (243)
excitonic condensation at ν = 1 in non-relativistic bi-
layer quantum Hall systems (Ezawa and Iwazaki, 1993; which describe mass gaps. Indeed, we already encoun-
Fertig, 1989; Wen and Zee, 1992a). Its superfluid behav- tered a term of the form (243) in Sec. V.A.4 and showed
ior gives rise to a zero-bias anomaly in the tunneling con- that it lifts the valley degeneracy of the n = 0 LL.
ductance between the two layers (Spielman et al., 2000) Whereas such a term arises naturally in the context of
as well as to a simultaneous suppression of the longitudi- an out-of-plane distortion of the graphene lattice, here,
nal and the Hall resistance in a counterflow experiment it is generated dynamically via the repulsive electron-
(Kellogg et al., 2004; Tutuc et al., 2004). The bilayer ex- electron interaction. The difference between the two
citonic condensate may be described as an easy-plane mass terms Ms and Mt stems from the residual sym-
quantum Hall ferromagnet (Moon et al., 1995), where metry of the SU(2)σ groups. The term (242), which may
the spin mimics the layer index. The origin of this easy- be viewed as a singlet mass term explicitly breaks this
plane anisotropy stems from the difference in the inter- symmetry, whereas the term (243) has been coined triplet
actions between electrons in the same layer as compared mass (Gorbar et al., 2008; Gusynin et al., 2006).
to the weaker one for electron pairs in different layers. In Sec. V.A.4, we have argued that mass terms of the
This comparison with non-relativistic 2D electrons in above form only lift the valley degeneracy in the zero-
bilayer systems indicates that there may exist a close re- energy LL n = 0, whereas they simply renormalize the
lation between the quantum Hall ferromagnetism and the LL energy for n 6= 0. Furthermore we have seen that as
A
scenario of the magnetic catalysis also in graphene in a a consequence of the vanishing spinor components ψK,σ
B
strong magnetic field. Notice, however, that the excitonic and ψK ′ ,σ , the mass term Mt is indistinguishable, in
46
n = 0, from a valley-pseudospin ferromagnetic state, an additional consideration of the subleading external

symmetry-breaking terms discussed in Sec. V.A.4. The
Mn=0
t = Ψ† (1valley ⊗ τAB
z
⊗ 1spin ) Ψ two-stage picture, which we adopt here based on the
∼ Ψ τvalley ⊗ 1AB ⊗ 1spin Ψ , (244)
† z above discussions, may be summarized as follows. (a)

The quantum Hall ferromagnetic states are formed to
whereas the singlet mass term simply renormalizes the minimize the leading energy given by the Coulomb in-
overall chemical potential, teraction. However, because of the (approximate) SU(4)
symmetry of the interaction, the orientation of the quan-
Mn=0 = Ψ† τvalley
z z
⊗ 1spin Ψ

s ⊗ τAB tum Hall ferromagnets is not fixed – a polarization in
∼ Ψ† (1valley ⊗ 1AB ⊗ 1spin ) Ψ . (245) the spin channel is as probable as one in the valley chan-
nel, and this yields the high degeneracy of the Goldstone
These arguments lead to the conclusion that the mag- modes described in Sec. V.B.1. (b) Therefore, in spite
netic catalysis in n = 0, i.e. the spontaneous genera- of the small energy scale of the external fields, the latter
tion of a mass gap due to electron-electron interactions, are relevant for the orientation of the ferromagnets and
may be fully described in the framework of the SU(4) for the degeneracy lifting of the Goldstone modes.
quantum Hall ferromagnetism. Furthermore, the recent
observation of a fully lifted spin-valley degeneracy in the
n = 1 graphene LL (Dean et al., 2011) is naturally under- a. The quantum Hall effect at νG = ±1. For νG = −1,
stood in the framework of quantum Hall ferromagnetism, only one spin-valley branch is completely filled by elec-
whereas the mass terms (242) and (243), obtained from trons.35 The Zeeman effect would give a small ener-
magnetic catalysis, would not provide a fully lifted spin- getic advantage to spin-↓ electrons, such that the two
valley degeneracy. spin Goldstone modes associated with collective excita-
tions to the spin-↑ acquire a q = 0 gap, given by ∆Z .
In contrast to the spin excitations, the Goldstone mode,
4. The quantum Hall effect at ν = ±1 and ν = 0 which couples the two valleys in the spin-↓ branch of
n = 0, remains gapless, and the ground state may thus
Before discussing the experimental results on the quan- be viewed as a valley-pseudospin ferromagnet in the spin-
tum Hall effect, a clarification on the filling factor is re- ↓ branch. The activation gap would be given by Eq.
quired. In the preceding parts of this section, which were (233) for pseudospin skyrmion–anti-skyrmion pairs, and
concerned with general aspects of the quantum Hall fer- √
its associated scaling e2 /εlB ∝ B has indeed been ob-
romagnet in LLs with internal degrees of freedom, the
served experimentally (Jiang et al., 2007b). The resid-
filling factor ν = k has been defined with respect to the
ual valley SU(2) symmetry may be broken by the lat-
bottom of the partially filled LL. However, in graphene,
tice distortions, which we have discussed in Sec. V.A.4.
this is at odds with the natural definition of the filling
Whereas an out-of-plane lattice distortion would yield a
factor (101) in terms of the electronic density measured
gapped valley-pseudospin-wave mode, a Kekulé-type in-
from the charge neutrality point in undoped graphene – a
plane distortion orients the pseudospin ferromagnet in
zero filling factor therefore corresponds to two completely
the XY -plane, associated with a gapless U(1) superfluid
filled spin-valley subbranches (k = 2) of the n = 0 LL. In
mode (Nomura et al., 2009). Notice that the lattice dis-
the remainder of this section, we therefore make a clear
tortion, characterized by the energy scale ∆kek is not
distinction between the two filling factors, and ν denotes
in competition, at νG = ±1, with the Zeeman effect,
the filling of the n = 0 LL measured from the bottom of
such that the resulting ferromagnetic state is the same
the level, whereas the natural filling factor (101) is from
for ∆Z > ∆kek as for ∆Z < ∆kek . In the remainder
now on denoted by νG . Explicitly, the relation between
of this section we restrict the discussion of the valley-
the two filling factors reads
pseudospin degeneracy lifting to in-plane distortions that
ν = νG + 2. (246) seem to be energetically more relevant than out-of-plane
distortions, but the overall picture remains unchanged if
Early transport measurements in exfoliated graphene the latter are more relevant.
on a SiO2 have revealed broken spin-valley-symmetry
states at at νG = 0, ±1 and ±4 (Jiang et al., 2007b;
Zhang et al., 2006), where the latter corresponds to the b. The quantum Hall effect at νG = 0. The situation is
LLs ±1. More recent experiments on exfoliated graphene more subtle at νG = 0, where it is not possible to fully po-
on a h-BN substrate have furthermore revealed quantum larize both the spin and the valley pseudospin and where
Hall states at νG = ±3 (Dean et al., 2011), thus com- the Zeeman effect is in competition with a lattice distor-
pleting the full resolution of the spin-valley quartet, not
only in n = 0, but also in ±1.
The observed states may generally be understood in
the framework of the quantum Hall ferromagnetism, but 35 For νG = +1, the same arguments apply in terms of holes due
the understanding of the situation at νG = 0 requires to particle-hole symmetry.
47
(a) (b) state at νG = 0 that is characterized by a bulk gap as-

