See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303699127

Quantum Hall effect- A brief introduction

Article · December 2015

CITATIONS READS

0 7,045

1 author:

Rishabh Khare
Purdue University
6 PUBLICATIONS   12 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Rishabh Khare on 01 June 2016.

The user has requested enhancement of the downloaded file.

 Quantum Hall Effect

PH5107 Term Paper

Advanced Condensed Matter

Rishabh Khare
11MS034
IISER Kolkata

Autumn 2015
Course Instructor: Dr. Siddhartha Lal
Acknowledgements

I would like to thank Dr. Siddhartha Lal for giving me this wonderful topic to study
and work on it.

This term paper has been mostly worked through Professor Anthony Leggett’s
lectures on Integer Quantum Hall effect and Fractional Quantum Hall effect.

2
Contents

1 Abstract 4

2 Integer Quantum Hall Effect 5

2.1 Landau Levels in symmetric gauge . . . . . . . . . . . . . . . . . . . 5
2.2 Quantized Hall Conductance . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Role of impurities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.4 Edge States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3 Fractional Quantum Hall Effect 11

3.1 Plateaus at fractional ν . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 Laughlin’s wavefunction . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3 Fractional Charges . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 Fractional Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.5 Chern-Simons Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3
1

Abstract

This report is about the Quantum Hall effect, both Integer and Fractional. The
Integer Quantum hall effect is explained in a brief manner with focus on Landau
levels in symmetric or radial gauge, Halperin’s arguement with a Corbino disk and
a little about the immportance of edge states in Quantum Hall systems. The main
focus is the Fractional Quantum Hall effect with a detailed discussion on Laughlin’s
wavefunction. The concept of fractional charges is invoked by giving a hole picture.
The role of Berry phases in particle statistics is discussed briefly with a qualitative
picture of particles obeying fractional statistics. The final part deals with a step
into Chern-Simons field theory devised to explain the fictitous flux attached to the
elctrons.

4
2

Integer Quantum Hall Effect

The integer quantum Hall effect is characterized by plateaus at quantized integer

values in the Hall resistance vs magnetic field plot. The Hall conductance σxy =
ne2 /h and σxx = 0 were observed in the experiment conducted by Von Klitzing in
1980.

Figure 2.1: Von Klitzing et al, 1980

2.1 Landau Levels in symmetric gauge

We take 2D electrons in a magnetic field, this leads to the formation of well known
Landau levels.
Generally, for explaning the Landau level formation, we use Landau gauge. We
will take a different route using symmetric gauge i.e. A = (−y, x, 0) or in the polar
coordinates Ar = 0 and Aθ = 12 rB.

Writing the Schrodinger’s equation for this particular choice of gauge,

h̄2 1 ∂ ie r2 2
 
∂ 1 ∂
− (r ) + 2 ( − B) ψ(r, θ) = Eψ(r, θ)
2m r ∂r ∂r r ∂θ 2 h̄

5
 Figure 2.2: Landau levels

We can see that there is no θ component explicitly and hence the θ direction becomes
free. We can rewrite the wavefunction ψ(r, θ) = eilθ Rl (r) and so

h̄2 1 ∂ ∂Rl
 
1 2 2 2
− (r ) + 2 (l − r /2lm ) Rl (r) = ERl (r)
2m r ∂r ∂r r

It is intuitive that for large l, the wavefunction would be confined to a radius rl

given by
2 1/2
rl ≡ (2 l lm )
2
Also, the area enclosed is 2πl lm i.e. it encloses exactly l quanta of flux.
2
For large l, the derivative term can be approximated as ddrR2l and we can expand
the other term upto second order in orders of r − rl ,

h̄2 d2 Rl 1
 
2 2
− + mωc (r − rl ) Rl (r) = ERl (r)
2m dr2 2

We get a harmonic oscillator in the radial direction and a free particle in the θ
direction.

Figure 2.3: Lowest Landau Level in symmetric gauge

6
 Consider a Corbino disk type geometry with a circular hole punched through it
and a Aharanov-Bohm flux Φ going through it. The Schrodinger equation can be
generalized in the same way as we do for a particle on a ring with AB-flux.

h̄2 1 ∂ iΦ ie r2 2
 
∂ 1 ∂
− (r ) + 2 ( − − B) ψ(r, θ) = Eψ(r, θ)
2m r ∂r ∂r r ∂θ Φ0 2 h̄

For this case, the lth state still enclosed l quanta of flux but we now get spetral
flow by tuning the AB flux.
2 1/2
rl = (2 (l − Φ/Φ0 ) lm )

2.2 Quantized Hall Conductance

It was found that in the case of QHE, the Hall conductance is robust against small
changes if material parameters, geometry, etc. This suggest that the origin of the
effect is topological. To see this, let us start with a thought experiment.

