⢉ো㓆ࢢࣂ

• ӶѾᱥćࠓ㚐ᘷ⢟䍞Ĉϋ
֠Ѽࢢࣂ ᡇⲺᇐѿφ㋈ᆆᇼᓜѰ~ 1023/cm3䠅㓝Ⲻ⢟䍞
ѱ㾷ऻᤢര։ૂ⏨։θԛ਀ԁӄര⏨ҁ䰪Ⲻ⏨ᲬᘷȽ⧱⪹ᘷȽࠓ㜬ᘷㅿ
⧁Ӑᝐ
␻঄ཝᆜ⢟⨼㌱ •ćࠓ㚐ᘷ⢟⨼Ĉѱ㾷⹊ガӶѾϋ
ࠓ㚐ᘷ⢟䍞Ⲻ㔉ᶺθ࣑Ƚ✣Ƚ༦Ƚ‫ݿ‬Ƚ⭫Ƚ⻷ㅿ⧦䊗θԛ਀▒൞Ⲻᓊ⭞

•ćࠓ㚐ᘷ⢟⨼Ĉ⹊ガⲺ⢯㢨ϋ
ѦሂཐṭⲺ⹊ガᯯੇȽ⨼䇰фᇔ僂Ⲻ㍝ᇼ㔉ਾȽфᢶᵥᓊ⭞ᇼ࠽⴮ީ

• ᴿཐቇćࠓ㚐ᘷ⢟⨼Ĉᆜᇬϋ
ཝ㓜1/3Ⲻ⢟⨼ᆜᇬᐛ֒൞ࠓ㚐ᘷ⢟⨼亼ตθ⧦ԙ⢟⨼ᴶཝⲺ࠼᭥

֠Ѽࢢࣂ ੿सफ़я
ș ӶѾᱥര։ϋ ㆶঋⲺ䈪θᱥޭᴿരᇐ։〥ૂᖘ⣬Ⲻ⢟։ ㅜа⅑䈮˖

ș ྸ֋⭞жѠ⢟⨼ᆜ৸䠅ᶛ㺞ᖷര։ϋ • 䠁኎ѝ⭥ᆀⲴสᵜ㹼Ѫ˖Drude Sommerfeld model

ࢠ࠽⁗䠅δShear ModulusεуѰ䴬Ⲻ⢟։ • ᐳ⍋䎛ᇊ⨶઼㜭ᑖⲴᖒᡀ

• ᲦṬᥟࣘо༠ᆀ ৲㘳Җljപփ⢙⨶ᆖNJˈ哴ᰶǃ丙⊍⩖㪇
ș Ԅ㔉ᶺⲺ䀈ᓜര։࠼Ѱࠖ㊱ϋ
• പփѝⲴ䗃䘀ᙗ䍘ㆰӻ ㅜаㄐǃㅜйㄐǃㅜഋㄐǃㅜӄㄐǃㅜ‫ޝ‬ㄐ
Წ։δޭᴿઞᵕᙝᲬṲ㔉ᶺεȽ䶔Წ։Ƚ߼Წ։

ㅜҼ⅑䈮˖
ș ᡇԢ䇨䘦Ⲻ޻ᇯቶ䲆ӄᲬ։
• ᐘ⻱䱫᭸ᓄ˄2007ᒤ䈪䍍ቄ⢙⨶྆˅
ș ⿱ᆆᖘᡆޭᴿઞᵕᙝⲺᲬṲ
• ᮤᮠ䟿ᆀ䴽ቄ᭸ᓄ˄1985ᒤ䈪䍍ቄ⢙⨶྆˅
ș ᡇԢ⵶䠃䇨䘦ެѣ⭫ᆆⲺ㺂Ѱ
• ࠶ᮠ䟿ᆀ䴽ቄ᭸ᓄ˄1998ᒤ䈪䍍ቄ⢙⨶྆˅

Review of the basic properties of basic properties of normal metals at low T

Electrons and Phonons electrical resistivity magnetic susceptibility heat capacity

U F = ∂M/∂H cv
• Basic physical properties of normal metals

• The Drude model

• Non-interacting Fermi gas model

• Bloch theorem and Band theory T (K) T (K) T (K)

• Basic behavior of phonons U increases with T F ~ const. cv ~ T

Physical picture of electrons in metal The Drude model of electrons in metals

General Hamiltonian of electrons in normal metals: • ignore all interactions (except instantaneous collisions)

• only consider kinetic energy

• electrons were treated as independent classical particles

• Maxwell-Boltzmann distribution

Drude model: two important approximations

e- e-
e- e-
e-
e- e-
1. free electron (ignore interaction with lattice) e- e- e- e-
e- e- e-
e- e-
e-
e- e-
e- e- e-
2. independent electron (ignore interactions between electrons) e-

electrons form ideal gas with charge e

Successes and problems of the Drude model The Sommerfeld Model (1928)
Successes:
• Sommerfeld model is the quantum mechanical version of the
• explanation of electrical conductivity and thermal conductivity
Drude model
• quantitative explanation of the Wiedemann-Franz law (by luck!)
• the Hall effect and magnetoresistance
Sommerfeld model = Drude model + quantum mechanics
• ac conductivity of metals
= Free Electron Gas + Schrodinger equation + Fermi statistics
Problems:
• heat capacity puzzle:
• Sommerfeld kept Drude’s concept of Free Electron Gas, and
Classical particle: u = 3/2 nkBT, cv = 3/2 nkB = const.; experiment: cv ~ T
• The susceptibility puzzle: incorporate quantum mechanics into it.
Classical Curie behavior: F ~ 1/T ; experiment: F does not change with T
• Free Electron Gas model is also called Drude-Sommerfeld Model

Why Drude model fails? How to solve these problems?

Schrodinger equation of Free Electron Gas Electrons Obey Fermi Statistics

Since all electrons are independent, we can solve for electron i:
Fermi distribution function at T = 0
Schrodinger equation:
f (H)

1
Eigenfunction: T = 0:

EF H
Eigenvalue:
EF = μ at T = 0 EF: Fermi Energy

electrons are plane waves with wavevector k traveling in free space

The Fermi Surface Density of state (DOS)

• At T = 0 (the ground state), electrons fill all
DOS: density of states with energy H that are available to be occupied
the states from zero energy till EF

• in momentum space (k-space), all the states

Definition of DOS of a system with volume V:
below kF are filled with both ↑ and ↓ electron

• Fermi Surface (FS): the surface in k space

that separates the occupied and empty states DOS for a 3D free electron gas:
State filling in k-space at T = 0

• in 3D, EF ~ n2/3, n: electron density

• typical values in normal metals:

Cu: EF ~ 7 eV, TF ~ 8h104 K, kF ~ 1/Å, vF ~ 2h108 cm/s DOS:

