lOMoARcPSD|3449400

Solid State Physics Notes

Solid State Physics (University College London)

StuDocu is not sponsored or endorsed by any college or university

Downloaded by Roy Vesey (rdv0044@googlemail.com)
 lOMoARcPSD|3449400

Crystal structures
We begin with a brief recap of the basics of crystal structures that you should remember
from PHAS1228.

An ideal crystal is an infinite repetition of an identical group of atoms. The crystal structure
can always be decomposed into a lattice and a basis1. The lattice describes the underlying
periodicity of the structure, whereas the basis is the group of atoms that is being repeated.
For example, the two-dimensional lattice of graphene (honeycomb arrangement of carbon
atoms) can be decomposed into a 2D hexagonal lattice and a two-atom basis.

+ =

Lattice Basis Structure

The Bravais lattice

The crystal lattice, also known as the Bravais lattice, is a periodic array of (lattice) points that
looks the same from every lattice point. That is to say that the crystal lattice has translational
symmetry—it is invariant under the translation + by any lattice vector . A lattice
vector is any linear combination of the (non-coplanar) primitive lattice vectors , and ,
i.e.

= +n +

where , , are integers . The primitive lattice vectors thus generate the lattice
and the parallelepiped formed by , and defines the primitive unit cell.

In 2D, there are five distinct Bravais lattices, shown below, all of which exhibit translational
invariance, in addition to other symmetries.

1
Strictly speaking, a crystal structure is a convolution of a lattice with the basis.

1-1

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Note that the honeycomb structure of graphene mentioned ealier is not itself a Bravais lattice
as any two neighbouring carbon atoms are not in exactly the same environment.

In 3D, there are 14 different Bravais lattices (first correctly enumerated by Bravais in 1845).
These are grouped into 7 different crystal classes, as shown below. In this course, we will be
mainly concerned with the cubic, hexagonal and sometimes tetragonal structures.

1-2

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Point groups and space groups

Once we add the basis, there is no limit on the number of possible atomic patterns.
However, the type of pattern is constrained to be consistent with the symmetry of the lattice.

The symmetry of the basis is called the point group, because it consists of symmetry
operations that leave a point in space fixed (i.e. these are non-translational operations,
namely rotations, rotation-reflections and rotation-inversions). There are 32 distinct point
groups that can be embedded into 3D lattices.

There are then 230 distinct ways of combining the 32 point groups with the 14 Bravais
lattices, giving rise to 230 space groups. All crystalline materials belong to one of these 230
space groups.

Primitive and conventional unit cells

A primitive unit cells is a repeat unit with the minimum volume that, when translated by all
reciprocal lattice vectors, just fills space. It is not uniquely defined but all primitive cells have
the same volume =| ( )|. A primitive unit cell always contains one and only
one lattice point.

The primitive unit cell does not always reflect the symmetry of the lattice, so it is often more
convenient to define a conventional unit cell that does. Conventional unit cells may contain
more than one lattice point (and therefore have a larger volume). The dimensions of the
conventional cells are known as the lattice parameters (or lattice constants), usually denoted
by the letters and (for cubic crystals the three dimensions are the same and hence only
one lattice parameter, , is specified). The conventional unit cells for the three cubic Bravais
lattices are shown below.

SC BCC FCC
Lattice points per conventional cell 1 2 4
Volume of conventional cell = =2× =4×
Coordination number2 6 8 12
Packing fraction (%)3 52 68 74
Examples Po Cr, Mo Al, Cu

2
The coordination number is the number of nearest neighbours
3
See section on close packing

1-3

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

The primitive unit cell of the fcc structure is a rhombohedron with an angle of 60 between
the primitive lattice vectors

= ( )  = ( )  = ( )
2 2 2

For the bcc, structure, the primitive unit cell is a rhombohedron with an angle of 109.47
between the primitive lattice vectors

= ( )  = ( )  = ( )
2 2 2
As we will see later in our discussion of diffraction from crystals, it is sometimes convenient
to decompose the bcc and fcc structures into a simple cubic lattice and a basis. In terms of
the lattice constant the coordinates of the basis atoms are (0,0,0) and , , for the bcc
structure, and (0,0,0), , , , 0, and , for fcc.

The Wigner-Seitz cell

The Wigner-Seitz cell is a primitive unit cell defined around a given lattice point as the
volume that includes all points in space which are closer to that given lattice point than to
any other lattice point. The Wigner-Seitz cell is therefore the smallest volume enclosed by
the planes that perpendicularly bisect the lines connecting a chosen lattice point to all nearby
lattice points (i.e. not just nearest neighbours, but in practice only a few near neighbours are
sufficient). For a 2D lattice, the construction is shown below.

The Wigner-Seitz cell is uniquely defined and displays the full symmetry of the lattice. It
plays a special role in the theory of wave propagation in crystals, as we shall see later.

Close packing
Solids with non-directional bonding (e.g. van der Waals solids, elemental metals) favour
structures with the closest packing of atoms. In 2D, the closest packing is achieved by a
hexagonal arrangements of spheres (atoms). In 3D, there are two different ways in which
these hexagonal layers can be stacked to achieve close packed structures. If spheres in the
second layer go on top of the B sites, then spheres in the third layer can go on either the A

1-4

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

or the C sites. The ABAB….arrangement leads to the hexagonal close packed (hcp)
structure, whereas the ABCABC… stacking gives rise to the fcc structure.

A
B
A A
C C
A A A
B B
C
A A
AB
B

B B

A A

C C
A A A
ABA ABC
A A C

Hexagonal close Face centred cubic

packed (HCP) (FCC)

You should convince yourself that the packing fraction (i.e. the ratio of the volume occupied
by the spheres to the total volume of the unit cell) for both these structures is 74%.

For the fcc structure, the close packed planes correspond to the (111) cubic planes (i.e.
those perpendicular to the body diagonal of the cube).

ABCA

Unlike the fcc structure, the hcp structure is not a Bravais lattice; it is formed from two
interpenetrating simple hexagonal lattices displaced from each other by , , . For an

ideal hcp structucture, the ratio of lattice parameters = .

1-5

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Examples of some important crystal structures

Caesium chloride (CsCl)

Lattice: simple cubic

Basis: Cs at (0,0,0), Cl at (½, ½, ½)

Examples: AlNi, CuPd, CuZn ( -brass)

Sodium chloride (NaCl)

Lattice: fcc

Basis: Na at (0,0,0), Cl at (½, ½, ½)

Examples: MgO, KCl, KBr, PdS

1-6

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Diamond (C)

Lattice: fcc

Basis: C at (0,0,0), C at (1/4,1/4,1/4)

Exemples: Si, Ge

Zinc blende (ZnS)

Lattice: fcc

Basis: Zn at (0,0,0), S at (1/4,1/4,1/4)

Examples: ZnS, GaAs, InSb

1-7

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Diffraction from crystals

Recap of Bragg’s law and Miller indices
Crystal structures are determined using diffraction techniques. The wavelength of the probe
(X-rays, neutrons, electrons) must be comparable to the interatomic spacing 漢 なD 噺 など貸怠待 m
= 0.1 nm.

Bragg’s law
W. L. Bragg (1915 Nobel Prize; UCL Quain Professor 1915-25) proposed a simple
geometrical construction for deriving the condition for constructive interference of X-rays.

Lattice planes are assumed to reflect X-rays (specular geometry, i.e. angle of incidence 肯 噺
angle of reflection). Interference is constructive when the path difference AO+OB 噺 に講券,
where 券 is an integer, i.e. when

に穴 œÆº 肯 噺 券膏

This is Bragg’s law and it is valid also for diffraction of neutrons and electrons.

