Академический Документы
Профессиональный Документы
Культура Документы
Franz Utermohlen
Contents
1 Introduction 2
2 Tight-binding Hamiltonian 2
2.1 Energy bands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Hamiltonian in terms of Pauli matrices . . . . . . . . . . . . . . . . . . . . . 4
2.3 σz term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1
1 Introduction
The unit cell of graphene’s lattice consists of two different types of sites, which we will call
A sites and B sites (see Fig. 1).
Figure 1: Honeycomb lattice and its Brillouin zone. Left: lattice structure of graphene, made out
of two interpenetrating triangular lattices a1 and a2 are the lattice unit vectors, and δi , i = 1, 2, 3
are the nearest-neighbor vectors. Right: corresponding Brillouin zone. The Dirac cones are located
at the K and K’ points.
Note: This figure and caption are from Castro et al. [2].
2 Tight-binding Hamiltonian
Considering only nearest-neighbor hopping, the tight-binding Hamiltonian for graphene
is X †
Ĥ = −t (âi b̂j + b̂†j âi ) , (2)
hiji
2
where i (j) labels sites in sublattice A (B), the fermionic operator â†i (âi ) creates (annihilates)
an electron at the A site whose position is ri , and similarly for b̂†j , b̂j . We can rewrite the
sum over nearest neighbors as
X † XX †
(âi b̂j + b̂†j âi ) = (âi b̂i+δ + b̂†i+δ âi ) , (3)
hiji i∈A δ
where the sum over δ is carried out over the nearest-neighbor vectors δ1 , δ2 , and δ3 , and the
operator b̂i+δ annihilates a fermion at the B site whose position is ri + δ. Using
1 X ik·ri †
â†i = p e âk , (4)
N/2 k
where N/2 is the number of A sites, and similarly for b̂†i+δ , we can write the tight-binding
Hamiltonian for graphene (Eq. 2) as
t X X i(k−k0 )·ri −ik0 ·δ †
Ĥ = − [e e âk b̂k0 + H.c.]
N/2 i∈A δ,k,k0
X
= −t (e−ik·δ â†k b̂k + H.c.)
δ,k
X
= −t (e−ik·δ â†k b̂k + eik·δ b̂†k âk ) , (5)
δ,k
where
âk
Ψ† = â†k b̂†k ,
Ψ≡ , (8)
b̂k
and
0 ∆k
h(k) ≡ −t (9)
∆∗k 0
is the matrix representation of the Hamiltonian and
X
∆k ≡ eik·δ . (10)
δ
3
2.1 Energy bands
p
The eigenvalues of this matrix are E± = ±t ∆k ∆∗k . We can compute this by writing ∆k
out more explicitly:
∆k = eik·δ1 + eik·δ2 + eik·δ3
= eik·δ3 [1 + eik·(δ1 −δ3 ) + eik·(δ2 −δ3 ) ]
h √ √ i
= e−ikx a 1 + ei3kx a/2 ei 3ky a/2 + ei3kx a/2 e−i 3ky a/2
h √ √ i
−ikx a i3kx a/2 i 3ky a/2 −i 3ky a/2
=e 1+e (e +e )
" √ #
−ikx a i3kx a/2 3
=e 1 + 2e cos ky a . (11)
2
The energy bands are therefore given by
s √ √
3 3 2
3
E± (k) = ±t 1 + 4 cos kx a cos ky a + 4 cos ky a , (12)
2 2 2
or, as it is sometimes written,
p
E± (k) = ±t 3 + f (k) , (13)
where √
√
3 3
f (k) = 2 cos( 3ky a) + 4 cos kx a cos ky a . (14)
2 2
These are two gapless bands that touch at the Dirac points K and K0 (see Fig. 2). In other
words, the Dirac points are the points in k-space for which E± (k) = 0.
Figure 2: Energy bands for graphene from nearest-neighbor interactions. The bands meet at the
Dirac points, at which the energy is zero.
4
in terms of Pauli matrices by expressing ∆k as
X X
∆k = eik·δ = [cos(k · δ) + i sin(k · δ)] , (16)
δ δ
2.3 σz term
it is natural to ask whether we can have also have a σz term. Let’s see what happens when
we do:
X
hM (k, M ) = −t [cos(k · δ)σx − sin(k · δ)σy + M σz ]
δ
X M cos(k · δ) + i sin(k · δ)
= −t . (20)
cos(k · δ) − i sin(k · δ) −M
δ
From the matrix form of the Hamiltonian we can see that this additional σz term raises the
energy of A sites and lowers the energy of B sites, thereby gapping the bands. We can think
of this as being caused by on-site potential terms of the form
!
X X
Ĥpotential = M n̂a,α − n̂b,β , (21)
α β
{A sites} {B sites}
where n̂a,α = â†α âα and n̂b,β = b̂†β b̂β . This could arise when there are two different types of
atoms on the A and B sites, such as in boron nitride, which has the same lattice structure
as graphene, but whose A and B sites correspond to boron atoms and nitrogen atoms.
5
3 Behavior near the Dirac points
3.1 Near K
Let’s look at the behavior of ∆k about the Dirac point K. Defining the relative momentum
q ≡ k − K, we can write ∆k in terms of q as
" √ #
3(K y + q y )a
∆K+q = e−iKx a e−iqx a 1 + 2ei3(Kx +qx )a/2 cos
2
" √ #
π 3a
= e−iKx a e−iqx a 1 − 2e3iaqx /2 cos + qy . (22)
3 2
3.2 Near K0
Similarly, defining the relative momentum q ≡ k − K0 and expanding ∆k about the Dirac
point K0 , we find
3a
∆K0 +q = − (qx − iqy ) , (30)
2
6
so the Hamiltonian about this Dirac point is
0 0 qx − iqy
h(K + q) = vF = vF (qx σx + qy σy ) = vF q · σ . (31)
qx + iqy 0
From the matrix form of the Hamiltonian near the Dirac points (Eqs. 25, 31) we find that
the energy bands near the Dirac points are given by
Near a Dirac point, the dispersion relation is therefore linear in momentum, so the energy
bands form cone (called a Dirac cone; see Fig. 3) with the vertex lying on the Dirac point.
Figure 3: Conical behavior of the bands near a Dirac point, known as a Dirac cone, where the
energy is linear in momentum.
If we add the σz term that we introduced in Section 2.3, the Hamiltonian near the Dirac
points will be of the form
M qx − iqy
hM = vF (qx σx + qy σy + M σz ) = vF . (33)
qx + iqy −M
As we saw, this additional term gaps the energies, so that near the Dirac points, the energies
become q
EM,± = ± qx2 + qy2 + M 2 , (34)
See Fig. 4 for a plot of the band gap near a Dirac point.
7
Figure 4: Energy gap near a Dirac point produced by a σz term. Compare to Fig. 3.
References
[1] Anthony J. Leggett. Lecture 5: Graphene: Electronic band structure and Dirac fermions.
http://web.physics.ucsb.edu/~phys123B/w2015/leggett-lecture.pdf, 2010.