Consider a Corbino disk, with current leads attached to the inner and outer
edges, which has a uniform field B and an AB flux going through it. The total flux
through the hole could be an arbitrary function of the time.

Figure 2.4: Corbino disk with a magnetic field and AB flux

As it was showed earlier, rl will depend on the AB flux Φ. Now, if we slowly

increase the AB flux that will generate an emf V = − dΦ
dt
around the disk. When we
increase Φ to exactly 1 quantum flux Φ0 = h/e(c = 1), each state would have moved
outwards and replaced its nearest neighbour(one state added at the inner edge and
disappeared at the outer edge). If each state contains n electrons, then the result
would be net transport of n electrons from the inner edge to the outer edge. This
means
ne
I=
τ
τ = Φ0 /V

and hence a Hall conductance of ne/Φ0 ≡ ne2 /h which gives IQHE.

7
2.3 Role of impurities
The question remains still unanswered: Why do we get plateaus on the IQHE plot?
The explanation lies in the existence of disorder and hence the localized states.

As the QHE is topological, Halperin took disorder to be localized in the central

part of the Corbino disk.

Figure 2.5: Corbino disk with impurities

Because of the impurities, there are a mixture of localized and extended states
created in the disordered region. The elctrons either get stuck in closed loops(localized
states) or are extended throughout(delocalized states).

Figure 2.6: Contours created by impurities

Due to this, the DOS(density of states) of each Landau level gets smeared out.
There are extended states at the centre(COM) and localized states at the ends.

Figure 2.7: Density of states due to disorder

8
 As we increase the magnetic field B, the spacing between Landau levels increases
and the number of Landau levels below the fermi level starts decreasing. When the
fermi level is in between two Landau levels, there are no delocalized states available
for spectral flow and hence the Hall conductance remains constant. After increasing
B, it starts crossing the extended state region, the conductance starts rising and
ultimately saturates again on the encounter of localized states. This explains the
plateaus in the IQHE plot.

Figure 2.8: Landau level shifting with the change in B

2.4 Edge States

Although the simple example of Corbino disk was able to explain the nature of
IQHE, but it is very difficult to measure the Hall resistance in that setup. Let us
look at a more realistic setup.

Figure 2.9: Four probe measurement

The above setup is used in labs worldwide for measuring the Hall resistance. The
Hall resistance is measure along AB.

Since the edges are not connected together, we would have edge current in the
system with a current of magnitude eµ/h circulating around the edges, with edges
being at a constant potential µ. In a semiclassical way, the edge current can be
thought of due to the electron’s cyclotron motion in the magnetic field. At the
edges the electron hop and carry a current. The direction of the current at the
edges is opposite to each other.

9
 Figure 2.10: Semiclassical picture for edge currents

Now, if we give a potential difference V = µ/e to the edges A and B, there is a

current generated i.e.
 2
e e
I = (∆µA,b − ∆µB,b ) = V
h h

where ∆µA,b = electrochemical potential difference between lead A and bulk, ∆µB,b =
electrochemical potential difference between lead B and bulk.

This has the same form that we calculated using the Corbino disk and tells us
that actually the current in QHE is actually carried by the edge states, while the
bulk remains inert.

10
3

Fractional Quantum Hall Effect

3.1 Plateaus at fractional ν

The puzzling thing that was observed in the experiments done by Stormer, Tsui and
Gossard1 was the presence of plateaus at fractional values of the filling factor(ν).
This puzzle was solved by Laughlin in his seminal paper2 on the theory of Fractional
Quantum Hall effect.

Figure 3.1: Plateaus at fractiona ν

The presence of plateus at fractional values means that there are delocalized
states available even at fractional filling which is counter-intuitive to our earlier
understanding of Integer Quantum Hall effect. We know that the existence of the
impurity potential leads to the formation of the contours that have mountains and
1
Tsui, D. C., Stormer, H. L., & Gossard, A. C. (1982). Two-dimensional magnetotransport in
the extreme quantum limit. Physical Review Letters, 48(22), 1559. Chicago
2
Laughlin, R. B. (1983). Anomalous quantum Hall effect: an incompressible quantum fluid
with fractionally charged excitations. Physical Review Letters, 50(18), 1395.