X-rays are EM waves and are scattered primarily by the electrons in the solid so the
scattering power of atoms increases with the atomic number 傑 (i.e. number of electrons). It is
more difficult to detect light atoms, especially when together with heavy atoms in the solid.
朕頂
X-rays with 膏 漢 な"D have energies 継 噺 月荒 噺 漢 など"Ł̋8 and a penetration depth of 漢 航m.
碇

Neutrons are scattered by nuclei via the strong nuclear force. Scattering strength does not
have a monotonic dependence on atomic number; neutrons are therefore good for studying
light elements. Neutrons also interact magnetically with magnetic moments of atoms and can
be used to determine magnetic structure of materials.

Neutrons typically penetrate millimetres into the materials and have energies of order
椎鉄 朕
漢 などど"Œ̋8 for 漢 な"D 岾継 噺 膏 噺 峇.
態陳 椎

Electrons are scattered by atomic electrons much more strongly than X-rays and therefore
penetrate only a few nanometres into the material. They are useful for studying surface
structures. For 膏 漢 な"D, the electron energy 継 蛤 などど"̋8.

2-1

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Miller indices
To understand diffraction data we need to identify and label different families of lattice
planes. Each family of lattice planes:

- contains all lattice points

- has the same density of lattice points per plane
- has a characteristic inter-plane spacing 穴

We label families of planes using Miller indices (月倦健) determined in the following way:

- take any lattice plane and find its intercepts 兼怠 , 兼態 and 兼戴 with the crystallographic
axes in terms of the corresponding lattice parameters
怠 怠 怠
- take reciprocals , ,
陳迭 陳鉄 陳典
- scale all reciprocals by a common factor to get a set of three smallest integers; these
are the 月 倦, 健 values1

Example:

planes

兼怠 = 1 1/兼怠 噺 な 抜に蝦 月 噺に
兼態 = 2 1/兼態 噺 な に 抜 に 蝦 倦 = 1
兼戴 噺 タ な 兼戴 噺 ど 抜 に 蝦 "健 = 0

i.e. these are the (210) planes.

Crystallographic directions are labelled as [憲懸拳] where 憲, 懸 and 拳 are the smallest integers
such that the vector 憲軍怠 + 懸軍態 + 拳軍戴 points in desired direction (e.g. see vector [210]
above).
{月倦健} denotes sets of planes equivalent by symmetry.
極月倦健玉 denotes sets of directions equivalent by symmetry.
e.g., for a cubic system the planes (110), (101), 盤110匪 樺 岶ななど岼 and the directions
[210], [120], [102] 樺 極210玉. Negative numbers are denoted with a bar, e.g. 1博 噺 伐な.

1
Note that the Miller indices depend on the particular choice of primitive vectors. By convention, the body-
centred and face-centred cubic Bravais lattices are described in terms of their conventional cubic cells, i.e. as
simple cubic lattices with bases.

2-2

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Reciprocal space and the reciprocal lattice

As with many problems involving periodic functions, it is often convenient to transform the
problem from real space to Fourier space or “reciprocal space”. We will see later that the
diffraction pattern of any solid is simply proportional to the square of the Fourier transform of
its electron density. It is therefore convenient to define the Fourier space analogue of the
crystalline lattice, known as the reciprocal lattice.

To see its usefulness, consider a plane wave

畦結 沈圭糾慶 . We can represent its direction and
plane of
periodicity by a vector 圭 = (倦掴 , 倦槻 , 倦佃 ) in k- constant
phase
space. This “wavevector space”, spanned by
all possible values of 倦掴 , "倦槻 and 倦佃 is known
as reciprocal space.

Note that the phase factor 結 沈圭糾慶 is the same

for any set of parallel planes perpendicular to
態訂
圭 and separated by 膏 = . We could therefore associate a point 圭 in reciprocal space with
|圭|
態訂
sets of parallel planes perpendicular to 圭 and with spacing 穴 = .
|圭|

The set of all wavevectors 圭 that yield plane waves with the periodicity of a given Bravais
lattice is known as its reciprocal lattice. We will label all such wavevectors 屑 and they
correspond to all wavevectors satisfying 結 沈屑糾慶 = 結 沈屑糾(慶袋栗) for all lattice vectors, i.e.

結 沈屑糾栗 = 1 or""""屑 糾 栗 = 2講 × integer

In practice, for a given Bravais lattice with 栗 = 券怠 軍怠 + 券態 軍態 + 券戴 軍戴, the reciprocal lattice (RL)
is generated by all 屑 噺 月郡怠 + 倦郡態 + 健郡戴 where 月 倦 and 健 are integers from 伐タ to タ and the
primate lattice vectors 郡沈 are given by

2講 2講 2講
郡怠 = 軍態 × 軍戴 """""""""""郡態 = 軍戴 × 軍怠 """"""""郡戴 = 軍 × 軍態
撃 撃 撃 怠

and 撃 岩 軍怠 糾 (軍態 × 軍戴 ), its magnitude |撃| being the volume of the primitive unit cell of the
direct lattice.

Check for yourself that 軍沈 and 郡珍 satisfy "軍沈 糾 " 郡珍 = 2講絞沈珍 (where 絞沈珍 = 1 if 件 = 倹 and 0 if 件 塙 倹)
and therefore that 屑 糾 栗 = 2講 × integer is satisfied for all 屑 and 栗.

The reciprocal lattice of a Bravais lattice is also a Bravais lattice and its reciprocal is the
direct lattice.

(態訂)典
The volume of the primitive unit cell of the reciprocal lattice 郡怠 糾 (郡態 × 郡戴 ) = , where 撃椎頂
蝶妊迩
is the primitive cell volume of the direct lattice.

The units of reciprocal space vectors [屑] = [栗]貸怠 = m貸怠.

When 月 倦 and 健 have no common factor, 屑 噺 月郡怠 + 倦郡態 + 健郡戴 is perpendicular to lattice
態訂
planes with Miller indices (月倦健) and |屑| corresponds to the interplane spacing 穴; i.e. the Miller

2-3

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

indices of a lattice plane correspond to the 月, 倦, 健 coordinates of the shortest reciprocal

lattice vector normal to that plane.

For 屑朕賃鎮 with 月 倦 and 健 that do have a common factor, the planes with spacing

2講
穴朕賃鎮 =
|屑朕賃鎮 |

are not all lattice planes, but there will be a set of lattice planes with 穴 = 券 糾 穴朕賃鎮 where 券 is
an integer.

planes with spacing

lattice planes with

spacing

鳥
We can then rewrite Bragg’s law as 2 岾 峇 sin 肯 = 膏" or
津

2穴朕賃鎮 sin 肯 = 膏
態訂
Since 穴朕賃鎮 = |屑 , we have an easy way of calculating the interplane distances, e.g., for
廿入如 |
迭
態訂
cubic crystals |屑朕賃鎮 | = (月態 + 倦 態 + 健 態 )鉄 ( ) and
銚

頂通長沈頂
欠
穴朕賃鎮 =
ヂ月態 + 倦 態 + 健態

2-4

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Examples of reciprocal lattices:

3D orthorhombic lattice

Direct lattice: 軍怠 吃 軍態 吃 軍戴 , 欠怠 塙 欠態 塙 欠戴

Reciprocal lattice:
態訂"軍鉄 ×軍典 態訂 態訂"軍典 ×軍迭 態訂 態訂"軍迭 ×軍鉄 態訂
郡怠 = = 軍賦 , 郡態 = = 軍賦 , 郡戴 = = 軍賦
|軍迭 糾 (軍鉄 ×軍典 )| 銚迭 怠 |軍迭 糾 (軍鉄 ×軍典 )| 銚鉄 態 |軍迭 糾 (軍鉄 ×軍典 )| 銚典 戴

where 軍賦沈 is a unit vector in the direction of 軍沈 .