Fermi Surface in 2D and 1D Fermi distribution function at finite T

f(H)
kBT
1

two dimensional plane one dimensional chain

ky EF H
kF
electronic states spill over EF

kx
kx 0 A particle-hole excitation:
-kF +kF an electron jumps from a
state below EF to an empty
“Fermi surface” at 1D: two points one above it
“Fermi surface” at 2D: a circle
 Linear T heat capacity Pauli paramagnetism
f(H) D(H) Electron magnetic susceptibility:
Electron specific heat:
kBT M ~ H, paramagnetic, F = ∂M/∂H ~ const. (Pauli paramagnetism)
1
cv ~ T H=0 H>0 H>0
H H H

EF EF
EF H

When T raises from 0 to T, μBH

D(H) μBH D(H) D(H)
electrons near EF increase energy by kBT
T (K)
number of such electrons ~ D(EF)kBT
increase of total energy: ∆u ~ kBT(D(EF)kBT) ~ T2 • M = M+ - M- = PB (N+ − N-) = PB · PBH·D+(EF) + PB · PBH·D-(EF) = PB2H·D(EF)
electron specific heat: cv = ∂u/∂T ~ D(EF).T • M ~ H, paramagnetic
• D(EF) is T-independent at low T, so F = μB2D(EF) = const.
• DOS at the Fermi level D(EF) is one of the most important parameters for a metal!

Electrons moving in a periodic potential

Question:

• The Drude-Sommerfeld model tell us how to describe the

electronic properties of metals.

• But we don’t know in the first place why some materials

Single electron
are metals, and some are not. Schrödinger equation: V(r) = V (r+R).

How to solve this complicated equation?

Because we ignored the effect of the lattice!
Bloch in 1928: “when I started to think about it, I felt that the main problem
was to explain how the electrons could sneak by all the ions in a metal …….”

The Bloch Theorem Visualization of a Bloch wave

plane wave: eikr

Schrödinger equation: V(r) = V (r+R).

“By straight Fourier analysis I found to my delight that the wave differs
periodic function uk(r)
from the plane wave of free electrons only by a periodic modulation”

Bloch’s theorem:
The eigenstates of the one-electron Hamiltonian H = -ħ2’2/2m + V(r), where
Bloch wave eikruk(r)
V(r + R) = V(R) for all R in a Bravais lattice, can be chosen to have the form
of a plane wave times a function with the periodicity of the Bravais lattice:

where
Bloch waves are plane waves with a periodic modulation.

Properties of the Bloch wave Properties of the Bloch wave

where where

• Bloch waves are plane waves with a periodic modulation. • For each k there is an infinite family of solutions with discretely
There is no decay term despite interactions with the ions! spaced eigenvalues. Why?

• This can be seen from the periodic boundary condition for u(r). The
• Momentum is no longer a good quantum number, why?
problem is like solving the equation in a fixed volume.

• The wave vector k is not proportional to the electronic • There is an index n, called Band index. \nk and Hnk express the
momentum p ≠ ħk like in free electrons. eigenfunction and eigenvalues for specific k in the nth band.

• For a given n, the eigenstates and eigenvalues are periodic functions

• k is called the crystal momentum, in certain circumstances it
of k in the reciprocal lattice.
has a momentum-like behavior, but in general it is only a
quantum number characteristic of the periodicity of the lattice.
 Formation of energy band Electron states of the Bloch waves
• Since energy levels of Bloch electrons are a family of continuous,
periodic function in k-space, it must form separated band. Why?

Free electrons Electrons in a periodic potential

Hamiltonian

wavefunction

energy ?
• Energy band formation is a natural consequence of quantum
mechanical waves in a periodic lattice. What is the H vs. k relation for Bloch electrons moving in a periodic potential?
This is the electronic dispersion, or electronic structure, or band structure.
• The Hn(k) vs. k relation is called the band structure of the solid. The most important intrinsic (DNA-like) properties of an electronic system.

Nearly Free Electron model Nearly Free Electron model in 1D

• If the periodic potential V(r) is very weak, it can be treated as

a perturbation to the free electron plane waves, i.e., the system
is a Nearly Free Electron gas. Perturbation to the free electron plane wave states:

• The NFE model works well for good metals, such as the V(x) = V(x + na)

group I and II elements, where the electron wavefunctions are

highly-extended. The unperturbed wavefunction and energy levels are:

• The efficient screening of the Coulomb interaction between

itinerant electrons and lattice ions in good metals also reduces
the effective potential.
The first order correction to energy:
• Note that the electron-electron interaction is still ignored.
It is a small constant shift (the average potential) that can be ignored.

Nearly Free Electron model in 1D Nearly Free Electron model in 1D

The 2nd order perturbation is more

complicated:

The matrix element is not zero only when

k’ = k – g, where g = 2nS/a, the reciprocal
lattice vector

• Degenerate perturbation, the mixing of degenerate wavefunctions

For non-degenerate case, i.e., Hk0 ≠ Hk-g0,
it is a small correction ~ |Vg|2 • level above it is pushed up, level below it is pushed down: level repulsion
• splitting of energy levels at k = nS/a, energy gap Eg = 2 |Vn| opens at k = nS/a
For degenerate case, i.e., Hk0 = Hk-g0
(k = nS/a, Brillouin zone boundary), Hf = H0 + V f |Vn|
this term diverges, degenerate • Far from k = nS/a, the E vs. k relation is similar to free electron gas
perturbation has to be used. • An energy gap opens at the Brillouin zone boundary.

Origin of the band gap Origin of the band gap

The standing waves are commensurate with the lattice.
There are two different kinds with opposite phase relative to the lattice.

• The degenerate perturbation is the maximum mixing between electron states

with the same energy.
• Level repulsion induces an energy gap at the Brillouin zone boundary.

We can view the problem from a totally different, phenomenological view.

• The electron waves ψk = eikx get reflected by the periodic lattice as ψ-k = e-ikx.
• When k = nS/a, it satisfies the Bragg diffraction condition.
• There is constructive interference between the reflected waves, and thus form
Coulomb interaction between electron density wave with the lattice
a standing electron wave.
causes the splitting of the energy level.
 Significance of band theory Energy gap and band filling
• Each band can accommodate 2N electrons, N is the number of unit cells
• When the electronic energy is within the gap, no (number of energy level equals the number of unit cells)
Bloch state is allowed
so naively speaking:
• No electronic DOS within the energy gap.
• if each unit cell has an even number of electrons, the solid is an insulator
• If EF lies in the gap, no Fermi surface, it is an
insulator. • if each unit cell has an odd number of electrons, the solid is a metal (half-filling)

• If EF lies within a band, well-defined Fermi • Question: why Be, Mg, Ca, etc. are metals?
surface, it is a metal.

• So the band structure and the existence of band

• Band structure of a metal with 2 valence electrons
gap explains the question we asked: why some
materials are metals and some are not. • Due to band overlap, the bands are partially filled

• This is one of the biggest successes of the • It is a metal!

electronic band theory.