sociated with two counter-propagating edge states [Fig.
K, K, 27(a)]. In the bulk, where M (y) = 0, both valley sub-
∆ kek K’, ∆Z levels of the spin-↓ branch are filled (spin ferromagnet).
K,
Energy
Energy
When approaching the edge, however, the energy term
∆Z ∆ kek
p
y K’, y ∆2kek + M (y)2 is enhanced by the rapidly increasing
K,
contibution from M (y), and the (K, ↑) level eventually
K’, K’, crosses the (K ′ , ↓) one at the edge at the Fermi energy
(Abanin et al., 2007b). This situation corresponds to
a quantum Hall state with a bulk insulator and (two
Figure 27 Possible scenarios for the lifted spin-valley degeneracy counter-propagating) conducting channels. In contrast
at νG = 0. (a) ∆Z > ∆kek in the bulk. When approaching to usual quantum Hall states, the edge states are not chi-
the edge, the energy difference between the two valleys increases
ral, but the chiralitiy, i.e. the transport direction, of each
drastically, and two levels (K ′ , ↑) and (K, ↓) cross the Fermi energy
at the edge depicted by the dashed line (Quantum Hall state). (b) channel is linked to its spin orientation.38 The quantum
∆kek > ∆Z in the bulk. The K subbranches are already located Hall state therefore remains stable unless magnetic impu-
above the Fermi energy, and those of K ′ below, such that the energy rities couple the two chiralities (Shimshoni et al., 2009).
difference is simply increased when approaching the edge with no One notices furthermore a change in the spin polariza-
states crossing the Fermi energy (Insulator).
tion at the edge; whereas the spin polarization in the
bulk is complete, the system becomes spin unpolarized
at the edge. If one takes into account the exchange inter-
tion that orients the valley pseudospin. For ∆Z > ∆kek , action, the change in the polarization takes place over a
it is favorable to fill both valley sublevels of the spin- certain distance, and the conducting properties may be
↓ branch and the resulting state is a spin ferromagnet described in terms of spin-carrying one-dimensional edge
with gapped spin-wave excitations. For ∆Z < ∆kek , a excitations (Shimshoni et al., 2009).
pseudospin-ferromagnetic ground state is favored with
In the opposite limit with ∆Z < ∆kek in the bulk [Fig.
both spin sublevels completely filled. The two different
27(b)], the system atpνG = 0 is already valley-polarized,
situations are depicted in Fig. 27. Most saliently, the two
phases reveal drastically different transport properties as and an increase of ∆2kek + M (y)2 when approaching
one may see from their behavior at the sample edges. the edge does not induce a level crossing at the Fermi
energy. Thus, there are no zero-energy states at the edge,
The electronic behavior at the edges may be described
and the system would be insulating both in the bulk and
within a model of electron confinement, in which the
at the edge.
sample edge is described via a mass confinement term
M (y)τABz
in the Hamiltonian, which has the symmetry From an experimental point of view, it is not fully set-
of the term (218) or else, in n = 0, that of a valley- tled which of the two phases describes the state at νG =
Zeeman term (217), as argued in Sec. V.A.4. The pa- 0. Whereas early experiments in exfoliated graphene on
rameter M (y) is zero in the bulk and increases drasti- SiO2 samples were discussed in the framework of a dom-
cally at the edge at a certain value of the coordinate inant Zeeman effect (Abanin et al., 2007b; Jiang et al.,
y.36 Although the model is a simplification to treat the 2007b; Zhang et al., 2006), more recent experiments at
graphene edges in the continuum description of the Dirac very large magnetic fields (Checkelsky et al., 2008, 2009)
equation, a more sophisticated treatment that takes into and on suspended graphene samples with increasing mo-
account the geometry of the edges yields, apart from a bility (Du et al., 2009) favor the insulating scenario of
fine structure of the levels at the edge, qualitatively simi- Fig. 27(b) with a dominant valley degeneracy lifting.
lar results (Brey and Fertig, 2006). The mass term M (y) Especially the high-field measurements hint at an easy-
modifies the valley coupling due p to the lattice distortion plane or XY (valley-pseudospin) ferromagnetic ground
and yields a y-dependent term ∆2kek + M (y)2 , which state because the transition between the metallic and the
therefore equally diverges at the sample edge.37 insulating state is reminiscent of a Kosterlitz-Thouless
These preliminary considerations on the gap behav- phase transition (Kosterlitz and Thouless, 1973) if one
ior at the edges allow us to appreciate the difference in replaces the temperature by the magnetic field as the
the expected electronic transport between a spin ferro- parameter driving the transition (Checkelsky et al., 2008,
magnet and a valley-pseudospin ferromagnet at νG = 0. 2009; Hou et al., 2010; Nomura et al., 2009). However, it
Indeed, for ∆Z > ∆kek , one obtains a quantum Hall has also been argued that this effect may be understood
within the above scenario of a Zeeman-dominated quan-
tum Hall ferromagnet in the bulk, in the framework of a
Luttinger-liquid description of the domain wall seperat-
36 For the present argument, we consider translation invariance in
the x-direction.
37 In the case of an out-of-plane distortion, the term M (y) simply
adds up to the energy scale ∆valley
Z [see Eq. (217)], but the 38 These helical edges are the signature of a quantum spin Hall
physical picture remains unaltered. effect (Hasan and Kane, 2010; Qi and Zhang, 2011).
48
ing the polarized from the unpolarized region at the edge strate (Dean et al., 2011).
(Shimshoni et al., 2009). Before commenting in more detail on these first exper-
One notices that both the Zeeman effect and the imental results (indeed, this part of graphene research
Kekulé-type lattice distortion are very close in energy has just started), we introduce the theoretical four-
(see Tab. II) such that one may speculate that other component or SU(4) picture of the FQHE in graphene,
effects, as e.g. impurities, strongly influence the forma- in terms of generalized Halperin wave functions. These
tion and the orientation of the quantum Hall ferromag- wave functions, which may be viewed as multi-component
net. Further experimental and theoretical studies there- generalizations of Laughlin’s wave function, provide the
fore seem to be necessary to clearly identify the leading natural framework for the description of the phenomenon
symmetry-breaking mechanisms, which need not be uni- in view of the model of electrons restricted to a single rel-
versal, in the zero-energy LL at νG = 0 and ±1. ativistic LL (Sec. V.A)
We finally mention scanning-tunneling spectroscopic
results for the level splitting at νG = 0 that were per-
formed on graphene on a graphite substrate (Li et al., 1. Generalized Halperin wave functions
2009a). Although a gap has been observed as one may
expect in the framework of the above scenario, it satu- The theoretical study of the FQHE is intimitely related
rates as a function of the√magnetic field. This is in dis- to trial N -particle wave functions. In 1983, Laughlin pro-
agreement with both the B-behavior of an interaction- posed a one-component wave function (Laughlin, 1983),
dominated gap as well as with the linear dependence of
N PN
the Zeeman or lattice-distortion effects. A probable ori- Y m |zk |2 /2
φL
m ({zk }) = (zk − zl ) e− k , (247)
gin of this gap is the commensurability of the graphene
k<l
lattice with the graphite substrate that may break the in-
version symmetry between the two sublattices by a term which allows for an understanding of incompressible
of the type (218). The coupling to the substrate being es- FQHE states at the filling factors ν = 1/m that are de-
sentially electrostatic, one would then expect no or only termined by the exponent m for the particle pairs k, l in
a weak magnetic-field dependence of the splitting, as ob- Eq. (247). The variable zk = (xk −iyk )/lB is the complex
served in the experiment (Li et al., 2009a). position of the k-th particle, and the form of the Laughlin
wave function (247) is dictated by the analyticity condi-
tion for wave functions in the lowest LL.40 Furthermore,
C. Fractional Quantum Hall Effect in Graphene the exponent m must be an odd integer as a consequence
of the fermionic statistics imposed on the electronic wave
The most salient aspect of strongly-correlated 2D elec- function. Even if Eq. (247) describes only a trial wave
trons in partially filled LLs is certainly the FQHE, which function, one can show that it is the exact ground state
is due to the formation of incompressible liquid phases at for a class of model interactions that yield, with the help
certain magical values of the filling factor. As we have of Eq. (215), the pseudopotentials (Haldane, 1983)
already argued in Sec. V.A.3 on the basis of the pseu-
dopotentials, the FQHE is expected to be present in the Vℓ > 0 for ℓ < m and Vℓ = 0 for ℓ ≥ m. (248)
graphene LLs n = 0 and n = 1, and the main differ-
ence with respect to non-relativistic 2D electron systems Although the Coulomb interaction does not fulfill such
should arise from the internal SU(4) symmetry [for a re- strong
√ conditions, the pseudopotentials decrease as
cent theoretical review see Ref. (Papić et al., 2009)]. 1/ m for large values of m. Because the incompressible
ground state is protected by a gap that is on the order
On the experimental level, recent progress in the fabri-
of V1 , one may view the pseudopotentials Vℓ≥m as an
cation of high-mobility samples, e.g. via current anneal-
irrelevant perturbation that does not change the nature
ing (Bolotin et al., 2008; Du et al., 2008), has allowed for
of the ground state. Indeed, exact-diagonalization calcu-
the observation of several FQHE states in graphene. The
lations have shown that, for the most prominent FQHE
first observations of a state at νG = ±1/3 were reported
at ν = 1/3, the overlap between the true ground state
in 2009 on current-annealed suspended graphene samples
and the Laughlin state (247) is extremely large (> 99%)
in the two-terminal configuration (Bolotin et al., 2009;
(Fano et al., 1986; Haldane and Rezayi, 1985).
Du et al., 2009).39 More recently, in 2010, the FQHE has
Soon after Laughlin’s original proposal, Halperin gen-
also been observed in the four-terminal geometry, which
eralized the wave function (247) to the SU(2) case of
allows for the simultaneous measurement of the longi-
tudinal and the Hall resistance, in suspended graphene
(Ghahari et al., 2011) and on graphene on a h-BN sub-
40 The lowest-LL condition of analytic wave functions may seem a
very strong restriction when discussing FQHE states in higher
LLs. However, the model (207) indicates that all LLs can be
39 There are also some weak indications for FQHE states at other treated as the lowest one, n = 0, if the interaction potential is
filling factors than νG = ±1/3 in these samples. accordingly modified. We adopt this point of view here.
49
electrons with spin, in the absence of a Zeeman effect where ν = ν1 + . . . + νN is the total filling factor mea-
(Halperin, 1983) – one has then two classes of particles, sured from the bottom of the lowest LL. Notice that Eq.
N↑ spin-↑ and N↓ spin-↓ particles, which are described (254) is only well-defined if the exponent matrix M is in-
(↑) (↓) vertible. In this case, all component filling factors νj are
by the complex positions zk↑ and zk↓ , respectively. In
the (theoretical) absence of interactions between elec- completely determined, whereas otherwise some of the
trons with different spin orientation, the most natural component fillings remain unfixed, e.g. ν1 and ν2 for the
ground-state candidate would then be a simple product sake of illustration, although the sum of them (ν1 + ν2 )
of two Laughlin wave functions (247), is fixed. This is nothing other than a consequence of the
underlying ferromagnetic properties of the FQHE state
that, similarly to the states at ν = k discussed in Sec.
n o n o
(↑) (↓)
φL
m↑ z k↑ φL
m↓ z k↓ , (249)
V.B, are described by subgroups of SU(N ).
Finally, we notice that not all SU(N ) wave functions
one for each spin component with the exponents m↑ and
(251) describe incompressible quantum liquids with a ho-
m↓ , respectively, that need not necessarily be identical.
mogeneous charge density for all components. A general-
Inter-component correlations may be taken into account
ization of Laughlin’s plasma picture, according to which
by an additional factor
the modulus square of the trial wave function corre-
N↑ N↓ n sponds to the Boltzmann weight of a classical 2D plasma
(↑) (↓)
Y Y
zk↑ − zk↓ , (250) (Laughlin, 1983), shows that all eigenvalues of the ex-
k↑ k↓ ponent matrix M must be positive (or zero for states
with ferromagnetic order). Otherwise, some of the differ-
where the exponent n can now also be an even integer be- ent components phase-separate in the 2D plane because
cause the fermionic anti-symmetry condition is concerned the inter-component repulsion between them exceeds the
only with electrons in the same spin state. intra-component repulsion (de Gail et al., 2008).
Halperin’s idea is easily generalized to the case of more
than two components, and the corresponding trial wave
function for an SU(N ) quantum Hall system with N com- 2. The use of generalized Halperin wave functions in graphene
ponents reads (Goerbig and Regnault, 2007)
SU(N ) These general considerations allow us to define the
ψm 1 ,...,mN ;nij
= φL inter
m1 ,...,mN φnij , (251) framework for a basic description of the FQHE in
in terms of the product graphene where the SU(4) spin-valley symmetry imposes
N = 4. Four-component Halperin wave functions are
Nj
N Y mj PN PN j (j)
therefore expected to play an equally central role in the
Y (j) (j) − |zk |2 /4
φL
m1 ,...,mN = zkj − zlj e j=1 kj =1 j description of the graphene FQHE as Laughlin’s in a one-
j=1 kj <lj component or Halperin’s in two-component systems. In
(252) the remainder of this section, we attribute the four spin-
of N Laughlin wave functions and the term valley components as 1 = (↑, K), 2 = (↑, K ′ ), 3 = (↓, K),
and 4 = (↓, K ′ ).
N Y Nj
Ni Y nij
(i) (j)
Y
φinter
nij = zki − zkj , (253)
i<j ki kj
a. Fractional SU(4) quantum Hall ferromagnet. In a first
which describes inter-component correlations. As in step, we consider a four-component Halperin wave func-
the case of Halperin’s two-component wave function tion in which all components are equal (odd) integers,
(Halperin, 1983), the exponents mj must be odd inte- mj = nij = m, regardless of whether they describe
gers for fermionic particles whereas the exponents nij intra- or inter-component correlations. One obtains then
may also be even integers. These exponents define a a completely anti-symmetric orbital wave function that
symmetric N × N matrix M = nij , where the diago- is accompanied by a fully symmetric SU(4) spin-valley
nal elements are nii ≡ mi . This exponent matrix encodes wave function.
the statistical properties of the quasi-particle excitations, As we have argued in Sec. V.B.1, this situation repre-
such as their (fractional) charge and their statistical an- sents precisely a perfect SU(4) quantum Hall ferromagnet
gle (Wen and Zee, 1992a,b). – indeed, for m = 1, the generalized Halperin wave func-
Moreover, the exponent matrix M determines the com- tion (251) is nothing other than the orbital wave function
ponent densities ρj – or, equivalently, the component fill- of the state at ν = 1, i.e. when one of the subbranches is
ing factors νj = ρj /nB , completely filled. The SU(4) symmetry is then sponta-
neously broken, and the group-theoretical analysis pre-

