You are on page 1of 8

Curvatronics with bilayer graphene in an effective 4D spacetime

Marco Cariglia
Departamento de Fı́sica, Universidade Federal de Ouro Preto, 35400-000 Ouro Preto MG, Brazil
School of Pharmacy, Physics Unit, Università di Camerino, 62032 - Camerino, Italy and
Dipartimento di Fisica e Astronomia, Università degli studi di Padova, via F. Marzolo 8, 35131 Padova, Italy

Roberto Giambò
School of Science and Technology, Mathematics Division,
University of Camerino, 62032 - Camerino, Italy
arXiv:1611.06254v3 [cond-mat.mes-hall] 28 Apr 2017

Andrea Perali
School of Pharmacy, Physics Unit, University of Camerino, 62032 - Camerino, Italy and
INFN, Sezione di Perugia, 06123 - Perugia, Italy

We show that in AB stacked bilayer graphene low energy excitations around the semimetallic
points are described by massless, four dimensional Dirac fermions. There is an effective reconstruc-
tion of the 4 dimensional spacetime, including in particular the dimension perpendicular to the
sheet, that arises dynamically from the physical graphene sheet and the interactions experienced
by the carriers. The effective spacetime is the Eisenhart-Duval lift of the dynamics experienced
by Galilei invariant Lévy-Leblond spin 21 particles near the Dirac points. We find that changing
the intrinsic curvature of the bilayer sheet induces a change in the energy level of the electronic
bands, switching from a conducting regime for negative curvature to an insulating one when
curvature is positive. In particular, curving graphene bilayers allows opening or closing the
energy gap between conduction and valence bands, a key effect for electronic devices. Thus using
curvature as a tunable parameter opens the way for the beginning of curvatronics in bilayer graphene.

Keywords: Bilayer Graphene, Lévy-Leblond equations, non-relativistic fermions, Eisenhart lift,


curved systems.

I. INTRODUCTION electronic and polaronic properties associated to these ef-


fects have been discussed extensively in Refs.[10–13]. In
the study of superconductors, which also represent con-
The investigation of condensed matter systems, par- densed matter systems with high potential for technolog-
ticularly in the nanometric regime and from the point ical breakthroughs, there is some experimental evidence
of view of new materials with tunable electronic prop- for fractal geometries in cuprates [14–17], however no ex-
erties, has assumed a primary role in physics to under- periment in a curved geometry setting, while a recent
stand many-body systems from first principles and to de- theoretical work investigates the effects of curvature on
vise new technological applications. In the last few years the superconducting pairing in the presence of spin-orbit
a series of advances, both experimental and theoretical, coupling, predicting non-trivial spin-triplet textures of
have made evident that new materials in the quantum the pairs [18]. In the case of carbon nanotubes effects of
regime present a range of phenomena that are well un- curvature on the electronic structure and transport have
derstood using concepts of relativistic particle physics, been studied [19–21].
until recently thought to be removed from practical appli-
cations. Nowadays related activities are relevant topics In this work we extend to bilayer graphene the rela-
in the physics community, and as an important exam- tivistic approach and the methods of effective geometry
ple we cite that of studying massless and massive Dirac for the massless Dirac fermion of monolayer graphene,
fermions in graphene. Here the state of the art is that generalising it by showing that the geometry can include
of using the formalism of 3D quantum field theory on extra dimensions that are related to an effective recon-
curved spacetime to describe electronic properties of the struction of the full ambient spacetime. We show that
material, where the effective curved geometry is origi- there is an effective 4D spacetime, in general curved,
nated by properties of the structure such as interaction where solutions of the massless Dirac equation are in cor-
with a substrate, topological defects such as disclinations respondence to low energy solutions of the original tight-
and dislocations, or ripples [1–9]. Distortions of the bi- binding model, in the continuum limit. This approach
layer graphene lattice, induced for instance by an ap- shows explicitly that even for bilayer graphene there are
plied curvature, have been considered as a mechanism massless excitations, which is of interest on its own, and
to tune and bend the quasiparticle energy dipersion, be- moreover is particularly powerful since properties of the
cause they generate new phonons interacting with con- system can be inferred by known geometrical methods.
duction electrons in a non trivial way. Corresponding For example we show in a simple way how the local 2D
2

