An Introduction to the Quantum Hall Effect

Thomas DeMastri
(Dated: December 2021)
In this paper, the Hall Effect, Integer Quantum Hall (IQH) Effect, and Fractional Quantum Hall
(FQH) Effect are discussed. The Hall Effect is explained in terms of the Drude Model of metals
and IQH states through a band structure of non-interacting electrons. We introduce interactions
between our electrons to explain FQH states of odd-integer fillings.

I. INTRODUCTION
d m
The Hall Effect is a resistance in the transverse direc- p⃗ = f⃗E + f⃗B − p⃗ (2)
dt ρo ne2
tion of the current in a 2D conductor that emerges when a
magnetic field is applied perpendicular to the conductor. with ρo being the resistivity of the material in the ab-
The Integer Quantum Hall (IQH) Effect shows that this sence of magnetic fields and n the density of charge carri-
resistivity is quantized to extraordinary precision in inte- ers. We desire a steady state solution (after enough time
ger multiples of fundamental flux when the conductor is has passed, momentum becomes constant), which leads
brought to very low temperatures. A satisfactory model to the following set of coupled equations:
of IQH states emerges from a non-interacting model of
the electrons in the conductor and a band structure of
these states can be found. The Fractional Quantum Hall e ρo ne2
0 = −eEx − Bpy − px
(FQH) Effect, or the emergence of Hall Resistances quan- m m
tized in fractional multiples of flux quantum, requires
us to treat the interactions between electrons, which is
e ρo ne2
tractable in this paper for a few simple cases. 0 = −eEy + Bpx − py
m m

We make the substitution ⃗j = − ne

mp ⃗ and collect our
II. HALL EFFECT terms in a matrix
B
    
Ex ρo ne jx
In a 2D conductor carrying current with density ⃗j = = B (3)
Ey − ne ρo jy
j⃗x, we define the x̂ direction as the direction of the elec-
tromotive force and ŷ transverse to this. We can define then the resistivity tensor:
An experiment conducted by Edwin Hall to determine
whether a magnetic field acts on the whole of such a con- B
   
ρxx ρyx ρo ne
ductor or just on the current itself (this was 20 years = B (4)
ρxy ρyy − ne ρo
prior to the discovery of the electron) gave rise to what
we now call the Hall Effect. Hall supposed that a mag- Note in particular that ρxy = −ρyx (ρxy is called the Hall
netic field perpendicular to a conductor would cause cur- resistivity). One reads easily that:
rent to collect on the edges of the surface, leading to a
transverse field Ey emerging [1]. This field after a large
time has passed negates the transverse current emerging −B
ρxy = = RH B (5)
from the magnetic field and produces a transverse resis- ne
tivity which depends on the strength of the field applied.
Where RH = −1 ne , commonly known as the Hall Coeffi-
This resistance is well motivated by the Drude theory
cient, is a function only of the density of charge carriers
of metals and well documented by experiment (following
in the material. As B can be fixed and ρxy may be mea-
discussion based on [2]).
sured to high precision, whether values of RH measured
If we apply a magnetic field in the ẑ direction, electrons
in this manner agree with theoretical expectations is an
undergo a force (working in units where c=1 and e is
excellent test of our understanding of metals.
positive):
One may see in Table 1 that the expected value of RH
coincides relatively well with experimental data for ele-
e ments with a small number of valence electrons, but the
f⃗B = − (⃗ ⃗
p × B) (1)
m relation given by Equation 5 breaks down for elements
with a high density of charge carriers. This is somewhat
in addition to the electromotive force f⃗E = −eE ⃗ = expected, as the Drude Model no longer applies in this
−eE x̂, that drives the current. The first order Drude limit.
Model relation for electron momentum gives: Largely however, one sees that in a metal the Drude
 2

Figure 1: Response of ρxy (red) and ρxx (green) to increasing

magnetic field in a GaAs material, sub-Helium temperatures [5]