ν1
  
1 sented in Sec. V.B.1 yields 3 degenerate Goldstone modes
 ..  −1  ..  that are generalized spin waves.
 .  = M  . , (254) The situation is exactly the same for any other odd
νN 1 exponent m, but the orbital wave function (251) is then
50
a Laughlin wave function (247) in terms of the particle may be viewed as an SU(4) singlet where the filling fac-
positions zk regardless of their internal index j = 1, ..., 4. tors of all spin-valley components are 1/(4m − 3). Exact-
The ferromagnetic properties of these wave functions may diagonalization calculations for N = 4 and 8 particles
be described by the same equations as the spin-wave and have shown that the [m; m − 1, m − 1] wave function with
skyrmion modes derived in Sec. V.B if one takes into ac- m = 3 (at ν = 4/9) describes to great accuracy the
count a renormalization of the spin stiffness, as it has ground state for a Coulomb interaction (205), with over-
been discussed extensively in the literature for SU(2) laps ON =8 = 0.992 in n = 0 and ON =8 = 0.944 and in
quantum Hall ferromagnets (Ezawa, 2000; Moon et al., the n = 1 graphene LL (Papić et al., 2009).
1995; Sondhi et al., 1993). States described by such a
wave function are ground-state candidates for the filling
factors ν = 1/m, which correspond to the graphene fill-
ing factors [see Eq. (246)] νG = −2 + 1/m or hole states
b. A route to understanding the graphene FQHE at νG =
at νG = 2 − 1/m. ±1/3. The discussion of the above-mentioned states was
There are now two different manners to break the in-
based on the understanding acquired from quantum Hall
ternal SU(4) symmetry explicitly. The simplest one is systems in semiconductor heterostructures, where the fill-
the same as for the quantum Hall ferromagnetism at ing factor is defined with respect to the bottom of the
νG = 0 or ±1, in terms of external symmetry-breaking n = 0 LL. First experimental observations, however, in-
fields such as those discussed in Sec. V.A.4. However, dicated a prominent FQHE at νG = ±1/3, which cor-
one may also change some of the exponents in the gen- responds to two completely filled spin-valley sublevels of
eralized Halperin wave function (251), in which case one the graphene n = 0 LL, and a third one that is 1/3 filled,
also changes the filling factor. One may for instance con-
ν = 2 + 1/3. Such a state would naturally arise in a
sider the [m; m− 1, m] wave function with mj = m for all system where the SU(4) symmetry is strongly broken,
j, n13 = n24 = m − 1, and n12 = n14 = n23 = n34 = m,
e.g. by a strong Zeeman effect. However, as argued in
which correspond to a filling factor41 Sec. V.A.4, these external fields are weak as compared
2 2 to the leading interaction energy scale, and it is therefore
ν= or νG = −2 + . (255) natural to ask how such a state may arise from the inter-
2m − 1 2m − 1
action point of view in the framework of four-component
Indeed, the difference in the inter-component exponents Halperin wave functions.
explicitly breaks the spin-valley symmetry – electrons in A Halperin wave function that describes the above-
different valleys are more weakly correlated (with an ex- mentioned situation is (Papić et al., 2010)
ponent m − 1) than electrons in the same valley (expo-
nent m), regardless of their spin orientation. As a con- 3 Y
Y Y ↑,ξ ′ 3