Metal, semiconductor, and insulator Lattice vibrations and Phonons

• Metal: no energy gap around Fermi level, well-defined FS, finite DOS at EF

• Semiconductor: small energy gap around Fermi level

• So far we treat lattice as purely static, why is it ok?
• Insulator: big energy gap around Fermi level

• Semiconductor and insulator are different only in quantitative sense. We • In fact the lattice dynamics is very interesting and important, but
usually just call them insulators. in general it is very difficult to describe lattice vibrations in solids

Lattice vibrations and Normal modes Lattice vibrations and Normal modes

The problem can be significantly simplified by two approximations: • In classical wave mechanics, with Harmonic approximation, the lattice
vibrations can be represented by independent normal modes, in which
all the lattice ions oscillate with the same frequency Z.
• we only consider nearest neighbor interaction and ignore long
range interaction • Any vibrational motion of a lattice can be considered as a superposition
(linear combination) of normal modes with various frequencies
• we treat the interaction as harmonic potential:

V(a+δx) = V0 + ½ β(δx)2

• The resulting lattice can be visualized as a system of balls • The motion of the N ions is represented as a superposition of 3N
connected by springs normal modes of vibration, each with its characteristic frequency

Harmonic oscillators Phonon

• In quantum mechanics, a normal mode vibration with frequency Z • The quantum of the lattice vibration is called “phonon”, in analogy to
corresponds to a harmonic oscillator with frequency Z, which is the photon
solution of the harmonic potential.
• For harmonic oscillator with EZ = (nk + ½ )ħZ, we say nk phonons each
with energy ħZ are excited.
• The energy of the harmonic oscillator is: EZ = (nk + ½) ħZ, where
nk is the excitation number of the oscillator. • Phonon obeys Bose-Einstein distribution, like photon.

• The energy is quantized in unit of ħZ with a zero point energy ½ ħZ • The average number nk of phonons with energy ħZ at T is:

• The total energy of the lattice vibrations is just the sum of the
energies of the individual harmonic oscillators:

• The number of phonon is not conserved, it can be created or annihilated,

like photon.
 Phonons Lattice waves in mono-atomic 1D chain

• Phonon is an important example of collective excitations.

• The collective motion of many lattice sites is represented as a particle.

• It is also a type of quasiparticle (in a more general sense than Landau’s

Equation of motion:
electron-like qp).

• The language of phonon is very convenient in describing energy

exchange between different vibrational modes, and with other Solution: superposition of normal modes with different Z and q
excitations such as electrons and photons.

• In this sense it is also very similar to photon, which was invented to

describe the interaction between light and electrons. Apply the periodic boundary condition, solve the equation
of motion, then we can find the dispersion relation ω(q)

Phonon dispersion in mono-atomic 1D chain Materials’ specific heat

Dispersion relation (Z vs. q relation):

Long wavelength limit:

• Dispersion relation of a mono-atomic

1D chain
• At long wavelength, the speed is a • Experimental data of specific heat vs. T, both for metals and insulators
constant independent of q
Speed: • They all collapse in one universal curve when the T is scaled
• It is a typical behavior of sound wave
• What does this mean? Where is the specific heat from?
• This phonon dispersion is called
How fast is the speed? acoustic phonon • What is the characteristic behavior? Why decrease at low T?

Phonon specific heat Debye model

• The total energy of all the phonons • Assume linear dispersion for all branches: Z = cq, like long
wavelength acoustic phonon

• Does this formula look familiar?

Integral over all q and all branches

• The specific heat of the phonons:
• Define a cutoff in the integral to satisfy the 3N counting:
• ħcqD = ħZD = kBTD
• ZD: Debye frequency, a measure of the maximum phonon frequency
• How to calculate this summation over all phonon modes in
• TD: Debye temperature, a measure of the temperature above which all
different branches? modes begin to be excited (100 K to 1000 K).

Phonon specific heat: Debye model

Transport Phenomena
• basic notions of transport phenomena

• electrical conductivity

• the Boltzmann equation

• thermal conductivity

• thermoelectric transport coefficients

• At T > TD, cv ~ 3NkB, Dulong-Petit law, lattice ions behave like classical particles • electron transport in magnetic field

• At T → 0, cv ~ T3, Stefan-Boltzmann law, blackbody radiation is the low T limit Reference:

of phonon specific heat! “Electrons and Phonons: The Theory of Transport Phenomena in Solids” by J. M. Ziman
 Basic notions of transport phenomena Transport phenomena in condensed matter
• A transport phenomenon is a physical mechanism by which External “motive forces” drive a solid out of equilibrium, and create a
particles or quantities move from one place to another. “steady current” associated with “electrons and phonons”.

• It’s a non-equilibrium, steady state phenomenon. Common motive force F: Common current J:
Electrical: E Charge: Je
• It’s a major branch of research in kinetics of ideal gas, fluid Thermal: -’T Heat: JQ
dynamics, and condensed matter physics. Magnetic: H Spin: JM

• Strong analogies between different transport phenomena,

usually they are closely related. L is the transport coefficient

The transport coefficients measure the ability of electrons and phonons

• Transport laws connect a transport flux with external influence
to “transport” a “steady current” under the influence of an external
flux = (coefficient) u (driving force) “motive force”.

Phenomena discovered by transport Electrical transport: electrical conductivity

• discovery of superconductivity In a conductive material, an applied electrical field E will generate

an electrical current Je.
• discovery of the Kondo effect

• invention of transistor Ohm’s Law, V is electrical conductivity

• discovery of organic metals

I
• discovery of integer quantum Hall effect

• discovery of fractional quantum Hall effect

V
• discovery of giant magnetoresistance effect

Transport measurements are very powerful, especially when applied Experiments: U = E/J
to novel materials and performed in extreme physical conditions. Measurements: four-probe, low frequency ac lock-in method

Origin of electrical resistivity Drude model for electron transport

Drude boldly applied the kinetic theory of ideal gas to transport in solids.

The basic assumptions are:

• in the absence of external field, electrons in E
solids move uniformly in a straight line, until
they get scattered by collisions.

• In external fields, electron motion is

• The electrical resistivity originates from scattering of electrons determined by Newton’s laws.

• Will a perfectly periodic lattice scatter electrons? • Collisions are instantaneous events that abruptly alter the velocity of electrons.
Electrons achieve thermal equilibrium with environment via collisions.
• scattering of electrons by non-perfect lattice (impurity, boundary,
• An electron experiences a collision with a probability per unit time 1/W, W is
vibrations), by other electrons, or by other excitations.
called the relaxation time or mean free time.