curvature affects the local energy density and the elec- tion Hλ = Eλ are
tronic structure. This is a kinematical effect that arises ~vF κ̄ ~vF κ
from the bound motion in a curved space, and is differ- λ1 = 1 , λ2 = , λ3 = σ , λ4 = σ , (2)
ent in nature from the electrical gating effect, that is due E E
to interaction with external fields. This provides a new where κ̄ is the complex conjugate of κ, λj (j = 1, . . . , 4)
mechanism to generate a gap in the energy levels. are the components of the eigenvectors, and the energy
The outline of the paper is the following. In Section II E satisfies the consistency condition
we begin showing that the quasi–free excitations in AB
stacked bilayer graphene obey, in the low energy limit, E 2 − |~vF κ|2 = σγE , σ = ±1 . (3)
the Galilei invariant Lévy-Leblond equations for a spin For each value of τ there are 4 solutions: for σ = ±1
1
2 particle. In Section III we demonstrate that these so- there are two eigenvalues of the energy. The spectrum
lutions can be lifted to solutions of the massless Dirac is valley degenerate, while the spinors (2) are not. We
equation in 4D Minkowski space. Next, we generalize this group the eigenvalues in two families:
geometrical construction first in Section IV by addition
of a transverse, constant magnetic field, and then in Sec-
" r #
(±) iγ γ2 2
tion V, considering a curved 2D sheet of bilayer graphene Ei = ± (−1) + + |~vF κ| , i = 1, 2. (4)
2 4
and discussing its consequences on the electronic proper-
ties. In particular, the energy band gap is evaluated as (±)
a function of the curvature radius. Positive (resp. nega- The E1 bands touch at |κ| = 0 and make bilayer
(±)
tive) curvature of the bilayer graphene is associated with graphene a semi-metal, while the E2 bands are sep-
insulating (resp. metallic) behavior of the system. Pos- arated by a distance 2γ.
sible applications of curvatronics are finally outlined in For low values of κ all bands grow quadratically, which
the concluding Section VI. indicates non-relativistic behavior. In fact as we show in
the rest of this section the low energy excitations satisfy
the Lévy-Leblond equations: non-relativistic, Galilei in-
II. LOW ENERGY SOLUTIONS OF THE variant equations for a spin 12 particle of mass m [24].
EFFECTIVE HAMILTONIAN IN AB BILAYER These were written in 1967 as a non-relativistic limit of
GRAPHENE the Dirac equation, and proved that the g = 2 Landé fac-
tor for the electron is not a relativistic property. In our
The electronic band structure of graphite was stud- case the mass is proportional to the hopping parameter
ied in 1947 using a tight-binding model by Wallace [22]. γ.
Nowadays we know that for graphene there exist pairs of We expand the solutions to leading order in the dimen-
sionless parameter  = ~vFγ |κ| . For the E1 bands
(±)
Dirac points K and K 0 at the corners of the first Brillouin
zone in momentum space, such that excitations with mo- γ γ
mentum sufficiently close to K or K 0 display a linear dis- λ2 = ± + O () , λ3 = ±σ + O () , (5)
~vF κ ~vF κ̄
persion relation, and we call these points ’valleys’. The
distinct honeycomb lattice of graphene decomposes into while for the E2
(±)
bands
two inequivalent A and B triangular lattices and these
excitations are described by a massless, covariant, con- ~vF κ̄ ~vF κ
+ O 3 , + O 3 . (6)
 
tinuum theory for two degrees of freedom in two dimen- λ2 = ± λ3 = ±σ
γ γ
sions, obtained from the Dirac equation in the plane.
The recent reference [23] contains technical details and a The Lévy-Leblond equations are written in terms of two
literature overview. time-dependent spinors χ1 (t), χ2 (t) with two compo-
We start from the low energy limit tight-binding model nents:
Hamiltonian of AB stacked bilayer graphene close to the i~ ∂t χ2 + i~vF Dχ1 = 0 , (7)
K or K 0 points 2mvF
Dχ2 − i χ1 = 0 , (8)

0 ~vF κ 0 γ
 ~
 ~v κ̄ 0 0 0  where D = iσ j kj is the 2–dimensional Dirac operator
HK,K 0 = F , (1)
0 0 0 ~vF κ  in phase space and the σ j are the Pauli matrices in the
γ 0 ~vF κ̄ 0 standard basis. We replaced the speed of light c in the
c
original work with the relevant speed vF ∼ 300 here.
where κ = τ kx +iky is the wave number of the excitation. Solutions of (7), (8) can be obtained from (5) and (6).
τ = ±1 denotes the Hamiltonian relative to the K or K 0 (±)
For E = E1 and τ = 1
point, γ ∼ 0.4eV is the hopping parameter between A1
|~vF κ|2
and B2 sites, while vF ∼ 106 ms−1 is the Fermi velocity χ1 (t) = e∓i ~γ t T
(λ1 , σλ4 ) , (9)
in a graphene monolayer close to the Dirac points. |~vFκ|2
∓i ~γ t T
The eigenvectors associated with the eigenvalue equa- χ2 (t) = e (λ2 , σλ3 ) , (10)
3

2D dynamics. For a massless particle the null cone is


g µν pµ pν = 0, where pµ are the momenta. Since the met-
ric does not depend on v, then the condition pv = mvF
can be imposed. The intersection of this plane with the
null cone yields the non-relativistic parabolic dispersion
relation, as showed in Fig.1. 4D Gamma matrices Γµ
satisfy the algebra {Γµ , Γν } = 2g µν . Spinors in 4D have
4 components, spinors in 2D have 2. We adopt a decom-
position of Dirac matrices suitable for the null form of
the metric:
     