11
valleys. The closed contours signify that the electron is trapped and the states are
localized whereas the open contours signify delocalized states.

Figure 3.2: Contours in the presence of impurity potential

There must be a competing interaction that fights with the impurity potential
to give extended states. Coulomb interactions do the job here. An electron feels the
sum of Coulomb and Impurity potentials. When the Coulomb potential wins over
impurity potential, we get delocalized states even at fractional ν.

The typical spacing between electrons can be calulated as

√
1 2πl
r0 ' = √
ne ν
The Coulomb interaction can then be calculated as
√
e2 ν e2
U0 = '√ .
4πr0 2π 4πl
Also, the strength of the impurity potential is proportional to the spread in the
Landay levels τ . So, when U0 >> τ , we are in the fractional quantum Hall regime.
√
Since, the magnetic length lm ∝ √1B and hence U0 ∝ B. The spacing between
the landau levels h̄ωc is proportional to B. This means that in the limit of strong
magnetic field, the spacing between the landau levels would be huge and thus, we
would concentrate only on the lowest landau level(LLL).

3.2 Laughlin’s wavefunction

Laughlin gave an ansatz for the ground state wavefunction of the fractional Hall
state.

Y X
Ψ(z1 , z2 , ..., zN ) = (zj − zk )q exp( −|zi |2 /4lm
2
)
j<k i

Although it is an ansatz, it is a very well thought guess. Let us start from the
single electron wavefunction of LLL in symmetric gauge.

12
 Ψl (z) = z l exp(−|z|2 /4lm
2
)

For N-electrons, it is natural to generalize the wavefunction like

Y
Ψ(z1 , z2 , ..., zN ) = f (z1 , z2 , ..., zN ) exp(−|zi |2 /4lm
2
)
i

We have a fermionic system which is spin-symmetric. This means that the spatial
wavefunction has to be antisymmetric.

f (zi , zj ) = −f (zj , zi ) (3.1)

With the Coulomb repulsion making the wavefunction smaller when the elctrons
approach wach other, the function should depend on the relative coordinates of the
electrons.

Y
f (z1 , z2 , ..., zN ) = g(zi − zj ) (3.2)
i<j

The system is also rotationally invariant and hence the total angular momentum
∂
should be a good quantum number. This means that if we operate −ih̄ ∂φ , we should
get a number.

∂
−ih̄ Ψ = MΨ (3.3)
∂φ

If we combine (3.1),(3.2) and(3.3), the only valid choice seems to be

Y
f (z1 , z2 , ..., zN ) = (zi − zj )q
i<j

where q is odd, respecting the antisymmetry. Hence, we can see that Laughlin’s
wavefunction is no shot in the dark, but a well thought argument.

From a qualitative point of view, we can see that q has to be related to the
filling factor. Suppose we expand f in powers of zi . The highest power that we get
is
pN q. So, the radius of the biggest circle that encloses the wavefunction in LLL is
2(N q + 1)lm . The filling factor can then be calculated as the number of electrons
per quantum flux,

2
2πlm N N 1
ν= 2
= '
π2(N q + 1)lm Nq + 1 q

in the thermodynamic limit.

13
 Laughlin gave numerical results to support his theory.

Figure 3.3: Comparison with the exact diagonalization results(Ne = No. of electrons,
NH = Hamiltonian matrix dimension, Ψ0 = Exact GS of Coulomb interacting system
and Ψ3 = Laughlin’s GS for ν = 1/3)

We can see that the agreement is more than 99% and that was an evidence that
Laughlin’s theory is indeed correct.

We still need to see that whether this theory is able to explain the experimental
results by Stormer, Tsui and Gossard. Consider the Corbino disk of our earlier
example with impurity shielded regions and a magnetic flux through it.

Figure 3.4: Corbino disk with impurities

When we change the magnetic flux by q flux quanta, q orbits move from inner
and outer edge. But, since the filling factor is 1/q, only a single elctron moves out
in this process.

e
Current I =
T
qφ0
Voltage V =
T
The Hall conductance is then
I e e2
= =ν
V qφ0 h
The plateaus are again due to the impurities.