2D hexagonal lattice

怠 ヂ戴
Direct lattice: 軍怠 = 欠景, 軍態 = 欠 岾伐 景 + 桂峇
態 態

Reciprocal lattice (assume 軍戴 吃 軍怠 , "軍態 ):

態訂"軍鉄 ×軍典 態訂 態 ヂ戴 怠
郡怠 = = "岾 景 + 桂峇
|軍迭 糾 (軍鉄 ×軍典 )| 銚 ヂ戴 態 態

態訂"軍鉄 ×軍典 態訂 態
郡怠 = = "桂
|軍迭 糾 (軍鉄 ×軍典 )| 銚 ヂ戴

i.e., the reciprocal lattice of a 2D hexagonal lattice is a 2D hexagonal lattice rotated

by 30Δ with respect to the direct lattice.

1D lattice

Direct lattice primitive vector: 軍怠

Reciprocal lattice: (assume 軍戴 吃 軍怠 吃 軍態 )

態訂
郡怠 = 軍賦
銚迭 怠

BCC/FCC lattice

The reciprocal lattice of BCC is FCC and vice versa (see homework exercise).

2-5

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Laue condition for diffraction

The Bragg formulation of diffraction assumed specular reflection (angle of incidence = angle
of reflection) of X-rays from lattice planes. The von Laue approach assumes the crystal is
made of identical atoms (or sets of atoms) at lattice sites 栗 that scatter X-rays elastically, i.e.
without change in energy (wavelength) of the X-rays.

Consider elastic scattering of a plane

EM wave 結 沈圭糾慶 by an atom at 慶. Relative
to the wave scattered by an atom at the O
origin, it picks up an additional phase
態訂
difference 剛 = (健怠 + 健態 ). wave
碇
fronts
態訂
Since |圭| = |圭 嫗 | = (elastic
碇
scattering), 剛 = (圭 伐 圭 嫗 ) 糾 慶 岩 伐粂 糾 慶,
where 粂 = 圭 嫗 伐 圭 is the scattering vector (aka momentum transfer vector2).

If the atoms are separated by lattice vectors 栗 沈 , scattering from all atoms will be in phase if
粂 糾 栗 = 2講 × integer for all 栗 沈 . This is satisfied for all

粂=屑

This is the Laue condition for diffraction.

Equivalence of Bragg and Laue conditions

屑 is always perpendicular to the (月倦健) planes
態訂
with spacing 穴 = . Therefore, when the Laue
弁屑弁
condition is satisfied, 粂 is also normal to the
scattering planes. Since |圭| = |圭 嫗 |, we have the
usual Bragg (specular) geometry. Thus,
starting from the Laue condition, we have

2講
|屑| = |粂| = 2|圭| sin 肯 =
穴
態訂
Using |圭| = and rearranging gives 2穴 sin 肯 = 膏, i.e. Bragg’s condition.
碇

Ewald sphere

The Laue condition for diffraction is very restrictive. In practice, to access the different lattice
points in reciprocal space, we have to rotate the crystal (and/or the beam). The Ewald
construction gives a useful visual representation of what’s required to satisfy the Laue
condition.
Let’s fix the incoming beam direction to be horizontal, as in the figure below. As we rotate
the crystal around the different axes, its lattice and therefore its reciprocal lattice rotate with
it. For a certain X-ray wavelength 膏, the magnitude of 圭 (and 圭 嫗 ) is fixed and therefore all

2
The name momentum transfer comes from the fact that 屋圭 corresponds to the momentum of the X-ray
photon and therefore 屋粂 is the change in the momentum of the scattered photon.

2-6

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

possible scattering vectors 粂 噺 圭 嫗 伐 圭 lie on the surface of a sphere—the Ewald sphere—of

radius |圭|. The Laue condition can then be satisfied if the orientation of the crystal with
respect to the incident beam is such that a reciprocal lattice point at some 屑朕賃鎮 lies on the
Ewald sphere. Then, for an appropriate position of the detector (i.e. when the detector is
positioned at an angle に肯 that satisfies the Bragg law for 穴朕賃鎮 噺 に講 |屑朕賃鎮 |), diffraction will be
observed.

Calculation of scattered intensities

The Bragg and Laue conditions give us the general requirement for constructive interference
of waves diffracted by a crystalline lattice. To calculate the scattered intensities, we need to
add up the scattering amplitudes from each atom in the basis, or more precisely, from the
electron distributions of each atom, as it is the electrons that interact with the X-rays.

solid

incident
wave
O
scattered
wave

Consider a general solid. Assume amplitude of wave scattered from volume 穴撃 is

proportional to 券(慶)穴撃 where 券(慶) is the local electron density. As before, the additional
phase difference picked up by the wave (compared to that scattered at the origin) 剛 噺
(圭 伐 圭 嫗 ) 糾 慶 噺 伐粂 糾 慶, therefore the total amplitude of the EM wave scattered in direction 圭 嫗 is
proportional to the integral

繋(粂) 噺 豹 券(慶)結 貸沈粂糾慶 穴撃

鎚墜鎮沈鳥

2-7

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

This integral is known as the scattering amplitude. It is just the Fourier transform of the
electron density function 券(慶). If we could measure 繋(粂) for all the different 粂 (i.e. different
direction of 圭 伐 圭 嫗 ) we could obtain the electron distribution directly by inverse Fourier
transform of 繋(粂). Unfortunately, 繋(粂) is a complex function and experimentally we can only
measure the diffracted intensity 荊 苅 |繋|態 , thereby losing the information about the phase of
繋(粂).

If our solid is a crystal, 券(慶) will have the periodicity of the crystal, i.e. 券(慶) = 券(慶 + 栗).
Additionally, when the diffraction condition is satisfied 粂 = 屑 and

繋(屑) = 軽 豹 券(慶)結 貸沈屑糾慶 穴撃 岩 軽鯨屑

頂勅鎮鎮

Where 軽 is the number of unit cells in the crystal and 鯨屑 is known as the structure factor.
The entire function 券(慶) of a crystal is therefore encoded in a set of discrete complex
numbers 鯨屑 .

It is convenient to separate 券(慶) into contributions from individual atoms at positions 慶珍 , i.e.
券(慶) = デ長銚鎚沈鎚
珍 券珍 盤慶 伐 慶珍 匪 = デ長銚鎚沈鎚
珍 券珍 盤素珍 匪 where 素珍 = 慶 伐 慶珍 is the position vector relative to the
centre of atom. Then,

長銚鎚沈鎚

鯨屑 = 布 豹 券珍 盤素珍 匪結 貸沈屑糾素 "結 貸沈屑糾慶乳 穴撃

珍 頂勅鎮鎮

長銚鎚沈鎚

= 布 結 貸沈屑糾慶乳 豹 券珍 盤素珍 匪結 貸沈屑糾素 "穴撃

珍 頂勅鎮鎮

or
長銚鎚沈鎚

鯨屑 = 布 血珍 結 貸沈屑糾慶乳
珍

where 血珍 = 完頂勅鎮鎮 券珍 盤素珍 匪結 貸沈屑糾素 "穴撃 is the atomic form factor, the amplitude scattered by the
atom.