TABLE 1: Metals that form a cubic lattice have had their Hall
Coefficients measured to high precision. Shown here are the A. Landau Levels and IQH Wavefunctions
values for different groups of the periodic table. Column 1 shows
expected RH in Drude Model, Column 2 the measured value
If we ignore electron-electron interactions, we need
(units of RH in 10−11 m3 C−1 and c = 1) with a low applied field
only one electron wavefunction to solve the system (fol-
at room temperature. Note that Group 1 Metals, for which the
lowing discussion based on [6] and [7]). The Hamilto-
Drude Model applies well at room temperature, have a very good
nian of a charged particle in a magnetic field with c=1 is
experimental agreement with theory [3].
known:

Model approximates decently, applying a magnetic field 1 ⃗ 2 + U (⃗r)

H= (⃗
p + eA) (7)
perpendicular to the conducting surface produces a trans- 2m
verse resistivity linear to the field strength.
We include some potential U (⃗r) since we know that
near the edges of the material, some large potential must
exist to confine our electrons [8]. We deceive ourselves
III. INTEGER QUANTUM HALL EFFECT for now and assume we are far from the edges of our
material, neglecting U (⃗r) for now.
This linearity breaks down in the very low temperature We are free to fix the gauge of our field so long as
regime, giving way to the IQH Effect. ∇×A ⃗ = B ẑ. We choose the Landau Gauge:
In 1980, a group of researchers lead by Klaus von Klitz-
ing published their measurements of the Hall resistivity
⃗ = Bxŷ
A (8)
of an oxidized silicon transistor at very low temperatures
(≈ 1 K, achieved using liquid Helium) and strong mag-
netic fields (>10 Tesla). They found that in this regime, This leads to the Hamiltonian:
the Hall resistivity (the original paper expressed it as the
Hall conductivity, but we continue with our scheme from 1 2 1
section II) is quantized according to: H= px + (py + eBx)2 (9)
2m 2m
We see that while [px , H] ̸= 0, [py , H] = 0, motivating
2πℏ 1 Φo the guess for our wave function:
ρxy = − 2
=− (6)
e n en
where n ∈ Z and we define a fundamental quanta of
ψky = eiky y f (x) (10)
magnetic flux Φo [4]. As one may observe in Figure 1,
increasing the magnetic field strength still leads to higher
Thus, H ψky = E ψky depends only on ky and
Hall resistivity, but the increases are not smoothly linear
x and in particular can be rearranged to be the
and take on a discrete nature.
Schröedinger Equation for a one dimensional quantum
This quantization can be explained through an analyi- harmonic oscillator:
sis of the magnetic field’s interaction with the conductor
when we ignore electron-electron interactions (plausible
in the low temperature limit) as we demonstrate in this 1 2 e2 B 2 ℏ
section. px f (x) + (x + ky )2 f (x) = Ef (x) (11)
2m 2m eB
 3

with the frequency ωL = eB m centered around xo =

− eB
ℏ
ky . Thus, we can immediately say f (x) = ϕn (x−x0 ),
where ϕn is the nth excited wavefunction of the harmonic
potential with frequency ωL . Thus, our wavefuntions
have the form:

ψky ,n = eiky y ϕn (x − x0 ) (12)

and energy spectrum En = ℏωL (n + 21 ).