14
3.3 Fractional Charges
After seeing the earlier exercise of changing the flux to q quantum fluxes, the natural
question that arises is what would happen if we changed the flux to 1 quantum flux?

Suppose we take a uniform circular disk and apply a magnetic field B such that
filling is exactly ν = 1/q, where q is an odd integer. The groundstate would be ap-
proximately Laughlin’s wavefunction. Now, imagine if we increase B so as to allow
an additional quantum flux in one of the area of the disk.

The maximum power which was related to the number of quantum fluxes enclosed
now increases by one i.e. lmax + 1 = N q + 1. To find the wavefunction of the new
state, we multiple the previous wavefunction by a symmetric term

" N
#
Y
Ψ0 = zi Ψ0 (z1 , z2 , ...., zN )
i

When we defined Laughlin’s GS, one should have noticed that it looked like the
GS favours the origin because of the term in the exponential. But, we can rewrite
that in terms of relative coordinates and a centre of mass

X 1 X
|zi |2 = |zi − zj |2 + N |Z|2
i
N i<j

This means that there is no favouring of origin and the single particle densities
are uniform. But, after multiplying, the new state has zero probability of finding
an electron at the origin. The extra quantum flux added has no electron associated
with it. We have created a hole at the origin!

We could have created a hole at z0 too.

"N #
Y
Ψ0 (z0 ) = (zi − z0 ) Ψ0 (z1 , z2 , ...., zN )
i

The next question to ask is what is the effective charge of the added hole? The
simple but tedious way to calculate this is to calculate the charge density with hole
and subtract the original charge density. But, there is a more elegant way to do this
and we will do the same.

Imagine we have added q such holes at the same point and at the same time we
have increased B so that q quanta of flux are added.

"N #
Y
Ψ0 (z0 ) = (zi − z0 )q Ψ0 (z1 , z2 , ...., zN )
i

Then we add an (N + 1)th electron to the system at z0 with the waveunction

δ(zN +1 − z0 ) exp(−|zN +1 |2 /4lm
2
).

15
 The wavefunction for N+1 electrons can then be calculated by integrating z0 ,
Z YN
N +1
Ψ0 ≡ dz0 (zi − z0 )q Ψ0 (z1 , z2 , ...., zN )δ(zN +1 − z0 ) exp(−|zN +1 |2 /4lm
2
)
i

But this is actually Laughlin’s wavefunction for N+1 electrons

Y N
X +1
Ψ0 (z1 , z2 , ..., zN +1 ) = (zj − zk ) exp(− |zi |2 /4lm )
j<k i=1

This means that total q holes have been neutralised by an electron and hence
the effective charge of a hole is e∗ = − qe . The physical meaning of this expression is
that the probability of finding the electron at the origin gets reduced by 1/q when
a hole is created.

3.4 Fractional Statistics

The probability density of the particle is same even if two particles exchange their
positions.
|ψ(r1 , r2 )|2 = |ψ(r2 , r1 )|2
This means that for any particle, be it fermions, bosons ,etc they would acquire a
phase when they swap positions.
ψ(r1 , r2 ) = eiα ψ(r2 , r1 )

α = 0, for Bosons
α = π, for Fermions

What happens when a two hole swap positions? What is the phase acquired?

Let us look at a simple thought experiment. A particle moves and circles another
particle and in the process acquires a phase.

Figure 3.5: (a) Process for a particle A to follow a path C around a second particle.
In 3D, the path can be lifted off from the plane and thus can be reduced to a
point(gray curves)). (b) Equivalent processus consisting in two successive exchanges
of particles A and B.

Suppose the two particles are holes that are added to the system. The wave-
function looks like
Y
Ψza ,zb = (zi − za )(zi − zb )Ψ0 (z1 , z2 , ..., zN )

16
We can see that the total wavefunction is symmetric with respect to the exchange
of holes at za and zb . Now, the Berry phase collected when the hole moves round
another is given by3
Z
γ(C) = 2π d2 z ρ(z)
C

where C is the contour enclosed and ρ is the charge density.