sequence, the filling factors in each of the two valleys, ψ2+1/3 = zi − zj↑,ξ zi↑,K − zj↑,K
νK = ν1 + ν3 and νK ′ = ν2 + ν4 , respectively, are fixed, ξ=K,K ′ i<j i,j
νK = νK ′ = 1/(2m − 1), and one may view the wave Y Y
function as a state with ferromagnetic spin ordering, but × zi↓,ξ − zj↓,ξ , (257)
that is valley-pseudospin unpolarized. Alternatively, the ξ=K,K ′ i<j
[m; m−1, m] wave function may be interpreted as a tensor
product of an SU(2) Halperin (m, m, m − 1) pseudospin- or any permutation of the spin-valley components. One
singlet wavefunction (Halperin, 1983) and a completely notices that this wave function implicitly breaks the
symmetric (ferromagnetic) two-spinor that describes the SU(4) spin-valley symmetry and, moreover, is not an
physical spin. The relevance of the [m; m − 1, m] wave eigenstate of the full SU(4) pseudospin, such that it can-
function, with m = 3 (ν = 2/5) has been corroborated in not describe the ground state in the total absence of
recent exact-diagonalization studes, both in the graphene an external symmetry-breaking field. However, exact-
LLs n = 0 and n = 1 (Papić et al., 2009; Töke and Jain, diagonalization calculations have shown that even a
2007). tiny external Zeeman field is capable to stabelize the
The SU(4) spin-valley symmetry is fully broken, e.g., state (257), which becomes the ground state for ∆1Z ≃
in the case of the [m; m − 1, m − 1] wave function with 0.01e2/εlB (Papić et al., 2010). Furthermore, the state
all mj = m and off-diagonal nij = m − 1. This wave (257) possesses, in addition to the valley-pseudospin-
function, which describes a state at wave Goldstone mode in the spin-↑ branch, low-lying
spin-flip excitations for moderately small Zeeman fields,
4 4
ν= or νG = −2 + , (256) even if the charge (activation) gap is the same as for the
4m − 3 4m − 3 usual 1/3 Laughlin state. This particular interplay be-
tween the leading Coulomb energy and subordinate ex-
ternal spin-valley symmetry breaking terms, illustrated
41 We only discuss electronic states here, but the arguments are at the νG = 1/3 example, shows the complexity of the
equally valid for the particle-hole symmetric states at νG = 2 − graphene FQHE, and further surprises may be expected
2/(2m − 1). in future experiments.
51
3. Experiments on the graphene FQHE the n = 1 LL. The estimation of the activation gap at
νG = 4/3 agrees again reasonably well with the theo-
We terminate this section on the graphene FQHE with retical expectation for the 1/3 state. The perhaps most
a short discussion of experimental observations in the salient (and unexpected) feature of the transport mea-
light of the above-mentioned theoretical four-component surement is the absence (or extreme weakness) of the
picture. νG = ±5/3 representative of the 1/3 family, which would
correspond to the Laughlin state (ν = 1/3 ↔ νG = −5/3
and the corresponding hole state) with a full SU(4) spin-
a. Two-terminal measurements. In the first observations valley ferromagnetic order, as argued in Sec. V.C.2.
of the FQHE, the two-terminal configuration was used, Whereas the absence of this state remains to be un-
where the voltage (and thus the resistance) is measured derstood, these findings corroborate the theoretical four-
between the same two contacts used to drive the elec- component picture of the graphene FQHE. Indeed, it
tric current through the sample (Bolotin et al., 2009; clearly shows that the SU(4) symmetry of the n = 0
Du et al., 2009). In this two-terminal configuration, it LL is essential because the only correspondence between
is not possible to measure simultaneously the Hall and the FQHE states is particle-hole symmetry that maps
the longitudinal resistance. It is nevertheless possible to νG ↔ −νG . If the SU(4) symmetry were broken, e.g.
extract the Hall and the longitudinal conductivities from by a sufficiently strong Zeeman effect, the only symme-
the two-terminal resistance with the help of a conformal try would be the valley-SU(2) symmetry in each spin
mapping, as a consequence of the 2D nature of the quan- branch of the n = 0 LL, in which case there exist the
tum transport in these systems (Abanin and Levitov, further mappings −2 + ν ↔ −ν in the spin-↓ branch and
2008; Williams et al., 2009). This technique has been ap- ν ↔ 2 − ν in the spin-↑ branch. However, the (observed)
plied to obtain insight into the longitudinal conductivity ±1/3 state would than be mapped on the (unobserved
the expected activated behavior of which yields a rough or extremely weak) ±5/3 state, and the strong difference
estimate of the activation gap at νG = 1/3 (∆1/3 ∼ 4.4 K in the visibility between these two states is therefore dif-
at B = 12 T) (Abanin et al., 2010), which is an order of ficult to understand. This is also the case if the SU(4)
magnitude smaller than the theoretically expected value symmetry is fully broken by strong external spin and
(Apalkov and Chakraborty, 2006; Töke et al., 2006).42 valley Zeeman fields, such that all spin-valley sublevels
are completely resolved, and ±5/3 would be mapped on
±4/3, in the same manner as ±1/3 on ±2/3.
b. Four-terminal measurements. The activation gap of
the 1/3 FQHE state has also been measured in suspended
graphene in the four-terminal configuration, in which VI. CONCLUSIONS AND OUTLOOK
the longitudinal resistance can be measured directly and
independently from the Hall resistance (Ghahari et al., We have reviewed the quantum-mechanical properties
2011). In this case, the activation gap has been estimated of relativistic 2D electrons in monolayer graphene ex-
to be ∆1/3 ∼ 26...50 K at B = 14 T, a value that agrees posed to a strong magnetic field. The main parts of this
reasonably well with the theoretically expected value review are concerned with the role of electronic interac-
(Apalkov and Chakraborty, 2006; Töke et al., 2006) if tions in graphene LLs. Whereas we have argued that
one considers the energy scale e2 /εε∞ lB , which takes into these interactions may be treated perturbatively in the
account the RPA dielectric constant ε∞ for graphene in regime of the relativistic (integer) quantum Hall effect
vacuum (see Sec. III.B.4). (RQHE), they constitute the relevant energy scale in par-
Finally, we would mention very recent high-field trans- tially filled graphene LLs due to the quenching of the ki-
port measurements in the four-terminal configuration on netic energy. This is reminiscent of partially filled LLs in
graphene on a h-BN substrate (Dean et al., 2011). These non-relativistic 2D electron systems, and the most promi-
experiments allowed for a clear identification of several nent consequence of this quenched kinetic energy and the
states of the 1/3 family, at νG = ±1/3, ±2/3, and ±4/3 macroscopic LL degeneracy is certainly the FQHE. The
corresponding to the zero-energy LL n = 0, as well as graphene FQHE is expected to be reminiscent of that of
at νG = ±7/3, ±8/3, ±10/3, and ±11/3 which reside in non-relativistic 2D electrons but it is governed by a larger
internal degeneracy described to great accuracy by the
SU(4) group. The experimental study of the FQHE in
graphene is still in its infancy, and novel surprises may
42 Notice that the theoretical estimates have been obtained within be expected. Only recently have been reported mea-
a simplified two-component model, with a completely frozen surements in the four-terminal geometry which allow for
spin degree of freedom. In spite of this simplification, the an analysis of prominent characteristics of FQHE states,
above-mentioned exact-diagonalization calculations with an im- such as the activation gaps. In view of the generally
plemented SU(4) symmetry have shown that the charge gap,
which is responsible for the activated behavior, coincides indeed accepted universality of the quantum Hall effect, it will
with that obtained in the two-component model (Papić et al., certainly be interesting to make a systematic compari-
2010). son with the activation gaps of related FQHE states in
52
conventional 2D electron gases with a parabolic band. Appendix A: Matrix Elements of the Density Operators
In the perturbative regime of the RQHE, the theoret-
ical study of electron-electron interactions indicates the The matrix elements that intervene in the expres-
presence of fascinating novel collective modes, such as sion for the density operators (139) are of the form
linear magneto-plasmons, that are particular to graphene hn, m| exp(−iq · r)|n′ , m′ i and may be calculated with
and do not have a counterpart in non-relativistic 2D elec- the help of the decomposition of the cyclotron variable η
tron systems in a perpendicular magnetic field. Also the and the guiding centre R into the ladder operators â and
upper-hybrid mode, which is the magnetic-field counter- b̂, respectively [see Eqs. (78), (91) and (94)]. We further-
part of the usual 2D plasmon, is expected to behave in more define the complex wave vectors q ≡ (qx + iqy )lB
a particular manner in graphene as a consequence of the and q̄ = (qx − iqy )lB ,43 One finds
linear disperison relation and the vanishing band mass.
Whereas these studies are at present only theoretical,
these collective modes may find an experimental verifica- hn, m|e−iq·r |n′ , m′ i = hm|e−iq·R |m′ i ⊗ hn|e−iq·η |n′ i
tion in inelastic light-scattering measurements. − √i2 (qb̂† +q̄ b̂)
= hm|e |m′ i (A1)
Similarly to the role of electron-electron interactions − √i2 (q̄â† +qâ) ′
in the RQHE regime, the electron-phonon coupling ⊗hn|e |n i.
yields exciting resonance phenomena in graphene in a
strong magnetic field. The electron-phonon interaction The two matrix elements may be simplified with the
in graphene LLs has been discussed in the framework of help of the Baker-Hausdorff formula exp(A) exp(B) =
a perturbative approach. Indications for the magneto- exp(A + B) exp([A, B]/2), for the case [A, [A, B]] =
phonon resonance, e.g., have recently been found in Ra- [B, [A, B]] = 0 (Cohen-Tannoudji et al., 1973). The sec-
man spectroscopy of epitaxial graphene. ond matrix element thus becomes, for n ≥ n′
The present review has been limited to monolayer
graphene, and it is definitely a reasonable research pro- − √i2 (q̄â† +qâ)
gram to ask how the effects described here manifest them- hn|e−iq·η |n′ i = hn|e |n′ i (A2)
selves in bilayer graphene. For example, the particular 2 − √i2 q̄â† − √i2 qâ
= e−|q| /4
hn|e e |n′ i
collective excitations described in Sec. III have been at- 2 X i
− √ q̄â †
− √i qâ
tributed to the lack of equidistant LL spacing and the = e−|q| /4
hn|e 2 |jihj|e 2 |n′ i
presence of two bands. Whereas bilayer graphene also j
consists of two (particle-hole-symmetric) bands in the
r n−n′
n′ !