Our reader may be wondering why we have only con-
sidered the interaction of the magnetic field with elec-
trons in our conductor, and not the electromotive force
Figure 2: Top: An exaggerated example that U (⃗ r) can vary
driving the current in the x̂ direction. One sees that this
between the end points, so long as its variations are small
term contributes a −eEx term to the Hamiltonian. Note
compared with the dimensions of the conductor. The difference in
that py still commutes with this, and one can complete
EF on the left and right stems from an electromotive force
the square to create an analogous relation to Equation
driving current [7]. Bottom: A smoothly varying U (⃗ r) and a |ψ⟩
11. All that this electromotive force amounts to in our
mE worked out in III.a shown near the edge - it is clear that a
wavefunction is a shift in the value of xo = − eB
ℏ
ky − eB 2.
different wavefunction emerges on the edge states [9].
Thus, our wavefunctions take the form of Equation 12
whether this force is considered or not.
This energy spectrum and the associated wavefunc- in their analysis). It is an excellent approximation then
tions are known as Landau Levels. Each Landau Level in to take ∆k = L 2π
, as one may approximate U (⃗r) as an
the absence of an electromotive force may be interpreted y
infinite square well potential of width Ly , for the pur-
as a flat band structure, meaning that changing the mo-
poses of calculating the degeneracy, given our conditions
mentum does not affect the energy value. Thus, each k
on U (⃗r).
value is degenerate and the degeneracy of these Landau
Levels is of particular importance to understanding the We define the dimensions of our conductor as having
IQH Effect. We discuss this in the next subsection. transverse length Ly and longitudinal length Lx . We
discuss ψky ,n in absence of an electromotive force for
the moment, as it only adds more to the algebra while
B. Degeneracy of En revealing no new physics.
We choose our origin x = 0 on the right side of the
material and x = −Lx on the left side. Note that ψky ,n
The degeneracy of a Landau Level is determined by
is highly localized in x, as it falls off exponentially away
how many allowed values of k there are. We may count
Pkmax from xo = − eB ℏ
ky . Thus, any ψky ,n with an xo not
the number of allowed k values using g = kmin 1 in on the interval [−Lx , 0] has almost no probability to be
integral form:
in the conductor. This naturally defines kmin = 0 and
kmax = eB ℏ Lx as the range of allowed ky values.
1
Z kmax We thus read off the degeneracy of one Landau Level
g= dk (13) from Equation 13:
∆k kmin

Calculating this integral requires us to examine the Ly Lx eB Ly Lx B

boundary conditions of the problem that influence k, g= = (14)
2πℏ Φo
namely U (⃗r) from Equation 7. U (⃗r) must satisfy two
conditions (following discussion based on [9])): This result claims that the degeneracy of a Landau
1. As electrons must be confined to the material - thus Level depends only on the external field applied and the
we posit that at the edges of the material, U (⃗r) area of the conductor. Understanding this result is key
must increase rapidly and go towards infinity. to our coming derivation of Equation 6.

2. In the low temperature limit, internal interactions

between electrons are weak. In our model of the C. The Quantization of IQH Flux from Landau
IQH effect, we choose to neglect them altogether. Levels
Thus, U (⃗r) must vary slowly in the interior of the
metal, else the interactions between electrons be-
On inspection, there are features of ψky ,n and its
comes too significant to ignore.
energy spectrum that indicate a ρxy response to B like
Two examples of allowable U (⃗r) are shown in Figure that shown in Figure 1. Firstly, there are large gaps in
2 (one may even take a trigonometric function as [8] do the band spectrum on the order of ωL which would ac-
 4

count for the spikes shown in Figure 1. Secondly, there Though ψky ,n px ψky ,n vanishes for all n. Thus, col-
are many allowed states at each energy level (one for each lecting our terms in a matrix:
allowed ky ), which would account for the “plateaus” in
ρxy . Intuition aside, we now show explicitly that this     
resistivity is quantized as per Equation 6 from our wave- Ex ρ −ρxy Jx
= xx
functions in Equation 12 (following discussion based on Ey ρxy ρyy Jy
[7]).
To discuss the resistivity of this state, we must discuss     
the current density: 1 ρxx −ρxy 0
= eν (23)
0 ρxy ρyy Φ0

ν
e XX d We may thus read off immediately from Equation 23.
J⃗ = I/A = − ψky ,n ⃗x ψky ,n (15) that ρxy = − Φ eν where ν is an integer. Thus, the non-
o
Lx Ly n=1 dt
ky interacting Hamiltonian given by Equation 7 is able to
predict the proper degeneracies, proper energy levels, and
where we assume the first ν Landau Levels are occu- even the proper quantized resistivities of IQH states.
pied. Making the canonical substitution for velocity:

