Вы находитесь на странице: 1из 80

Course Reader

for
Introduction to Solid State Physics
Di-Jing Huang
c Draft date October 6, 2014

Contents
1 Specific Heat of Solids
1.1 Einsteins calculation: simple harmonic oscillator . . . . . . .
1.2 The T 3 dependence . . . . . . . . . . . . . . . . . . . . . . . .
2 Electrons in Metals
2.1 The Drude model . . . . . . . . .
2.2 Free electron Fermi gas . . . . . .
2.3 Electronic heat capacity . . . . .
2.4 Screening and the Mott transition
2.5 Thermionic emission . . . . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

3 Vibrations of One-Dimensional Lattices


3.1 One-dimensional monatomic solids . . .
3.2 Reciprocal lattice . . . . . . . . . . . . .
3.3 Diaomic chain . . . . . . . . . . . . . . .
3.4 Quantized waves: phonons . . . . . . . .
3.5 Anharmonic effect in crystals . . . . . .
3.5.1 Thermal expansion . . . . . . . .
3.6 Heat conduction by phonons . . . . . . .

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

3
3
6

.
.
.
.
.

11
11
13
18
21
24

.
.
.
.
.
.
.

29
29
31
34
36
38
38
38

4 Crystal structure
39
4.1 Lattices and unit cells . . . . . . . . . . . . . . . . . . . . . . 39
4.2 Symmetry of 3D crystals . . . . . . . . . . . . . . . . . . . . . 41
4.3 The reciprocal lattice in 3D . . . . . . . . . . . . . . . . . . . 45
5 Wave Scattering by Crystals
51
5.1 The Laue and Bragg Conditions . . . . . . . . . . . . . . . . . 51
6 Electrons in a Periodic Potential
57
6.1 Electron band . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.2 Tight-binding approximation . . . . . . . . . . . . . . . . . . . 57
i

ii

CONTENTS
6.3
6.4
6.5

The Translational Symmetry Blochs theorem . . . . . . . . 61


Nearly Free Electrons . . . . . . . . . . . . . . . . . . . . . . . 65
Symmetry in Electronic Band Structure . . . . . . . . . . . . 71

7 Electron-Electron Interactions
7.1 The Hatree-Fock approximation . . . .
7.2 Electron-electron interaction: Screening
7.3 Fermi liquid and quasiparticles . . . . .
7.4 Electron-phonon interaction . . . . . .
7.5 Electrons in a magnetic Field . . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

73
73
73
73
73
73

8 Semiconductor Physics
75
8.1 Electrons and Holes . . . . . . . . . . . . . . . . . . . . . . . . 75
9 Magnetism

77

Introduction
Solid-state physics is the field of physics that deals with the macroscopic
and microscopic physical properties of solids, state of matter in which a large
number of atoms are chemically bound to produce a dense aggregate. These
properties are broad, rich, and deep because they emerge from collective
phenomena of enormously large number of particles. This many-body physics
is often beyond reductionism and most of the time is closely connected to
our daily life.
This course reader is prepared to cover what will be presented in the
lecture of Introduction to Solid State Physics (I). Instead of surveying many
phenomena of solids, this course focusses on the basic concept of solid-state
physics, including crystalline lattice structure, electronic properties, semiconductor physics, and magnetism. The contents of this course follow Simons
text closely; symbols and notations used here are almost the same as those
in the textbook.
Textbook: The Oxford Solid State Basics, by S. Simon, Oxford University
Press (2013)
References:
Introduction to Solid State Physics, 8th ed, by Charles Kittel, Wiley
(2004).
This is a very popular text. It collects many useful informations and is
quite handy as a reference material after you have some background in
solid state physics. In my opinion, it is not a good introductory book.
Its theme may seem unclear to many beginners, particularly those who
find themselves struggling to follow its presentation.
Solid-State Physics, 4th ed, by H. Ibach and H. Luth, Springer-Verlag
(2009).
Another very popular book on the subject, with quite a bit of information in it.
1

CONTENTS

Two graduate-level references:


Solid State Physics by N. W. Ashcroft and D. N. Mermin, Cengage
Learning (1976).
This is an elegantly written and standard complete introduction to
solid state physics although it is a dated book, published 38 years ago.
Still it is a favorite text of many solid-state-physics courses.
Condensed Matter Physics, 2nd ed, by M. Marder, Wiley (2010).
It is a good reference book which is complementary to the classic
Ashcroft-Mermin text; some parts of the book practically echo AshcroftMermin. The author attempts to provide a great deal of breadth on
the modern condensed matter physics. The derivations are complete,
although difficult for some beginners to follow.

Chapter 1
Specific Heat of Solids
Historically the research of solid state physics began with the study of
the thermal properties of solids without considering microscopic structure.
In 1819 Dulong and Petit found experimentally that for many solids at room
temperature the heat capacity per atom Cv is approximately 3N kB , i.e., the
Law of Dulong-Petit, where kB is Boltzmanns constant. While this law is
not always correct, it frequently is correct for most of materials. An exceptional example is diamond whose molar heat capacity at room temperature
is 0.735R, much smaller than 3R. Particularly this law does not hold at low
temperatures; for diamond, room temperature appears be low temperature.

1.1

Einsteins calculation: simple harmonic


oscillator

Assume that each atom vibrates independently of each other and every
atom has the same vibration frequency. Consider an 1D harmonic oscillator
in equilibrium with a heat bath at temperature T . The oscillator can not be
fixed at a quantum state n with energy En = ~(n + 1/2), where ~ and are
the Planck constant divided by 2 and the angular frequency, respectively.
Instead the probability that the oscillator is in state n is Pn = eEn in
which is a normalization constant and is defined as 1/kB T with kB being
the Boltzmann constant. As the oscillator must be in one of the possible
3

CHAPTER 1. SPECIFIC HEAT OF SOLIDS

states,
X

Pn = 1,

~(n+1/2)

n=0

e~/2
=
= 1,
1 e~

= (1 e~ )e~/2 .
Therefore we have

Pn = (1 e~ )en~

(1.1)

and the average energy hEi of the system is


hEi =
=

X
n=0

~(n + 1/2)(1 e~ )en~

X
~
+ ~(1 e~ )
n(e~ )n
2
n=0

~
e~
+ ~(1 e~ )
2
(1 e~ )2


1
1
+
= ~ ~
.
e
1 2
=

(1.2)

P
x
n
1
Here we use
n=0 nx = (1x)2 to calculate the summation. This expression
has the form of En = ~(nB + 1/2) of a single harmonic oscillator with the
Bose occupation factor nB as
nB (~) =

1
.
1

e~

and the expectation energy of the solid is then2




hEi = ~ nB (~) + 1/2 .
The heat capacity C is defined as

C=

hEi
.
T

P
x
n
= (1x)
It is straightforward to show
2 by differentiating the geometric
n=0 nx
P n
1
series n=0 x = (1x) .
2
see Simons text for using the 1D partition function Z1D =
P Alternatively
~(n+1/2)
1D
and to obtain hi, because the average energy is hi = lnZ
n>0 e
.
1

1.1. EINSTEINS CALCULATION: SIMPLE HARMONIC OSCILLATOR5


Differentiating hEi with respect to temperature T and defining x~, we
then obtain the heat capacity
C = ~

x
1
e~
2
=
k
(~)
.
B
T x ex 1
(e~ 1)2

(1.3)

In the high-temperature limit, (e~ 1) approaches ~ because ex 1 + x,


we then have C = kB , consistent with the Law of Dulong-Petit.
For the 3D case,

(1.4)
Enx ,ny ,nz = ~ nx + 12 + ny + 21 + nz + 12 ,
and

hE3D i = 3hE1D i.

(1.5)

Consequently the heat capacity is


C = 3kB (~)2
10

e~
.
(e~ 1)2

(1.6)

CHAPTER 2. SPECIFIC HEAT OF SOLIDS: BOLTZMANN, EINSTEIN, AND DEBYE

Figure 2.2: Plot of Specific Heat of Diamond from Einsteins original 1907 paper. The fit is to
the Einstein theory of heat capacity. The x-axis is kB T in units of ~ and y axis is C in units of
cal/(K-mol). In these units, 3R 5.96).

2.2

Figure 1.1: C vs T, from Simons text


Debyes Calculation

Einsteins theory of specific heat was remarkably successful, but still there were clear deviations
from the predicted equation. Even in the plot in his first paper (Fig. 2.2 above) one can see that
at low
lies above the calculation
theoretical curve.reasonably
This result turnsaccurately
out
Astemperature
plotted the
inexperimental
Fig. 1.1,dataEinsteins
exto be rather important! In fact, it was known that at low temperatures most materials have a heat
plained
the
behaviorto of
the the
heat
capacity
of diamond
as ato function of
capacity that
is proportional
T 3 (Metals
also have
a very
small additional
term proportional
T which we will discuss
in section
3.2.2. Magnetic
materials
may have ,
otherthe
additional
terms frequency.
temperature
withlater
only
a single
fitting
parameter
Einstein
as well. Nonmagnetic insulators have only the T 3 behavior). At any rate, Einsteins formula at
His
result was
remarkable
told
us that
quantum
low temperature
is exponentially
smallas
in Tit, not
agreeing
at all with
the actualmechanics
experiments. is important

to correctly
explain
the temperature
dependence.
In 1912 Peter
Debye5 discovered
how to better treat
the quantum mechanics of oscillations
of atoms, and managed to explain the T 3 specific heat. Debye realized that oscillation of atoms is
the same thing as sound, and sound is a wave, so it should be quantized the same way as Planck
quantized light waves. Besides the fact that the speed of light is much faster than that of sound,
there is only one minor difference between light and sound: for light, there are two polarizations for
each k whereas for sound, there are three modes for each k (a longitudinal mode, where the atomic
motion is in the same direction as k and two transverse modes where the motion is perpendicular
to k. Light has only the transverse modes.). For simplicity of presentation here we will assume that
the transverse and longitudinal modes have the same velocity, although in truth the longitudinal
velocity is usually somewhat greater than the transverse velocity.
We now repeat essentially what was Plancks calculation for light. This calculation should
also look familiar from your statistical physics course. First, however, we need some preliminary
information about waves:

CHAPTER 1. SPECIFIC HEAT OF SOLIDS

1.2

The T 3 dependence

Einstein successfully explained the molar heat capacity of diamond, but


still there were clear deviations from the prediction at low temperatures.
For example, Fig. 1.2 shows that the low-temperature heat capacity of solid
argon below 2 K is proportional to T 3 .

Figure 9 Low temperature heat capacity of solid argon, plotted against T3. In this temperature
region the experimental results are in excellent agreement with the Debye T3law with B = 92.0 K.
of L. Finegold and N. E.heat
Phillips.)
Figure(Conrtesy
1.2: Low-temperature
capacity of solid argon plotted against T 3 ,

from Kittels

text

In 1912 Peter Debye discovered how to better treat the quantum mechanics of oscillations of atoms, and managed to explain the T 3 specific heat. In
Debyes model, we have the following assumptions:
The thermal energy results from sound wave, i.e., vibration of atoms
with long wavelength.
Sound waves in solids are quantized the same way as Planck quantized
light10waves.
Figure
To obtain a qualitative explanation of the Debye T3law, we suppose that all phonon
modes of wavevector less than K , have the classical thermal energy k,T and that modes between
K, and the Debye cutoff K, are not excited at all. Of the 3N possible modes, the fraction excited is
= (T/O)3,
because collectively
this is the ratio of thein
volume
of the inner sphere
to the outer
(KdKDJ1
Atoms
vibrate
a wave-like
fashion.
Asphere.
linear
Tne e n e r a i s U k,T . 3N(T@, and the heat capacity is C, = JU/aT= 12NkB(T/B)3.

dispersion
of frequency versus wave vector k is assumed, i.e., (k) = v|k|; here
v is the sound velocity.

1.2. THE T 3 DEPENDENCE

Vibration of atoms is isotropic; the transverse and longitudinal modes


have the same velocity, vl = vt .
There is a maximum frequency to obtain a total of 3N degrees of freedom of of the system.
Now we have the expectation of thermal energy
hEi = 3

X
k



1
.
~(k) nB (k) +
2

(1.7)

To sum over all possible values of k, we need to replace the summation


R
with an integral dk.

P
k

periodic boundary condition


Consider waves in an 1-D system of length L, e.g., 1D chain of atom
with an interaomic spacing a. If the end effects are ignored, the precise way
of treating the atoms at the ends is unimportant and we may choose the
approach on grounds of mathematical convenience. The most convenient
choice is the Born-von Barman periodic boundary condition. Any wave in
the sample eikx is required to have same value for a position x as it for x + L,
i.e., eikx = eik(x+L) . This requires eikL = 1 and then restricts k to be certain
discrete values
2n
k=
L
for n an integer. The spacing between allowed k point in k space is 2
. If
L
L  a, k is nearly continuous and we can replace a sum over k by an integral.
Because k = 2
, we have the following replacement
L
X
k

dk.

In three dimensions, we can extend this discretization of values of k for a


sample of size L3 and obtain
X
k

L3

(2)3

dk.