0 I 0 0 σi 0
Γ+ = , Γ− = 2 , Γi = .
0 0 I 0 0 −σi
FIG. 1. The 4D massless Dirac cone p2x + p2y + 2pu pv is cut
by the plane pv = mvF .  The result is the non-relativistic The 4D Dirac operator decomposes as
1
parabola E = 2m p2x + p2y , where E = −vF pu . In our nota-
1 vF  
tion pZ = 2 (pu + pv ), pT = √

2
(pu − pv ). D ∂u
D̂ = , (12)
2∂v −D
(±)
where the mass is given by m = ± 2vγ2 . For E = E2 the where D = σ i ∂i is the 2D Dirac operator. The matrix
F
above is not symmetric under the exchange of the u, v
solution is obtained swapping χ1 ↔ χ2 above, while for
coordinates. The factor of 2 has been chosen in order to
τ = −1 the solutions are generated by the substitution
recover the Lévy-Leblond equations, as discussed below.
κ → −κ̄. Let us recall that λ1 and λ4 are related to
Solutions of the massless Dirac equation in 4D, D̂ Ψ̂ = 0,
components of the wavefunction on two stacked sites of
are compatible with the light-like projection
type A − B where up-down hopping is allowed, while the
λ2 , λ3 components are associated to sites which are not mvF
directly overlapping and for which hopping is negligible. ∂v Ψ̂ = i Ψ̂ , (13)
~
since the Minkowski metric is independent of v. Upon
III. MASSLESS 4D FERMIONS
using this projection one immediately sees that (12) in-
duces the Lévy-Leblond equations (7), (8). Therefore we
reach the important conclusion that the 4D spinor
The massive Lévy-Leblond equations can be obtained
from the massless Dirac equation in a spacetime with
 
mv
i ~F v χ1 (u)
2 extra dimensions [25–28]. The construction is based Ψ̂ = e (14)
χ2 (u)
on the Eisenhart-Duval lift of dynamics, which was first
discussed by Eisenhart [29] in the first half of the pre- satisfies the massless 4–dimensional Dirac equation in flat
vious century, and then independently rediscovered by space. The four possible spinors are in one to one cor-
Duval and collaborators [30, 31]. The lift establishes respondence with the low energy solutions of the tight-
a correspondence between classical, non-relativistic mo- binding model of bilayer graphene close to the Dirac
tions in the presence of a scalar and vector potential and points.
null geodesics in a higher dimensional spacetime, and ex- To conclude this section, we define 4D variables Z, T
tends to quantum mechanics relating the non-relativistic using u = Z+v √ F T , v = Z−v
2
√ F T , such that 2dudv =
2
Schrödinger equation with the higher dimensional Klein- dZ 2 − vF2 dT 2 . To lowest order in  the T, z dependence
Gordon equation, and the Lévy-Leblond with the Dirac i γ Z−vF T

equation. The technique has been successfully used for of our solutions is of the kind e ~vF 2 2 from which
several applications, as for example higher derivative we√ infer a wavelength along the Z direction with value
systems [32] and non-relativistic holography [33–35], in- 4 2π ~vγF = 29 nm, large compared to the real thick-
spired by previous results in holography [36]; see [37] for ness of the bilayer. Therefore the effective description
a review of the associated geometry. adopted here, where an effective flat space appears that
As an example, the trajectory of a classical free point is infinitely extended in all directions, is compatible with
particle with unit mass moving on a plane can be lifted, the real bilayer electronic structure: the wavefunction Ψ̂
using Eisenhart-Duval correspondence, to a null geodesic is not able to resolve the real finite thickness.
of the Minkowski metric in 4D written in double null
coordinates u and v:
IV. TRANSVERSE MAGNETIC FIELD
gµν dxµ dxν = dx2 + dy 2 + 2dudv . (11)
The results obtained so far are not limited to the spe-
The 4D variable u corresponds, in a standard identifi- cial case of free bilayer graphene. The only constraint
cation, with vF t, where t is the time coordinate in the is that the Eisenhart-Duval lift cannot describe external
4

fields that depend on the v direction: a dependence on graphene become


the 2D time variable t is allowed but not one on Z. So
the case of an electric field transverse to the plane, and (±) 2~eBvF2 (n + 1/2) (±) (±)
E1 ∼± , E2 ∼ ±γ + E1 , (18)
hence to the bilayer graphene, cannot be treated in this γ
framework.
On the other hand the solutions of the Lévy-Leblond
In this section we show that our results continue being (±)
valid in the presence of a constant, transverse magnetic equations for the E1 bands are:
~ = B êZ . The tight binding Hamiltonian is ob- √
field B eB !
e∓in m t 2mv
n2~eB
ψn
tained from (1) with the substitution χ1 (t) = √ F
, (19)
eB (n+1)2~eB
  e∓i(n+1) m t 2mvF ψn
eBx T
~k → π = −i~ τ ∂x + i∂y +