Drude model for electrical conductivity Thermal transport

E
An applied temperature gradient ’T generates a thermal current JQ

• Fourier’s Law
• In an electrical field E, the electron is accelerated with a = Ee/m • κ is thermal conductivity
• It gains a drift velocity v = a·t in time interval t
• With mean free time W, the average velocity gained between two Thermal conductivity is more complex than electrical conductivity.
collisions is: v = a·τ = Eeτ/m
• The electrical current is J = nev = nEe2τ/m = VE, so the electrical Thermal conductivity may come from electrons, phonons and any
conductivity: V = ne2τ/m = neP (P is called mobility) other excitations that carry heat.
 N e)
Electron Thermal Conductivity (N Drude model for Ne

• Boundary condition: no net flow of electrons • In simple 1D model, at each location x half
(sample is electrically isolated) e- from warm side and half from cool side.

-’Tx -’Tx
• the energy flux density at x:
• At each location, equal number of electrons from
x warm side and cool side x JQ = ½nvx[ε(Tx-vτ) - ε(Tx+vτ)]

• electrons from warm side have higher energy = ½nvx·dε/dT·dT/dx·(-2vxτ)

ε(Tx-vτ), due to a higher T they gained in the last = nvx2τ·cv·(-’Tx)
collision
= n·1/3v2·τ·cv·(-’Tx)
• electrons from cool side have lower energy ε(Tx+vτ), = 1/3n·v·l·cv(-’Tx)
due to a lower T they gained in the last collision

• Thus heat conducted from warm to cool side Electron thermal conductivity:

Revisit the Ne argument

Now we know:
• no net flow of electrons (electrically isolated)
-’Tx • At each location, equal number of electrons from
warm side and cool side
x
These are pure electrical and thermal transports • electrons from warm side have higher energy ε(Tx-vτ)
• electrons from cool side have lower energy ε(Tx+vτ)
Questions: • Thus heat is conducted from the warm side to the
is there mixture between electrical and thermal transports? cool side

The answer is YES! Questions:

• electrons of the warmer end should have faster velocity than those from the
cooler end.
• So at location x, there will be more electrons moving down than moving up.
Thermoelectric Transport
• How to explain the discrepancy with the boundary condition?

Electric field induced by temperature gradient Thermoelectric power (Thermopower)

Imagine an equilibrium system, turn on T gradient:
-’Tx • In conducting materials, the electric
• Initially there will be more electrons moving down field generated by a temperature
x than moving up gradient is called thermo-electric power,
heater
• But they cannot get out of the sample, so the extra or Thermopower.
thermometer
downwards electrons get piled up at the cooler end E -’T
• This effect was discovered by Seebeck
V 'T
(1821), also called Seebeck effect.
• An electric field is set up by the piled up electrons
thermometer
• The field generates a counter electron current to • Thermopower (also called Seebeck
+ + + + + ’T,
coefficient) is defined as: S = E/’
balance the extra downwards electrons
heat sink electrical field generated by certain
-’Tx
• So after a short settling time, the system reach a temperature gradient
x steady state with no net flow of current
E Thermopower measurement • Thermopower is the most commonly
• The boundary condition is satisfied, the consequence studied thermoelectric transport effect.
- - - - -
is an electric field induced by T gradient

Drude Model for Thermopower Peltier effect

• assume temperature gradient (-’xT) and relaxation time τ + Je
+
• electrons from warm end have average speed vx(x-vτ), + + JQ
electrons from cool end have average speed vx(x+vτ).
• the mean velocity at x due to thermal gradient is:
vT = ½(vx(x-vτ) - vx(x+vτ))
Imagine the following situation:
E -’T • For a metal, keep it in an isothermal condition, so ’T = 0.
= ½ (-2vxτ·dvx/dx) = vxτ·dvx/dT(-’T)
= τ (-’T) ·d(vx2/2)/dT x • apply an E along the sample, so there is an electrical current Je.
= 1/(3m) τ (-’T) ·dH/dT • since each charge carrier also carry a piece of heat, there is also
= 1/(3m) τ (-’T) ·cv/n + + + + +
a heat current JQ.
• the drift velocity due to electric field is: vx = Eeτ/m
• In the steady state the mean velocity is zero to maintain Je = 0 • the heat current is proportional to the electrical current, JQ = 3 Je.
• So the thermal power S = E/’T = cv/3ne • this is called the Peltier effect, 3 = JQ /Je is the Peltier coefficient.
 Thermoelectric applications Thermal couple

The Peltier effect is a reciprocal effect of the Seebeck effect.

• A thermal couple is made of two different
Tx (dissimilar) metals, why dissimilar?
electrical → thermal thermal → electrical
metal A metal B • One end is kept at a known temperature T0
SA SB
• They are closely related: 3 = TS, called Kelvin relation. • The voltage across the junction is
proportional to the difference between Tx
T0 V T0 and T0: V = (SB-SA)(Tx-T0)
• Thermoelectric effect can realize the conversion between heat and
electricity. • If we know T0, then we can get Tx.

• This is a widely used thermometer.

• Thermoelectric effect and materials have very unique applications
in power generation and refrigeration.

Thermoelectric power generation thermoelectric refrigeration

• 3-junction consisting of N type (electron carrier)
• 3-junction consisting of N type (electron carrier)
and P type material (hole carrier) with 3N and 3P
and P type material (hole carrier) with SN and SP
• An external current Je flowing through the two legs
• The two materials are fixed between two thermal will take away heat JQ = Je (|3N| + |3P|)
bath with different temperatures (Th and Tc).
• This makes a refrigerator that cools the top surface!
• The voltage difference between the two ends of • It’s called thermoelectric refrigeration.
the junction is: V = (|SN|+|SP|)(Th-Tc) • Advantages?
• Thermoelectric power generation!

Advantages: it is a solid state device, so:

• no moving part: quiet, no mechanical fatigue, reliable.
• can be made into arbitrary shape and size.
Chip cooler, local cooling
• clean, environmental friendly, green energy source! Wine cellar, no vibration
Portable fridge, small size

Giant Magnetoresistance (GMR)

• Most MR devices are used to measure the external magnetic field by
֠Ѽࢢࣂ monitoring the change of resistance.

• For practical MR applications, we need to find materials with large MR in

⧁Ӑᝐ weak magnetic field at room temperatures.

␻঄ཝᆜ⢟⨼㌱
• Giant magnetoresistance (GMR) effect was discovered in 1988 in Fe/Cr
multilayers by a research team led by Albert Fert.

• It was also simultaneously but independently discovered in Fe/Cr/Fe trilayer

by the group of Peter Grünberg.

• GMR revolutionized the IT industry and have huge impact in our daily life.

• For their discovery of the GMR effect, Albert Fert and Peter Grünberg shared
the 2007 Nobel Prize in physics.