(1.8)

CHAPTER 1. SPECIFIC HEAT OF SOLIDS

Debyes calculation
Following Eqs. 1.7 and 1.8, we obtain the expectation of thermal energy


Z
L3
1
.
(1.9)
hEi = 3
dk~(k) nB (k) +
(2)3
2
For k = /v and using the spherical symmetry,3


Z
4L3 1 2
1
hEi = 3
d(~) nB (~) +
.
(2)3 v 3 0
2

(1.10)

Now we introduce the density of states g() of vibration such that g()d
gives the total number of vibration modes with frequency between and
+ d. Then the expectation of thermal energy can be expressed as


Z
1
,
(1.11)
hEi =
d(~)g() nB (~) +
2
0
and the density of states is
g() =

12L3 2
12 2
9 2
=
N
=
N
(2)3 v 3
(2)3 n v 3
d3

(1.12)

with N the total number of atoms and n the density of atoms. d (6 2 n)1/3 v
is known as the Debye frequency. Therefore the heat capacity is
Z
9N ~ 2
3
C =
d
d3 T 0 e~/kB T 1
Z
9N ~ 2 kB 4 4 x3
=
T
dx
x
d3 ~4 T
0 e 1
12 4 (kB T )3
= N kB
.
(1.13)
5 (~d )3
P
p
Note that one can use the Riemann zeta function (p) =
to obn=1 n
R x3
4
tain 0 ex 1 dx = 15 . The Debye frequency is often replaced by the Debye
temperature
kB T = ~d
(1.14)
so that we obtain the Debye T 3 law
C = N kB
3

Note that

dk4

R
0

k 2 dk.

12 4
T3
.
5 (TDebye )3

(1.15)

1.2. THE T 3 DEPENDENCE

Debye encountered a problem that the heat capacity does not level off
to 3kB N at high T because his approximation allowed an infinite number of
sound wave modes. To reconcile this discrepancy Debye introduced a cutoff
frequency cutoff to ensure that the number of sound wave modes is the same
as the number of degrees of freedom 3N . That is
Z cutoff
dg()
3N =
0
Z cutoff 2

d
= 9N
d3
0
(1.16)

3
cutoff
= 3N
d
Clearly the cutoff frequency is exactly the same as the Debye frequency,
i.e., cutoff = d . Note that the T 3 dependence at low temperature is still
valid even a cutoff frequency is introduced. In the high temperature limit
T  TDebye ,
kB T
1

,
(1.17)
nB (~) = ~
e
1
~
and the heat capacity is
Z cutoff
kB T

dg()~
C =
T 0
~
(1.18)
= 3N kB .
We then obtain the Dulong-Petit law.
shortcomings of Debyes theory
Debyes calculation successfully explains C3N kB at high T and CT 3
at low T . In addition, at very low temperatures, metals have a linear term
in the heat capacity and the overall specific heat is C = T + T 3 and at
low enough T the linear term dominates. Figure 1.3 shows the evidence for
the existence of the linear term. As we will discuss in the next chapter, the
linear dependence results from the contribution of electrons in metals.
Homework #1
1.1 Consider a quasi-2D layered material in which the coupling between
atoms in different layers is much weaker than that between atoms in

their measur
paramagnetic
4
posed a ne
pressure and
I
I4
l6
Q
2
4
6
8
IO
I2
I8
comparison
',
(oeg'.
)
r,
thermometer.
CHAPTER 1. SPECIFIC HEAT OF SOLIDS
FIG. 5. Atomic heat of copper.
gest that th
Figure 8 sho
tematic dev
'
'
'
4.0
'
'
'
although the
capacities, s
3.0
probable tha
2.0
temperature
.8

10

a
I
I

'D

'~ Although

1.0
ct

t,

8
IO
T', (oeg'. )

l2

FIG. 6. Atomic heat of silver.

t,

I4

I6

I8

Figure 1.3: Low-T heat capacity of Ag, from Corak et al., Phys. Rev. 98, 1699 (1955).

the abstract fo
factor of 10' too
the temperature
incorrectly in t
'6 R. A. Erick
'~ R. Berman

the same layer. Use the Debye theory to show that the heat capacity
of 2D materials is proportional to T 2 at low temperatures.

Chapter 2
Electrons in Metals
Metals are good conductors of heat and electricity. Electrons involved
with thermal and electrical conduction in metals are mobile. The conduction electrons of a metal are detached from the ionic ion and wander freely
through the metal. Between collisions the interaction of a given electron, both
with the others and the ions, is neglected. The neglect of electron-electron
interactions between collisions is known as the independent electron approximation. The corresponding neglect of electron-ion interactions is known as
the free electron approximation.

2.1

The Drude model

In 1900 Paul Drude applied Boltzmanns kinetic theory of gases to understanding electron motion within metals. The basic assumptions of the Drude
model are:
Between collisions and in the absence of external fields, conduction
electrons move uniformly in a straight line. On the average, an electron
travels for a time between two consecutive collisions. The probability
of collision per unit time is 1/ .
Immediately after each collision, an electron is randomly redirected
and with a speed appropriate the temperature prevailing at the place
where the collision occurred. Technically one can assume that once a
scattering (collision) event occurs, the electron returns to momentum
p = 0.
Between scattering events, the electrons respond to external E and B
fields, following Newtons law of motion.
11

12

CHAPTER 2. ELECTRONS IN METALS

Figure 2.1: Trajectory of a conduction


electron scattering off the ion according the naive picture of Drude. (from
Ashcroft and Mermins text)

Consider an electron with momentum p at time t and under an externally


applied force F(t). At time t+dt the probability that the electron will scatter
to p = 0 is dt/ , and the probability that the electron does not scatter to
p = 0 is (1 dt/ ); the corresponding momentum change is Fdt. Combining
these cases we then have
p(t + dt) =

dt
dt
0 + (1 )(p(t) + Fdt)

Keeping terms only up to the linear term in dt, dp = Fdtp(t)dt/ . Dividing


this by dt and taking the limit as dt0, we then obtain the equation of motion
dp(t)
= F(t) p(t)/.
dt
The effect of individual electron collisions induces a fractional damping term
providing a drag force to the electron. In the absence of external field F = 0,
the electron momentum exponentially decays
p(t) = p0 et/ .
electrical conductivity
If electrons are under an electric field E, then F = eE with the electron
charge e. In steady state dp/dt = 0, we have
p = mv = e E,
with m the electron mass and v its velocity. For electrons of density n, the
electrical current is
e2 n
j = env =
E.
m
The electrical conductivity 0 , defined via j = 0 E, is then
0 =

e2 n
.
m

(2.1)

From Eq. (2.1) we can extract the Drude scattering time to be in the range
of 1014 seconds for most metals near room temperature.

2.2. FREE ELECTRON FERMI GAS

2.2

13

Free electron Fermi gas

For many years the electronic velocity distribution in solids was given in
equilibrium at temperature T by the Maxwell-Boltzmann distribution

fB (v) = n

m
2kB T

3/2

emv

2 /2k T
B

In conjunction with the Drude model this leads to a wrong prediction


that the contribution of an electron to the specific heat of a metal is 32 kB .
Later Sommerfeld generalized Drudes theory of metals to incorporate Fermi
statistics which satisfies the Pauli exclusion principle.
Fermi-Dirac distribution
To obtain the expression of the Fermi-Dirac distribution, we begin with
an N -electron system in thermal equilibrium at finite temperature.1 The
probability PN (E) of finding the N -particle system of energy E is proportional
to the Boltzmann factor eE/kB T , i.e,
PN (E) =
1

eE/kB T
,
ZN

ZN =

eE /kB T ,

The proof shown below is from Ashcroft and Mermins text.

Figure 2.2: Illustration of 4 different N -electron stationary states. The filling of N oneelectron states satisfies the Pauli exclusion principle. This illustration is an example of
N = 3.

14

CHAPTER 2. ELECTRONS IN METALS

where ZN is the partition function and is related to the Helmholtz free energy
FN as2
ZN = eFN /kB T
(2.2)
Thus the probability PN (E) is
PN (E) = e(EFN )/kB T .

(2.3)

Now consider a system of N electrons occupying N one-electron states. Assuming that the energy of an accessible N -electron state is labeled E , the
probability fiN of there being an electron in the ith one-electron level is simply
the sum of the independent probabilities PN (EN ) of finding any N -electron
system in which the ith electron state is occupied, i.e.,
X
X
N
fiN =
PN (EN ) =
e(E FN )/kB T ,
(2.4)

where the summation is over all N -electron states .


The expression of fiN can be obtained through the following three steps.
1. The Pauli exclusion principle requires that any one-electron state is
either occupied or unoccupied. If the probability of finding an N electron state in which no electron being in the ith one-electron state
is PN (EN ), we could equally well write Eq. 2.5 as3
X
PN (EN )
(2.5)
fiN = 1

2. An N -electron state in which there is no electron in the one-electron


state i of energy i can be constructed by removing the electron in the
ith one-electron state of any (N + 1)-electron state in which there is an
electron in the ith one-electron state, i.e.,
X
fiN = 1
PN (EN +1 i ).
(2.6)

Using Eqs. 2.3, we have


N +1

PN (EN +1 i ) = e(E

i FN )/kB T

N +1
= e(i FN +1 +FN )/kB T e(E FN +1 )/kB T
= e(i )/kB T PN +1 (EN +1 ),

(2.7)

For S being the entropy of the system, the Helmholtz free energy FN is defined as
FN E T S = kB T lnZN .
P
P
3
The Pauli exclusion principle requires PN (EN ) + PN (EN ) = 1

2.2. FREE ELECTRON FERMI GAS

15

where = FN +1 FN is known as the chemical potential. Therefore


we could rewrite Eq. 2.6 as
X
PN +1 (EN +1 ) = 1 e(i )/kB T fiN +1 , (2.8)
fiN = 1 e(i )/kB T

3. When N is very large (of the order of 1023 ), fiN +1 can be replaced by
fiN , giving rise to
1
fiN = (i )/k T
(2.9)
B
e
+1
Now we have proved the Fermi-Dirac distribution. In summary, given a
system of free electrons with chemical potential , the probability of an
eigenstate of energy E being occupied is given by the Fermi function
f (, , T ) =

1
e()/kB T

+1

(2.10)

At T = 0 the Fermi function becomes a step function and the chemical potential defined is the Fermi energy. States below the chemical potential are
filled and those above the chemical potential are empty, whereas at higher
temperatures the Fermi function become more smeared out. For materials
with an energy gap, the chemical potential is precisely halfway between the
highest-energy occupied eigenstate and the lowest-energy unoccupied eigenstate. (Homework #?).
1.2
1.0

= 5 eV

f()

0.8

Temperature (kBT)
1000 K (86.2 meV)
300 K (25.8 meV)
100 K (8.62 meV)
0K

0.6
0.4
0.2
0.0
4.6

4.8

5.0
Energy (eV)

5.2

5.4

Figure 2.3: Sketch of the Fermi function for various temperatures and values of kB T .

16

CHAPTER 2. ELECTRONS IN METALS

The free electron model


Much of solid-state physics is determined by the Hamiltonian of the system
N
X
1 X ql q l 0
Pl

+
.
(2.11)
H=
2m
2
r
l
l rl 0
0
l=1
l6=l

The summation runs over all electrons and nuclei of charge ql at position rl
in solids. However, it cant be solved without approximations because of the
enormous number of particles in solids. The simplest approximation is the
free electron gas model. This model only considers noninteracting electrons
which freely move about, subject only to the Pauli exclusion principle no
two electrons occupy the same quantum mechanical state. Since electrons are
assumed to have no interactions, the one-electron wave function associated
with an energy level satisfies

~2 2
(r) = (r).
2m

(2.12)

The eigenfunction of a many-electron system is simply the product of the


one-electron wave functions. The energy of the many-electron system is the
sum of all one-electron energies. Assume that electrons are confined in a
square box of side length L, L3 = V . We then have the Born-von Karman
boundary condition
(x + L, y, z) = (x, y, z),
(x, y + L, z) = (x, y, z),
(x, y, z + L) = (x, y, z).

(2.13)

The eigenfunction of the Schrodinger equation Eq. 2.12 is


1
k (r) = eikr ,
V

(2.14)

with energy

~2 k 2
.
2m
Now we invoke the boundary condition Eq. 2.13 and obtain
(k) =

eikx L = eiky L = eikz L = 1.