−|q|2 /4 −iq̄
low-energy regime, the approximate parabolicity there = e √
n! 2
yields almost equidistant LLs. The presence of additional ′
n n′ −j
high-energy bands (in the 300 meV range) certainly also |q|2

X n!
affects the plasmonic modes. × −
j=0
(n − j)!(n′ − j)!j! 2
r n−n ′ 2
n′ ! −iq̄

2 ′ |q|
= e−|q| /4 √ Lnn−n
′ ,
n! 2 2
Acknowledgments
where we have used
I would like to express my deep gratitude to nu-
merous collaborators without whom the realisation of 
this review would not have been possible. Above all, 0 for j > n
− √i q̄â†
I must acknowledge the very fruitful long-term col- hn|e 2 |ji = q n! 1 i n−j
 j! (n−j)! − √ q̄ for j ≤ n
2
laborations with Jean-Noël Fuchs, on electron-electron
interactions in the IQHE regime and electron-phonon
coupling, and with Nicolas Regnault on the FQHE in in the third line and the definition of the associated La-
graphene. I would furthermore thank my collabora- guerre polynomials (Gradshteyn and Ryzhik, 2000),
tors Claire Berger, Raphaël de Gail, Benoı̂t Douçot,
Volodya Fal’ko, Clément Faugeras, Kostya Kechedzhi, ′
n
Pascal Lederer, Roderich Moessner, Gilles Montambaux, ′ X n! (−x)m
Cristiane Morais Smith, Zlatko Papić, Frédéric Piéchon, Lnn−n
′ (x) = .
m=0
(n′ − m)!(n − n′ + m)! m!
Paulina Plochocka, Marek Potemski, Rafael Roldán, and
Guangquan Wang. Many thanks also to my colleagues
Hélène Bouchiat, Antonio Castro Neto, Jean-Noël Fuchs,
Christian Glattli, Paco Guinea, Anuradha Jagannathan, 43 We use this notation solely in the present appendix. Throughout
Philip Kim, and Bernhard Wunsch for fruitful discussions the main text, q denotes the modulus of the wave vector q, q =
and a careful reading of this review. |q|.
53
In the same manner, one obtains for m ≥ m′ Bena, C., and G. Montambaux, 2009, New J. Phys. 11,
095003.
− √i2 (qb̂† +q̄b̂)
hm|e−iq·R |m′ i = hm|e |m′ i Berger, C., Z. Song, T. Li, A. Y. Ogbazghi, R. Feng, Z. Dai,
r m−m′ A. N. Marchenkov, E. H. Conrad, P. N. First, and W. A.
m′ ! −iq

−|q|2 /4 de Heer, 2004, J. Phys. Chem. 108, 19912.
= e √
m! 2 Berman, O., G. Gumbs, and Y. E. Lezovik, 2008, Phys. Rev.
2 B 78, 085401.
′ |q|
×Lm−m
m′ . (A3) Bolotin, K. I., F. Ghahari, M. D. Shulman, H. L. Stormer,
2 and P. Kim, 2009, Nature 462, 196.
Bolotin, K. I., K. J. Sikes, Z. Jiang, M. Klima, G. Fudenberg,
With the help of the definition J. Hone, P. Kim, and H. L. Stormer, 2008, Solid State
r n−n′ 2 Commun. 146, 351.
n′ ! −iq |q|

′
Gn,n′ (q) ≡ √ Lnn−n , Bostwick, A., T. Ohta, T. Seyller, K. Horn, and E. Rotenberg,
′
n! 2 2 2007, Nat. Phys. 3, 36.
Brey, L., and H. Fertig, 2006, Phys. Rev. B 73, 195408.
one may rewrite the expressions without the conditions Brey, L., H. A. Fertig, R. Côté, and A. H. MacDonald, 1995,
n ≥ n′ and m ≥ m′ , Phys. Rev. Lett. 75, 2562.
Büttiker, M., 1992, The Quantum Hall Effect in Open Con-
hn|e−iq·η |n′ i = [Θ(n − n′ )Gn,n′ (q̄) (A4) ductors, in M. Reed (Ed.) Nanostructured Systems (Semi-
′ −|q|2 /4 conductors and Semimetals, 35, 191) (Academic Press,
+Θ(n − n − 1)Gn′ ,n (−q)] e
Boston).
and Bychkov, Y. A., and G. Martinez, 2008, Phys. Rev. B 77,
125417.
hm|e−iq·R |m′ i = [Θ(m − m′ )Gm,m′ (q) (A5) Castro Neto, A. H., and F. Guinea, 2007, Phys. Rev. B 75,
045404.
′ −|q|2 /4
+Θ(m − m − 1)Gm′ ,m (−q̄)] e . Castro Neto, A. H., F. Guinea, N. M. R. Peres, K. S.
Novoselov, and A. K. Geim, 2009, Rev. Mod. Phys. 81,
109.
References Checkelsky, J. G., L. Li, and N. P. Ong, 2008, Phys. Rev.
Lett. 100, 206801.
Abanin, D. A., P. A. Lee, and L. S. Levitov, 2007a, Phys. Checkelsky, J. G., L. Li, and N. P. Ong, 2009, Phys. Rev. B
Rev. Lett. 98, 156801. 79, 115434.
Abanin, D. A., and L. S. Levitov, 2008, Phys. Rev. B 78, Chiu, K. W., and J. J. Quinn, 1974, Phys. Rev. B 9, 4724.
035416. Cohen-Tannoudji, C., B. Diu, and F. Laloë, 1973, Quantum
Abanin, D. A., K. S. Novoselov, U. Zeitler, P. A. Lee, A. K. Mechanics (Hermann, Paris).
Geim, and L. S. Levitov, 2007b, Phys. Rev. Lett. 98, Coleman, P., 2003, An.. Henri Poincaré 4, S559.
196806. Cooper, K. B., M. P. Lilly, J. P. Eisenstein, L. N. Pfeiffer, and
Abanin, D. A., I. Skachko, X. Du, E. Y. Andrei, and L. S. K. W. West, 1999, Phys. Rev. B 60, R11285.
Levitov, 2010, Phys. Rev. B 81, 115410. Côté, R., D. B. Boisvert, J. Bourassa, M. Boissonneault, and
Abergel, D. S. L., V. Apalkov, J. Berashevich, K. Ziegler, and H. A. Fertig, 2007, Phys. Rev. B 76, 125320.
T. Chakraborty, 2010, Advances in Physics 59, 261. Côté, R., J.-F. Jobidon, and H. A. Fertig, 2008, Phys. Rev. B
Abergel, D. S. L., and V. I. Fal’ko, 2007, Phys. Rev. B 75, 78, 085309.
155430. Damascelli, A., 2004, Phys. Scripta T109, 61.
Abramowitz, M., and I. Stegun, 1970, Handbook of Mathe- de Gail, R., N. Regnault, and M. O. Goerbig, 2008, Phys.
matical Functions (9th Ed.) (Dover Publications, Dover). Rev. B 77, 165310.
Ajiki, H., and T. Ando, 1995, J. Phys. Soc. Jpn. 64, 260. Dean, C., A. Young, P. Cadden-Zimansky, L. Wang, H. Ren,
Aleiner, I. L., and L. I. Glazman, 1995, Phys. Rev. B 52, K. Watanabe, T. Taniguchi, P. Kim, J. Hone, and K. Shep-
11296. ard, 2011, Nature Phys. 7, 693.
Alicea, J., and M. P. A. Fisher, 2006, Phys. Rev. B 74, Dietl, P., F. Piéchon, and G. Montambaux, 2008, Phys. Rev.
075422. Lett 98, 236405.
Ando, T., 2006a, J. Phys. Soc. Jpn. 75, 074716. Dillon, R. O., I. L. Spain, and J. W. McClure, 1977, J. Phys.
Ando, T., 2006b, J. Phys. Soc. Jpn 75, 124701. Chem. Solids 38, 635.
Ando, T., 2007a, J. Phys. Soc. Jpn. 76, 024712. Doretto, R. L., and C. Morais Smith, 2007, Phys. Rev. B 76,
Ando, T., 2007b, J. Phys. Soc. Jpn. 76, 024712. 195431.
Andrei, E. Y., G. Deville, D. C. Glattli, F. I. B. Williams, Douçot, B., M. O. Goerbig, P. Lederer, and R. Moessner,
E. Paris, and B. Etienne, 1988, Phys. Rev. Lett. 60, 2765. 2008, Phys. Rev. B 78, 195327.
Apalkov, V. M., and T. Chakraborty, 2006, Phys. Rev. Lett. Drut, J. E., and T. A. Lähde, 2009a, Phys. Rev. Lett. 102,
97, 126801. 026802.
Arikawa, M., Y. Hatsugai, and H. Aoki, 2008, Phys. Rev. B Drut, J. E., and T. A. Lähde, 2009b, Phys. Rev. B 79, 165425.
78, 205401. Du, R. R., D. C. Tsui, H. L. Stormer, L. N. Pfeiffer, K. W.
Arovas, D. P., A. Karlhede, and D. Lilliehöök, 1999, Phys. Baldwin, and K. W. West, 1999, Solid State Comm. 109,
Rev. B 59, 13147. 389.
Banerjee, S., R. R. P. Singh, V. Pardo, and W. E. Pickett, Du, X., I. Skachko, A. Barker, and E. Y. Andrei, 2008, Nat.
2009, Phys. Rev. Lett. 103, 016402. Nanotech. 3, 491.
54
Du, X., I. Skachko, F. Duerr, A. Luican, and E. Y. Andrei, 241405.