, (15) eB eB
χ2 (t) = e∓i(n+1) m t ψn+1 , e∓in m t ψn−1 (20)
~
(±)
arising from the standard definition of covariant momen- while for the E2 bands
~ and in a gauge where the only nonzero com-
tum p~ − eA, √ !
eB (n+1)2~eB
ponent of the gauge potential is Ay = Bx. This Hamil- e∓i(n+1) m t ψn+1
χ1 (t) = √ 2mvF , (21)
tonian applies if the latticeqspacing is much smaller than e
eB
∓in m t n2~eB
2mvF ψn−1
~
the magnetic length lB = eB . We look for solutions to  eB eB
T
the eigenvalue equation in the form λ = eiky y ϕ(x). In the χ2 (t) = e∓in m t ψn , e∓i(n+1) m t ψn . (22)
rest of the section we will identify the operator π with  
its reduction on the ϕ type χ1 (t)
 of spinors, i.e. we will write In the former case for low energy ∼
χ2 (t)
π = −i~ τ ∂x − ky + eBx ~ . The problem reduces to the √ (±)
quantum harmonic oscillator since the rescaled operators (n+ 12 )2~eB ∓i E1 t T
e ~γ (λ1 , σλ4 , λ2 , σλ3 ) , and in the lat-
1
a = √2~eB π = − √i2 (τ ∂ξ + ξ), and similarly for a† satisfy 2mvF

χ1 (t)
 (E2(±) ∓γ )
the Heisenberg algebra. Here ξ = lxB − ky lB is a dimen- ter
T
∼ e∓i ~γ t (λ2 , σλ3 , λ1 , σλ4 ) , which
sionless quantity. We employ the ansatz ϕ1 = ψn (ξ), χ2 (t)
where ψn is the n–th level normalized eigenfunction of are the same relations found in the free case. The mass
the quantum mechanical harmonic oscillator. For non- of the low energy excitations is still given by m = ± 2vγ2 .
F
zero energy we find Therefore, in this section we have demonstrated that the
Lévy-Leblond equations describe the low energy limit of
p ! 12 the electronic spectrum, splitted in Landau levels, of bi-
2n + 1 + γ̃ 2 ± γ̃ 4 + (4n + 2)γ̃ 2 + 1
Ẽ1,2 = ± layer graphene systems in the presence of an external
2 magnetic field perpendicular to the layers.

E
where Ẽ = √2~eBv and similarly for γ, which agrees
F
V. EXTENSION TO CURVED SPACE:
with the formulae reported in the literature, see e.g. [23, CURVATRONICS
Eqn. (2.49)]. In particular for the n = 0 level there is
no gap opening: this effect is the same in the case of the
pseudo-magnetic fields arising from strain, when present. To extend the example of the classical free point par-
In the next section we will show that instead intrinsic ticle examined above, let us consider a generic 2D Rie-
curvature of the surface can open a gap: this underlies mannian space M with metric
the difference between the effects of strain and those of
g = gij (x)dxi dxj ,
curvature. The remaining spinor components are
√ and the classical theory of a particle of mass m and elec-
n+1 √ Ẽ 2 − (n + 1) tric charge e on M, interacting with a scalar potential
ϕ2 = ψn+1 , ϕ3 = n ψn−1 , (16)
Ẽ Ẽ 2 γ̃ V and an electromagnetic field with vector potential Ai ,
Ẽ 2 − (n + 1) both possibly depending on position and time. In this
ϕ4 = ψn . (17) case the Hamiltonian is given by
Ẽγ̃
Now we examine the low energy limit and show that is g ij
is again described by the Lévy-Leblond equations. For H= (pi − eAi )(pj − eAj ) + V .
† † 2m
B = 0 it must be ππ +π 2
π
= |~κ|2 , while for finite B
ππ † +π † π Then Eisenhart-Duval lift is given by the 4D Lorentzian
we have = 2~eB(N + 1/2), where N = a† a is
2 metric
the number operator. Therefore we take the limit B →
0 together with n → +∞ so that 2~eBvF2 (n + 1/2) ∼ 2e 2V
ds2 = gij dq i dq j + Ai dq i du + 2dudv − du2 ,
|~vF κ|2 << γ 2 . In this limit the Landau levels of bilayer mvF mvF2
5

where q i = (x, y), Ai (q, u) and V (q, u) are the vector and
scalar potential. Note that, the external potentials do
not depend on the transverse v direction, as pointed out
above. To see that this is correct one can calculate the
geodesic Hamiltonian from the metric above obtaining
a) R < 0 b) R = 0 c) R > 0
g ij e pv e pv pu pv V
H= (pi − Ai )(pj − Aj ) + + 2 2 p2v .
2m mvF mvF m m vF FIG. 2. Structures of bilayered graphene for different values
of the 2D curvature. R < 0 corresponds to hyperbolic geom-
Setting pv = mvF , and H = 0 for null geodesics we obtain etry, R = 0 to the flat graphene bilayer, R > 0 to spherical
the condition geometry.

vF pu = −H . (23) ER (meV)