(2.15)

(2.16)

This permits certain discrete values of k. In a 3D space with Cardtesian axes


kx , kx and kx (known as k-space), the allowed ks are those whose coordinates

2.2. FREE ELECTRON FERMI GAS

17

along the three axes are given by integral multiples of 2/L. The wave vector
k must be of the form
2
(2.17)
k = (kx , ky , kz ) = (nx , ny , nz ) ,
L
with nx , ny , nz integers. In k-space, if there is a very large number of states,
the volume of each k point is (2/L)3 . Therefore a region of k-space of
volume will contain

V
= 3
(2.18)
3
(2/L)
8
allowed value of k. The k-space density of states per unit volume is 1/8 3 .
In building up the N -electron ground state, we begin with by placing
two electrons in the one-electron level k = 0 for spin up and spin down,
which has the lowest possible one-electron energy = 0. We then continue
to add electrons, successively filling the one-electron levels of lowest energy
that are not already occupied. Since is proportional to k 2 and N  1, the
occupied region will be distinguishable from a sphere (the Fermi sphere) with
a radius called kF (the Fermi wave vector ) and its volume is = 4kF3 /3.
The surface of the Fermi sphere is known as the Fermi surface, the energy
surface in k-space dividing filled from unfilled states at zero temperature.
The total number of electrons is
k3
4k 3 /3
(2.19)
N = 2 3F = F2 V
8 /V
3
and the electron density
N
k3
= F2 ,
(2.20)
V
3
i.e., kF = (3 2 n)1/3 . Since mv = ~k, we then have the Fermi velocity
n=

vF =

~kF
~(3 2 n)1/3
=
.
m
m

Figure 2.4: The ground state of the


free electron gas is constructed by
occupying a sphere of states in kspace, whose radius is kF . (from
Marders text)

(2.21)

18

CHAPTER 2. ELECTRONS IN METALS

In addition, for a metal, the Fermi energy is the highest occupied one-electron
energy
~2 kF2
2m
~2 (3 2 n)2/3
=
.
2m

F =

2.3

(2.22)

Electronic heat capacity

The heat capacity of electronic conduction caused a great difficulty in the


early development of the theory of electrons in metals. In the viewpoint of
classical statistical mechanics one would expect that a free electron should
have a heat capacity of 32 kB and the electronic contribution to the heat capacity should be 32 N kB for a system of N atoms and each atom giving one
electron which freely moves around. However the observed electronic contribution at room temperature is often less than 1% of this value.
This puzzle can be reconciled by including the Pauli exclusion principle
and the Fermi distribution function. When a free electron gas is heated
from 0 K, not every electron gains an energy kB T as expected classically,
but only those within an energy range kB T about the Fermi surface are
thermally excited. That is, a fraction kBFT of the total electrons are excited
and each excited electrons has an energy kB T . So total electronic thermal
energy is
kB T
N kB T,
Eel
F
and thus the heat capacity is
C=

2
Eel
N kB
T,

T
F

directly proportional to T .
To exactly calculate the heat capacity of electrons in a metal, we first
find out the temperature dependence of the total electronic energy by using
the Fermi function. The total energy is
X
Eel = 2
(k)f (, , T )
(2.23)
k
R
V
Replacing the summation by (2)
dk,
3
Z
2V
Eel =
dk (k) f (, , T )
(2)3
Z
2V
=
4
k 2 dk (k) f (, , T ).
(2)3
0

2.3. ELECTRONIC HEAT CAPACITY


Because k =

19

p
2m/~2 , we have
r
Z
Z
m
dk = d
,
2~2

and
Eel

Z r
2V
m 2m
=
d
4
f (, , T ),
3
(2)
2~2 ~2
0
Z
(2m)3/2
= V
d
f (, , T ),
2 2 ~3
0
Z
= V
g()f (, , T )d,
0

where
(2m)3/2
=
g()
2 2 ~3

2m
~2

3/2

2 2

is the density of states per unit volume, giving the number of eigenstates
(including both spin states) with energies between and + d. Note that
the density of states can also be expressed as
r
3n

g() =
,
(2.24)
2F F
because
F =

~2 (3 2 n)2/3
.
2m

and

2m
(3 2 n)2/3
=
.
~2
F
Similarly the total number of states is
Z
N =V
g()f (, , T )d.
0

Assuming that the system is heated from 0 to T and the temperature is


sufficiently low, kB T F , we now calculate the increase in the total energy
of an N -electron system
4U = U (T ) U (0)
= V

g()f (, , T )d V

g()d,
0

(2.25)

20

CHAPTER 2. ELECTRONS IN METALS

since the Fermi function f (, , T ) is a step function at T=0. In addition,


the total number of electron is conserved during the change of temperature,
i.e., N (T ) = N (T = 0). We obtain the identity
Z
Z F
F
g()f (, , T )d = F
g()d.
0

With this identify, Eq. 2.25 can be rewritten as


Z F
Z
g()f (, , T )( F )d V
g()( F )d.
4U = V
0

(2.26)

The next step is to find U/T . The only temperature-dependent term in


Eq. 2.26 is f (, , T ), so the heat capacity is
Z
df (, , T )
dU
g()
=V
( F )d.
C=
dT
dT
0

For low temperature, f is a step-like function and df /dT is large only at


energies near F . A good approximation to the calculation of heat capacity
is to replace and g() in the integrand by F and g(F ), respectively, and
take g(F ) outside of the integral, i.e.,
Z
df
d ( F ).
C = V g(F )
dT
0
Note that, for F ,

F
df
e(F )/kB T
.

dT
kB T 2 (e(F )/kB T + 1)2

With x ( F )/kB T , we then have


Z
2
C = V g(F )kB T

F /kB T

x2 e x
dx
.
(ex + 1)2

(2.27)

(2.28)

From Eq. (2.24) we know that g(F ) = 3n/2F , so4


Z
3N 2
x2 ex
C
kB T
dx
2F
(ex + 1)2

2 N kB kB T
(2.29)
2
F
Therefore, at low temperatures, electronic heat capacity is linearly dependent
on T . The overall low-temperature specific heat including the contributions
of lattice vibration and electron interaction is C = T + T 3 , and at very
low enough T the linear term dominates.
=

2 x

e
dx (exx +1)
2 =

2
3

2.4. SCREENING AND THE MOTT TRANSITION

2.4

21

Screening and the Mott transition

If a positive charge q is introduced into an electron gas, the surrounding


electrons tend to gather around and the electric field of the positive charge
falls off with increasing r faster than 1/r. In other words, the electric field
of the charge q is screened and there is a perturbation in the electron concentration in the vicinity of this charge, resulting in a local perturbation
potential which produces a local raising of the density of states parabola
g() by e. If one imagines to be switched on, some electrons will immediately leave this region in order for the Fermi level to constant throughout
the crystal.
Assuming e||  F , the change in the electron density is given region
is
n(r) = g(F )e|(r)|.
From the Poisson equation 2 (r) = 4(r) which (r) is the charge
density, we have
2 (r) = 4(e)n(r) = 4e2 g(F )(r).
Defining 2 4e2 g(F ), the Poisson equation is simplified to

2 2 (r) = 0.

(2.30)

e
g()

g()

Figure 2.5: Effect a local perturbation potential (r) on the density of states of a free
electron gas (from Ibach and Luth).

22

CHAPTER 2. ELECTRONS IN METALS

To solve this differential equation, one can express (r) in terms of Fourier
transform,
Z
1
ikr

dk (k)e
..
(2.31)
(r) =
(2)3
Combining Eq. (2.30) and the Fourier transform of (r) one can obtain

(k).
Substituting (k)
into Eq. (2.31), we then have5
(r) =

er
r

(2.32)

with being a constant. If there is no screening, the electric potential is q/r.


That is, the boundary condition requires (r) qr as 0. The solution
of the perturbation potential is
er
(r) = q
.
r
This is the screened Coulomb potential of conduction electrons. The quantity 1/ is known as the Thomas-Fermi
screening length rT F .
q
2
2 2/3
3n

Recall that g() = 2F F and F = ~ (32mn) . So we have


2 = 4e2 g(F )

3n
2m
2
2 ~ (3 2 n)2/3
 1/3  1/3
3
n
= 4
,

a30

= 4e2

~2

where a0 = me
2 0.53 A is the Bohr radius. The Thomas-Fermi screening
length is rT F
 3 1/6
a
rT F 0.5 0
n

For example, Cu has an electron concentration of n = 8.5 1022 cm3 and a


screening length of rT F = 0.55
A.
Although the above derivation of the screened Coulomb potential is not
rigorous, this screening process explains the fact hat the valence electrons of
highest energy in a metal are not localized. For a metal of high electron concentration, the Coulomb potential is well screened and valence electrons are
5

Homework: Prove that (k)

1
(k2 +2 )

and (r) = e r .

2.4. SCREENING AND THE MOTT TRANSITION

23

1
r
r

e
r

Figure 2.6:
screened potential
(from Ibach and Luth)

itinerant and conducting. With the decrease of electron density, the screening length increases, i.e., decreases. Below a critical electron concentration
the potential of the screened field extends far enough for a bound state to be
possible. The electron then is localized in a covalent or ionic bond. When
a bound state becomes possible in a screened potential, the screening length
must be significantly larger than the Bohr radius a0 ,
1/2

rT F

1 a0
 a0 .
2 n1/6

i.e.,
n1/3  4a0 .
This hand-weaving argument, originally proposed by N. Mott, predicts that,
when the average electron separation n1/3 is significantly larger than 4a0 , a
metal will lose its metallic character and an abrupt transition to insulating
properties is expected. This is the so-called Mott transition.

24

CHAPTER 2. ELECTRONS IN METALS

Figure lob Sclnilog plot of observed "zero temperature" condiictivity m(D) versus donor concentration n for
phosphorous donors in silicon. (After T F. Roscnbamii,
ct al.)
Figure 2.7: Mott transition (from

n, in 1018 cm-3

Kittels text)

where k,2 = 3.939nhi3/ao,as in (34), where no is thc electron concentration. At


high concentrations k, is large and the potential has no hound statc, so that we
must have a metal.
The potential is known to have a bound state when k, is smaller than
For a metal at sufficiently high temperature, it will emit conduction elec1.19/ao. With a bound state possible the electrons may condense about the
tronsprotons
from the
surface. Lets consider a vacuum tube which consists of a
to form an insulator. The inequality rnay be written in terms of no as

2.5

Thermionic emission

cathode and an hot anode. If the cathode is connected to the anode through
biased voltage and a current meter, one can observe a temperature-dependent
saturation
in athe
current-voltage
charateristic.
This phenomenon
With n,current
= l/a3 for
simple
cnhic lattice, wc
may have an insulator
when a, > is
called2.78a0,
thermionic
can result
use the
free
electron
gas model
which is emission.
not far fromOne
the Mott
4.5ao
found
in a diffcrcnt
way. to calculate the
current. transition
Assuming
drift situations
velocity where
of charge
Thethermionic
terrn metul-insulator
has that
come the
to denote
N
thewith
electrical
conductivity
rriaterial
from
metalsurface
to insulator
as a the
andchanges
that the
metal
is along
carrier
a density
n = Vofisa v(k)
function
of
some
external
parameter,
which
may
be
composition,
pressure,
x-axis, the current density is

strain, or magnetic field. The metallic phase may usually be pictured in terms
X
of an independent-clcctron model; the
insulator phase may suggest important
Js Sites
= erandomly
n vxoccupied
(k),
electron-electron interactions.
introduce new and ink
teresting aspects to the problem, aspects
that lie within percolation theory.
The percolation transition is beyond the scope of our book.
2
When
a semiconductor
is doped
with requirement
increasing concentrations
where the
summation
is subject
to the
of ~k
> ofF donor
+ and
2m
(or
acceptor)
atoms,
a
transition
will
occur
to
a
conducting
metallic
phase.
v > 0. Since all states of the free-electron gas are doubly degenerate and
x

2.5. THERMIONIC EMISSION

25

mv = ~k,

2e~
=
(2)3 m

Js

dk f (, , T ) kx .
2

When the work function is much larger than kT , we have f emv /2kB T
and
Z
 2 2

k
e~
~2m
F /kB T
dk kx e
Js =
4 3 m


Z
Z
Z
~2 k 2
2mx F /kB T
e~
~2 ky2 /2mkB T
~2 kz2 /2mkB T
.
dky e
dkz e
dkx kx e
=
4 3 m

kmin
Because

dxe

ax2

we have
Js

ekB T
=
2 2 ~

dkx kx e

2
~2 kx
F
2m

,
a


/kB T

kmin

 Z
ekB T 2mkB T F /kB T 1
2
d(x2 )ex
=
e
2
2
2
2 ~ ~
2
x=~2 kmin
/2mkB T

4me
2 2
(kB T )2 e(~ kmin /2mF )/kB T
3
h

2
4mekB
=
T 2 e/kB T
3
h
This is the so-called Richardson-Dushman formula for the saturation current
density.
Note that there are two corrections one needs to make for the RichardsonDushman equation. First it is incorrect to assume that electrons at the
2
> F + have a probability of 100% to escape from
surface with an energy ~k
2m
the solid. The transmission of the potential barrier needs to treated quantum
mechanically. Second, the work function is not constant; an electric field E
and the image charge
of an electron lying outside the surface will reduce the
work function by e3 E. Hence we can express the saturation current density
as

Js = T 2 e(

e3 E )/kB T

where is a constant. Through plotting Js /T 2 versus 1/T , one can determine


the work function of metal.

26

6.6 Thermionic
Emission IN
of METALS
Electrons from Metals
CHAPTER
2. ELECTRONS

6.12.
(a) Schematic
a diode
circuittube
for for
observing
FigureFig.
2.8: (a)
Schematic
illustration ofdrawing
an electricof
circuit
and a vacuum
observing thermionic
electrons
from
heated cathode
(A =
anode).
(b) Qualitative behavior of
thermionic
emission,
and the
(b) quantitative
results. C
(from
Ibach
and Luth)

em
6.6 Thermionic Emission of Electrons from Metalsthe
voltage curve at two dierent temperatures T1 and T2 > T1. As a consequence of th
mal energy, electrons can even overcome a countervoltage (A negative with respect

The existence of this eect demonstrates that the assumption of


nite square well to describe metal electrons is too simple. The potent
clearly has only a finite barrier height. The energy dierence EvacEF
known as the work function. This is the energy barrier that an electro
overcome in order to reach the energy level of the vacuum (far awa
the metal) from the ``Fermi sea''. If the electron also possesses a su
momentum perpendicular to the surface, it can leave the metal and w
tribute to the saturation current js.
We will calculate the temperature-dependent saturation current
free-electron-gas model. If the drift velocity v of the charge carriers is
geneous, the current density is given by j = n e v, where n is the concen
of Fig.
charge
speaking
minus signemission
is alsoofrequired,
but t
6.13.carriers.
Schematic(Strictly
representation
of thea thermionic
free electrons
Figure 2.9: Schematic representation of the thermionic emission of free electrons from a
benomitted
for
our present purposes.) We can generalize this expression
) from
metal.
metal. (from
Ibach a
and
Luth)An electron in the potential well must overcome the work function
in electron
order to velocity
reach the is
energy
level Evac
vacuum
and kescap
U =where
EvacEFthe
case
a function
of of
thethewave
vector
:
the crystal. An important part of the work function is assumed to be the C
X
e electron and its positive image charge in the metal
Homework
#2ebetween the escaping
potential
j

v
k

visx k
dk : then U is reduced by an amou
x
x
.
3
potentialV ). If an external
electric
field
applied,
2p
2.1 Exercise (3.2) kof Simons text.

