Quantum Mechanics - I

Lecture-18
Landau Levels and Integer Quantum
Hall Effect
Dipan Kumar Ghosh
Centre for Excellence in Basic Sciences
Kalina, Mumbai 400098
June 6, 2020

1 Introduction
In this lecture we will discuss the classic work of Lev Landau (1908-1968)
on the quantum mechanics of the motion of an electron in the presence of a
magnetic field. The work became all the more important after the discovery
by Klitzing in 1980 of what came to be known as the Quantum Hall Effect
(QHE). We will begin by a brief description of the classical Hamiltonian and
then discuss the quantum mechanical formulation. At the end, we will give
a very brief description of the quantum Hall effect and the application of
the quantum mechanical solution of the electrons in a magnetic field for this
problem.

2 Classical Hamiltonian
We take the magnetic field to be in the −z direction and e, the electron charge
to be negative, so that eB > 0. For the moment we take the electron to be
confined in the x-y plane, the motion in the z-direction can be subsequently
superposed. As the magnetic field does no work on the electron, the speed

1
of the electron does not change and is given by equating the centrifugal force
to the Lorentz force acting on the electron

mv 2 qvB
=
R c
which gives the “cyclotron radius” R to be given by
mvc
R= (1)
qB
The electron moves in a circular path of radius R with a frequency known as
the “cyclotron frequency” given by
eB
ωc = (2)
mc
In the presence of an electromagnetic field defined by the scalar potential φ
~ the classical Hamiltonian is given by
and a vector potential A,

1 eA~
H= (~p − )2 + eφ (3)
2m c
The choice of φ and A ~ are not unique. The problem clearly has translational
symmetry in x and y directions. In addition, it has rotational symmetry for
rotation about the z-axis. However, no single choice of a gauge brings out
all these invariances. A popular choice is the symmetric gage in which
~ = −y î + xĵ
A (4)

which preserves the rotational symmetry but not the translational symme-
try. The Landau gauge, which was originally used by Landau in solving the
problem is
~ = Bxĵ
A (5)
which preserves translational symmetry in the y-direction and the gauge
~ = −By î
A

maintains translational symmetry in the y-direction. [ The gauges differ by

gradient of a scalar function. For instance, the symmetric gauge may be
B
obtained from Landau gauge by adding a terms ∇(xy).] The classical
2
2
Hamiltonian given by (3) can be used to provide solution for the Hamilton’s
equations of motion
∂H
q̇i = (6)
∂pi
∂H
ṗi = − (7)
∂qi

where ~q = {qi } = (x, y, z) and p~ = {pi } = (px , py , pz ). Taking the x−

component, we have
 
∂H 1 eAx
ẋ = = px −
∂px m c
     
∂H e eAx ∂Ax e eAy ∂Ay e eAz ∂Az ∂φ
ṗx = − =− px − − py − − pz − −e
∂x mc c ∂x mc c ∂x mc c ∂x ∂x
!
e eA~ ~
∂A ∂φ
=− p~ − · −e
mc c ∂x ∂x

Similar equations can be obtained for the other components as well and we
have
 
1 eAi
q̇i = pi − (8a)
m c
!
e eA~ ~
∂A ∂φ
ṗi = − p~ − · −e (8b)
mc c ∂qi ∂qi

Taking a second derivative, we get for the acceleration

 
1 e dAi
ai = q̈i = ṗi −
m c dt

Substituting for ṗi we have

!
e eA~ ~ e dAi
∂A ∂φ
mai = p~ − · − −e (9)
mc c ∂qi c dt ∂qi

3
We will rewrite the second term containing the total time derivative of Ai as
follows:
dAx ∂Ax dx ∂Ax dy ∂Ax dz ∂Ax
= + + +
dt ∂x dt ∂y dt ∂z dt ∂t
∂Ax ∂Ax ∂Ax ∂Ax
= vx + vy + vz +
∂x ∂y ∂z ∂t
∂Ax
= ~v · ∇Ax +
∂t
Substituting the above in the expression (9), we can rewrite it as
!
e eA~ ~ e
∂A
mai = p~ − · − (~v · ∇)Ai + eEi (10)
mc c ∂qi c

where we have substituted

~
~ = −∇φ − 1 ∂ A
E
c ∂t
Substituting the expression for velocity from (8a), we get, for the acceleration

e ∂A ~ e
mai = ~v · − (~v · ∇)Ai + eEi (11)
c ∂qi c
Consider the first two terms of the above equation. Consider qi = x