ν X IV. FRACTIONAL QUANTUM HALL EFFECT

e X
J⃗ = − ⃗ ψk ,n
ψky ,n p⃗ + eA y (16)
mLx Ly n=1
ky Just a short two years after the IQH effect was es-
tablished and explained through Landau Levels, research
We are interested in the transverse current density, or teams at Bell Laboratories and Princeton discovered the
the y-component of J.⃗ Using our Landau Gauge once FQH Effect, which had not been predicted or expected
more: by theorists. They measured a Hall resistivity across a
semiconductor cooled to ≈ 0.1 K under an applied mag-
ν X netic field of 30 Tesla of ρxy = −Φ o 1
e 1/3 [10].
e X
Jy = − ψky ,n ℏky + eBx ψky ,n (17) This does not mesh immediately with the scheme we
mLx Ly n=1 developed to explain IQH states. To make matters worse,
ky
this value of ν = 1/3 is not the only extension of our
We know that the expectation value of x for ψky ,n theory we need to make. Since its discovery in 1982,
must be xo : dozens of FQH plateaus have been found, five of which
are plotted in Figure 3 [7, 11].
The scheme we have developed for IQH states relies
−ℏ mEx on the absence of interactions between electrons. If this
eB ψky ,n x ψky ,n = eBxo = eB( ky − ) (18)
eB eB 2 has proved inadequate, the natural next step is to include
interactions between electrons. We follow the derivations
mEx of [7, 9, 12] that produce the ground state of any FQH
eB ψky ,n x ψky ,n = −ℏky − (19) state with ρxy = Φeo 1/ν
1
, where ν is an odd integer.
B
Plugging Equation into Equation 17 gives:
A. Symmetric Gauge and Single Particle
ν X
Wavefunctions
X Ex
Jy = e (20)
n=1 ky
B In order to incorporate Coulomb Interactions, a change
in gauge will make calculations easier. We use the same
There is no dependence left in our sum over n or ky , symmetric gauge as Laughlin does in his original deriva-
so the result is: tion: A⃗ = B (xŷ − yx̂). The kinetic energy operator is
2
the Hamiltonian given by Equation 7 if we neglect U (⃗r):
e E 1
Jy = νg = eEx ν (21) 2
Lx Ly B Φo 1 ⃗ 2 = 1 p⃗2 + e (⃗ ⃗ + e A⃗2
T = (⃗
p + eA) p · A) (24)
2m 2m m 2m
We now calculate Jx :
Capitalizing on our choice of gauge:
ν X
e X
Jx = − ψky ,n px ψky ,n (22) 1 2 Be e2 B 2 2
mLx Ly n=1
ky
T = (px +p2y )+ (xpy −ypx )+ (x +y 2 ) (25)
2m 2m 8m
 5

B. Laughlin’s Trial Function

From our single, free-particle wavefunctions, adding in-

teractions between our electrons gives a grand-ensemble
wavefunction:

( N
)
1 X
Ψ(z1 , ..., zn ) = f (z1 , ..., zN ) exp − 2 |zi |2 (28)
4l i=1

where f is antisymmetric under particle exchange as

electrons are fermions.
Solving for f analytically or numerically is not possible.
However, Laugliln was able to guess a trial function that
gives the proper ρxy ∝ ν, ν odd:

( N
)
1 X Y
Figure 3: FQH plateaus can be observed in this graphene sample. ψ(zi ) = exp − 2 |zi |2 (zi − zj )ν (29)
4l i=1 i<j
Note that these are measured in conductivity, the inverse of
resistivity (units e2 /ℏ). Sample was cooled to 0.30 mK., under
field strength of a. 15 T, b. 21.5 T [11] One sees that this form of f (z1 , ...zn ) is perhaps the
simplest guess Laughlin could make that could reason-
ably approximate interactions. We understand that the
This first and last term form a two dimensional quan- interactions are dependent on the distances between par-
tum harmonic oscillator, and the middle term can be ticles, and the displacement between each pair of parti-
Be ˆ cles appears in Equation 29 exactly once. Furthermore,
written as 2m Lz .
it is antisymmetric under particle exchange as ν is odd.
The power law dependency on ν is the only feature that
T = HHO + αLz (26) defies an intuitive explanation. However, this relation
can be shown to have to the properties we expect of FQH
We can thus write our single free-particle wavefunc- states.
tions as |ψmn ⟩, with mℏ being the z projection of the It has been calculated that Laughlin’s trial function
angular momentum of the nth excited state of the har- very closely matches the true ground state energy for
monic potential given by ωL = eB m . We are interested the odd ν case in computational tests at low N (as N
in FQH states with low fillings, so we assume n=0 and becomes large, computational tests become impossible)
suppress this index. [12]. We illustrate briefly how it produces the proper Hall
It can be shown through some algebra that |ψm ⟩ takes resistivity.
the form:
Consider z1 - the leading power of z1 is approximately
νN , which is thus the largest projection of Lz for z1 . In