血珍 (屑) contains information about the structure of the atom (remember, 券(慶) = | |態 ). Most of
the electron density remains unchanged when atoms are integrated into a crystal and
therefore we can use the 血珍 calculated for isolated atoms. Atoms have spherical symmetry
and so do the 血珍 . If we assume atoms are point-like, i.e. its electron density distribution can
be represented as a delta function 券(素) = 傑絞(素), then 血 = 傑, the atomic number. In reality,
血珍 (屑) decreases with |屑| and tends to zero as |屑| 蝦 タ (e.g., see Kittel’s book). To
determine the crystal structure accurately we need to measure as many (月倦健) reflections as
possible but this becomes difficult for larger (月倦健) (larger with |屑|) since 血珍 (屑) 蝦 ど.

2-8

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Example

Diffraction from the FCC lattice

We are now ready to treat the diffraction from a structure with a multi-atom basis. The
FCC structure has four lattice points per conventional unit cell. We can decompose
this structure into a simple cubic lattice with a four atom basis:
銚 銚 銚 銚 銚 銚
(0,0,0), 岾 , , 0峇 , 岾 , 0, 峇 and 岾0, , 峇
態 態 態 態 態 態

The reciprocal lattice of a simple cubic lattice is also a simple cubic lattice with
態訂
reciprocal lattice vectors 屑 = (月 倦, 健). The structure factor is therefore
銚

庁寵寵
鯨朕賃鎮 = 布 血珍 結 貸沈屑糾慶乳 = 血盤1 + 結 貸沈訂(朕袋賃) + 結 貸沈訂(賃袋鎮) + 結 貸沈訂(朕袋鎮) 匪
珍

= 血盤1 + (伐な)朕袋賃 + (伐な)賃袋鎮 + (伐な)朕袋鎮 匪

4血""""""if 月倦健"are all even or all odd

=峽
0 otherwise

So, in order of increasing 月態 + 倦 態 + 健 態 (i.e. increasing |屑|), we have

月倦健 庁寵寵
鯨朕賃鎮 月態 + 倦 態 + 健 態
100 0 1
110 0 2
111 4f 3
200 4f 4
210 0 5
211 0 6
220 4f 8
221 0 9
300 0 9

Note that the 屑 vectors with non-zero intensity generate a BCC lattice (as expected,
since the reciprocal lattice of FCC is BCC!).

Similarly, for BCC we find that

喋寵寵 2血"""if 月 髪 倦 + 健 = even

鯨朕賃鎮 =峽
0 if 月 髪 倦 + 健 = odd

The missing peaks for certain (月倦健), e.g. (100) for FCC and BCC, are known as
missing orders. They are due to destructive interference by atoms in the basis. For
example, for the FCC structure, when Bragg’s law is satisfied for the (100) planes, the
atoms in the planes half-way between contribute a phase difference of 講 and thus
contributes scattered amplitude that is exactly out-of-phase.

2-9

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Structure determination from powder diffraction

X-ray powder diffraction is one of the most common methods for determining the structure of
materials.

The powder sample ensures that as long as に肯 is chosen to satisfy Bragg’s law for some
(月倦健), there will be some powder grains oriented correctly for specular ‘reflection’ of the X-
ray beam. The experiment then consists of scanning the angle に肯 and recording the
scattered intensity on the detector. Often neither the structure nor the lattice parameters are
known and therefore the simple procedure is to apply Bragg’s law on a trial and error basis
until a good fit is obtained.

Example

Let’s assume we want to find out whether a particular diffraction pattern corresponds to a
simple cubic or BCC structure.
銚
For a cubic material, we have 穴 噺 . Therefore, Bragg’s law に穴 œÆº 肯 噺 膏 gives
ヂ朕 鉄 袋賃 鉄 袋鎮鉄

膏態 態
sin態 肯 噺 (月 髪 倦 態 髪 健 態 )
ね欠態

If experimentally, the peaks in intensity are observed at に肯 values に肯怠 に肯態 に肯戴 …etc, then we
first need to index the observed reflections (i.e. identify their 月倦健 values). Since the lattice
parameter 欠 is also unknown, we eliminate it by taking ratios of the nth to 1st peak, which
according to the equation above should be given by

sin態 肯津 (月態 髪 倦 態 髪 健 態 )津
=
sin態 肯怠 (月態 髪 倦 態 髪 健 態 )怠

To distinguish between simple cubic and BCC, we calculate the (月態 髪 倦 態 髪 健 態 ) ratios for all
allowed (月倦健) reflections for the two structures and then compare these to the experimental
sin態 肯 ratios. For the simple cubic structure, all (月倦健) are allowed, whereas for BCC, only
those with 月態 髪 倦 態 髪 健 態 = even give non-zero intensity, therefore we have

2-10

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Simple cubic BCC

月倦健 月態 + 倦 態 + 健 態 (月態 + 倦 態 + 健 態 )津 月倦健 月 + 倦 態 + 健態
態 (月態 + 倦 態 + 健 態 )津
(月態 + 倦 態 + 健 態 )怠 (月態 + 倦 態 + 健 態 )怠
100 1 1 110 2 1
110 2 2 200 4 2
111 3 3 211 6 3
200 4 4 220 8 4
210 5 5 310 10 5
211 6 6 222 12 6
220 8 8 321 14 7

We see that for the two structures differ in the value of the 7th (月態 + 倦 態 + 健 態 ) ratio and
therefore experimentally one could distinguish between them by the absence or presence of
坦辿樽鉄 提韮
= 7.
坦辿樽鉄 提迭

For structures of lower symmetry than cubic, other peaks appear as some of the {月倦健}
planes that were equivalent by symmetry in the cubic structure are no longer symmetry
equivalent (i.e. the planes do not have the same spacing). For example, for the lattice below,
穴怠怠待 塙 穴拍怠怠待 and therefore the (110) and (1博 10) reflections will appear at different 2肯 angles.

Once the 月倦健 values for all the reflections are known, we can calculate the lattice
parameter(s) using Bragg’s law from a given reflection (or subset of reflections).

2-11

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Diffraction from powdered samples leads to rings of intensity at specific values of に肯, known
as Debye-Scherrer rings.

2-12

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Examples of powder diffraction patterns from cubic crystals:

(011)
膏 = 0.154 nm, 欠 = 0.412 nm

(112)
(001)

Simple
(002)

(012)
(111)

(013)
(022)
cubic

(122)

(113)
(no restrictions)

膏 = 0.154 nm, 欠 = 0.564 nm

(002)

(022)

(024)

Face centred
(222)

cubic
(004)
(111)

h,k,l all even or all odd

(113)

(133)

only

2-13

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Brillouin zones
As for the direct lattice, we can define unit cells for the reciprocal lattice. The most important
unit cell is the Brillouin Zone (BZ), which is the Wigner-Seitz cell of the reciprocal lattice.
Recall that it is formed from perpendicular bisectors of lines from a reciprocal lattice point
(that we choose as the origin) to all nearest reciprocal lattice points. E.g., for a square
(reciprocal) lattice we have

Brillouin zone
boundary

1st Brillouin
zone

The first Brillouin zone has the same volume as the primitive unit cell of the reciprocal lattice.
Higher zones can also be defined, e.g., the 2nd BZ is formed by the perpendicular bisectors
to the next nearest neighbours, etc. All BZs have the same volume.

The Brillouin zone plays a very special role in the theory of propagation of waves in crystals
(be it electromagnetic waves, electron waves, lattice waves, etc.) and facilitates the
statement of a very general condition for diffraction of waves in crystals.

For the square lattice above, when the Laue condition 屑 = 粂 = 圭 嫗 伐 圭 is satisfied, 圭
terminates on the BZ boundary (BZB). This is a sufficient condition for diffraction, i.e., if a
wavevector terminates on the BZB, the wave will be diffracted.