GMR in multilayer thin film GMR in trilayer sandwich

Fe
Cr Fe
Fe
Cr Cr
Fe
Cr Fe
Fe

• 60 layers of Fe/Cr bilayered structure (Fert group, PRL 1988) • GMR in Fe/Cr/Fe trilayer sandwich, a very simple structure (Grünberg group,
published on PRB in 1989)
• MR decreases by ~ 50% at T = 4.2 K in a 2-tesla in-plane field, GMR!
• MR decreases by ~ 1.5% in an inplane magnetic field ~ 500 G at room T.
• The GMR effect (both the magnitude and the saturation field) show strong
variations with the thickness of the Cr layer • Hysteresis in the R-H loop was observed.
 GMR systems and mechanism Transport in ferromagnetic metals
• Some other trilayer and multilayer structures 0.50

(left column) that show GMR effect U

0.45

Uxx (m:cm)
• What is the common feature of all these
GMR systems? 0.40

FM paramagnetic
• Ferromagnetic metal layers (Fe and Co) 0.35
0 50 100 150 200
separated by non-magnetic metal layers T (K)
T
(Cu, Au, Cr)! TC • Ferromagnetic Cr doped Sb2Te3
• Data courtesy of Minhao Liu
• Mechanism of GMR: why are ferromagnetic metal layers separated by non-
magnetic metal layers, show GMR effect? • For many ferromagnetic metals, it was found that the resistivity decreases
suddenly as it is cooled to below TC, i.e., the ferromagnetic state has much
• In ferromagnetic metals the scattering rates of conduction electrons depend on
smaller resistivity than the paramagnetic state.
the relative orientation (parallel or antiparallel) of the electron spins with
respect to the local magnetization. • This can be explained convincingly using Mott’s two-current model.

Mott’s two-current model Mott’s two-current model

• The influence of spin on electron transport was first proposed by Mott in 1936 Key assumptions of Mott’s two-current model:
to explain the sudden decrease of resistivity in ferromagnetic metals at TC.
• spin is preserved (no spin-flip process)
• Mott’s two-current model: conduction • conduction is mainly by s-electrons (low
current in ferromagnetic metals is the sum of effective masses compare to the d-electrons)
two currents carried by the majority-spin
electrons and minority-spin electrons • The scattering processes include scattering
separately. of s-electrons into other s-electron states and
into the d-electron states with the same spin.
U↑
I↑ • The dominant scattering mechanism is into the d-electron states, WHY?
I
IĘ UĘ • For the band structure shown above, below TC only the minority spin electrons
can be scattered into d-electrons, WHY?
• Equivalent to two resistors in parallel:
• Above TC, both the minority and the majority spin electrons can be scattered into
The two current model 1/U = 1/U↑ + 1/UĘ
the d-electron states, WHY?

Mott’s two-current model RKKY (Ruderman, Kittel, Kasuya, Yosida) exchange

Uss↑ Usd↑ • In metals, conduction electrons can mediate magnetic interaction.
I↑ • A local magnetic moment in a host metal will polarize the conduction electrons
I around it, and in turn couples to another local moment at distance r away.
I↓
UssĘ UsdĘ • At large distance r, the coupling has the form JRKKY ~ cos(2kFr)/r3.

• Above TC, both the minority and majority spin electrons can be scattered into • It is long-range and has an oscillatory spatial distribution with cos(2kFr)
d-electron states.

Uss↑
I↑
I
I↓
UssĘ UsdĘ

• Below TC, only minority spin electrons can be scattered into d-electron states.

• Mott’s simple two-current model nicely explains the decrease of U at TC. distance (Å)

Oscillatory interlayer exchange coupling Transport in anti-parallel trilayer

• By summing over the RKKY interactions from all the spins
in the ferromagnetic sheets, the interlayer coupling between • For an appropriate spacer thickness around 10 to 20 Å two
the two ferromagnetic layers in the trilayer structure is: ferromagnetic layers can have anti-parallel alignment.
Fe Cr Fe
• How to achieve GMR using “Mott’s two current model”
Fe Cr Fe and the spin dependent transport picture?

• It has a oscillatory behavior as a function of the distance between the two layers.
• The scattering rates of conduction electrons depend on the relative orientation
(parallel or anti-parallel) of the electron spin with respect to local magnetization.
• The interlayer exchange could be either ferromagnetic or antiferromagnetic,
depending on the thickness of the spacer layer. • For simplicity, let’s assume the electrons will be strongly scattered only when the
electron spin and local magnetization has anti-parallel orientation.
• We can thus have parallel or antiparallel spin alignments between the two layers
by choosing the appropriate thickness of the spacer. • At zero field, both spin up and spin down electrons in the above structure will
be scattered, thus sample has a large resistance in Mott’s “two current model”.
 Transport in parallel layers Mechanism of GMR
• What if we apply a magnetic field along the film plane direction?

• The magnetic moments of the two ferromagnetic layers

will be aligned parallel.

• When conduction electrons travel in the sample, only

spin down electrons will be strongly scattered, spin up
electrons will not suffer spin-dependent scattering.
Fe Cr Fe
• In Mott’s two current model, the resistance will be
significantly smaller than the zero field resistance
B because now one conduction channel is not scattered
and acts as a short circuit.
• External magnetic field causes parallel alignments of the FM layers along the up
direction, so only spin down conduction electrons are strongly scattered.
• This explains the significant decrease of resistivity of the magnetic trilayer in a
• The spin up channel has small resistance, acting as short circuit, leading to a
magnetic field, which is the working mechanism of GMR!
much smaller total resistance than the zero-field anti-parallel aligned layers.

Spin valve Spin valve

Free ferromagnetic layer Free ferromagnetic layer

Non-magnetic spacer layer Non-magnetic spacer layer
Pinned ferromagnetic layer Pinned ferromagnetic layer

AF pinning layer AF pinning layer

• In the spin valve structure, one ferromagnetic layer is pinned (with fixed • Under the influence of external magnetic field, the magnetic moment of the
magnetization) by a natural AF layer (exchange biasing) free ferromagnetic layer can be parallel or anti-parallel to the pinned magnetic
layer, leading to GMR effect.
• On top this is a layer of non-magnetic metal as spacer
• In the spin valve the two ferromagnetic layers are magnetically decoupled.
• Above that is another soft ferromagnetic layer, called free ferromagnetic layer,
• The spacer layer is thicker (but need not be accurate) than in the original
whose magnetization can be easily changed by external magnetic field
GMR devices so the RKKY-like interlayer exchange coupling is not present.

GMR as hard disk drive read head GMR as hard disk drive read head

picture of a hard disk drive Increase with time of the areal bit
density in hard-disk recording
Schematic of a HDD read-write head
Schematic of magnetic recording and reading
• The GMR read head has a spin-valve structure: AF MnFe as the exchange-
• An electromagnet writing unit magnetizes the magnetic surface of the platter.
biasing layer; FM Co as the pinned layer; Non-magnetic Cu as the spacer layer;
• The direction of the magnetization corresponds to binary digit 0 and 1. FM NiFe20 as the free layer that can be flipped by the magnetic storage bit

• A GMR device is used to read the stored information (direction of the magnetic • The small size and high sensitivity of the GMR read head dramatically
field generated by the magnetic bits). improved the density of HDD, and make it available to consumer electronics.