|z|2
 Section IV.A.,√we discussed
√ that a particle is localized to
m
|ψm ⟩ ∝ z exp − 2 (27) a radius r = 2Lz l = 2νN l, meaning it occupies an
4l area A = πr2 = 2πνN l2 = νN ΦBo .
Where z = x − iy and l is a characteristic length of the Given the area that it occupies, we can calculate the
system 2πl2 = ΦBo (electron is relatively localized within number of states in this Landau Level from Equation 15:
√
a radius 2ml, m here the quantum number, not the
mass). We see intuitively that the Gaussian corresponds
to the HO ground state while z m gives |ψm ⟩ the proper AB
g= = νN (30)
angular momentum. Φo
It is worth noting that this is not a “new” wavefunc-
tion. If we desired, we could make the proper gauge If we were to rework our integral from section II.B., the
transformations and return our single particle eigenstates only value that should change is the value of ∆k picking
to their form in Equation 12, but the form given in Equa- up a scaling by 1/ν. Thus, the degeneracy matches a
tion 27 is much preferred for the coming analysis. Landau Level of a state with ρxy = Φeo 1/ν
1
.
 6

V. CONCLUSION anism of band structures does an excellent job of describ-

ing IQH states, but they can only get you so far in FQH
States. Much of the research done in FQH materials
involves topological orders, which are still not fully un-
The Hall Effect, IQH Effect, and FQH Effect are fas- derstood. Thus, while a comprehensive theory of FQH
cinating examples of how edges can lead to interesting states does not yet exist, that is why it is an exciting field
states in condensed matter. It is excellent that our mech- of study.

[1] E. H. Hall, On a new action of the magnet on electric cur- ture notes on QHE, Cambridge grad QHE class.
rents, American Journal of Mathematics 2, 287 (1879). [8] D. J. Thouless, M. Kohmoto, M. P. Nightingale, and
[2] N. W. Ashcroft and N. D. Mermin, Solid State Physics M. den Nijs, Quantized hall conductance in a two-
(Holt, Rinehart and Winston, New York, 1976). dimensional periodic potential, Phys. Rev. Lett. 49, 405
[3] W. W. Schulz, P. B. Allen, and N. Trivedi, Hall coefficient (1982).
of cubic metals, Phys. Rev. B 45, 10886 (1992). [9] S. M. Girvin, The quantum hall effect: Novel exci-
[4] K. v. Klitzing, G. Dorda, and M. Pepper, New method tations and broken symmetries, https://arxiv.org/
for high-accuracy determination of the fine-structure con- pdf/cond-mat/9907002.pdf (1998), lectures delivered at
stant based on quantized hall resistance, Phys. Rev. Lett. Ecole d’Et e Les Houches, July 1998.
45, 494 (1980). [10] P. Ball, Physicists rewarded for ‘fractional electrons’, Na-
[5] W. Bao, Electrical and mechanical properties of graphene ture 395 (1998).
(2012). [11] Y. Kim et. al, Even denominator fractional quantum hall
[6] K. Sun, Topological insulators part i: Phenomena, states in higher landau levels of graphene, Nature Physics
http://www-personal.umich.edu/~sunkai/teaching/ 15, 154 (2019).
Fall_2013/chapter3.pdf (2013), lecture notes on [12] R. B. Laughlin, Anomalous quantum hall effect: An in-
topological insulators, UMich grad condensed matter compressible quantum fluid with fractionally charged ex-
class. citations, Physical Review Letters 50, 1395 (1983).
[7] D. Tong, The quantum hall effect, http://www.
damtp.cam.ac.uk/user/tong/qhe/one.pdf (2016), lec-