Physics 141B, Spring 2010

Problem Set #2 Solutions

David Strubbe
1. (Graphene tight-binding model)
a) Let a be the C-C bond length. The lattice vectors,
defined by the diagram
 as 


3 3
in the problem set, are ~a1 = a 3, 0 , ~a2 = a 2 , 2 , and the basis vector


between sublattice A and B is ~ = a 3, 1 . Therefore the vectors to the 3
nearest neighbors of an A atom are
!
!

3 1
3 1
,
,

~1 = a (0, 1) ,
~2 = a
,
~3 = a
(1)
2 2
2 2
The nearest-neighbor vectors from a B atom are obtained by negating the
y-coordinate of each. Note that all the nearest neighbors of A atoms are B
atoms, and vice versa.
Since there are two atoms per unit cell, each with an associated pz orbital
(~r), our Hamiltonian is a 2 2 matrix. It is constructed from the matrix
elements between the wavefunctions on each atom in the unit cell, each of
which has the form (9.5):

1 X i~kR~ j 
~j
~kA (~r) =
e
~r R
N j

1 X i~kR~ j 
~ j ~
~kB (~r) =
e
~r R
(2)
N j
The Hamiltonian then is
D

  E
 
~A H ~A
k

H~k = D   k E
~B H ~A
k

  E
 
~A H ~B
k
 kE

D
  B
B
~ H ~

The diagonal elements are equal by symmetry, so

D   E
D   E
 
 
~kA H ~kA = ~kB H ~kB
1

(3)

(4)

The off-diagonal element is

ikx 23 a

= eiky a + e
=

  E
X ~
 
~kA H ~kB =
eik~m
m


eiky a/2 + e
eiky a/2
!
!

3
iky a/2
iky a
kx a e
e
+ 2cos
2
ikx 23 a

(5)

The off-diagonal
since the Hamiltonian
  Eare complex
D elements
D  conjugates,
E




is Hermitian: ~A H ~B = ~B H ~A , which can also be checked
k
k
k
k
explicitly.
The Hamiltonian is






eiky a + 2cos 23 kx a eiky a/2






eiky a + 2cos 23 kx a eiky a/2

(6)

The bands are the eigenvalues of this matrix, found by the condition


det ~k I H~k = 0
2

!




2
3
iky a/2 
2  iky a
kx a e
+ 2cos
= ~k e



2

(7)

Therefore the solution is



3
 iky a

+ 2cos
~k = e
kx a eiky a/2 


2

!



3


= eiky a/2 + 2cos
kx a 


2

(8)

In the (10) direction (in Cartesian coordinates, not reciprocal-lattice coordinates), ~k = (k, 0), and

"
!
!#



3
3


k = 1 + 2cos
ka  = 1 + 2cos
ka
(9)


2
2

With the aid of Mathematica, we can plot k from the point up to the
zone edge:
2

The x-axis in units of 1/a; the y-axis is in units of , with the energy zero
at . We plot from the point out to the edge of the Brillouin zone in each
direction.
In the (11) direction, ~k = (k, k), and

!



3

 ika/2
ka 
+ 2cos
k = e


2
v
!
!
u



u
3
3
1
t
2
= 1 + 4cos
ka + 4cos
ka cos
ka
(10)
2
2
2

To find the zone edges, we must determine the reciprocal lattice vectors
~b1 , ~b2 , given by the condition ~ai ~bj = 2ij . The volume of the cell V =

|~a1 ~a2 | = 3 2 3 a2 .


2
2
~b1 = 2 R90 ~a2 =
,
V
a 3 3a


~b2 = 2 R90 ~a1 = 0, 4
(11)
V
3a
The Brillouin zone is hexagonal like the Wigner-Seitz cell in real space and
has the same orientation, with the reciprocal-lattice vectors passing through
the center of each edge of the hexagon. With a bit of trigonometry, we see
. A line joining
that in the h10i direction the zone edge is when kx = 3a4
3

4
two corners of the hexagon is ky = 3a 3kx , which meets the (11) line
4
when k = 3a 1+
.
3)
(
b) With two orbitals per unit cell, we have two bands, i.e. the + and - branches
of ~k . With two electrons per unit cell, there should be one electron in each
band, so each is half-filled. The bands are symmetric about , so this must
be the Fermi energy. We see in the (10) direction that there are two points
where ~k = . We can find these points, and see whether there are others,
by solving




0 = eiky a/2 + 2cos


0 = eiky a/2 + 2cos

~k = 0
!

3

kx a 

2
!

3
kx a
2

(12)
 
k a
The imaginary part gives sin y2 = 0, so ky = 0, 2
a , ... The real part




k a
gives cos y2 +2cos 23 kx a = 0. When ky = 0, that is 1+2cos 23 kx a = 0.



cos 23 kx a = 12 . 23 kx a = 2
3 , ... The result is kx = 3a 3 , ky = 0.