e ∂A ~ e e

∂Ax ∂Ay ∂Az
 
∂Ax ∂Ax ∂Ax

~v · − (~v · ∇)Ax = vx + vy + vz − vx + vy + vz
c ∂x c c ∂x ∂x ∂x ∂x ∂y ∂z
    
e ∂Ay ∂Ax ∂Az ∂Ax
= vy − + vz −
c ∂x ∂y ∂x ∂z
e e ~ x
= [vy Bz − vz By ] = (~v × B)
c c
Thus, the expression (11) for the force on the charged particle in the electro-
magnetic field becomes
e ~ + eE
~
m~a = (~v × B) (12)
c
which is the correct expression for the Lorentz force. We will now go over to
the quantum mechanical formulation of the problem.

4
3 Going Over to Quantum Mechanics - Lan-
dau Levels
Having established that the classical Hamiltonian given by (3) describes the
behaviour of a charged particle in an electromagnetic field, we will be looking
for a solution of the corresponding quantum mechanical Schrödinger equa-
tion. We have !2
∂ψ 1 eA~
i~ = −i~∇ − ψ + eφψ (13)
∂t 2m c

We will choose a symmetric gauge in which A~ = B (−y, x, 0). For the time
2
being we will choose φ = 0. We define new variables
Be
Πx = px + y
2c
Be
Πy = py − x
2c
Πz = pz (14)
r r
c c
It may be checked that Πx and Πy are canonically conjugate be-
Be Be
cause
Be Be Be Be Be
[Πx , Πy ] = [px + y, py − x] = (−[px , x] + [y, py ]) = (i~+i~) = i ~
2c 2c 2c 2c c
Thus r r
c c
[ Πx , Πy ] = i~
Be Be
However Πz commutes with both Πx and Πy . In terms of these variables the
Hamiltonian becomes
Π2 Π2y Π2
H= x + + z (15)
2m 2m 2m
Π2z p2z
It may be observed that the Hamiltonian commutes with = . As
2m 2m
2
p
a result, the eigenstates of the Hamiltonian are also eigenstates of z , the
2m
~2 kz2
latter having energy . The motion in z- direction is that of a free particle.
2m

5
Let us define two new operators
c
b= (Πx + iΠy ) (16a)
2Be~
c
b† = (Πx − iΠy ) (16b)
2Be~
Using the commutation relations of Πx and Πy given above, it is easy to show
that
[b, b† ] = 1 (17)
It may be seen that
c c
bb† = (Πx + iΠy )(Πx − iΠy ) = (Π2 + Π2y − i[Πx , Πy ])
2e~B 2e~B x
c c e~B c 1
= (Π2x + Π2y ) + × = (Π2x + Π2y ) +
2e~B 2e~B c 2e~B 2
Thus the Hamiltonian (15) becomes

Π2 Π2y Π2 p2z
 
2e~B † 1
H= x + + z = bb − +
2m 2m 2m 2mc 2 2m
2
1 p
= ~ωc (b† b + ) + z (18)
2 2m
where
eB
ωc =
mc
is the electron cyclotron frequency. The energy levels, apart from the free
motion in the z-direction are given by those of one dimensional harmonic
oscillator,  
1
En = n + ~ωc (19)
2
These energy levels are known as Landau Levels. The cyclotron motion is
confined in the x-y plane while the z-direction is free. There is something
strange about the above result. We had a two dimensional problem to begin
with. However, we have mapped it to a one dimensional problem, thereby
losing one degree of freedom. The reason could be attributed to the fact
that unlike the case of the harmonic oscillator, each energy level does not
correspond to a unique state, rather there are degeneracies associated with
the energy levels, which we will explore in the next section.

6
A comment on the spin of the electrons is appropriate. In the above dis-
cussion, we have neglected the spin of the electrons. Electrons placed in a
magnetic field have different energies for different values of their spin and
orbital angular momenta. For free electrons, the g− factor is 2. The up spin
electrons are raised in the energy and the down spins lowered, with their
e~
energy difference being B = ~ωc . Thus the energy of up spin electrons
mc
in the n− th level is precisely equal to that of down spin electrons in the
(n + 1)− th level. This does not actually happen in real materials because of
band structure effects, which results in Zeeman splitting being much smaller.
Thus in practice, for GaAs the n = 0 spin up levels will fill up first, followed
by n = 0 spin down and alternating thereafter. This is schematically shown
in Fig. 1.

n=3 gµ B
Β

hω
n=2 gµ B
Β

hω
n=1 gµ B
Β
hω
n=0 gµ B
Β

Figure 1: Landau Energy levels including Zeeman splitting - in practice,

Zeeman splitting is much smaller.