Reductions of the work function E>E


of *1
eV as shown here can only be achieved wit
F U
8
v x k>0
V/cm
mely strong external fields of 10710
2.2 Exercise (4.6) of Simons text.

This form includes the fact that the density of states in k-space is V
fieldthe
in reducing
the potential
barrier. This
is illustrated
the superpo
Both
summation
and the integral
extend
only overbyoccupied
st
of the external
potential
x and thecan
Coulomb
image poten
dictated
by Fermiapplied
statistics.
This Econdition
be included
explic
shown in Fig.
6.13.
multiplying
by the
occupation probability given in (6.33). Thus
The Richardson-Dushman
formula can be used in this extended fo
1
1

determine
2 ethe
h work functions of metals. In order to do this, one mu

j

dky dkz emission


dkx kcurrent
: E = 0 by extrapolati
x
x fEk;j Tfor
determine the
saturation
s0
2 p3 m
2

2.5. THERMIONIC EMISSION

27

2.3 Exercise (4.7) of Simons text.

2.4 Prove (k)

1
(k2 +2 )

and Eq. (2.32) (r) = e r .

2.5 Show that in the presence


of an electric field E, the work function of
a metal is reduced by e3 E. (The charge of electron is assumed to be
e.)

28

CHAPTER 2. ELECTRONS IN METALS

Chapter 3
Vibrations of One-Dimensional
Lattices
In the first two chapters we discuss the thermal properties of solids without considering their microstructure seriously. Our simple models of solids,
and electrons in solids, are insufficient for correctly understanding the physics
of solids. In order to improve our understanding, we now take the periodic
crystal structures into account. To get a qualitative understanding of the
effects
of the periodic lattice,
we start with simple one dimensional systems.
58
CHAPTER 5. THE ONE-DIMENSIONAL MONATOMIC SOLID
We will be able to better understand the shortcomings of Einstein and Debye
models Let
of vibrations
in solids.
us consider a chain
of identical atoms of mass m where the equilibrium spacing between
atoms is a. Let us define the position of the nth atom to be xn and the equilibrium position of the
nth atom to be xeq
n = na.
Once we allow motion of the atoms, we will have x deviating from its equilibrium position,
3.1
One-dimensional monatomic solids
so we define the small variable
n

xn = xn xeq
n

Forthat
simplicity,
lets
imagine
true one-dimensional
of dimension
atoms of(i.e.,
Note
in our simple
model
we are aallowing
motion of the massessystem
only in one
mass
mallowing
wherelongitudinal
there is only
longitudinal
motion
and the equilibrium
we are
motionthe
of the
chain, not transverse
motion).
spacingAsbetween
is a. Define
nth atom
to be x
discussed atoms
in the previous
section, the
if theposition
system is of
at the
low enough
temperature
wen can
th
eq
consider
the potential holding
the atoms
to be to
quadratic.
our model
of a solid is
and
the equilibrium
position
of thetogether
n atom
be xn Thus,
= na.
The Taylor
chain of masses held together with springs as show in this figure

With this quadratic interatomic potential, we can write the total potential energy of the
29
chain to be
X
Vtot =
V (xi xi+1 )
i

= Veq +

X
i

(xi xi+1 )2

The force on the nth mass on the chain is then given by


Vtot

30

CHAPTER 3. VIBRATIONS OF ONE-DIMENSIONAL LATTICES

expansion of the potential V (x) between the atoms around its minimum is
3

V (x) V (xeq ) + (x xeq )2 + (x xeq )3 + . . . .


2
3!
In addition, atoms move with their position deviating from their equilibrium
position by
xn = xn xeq
n .
With this quadratic interatomic potential, we can write the total potential
energy of the chain to be
X
X
Vtot = Veq +
V (xi+1 xi ) = Veq +
(xi+1 xi )2 .
2
i
i
The force on the nth mass on the chain is then given by
Fn =

Vtot
= (xn+1 xn ) + (xn1 xn ).
xn

Thus we have the equation of motion


m(
x) = Fn = (xn+1 + xn1 2xn )
Note that, for any coupled system system, a normal mode is defined to be a
collective oscillation where all particles move at the same frequency. We now
find solutions to the equation of motion by using a trial solution (ansatz)
that describes the normal modes as waves
eq

xn = Aeitikxn = Aeitikna .

(3.1)

Plugging the ansatz into the equation of motion, we obtain




m 2 Aeitikna = Aeit eika(n+1) + eika(n1) 2eikan ,
i.e.,

m 2 = 2[1 cos(ka)] = 4 sin2 (ka/2).


Therefore the dispersion relation between frequency and wave vector k is
r  

ka
sin
.
(3.2)
=2
m
2

m 2 Aeitikna = Aeit eika(n+1) + eika(n1) 2eikan


or
m 2 = 2[1 cos(ka)] = 4 sin2 (ka/2)
We thus obtain the result
=2

(5.5)

 

ka
sin
m
2

(5.6)

In general a relationship between a frequency (or energy) and a wavevector (or momentum) is
known as a dispersion relation. This particular dispersion relation is shown in Fig. 5.2

3.2. RECIPROCAL LATTICE

31


5676589
 
2
3
3
32
3


2
01

02

03

45676589
3

Figure 3.1: Dispersion relation between and k for a 6 k 6

Figure 5.2: Dispersion relation of the 1d harmonic


. chain. The dispersion is periodic in k k+2/a

3.2

First Exposure to
the Reciprocal Lattice
Reciprocal
lattice

5.2.1

Note that in the figure Fig. 5.2 we have only plotted the dispersion for /a 6 k 6 /a. The

reason 3.1
for this
is obvious
Eq. 5.6
the dispersion
relation is for
actually
k+2/a.
Figure
plots
thefrom
vs.
k dispersion
relation
aperiodic
6 k in6k
because
a
In fact this is a very important general principle:
the system is periodic in real space with a periodicity a, corresponding to a
. It is obvious that the energy
reciprocal space (k space) with periodicity 2
a
and wave functions of vibrations are invariant under a kk + 2
translation
a
in reciprocal space, because
ei2n = 1

and
xn = Aeiti(k+2/a)na = Aeitikna .
In fact if we shift k by any vector 2
p with p an integer, we will have exactly
a
the same wave as well. Therefore 2
m in k-space (reciprocal space) with
a
m an integer are all physically equivalent to the point k = 0. This result
is quite profound. One must notice the statement that k and k + 2
m are
a
equivalent in describing lattice vibrations is valid ONLY if one measures the
wave at lattice point xn = na and not at arbitrary points x along the axis.
This phenomenon, that two waves with different wavelengths will appear the
same if they are sampled only the lattice points, is known as aliasing of
waves. See Fig. 3.2. If the physical waves is only defined at lattice points,
the two waves are fully equivalent.
A set of points Gm = 2
m are defined as reciprocal lattice, corresponding
a
to the real-space lattice of lattice constant a, the original periodic set of points

32

CHAPTER 3. VIBRATIONS OF ONE-DIMENSIONAL LATTICES

1.0

xn / A

0.5

n
k = (1/6) 2/a
k = (7/6) 2/a

0.0

-0.5

-1.0
0

1a

3a

2a

4a

5a

7a

6a

Figure 3.2: Plot of atom displacements x versus position x. These displacements are
2

described equally well by eitikx and eiti(k+ a )x for x = na. Here k = 3a


.

xn = na, i.e.,
real spacelattice xn =
reciprocal lattice Gn

...

2a,

a,

0,

a,

2a,

...

       
2
2
2
2
= ... 2
,
, 0,
,2
a
a
a
a

One important property of the reciprocal lattice points Gm and the real-space
lattice points xn is
eiGm xn = 1.
(3.3)
We also have the following terminologies:
Brillouin Zone: the periodic unit (the unit cell) in k-space.
First Brillouin Zone: a unit cell in k-space centered around the point
k = 0, i.e., /a6k < /a
Brillouin-Zone boundary: k= /a, points which are symmetric around
k = 0 and are separated by 2/a.

3.2. RECIPROCAL LATTICE

33

Sound waves
In the acoustic branch, the frequency is linearly proportional to wave
vector in the small k regime (long wavelength)
r

ka.
=
m
p
Sound wave is a vibration of long wavelength, and its velocity is m
a. In
contrast, the dispersion is no longer linear at large k. This is in disagreement
with what was assumed in Debyes model. At large k, the group velocity
d/dk, the speed at which a wavepacket moves, differs from the phase velocity
/k.
At the zone boundary k= /a, the displacement <[xn ] at time t is
described as


(3.4)
<[xn ] = < Aeitikna = A(1)n cos t,

so the alternate atoms oscillate in opposite phases. This is consistent with


that the group velocity becomes zero at the Brillouin zone boundaries k= /a
(i.e., the dispersion is flat), implying that no energy is propagated. In
other words a sound wave whose wavelength is twice the lattice spacing,
i.e., = 2a, will experience reflection from the periodic array of atoms, so
that the sound wave becomes a standing wave made up of two waves going
in opposite directions.
Number of normal modes
Assume that there are exactly N atoms and use the Born-von Barman periodic boundary condition eikN a = 1, This requirement restricts the possible
values of k to be of the form
2p
2p
=
.
k=
Na
L
where p is an integer and L is the total length of the system. Thus k becomes
discrete rather than a continuous variable; the spacing between two of these
consecutive points being 2/N a,
2p
2p
=
.
k=
Na
L
Thus the total number of normal modes is
Range of k the 1st BZ
2/a
=
= N.
space between neighboring
2/N a
As insightfully predicted by Debye in order to cut off his divergent integrals
in calculating heat capacity, there is precisely one normal mode per degree
of freedom in the whole system.

34

CHAPTER 3. VIBRATIONS OF ONE-DIMENSIONAL LATTICES

3.3

Diaomic chain

Consider a model
Chapter
6 of periodic arrangement of two different types of atoms
with masses m1 and m2 . The springs connecting the atoms have spring
constants 1 and 2 also alternate. We assume that m1 = m2 and identify
a unit cell with a length a known as the lattice constant. Note that the
definition of the unit cell is extremely non-unique.1 We could just as well
have chosen (for example) the unit cell shown by the box of dotted line in
Fig. 3.4. One can pick some reference point inside each unit cell to define a
lattice. We can set the reference lattice point rn in unit cell n as

The one-dimensional diatomic


solid
rn = na.

In the previous chapter we studied in detail a one dimensional model of a solid where every atom is

Analogous
the
case
of However,
monoatomic
chain, we
of
identical to to
every
other
atom.
in real materials
not can
every have
atom isthe
the equations
same (for example,
in
Sodium
Chloride,
we
have
two
types
of
atoms!).
We
thus
intend
to
generalize
our
previous
motion for a diatomic chain
discussion of the one dimension solid to a one dimensional solid with two types of atoms. Much of
this will follow the
outline
setin
thenprevious
but
we
willx
see
m(
xn ) =
xn )chapter,
+ 1 (y
2 (y
n1
n )that several fundamentally
new features will now emerge.

m( yn ) = 1 (xn+1 yn ) + 2 (xn yn ),

(3.5)

where xn and yn are displacements of atoms #1 and #2 in the unit cell,


6.1 Diatomic Crystal Structure: Some useful definitions
respectively. For trial solutions
itikna
Consider the following
x model
= A esystem
n

and yn = Ay eitikna ,

A unit cell is the repeated pattern which is the elementary building block of a periodic
crystal.

Fig. 6.1.1
1

68

m1

m2
m
m2
CHAPTER
6. THE1ONE-DIMENSIONAL
DIATOMIC SOLID

and it is labeled a.
Figure 3.3: Diatomic chain structure.
Which represents a periodic arrangement of two different types of atoms. Here we have
given them two masses m1 and m2 and the springs which alternate along the 1-dimensional chain.
The springs connecting the atoms have spring constants 1 and 2 and also alternate.
a
In this circumstance with more than one type of atom, we first would like to identify the
so-called unit cell which is the repeated motif in the arrangement of atoms. In this picture, we have
put a box around the unit cell. The length of the unit cell in 1d is known as the lattice constant
Fig. 6.1.2
67

Figure
Diatomicofchain
structure
with a unit
cell
Note however,
that 3.4:
the definition
the unit
cell is extremely
non-unique.
We could just as
well have chosen (for example) the unit cell to be as follows.

Fig. 6.1.3

3.3. DIAOMIC CHAIN

35

it is straightforward to obtain

 


Ax
1 + 2
1 eika 2
Ax
2
m
=
.
Ay
1 eika 2
1 + 2
Ay
So the dispersion is
=

1 + 2
1

m
m

q
21 + 22 + 21 2 cos(ka).

!+ =

2(1 + 2 )
m

frequency

optical

!+ =

! =

acoustic

-/a

wave vector k

(3.6)

2
m

1
m

/a

Figure 3.5: Dispersion of a diatomic chain.

Figure 3.5 plots the -vs.-k dispersion for the lattice vibration of this
diatomic chain system. There are two modes,2 acoustic mode ( ) and
optical mode (+ ), in the dispersion. The definition of an acoustic mode is
any mode that has linear dispersion as k 0. For small k, the frequency of
acoustic phonons is3
r
1 2
ka

2m(1 + 2 )
and the sound velocity is

vsound =
2

a2 1 2
.
2m(1 + 2 )

(3.7)

In three dimension, if there are n atoms per unit cell, there will be 3(n 1) optical
modes but always 3 acoustic modes.
2
3
Use cos x 1 x2 for x  1.