Direct lattice Reciprocal space

BZB

Lattice planes

2-14

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Proof:

The Laue condition says that diffraction occurs when 圭 嫗 伐 圭 = reciprocal lattice vector. If
when 圭 嫗 伐 圭 is a reciprocal lattice vector, then so is when 圭 伐 圭旺 (in the sketch above, both
屑怠待宋 and 屑怠待宋 satisfy the Laue condition). Let this reciprocal lattice vector be 屑, i.e. let
圭 伐 圭 嫗 = 屑, then

圭嫗 = 圭 伐 屑

|圭 嫗 |態 = |圭|態 + |屑|態 伐 に圭 糾 屑

For elastic scattering considered here, |圭 嫗 | = |圭| and therefore

2圭 糾 屑 = |屑|態

撫= 屑,
or, defining the unit vector 屑 |屑|

1
撫 = |屑|
圭糾屑
2
怠
So, when the component of 圭 parallel to 屑 is |屑| (i.e., when 圭 lies on the Brillouin zone
態
boundary), the Laue condition is satisfied and the wave will be diffracted. We will see later
the consequences of this for lattice waves and electron waves in crystals.

2-15

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Lattice dynamics
So far we have assumed the atoms to be stationary. In reality this is never he case as even
at zero temperature the atoms undergo zero point motion in accordance with Heisenberg’s
Uncertainty Principle. As the atoms are coupled through the interatomic potential, their
oscillations around the equilibrium positions are highly correlated and can be described in
terms of lattice waves. These lattice dynamics play an important role in the thermal
properties of solids (e.g. specific heat, thermal conductivity).

Elastic waves in a solid

Before addressing the coupled vibrations of atoms, we review the propagation of elastic
waves in the continuum approximation—the long wavelength limit where the solid is
considered to have a uniform density. The elastic distortions propagate at the speed of
sound 懸鎚 . Displacements of the material 掲 can be parallel or perpendicular to the
propagation vector 圭 (|圭| = 2講/膏).

displacements !

Longitudinal wave, L. 掲 勘 圭 Transverse wave, T. 掲 吃 圭

(compression wave) (shear wave)

The displacements 掲 are given by

掲 = 掲宋 結 沈(圭糾慶貸摘痛) in 3D

or

憲 = 憲墜 結 沈(賃掴貸摘痛) in 1D

The speed of sound 懸鎚 = 降/倦 is equal to the slope of the

dispersion relation (the plot of 降 vs 倦), and is constant
(independent of 倦 or ‘dispersionless’).
擢摘
The group velocity 懸直 = is equal to 懸鎚 . In general, the
擢賃
speed of sound for transverse and longitudinal waves is
not the same (懸鎚挑 塙 懸鎚脹 ).

3-1

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

The Harmonic approximation

To describe lattice vibrations we will model the lattice as atoms attached by ideal springs of
spring constant 系. This is equivalent to approximating the interatomic potential 戟(堅) by a
harmonic potential and is known as the Harmonic
approximation.

Expanding 戟(堅) about the equilibrium separation 欠"

擢腸 (追貸銚)鉄 擢鉄 腸
戟(堅) = 戟(欠) + (堅 伐 欠) 岾 峇 + 岾 鉄峇 髪橋
擢追 追退銚 態 擢追 追退銚

擢腸
Since we are expanding about the minimum, 岾 峇 =0
擢追 追退銚
and for small (堅 伐 欠) we can truncate the expression at the
quadratic term, leaving us with

(堅 伐 欠)態 項 態 戟
戟(堅) = 戟(欠) + 峭 態嶌
2 項堅 追退銚

This is the Harmonic approximation. The force 繋 is

項戟 項態戟
繋噺伐 噺 伐岫堅 伐 欠) 峭 態 嶌
項堅 項堅 追退銚

Comparing this to Hooke’s law 繋 噺 伐系(堅 伐 欠), shows that the spring constant 系 is the
curvature of the interatomic potential around the equilibrium separation

項態戟
系=峭 嶌
項堅 態 追退銚

Excitations of 1D monoatomic chain

Consider a 1D chain of identical atoms of mass 兼 connected by ideal springs of spring
constant 系 with equilibrium separation 欠.
1

If the n atom is displaced by 憲津 from its equilibrium position, it will experience a net force
th

繋津 = 系(憲津袋怠 伐 憲津 ) 伐 系(憲津 伐 憲津貸怠 )

= 系(憲津袋怠 伐 に憲津袋 + 憲津貸怠 )

1
The displacements of the masses in our 1D model could represent the displacements of atomic planes in a
crystal.

3-2

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

nd th
From Newton’s 2 law, the equation of motion for the n atom is

兼憲岑 津 = 系(憲津袋怠 伐 に憲津袋 + 憲津貸怠 )

轍
We look for wave-like solutions 憲津 = 憲待 結 沈盤賃掴韮 貸摘痛匪 where 捲津待 = 券欠 is the equilibrium position
of the n atom, 倦 = 2講/膏 and 膏 is the wavelength of the lattice wave.
th

Using 憲津袋怠 = 結 沈賃銚 憲津 and 憲津貸怠 = 結 貸沈賃銚 憲津 , the equations of motion become

伐兼降態 憲津 = 系盤結 沈賃銚 伐 に 髪 結 沈賃銚 匪憲津

怠
which, using cos 倦欠 = (結 沈賃銚 + 結 貸沈賃銚 ) gives
態

伐兼降態 = 2系(cos 倦欠 伐 な)

倦欠
噺 伐ね系 sin態
2

i.e.

4系 態 倦欠
降態 = sin
兼 2

This is the dispersion relation for a 1D monoatomic chain. It describes how the frequency of
the wave changes with wavevector (wavelength). Every 倦 represents a solution.

Note that the dispersion relation is independent of 券, i.e., we get the same relation starting
from any atom. Note also that although the motions of individual atoms are coupled (if one
atom starts vibrating it will transfer energy to adjacent atoms and set them into motion), our
wavelike solutions are uncoupled. Each 倦 has a definite frequency 降 and oscillates
independently of other modes, i.e., the solutions are normal modes. This independence is a
result of the harmonic approximation (i.e., the fact that we assumed ideal springs with spring
constants that are independent of how much the spring is extended or compressed by a
passing lattice wave).

3-3

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Finite chain of 軽 atoms

For an infinite chain of atoms, any value of 倦 is a solution, i.e. 降(倦) is a continuous function.
For a finite chain of 軽 atoms, we should expect to find 軽 modes corresponding to 軽
equations of motion. To do so, we need to take into account the boundary conditions of the
ends of the chain. There are two possibilities:
i) Consider the ends to be either fixed or free
ii) Periodic boundary conditions (a.k.a. Born-von Karman boundary conditions)
Both give the same physics as long as they are applied consistently. We will choose periodic
boundary conditions, which can be visualised as a chain with its ends joined to form a ring
(for fixed-end boundary conditions see Kittel, chapter 5).
Periodic boundary conditions mean 憲津 = 憲津袋朝 , i.e. 憲待 結 沈賃津銚 =
憲待 結 沈賃(津袋朝)銚 , so that 倦軽欠 = 2講喧, where 喧 = 0, ±1, … is an integer
2講
倦= 喧
軽欠
態訂
Therefore there are 軽 possible values of 倦 within a range of
銚
訂 訂
(e.g. for 伐 <倦判 ) spaced by
.銚 銚
2講
倦=
軽欠
In practice, 軽 is a very large number and therefore 降(倦) is essentially a continuous function
of 倦 for many intents and purposes.
Uniqueness of the 1 st Brillouin Zone
訂 訂
As in our discussion of X-ray diffraction, the first BZ, i.e., 伐 判 倦 判 has a very special role
銚 銚
as it contains all the physically unique solutions 倦.
From the periodicity of the dispersion relation we can see that waves with 倦 differing by a
態訂
reciprocal lattice vector 罫 = have the same frequency 降. In fact, they actually also
銚
represent the same pattern of atomic displacements, i.e. the same lattice wave:
2講 沈賃津銚 態訂
沈岾賃袋 峇津銚
倦蝦倦+ ,結 蝦結 銚 = 結 沈賃津銚 結 沈態訂津 = 結 沈賃津銚
欠
態訂
The displacements of discrete atoms represented by a high-frequency wave 倦 + and a
銚
low frequency wave 倦 are indistinguishable.