3
When ky = 2
= 0, and the solutions are kx =
a , 1 + 2cos
2 kx a

, ky = 2 . These six points are the corners of the hexagonal Brillouin

3a2
a
3
zone and are referred to as the K and K points.

The density of states at the Fermi energy is zero because there are only two
discrete points at the Fermi energy, whose length is zero, and the density of
states is proportional to the length of the curve of k-points at a given energy
(area of a surface, in 3D).
The energy dispersion is linear around the K and K points, as we can show
by differentiating
v
!
!
u

u
k
a
3k
a
3k
a
y
x
x
~k = t1 + 4cos
+ 4cos2
cos
(13)
2
2
2
to get




ky a
3kx a
3kx a
3
sin
+
2cos
cos
a
2
2
2

~k = r





kx
ky a
3kx a
3kx a
2
+
4cos
cos
1 + 4cos
2
2
2

k a

a cos 3k2 x a sin y2

~k = r




ky
ky a
3kx a
3kx a
2
1 + 4cos
+
4cos
cos
2
2
2

(14)

At K and K points, these first derivatives become the indeterminate form

0/0, but the limits are non-zero (complicated to show), demonstrating that
the dispersion is linear. Therefore the band structure has the form of a cone
around these points. A curve of constant energy on a cone is a circle,
 whose
~ ~ 
length (density of states) is proportional to the radius (k kK ). Since
the energy is also proportional to the radius, therefore the DOS is linear in
energy at these points.
2. (Kittel 9.10)
a) In the Hall geometry specified in Figure 6.14, the current is flowing in the
x
direction. Therefore the relation ~j = E~ becomes

Ex
Q
Q1 0
jx
0 = 0 Q1
s
0 Ey
Ez
0
0
1
0

2
1
Ex Q Ey Q
(15)
= 0 Ex Q1 + Ey s
Ez
5

The second line gives

0 = Ex Q1 + Ey s

(16)

Solving,
Ey =

Ex
sQ

(17)

b) The deflection of the charge carriers by the magnetic field changes the effective resistance along the sample.
xy
Ex
jx = xx Ex + xy Ey = xx Ex
sQ




Q1
2
= 0 Q +
Ex = 0 Q2 1 + s1 Ex
sQ
s+1
Ex
= 0 Q2
s

(18)

Compared to Ex = eff
x jx , what we would have if the resistance were not a
tensor, then we have
eff
x =

Q2 s
0 s + 1

(19)

3. (De Haas-Van Alphen effect)

a) The two oscillation frequencies come from two different extremal orbits. As
you can see in Figure 9.29, the small orbit is the neck and the large orbit
is the belly.
b) One oscillation has 2.8 periods in the plot, and the other has 83 periods. The
range in 1/B is (36540 gauss)1 (36784 gauss)1 = 1.82 107 gauss1 .
Therefore the periods
are 2.2 109 gauss1 and 6.5 108 gauss1 . Using


1
2e
S = hc / B , we get areas of 1.5 1015 cm2 (neck) and 4.3 1016 cm2
(belly). These results are close to those quoted in Kittels example on page
249.
c) Kittel gives kF = 1.20 108 cm1 for Au in the free-electron model (Table
6.1). Therefore the area of the orbit is A = kF2 = 4.5 1016 cm2 , not too
far from the belly orbit actually despite a very different path.
4. (Kittel 9.11)
6

a) The Hamiltonian is
2
h
H=
2m

2
2
+
y 2 z 2

1
+
2m



eyB 2

2
ih
x
c

(20)

The Ansatz for the wavefunction is = (y) ei(kx x+kz z) . Inserting into the
Schrodinger equation H = , we have


h2 2 i(kx x+kz z)

2
i(kx x+kz z)
+
e
+ kz e

2m y 2


1
eyB 2 i(kx x+kz z)
e
= ei(kx x+kz z)
(21)
i
h (ikx ) 2
2m
c
Removing the phase factors,
2
h

2m




eyB 2
2
1
2
hkx 2
=
+ kz +
y 2
2m
c

Rearranging and multiplying by -1,

"

 #
h2 2

h2 kz2
1
eyB 2
=0
+

hkx 2
2m y 2
2m
2m
c
"

2 #
 
h2 2

eB 2 chkx
h2 kz2 1

=0
+
m
y
2m y 2
2m
2
mc
eB
Substitute c =

eB
mc

and y0 =

(22)

(23)
(24)

c
hkx
eB :



2 2
h
2 kz2 1
h
2
2
+

m
(y

y
)
=0
0
c
2m y 2
2m
2

(25)

b) Multiplying by -1 again, and moving two terms to the right-hand side, we

have


h2 2 1

h2 kz2
2
2

+ mc (y y0 ) =

(26)
2m y 2
2
2m
This is a harmonic-oscillator equation, with wavefunction and energy =


2 2
kz
, with eigenvalues n = n + 21 hc . Therefore the actual energies
h2m


2 2
kz
are n = n + 12
.
hc + h2m
7