For this case,
1
γ(C) = 2π(ne S − )
q
We must attribute this extra phase as coming from a fictitious magnetic flux at-
tached to each quasihole. We consider quasiholes as anyonic(0 < α < π) by nature
and so if we want to treat them as bosons, they appear as composite particles with
flux attached to recover their anyonic nature. The total phase acquired comes out
to be 2π/q, so for swapping of particles, the phase is π/q. That is why quasiholes
are anyonic in nature.

We can also notice that there are no anyons in 3 dimensions. The reason is that
in d=3, the other particle can move a complete circle without enclosing the other
particle fully and hence gets a trivial Berry phase of either 0 or 2π. In Figure 3.5(b),
we see that there are two ways to encircle the particle: one gives a phase θ and the
other −θ. For d=3, we can turn over the second sheet upside down and both the
phases will then look identical. This means that

eiθ = e−iθ

The only solutions are 0 or 2π, which means that there are only bosons or fermions
in 3 dimensions.

3.5 Chern-Simons Theory

A field theory was developed to explain the attached flux to the electrons. Such
theories are known as ”Chern-Simons” theories. Lopez and Fradkin4 were first to
point out in 1991 their relevance for the FQHE, followed in 1993 by Halperin, Lee
and Read5 , who studied the compressible state at ν = 1/2.

A brief introduction to the field theory is given below.

The Hamiltonian for an electron in a magnetic field is written in second quantized

form as

H = H0 + Hint
3
D. Yoshioka, The Quantum Hall Effect
4
Lopez, A., & Fradkin, E. (1991). Fractional quantum Hall effect and Chern-Simons gauge
theories. Physical Review B, 44(10), 5246.
5
Halperin, B. I., Lee, P. A., & Read, N. (1993). Theory of the half-filled Landau level. Physical
Review B, 47(12), 7312.

17
where the kinetic term is
[−ih̄∇ + eA(r)]2
Z
H0 = d2 r ψ † (r) ψ(r)
2m
and Hint is the interaction term.

A Chern-Simons transofromation is a singular unity transformation,

d2 r0 θ(r−r0 )ρ(r0 )
R
ψ(r) = e−iφ̄ ψCS (r)

where θ(r) = tan−1 (y/x) is the angle formed by r and x-axis.

We can see that this transformation is singular as θ(r) is not defined at r = 0.

The momentum operator operates on the phase factor of the transformation,

 Z 
−iφ̄ d2 r0 θ(r−r0 )ρ(r0 )
R
2 0 0 0
−ih̄∇ψ(r) = e −ih̄∇ − φ̄h̄∇ d r θ(r − r )ρ(r ) ψCS (r)

The action is same as the one we had in the case of a gauge transformation and we
can define a new gauge field or the Cher-Simons vector potential,
Z
h̄
ACS (r) = − φ̄∇ d2 r0 θ(r − r0 )ρ(r0 )
e
Normally, the curl of a gradient is zero. But due to θ(r) being singular, this is not
the case.

[∇ × ∇θ(r)]z = (∂x ∂y − ∂y ∂x )θ(r) = ∆ ln(r) = 2πδ 2 (r)

The corresponding magnetic field can then be calculated as

Z
h̄ h
BCS = − φ̄ d2 r0 ∇ × ∇θ(r − r0 )ρ(r0 ) = − ρ(r)ez
e e
Two things to notice here are that:
(1) It is intimately connected to the electronic density.
(2) It is a quantum operator, contrary to the usual magnetic field.

Since, the elctronic density does not get affected by this transformation, the
interaction Hamiltonian Hint remains invariant.
†
ρ(r) = ψ † (r)ψ(r) = ψCS (r)ψCS (r)

In the mean field approximation, density operator is replaced by hρ(r)i = nel . The
field B gets renormalized to
h
B → B ∗ = B + hBCS i = B − φ̄nel
e
and in terms of the filling factor,

B → B ∗ = B + hBCS i = B(1 − φ̄ν)

18
 References

[1] Anthony J Leggett. Phys 769: Selected Topics in Condensed Matter Physics,
Lecture 7: The integral quantum Hall effect Summer, 2010

[2] Anthony J Leggett. Phys 769: Selected Topics in Condensed Matter Physics,
Lecture 8: The fractional quantum Hall effect Summer, 2010

[3] D. Yoshioka. The quantum hall effect. Springer, 1998.

[4] Pascal Lederer, Mark Oliver Goerbig. Lecture notes: Introduction to the quan-
tum hall effects, 2006.

19

View publication stats