3-4

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

st
Thus the wavevectors within the 1 BZ represent a complete set of solutions. The total
訂
number of modes within the 1 BZ is 岾2 峇 倦 = 軽, as expected.
st

Note, the modes ±倦 are distinguishable because of their time dependence 憲津 (建) =
憲待 結 沈(賃津銚貸摘痛) . It is a travelling wave and can have positive or negative phase velocity 降/倦.
訂
But at the BZ boundary where 倦 = ± , this sense of direction is lost
銚

憲津 (建) = 憲待 結 沈(±津訂貸摘痛) = 憲待 (伐な)津 結 貸沈摘痛

This is a standing wave with opposite displacements of adjacent atoms. Therefore the
訂
倦 = ± solutions are the same.
銚

Recall that a wave with 倦 on the BZB satisfies the Bragg condition and will therefore be
訂
diffracted. So we can think of the incident wave 倦 = ± being scattered into the opposite
銚
訂
direction to form a wave with 倦 噺 噛 . The incident and backscattered waves superpose to
銚
form a standing wave. When the Bragg condition is satisfied, a travelling wave cannot
propagate in a lattice.
When 倦 = 講/欠, 膏 = 2講/倦 = 2欠 and satisfies Bragg’s law
券膏 = 2欠 sin 肯 with 券 = 1 and 肯 = 90° (2肯 = 180°).

Group velocity, 懸直
The group velocity represents the velocity of wave packet propagation

項降 項 系 倦欠 系 倦欠
懸直 = = 嵜2俵 sin 崟 = 欠俵 cos
項倦 項倦 兼 2 兼 2

懸直 蝦 ど as 倦 蝦 "講/欠"(at BZB).
In the long-wavelength limit, 倦 蝦 ど

系 倦欠 系 倦欠
降(倦) = 2俵 sin 蛤 に俵
兼 2 兼 2

降 系
懸椎 = = 欠俵 = 懸直 = 懸鎚
倦 兼

The phase velocity 懸椎 equals the group velocity 懸直 and is equal to the speed of sound 懸鎚 .
The linear (“dispersionless”) region near 倦 = 0 corresponds to the long-wavelength elastic
waves in a solid; i.e. when 膏 伎 interactomic spacing, the detailed atomic structure becomes
irrelevant and we recover the same behaviour as in a continuous medium.

3-5

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Excitations of 1D diatomic chain

We now consider a 1D chain with a two-atom basis, the two types of atom having masses 兼
and 警 (兼 < 警).

The equations of motion are

警憲岑 津 = 系(懸津 伐 に憲津 + 懸津貸怠 )
兼懸岑津 = 系(憲津袋怠 伐 に懸津 + 憲津 )
Try travelling wave solutions
憲津 = 憲待 結 沈(賃津銚貸摘痛)
懸津 = 懸待 結 沈(賃津銚貸摘痛)
Substituting into equations of motion gives
伐降態 警憲待 = 系盤懸待 + 懸待 結 貸沈賃銚 伐 に憲待 匪
伐降態 兼懸待 = 系盤憲待 結 沈賃銚 + 憲待 伐 に懸待 匪
We can express these coupled equations as a matrix equation
2系 伐 降態 警 伐系"盤1 + 結 貸沈賃銚 匪 憲待
峭 嶌 岾懸 峇 = 宋
伐系盤1 + 結 沈賃銚 匪 2系 伐 降態 兼 待

which only has solutions if the determinant of the matrix is zero, giving
降替 警兼 伐 降態 2系(警 + 兼) + 4系 態 伐 系 態 婀娶 沈賃銚
娶娶結娶娶娶婬娶
盤1 + 匪盤な娶伐 沈賃銚
娶娶結娶娶婉匪 = 0
態袋態 達誰坦 賃銚

降替 警兼 伐 降態 2系(警 + 兼) + 2系 態 (な 伐 ̊æœ 倦欠) = 0

Solving for 降態 we get the dispersion relation

2系(警 + 兼) ± 紐4系 態 (警 + 兼)態 伐 ぱ系 態 警兼(な 伐 ̊æœ 倦欠)

降態 =
2警兼
Note, two atoms per primitive cell gives us 2 branches in the dispersion relation.
Let’s examine the limiting cases:
訂 怠
1) Small 倦, 倦 企 蝦 な 伐 ̊æœ 倦欠 蛤 倦 態 欠態
銚 態
怠
態
2系(警 + 兼) 2系(警 + 兼) 2警兼 1 態
降 = ± 磐な 伐 態
糾 "倦 態 欠態 卑
2警兼 2警兼 (警 + 兼) 2
婀娶娶娶娶娶娶娶婬娶娶娶娶娶娶娶婉
怠 暢陳
蛤磐怠貸 賃 鉄 銚 鉄 "卑
態(暢袋陳)鉄
so
2系(警 + 兼) 1 1 2系
降態 蛤 = 2系 磐 + 卑 =
警兼 警 兼 航
where 航 is the “reduced mass”; or

3-6

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

系
降態 蛤 倦 態 欠態
2(警 + 兼)
態寵(暢袋陳)
i.e. 降 = 謬 – independent of 倦
暢陳
寵 寵
or 降 = 欠謬 "倦 – linear in 倦 with 懸鎚 = 懸直 = 懸椎 = 欠謬 ;
態(暢袋陳) 態(暢袋陳)

寵
(compare this with 降 = 懸鎚 倦, 懸鎚 = 欠謬 for monoatomic chain).
陳

訂
2) At zone boundary, 倦 = , な 伐 ̊æœ 倦欠 噺 な 伐 (伐な) = 2, therefore
銚
系(警 + 兼) 2系 怠
降態 = ± ((警 + 兼)態 伐 ね警兼)態
警兼 2警兼
迭
The last term in brackets is equal to ((警 伐 兼)態 )鉄 = (警 伐 兼) and therefore
系(警 + 兼) ± 系(警 伐 兼) 2系 2系
降態 = = or
警兼 兼 警

optical branch

gap (=0 if M=m)

acoustic branch

3-7

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Repeated zone scheme

Reduced zone scheme

Extended zone scheme

3-8

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Knowing the dispersion relation, we can use the matrix equation to find the relationship
between 憲待 and 懸待 :
For optical modes, 憲待 and 懸待 have opposite sign
2
-
- For acoustic modes, 憲待 and 懸待 have the same sign
The atomic displacements are easier to visualise for the transverse (rather than longitudinal)
modes

(zone centre modes)

Acoustic
(low freq.)
Rigid displacement of the chain

Optical
(high freq.)

(zone boundary modes; only or is moving)

Generalisation to 3D
In 3D we always have 3 acoustic branches: two transverse (TA) and one longitudinal (LA)
For a two-atom basis, there is then a total of 2 × 3 = 6 branches: 3 acoustic and 3 optical.
For a 喧-atom basis, there is a total of 3喧 branches: 3 acoustic and 3(喧 伐 な岻 optical.
Again, for each longitudinal optical branch, there are two transverse.