400 1.16nm
If we define a new variable t by u = vF t then the equa-
tion above says that H, the generator of time translations 5
300
for the original dynamical system in 2D, can be identified 4

with minus the momentum along the t direction, which 3

justifies identifying the variable t with the time parame- 200 2

ter in 2D. 1 23nm

As a further generalization, g can be considered as a 100 0


10 12 14 16 18 20 22 24
time dependent 2D metric, which implies that gij are also
functions of (q, u). Explicitly time dependent systems 0 lR (nm)
1 2 3 4 5 6 7
have been recently discussed in [38], where it has been
shown that the formalism of the Eisenhart lift is com-
patible with the time dependence, both classically and FIG. 3. Energy band gap between conduction and valence
bands ER as a function of the (constant) curvature radius lR .
quantum mechanically. To generalize the projection of
The arrows indicate the validity range of our approach.
the 4D Dirac operator (12) described above to this case,
the curvature of the Riemannian manifold M must be
taken into account, and this can be accomplished using inducing Landau levels. In particular, there always exists
the covariant spinorial derivatives defined in [28], retriev- a zero Landau level and no gap in the energy spectrum
ing the operator – with a slight correction with respect can be opened in this way. As we are going to show next,
to [28, Eqn. (4.11)] our geometrical analysis shows that the local curvature of
 i the sheet can be used to tune a gap opening: since this
Π ∂u + i ~vVF

D̂ = ~ . (24) effect is different in nature from that of strain, in this
2i mv~ F − ~i Π work as a first step in the description of curved bilayer
graphene we do not include the effects of strain. These
Here Π = σ j Πj and Πj = −i~∇j − eAj is the U (1) co- are important and will be included in a future work. The
variant momentum including the spin connection, and formalism we use has the important advantage of linking
we used the projection (13). It can be seen from (24) directly the curvature of the 2D geometry to the energy
that the massless, 4D Dirac equation D̂Ψ̂ = 0 induces of states. We realize the geometrical description with
the curved version of the Lévy-Leblond equations (7),(8) a locally Minkowski 4D metric, describing soft deforma-
when Ψ̂ is of the form (14). The equations obtained au- tions of the bilayer structure that maintain the first or-
tomatically include the presence of a scalar and vector der structure of the hexagonal cells, without local lattice
potential, and of a curved 2D metric gij . In particular strains or compressions of the bonds within the cells. For
they incorporate the results above discussed for magnetic a non-trivial gij the energy can be considered low if at
fields. Therefore non-relativistic Lévy-Leblond fermions, all points ~2 vF2 R << γ 2 , where R is the Ricci scalar of
arising in the continuum limit of the tight binding theory g.
in bilayer graphene in presence of potentials independent From Eq. (24) one obtains the curved version of the
of the variable v, are described by a massless Dirac equa- Lévy-Leblond equations with a scalar potential. Taking
tion in the effective 4D geometry. Bilayer sheets with two derivatives of χ2 we obtain the Schrödinger equation
constant curvature are represented in Fig.2. In the lit-
g ij ~2
 
erature on monolayer graphene it is known that, using a e~
E− Πi Πj − V + B− R χ2 = 0 , (25)
covariant version of the Dirac equation, one should con- 2m 2m 8m
sider effects of strain of the atomic lattice that produce
pseudo-magnetic effective fields [39–42]. It is also known where E = i~ ∂t . For a surface of constant radius of curva-
that the effects of strain are important, as they induce ture lR then |R| = l22 , and if the R term in (25) is smaller
R
strong effective magnetic fields. The effective magnetic than γ, we obtain lR >> 1.16 nm, in agreement with re-
fields arising from strain modify the energy spectrum by quirements of smooth deformations on scales larger than
6

the hexagonal cell. Experimental values of the energy (25) arises directly from the term
contribution that can be measured by ARPES photoe-
mission spectroscopy are of the order of EA = 10−3 eV : 1
[∇i , ∇j ] = Rijlm Γlm (27)
requiring that the curvature effects are measurable with 4
photoemission results in the constraint lR ≤ 23nm. This
is well within the typical curvature scale of interest in that expresses the non-commutation of (spinorial) covari-
graphene systems [43–45]. The scalar curvature of the ant derivatives in curved space. In other words the gap
surface has been discussed in the context of curved mono- arises from the properties of propagation of fields in the
layer graphene in [46], where it has been associated to an curved space, that are required by covariance and by con-
effective pseudomagnetic field. sistency with the bound motion.
In the literature for monolayer graphene, curvature
The case V = 0, B = 0 can be studied in terms of
is in connection with two opposite effects on electronic
the eigenvalues p of the spinorial momentum operator Π,
p2
states: in an earlier work using a continuum model pos-
proportional to the Dirac operator: then E = 2m . For ex- itive curvature was found to repel charges, and negative
ample in the case of a sphere of radius lR the eigenvalues curvature to attract them [50], while more recent work
are known [47], and the quantized energy is that uses the Dirac equation on a curved background
found that positive curvature conical defects are associ-
~2 vF2 ated to an increase in the DOS, and viceversa for nega-
E = ±n2 n ∈ Z∗ , (26)