The Hall Effect

Quantum Hall Effect
Discovered in 1878 by Edwin Hall, a graduate student at Johns Hopkins.

• The Hall effect

B Vy
• Phenomena of Quantum Hall Effect
Ix
• Landau Levels and localization

• Edge state Vx

Experimental Configuration: Hall bar Hall’s original data on Au leaf

• Apply Ix along longitudinal and Bz along perpendicular directions

Reference: • The transverse Hall voltage increases linearly with magnetic field
“The Quantum Hall Effect” by Daijiro Yoshioka (Springer Series in Solid-State Sciences)
• Same behavior in most normal metals.
 MR and Hall effect MOSFET
• MOSFET ( Metal Oxide Semiconductor Field Effect Transistor )
• Si MOSFET is the most commonly used semiconductor device.
For Free Electron Gas,
the equation of motion: • It combines Si, a 3D semiconductor, SiO2 , an insulator, and a metal.

Boundary condition: Jy = 0, so in steady state vy = 0.

so no magnetoresistance

Picture of an “ancient”
so Hall coefficient RH = 1/ne Si MOSFET
Schematic of Si MOSFET

2D electron gas (2DEG) in MOSFET Quantum Hall Effect (1980)

Uxx Uxy

k:/sq h/e2

• Naively the MOSFET device can be viewed as a plane capacitor.

Magnetic field (B)
• When a positive voltage (VG) is applied on the gate, electrons in the semi-
conductor are drawn towards the interface between p-Si body and SiO2 layer. For fixed gate voltage (carrier density), with increasing magnetic field:

• The density of the electrons (ne) can be controlled by the gate voltage, HOW? • step-like Hall resistance, broad plateau at quantized value Rxy = h/ie2
• MR almost vanishes at the “Hall plateaus” and peaks sharply between plateaus.
• When a small bias voltage (Vsd) is applied between source and drain, electrical
current will flow in the 2-dimensional electron gas (2 DEG). • The effect is more dramatic at high field. What is the low field behavior?

The QHE puzzle Hall effect in the large Hall angle regime
• Why the Hall Effect form quantized plateau?
• Why MR becomes 0 at the Hall plateau? jx Bz

• It happens for 2DEG in strong magnetic field at low temperature

• QHE appears in 2D materials at low T and high B.

jx Bz E • In this regime, ZcW >> 1, electrons finish many cycles of motion

before they are scattered.

• In a magnetic field, the direction of electrical current (J) generally does not
• This is the quantum limit of charge transport, where the QHE
coincide with that of the electrical field (E). happens.

• There is an angle between them (called Hall angle): • Next we will consider the motion of electrons and the electronic
tan T = Ey/Ex = BeW/m = ZcW (Zc = Be/m is cyclotron frequency) energy levels in this regime.

Classical electron motion in B and E Classical electron motion in B and E

In the absence of scattering (W o f), such In the absence of scattering (W o f), such y
as in free space, the equation of motion is: as in free space, the equation of motion is: x
Ex
Bz Ex z
d 2vr d 2vr
m e( E  v u B ) m e( E  v u B )
dt 2 dt 2

The solution is: The solution is:

vx = r0Zcsin(Zct) vx = r0Zcsin(Zct)
vy = r0Zccos(Zct) + v0, where v0 = -Ex/Bz vy = r0Zccos(Zct) + v0, where v0 = -Ex/Bz v0 Bz

• The motion of electrons in the presence of orthogonal B and E consists of two

parts: a fast cyclotron motion and a slow drift with constant v = EhB/B2
perpendicular to both B and E.

• The motion is along the equi-potential contours defined by the total potential V.
 Quantum mechanics of 2DEG in B Landau levels (LL)
B Hamiltonian of electrons moving within a 2D
plane in a perpendicular magnetic field:

1 ur ur
This is exactly the Hamiltonian for quantum harmonic oscillator along the x-axis,
H ( p  e A) 2
2m and the energy eigenvalues are:

Choose Landau gauge: A = (0, Bx, 0), so that B = Bez

• The cyclotron motions of 2DEG in a perpendicular magnetic field can be

simplified as 1D Harmonic oscillator.

Operator py commutes with this Hamiltonian since the operator y is absent in • The continuous energy levels in zero field split into discrete Landau Levels
Landau gauge. Then the operator can be replaced by its eigen value ħky. (LLs) by magnetic field.

Quantum mechanics of 2DEG in E and B DOS of Landau Levels

In the presence of an electrical field E along the x axis and B along the z axis,
the Hamiltonian: 1 ur ur
H ( p  e A) 2 - eEx
2m

Energy
The eigenenergy of the electrons are:

DOS: Increasing B

X is a constant displacement relative to the center of the Harmonic oscillator. • For a 2D free electron gas, the DOS per unit area is: m/Sħ2

• Just like in the classical case, the electrons have a cyclotron motion energy (1st • In a B field, the 2D electrons will occupy the discrete Landau Levels.
term), a drift motion energy (3rd term), and a constant potential energy (2nd term).
• The total number of electronic states contained by each LL is:
• The continuous energy levels in zero field split into discrete Landau Levels by B. D = m/Sħ2 ˜ S ˜ ħZc = B˜ S˜ 2e/h, D is called the degeneracy of each LL

• The wavefunction is extended along an equipotential line perpendicular to E. • What’s the trend with increasing B?

DOS of Landau Levels Filling of Landau Levels

EF
Energy

DOS: Increasing B

• So far we have only considered the orbital motion in a B field. DOS DOS

• In a B field, the spin degeneracy is lifted by the magnetic field (Zeeman splitting). • In realistic sample, the LLs are broadened (by what?)
• If up to EF exactly i LLs are filled, then we say the LL Filling factor is i.
• The degeneracy for each spin-polarized LL is then: D = B˜ S˜ e/h (any meaning?)
• The total number of electrons contained by the filled LLs is:
• We will always refer to this spin-polarized LLs without mentioning it explicitly.
ieBS Ne ieB n is the areal density
Ne iD n
• What about Pauli exclusion principle? h S h of the 2D electrons.

The potential energy The role of impurities

• So far we have only considered the kinetic energy of electrons, i.e., the LLs • It is very remarkable that impurities are necessary for the existence of QHE,
of cyclotron orbital motion.
because we always focus on how to get rid of them (to find clean 2DEG).
• What about the potential energy?

• From our discussion, we know that the electrons’ path will follow the
energy

equi-potential lines while doing fast cyclotron motion.

• What defines the equi-potential line in the 2DEG?

• Electrical field, charged impurities, and the electrons themselves.

No impurity: flat energy landscape With impurities: complex energy landscape
• In microscopic scale, the potential by electrical field is almost a constant.

• We ignore the electron-electron Coulomb interaction for now, i.e., we adopt a broadened by impurities
non-interacting single-particle picture. Landau Levels
• The only thing left is the impurity potential!
 Equipotential contour in disordered 2DEG The sample edge
• These questions can be resolved when we consider the edge of the 2DEG.