2
Optical modes are so called because if the two different masses are ions of opposite charge then optical
modes lead to creation of dipoles as the positive and negative ions move in opposite direction. Such modes
can be excited by the electric field of an electromagnetic wave of optical frequency, hence optical modes.
Acoustic modes are so-called because long wavelength acoustic modes correspond to sound waves.

3-9

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Some examples of phonon dispersion relations

Solid neon (monoatomic, fcc)

Reproduced from Y. Endoh et al., Physical Review B 11, 1681 (1975)

- All modes are acoustic (just one type of atom)
- Sinusoidal dispersion
- Transverse modes are degenerate (i.e. have the same frequency and hence energy)
along high symmetry directions (e.g. [100], [111])
- More complex along [110] (next-nearest neighbour interactions)

NaCl (fcc with 2-atom basis)

Reproduced from G. Raunio et al., Physical Review 178, 1496 (1969)

- 2-atom basis so now have acoustic and optical modes
- Along [111] we have alternating planes of Na and Cl atoms, hence the dispersion
along reciprocal space direction labelled 岫行 行 行岻 is most like our linear chain model

3-10

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Quantisation of lattice vibrations

So far we have considered the lattice vibrations from a purely classical point of view but just
as for EM waves, the energy of lattice vibrations is quantised. By analogy with photons, a
quantum of energy of a mechanical excitation is called a phonon. In the quantum theory of
phonons, each elastic mode of frequency 降 is treated as a quantum harmonic oscillator
(QHO) and has energy
1
竺 噺 磐券 + 卑 屋降
2
怠
The 屋降 is the zero-point energy due to Heisenberg’s Uncertainty Principle. When a mode
態
is excited to quantum number 券, we say it is occupied by 券 phonons. Classically, we can
think of larger 券 as corresponding to a larger amplitude of the wave.
Each mode 倦 corresponds to a QHO with frequency 降(倦), so the total energy is
1
継 = 布 竺(倦) = 布 磐券(倦) + 卑 屋降(倦)
2
賃 賃

Phonons are boson (spin 0) so there is no restriction on the number 券 of phonons in a mode.
They obey Bose-Einstein statistics and 券 is given by
1
券(降) =
結 屋摘/賃遁畷 伐な
For many purposes we can think of phonons as a sort of particle (a “quasiparticle”) that
carries energy 屋降 and momentum 屋倦 known as crystal momentum. It is not a true
momentum as phonons involve only relative displacements of atoms rather than the
displacement of the crystal as a whole but crystal momentum can be exchanged during
interactions with external probes (e.g. EM waves, neutrons) and is conserved within a
reciprocal lattice vector 屑.
Inelastic scattering from phonons
Phonon dispersion curves can be measured using neutron scattering. When neutrons are
scattered inelastically by a crystal (i.e., final energy 塙 initial energy), phonons are created or
destroyed. Both energy and momentum are conserved, however, as the crystal momentum
is only defined to within a reciprocal lattice vector, momentum conservation involves a 屑"that
brings 圭 丹竪誰樽誰樽 to the 1 BZ.
st

Energy conservation
罰屋降椎朕墜津墜津 = 継沈 伐 継捗
Momentum conservation neutron

±圭 丹竪誰樽誰樽 = 圭 辿 伐 圭 脱 + 屑
A neutron spectrometer measures 継捗 and
圭 脱 so that the 圭 and 降 of the phonons can
be reconstructed to give the dispersion
curve.

3-11

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Phonons: Thermal Properties

Phonons can be created by thermal excitation of the lattice and play a very important role in
the thermal properties of materials. We will now look at two such properties:
1. Heat capacity, which can be treated within the harmonic approximation.
2. Thermal conductivity, which requires us to go beyond the harmonic approximation
and consider interactions between phonons (anharmonic effects)

Heat capacity
The total heat capacity of a crystal can be written as
椎朕墜津墜津鎚 陳銚直津勅痛沈頂
系蝶 = 婀娶
系蝶娶婬娶 系蝶勅鎮勅頂痛追墜津鎚
娶婉 + 婀娶 娶婬娶 娶婉 + 系
婀娶
蝶娶婬娶
娶婉
銚鎮鎮"鎚墜鎮沈鳥鎚 陳勅痛銚鎮鎚"墜津鎮槻 陳銚直津勅痛鎚"墜津鎮槻

For a non-magnetic insulator, 系蝶 is only due to phonons

項竺
系蝶 = 磐 卑
項劇 蝶
Note, for solids 系蝶 蛤 系牒 .

Dulong & Petit law

Classically, if we think of a solid as 軽 atoms, equipartition theorem would give
1
竺 噺 は軽 × 倦喋 劇
2
Where the factor of 6 comes from 6 degrees of freedom per atom (KE + PE × 3 spatial
degrees of freedom each). Hence
系蝶 = 3軽倦喋
This is known as the Dulong and Petit law (~1819). It holds for some solids at room
temperature but fails at low temperature where a quantum mechanical treatment is
necessary.

Simple harmonic oscillator

Let’s first take a look at the heat capacity of a single quantum harmonic oscillator (QHO).
The energy eigenvalues are
1
竺津 = 磐券 + 卑 屋降
2
To find the average thermal equilibrium occupancy 極券玉 of this QHO (i.e., how many phonons
are in mode 降 at a given temperature), we use the usual statistical mechanics approach.
結 貸竺韮 /賃遁 脹 デ津 券"盤結 貸屋摘/態賃遁 脹 匪結 貸津屋摘/賃遁 脹
極券玉 = 布 券"喧津 = 布 券 =
傑 デ津 (結 貸屋摘/態賃遁 脹 )結 貸津屋摘/賃遁脹
津 津

4-1

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

貸竺韮 /賃遁 脹
where 喧津 is the probability of state 券 being occupied and 傑 = デ津 結津 is the partition
function. Cancelling out the two terms in brackets,
デ津 券"結 貸津屋摘/賃遁 脹 倦喋 劇 態 項 ln 畦
極券玉 = =
デ津 "結 貸津屋摘/賃遁 脹 屋降 項劇
貸怠
where 畦 = デ著津退待 結
貸津屋摘/賃遁 脹
= 1 + 結 貸屋摘/賃遁脹 + 結 貸態屋摘/賃遁 脹 髪 橋 噺 盤な 伐 結 貸屋摘/賃遁脹 匪 (geometric
series). Now,
倦喋 劇 態 項 岫伐غ岫な 伐 結 貸屋摘/賃遁脹 )) 倦喋 劇 態 1 屋降 貸屋摘/賃 脹
極券玉 = = 磐伐 結 遁 卑
屋降 項劇 屋降 結 貸屋摘/賃遁脹 倦喋 劇
which simplifies to
1
極券玉 =
結 屋摘/賃遁 脹 伐な
This is just the Bose-Einstein factor, the same as the well-known expression for the Planck
distribution for black body radiation. Therefore, the average energy of this oscillator is
1 1 屋降
竺違 = 磐極券玉 + 卑 屋降 = 屋降 + 屋摘/賃 脹
2 2 結 遁 伐な

We can now calculate the heat capacity of the QHO

項竺違 1 屋降
系蝶 = 峭 嶌 噺 屋降 磐伐 屋摘/賃 脹 卑 磐伐 卑 結 屋摘/賃遁 脹
項劇 蝶 結 遁 伐な 倦喋 劇 態

屋降 態 結 屋摘/賃遁脹
系蝶 = 倦喋 磐 卑
倦喋 劇 (結 屋摘/賃遁脹 伐 な)態
屋摘
or, defining a characteristic temperature 岩
賃遁
態
結 誕/脹
系蝶 = 倦喋 磐 卑
劇 (結 誕/脹 伐 な)態
誕 態
In the low-temperature limit, as 劇 蝦 ど, 結 誕/脹 伎 な and 系蝶 蝦 " 倦喋 岾 峇 結 貸誕/脹 .
脹

誕 誕 態 怠
In the high-temperature limit, as 劇 蝦 タ, 結 誕/脹 蛤 な 髪 髪 橋 and 系蝶 蝦 " 倦喋 岾 峇 鉄 = 倦喋 ,
脹 脹 到
岾怠袋 貸怠峇
畷
as expected from equipartition.