lR tive curvature [1, 51]. On the one hand the results of [50]
should provide a refinement, smaller and in the opposite
where no zero-modes of the Dirac operator exist on the direction, of analyses based on the connectivity of single
sphere. The expression is valid for lnR > ~vγF = 1.16nm, sites. On the other hand it is unclear if the results of
and describes the two touching energy bands, as well as [1] are due to the singularity or to the curvature: conical
the departure from γ of the non-touching bands. defects are singular points with zero intrinsic curvature,
and the effects described are very localized, disappear-
Our results imply that for a positive curvature surface ing a few lattice constants away from the defect. Our
the energy of + bands will be shifted higher, while the approach describes long range curvature and therefore is
energy of the − bands will be shifted lower, due to the complementary to that of [51]. In fact it is a second order
sign change in the effective mass. Therefore the shifted effect: the energies in [1] are of the order of Eloc = ~v
lR ,
F
E1 bands will make the bilayer graphene a semiconduc-
while those in our model, using the value of m in (25),
tor with a tunable band–gap proportional to R. In Fig. ~2 v 2
(+) (−)
3 the energy band gap ER = E1 − E1 is plotted are Egeom = 2γl2F . In fact these are the two allowed
R
as a function of the curvature radius lR , for positive combinations of parameters with which one can build
constant curvature. Considering for instance the range an energy. Then our earlier requirement of low energy
3nm< lR < 4nm, Fig.3 shows that the band gap is in ~2 vF2 R << γ 2 implies Egeom << Eloc .
the range ER ' 40 ÷ 60meV , already enough to suppress These results are important in the development of
thermal broadening and thermal excitations across the graphene based curvatronics as they give a powerful tool
bands at or below room temperature. Band gap open- to describe the local effect of curvature on electronic
ing in bilayer graphene of energies of the order shown states in (25). They are also important in the fundamen-
in Fig.3 have been obtained by electric field gating as tal understanding of bilayer graphene and can be applied
measured in Ref.[49], following the earlier prediction of to other 2D materials with massive quasiparticles.
Ref.[48]. On the other hand, negative curvature makes
the material a conductor to leading order. Negative cur-
vature can be applied to 2D semiconducting systems, as VI. CONCLUSIONS
bilayer graphene with a band gap induced by an exter-
nal potential, to reduce or close the band gap, increasing We have shown that the low energy limit of the contin-
progressively their metallic behavior. uum tight-binding model for AB stacked bilayer graphene
The reader might wonder what is the physical reason is given by the Galilei invariant Lévy-Leblond equations.
behind the opening of a gap due to curvature, and if there Using the Eisenhart-Duval lift we proved that the low
is any relation with the previously known phenomenon energy excitations satisfy the massless Dirac equation in
of tunable gap opening by electrical gating. As remarked an effective 4D Lorentzian geometry that reconstructs
in [49], a crucial reason why electrical gating opens a the full space. The parabolic dispersion relations of bi-
gap is that it breaks the inversion symmetry of bilayer layer graphene look conical from a 4D perspective. We
graphene. We have investigated this issue and found presented detailed evidence for free bilayer graphene and
that curvature does not break the inversion symmetry. bilayer graphene with a transverse, constant magnetic
Rather, while electrical gating is a dynamical effect, i.e. field. Application to a curved 2D sheet yields a simple
due to the interaction with an external field, the gap and powerful relationship between the Ricci curvature of
opening that we discuss in this work is kinematical in the surface and the local energy of the excitations, that
nature. The R term present in the Schrödinger equation arises from kinematical effects. The theory models the
7

effect of long range curvature and is complementary, but ferent effects and generate topological transitions [53].
with opposite behavior, to the theory of curvature gen-
erated by local defects. Our results open the way to cur-
vatronics for tuning the electronic properties of graphene ACKNOWLEDGMENTS
systems by local, smooth deformations, in such a way to
allow and control a continuous crossover from metallic to We would like to thank L. Covaci, L. Dell’Anna, M.
semiconducting behavior and viceversa. Our geometrical Doria, C. Duval, P. Horváthy, A. Marcelli, D. Neilson
approach can be also applied to other ultrathin materials and M. Zarenia for useful discussions. M. Cariglia ac-
and tested on naturally curved systems, as fullerens with knowledges CNPq support from project (205029/2014-
their number of carbon atoms controlling the curvature, 0) and FAPEMIG support from project APQ-02164-14.
including fullerens with concentric onion-like structures A. Perali acknowledges financial support from the Uni-
having a spherical bilayer of carbon atoms generating a versity of Camerino under the project FAR “Control
band gap [52]. Geometrical effects are also relevant for and enhancement of superconductivity by engineering
metamaterials with interesting topological properties, in materials at the nanoscale”. We acknowledge the col-
which positive or negative curvature may induce very dif- laboration within the MultiSuper International Network
(http://www.multisuper.org) for exchange of ideas and
suggestions.