4 The Ground State

We define the ground state of the Landau levels by the action of the ladder
operator Π on the ground state
b | 0i = 0 (20)
Inserting the definition of b in terms of Πx and Πy vide equations (16b)
c
(Πx + iΠy ) | 0i = 0
2Be~
7
Reverting to the momentum and position variables, using (14)

c Be Be
(px + y + ipy − i x) | 0i = 0 (21)
2Be~ 2c 2c
It is convenient at this stage to go over to complex variables z = x + iy, z̄ =
x − iy. This gives the following relations

z + z̄ z − z̄
x= ; y=
2 2i
∂x 1 ∂x 1
= ; =
∂z 2 ∂ z̄ 2
∂y 1 ∂y 1
= ; =−
∂z 2i ∂ z̄ 2i
∂ 1 ∂ 1 ∂ ∂ 1 ∂ 1 ∂
= − i ; = + i
∂z 2 ∂x 2 ∂y ∂ z̄ 2 ∂x 2 ∂y
∂ ∂ ∂ ∂ ∂ ∂
= + ; =i −i
∂x ∂z ∂ z̄ ∂y ∂z ∂ z̄

Using the above variable transformation (21) becomes

r
c Be Be
(px + y + ipy − i x) | 0i = 0
2Be~ 2c 2c
 
∂ ∂ Be
−i~ +~ + (y − ix) ψg (z, z̄) = 0
∂x ∂y 2c
 
∂ ∂ Be
+i + (x + iy) ψg (z, z̄) = 0
∂x ∂y 2c~
 
∂ Be
2 + ψg (z, z̄) = 0 (22)
∂z 2c~

The equation (22) can be solved by an ansatz

eB
ψg (z, z̄) = exp(− z z̄)f (z, z̄) (23)
4c~
Substituting the above ansatz into (22), we get

∂ eB eB eB ∂
ψg (z, z̄) = − z exp(− z z̄)f (z, z̄) + exp(− z z̄) f (z, z̄) = 0
∂ z̄ 4c~ 4c~ 4c~ ∂ z̄

8
Substituting this in (22), we get

eB eB eB ∂ eB
− z exp(− z z̄)f (z, z̄) + exp(− z z̄) f (z, z̄) + zψ(z, z̄) = 0
4c~ 4c~ 4c~ ∂ z̄ 4c~
The first and the last terms of the above cancel, resulting in f (z, z̄) satisfying

∂
f (z, z̄) = 0 (24)
∂ z̄
Thus, f (z, z̄), as a function of z̄ is a constant. Assuming that as a function
of z itself, it is an analytic function of z. For instance, any polynomial
in z would satisfy this equation. This is different from the solution of the
harmonic oscillator, where it is a Hermite polynomial. Take the ground state
to be labeled by an index m, so that the wavefunction ψg (z, z̄) may be written
as
eB
ψg (z, z̄) = z m exp(− | z |2 ) (25)
4c~
Note that the values of m must only be non-negative integers, as negative val-
ues of m would make the function ill behave near origin and fractional values
would make it multi-valued. Thus the ground state has infinite degeneracies
corresponding to different values of m. We will see later that confinement in
the x-y plane gives rise to finite degeneracies.
Recalling that Rthe motion is two dimensional, we can normalize the wave-
function using ψ ∗ (z, z̄)ψ(z, z̄)d2 r = 1. Taking

eB
ψg (z, z̄) = Nm z m exp(− | z |2 )
4c~
we have
Z ∞ Z ∞  m+1
2 2m eB 2 2c~
Nm r exp(− 2πrdr = Nm π m! = 1
−∞ −∞ 2c~ eB

Thus the normalization constant is given by

"  m+1 #−1/2
2c~
Nm = πm! (26)
eB

9
5 Landau Gauge Formulation
While the physics of choosing one gauge or another remains the same, the
intermediate calculations do not remain invariant. Note that the magnetic
field in the z-direction is invariant under rotational symmetry in x-y plane and
has a translational symmetry, the choice of gauges do not. The symmetric
gauge preserves only the rotational invariance. A choice which preserves
translational invariance along the y-direction is “Landau gauge” in which
A~ = (0, Bx, 0). Because of this, we can look for energy eigenstates which
are simultaneous eigenstates of py , i.e which have the form eiky y f (x, ky ). The
Schrödinger equation for this case may be written as
"  2 #
1 eBx
H= p2 + py − + p2z − eφ(z)
2m x c

p2
Apart from py , H also commutes with z − eφ(z). We look for simultaneous
2m
eigenstates for all the three,
 2 
pz
Hψ = Eψ, py ψ = ~ky ψ, − eφ(z) = Ez ψ
2m
We thus rewrite the Schrödinger equation as
"  2 #
1 eBx
p2 + ~ky − ψ = (E − Ez )ψ
2m x c