• In quantum mechanical situation (ZcW >> 1), electrons exhibit cyclotron-like

energy

motion along the equipotential contour.

• In the presence of E and B fields, the electron motion is along the

equipotential contours defined by the total potential V.
• The equipotential contour of summit and basins are both closed (localized),
only that of the flat land is open (extended).
For a macroscopic sample, all electrons in the bulk are localized
• That means only states with potential energy near the average value of the
because the equipotential orbitals are closed.
total potential energy are extended.

The sample edge Extended state along the edge

vacuum

edge edge

edge state Skipping orbital

potential
energy • In 2DEG in a magnetic field, all states in the bulk are localized, the only extended
states are the states along the sample edge, or the edge states

• At the sample edge, the potential energy increases steeply • In classical picture, only electrons moving along the edge are extended because
they cannot finish full cyclotron motion due to bounce-back from the edge.
• The equipotential contour are lines parallel to the edge, the electrons thus
travel along the edges, forming the edge states • Therefore, electrons move parallel to the edge in skipping orbits.

Energy diagram at the sample edge Current carried by the edge states

PA P1 PC
A C B
I1
x
y I2 I
I
PB B P2 D PD

• The edge states are protected from back-scattering, it is dissipationless.

• If we adopt an infinite potential wall at the sample edge, the energy of an
• Each LL has a quantum conductance: G = e2/h
electron increases rapidly at the edge.
• We only need to count the number of LLs that are occupied (thus the edge
• Each LL now crosses EF at the edge, thus contributes to current carrying.
state crosses the EF) to get the total conductance G = ie2/h
• In the bulk, all states are localized and there is energy gap between LLs.
• The edge current is determined by the chemical potential (P) of the electrode
• Therefore, the current is carried solely by the 1D edge states. from where the electron leave.

Edge State Current and Voltage Vanishing longitudinal resistance

PA PC PA PC

PL PR PL PR

PB PD PB PD

• Assume now that in the bulk EF lies in the energy gap between two LLs, i.e., • Along the upper edge, electrical current IL leave source electrode with PL.
i LLs are occupied and i edge states contribute to conductance.
• The first voltage electrode A does not draw net current, the current flow out
• The total conductance between source and drain is: G = ie2/h equals the current flow in, so its chemical potential PA = PL.
• The same thing for the second voltage electrode C, so PC = PL = PA.
• Assume there is a small bias between source (left) and drain (right) electrodes
so that PL - PR = eV. • So there is no chemical potential difference, and no voltage between A and C.

• Then the edge state can carry a current I = V˜ ie2/h • This explains the vanishing resistance!
 Quantized Hall resistance Quantized Hall resistance
PA PC PA PC

PL PR PL PR

PB PD PB PD

• Same to the lower edge, where electrical current IR leave drain electrode with PR. • The crucial point is the whole upper edge has the same potential as the source
• The two voltage electrodes D and B have the same chemical potential as the electrode, the whole lower edge has the same potential as the drain electrode.
drain, since no net current is drawn into them: PB = PD = PR.
• This is a peculiar feature due to the unique properties of quantum edge state.
• The voltage difference between A and B, which is the Hall voltage measured in
the Hall effect, is thus eVAB = PA - PB = PL - PR = eV • This is why there is no longitudinal resistance and the Hall angle is 90 degree.

• The measured Hall resistance is thus: Ryx = VAB/I = V/(V˜ ie2/h) = h/ie2 • The potential only changes abruptly at the boundary where electrons flow into
• This explains the quantized Hall resistivity! the source or drain electrodes.

Broad Hall Plateau Question: what about even higher field?

PA PC

PL PR

PB PD

• As long as the Fermi level lies in between two bulk LLs, the number of edge
states that cross EF does not change.

• Thus the quantized Hall resistivity and zero longitudinal resistivity do not change.

• This is why there is a broad plateau!

• The Hall resistivity and longitudinal resistivity changes only when EF is varied • For the maximum field here, only the 1st LL is occupied so that:
through a bulk LL, then there is a change of the number of edge state by one. Uxy = h/e2 and Vxy = e2/h

• Why there is a finite Uxx across the transition? • What if we further increases the magnetic field?

Fractional Quantum Hall Effect Fractional Quantum Hall Effect ( 1982 )

• In early 1980s, Tsui and Stormer studied the Hall effect in very clean 2DEG • The original published data of the
grown by modulation doping. Fractional QHE (Tsui et al., 1982)
• At very low T and high magnetic field, they found something very strange • At low fields (large filling factors),
Uxy pronounced signatures of the
integral QHE
4.2 K
• New discovery: Hall plateau at
1.5 K fractional filling factor Q = 1/3
with Uxy = 3h/e2 or Vxy = 1/3.e2/h

• Resistivity minimum at the same

regime

• The features become more

1st LL pronounced at low T
B
• Other special filling factors?
• Hall plateau and resistivity minimum at filling factor Q = 1/3!

Much richer FQHE structure What is the origin of the FQHE?

• Measurements on very clean
• In the FQHE regime, the kinetic energy becomes an irrelevant constant, WHY?
sample, at extremely low
temperature and super strong • The dominant term in the Hamiltonian is the Coulomb interaction between
magnetic field electrons:
• Very rich structure of FQHE
• Hall plateau and resistivity
minimum in many fractional • We need to solve this many-body Schrodinger equation which we don’t
filling factors know the exact solution.

• Any rule for the quantization • Remarkably, soon after it was discovered, Bob Laughlin came up with the
filling factors? right answer (Laughlin, PRL (1983)).

• By a stroke of genius, he guessed the FQHE wavefunction and was able to

• Q = p/(2q+1) explain all the key features of the Q = 1/3 FQHE state.

• Other pattern? • Here we will proceed through a pure phenomenological path.

 Effect of strong electron correlations Delocalization of electrons by correlation
• All these discussions about the Integer QHE are based on a single particle
• If the electron correlation is significant, the total potential is determined by
picture without considering electron-electron interactions.
both the impurity potential and the Coulomb repulsion between electrons.
• In the presence of strong electron-electron Coulomb interaction, are these
discussions still valid? • Then the total potential is no longer static, but changes with the position of
other electrons.
• Most importantly, are the bulk states still localized?
• When the Coulomb potential is larger than the impurity potential, the
equipotential contours in the bulk are not static closed loops, thus the electrons
states are delocalized.

• This oversimplified argument illustrates the competition between electron-

electron Coulomb potential and the impurity potential.

• It is kind of counter-intuitive because we showed before that correlation usually

causes electron localization (Mott insulator and Wigner crystal).

Classical view of electrons in 2D Quantum view of electrons in 2D

• In the classical picture, electrons

are distinguishable, charged
particles moving in a plane.