[1] M. A. H. Vozmediano, M. I. Katsnelson, F. Guinea, [14] H. Büttner, A. Blumen, Possible explanation for the su-
Gauge fields in graphene, Phys. Rep. 496, 109 (2010). perconducting 240-K phase in the YBaCuO system, Na-
[2] W. Yan et al., Strain and curvature induced evolution ture 329, 700 (1987).
of electronic band structures in twisted graphene bilayer, [15] M. Fratini, N. Poccia, A. Ricci, G. Campi, M. Burgham-
Nat. Comm. 4, 2159 (2013). mer, G. Aeppli, A. Bianconi, Scale-free structural organi-
[3] S. Ulstrup et al., Ultrafast Dynamics of Massive Dirac zation of oxygen interstitials in La2 CuO4+y , Nature 466,
Fermions in Bilayer Graphene, Phys. Rev. Lett. 112, 841 (2010).
257401 (2014). [16] J. M. Tranquada, H. Woo, T. G. Perring, H. Goka, G. D.
[4] O. Boada, A. Celi, J. I. Latorre, M. Lewenstein, Dirac Gu, G. Xu, M. Fujita, K. Yamada, Quantum magnetic
equation for cold atoms in artificial curved spacetimes, excitations from stripes in copper-oxide superconductors,
New J. Phys. 13, 035002 (2011). Nature 429, 534 (2004).
[5] N. Szpak, A Sheet of Graphene Quantum Field in a Dis- [17] N. Poccia, A. Ricci, A. Bianconi, Fractal structure
crete Curved Space, in Relativity and Gravitation, edited favouring superconductivity at high temperatures in a
by J. Bičák and T. Ledvinka (Springer International Pub- stack of membranes near a strain quantum critical point,
lishing, 2014) , p.583. J. Supercond. Nov. Magn. 24, 1195 (2011).
[6] V. Atanasov, A. Saxena, Electronic properties of corru- [18] Z.-J. Ying, M. Cuoco, C. Ortix, P. Gentile, Tuning
gated graphene, the Heisenberg principle and wormhole Pairing Amplitude and Spin-Triplet Texture by Curv-
geometry in solid state, J. Phys: Cond. Matt. 23, 175301 ing Superconducting Nanostructures, arXiv:1704.00578
(2011). (2017).
[7] A. Iorio, Graphene: QFT in curved spacetimes close to [19] A. Kleiner, S. Eggert, Curvature, hybridization, and STM
experiments, J. Phys. Conf. Ser. 442, 012056 (2013). images of carbon nanotubes, Phys. Rev. B. 64, 113402
[8] A. Cortijo, M. A. Zubkov, Emergent gravity in the cubic (2001).
tight-binding model of Weyl semimetal in the presence of [20] C. L. Kane, E. J. Mele, Size, shape, and low energy elec-
elastic deformations, Ann. Phys. 366, 45 (2016). tronic structure of carbon nanotubes, Phys. Rev. Lett. 78,
[9] C. G. Beneventano, P. Giacconi, E. M. Santangelo, R. 1932 (1997).
Soldati, Planar QED at finite temperature and density: [21] K. Falk, F. Sedlmeier, L. Joly, R. R. Netz, L. Bocquet,
Hall conductivity, Berrys phases and minimal conductiv- Molecular origin of fast water transport in carbon nan-
ity of graphene, J. Phys. A 42, 275401 (2009). otube membranes: superlubricity versus curvature depen-
[10] Z. Li, L. Covaci, and F. Marsiglio, Impact of Dressel- dent friction, Nano lett. 10, 4067 (2010).
haus versus Rashba spin-orbit coupling on the Holstein [22] P. R. Wallace, The band theory of graphite, Phys. Rev.
polaron, Phys. Rev. B 85, 205112 (2012). 71, 622 (1947).
[11] Z. Li, C. J. Chandler, and F. Marsiglio, Perturbation the- [23] M. Zarenia, Confined States in Mono-and Bi-layer
ory of the mass enhancement for a polaron coupled to Graphene Nanostructures, Ph.D. Thesis, Universiteit
acoustic phonons, Phys. Rev. B 83, 045104 (2011) Antwerpen, 2013.
[12] Z. Li and J. P. Carbotte, Conductivity of Dirac fermions [24] J. M. Lévy-Leblond, Nonrelativistic particles and Wave
with phonon-induced topological crossover, Phys. Rev. B Equations, Comm. Math. Phys. 6, 286 (1967).
88, 195133 (2013) [25] C. Duval, The Dirac & Lévy-Leblond Equations and Ge-
[13] Z. Li and J. P. Carbotte, Electron-phonon correlations on ometric Quantization, in Differential Geometric Methods
spin texture of gapped helical Dirac fermions, Eur. Phys. in Mathematical Physics, edited by P. L. Garcı́a and A.
J. B 88, 87 (2015) Pérez-Rendon (Springer Berlin, Heidelberg, 1987), p.205
8