2009, Nature 462, 192. González, J., F. Guinea, and M. A. H. Vozmediano, 1994,
Eisenstein, J. P., K. B. Cooper, L. N. Pfeiffer, and K. W. Nucl. Phys. B 424, 595.
West, 2002, Phys. Rev. Lett. 88, 076801. González, J., F. Guinea, and M. A. H. Vozmediano, 1999,
Esaki, K., M. Sato, M. Kohmoto, and B. I. Halperin, 2009, Phys. Rev. B 59, R2474.
Phys. Rev. B 80, 125405. Gorbar, E. V., V. P. Gusynin, V. A. Miransky, and I. A.
Ezawa, M., 2007, J. Phys. Soc. Jpn. 76, 094701. Shovkovy, 2002, Phys. Rev. B 66, 045108.
Ezawa, Z. F., 2000, Quantum Hall Effects – Field Theoretical Gorbar, E. V., V. P. Gusynin, V. A. Miransky, and I. A.
Approach and Related Topics (World Scientific, Singapore). Shovkovy, 2008, Phys. Rev. B 78, 085437.
Ezawa, Z. F., and K. Hasebe, 2002, Phys. Rev. B 65, 075311. Gradshteyn, I. S., and I. M. Ryzhik, 2000, Table of Inte-
Ezawa, Z. F., and A. Iwazaki, 1993, Phys. Rev. B 47, 7295. grals, Series and Products (6th Ed.) (Academic Press, San
Ezawa, Z. F., G. Tsitsishvili, and K. Hasebe, 2003, Phys. Rev. Diego).
B 67, 125314. Graf, D., F. Molitor, K. Ensslin, C. Stampfer, A. Jungen,
Fano, G., F. Ortolani, and E. Colombo, 1986, Phys. Rev. B C. Hierold, and L. Wirtz, 2007, Nano Lett. 7, 238.
34, 2670. Greiter, M., X.-G. Wen, and F. Wilczek, 1991, Phys. Rev.
Farjam, M., and H. Rafii-Tabar, 2009, Phys. Rev. B 80, Lett. 66, 3205.
167401. Gupta, A., G. Chen, P. Joshi, S. Tadigadapa, and P. Eklund,
Faugeras, C., M. Amado, P. Kossacki, M. Orlita, M. Sprinkle, 2006, Nano Lett. 6, 2667.
C. Berger, W. A. de Heer, and M. Potemski, 2009, Phys. Gusynin, V. P., V. A. Miransky, S. G. Sharapov, and I. A.
Rev. Lett. 103, 186803. Shovkovy, 2006, Phys. Rev. B 74, 195429.
Ferrari, A. C., J. C. Meyer, V. Scardaci, C. Casiraghi, Gusynin, V. P., and S. G. Sharapov, 2005, Phys. Rev. Lett.
M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K. S. 95, 146801.
Novoselov, S. Roth, and A. K. Geim, 2006, Phys. Rev. Gusynin, V. P., and S. G. Sharapov, 2006, Phys. Rev. B 73,
Lett. 97, 187401. 245411.
Fertig, H. A., 1989, Phys. Rev. B 40, 1087. Haldane, F. D. M., 1983, Phys. Rev. Lett. 51, 605.
Fogler, M. M., A. A. Koulakov, and B. I. Shklovskii, 1996, Haldane, F. D. M., 1988, Phys. Rev. Lett. 61, 2015.
Phys. Rev. B 54, 1853. Haldane, F. D. M., and E. H. Rezayi, 1985, Phys. Rev. Lett.
Fuchs, J.-N., and P. Lederer, 2007, Phys. Rev. Lett. 98, 54, 237.
016803. Halperin, B. I., 1983, Helv. Phys. Acta 56, 75.
Fukuda, Y. et al. (Super-Kamiokande Collaboration), 1998, Harrison, W. A., 1981, Phys. Rev. B 24, 5835.
Phys. Rev. Lett. 81, 1562. Hasan, M. Z., and C. L. Kane, 2010, Rev. Mod. Phys. 82,
Fukuyama, H., 1975, Solid State Commun. 17, 1323. 3045.
Gangadharaiah, S., A. M. Farid, and E. G. Mishchenko, 2008, Hasegawa, Y., R. Konno, H. Nakano, and M. Kohmoto, 2006,
Phys. Rev. Lett. 100, 166802. Phys. Rev. B 74, 033413.
Geim, A. K., and K. S. Novoselov, 2007, Nat. Materials 6, de Heer, W. A., C. Berger, X. Wu, P. N. First, E. H. Conrad,
183. X. B. Li, T. B. Li, M. Sprinkle, J. Hass, M. Sadowski,
Ghahari, F., Y. Zhao, P. Cadden-Zimansky, K. Bolotin, and M. Potemski, and G. Martinez, 2007, Solid State Comm.
P. Kim, 2011, Phys. Rev. Lett. 106, 046801. 143, 92.
Girvin, S. M., 1999, The Quantum Hall Effect: Novel Excita- Herbut, I. F., 2006, Phys. Rev. Lett. 97, 146401.
tions and Broken Symmetries, in A. Comptet, T. Jolicoeur, Herbut, I. F., 2007a, Phys. Rev. B 76, 085432.
S. Ouvry and F. David (Eds.) Topological Aspects of Low- Herbut, I. F., 2007b, Phys. Rev. B 75, 165411.
Dimensional Systems – École d’Éte de Physique Théorique Herbut, I. F., 2008, Phys. Rev. B 78, 205433.
LXIX (Springer, Berlin). Herbut, I. F., V. Juričić, and B. Roy, 2009a, Phys. Rev. B
Girvin, S. M., A. H. MacDonald, and P. M. Platzman, 1986, 79, 085116.
Phys. Rev. B 33, 2481. Herbut, I. F., V. Juričić, and O. Vafek, 2009b, Phys. Rev. B
Giuliani, G. F., and G. Vignale, 2005, Quantum Theory of 80, 075432.
Electron Liquids (Cambridge UP, Cambridge). Hou, C.-Y., C. Chamon, and C. Mudry, 2010, Phys. Rev. B
Goerbig, M. O., 2009, Quantum Hall Effects (Lecture notes of 81, 075427.
the Les Houches Summer School 2009 (Singapore Session), Hou, J.-M., W.-X. Yang, and X.-J. Liu, 2009, Phys. Rev. A
arXiv:0909.1998). 79, 043621.
Goerbig, M. O., B. Douçot, and R. Moessner, 2006, Phys. Hwang, E. H., and S. Das Sarma, 2007, Phys. Rev. B 75,
Rev. B 74, 161407. 205418.
Goerbig, M. O., J.-N. Fuchs, K. Kechedzhi, and V. I. Fal’ko, Imry, Y., 1997, Introduction to Mesoscopic Physics (Oxford
2007, Phys. Rev. Lett. 99, 087402. UP).
Goerbig, M. O., J.-N. Fuchs, G. Montambaux, and Ishikawa, K., and T. Ando, 2006, J. Phys. Soc. Jpn. 75,
F. Piéchon, 2008, Phys. Rev. B 78, 045415. 084713.
Goerbig, M. O., J.-N. Fuchs, G. Montambaux, and Iyengar, A., J. Wang, H. A. Fertig, and L. Brey, 2007, Phys.
F. Piéchon, 2009, EPL 85, 57005. Rev. B 75, 125430.
Goerbig, M. O., P. Lederer, and C. Morais Smith, 2003, Phys. Jackson, J. D., 1999, Classical Electrondynamics (Wiley, 3rd
Rev. B 68, 241302. ed., New York).
Goerbig, M. O., P. Lederer, and C. Morais Smith, 2004, Phys. Jiang, Z., E. A. Henriksen, Y.-J. W. L. C. Tung, M. E.
Rev. B 69, 115327. Schwartz, M. Y. Han, P. Kim, and H. L. Stormer, 2007a,
Goerbig, M. O., and N. Regnault, 2007, Phys. Rev. B 75, Phys. Rev. Lett. 98, 197403.
Jiang, Z., Y. Zhang, H. L. Stormer, and P. Kim, 2007b, Phys.
55
Rev. Lett. 99, 106802. Martin, J., N. Akerman, G. Ulbricht, T. Lohmann, J. H.