36

CHAPTER 3. VIBRATIONS OF ONE-DIMENSIONAL LATTICES

In addition, as k 0, the eigenvalue equation Eq. 3.6 takes the simple form





1 + 2
Ax
1 1
Ax
2
=
.

Ay
1 1
Ay
m
Therefore the acoustic mode has a zero frequency at k = 0 and its eigenvector,
i.e., amplitude,
6.2. NORMAL
MODES OFisTHE DIATOMIC SOLID
73

  
Ax
1
6.2. NORMAL MODES OF THE DIATOMIC
73
=SOLID ,
Ay
1
implying that both atoms in the unit cell move together in the same direction
as depicted in Fig. 3.6.
Fig. 6.2.2
2

A1 long wavelength
2
1 mode
acoustic

Fig. 6.2.2

A long wavelength acoustic mode


2(1 +2 )
On the other hand,
the3.6:
optical
mode
at k = 0, having
frequency
, has the
Figure
A long
wavelength
acoustic
mode. 2 =
m
eigenvector

 

pfrequency 2 = 2(1m+2 ) , has the
Ax at k = 0,1 having
On the other hand, the optical mode
= goes to
For
the
optical
mode,
its
frequency
2(1 + 2 )/m at k = 0,
eigenvector
 Ay   1 
A
1
and
itsdescribed
eigenvector
x
which
the twoismasses in 
the unit 
cell moving
in opposite
directions, for the optical mode.

=
Ay figure that
1
This is depicted in Figure 6.2.3. Note A
in xthe
1 the amplitude of the compression is slowly
=
.almostdirections,
modulated,
but the
always
two in
atoms
cell1
move
exactly the
way.
which
described
two the
masses
the unit
cellunit
moving
in opposite
for opposite
the optical
mode.
Ainy the
This is depicted in Figure 6.2.3. Note in the figure that the amplitude of the compression is slowly
modulated, but always the two atoms in the unit cell move almost exactly the opposite way.

Fig. 3.7 shows that both masses in the unit cell move in the same direction.

Fig. 6.2.3
2

A1 long wavelength
2
1
opticalmode

Fig. 6.2.3

Figure 3.7: A long wavelength optical mode.


A long wavelength optical mode
In order to get a better idea of how motion occurs for both the optical and acoustic modes, it
is useful to see animations, which you can find on the web. Another good resource is to download the
In order
to get afrom
better
idea of how
occurs
for both
the optical and acoustic modes, it3
program
ChainPlot
Professor
Mikemotion
Glazers
web site
(http://www.amg122.com/programs)
is useful to see animations, which you can find on the web. Another good resource is to download the
In this example we had two atoms per unit cell and we obtained two modes per distinct
program ChainPlot from Professor Mike Glazers web site (http://www.amg122.com/programs)3
value of k. One of these modes is acoustic and one is optical. More generally if there are M atoms
In cell
this(in
example
we had
two
perMunit
cell per
anddistinct
we obtained
two
per
distinct
perEinstein
unit
one dimension)
we atoms
willofhave
modes
value
of k,modes
of which
one
mode
used
the
concept
harmonic
oscillator
to calculate
the
heat
value
of acoustic
k. One of
thesetomodes
is acoustic
optical.
generally
if there
are M (do
atoms
will be
(goes
zero energy
at k and
= 0)one
andisall
of the More
remaining
modes
are optical
not
capacity
ofenergy
solids
with
frequency,
Einstein
frequency.
Debye
extended
per
unit
cell
(in
one
we will
have M modes
per distinct
value of k,
of which
one mode
go to
zero
at dimension)
k = 0).a fixed
willidea
be acoustic
(goes to zero
energy
at k = 0) and all of the remaining modes are optical (do not
his
with frequency
= vk.
Caution: We have been careful to discuss a true one dimensional system, where the atoms are allowed
go The
to zerodynamics
energy at k =
0).
of the lattice is governed by the classical Hamiltonian:

3.4

Quantized waves: phonons

to move only along the one dimensional line. Thus each atom has only one degree of freedom. However, if we
Caution:
We
been directions
careful to (transverse
discuss a true
one 1d
dimensional
system,
the atoms
are allowed
allow atoms
to+move
in other
to the
line) we will
have where
more degrees
of freedom
per
H=P2/2m
Vhave
to
moveWhen
only along
Thus
each3 atom
hasofonly
one degree
of freedom.
However,
we3
atom.
we getthe
to one
the dimensional
3d solid we line.
should
expect
degrees
freedom
per atom.
And there
shouldifbe
allow
atoms
to move
in other
directions
(transverse
to the
we will
moreper
degrees
of freedom
different
acoustic
modes
at each
k at long
wavelength.
(In 1d
3d,line)
if there
are have
n atoms
unit cell,
there willper
be
atom.
When
we get
to the
solid3we
should modes
expecttotalling
3 degrees3nofdegrees
freedomof per
atom.per
And
there
3(n 1)
optical
modes
but 3d
always
acoustic
freedom
unit
cell.should be 3
different
acoustic modes at each k at long wavelength. (In 3d, if there are n atoms per unit cell, there will be
3 Note in particular the comment on this website about most books getting the form of the acoustic mode
3(n
1) optical modes but always 3 acoustic modes totalling 3n degrees of freedom per unit cell.
incorrect!
3 Note in particular the comment on this website about most books getting the form of the acoustic mode
incorrect!

3.4. QUANTIZED WAVES: PHONONS

37

From the viewpoint of quantum mechanics, those waves become quantized


with energy


1
= ~ n +
.
(3.8)
2
Analogous to defining a single quantum of light as a photon, the quantum of
energy of lattice vibration is called phonon. Characteristics of phonons are
summarized below:
1. Both phonons and photons are bosons and are categorized as quasiparticle 4 .
2. In the energy expression Eq. 3.8 of phonon, the energy level is occupied
by n phonons of energy ~; 21 ~ is the zero-point energy. The existence
of zero-point energy supports Heisenbergs uncertainty principle.
3. The occupation number is temperature dependent
nB (, T ) =

1
e~/KB T

(3.9)

4. A phonon can interact with other particles such as photons, neutrons,


electrons and so on, as it has a momentum which is the momentum
modulo the reciprocal lattice. In other words, the wave vector k of
phonons must be within the first Brillouin zone and ~k is known as the
crystal momentum.
The total vibration energy5
Utot

Na
=
2



1
dk~(k) nB (k) +
,
2
/a
/a

(3.10)

from which we could calculate the heat capacity as dU/dT . This is a 1D


version of Eq. 1.9. In comparison with Einstein and Debye models, the
frequency dispersion is no longer linear in the 1D-monoatomic-chain model
discussed here and we, instead of using a cutoff frequency, use the continuum
integral to count the total number of phonon modes,
!
Z /a
2
dk /
= N.
(3.11)
Na
/a
4
5

A quasiparticle
Ris
P
a
Using
N
2 dk
k

38

CHAPTER 3. VIBRATIONS OF ONE-DIMENSIONAL LATTICES

Although the linear dispersion is invalid, it is useful to further replace integrals over k with integrals over frequency to obtain density of states g()
2
Na
where

3.5

/a

dk =
/a

dg(),


N a dk
g() =
.
d

(3.12)

(3.13)

Anharmonic effect in crystals

If there is no anharmonic effect, then There is no thermal expansion A


crystal would vibrate forever There is no phonon-phonon interaction Thermal conductivity would be infinite

3.5.1

3.6

Thermal expansion

Heat conduction by phonons

Chapter 4
Crystal structure
After learning some macroscopic properties of solids, we need to understand their atomic arrangements for further studies of the rich physics of
solids. For almost all the elements and for a vast array of compounds, the
lowest-energy state is crystalline. Crystalline order is the simplest way that
atoms could be possibly be arranged to form a macroscopic solid. We here
survey some of the most important geometrical properties of periodic arrays
in three dimensional space.

4.1

Lattices and unit cells

Lattices
A lattice is a set of points where the environment of any given point is
equivalent to the environment of any other given point. Mathematically this
set of points R is defined by integer sums of a set of linearly independent
primitive basis vectors ai with i = 1, 2, and 3,
R = n1 a1 + n2 a2 + n3 a3 ,

Figure 4.1: A hexagonal lattice,


also called a triangular lattice, is
symmetric under reflection about x
& y axes, and 60 -rotation, from
Marders text

39

(4.1)

40

CHAPTER 4. CRYSTAL STRUCTURE

where ni are integers. The choice of primitive basis vectors is not unique.
One should note that not all periodic arrangements of points are lattices. For
example a 2D honeycomb shown in Fig. 4.2 is not a lattice. The environment
of point P and point R are actually different point P has a neighbor directly
above it (the point R), whereas point R has no neighbor directly above.
Nevertheless all equivalent points form a triangular lattice, as demonstrated
in Fig. 4.3.

Figure 4.2: The honeycomb is not a lattice. Points P and R are inequivalent. (Points P
and Q are equivalent), from Simons text

Figure 4.3: The honeycomb is not a lattice, but all equivalent points form a triangular
lattice.

4.2. SYMMETRY OF 3D CRYSTALS

41

Unit cells
A unit cell is a region of space, such that when many identical units
are stacked together it completely fills all of space and reconstructs the full
structure. Because lattices are created by repeating basic units over and over
throughout space, the full information of a crystal can be obtained in a small
region of space. Such a region, chosen to be as small as it can be, is called
primitive unit cell. In other words, a primitive cell is a volume of space that,
when translated through all the vectors in a lattice, just fills all of space
without either overlapping itself or leaving voids. So a primitive unit cell for
a periodic crystal is a unit cell containing only a single lattice point.

Figure 4.4: Some unit cells for the triangular lattice, from Simons text.

One can also fill space up with nonprimitive cells (known as conventional
cells). The conventional unit cell is a region that just fills space without any
overlapping when translated through some subset of the vectors of a lattice.
The conventional unit cell is generally chosen to be bigger than a primitive
cell and to have the required symmetry. The Wigner-Seitz cell about a
lattice point is the region of space that is closer to any other lattice point.
The Wigner-Seitz cell is a primitive cell with full symmetry of the lattice.
One way to construct the Wigner-Seitz cell is to draw the perpendicular
bisector of all the lines between a lattice point to each of its neighbor.

4.2

Symmetry of 3D crystals

The set of points specified in Eq. 4.1 is often called the Bravais lattice.
From the viewpoint of symmetry, a Bravais lattice is characterized by the

42

CHAPTER 4. CRYSTAL STRUCTURE

specification of all rigid body operations that take the lattice to itself. This
set of operations is known as the symmetry group or space group of the Bravais lattice. The operations in the symmetry group include all translations
through lattice vectors. In addition, there will be rotations, reflections, and
inversions that take the lattice to itself. For example, a simple hexagonal lattice is taken to itself by a rotation through 60 about a line of lattice point
parallel to the c-axis, and reflection in the lattice plane perpendicular to the
c-axis.
Any symmetry operation of a Bravais lattice can be combined out of a
translation TR through a lattice vector R and rigid body operation leaving at
least one lattice point fixed. The point group of a lattice is defined as a subset
of the full symmetry group which leaves a particular point fixed. There are
only 7 distinct point group that a Bravais lattice can have, so there are 7
three-dimensional crystal systems. When the full symmetry is considered,
there are 14 distinct space groups that a Bravais lattice can have. Thus,
from the view of point of symmetry, there are 14 different kinds of Bravais
lattice.1 The general one is the triclinic lattice. The general triclinic lattice
Fourteen Bravais Lattices and Seven Crystal
is a unit cell, which is a parallelepiped with all sides different lengths, and all
18
angle different from 90 or 120 .Systems
The other 13 lattices have some symmetry,
such as relations between the sides or angles.




90




Rhombohedral
a b c
90

Hexagonal
a b c
90
120

Triclinic
a b c

Monoclinic
a b c
90
90


90

90

Orthorhombic
a b c
90

Tetragonal
a b c

Cubic
a b c

b
c

b
Simple

BaseCentered
BodyCentered

FaceCentered

Figure 4.5: The 14 three-dimensional Bravais lattices. The two centered cubic lattices and
hexagonal lattice are particularly important in solid state physics. (from Marders text)
16th May 2003
c 2003, Michael Marder

For a general crystal structure, there are 32 point groups and 230 space groups that
a lattice with a basis can have.

4.2. SYMMETRY OF 3D CRYSTALS

43

Cubic crystals
There are 3 lattices in the cubic system: the simple cubic (sc) lattice, the
body-centered cubic (bcc) lattice, and the face-centered cubic (fcc) lattice.
The simplest lattice in three dimensions is the simple cubic lattice. In fact,
real crystals of atoms are rarely simple cubic as its packing fraction is only
/6. The bcc lattice is a simple cubic lattice where there is an additional
point in the very center of the cube. A primitive unit cell of the bcc lattice
with the cube edge a can be obtained in terms of primitive basis vectors
a
z);
x+y
a1 = (
2

a
+ z);
a2 = (
x+y
2

a
z),
a3 = (
xy
2

(4.2)

, z, x
are the Cartesian unit vectors. The coordination number of a
where x
lattice (frequently called Z or z) is the number of nearest neighbors any point
of the lattice has. For the bcc lattice the coordination number is Z = 8.

10928

Figure 4.6: Left: primitive basis vectors


of the bcc lattice. Right: The bcc lattice with a
rhombohedra primitive unit cell of edge 23 a.