~ky c
Define x0 = the equation becomes
eB
e2 B 2
 
1 2 2
p + 2 (x − x0 ) ψ = (E − Ez )ψ
2m x c
eB
Using ωc = ,
mc
1  2
px + ωc2 (x − x0 )2 ψ = (E − Ez )ψ

(27)
2m
The above equation represents a one dimensional harmonic oscillator with its
centre at x = x0 . The presence of x0 has no effect on the energy eigenvalues

10
because both x and x − x0 are canonically conjugate to px . Thus The energy
eigenvalues are
1
E = Ez + ~ωc (n + )
2
where n = 0, 1, 2 . . .. the ground state would be
2
ψg = Nm e−α(x−x0 ) (28)

6 Degeneracy of the Ground State

We have seen that the energy is independent of the position of the centre of
the oscillator, thereby making it independent of ky . This leads to degeneracy.
Let us set kz = 0. Consider a two dimensional system of dimension Lx × Ly .
Translational symmetry in the y-direction requires eiky y = 1 i.e
2π
ky = ny
Ly
where ny = 0, 1, 2 . . .. However, ny has to have a maximum because ky has
a maximum value corresponding to the maximum number of states. There
is no translational symmetry in the x-direction. The centres of the cyclotron
orbits are restricted to lie within the sample, i.e 0 ≤ x0 ≤ Lx . Thus
~ky c
0≤ ≤ Lx
eB
Thus the number of degenerate states is
Ly Ly eBLx BA
N = (ny )max = (ky )max = × =
2π 2π ~c ch/e
We define flux quantum Φ0 by
ch
Φ0 = (29)
e
in terms of which the ground state degeneracy works out to
BA
N= (30)
Φ0
Flux quantum is a constant and has a numerical value given by
Φ0 = 4.13 × 10−15 T − m2 = 4.13 × 10−7 G − cm2

11
6.1 Excited States
The excited states of the system are obtained by action of the raising oper-
ators (b† )n on the ground state of the system. The energies would be found
to be given by given by the same expression as for the harmonic oscillator.
It is also easy to show that the degeneracy of each Landau level is the same.

7 Hall Effect
The basic geometry of a Hall effect experiment is shown in Fig. 2. Consider
a thin slab of width w and thickness t. An electric field E~ = Ex î is applied
along the length of the sample. This causes the charge carriers q (assumed
~ = B k̂ is applied
positive) to drift along the x-direction. A magnetic field B
perpendicular to the surface of the sample, causing the charge carrier to
bend along −y direction. This results in an accumulation of positive charges
along one edge and negative charges along the opposite edge. Since charges
cannot move out of the sample, the force on charge carriers due to the electric
field created by the charge separation must exactly balance the Lorentz force
q ~ on the carriers. Thus if the resulting transverse voltage (called the
~v × B
c
Hall voltage) is denoted by VH , we must have
qVH qvB
=
w c

x
z
B

+
+
y +
+
+
+
+ +
+
t

V
H

Figure 2: Geometry of Hall Effect

12
 Thus the Hall voltage is given by
vBw
VH = (31)
c
The Hall resistance is the ratio of the transverse voltage and the current.
In three dimensions, if the density of the charge carriers is n, the current is
the product of the current density qnv with the cross sectional area wt, i.e.
I = qnv(wt), so that
VH B
Rxy = = (32)
Ix cqnt
The Hall resistivity is then obtained by dividing the Hall resistance by ap-
propriate geometrical factors.
However, in two dimensions, the Hall resistance is devoid of any geomet-
rical factor because the current density is current per unit length so that
I = (qvn)w
VH B
Rxy = = (33)
Ix cqn
Because of this reason, in two dimension, Hall resistance and resistivity are
used synonymously. The longitudinal resistivity ρxx however, has a geo-
metrical factor, though, in the absence of scattering ρxx = 0. Thus in the
classical Hall effect, a plot of the Hall resistivity against the magnetic field is
a straight-line of unit slope, as shown in Fig. 3. The non-zero value of ρxx
is due to the finite life time of scattering.