[26] C. Duval, P. A. Horváthy and L. Palla, Spinor vortices [41] F. de Juan, M. Sturla, M. A. Vozmediano, Space depen-
in nonrelativistic Chern-Simons theory, Phys. Rev. D 52, dent Fermi velocity in strained graphene, Phys. Rev. Lett.
4700 (1995). 108, 227205 (2012).
[27] C. Duval, P. A. Horváthy and L. Palla, Spinors in nonrel- [42] F. de Juan, J. L. Manes, M. A. Vozmediano, Gauge fields
ativistic Chern-Simons electrodynamics, Ann. Phys. 249, from strain in graphene, Phys. Rev. B 87, 165131 (2013).
265 (1996). [43] J. C. Meyer, A. K. Geim, M. I. Katsnelson, K. S.
[28] M. Cariglia, Hidden symmetries of Eisenhart-Duval lift Novoselov, T. J. Booth, S. Roth, The structure of sus-
metrics and the Dirac equation with flux, Phys. Rev. D pended graphene sheets, Nat. 446, 60 (2007).
86, 084050 (2012). [44] A. C. Neto, F. Guinea, N. M. Peres, K. S. Novoselov,
[29] L. P. Eisenhart, Dynamical trajectories and geodesics, A. K. Geim, The electronic properties of graphene, Rev.
Ann. Math. 30, 591 (1928). Mod. Phys. 14, 109 (2009).
[30] C. Duval, G. Burdet, H. P. Künzle and M. Perrin, [45] M. Neek-Amal, P. Xu, J. K. Schoelz, M. L. Ackerman,
Bargmann structures and Newton-Cartan theory, Phys. S. D. Barber, P. M. Thibado, A. Sadeghi, F. M. Peeters,
Rev. D 31, 1841 (1985). Thermal mirror buckling in freestanding graphene locally
[31] C. Duval, G. W. Gibbons, P. Horváthy, Celestial mechan- controlled by scanning tunnelling microscopy, Nat. Com-
ics, conformal structures and gravitational waves, Phys. mun. 5, 4962 (2014).
Rev. D 43, 3907 (1991). [46] E. Arias, A. R. Hernández, C. Lewenkopf, Gauge fields
[32] A. Galajinsky, I. Masterov, Eisenhart lift for higher in graphene with nonuniform elastic deformations: A
derivative systems, arXiv:1611.04294. quantum field theory approach, Phys. Rev. B 92, 245110
[33] K. Jensen, A. Karch, Revisiting non-relativistic limits, (2015).
JHEP 2015.4, 1 (2015). [47] J. C. Várilly, An Introduction to Noncommutative Geom-
[34] M. Geracie, K. Prabhu, M. M. Roberts, Curved non- etry (European Mathematical Society, 2006).
relativistic spacetimes, Newtonian gravitation and mas- [48] H. Min, B. Sahu, S.K. Banerjee and A.H. MacDonald, Ab
sive matter, J. Math. Phys. 56, 103505 (2015). initio theory of gate induced gaps in graphene bilayers,
[35] X. Bekaert, K. Morand, Connections and dynamical tra- Phys. Rev. B 75 155115 (2007).
jectories in generalised Newton-Cartan gravity I. An in- [49] Y. Zhang, T.-T. Tang, C. Girit, Z. Hao, M.C. Martin, A.
trinsic view, J. Math. Phys. 57, 022507 (2016). Zettl, M.F. Crommie, Y.R. Shen, F. Wang, Direct obser-
[36] R. A. Davison, K. Schalm, J. Zaanen, Holographic duality vation of a widely tunable bandgap in bilayer graphene,
and the resistivity of strange metals, Phys. Rev. B 89, Nature 459, 820 (2009).
245116 (2014). [50] S. Azevedo, C. Furtado, F. Moraes, Charge localization
[37] M. Cariglia, Hidden symmetries of dynamics in classical around disclinations in monolayer graphite, Phys. Stat.
and quantum physics, Rev. Mod. Phys. 86, 1283 (2014). Sol.(b) 207, 387 (1998).
[38] M. Cariglia, C. Duval, G. W. Gibbons, P. A. Horváthy, [51] A. Cortijo, M. A. Vozmediano, Electronic properties
Eisenhart lifts and symmetries of time-dependent sys- of curved graphene sheets, Europhys. Lett. 77, 47002
tems, Ann. Phys. 373, 631 (2016). (2007).
[39] F. de Juan, A. Cortijo, M. A. Vozmediano, Charge in- [52] R. Pincak, V. V. Shunaev, J. Smotlacha, M. M.
homogeneities due to smooth ripples in graphene sheets, Slepchenkov, and O. E. Glukhova, Electronic Properties
Phys. Rev. B 76, 165409 (2007). of Bilayer Fullerene Onions, arXiv:1612.01415
[40] F. Guinea, M. I. Katsnelson, A. K. Geim, Energy gaps [53] H. N. S. Krishnamoorthy, Z. Jacob, E. Narimanov, I
and a zero-field quantum Hall effect in graphene by strain Kretzschmar, V. M. Menon, Topological Transitions in
engineering, Nat. Phys. 6, 30 (2010). Metamaterials, Science 336, 205 (2012).