The fcc lattice is a simple cubic lattice where there is an additional point
in the center of every face of every cube. Analogous to the bcc case, it is
sometimes convenient to think of the fcc lattice as a simple cubic lattice
with a basis of four atoms per conventional cell. The rhomboheoral primitive
cell of the face-centered cubic crystal can be constructed by three primitive
translation vectors a1 , a2 and a3 which connect the lattice point at the origin
with lattice points at the face centers, i.e,
a
);
a1 = (
x+y
2

a
a2 = (
y + z);
2

a
).
a3 = (z + x
2

(4.3)

The angles between the axes. are 60 . The fcc Bravais lattice has a cubic

44

CHAPTER 4. CRYSTAL STRUCTURE

1 Crystal Stru

IJ

Figure 4.7: The rhomboheoral primitive cell


of the face-centered cubic crystal.

Figur~

11 The rhomboheoral primitive cell of the face-centered


cubic crystal. The primitive translation vectors a 1, a 2, a 3 connect
the lattice point at the origin with lattice points at the face centers .
As drawn, the primitive vectors are:

Figure 12 Relation o

in the hexagonal syste

a prism of hexagona

a =n
a
closest packed structure. The 2D lattice in the plane perpendicular to the
al = ~ the
a(X. + y)The
; a'.l =lattice
~ a(y + i) ; points
~ = ~ a(z + of
x) . the
[111] direction forms a triangular structure.
The angles between the axes. are 600.
second layer is located on half of the holey sits in the first layer. The points
of the third layer directly overlay the other half of the first-layer holey sites.
The repeating order of the layers is the ABCA... stacking as illustrated in
The position of a point in a c~ll is specified by (2) in terms of the atomi
Fig. 4.8.
coordinates x, y, z . Here each coordinate is a fraction of the axial length al> a
1

2 =ft.

3.

a 3 in the direction of the coordinate ax.1s, with the origin taken at qne corner o
the cell. Thus the coordinates of the body center of a cell are ~~t and the fac
centers include ~iO, ~ ~; ~0-k . In the hexagonal system the primitive cell is
right prism based on a rhombus with an included angle of 120. Figure 1
shows the relationship of the rhombic cell to a hexagonal prism.

INDEX SYSTEM FOR CRYSTAL PLANES

The orientation of a crystal plane is determined by three points in th


plane , provided they are not collinear. If each point lay on a dillerent crysta
axis, the plane could be specified by giving the coordinates of the points i
terms of the lattice constants a 1 , a 2 , a 3 . However, it turns out to be more usefu
for structure analysis to specify the orientation of a plane by the indices deter
mined by the following rules (Fig. 13).

A on the axes in terms of the lattice constants al> a , a


Find the intercepts
2
The axes may be those
B of a primitive or nonprimitive cell.
C

Figure 4.8: ABCA.... stacking of the fcc lattice. Small red and blue circles denote the
locations of layer B and layer C, respectively.

Only slightly more complicated than the simple cubic lattice are the

4.3. THE RECIPROCAL LATTICE IN 3D

45

tetragonal and orthorhombic lattices where the axes remain perpendicular,


but the primitive lattice vectors may be of different lengths. The orthorhombic unit cell has three different lengths of its perpendicular lattice primitive
basis vectors, whereas the tetragonal unit cell has two lengths the same and
one different.
Examples of crystal structure

4.3

The reciprocal lattice in 3D

Definition of reciprocal lattice


The important physics of waves such as vibrational waves, electron waves
or electromagnetic waves in solids is best best described in reciprocal space
as atoms form crystalline order in solids. Consider plane waves eikr traveling
in a Bravais lattice composed of points R. The set of all vectors G which
yield plane waves eiGr with the periodicity of the Bravais lattice is known
as its reciprocal lattice. In other words, if
eiG(r+R) = eiGr ,

(4.4)

then G belongs to the reciprocal lattice of the Bravais lattice. Factoring


out eiGr , we therefore have the definition of the reciprocal lattice of a given
direct lattice composed of lattice points R:
G is a point in the reciprocal lattice if and only if
eiGR = 1.

(4.5)

Recall that R = n1 a1 + n2 a2 + n3 a3 . The primitive basis vectors of the


reciprocal lattice bi (i = 1, 2, and 3) can be chosen through the following
conditions:
ai bj = 2ij ,
(4.6)
where ij is the Kronecker delta.2 It is obvious that we can generate the
primitive basis vectors bi as follows:
b1 = 2

a2 a3
,
a1 (a2 a3 )

b2 = 2

a3 a1
,
a1 (a2 a3 )

b3 = 2

a1 a2
,
a1 (a2 a3 )

where a1 (a2 a3 ) is the volume of the rhombohedron defined by ai . For an


arbitrary point in reciprocal space written as
G = m1 b1 + m2 b2 + m3 b3
2

The Kronecker delta is defined as ij = 1 for i = j and ij = 0 otherwise.

(4.7)

46

CHAPTER 4. CRYSTAL STRUCTURE

with m1 , m2 , and m3 integers, G is a point of the reciprocal lattice because


eiGR = ei(m1 n1 +m2 n2 +m3 n3 )2 = 1 is satisfied.
The Reciprocal Lattice as a Fourier Transform
Generally one can think of the reciprocal lattice as being the Fourier
transform of a direct lattice. We first prove this statement for the onedimensional case of which the direct lattice is given by Rn = na and the
density (x) of lattice points in one dimension can be expressed as a deltafunction of density at these lattice points,

(x) =

n=

(x na).

(4.8)

We can express (x) in terms of the Fourier series as


(x) =

cm ei a mx ,

(4.9)

m=

and the coefficient cm is


1
=
a

cm

1
a

a/2

dx(x)ei a mx

a/2

a/2

dx

a/2

n=

1
.
=
a

(x na)ei a mx

Then the Fourier transform of the direct lattice (x) is3


Z
dxeikx (x)
F[(x)] =
1
=
a
=
=
3

Note that

m=

ei

2
a mx

x=

dxe
x=

1
a m=
2
a
P

ikx

m=

m=

ei

ei a mx

m=

dxei(k a m)x
x=

(k
2
a mx

2
m).
a

and (k) =

1
2

x=

dxeikx

4.3. THE RECIPROCAL LATTICE IN 3D

47

On the other hand,


F[(x)] =

dxeikx (x)
x=

Z
X

=
=

n=

x=

dxeikx (x na)

eikna .

n=

So
F[(x)] =

eikna =

n=

2 X
2
(k
m).
a m=
a

(4.10)

Thus one obtains that delta function in reciprocal space peaks precisely at the
positions of reciprocal lattice vectors. The Fourier transform of the lattice
function of the direct lattice is proportional to the lattice function of the
reciprocal lattice with the reciprocal constant G = 2/a. This proof can be
generalized to the 3D case,4
F[(r)] =

X
R

ikR

(2)3 X 3
(k G)
=
v
G

(4.11)

where v is the volume of the unit cell. Note that the reciprocal lattice of an
fcc direct lattice is a bcc lattice in reciprocal space. Conversely, the reciprocal
lattice of a bcc direct lattice is an fcc lattice in reciprocal space.
Brilloun zone
In order to describe waves in solids, it is important to understand the
structure of reciprocal space. A unit cell of the reciprocal lattice is called
a Brillouin zone. The Winger-Seitz primitive cell of the reciprocal lattice is
known as the first Brillouin zone. Note that term first Brillouin zone is
only applied to the reciprocal space. In particular when the first Brillouin
zone of a particular r-space lattice is referred, what is meant is the WingerSeitz primitive cell of the associated reciprocal lattice. For example, the first
Brillouin zones of the fcc lattice is just the bcc Winger-Seitz cell, and the first
Brillouin zones of the bcc lattice is just the fcc Winger-Seitz cell, as shown
Fig. 4.9.
4

Note that 3 (k G) = (kx Gx )(ky Gy )(kz Gz )

48

CHAPTER 4. CRYSTAL STRUCTURE

Figure 4.9: First Brillzoun zones of the fcc lattice (left) and the bcc (right) lattice.

Lattice planes
Another way to understand the reciprocal lattice is via families of lattice planes of the direct lattice. We first consider a reciprocal lattice vector
Ghkl hb1 + kb2 + lb3 . In order to satisfy the definition of reciprocal lattice
eiGR = 1, we have
Ghkl r = 2m,

where m is an integer. Thus a set of collinear reciprocal lattice vectors Ghkl


define a family of equally spaced lattice planes to which Ghkl are normal. This
family of planes together contain all the points of the 3D Bravais lattice. For
r1 and r2 being separately in two adjacent planes of this family of parallel
planes and G = G
n with n
being its unit vector, we have
G
n (r1 r2 ) = 2m,
where n
(r1 r2 ) dhkl is the spacing between these two adjacent planes.
The shortest reciprocal lattice vector Gmin in the direction n
corresponds to
m = 1 and gives rise to
2
,
(4.12)
dhkl =
|Gmin |

with Gmin = hb1 + kb2 + lb3 in which h, k, and l do not have a common
integer factor other than 1.
The correspondence between reciprocal lattice vectors and families of lattice plans provides a convenient way to specify the orientation of a lattice
plane. For a given direction n
, the indices (h, k, l) which define the shortest reciprocal lattice vector Gmin in the direction are general used to specify
the plane orientation. These conventional notations (h, k, l) which have no

4.3. THE RECIPROCAL LATTICE IN 3D

49

common factor are known as Miller Indices. For fcc and bcc lattices, Miller
indices are usually stated using the primitive basis vectors of the cubic lattice
rather than the primitive basis vector of the fcc or bcc. It is straightforward
to show that the interlayer spacing for an orthorhombic lattice is
s
h2 k 2
l2
1
=
+ 2 + 2.
(4.13)
2
dhkl
a1 a2 a3
A useful shortcut for figuring out the geometry of lattice planes is to look
111
at the intersection of a plane with the three coordinate axes.
The plane
intercepts the axes at the points x1 a1 , x2 a2 , and x3 a3 . Since G(x1 a1 ) = 2m,
the Miller indices are inversely proportional to the xi

9.1. THE RECIPROCAL LATTICE IN 3D

1 1 1
:
:
= h : k : l.
x1 x2 x3

(4.14)

This is the crystallographic definition of the Miller indices as demonstrated


in Fig. 4.10.

1234567892
62656 

96  6
 
896 562 5
 96
6
465
56  896

 6
956626365
9 23
96
65 2 56 
94
962 65226
96
6 56
Figure 4.10: An illustration of the crystallographic definition of the Miller indices.

Figure 9.2: Determining miller indices from the intersection of a plane with the coordinate axes.
1
22
32
32
The spacing between lattice planes in this family would be |d(2,3,3)
|2 = a2 + b2 + c2 .
conventional unit cell, and the (100) lattice planes would not intersect. However,the (200) planes
would interesect these points as well, so in this case (200) represents a true family of lattice planes
whereas (100) does not!
From Eq. 9.8 one can write the spacing between a family of planes specified by Miller indices
(h, k, l)
2
2
d(h,k,l) =
= 2
(9.10)
|G|
h |b1 |2 + k 2 |b2 |2 + l2 |b3 |2

Where we have assumed that the coordinate axes of the primative basis vectors bi are orthogonal.
Recall that |bi | = 2/|ai | where ai are the lattice constants in the three orthogonal directions.
Thus we can equivalently write
1
|d(h,k,l) |2

h2
k2
l2
+ 2 + 2
2
a1
a2
a3

(9.11)

Note that for a cubic lattice this simplifies to


dcubic
(hkl) =

(9.12)

50

CHAPTER 4. CRYSTAL STRUCTURE

Chapter 5
Wave Scattering by Crystals
Propagation of electron or phonon waves in a crystal plays an important
Chapter
10 of materials. Due to the wave-like nature of both
role in underlying
physics
the electron and the phonon, they share a similarity in the energy dispersion
in reciprocal space. Much of the same physics occurs when crystals scatter
waves (or particles) that impinge upon a crystal externally. Furthermore one
can experimentally determine crystal structures from real-space microscopy
or from diffraction to obtain the lattice structures in reciprocal space. Exposing a solid to a wave such as X-ray, neutron or electron, provides us with
In the last chapter we discussed reciprocal space, and explained that the energy dispersion of
a great phonons
opportunity
to reveal
crystalline
structure.
In facthow
it can
and electrons
is plottedits
within
the Brillouin
zone. We understand
these hardly
are similar
to each other
to the wave-like
bothexperiment
the electron and
However, much of
be overstated
howdueimportant
thisnature
typeof of
is1the
tophonon.
science.

Wave Scattering by Crystals

5.1

the same physics occurs when crystals scatter waves (or particles ) that impinge upon a crystal
externally. Indeed, exposing a solid to a wave in order to probe its properties is an extremely useful
thing to do. The most common probe to use are X-rays. Another common, more modern, probe
to use is neutrons. In fact it can hardly be overstated how important this type of experiment is to
science.

The Laue and Bragg Conditions


The general setup that we will examine is shown in Fig.10.1.

Laue diffraction

Figure 5.1: A generic scattering experiment (from Simons text)


Figure 10.1: A generic scattering experiment.

51
1 Remember,

in quantum mechanics there is no real difference between particles and waves!

121

52

CHAPTER 5. WAVE SCATTERING BY CRYSTALS

For an X-ray being scattered a sample, we can treat the sample as being
some potential V (r) that the photon experiences as it goes through the sample, if we think of the incoming X-ray as a particle. According to Fermis
golden rule, the transition rate (k0 , k) per unit time for the particle scattering from k to k0 is given by
2
2 0

(k , k) =
hk | V (r) |ki (Ek0 Ek ),
~
0

where is the photon state is |ki = 1V eikr with V as being the volume, and
the matrix element is nothing but the Fourier transform of the scattering
potential because
0

hk | V (r) |ki =

eikr
1
eik r
dr V (r) =
V
V
V

dreiQr V (r).