Jobst, J., D. Waldmann, F. Speck, R. Hirner, D. K. Maude, Smet, K. von Klitzing, and A. Yacobi, 2008, Nat. Phys.
T. Seyller, and H. B. Weber, 2010, Phys. Rev. B 81, 4, 144.
195434. McClure, J. W., 1956, Phys. Rev. 104, 666.
Juričić, V., I. F. Herbut, and G. W. Semenoff, 2009, Phys. Meyer, J. C., A. K. Geim, M. I. Katsnelson, K. S. Novoselov,
Rev. B 80, 081405(R). T. J. Booth, and S. Roth, 2007, Nature 446, 60.
Kallin, C., and B. I. Halperin, 1984, Phys. Rev. B 30, 5655. Mikitik, G. P., and Y. V. Sharlai, 1999, Phys. Rev. Lett. 82,
Kane, C., and E. J. Mele, 2005, Phys. Rev. Lett. 95, 226801. 2147.
Katayama, S., A. Kobayashi, and Y. Suzumura, 2006, J. Phys. Min, H., J. E. Hill, N. A. Sinitsyn, B. R. Sahu, L. Kleinman,
Soc. Jpn. 75, 054705. and A. H. MacDonald, 2006, Phys. Rev. B 74, 165310.
Katsnelson, M., K. Novoselov, and A. Geim, 2006, Nat. Phys. Moessner, R., and J. T. Chalker, 1996, Phys. Rev. B 54, 5006.
2, 620. Montambaux, G., F. Piéchon, J.-N. Fuchs, and M. O. Goer-
Katsnelson, M. I., 2006, Phys. Rev. B 74, 201401. big, 2009a, Phys. Rev. B 80, 153412.
Kekulé, A., 1865, Bulletin de la Societe Chimique de Paris 3, Montambaux, G., F. Piéchon, J.-N. Fuchs, and M. O. Goer-
(2) 98. big, 2009b, Europhys. J. B 72, 509.
Kekulé, A., 1866, Annalen der Chemie und Pharmazie 137, Moon, K., H. Mori, K. Yang, S. M. Girvin, A. H. MacDonald,
(2) 129. I. Zheng, D. Yoshioka, and S.-C. Zhang, 1995, Phys. Rev.
Kellogg, M., J. P. Eisenstein, and L. N. P. an K. W. West, B 51, 5143.
2004, Phys. Rev. Lett. 036801. Moore, G., and N. Read, 1991, Nucl. Phys. B 360, 362.
Khveshchenko, D. V., 2001, Phys. Rev. Lett. 87, 206401. Morinari, T., T. Himura, and T. Tohyama, 2009, J. Phys.
Khveshchenko, D. V., 2007, Phys. Rev. B 75, 153405. Soc. Jpn. 78, 023704.
Klein, O., 1929, Z. Phys. 53, 157. Mucha-Kruczyński, M., O. Tsyplyatyev, A. Grishin, E. Mc-
v. Klitzing, K., G. Dorda, and M. Pepper, 1980, Phys. Rev. Cann, V. I. Fal’ko, A. Bostwick, and E. Rotenberg, 2008,
Lett. 45, 494. Phys. Rev. B 77, 195403.
Kobayashi, A., S. Katayama, Y.Suzumura, and H. Fukuyama, Nomura, K., and A. H. MacDonald, 2006, Phys. Rev. Lett.
2007, J. Phys. Soc. Jpn. 76, 034711. 96, 256602.
Kohn, W., 1959, Phys. Rev. Lett. 2, 393. Nomura, K., S. Ryu, and D.-H. Lee, 2009, Phys. Rev. Lett.
Kohn, W., 1961, Phys. Rev. 123, 1242. 103, 216801.
Kosterlitz, J. M., and D. J. Thouless, 1973, J. Phys. C 6, Novoselov, K. S., A. K. Geim, S. V. Morosov, D. Jiang, M. I.
1181. Katsnelson, I. V. Grigorieva, S. V. Dubonos, and A. A.
Kotov, V. N., B. Uchoa, and A. H. Castro Neto, 2007, Phys. Firsov, 2005a, Nature 438, 197.
Rev. B 78, 035119. Novoselov, K. S., A. K. Geim, S. V. Morosov, D. Jiang,
Kotov, V. N., B. Uchoa, V. M. Peirera, A. H. Castro Neto, Y. Zhang, S. V. Dubonos, I. V. Grigorieva, and A. A.
and F. Guinea, 2010, preprint , arXiv:1012.3484. Firsov, 2004, Science 306, 666.
Koulakov, A. A., M. M. Fogler, and B. I. Shklovskii, 1996, Novoselov, K. S., D. Jiang, T. Booth, V. V. Khotkevich, S. M.
Phys. Rev. Lett. 76, 499. Morozov, and A. K. Geim, 2005b, PNAS 102, 10451.
Laughlin, R. B., 1983, Phys. Rev. Lett. 50, 1395. Onsager, L., 1952, Philos. Mag. 43, 1006.
Lazzeri, M., and F. Mauri, 2006, Phys. Rev. Lett. 97, 266407. Papić, Z., M. O. Goerbig, and N. Regnault, 2009, Solid State
Lee, C., X. Wei, J. K. Kysar, and J. Hone, 2008, Science 321, Comm. 149, 1056.
385. Papić, Z., M. O. Goerbig, and N. Regnault, 2010, Phys. Rev.
Lee, K. L., B. Grémaud, R. Han, B.-G. Englert, and Lett. 105, 176802.
C. Miniatura, 2009, Phys. Rev. A 80, 043411. Partoens, B., and F. Peeters, 2006, Phys. Rev. B 74, 075407.
Lewis, R. M., Y. Chen, L. W. Engel, D. C. Tsui, L. N. Pfeiffer, Pauling, L., 1960, The Nature of Chemical Bonds (Cornell
and K. W. West, 2005, Phys. Rev. B 71, 081301(R). UP).
Lewis, R. M., Y. Chen, L. W. Engel, D. C. Tsui, P. D. Ye, Pereira, V. M., A. H. Castro Neto, and N. M. R. Peres, 2009,
L. N. Pfeiffer, and K. W. West, 2004, Phys. Rev. Lett. 93, Phys. Rev. B 80, 045401.
176808. Peres, N. M., F. Guinea, and A. H. Castro Neto, 2006, Phys.
Lewis, R. M., P. D. Ye, L. W. Engel, D. C. Tsui, L. N. Pfeiffer, Rev. B 73, 125411.
and K. W. West, 2002, Phys. Rev. Lett. 89, 136804. Peres, N. M. R., 2010, Rev. Mod. Phys. 82, 2673.
Li, G., and E. Andrei, 2007, Nat. Phys. 3, 623. Peres, N. M. R., and E. V. Castro, 2007, J. Phys.:Condensed
Li, G., A. Luican, and E. Y. Andrei, 2009a, Phys. Rev. Lett. Matter 19, 406231.
102, 176804. Pisana, S., M. Lazzeri, C. Casiraghi, K. S. Novoselov, A. K.
Li, G., A. Luican, and E. Y. Andrei, 2009b, Phys. Rev. Lett. Geim, A. C. Ferrari, and F. Mauri, 2007, Nature Mat. 6,
102, 176804. 198.
Lifshitz, I. M., and A. M. Kosevich, 1956, Sov. Phys. JETP Piscanec, S., M. Lazzeri, F. Mauri, A. C. Ferrari, and
2, 636. J. Robertson, 2004, Phys. Rev. Lett. 93, 185503.
Lilly, M. P., K. B. Cooper, J. P. Eisenstein, L. N. Pfeiffer, and Plochocka, P., C. Faugeras, M. Orlita, M. L. Sadowski,
K. W. West, 1999, Phys. Rev. Lett. 82, 394. G. Martinez, M. Potemski, M. O. Goerbig, J.-N. Fuchs,
Lukose, V., R. Shankar, and G. Baskaran, 2007, Phys. Rev. C. Berger, and W. A. de Heer, 2008, Phys. Rev. Lett 100,
Lett. 98, 116802. 087401.
Mahan, G. D., 1993, Many-Particle Physics (Plenum Press, Polini, M., R. Asgari, G. Borghi, Y. Barlas, T. Pereg-Barnea,
2nd Ed., New York). and A. H. MacDonald, 2008, Phys. Rev. B 77, 081411.
Mallet, P., F. Varchon, C. Naud, L. Magaud, C. Berger, and Ponomarenko, L. A., F. Schedin, M. I. Katsnelson, R. Yang,
J.-Y. Veuillen, 2007, Phys. Rev. B 76, 041403(R). E. W. Hill, K. S. Novoselov, and A. K. Geim, 2008, Science
56
320, 356. Weinberg, S., 1995, The Quantum Theory of Fields (Cam-
Poplavskyy, O., M. O. Goerbig, and C. Morais Smith, 2009, bridge UP).
Phys. Rev. B 80, 195414. Wen, X.-G., and A. Zee, 1992a, Phys. Rev. Lett 69, 1811.
Qi, X.-L., and S.-C. Zhang, 2011, Rev. Mod. Phys. 83, 1057. Wen, X.-G., and A. Zee, 1992b, Phys. Rev. B 46, 2290.
Reina, A., X. Jia, J. Ho, D. Nezich, H. Son, V. Bulovic, M. S. Willett, R. L., J. P. Eisenstein, H. L. Stormer, D. C. Tsui,
Dresselhaus, and J. Kong, 2009, Nano Lett. 9, 30. A. C. Gossard, and J. H. English, 1987, Phys. Rev. Lett.
Roldán, R., J.-N. Fuchs, and M. O. Goerbig, 2009, Phys. Rev. 59, 1776.
B 80, 085408. Williams, F. I. B., P. A. Wright, R. G. Clark, E. Y. Andrei,
Roldán, R., M. O. Goerbig, and J.-N. Fuchs, 2010, Semicond. G. Deville, D. C. Glattli, O. Probst, B. Etienne, C. Dorin,
Sci. Technol. 25, 034005. C. T. Foxon, and J. J. Harris, 1991, Phys. Rev. Lett. 66,
Sabio, J., J. Nilsson, and A. H. Castro Neto, 2008, Phys. Rev. 3285.
B 78, 075410. Williams, J. R., D. A. Abanin, L. DiCarlo, L. S. Levitov, and
Sadowski, M. L., G. Martinez, M. Potemski, C. Berger, and C. M. Marcus, 2009, Phys. Rev. B 80, 045408.
W. A. de Heer, 2006, Phys. Rev. Lett. 97, 266405. Wirtz, L., and A. Rubio, 2004, Solid Stat. Comm. 131, 141.
Saito, R., G. Dresselhaus, and M. S. Dresselhaus, 1998, Wojs, A., G. Moller, and N. R. Cooper, 2010, preprint ,
Physical Properties of Carbon Nanotubes (Imperial College arXiv:1007.3006.
Press, London). Wu, X., Y. Hu, M. Ruan, N. K. Madiomanana, J. Hankinson,
Salem, L., 1966, Molecular Orbital Theory of Conjugated Sys- M. Sprinkle, C. Berger, and W. A. de Heer, 2009, Appl.
tems (Benjamin, New York). Phys. Lett. 95, 223108.
Shen, T., J. J. Gu, M. Xu, Y. Q. Wu, M. L. Bolen, M. A. Wunsch, B., F. Sols, and F. Guinea, 2008, New Journal of
Capano, L. W. Engel, and P. D. Ye, 2009, Appl. Phys. Physics 10, 103027.
Lett. 95, 172105. Wunsch, B., T. Stauber, F. Sols, and F. Guinea, 2006, New
Shimshoni, E., H. A. Fertig, and G. Vanketeswara Pai, 2009, Journal of Physics 8, 318.
Phys. Rev. Lett. 102, 206408. Yan, J., Y. Zhang, P. Kim, and A. Pinczuk, 2007, Phys. Rev.
Shizuya, K., 2007, Phys. Rev. B 75, 245417. Lett. 98, 166802.
Shon, N. H., and T. Ando, 1998, J. Phys. Soc. Jpn. 67, 2421. Yang, K., S. Das Sarma, and A. H. MacDonald, 2006, Phys.
Shung, K. W. K., 1986, Phys. Rev. B 34, 979. Rev. B 74, 075423.
Sondhi, S. L., A. Karlhede, S. A. Kivelson, and E. H. Rezayi, Zhang, C.-H., and Y. N. Joglekar, 2007, Phys. Rev. B 75,
1993, Phys. Rev. B 47, 16419. 245414.
Song, Y. J., A. F. Otte, Y. Kuk, Y. Hu, D. B. Torrance, P. N. Zhang, C.-H., and Y. N. Joglekar, 2008, Phys. Rev. B 77,
First, W. A. de Heer, H. Min, S. Adam, M. D. Stiles, and 205426.
A. H. MacDonald, 2010, Nature 467, 185. Zhang, Y., Z. Jiang, J. P. Small, M. S. Purewal, Y.-W. Tan,
Spielman, I. B., J. P. Eisenstein, L. N. Pfeiffer, and K. W. M. Fazlollahi, J. D. Chudow, J. A. Jaszczak, H. L. Stormer,
West, 2000, Phys. Rev. Lett. 84, 5808. and P. Kim, 2006, Phys. Rev. Lett. 96, 136806.
Stern, F., 1967, Phys. Rev. Lett. 18, 546. Zhang, Y., Y.-W. Tan, H. L. Stormer, and P. Kim, 2005,
Tahir, M., and K. Sabeeh, 2008, J. Phys.:Condens. Matter Nature 438, 201.
20, 425202. Zhao, E., and A. Paramekanti, 2006, Phys. Rev. Lett. 97,
Töke, C., and J. K. Jain, 2007, Phys. Rev. B 75, 245440. 230404.
Töke, C., P. E. Lammert, V. H. Crespi, and J. K. Jain, 2006, Zhou, S. Y., G.-H. Gweon, J. Graf, A. V. Fedorov, C. D.
Phys. Rev. B 74, 235417. Spataru, R. D. Diehl, K. Kopelevich, D.-H. Lee, S. G.
Tsui, D. C., H. Störmer, and A. C. Gossard, 1982, Phys. Rev. Louie, and A. Lanzara, 2006, Nat. Phys. 2, 595.
Lett. 48, 1559. Zhu, S.-L., B. Wang, and L.-M. Duan, 2007, Phys. Rev. Lett.
Tutuc, E., M. Shayegan, and D. A. Huse, 2004, Phys. Rev. 98, 260402.
Lett 93, 036802.
Wallace, P. R., 1947, Phys. Rev. 71, 622.