ρ
xy
resistivity

ρ
xx

Magnetic Field B

Figure 3: Variation of Hall resistivity as a function of the magnetic field

strength - classical case

13
7.1 Integral Quantum Hall Effect
Before discussing the Quantum version of Hall effect, we will briefly recollect
some defining relations on resistivity. For a sample in two dimensions, the
conductivity tensor σ is defined by the following relations
jx = σxx Ex + σxy Ey
jy = σyx Ex + σyy Ey
Symmetry properties under time reversal requires σyx = −σxy . We may
also assume, by symmetry, σxx = σyy . The resistivity matrix is obtained by
inverting the matrix for the conductivity
 
1 σxx −σxy
ρ= 2 2
σxx + σxy σxy σxx
Thus
σxx
ρxx = 2 2
σxx
+ σxy
σxy
ρxy =− 2 2
σxx + σxy
Thus though it seems counter intuitive, if σxy 6= 0, then σxx = 0 implies
ρxx = 0, which implies perfect conductivity.
In 1980, K. von Klitzing1 found some interesting new facts about Hall
effect for high magnetic fields. He found that at liquid He temperatures and
at very high magnetic fields (typically ∼ 15 T or more), the Hall resistivity
ρxy does not change for a range of magnetic field strength, and jumps to the
next plateau as the range changes (see Fig. 4). On this plateau, the Hall
resistivity takes the values
h
ρH = ρxy = (34)
νe2
h
where ν is an integer, 1, 2, 3, . . .. The quantity 2 may be thought of as
e
the quantum of resistivity. Klitzing’s suggested this as an experiment for
accurate determination of the fine structure constant e2 /c~. This is known
as the Integer Quantum Hall Effect (IQHE). Klitzing was awarded the Nobel
prize in 1985 for this discovery.
1
K. v. Klitzing, G. Dorda and M. Pepper, Phys. Rev. Letters, 45, 494 (1980)

14
 4

3 ρ
xy
resistivity

R (h/e )
2 ρ

H
xx
1/2

2
1 1/3
1/4
1/5
1/6
0 5 B in Tesla 10 15

Figure 4: Variation of Hall resistivity as a function of the magnetic field

strength - quantum case

Subsequently, Tsui and Störmer also discovered what is known as the

Fractional Quantum Hall Effect where, instead of the integer ν in Eqn. (34)
for the Hall resistivity, one has a fraction f which can take various values,
the most common of which happen to be 1/3 and 2/3.

7.2 Landau Levels and IQHE

A somewhat simplified explanation of the integral quantum Hall effect is
based on considering the system of a two dimensional electron gas (2DEG) in
a strong magnetic field. The discussion requires an understanding of Landau
level degeneracies and the role of chemical potential (at low temperatures,
this is also the Fermi energy of the system). The chemical potential is defined
by
∂F
µ=
∂N
where F is the free energy of the system as a function of the number of
electrons N . Thus it is the change in the free energy of the system when
we add an additional electron to the system. In metallic solids, the energy
available to electrons are very close together; consequently the Fermi energy
is also the energy of the highest occupied level. Clearly, if there is a gap in
the energy levels, when the lower levels (or band in a solid) gets completely
full, the Fermi energy would jump to the higher level as one would require
higher energy to fill up such a level. This is what happens in our system

15
under discussion.
We had seen in (33) that the classical Hall resistivity is given by the expres-
B
sion ρH = , where we have replaced the charge q by the electronic charge
nec
e. If the electron density n is taken to be a value at which ν Landau levels
are filled, then, recalling that the degeneracy of a Landau level per unit area
Be
is given by (see Eqn. (30)) and the fact that each Landau level has the
ch
same degeneracy, , corresponding to a charge density for which exactly ν
Beν
states are filled, the charge density is given by n = . Substituting this
ch
into the expression for the Hall resistivity, we get
h
ρxy = (35)
νe2
When ρxy takes its non-zero constant value on a plateau, ρxx = 0 as well as
σxx = 0. The latter condition implies that no current flows in the longitudinal
direction while the former implies that it behaves like a perfect conductor
with no dissipation of energy. One can understand this result as follows.
When a certain number of Landau levels are just getting filled, there occurs
an energy gap ~ωc in the spectrum. Since the thermal energy kB T  ~ωc ,
the electrons in these levels cannot move to another level in the presence of a
small electric field. This implies ρxx = 0 at the edge of a plateau. However,
this does not explain why ρxx = 0 even when the Landau levels are only
partially full and the electrons are sitting on a plateau. This is explained
on the basis of the existence of disorder in the system which gives rise to
localized states instead of extended states. However, a discussion of this an
other subtle points of the theory is not within the scope of our course.

16