Here Q k k0 is the wave vector change of scattering. If the sample is


periodic, the matrix element is zero unless Q is a reciprocal lattice vector,
Q = G. Assume V (r) periodic such that V (r + R) = V (r). The matrix
element is
Z
1
0
dreiQr V (r)
hk | V (r) |ki =
V
Z
1X
dreiQ(r+R) V (r + R)
=
V R
unitcell
"
# Z

1 X iQR
iQr
=
e
dre V (r)
V R
unitcell
The first term in the brackets must vanish unless Q is a reciprocal lattice
vector. If Q = G, the first term in the brackets is the total number of unit
cell N and

Z
1
iQr
0
(5.1)
hk | V (r) |ki =
dre V (r) ,

unitcell
where is volume of the unit cell in real space, = V/N . The condition
k k0 = G,

(5.2)

known as the Laue condition, is precisely the statement of the conservation of


crystal momentum. Also, for elastic scattering, |k0 | = |k| due to the energy
conservation, as enforced by the delta function in the Fermis golden rule.

5.1. THE LAUE AND BRAGG CONDITIONS

53

G1

G2
k

G3

Figure 5.2: Eward sphere of diffraction

Consider a sphere of radius |k| in momentum space and choose one reciprocal lattice point which lies on the sphere as the origin of the reciprocal
space. For any reciprocal lattice points, such as G1 , G2 and G3 , which lie
also on the sphere, there will be beams diffracted through constructive interference and emitted along k Gi and the Laue condition is satisfied. This
construction of reflection is known as the Ewald construction. This sphere is
called Ewald sphere.
Bragg reflection
Consider an incoming wave is reflected from two adjacent layers of atoms
separated by a distance d as plotted in Fig. 5.3 in which the incidence angle
with respect to the atom plane is , i.e., the wave is reflected by 2. It is obvious that there is a path difference between wave components reflected from
the two planes of atoms; the additional distance traveled by the component
of the wave that reflects from the further layer of atoms is
path difference = 2d sin .
In order to have constructive interference, this path difference must be equal
to an integer number of wavelengths. Thus we derive the Bragg condition
2d sin = n.

54

CHAPTER 5. WAVE SCATTERING BY CRYSTALS

k'

Figure 5.3: Bragg reflection (from Simons text)

is the normal unit vector of the atom planes,


As G
G
= sin
k

and

= sin .
k0 G

If the Laue condition k k0 = G is satisfied, the relation k k0 = nG


also gives rise to a non-vanishing matrix element hk0 | V (r) |ki, where n is an
integer. The Laue condition implies that
k0 )
|k|(k
2 0
(k k )

2 0
(k k ) G

2
2 sin

= nG
= nG

= nG G
= n

2
.
d

2
We use the relation d = |G|
in the last step. So we obtain 2d sin = n,
showing that the Laue condition is precisely equivalent to the Bragg condition.

Scattering amplitude
The transition rate of wave scattering by a periodic potential is proportional to the matrix element hk0 | V (r) |ki which can also be expressed as
"
#
X
1
hk0 | V (r) |ki
eiQR S(hkl)
V R
Z
S(hkl) = S(Q)
drV (r)eiQr ,
unitcell

5.1. THE LAUE AND BRAGG CONDITIONS

55

where S(hkl) is known as the structure factor. Therefore the intensity Ihkl
of scattering from the lattice planes defined by the reciprocal lattice vector
(hkl) is proportional to the square of the structure factor at this reciprocal
lattice vector, i.e.,

2
Ihkl S(hkl) .
A good approximation assuming the scattering potential is the sum over
the scattering potentials of individual atoms,
X
V (r) =
Vj (r rj )
atoms j

atoms j

fj (r rj )

(5.3)

So
S(Q) =
=

drV (r)eiQr

unitcell
X Z

atoms j

unitcell

drfj (r rj )eiQr

fj eiQrj .

atoms j unit cell

For neutron scattering, the atomic form factor fj represents the strength of
scattering from nucleus j; it varies rather erratically with atomic number and
is independent of Q. Also, the structure factor S(Q)
In contrast,

56

CHAPTER 5. WAVE SCATTERING BY CRYSTALS

Chapter 6
Electrons in a Periodic
Potential
The single most important fact that the ions in crystalline solids is that
they are in a periodic array. We thus need to consider electrons in a periodic
potential V (r + R) = V (r). To introduce the basic ideas of band structure,
we first discuss the tight-binding approximation.

6.1

Electron band

6.2

Tight-binding approximation

We consider a thought experiment in which a very large number of initially isolated atoms are gradually brought together. As a result of their
interaction with one another, the energy levels of electrons broaden when
the inter-atomic distance is decreased and the electron wavefunctions start
to overlap, forming an energy band. Under some circumstances, the formation
of energy band gives rise to a reduction in electronic energy and consequently
leads to chemical bonding. The bandwidth, i.e., the broadening, depends on
the overlapping of the wave functions concerned, as illustrated in Fig. 6.1
2
A = ~ 2 + VA (r Rn )
H
2m

A (r Rn ) = atomic (r Rn )
H
57

58

CHAPTER 6. ELECTRONS IN A PERIODIC POTENTIAL

1.1 The Periodic Table of the Elements

Figure 6.1: Broadening


of the energy
levels
as levels
a large
of identical
atoms atoms
are from the
Fig. 1.1. Broadening
of the
energy
as number
a large number
of identical
firsttogether.
row of the
periodic
table approach
one another
separation r0 corregradually brought
The
separation
r0 corresponds
to the(schematic).
equilibriumThe
separation
sponds
to the
approximate
separation of chemically bound atoms. Due to the
of atoms in a solid.
(from
Ibach
& L
uths equilibrium
text)
overlap of the 2 s and 2 p bands, elements such as Be with two outer electrons also become
metallic. Deep-lying atomic levels are only slightly broadened and thus, to a large extent,
they retain their atomic character

For the sake of simplicity, we assume all the atomic orbitals (r Rn ) are
orthogonal to each other and possess spherical symmetry,
one extreme, this overlap
may be limited to neighboring atoms; in other
D
E Z may be spread over many atoms. In the former
cases the wavefunctions

(r case,
Rm ) the
(rdegree
Rn )of overlap,
= drand
(r
Rmthe
)(r
Rn )of=the
m,nbonding,
.
thus
strength
is dependent not only on the separation of neighboring atoms, but also on the bond

The Hamiltonian
forThis
an electron
in to
theastotal
potential
of allorthe
atoms,
i.e.,
angles.
is referred
directional
bonding
covalent
bonding.
In its purestcan
form,
bonding is realized between a few elements
one-electron approximation,
be covalent
written as

of equal ``valence'', i.e. elements with the same outer electronic configura2
tion. However,
equal
configuration is neither a necessary nor a
2
= ~an
+ Velectronic
H
A (r Rn ) + V (r),
sucient condition
2m for covalent bonding. What is important is simply the
P
relative
extent of the wavefunctions in comparison to the interatomic
where V (r) separation.
nowwavefunctions
seek for the iseigenfunction
(r)
m ). We
m6=n VA (r
If
theRextent
of the
large compared
to the nearand the eigenenergy
E
of
the
Schr
o
dinger
equation
est-neighbor distance, then the exact position of the nearest neighbors plays
an insignificant role
the greatest possible overlap with many
in producing
H(r)
= E(r)
atoms. In this case, the packing density is more important than the position
of the next neighbors. Here one speaks of non-directional bonding. This
Multiplying the equation by and integrating over the whole crystal, we
regime in which the wavefunctions spread over a distance that is large in
obtain
E
D separation
comparison to the atomic
is characteristic of metallic bonding.

H

type of non-directional bonding with extreHowever, there is a further
D E ,
E =of wavefunctions;
mely small overlap
this is the ionic bond. It occurs in


cases where the transfer of an electron from one atom to another is suciently

6.2. TIGHT-BINDING APPROXIMATION

59

where the Dirac notion is used,


D E Z
D E Z



H = dr H and
= dr

Now we exploit the variational method to find ground state by choosing an


ansatz solution (trial function) from a linear combination of atomic orbitals
(r Rn ), i.e.,
X
(r) =
an (r Rn ).
n

to minimize

D E

H

D E .

Provided that there is a translational symmetry in the periodic structure


of lattice, one can choose the expansion coefficients as an = eikRn through
introducing with an index k, and have1
X
eikRn (r Rn )
k ' k =
n

and

Z
E X

ik(Rn Rm )
k k =
e
dr (r Rm ) (r Rn ).
n,m

For a sufficiently localized electron, (r Rn ) only has significant values


around Rn . To a first-order approximation, we get
D E XZ

k k '
dr (r Rn ) (r Rn ) = N,
n

where N is the number of atoms in the crystal. Therefore the eigenenergy is


Z
1 X ik(Rn Rm )
Ek '
e
dr (r Rm )[atomic + V (r)] (r Rn )
N n,m
Z
1 X ik(Rn Rm )
= atomic +
e
dr (r Rm )V (r) (r Rn )
N n,m
Z
1 X ik(Rn Rm )
= atomic + V0 +
e
dr (r Rm )V (r) (r Rn ),
N n6=m
1

After learning the Blochs theorem, we will be able to justify the choice of eikRn as
the expansion coefficients.

60

CHAPTER 6. ELECTRONS IN A PERIODIC POTENTIAL

R
where V0 dr (r Rn )V (r) (r Rn ). If we define the hopping integral
that depends on k and describes interaction of electrons at one atomic site
with neighbouring sites as
t

dr (r Rm )V (r) (r Rn ),

one can express the band energy in the tight-binding approximation as


Ek ' atomic + V0 t

eik ,

(6.1)

P
where Rn Rm and the summation runs only up to nearest neighbours. This expression shows that the band energy in the tight-binding
scheme is mainly given by the energy of original atomic orbitals, plus a constant correction due to interactions, and plus a hopping term proportional
the hopping integral t.
7.3 The Tight-Binding Approximation

171

Fig. 7.8 ac. Qualitative illustration of the result of a tight-binding calculation for a primi-

Figure 6.2:
of the electronic
structure
terms
of the
tight-binding
E2 in the
tiveQualitative
cubic latticeillustration
with lattice constant
a. (a) Position
of the in
energy
levels
E1 and
approximation
forVr
a primitive
cubic
lattice
of latticeand
instant
a. (from
Ibach
thsEtext)
potential
of the free
atom.
(b) Reduction
broadening
of the
levels&EL
1 uand
2 as
a function of the reciprocal atomic separation r 1. At the equilibrium separation a the
mean energy decrease is A and the width of the band is 12 B. (c) Dependence of the oneelectron energy E on the wave vector k (1, 1, 1) in the direction of the main diagonal [111]

Zone center ??
Zone boundary
where k 2/a
= k 2x + k 2y + k 2z . This k 2-dependence corresponds to that which

results from the quasi-free-electron approximation (Sect. 7.2).


ii) From (7.36 b) it follows that the energy width of the band becomes
greater as the overlap between the corresponding wavefunctions of neighboring atoms increases. Lower lying bands that stem from more strongly
localized states are thus narrower than bands originating from less
strongly bound atomic states whose wavefunctions are more extended.
iii) In the framework of the present one-electron approximation, the occupation of the one-electron band states is obtained by placing two of the
available electrons of every atom into each band, beginning with the

6.3. THE TRANSLATIONAL SYMMETRY BLOCHS THEOREM

6.3

61

The Translational Symmetry Blochs theorem


TR f (r) = f (r + R).


~2 2

(r) + V (r)(r
TR H(r) = TR
2m
~2 2
(r + R) + V (r + R)(r + R)
=
2m
~2 2
=
(r) + V (r)(r + R)
2m

= H(r
+ R)
TR (r)
= H
TR (r) = CR (r).

(6.2)

(6.3)
(6.4)

TR+R0 (r) = TR0 TR (r)


CR+R0 (r) = TR0 CR (r)
= CR0 CR (r)

h(r)|(r)i = h(r + R)|(r + R)i = 1

= CR
CR h(r)|(r)i

(6.5)

|CR |2 = 1.

(6.6)

CR = eikR

(6.7)

k (r + R) = eikR k (r),

(6.8)

for some real k. We the have

where is k an index. This is the Blochs theorem.


Multiplying Eq ?? by a phase factor eikr , we get
k (r + R)eik(r+R) = eikr k (r)

62

CHAPTER 6. ELECTRONS IN A PERIODIC POTENTIAL

we can define eikr k (r) as a function uk (r) with a periodicity R, i.e.,


uk (r) eikr k (r)
Therefore
uk (r + R) = eik(r+R) k (r + R)
= eik(r+R) eikR k (r)
= uk (r)
k (r) = eikr uk (r)

(6.9)

(6.10)

This is another version of Blochs theorem; k (r) is a modified plane wave


known as the Bloch function. Because uk (r) is a periodic function, it is can
be written as a sum over reciprocal lattice vectors,
X
uk (r) =
uk eiGr .
(6.11)
G

Thus the wave function can be expressed as


X
k (r) =
uk ei(G+k)r .

(6.12)

This implies that we can write each eigenstate as being made up of planewave states eikr which differ by reciprocal lattice vectors G.
Fourier analysis of Blochs theorem
In any given Bloch wave function, only plane waves with k that differ by
some reciprocal vector G can be mixed together. To further understand this,
we derive the Schrodinger equation in the momentum space. First, consider
the Schrodinger equation,


~2 2
+ V (r) (r) = E(r).

2m
0

We then expand (r) in terms of plane waves eik r and have2


X
0
(r) =
(k0 )eik r .
k0

(k0 ) is denoted as k0 in Simions text.