Wa0055

Uploaded by

Copyright:

Available Formats

Wa0055

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

Wa0055

Uploaded by

Copyright:

Available Formats

Electronic Properties of Graphene in a Strong Magnetic Field

(Dated: November 24, 2011)

We review the basic aspects of electrons in graphene (two-dimensional graphite) exposed to a

PACS numbers: 81.05.ue, 73.43.Lp, 73.22.Pr

Contents IV. Magneto-Phonon Resonance in Graphene 31

was the experimental evidence of an unusual quantum

Figure 3 (a) Honeycomb lattice. The vectors δ 1 , δ 2 , and δ 3

and accessible to electronic-transport measurements only

benzene. C. Electronic Band Structure of Graphene

and the overlap matrix

yield the electronic bands, and they may be obtained

matrix may then be written as3 tAB k = tγk∗ = (tBA ∗

and one obtains, thus, the secular equation

ηq clearly commutes with the massless 2D Dirac Hamil-

valence which takes into account the two-fold valley degeneracy,

tice components in the spinors when changing the valley.

whereas in quinoid-type graphene b′ = b(1 + ε/2), which

t′nnn = tnnn (1 − 2ε + bε/2d). (58)

The electronic properties of quinoid-type graphene

with (Goerbig et al., 2008)

2 +,n=1 non-zero. Remember that this component corresponds

1 1 which characterizes the filling of the different LLs.

– their consequences, such as the fractional quantum Hall

which is asymptotically correct in the large-n limit, where vF 3tnnn a

D. Landau Levels in Deformed Graphene

and the Fourier components are then simply the

for intra-valley and

2. SU(4) spin-valley symmetric Hamiltonian

The SU(4)-symmetric part of the interaction Hamiltonian (143) finally reads

where the symmetric interaction vertex is

in terms of the reduced density operators

m e−iq·R m′ c†λn,m;ξ,σ cλ′ n′ ,m′ ;ξ,σ ,

(a) (b) level of the random-phase approximation (RPA), which

in terms of the graphene form (a) (b)

may be viewed as the sum of the pseudomomenta asso- (a) (b)

of the pseudomomenta is naturally a constant of motion

where ∆R = R − R′ is the distance between the guiding

to a dielectric constant of ε ≃ 2. (a) (b)

Figure 19 Effective coupling constant α∗G as a function of the bare

(a) (b) (c)

1. Coupling Hamiltonian 2. Hamiltonian in terms of magneto-exciton operators

(a) 22 boson modes read (Ando, 2007b; Goerbig et al., 2007)

A. Electrons in a Single Relativistic Landau Level

(a) (b) describe the density fluctuations inside the LL λn that

which coincides with the scale-invariant zero-field cou- 1X

in terms of the Bessel function J0 (x), such that one

effective interaction potential

which breaks the lattice inversion symmetry and opens a

energy value for arbitrary B for B = 6 T for B = 25 T

tronic spin σ =↑, ↓ (N = 2) or bilayer quantum Hall sys-

one, and one obtains therefore k(N − k) different spin- (a)

wave excitations, in agreement with the above group-

which explores the surface of the Bloch sphere depicted

n = 0, from a valley-pseudospin ferromagnetic state, an additional consideration of the subleading external

(a) (b) state at νG = 0 that is characterized by a bulk gap as-

Du, X., I. Skachko, F. Duerr, A. Luican, and E. Y. Andrei, 241405.

Rev. Lett. 99, 106802. Martin, J., N. Akerman, G. Ulbricht, T. Lohmann, J. H.

You might also like