• Due to the Coulomb repulsion

• In the quantum mechanical view, electrons are inherently indistinguishable.
between electrons, they push and
avoid each other along the path
• They are quantum mechanical waves permeating in space.
of their motion.
• Coulomb repulsion affect the probability of finding an electron in a particular
• In the presence of external location if an electron has been detected in another location close by.
B
magnetic field, electrons move
in cyclotron orbitals. • The electrons thus form a strongly correlated, featureless quantum liquid that
is not easy to illustrate graphically.

The effect of magnetic field Magnetic field and flux quanta

B B

• In an external magnetic field, electrons participate an elaborate, mutual, • In the composite particle picture, the magnetic field and the correlated electrons
quantum-mechanical dance to avoid each other and find the energetically are treated separately.
favorable state. • The external magnetic field penetrating the 2DEG creates tiny whirlpools, called
• Describing their motion becomes a mission impossible. vortices, each carrying a magnetic flux quantum )0 = h/e (flux quantization).

• A big conceptual breakthrough is to disentangle this many-particle problem • The phase of electron wavefunction changes by 2S upon finishing one circle of
into an effective single-particle problem by introducing new kind of particles: motion around a vortex (the Aharonov–Bohm effect) to ensure single value.
the Composite Particles, combination of electrons and magnetic flux quanta. • At the vortex core there is a void of electron, or creation of a hole.

Binding electrons with fluxes Binding electrons with fluxes

B B B B

• Now let’s think about electrons. • Combining an electron and a vortex reproduces the real situation: featureless
liquid uniformly distributed in space.
• What is the energetically most favorable way to put electrons in this system?
• Each electron is at the center of a flux and at the same time is part of the pool
• It is energetically favorable to put electrons in the center of vortices. that creates the vortices surrounding all the electrons!

• Electrons and vortices are opposite objects, one representing a package of • This is characteristic of quantum many-particle physics.
charge the other the absence of charge (hole).
• Through this detour we can again think about the FQHE problem using the
• Putting electrons at vortex centers help to reduce electron mutual repulsion. single particle picture that we are familiar with.
 Composite Particles (CP) Filling factor and flux quanta
CP = electron + flux quanta

Energy
impinging magnetic field
DOS: Increasing B
creating vortices
• Recall the degeneracy of each spin-polarized LL:
each vortex carrying one flux quantum
eBS )
D the number of electrons in each LL = the number of flux quanta
Energetically favorable to bind electrons to flux quanta h )0

• For filling factor Q = 1, each electron state corresponds to one flux quantum
• Understanding correlated electrons in strong magnetic fields is a formidable task.
• The filling factor characterizes the ratio of charge and flux quanta of a CP.
• The CP picture provides an intuitive way of looking at this problem.

Filling factor Q = 1 For filling factor Q = 1/3

• For Q = 1/3, the number of flux quanta is 3 times of the number of electron.

• For Q = 1, each electron is bound to a flux quantum.

• Or in another word, only 1/3 of the vortex center is occupied by electrons if
• The CP here is “one electron + one flux quantum” we put one electron in each vortex.

• The electrons form a densely-packed liquid and interaction between them is

• In this case, what is the energetically most favorable state from the viewpoint
unimportant, just like in normal metals. of CP?
• It can be viewed as noninteracting CPs (1 electron + 1 flux) in zero magnetic field.

For filling factor Q = 1/3 Stormer’s illustration of the Q = 1/3 state

• The energetically most stable state is to form CPs = 1 electron + 3 flux quanta

• The three vortices provide three layers of screening of Coulomb potential

• The effect of magnetic field is totally absorbed into the flux quanta.
• Stormer’s famous artistic illustration of the Q = 1/3 CPs living in a background
• 2DEG can be viewed as non-interacting CPs (1 electron + 3 fluxes) in zero field!
with potential fluctuations.

Question: Statistics of CPs

Are the composite particles fermions or bosons?

What is the composite particle for filling factor Q = 1/2 ? With exchange of position:

electron part Me o Me

For each attached flux quantum: M flux o M flux

CP = 1 electron + 2 flux quanta
So the Q = 1 CP is a boson!

CPs can be either fermions or bosons, depending

on the number of attached flux quanta
In experiments there is no Hall plateau at Q = 1/2.
CP = 1 electron + m flux quanta:
What’s so different between Q = 1/2 and Q = 1/3 ?
m = odd CP = boson MCP 
P
o MCP
This is an important question regarding the CP picture. m = even CP = fermion MCP 
P
o MCP
 Anyon Composite Fermions (CF) at Q = 1/2

• The statistics of the CPs cannot be simply classified as fermions and bosons,
they are thus called Anyons.

• Anyon statistics is a special property of particles moving in 2D.

• The exchange of two electrons creates a phase S in the total wavefunction

(the wavefunction changes sign).
• CPs for Q = 1/2: 1 electron + 2 fluxes
• The binding of a magnetic flux quantum on an electron creates an additional
phase S for the exchange of electron positions. • TheQ = 1/2 CPs are composite fermions (CFs), they behave like a gas of
non-interacting fermions in zero field.
• This additional phase changes the statistical properties from fermions to
• The behavior is exactly like normal free electrons in zero magnetic field.
bosons.
• This is why there is no FQHE for Q = 1/2.
• Each time an additional flux quantum is attached to an electron, the CP’s
statistical property will switch once. • However, the Q = 1/2 CF state is actually the most important one because it
can be viewed as the parent state for other FQHE.

The Q = 1/2 CF Landau Level The Q = 1/3 FQHE from the CF picture

• When magnetic field deviates from the exact Q = 1/2 filling factor, the Q = 1/2
CFs move in an effective external magnetic field, which is the remaining field
• The Q = 1/3 CB can be viewed as a Q = 1/2 CF + 1 flux
after attaching two fluxes to each electron.

• Just like electrons, the Q = 1/2 CFs also form discrete Landau Levels, i.e., the • The Q = 1/3 state can be viewed as to fill the 1st LL of the Q = 1/2 CFs
LLs for CFs.
• Therefore the Q = 1/3 FQHE is the i = 1 IQHE of the Q = 1/2 CFs!
• The Q = 1/2 CFs thus can have their own integral QHE. i.e., IQHE for CFs.
• This naturally explains the FQHE at Q = 1/3.
• This provides a very straightforward explanation of all the FQHE states.

All those other FQHE states hierarchical structure and self similarity
Uxx

• All the other FQHE states can also be explained nicely by drawing analogy
between CFs and electrons!

• What about Q = 1/4?

• What about Q = 1/5?

• What about Q = 2/3?

hierarchical structure and self similarity Summary of FQHE

• It happens in dilute, extremely clean 2DEG at extremely low
temperatures and strong magnetic fields.

• Quantization of Hall effect at fractional filling factors.

• The composite particles idea reduces the strongly-correlated

many-body effect in a strong magnetic field to a non-interacting
one-particle problem in zero magnetic field.

• FQHE can be explained as the QHE of composite fermions.

Other types of quantum Hall effect?