(6.13)

6.3. THE TRANSLATIONAL SYMMETRY BLOCHS THEOREM

63

Similarly we express V (r) as


V (r) =

VG eiGr ,

where G is the reciprocal lattice vector. The Schrodinger equation becomes



X ~2 |k0 |2
X
0
0
+ V (r) (k0 )eik r = E
(k0 )eik r .
2m
k0
k0

(6.14)

Note that
X

V (r)(k0 )eik r =

k0

VG ei(G+k )r (k0 )

k0 G

X
k0 G

VG ei(G+k G)r (k0 G),

where the summation index k0 is dummy and can be changed from k0 to


k0 G in the last step. The modified Schrodinger equation is
(
)

X  ~2 |k0 |2
X
0
E (k0 ) +
VG (k0 G) eik r = 0.
2m
G
k0
0

Since the plane waves eik r which satisfy the Born-von Karman boundary
condition are orthogonal, the coefficient of each separate term inside { }
must vanish. We then have3


X
~2 |k|2
VG (k G).
(6.15)
(k) =
E
2m
G
This is a representation of the Schrodinger equation in the momentum space.
The wave vector k of the Fourier component (k) in Eq. 6.13 only assumes
the values k, k K, k K0 , and (k) couples only to (k G) whose
Alternatively one can prove this by multiplying the equation by V1 eikr and then
integrating over the entire volume V in r-space, i.e.,
(
)

X 1Z
X
~2 |k0 |2
i(k0 k)r
0
0
dre
E (k ) +
VG (k G) = 0.
V
2m
0
3

Using the identity

1
V

drei(k k)r = k,k0 , we get Eq. 6.15.

64

CHAPTER 6. ELECTRONS IN A PERIODIC POTENTIAL

k-values differ from one another by a reciprocal vector G. Therefore the


wave function is of the form
X
k (r) =
(k G)ei(kG)r
(6.16)
G

= eikr

X
G

ikr

(k G)eiGr

e uk (r)
(6.17)
P
It is obvious that uk (r) = G (k G)eiGr is a periodic function with
the same period as the that of the potential V (r). So the Blochs theorem is
proven through Fourier analysis.
In the above Fourier analysis, the index k describes symmetry properties
of translation. One can directly to label Bloch wave functions with ki which
span over the entire momentum space. This method of indexing Bloch wave
functions is called the extended zone scheme. If k is shifted by a reciprocal
vector G, the Bloch wave function is invariant because
X
0
(k + G G0 )ei(k+GG )r
k+G (r) =
G0

X
G00

00 )r

(k G00 )ei(kG

= k (r),

(6.18)

where the summation index G00 G0 G is dummy. For a given k, one


can find a set of wave functions k1 (r), k2 (r), k3 (r) with different
energies but the same eigenvalue eikR when acted upon by TR . One way
to ensure the set of all ki (r) is a complete and linearly independent set of
wave functions is by using the reduced zone scheme. In the reduced zone
scheme, we limit k to be within the first Brillouin zone. Wave vectors lying
outside of the first Brillouin zone will be shifted by a reciprocal vector G into
the aforementioned zone accordingly through ki = k + G. Then we need an
additional index n to label Bloch wave functions and associate energies as
n,k (r) and Enk .
In summary, the Blochs theorem leads to a profound consequence that
even though the potential that the electron feels from each atom is extremely
strong, the electrons still behave almost as if they do not see the atoms at
all. They still almost form plane wave eigenstates, with the only modification
being the periodic Bloch function uk (r) and the fact that momentum ~k is
now crystal momentum.
Homework:
(1) Prove k (r) = k (r)

6.4. NEARLY FREE ELECTRONS

65

extended zone scheme

reduced zone scheme

Figure 6.3: Illustration of the extended and reduced zone schemes in one dimension.

(2) Prove En (k) = En (k)


(3) Prove k En (k) = 0 at k = 0 &

G
,
2

i.e.,

directions.

6.4

En (k)
k

k=0, G
2

= 0 in all

Nearly Free Electrons

After learning the general features of electrons in a Bravais lattice, it


is instructive to consider the limiting case of a vanishingly small periodic
potential of the ion cores. At the opposite extreme from the tight-binding
approximation, the electron states in this approximation are almost the same
as free plane waves. This approach is called the nearly-free-electron model,
which can often explain the band structure of a crystal and answer almost
all the qualitative questions about the behaviour of electrons in metals.
For free electrons in a lattice, the requirement of translational symmetry
demands that the possible electronic states are not restricted to a single
parabola in k-space in the reduced zone scheme, but can be equally well
described by parabola shifted by any reciprocal vector G, i,e.,
k+G (r) = k (r)

(6.19)

For example, Fig. 6.4 shows the low-lying energy bands of free electron in the
empty simple cubic lattice for k along [100]. Parabolas of the free-electron

66

CHAPTER 6. ELECTRONS IN A PERIODIC POTENTIAL

energy intersect at the zone centre and zone boundaries. The true band
structures open up gaps at these k values. Below we explain this feature by
using the nearly-free electron model.
From the Fourier analysis of Blochs theorem, we know that the Bloch

164

7 The Electronic Bandstructure

Fig. 7.3. B
tive cubic
tion along
tential is
various br
ciprocal s
000; (
101, 110,

Figure 6.4: The band structure of free electron gas in a sc lattice along kx in the
first Brilloun zone. Various branches stemming from k 2 = kx2 + ky2 + kz2 of different k = (kx , ky , kz ) are plotted by various line symbols: () for k = (kx , 0, 0),
( ) for k =  kx 2
( )
a , 0, 0 , 
2
for k = kx , 2
,
0
and
k
,
0,

a
a , ( )
 x
2
2
2
2
for k = kx a ,
 a , 0 , kx a , 0, a ,
2
2
and kx , a , a

eikr

i(k G)r

e
and G/2
= p/a, where two parabo
of an electron with these k-values
two corresponding plane waves.
approximation) these waves are

eiG x=2

and eiG=2

Gx

iGx

G
Equation
(7.8) implies that waves w
i(kG)r into account. However, o
Figure 6.5: Dispersions of eikr be
and etaken
waves
lows that Ck is particularly large w
equal to 
h 2 k 2/2 m, and that the co
same absolute magnitude as Ck. Th
waves at the zone boundaries (7.17
can neglect contributions from oth
priate expressions for a perturbatio

G/2

6.4. NEARLY FREE ELECTRONS

67

wave function is composed of Fourier components ei(kG)r , i.e.,


X
k (r) =
(k G)ei(kG)r .
G

The nearly-free-electron approximation is equivalent to assuming the wave


function is nearly equal to a plane wave, eikr . This means that the leading
term in the expansion is (k) 1 and higher-order terms of the expansion
are negligible.
For a plane wave eikr travelling near the zone boundary, i.e., k G/2,
the counter-propagating Bragg-reflected wave ei(kG)r has approximately the
same energy and the magnitude as those of eikr . Therefore, in addition to the
leading term (k) 1, we must also keep the terms that include (k G).
In short, because of the presence of the vanishingly small periodic potential, the electron state is no longer a plane wave. Instead, it is a superposition of eikr and ei(kG)r . We can either use the Schrodinger equations in
the k-space or the degenerate perturbation theory to find the eigenstates of
nearly-free electrons in a lattice.
using Schrodinger equations in the k-space
Neglecting Fourier components other than (k) and (k G), the Schrodinger
equations in k-space, i.e., Eq. 6.15, become


~2 |k|2
(k) = VG (k G)
(6.20)
E
2m

2 !
~ 2 k G
E
(k G) = VG (k G),
2m
where the zero-order correction energy V0 is constant and can be set to 0.
That is, we have
!
!
~2 |k|2
(k)

E
V
G
2m
= 0.
~2 |kG|2
(k G)
VG

E
2m
We then obtain the secular equation for the energy value
2 2

~ |k|

VG
2m E


= 0.
2
2
~ |kG|
VG
E
2m

The solutions are

 4
1/2


~2
~
2
2
2
2
2 2
E =
|k| + |k G|
|k G| |k| + |VG |
, (6.21)
4m
16m2

68

CHAPTER 6. ELECTRONS IN A PERIODIC POTENTIAL

indicating that, near the Brillouin zone boundary, a gap opens up due to
scattering by a reciprocal lattice vector.
using degenerate perturbation theory
In this method, we need to solve the eigenvalues and eigenvectors of
|(r)i = E|(r)i,
H

= ~2 2 +V (r). Since the wave function of nearlywith the Hamiltonian H


2m
free electrons in a lattice is a superposition of eikr and ei(kG)r , it can be
expressed as




|(r)i = (k) eikr + (k G) ei(kG)r .

So the Schrodinger equation is












(k) eikr +(k G) ei(kG)r = E (k) eikr +(k G) ei(kG)r .
H

Multiplying both sides of equation with eikr , we have


i(kG)r
ikr

e
e
+ (k G) eikr H
= E(k)
(k) eikr H

Similarly through the multiplication with ei(kG)r , we have


ikr
i(kG)r

e
e
(k) ei(kG)r H
+ (k G) ei(kG)r H
= E(k G)
We can combine these into a matrix equation,

|eikr i E
|ei(kG)r i
heikr | H
heikr | H
|eikr i hei(kG)r | H
|ei(kG)r i E
hei(kG)r | H

(k)
(k G)

= 0.

Because
2
2
|eikr i = ~ |k|
heikr | H
2m
2
~
|k G|2
|ei(kG)r i =
hei(kG)r | H
2m
ikr i(kG)r
he | H |e
i = VG
i(kG)r ikr
he
| H |e i = VG ,

the matrix equation is


~2 |k|2
2m

VG

VG
~2 |kG|2

2m

(k)
(k G)

= 0.

Therefore we will obtain the same solutions as those shown by Eq. 6.21.

6.4. NEARLY FREE ELECTRONS

69

on the zone boundary


When k = G/2, free-electron waves of eikr and ei(kG)r have exactly the
same energy,
~2 |k G|2
~2 |k|2
=
2m
2m
and the energy solutions are
E =

~2 2
|k| |VG |
2m

(6.22)

An energy gap of |VG | at the zone boundary opens up as a result of Bragg


reflection.
Consider a one-dimensional problem of nearly-free electron gas. Lets assume the potential energy is very small and negative near the ion cores, e.g.,
V (x) = V cos(2x/a), where V < 0 and a is the lattice constant. Note that
VG < 0 in this one-dimensional system. Because of translational symmetry,
there is a degeneracy of the energy values at the zone edges, k = G2 = a ,
where two parabolas intersect. To a first-order approximation, the description of the electron states at k = a is at least a superposition of two
corresponding waves,

ei a x
and
ei a x .
If the contributions of wave vectors other than k = a are neglected, the

+|

|2

Figure 6.6: Left: (a) The potential energy V (x) of an electron in a one-dimensional lattice.
(b) &(c) Probability densities. Right: Splitting of the energy parabola of the free lectern
at the Brillouin zone edges. (from Ibach & L
uths text)

70

CHAPTER 6. ELECTRONS IN A PERIODIC POTENTIAL

energy solutions are


~2  2
|VG | E0 |VG |.
(6.23)
2m a
For E+ = E0 + |VG |, Eq. 6.20 leads us to have

|VG |( ) = VG ( )
a
a
and, because VG < 0, we have the standing wave

+ ei a x ei a x sin( x).
a
Similarly we the standing wave

ei a x + ei a x cos( x).
a
corresponding to E = E0 |VG |.
If we check the charge densities | |2 associated with these two eigenfunctions , we see that for an electron in the lower energy eigenstate ,
the charge density is concentrated mainly around the position of the ion cores
and minimum in between; for the higher energy eigenstate + , the charge
density is maximum between the cores.
So the general principle is that the periodic potential scatters between
the two plane waves eikr and ei(kG)r . If the energy of these two plane
waves are the same, the mixing between them is strong, and the two plane
waves can combine to form one state with higher energy (concentrated on
the potential maxima) and one state with lower energy (concentrated on the
potential minima).
E =

near the zone boundary


+ and G =
For the one-dimensional case, lets assume that k = n
a
and then we have
 n 2
|k|2 + |k G|2 2
+ 2 2
a
 n 2
2

|k|2 |k G|2 16
a
and
 4
1/2
 4
1/2

~
~ n 2
2
2
2
2 2
|k G| |k| + |VG |

+ |VG |
16m2
m2 a
1 ~4 2  n 2
|VG | +
.
|VG | 2m2 a

2n
a

6.5. SYMMETRY IN ELECTRONIC BAND STRUCTURE

71

Therefore the energy solutions of Eq. 6.21 are




1 ~2  n 2
~2  n 2
~2 2
1
E
|VG | +
2m2 a
2m
|VG | m a

6.5

Symmetry in Electronic Band Structure

72

CHAPTER 6. ELECTRONS IN A PERIODIC POTENTIAL

Chapter 7
Electron-Electron Interactions
7.1

The Hatree-Fock approximation

7.2

Electron-electron interaction: Screening

screened Coulomb potential

7.3

Fermi liquid and quasiparticles

7.4

Electron-phonon interaction

7.5

Electrons in a magnetic Field

73

74

CHAPTER 7. ELECTRON-ELECTRON INTERACTIONS

Chapter 8
Semiconductor Physics
8.1

Electrons and Holes

75

energy surface
219
5. Magnetoresistance with two carrier types 219

76

CHAPTER 8. SEMICONDUCTOR PHYSICS

Figure 1 Carrier concentrations for metals, semimetals, and semiconductors. The semiconductor
Figure 8.1: Trajectory of a conduction electron scattering off the ion according the naive
range
may b e extended upward by increasing the impurity concentration, and the range can be expicture of Drude.
tended downward to merge eventually with the insulator range.

Chapter 9
Magnetism

77

Вам также может понравиться