Einstein’s theory of heat capacity in solids

Einstein considered the solid as a collection of 軽 simple QHOs all with the same frequency
降帳 (corresponding to typical optical mode frequency ~10 Hz). If we define an Einstein
12

temperature as 帳 噺 屋降/倦喋 , then in 3D

態
醍 結 誕吐 /脹
系蝶 = 3軽倦喋 磐 卑
劇 (結 誕吐 /脹 伐 な)態
For 劇 伎 帳, 系蝶 蛤 ぬ軽倦喋 , i.e. we recover the law of Dulong and Petit.

4-2

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

For 劇 企 " 帳 , 系蝶 苅 結 貸誕吐 脹 . This explains qualitatively the observed decrease of 系蝶 at low
temperature, but experimental data do not decrease as quickly— to explain this we need low
frequency (acoustic) phonons.

Einstein theory fit (dashed line) to experimental heat capacity data for diamond (open
circles); 帳 = 1,320 K. Reproduced from Einstein’s seminal 1907 article in Annalen der
Physik [vol. 22, pp. 180-190].

Total lattice energy

To calculate the total lattice energy at a certain temperature more accurately we need to
sum up the contributions from all phonon modes, i.e., over all 圭 and for all branches 倹
1
竺 噺 布 布 屋降圭 珍 磐券盤降圭 珍 匪 + 卑
2
圭 珍

Since the spacing between the states 倦 噺 に講 詣" is very small compared to に講 欠, the width
1
of the Brilouin zone , we can replace the sum with an integral
著
1
竺 噺 豹 訣(降)屋降 磐券(降) + 卑 穴降
待 2
where 訣岫降岻 is the density of states such that the number of states between 降 and 降 髪 穴降 is
訣(降)穴降. In reality, 訣(降) is a complicated function of 降, so we need some approximations.

Debye theory
Einstein assumed a single frequency 降帳 . Debye took into account different frequencies
assuming a linear dependence of 降 on 倦, i.e., 降 噺 懸聴 倦. We now know these as acoustic
phonons (though Debye’s theory predates the phonon theory of lattice dynamics and was
based on classical elasticity). The assumptions of the model are:

態訂 挑
Typically 欠ｂなど貸怠待 m, 倦ｂなど貸態 m貸怠 so that ~10腿
1
態訂 銚

4-3

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

- The crystal is harmonic (the phonon modes are independent)

- Elastic waves are non-dispersive (降 噺 懸鎚 倦)
- The crystal is isotropic
- There is a high-frequency cut off 降帖 chosen in such a way that the total number of
摘呑
modes is correct, i.e. that per branch 完待 訣(降)穴降 噺 軽, the number of atoms in the
crystal

Density of states
We now need an expression for the density of states.
態訂
In 1D, the states are separated by 倦掴 = , where 詣 is the dimension (length) of the crystal.
挑
態訂
In 2D, the allowed states form a grid with spacing 倦掴 噺 倦槻 = .
挑

Consider an annulus of radius 倦 and thickness 穴倦. The area occupied by 1 state in k-space
態訂 態 挑 態
is 岾 峇 or, equivalently, there are 岾 峇 states per unit area. So within the annulus area
挑 態訂
挑鉄
に講倦"穴倦 there are に講倦"穴倦 (態訂)鉄 states, hence in 2D

詣態
訣(倦)穴倦 噺 倦"穴倦""""""岫ƺ"に&岻
に講
where 訣(倦) is the density of states (DOS).
態訂 戴
In 3D, we have to consider a shell of volume ね講倦 態 "穴倦. The volume of each state is 岾 峇 or
挑
挑 戴 挑典
there are 岾 峇 states per unit volume. Hence 訣(倦)穴倦 噺 ね講倦 態 (態訂)典 穴倦"or using 撃 噺 詣戴
態訂

撃 態
訣(倦)穴倦 噺 倦 穴倦"""""""岫ƺ"ぬ&岻
に講 態
The corresponding DOS per unit frequency 訣岫降岻 is given by

4-4

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

訣(降)穴降 = 訣(倦)穴倦
撃 態 穴倦
訣(降) = 倦 (in 3D)
2講 態 穴降
per branch of the dispersion curve (e.g., for a monoatomic crystal in 3D there are 2T and 1L
branches).
To calculate the Debye frequency 降帖 , we use
摘呑
軽=豹 訣(降)穴降 (per branch)
待

Using Debye’s approximation for the dispersion 降 = 懸鎚 倦

穴倦 1 撃 降態
= 蝦 訣(降) = 態 戴
穴降 懸鎚 2講 懸鎚
Therefore
摘呑
撃 降態 撃 1 降帖 戴 戴
6軽講 態 戴
軽=豹 穴降 = "磐 卑 蝦 """" 降 帖 = 峭 嶌 懸鎚
待 2講 態 懸鎚戴 2講 態 3 懸鎚 撃
With three acoustic branches, the total lattice energy
著 著
1
竺 噺 ぬ豹 屋降"訣(降)"穴降 + 3 豹 屋降"訣(降)券(降)穴降
婀娶娶娶娶娶婬娶娶娶娶娶婉
待 2 待
竺二鍋

can now be written as

摘呑
撃 降態 1
竺 噺 竺跳牒 + 3 豹 屋降" 態 戴"" 屋摘/賃 脹 "穴降
待 2講 懸鎚 結 遁 伐な

蝶 怠 戴朝
or, using =
態訂鉄 塚濡典 典
摘呑

9軽 摘呑 屋降戴
竺 噺 竺跳牒 + 戴 豹 ""穴降
降帖 待 結 屋摘/賃遁 脹 伐 な
Noting that the zero-point energy 竺跳牒 is independent of temperature and therefore doesn’t
contribute to the heat capacity, we can now calculate the heat capacity as
項竺 9軽 摘呑 屋降戴 屋降
系蝶 = = 戴豹 結 屋摘/賃遁 脹 磐伐 卑 """穴降
項劇 降帖 待 伐岫結 屋摘/賃 脹
遁 伐 な岻態 倦喋 劇 態
9軽 摘呑 屋態 降替 結 屋摘/賃遁脹
= 豹 穴降
降帖戴 待 倦喋 劇 態 (結 屋摘/賃遁脹 伐 な岻態
Letting 捲 噺 屋降/倦喋 劇, 穴降 = (倦喋 劇 屋岻穴捲 and the limits of integration become 降 噺 ど 蝦 捲 = 0
and 降 = 降帖 蝦 捲 噺 屋降帖 /倦喋 劇 噺 帖 /劇, so that
9軽 誕呑 /脹 屋 倦喋 劇 替 捲 替 結 掴
系蝶 = 戴 豹 磐 卑 穴捲
降帖 待 劇 屋 (結 掴 伐 な岻態
戴 誕呑 /脹
劇 捲替結 掴
系蝶 = 9軽倦喋 磐 卑 豹 穴捲
帖 待 (結 掴 伐 な岻態
屋摘
In the high-temperature limit, as 劇 蝦 タ, 捲 = 蝦 ど so that the integrand in the above
賃遁 脹
expression can be approximated to

4-5

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

捲 替 (1 + 捲 + 捲 態 /2 髪 橋 岻
蛤 捲態
(1 + 捲 + 捲 態 /2 髪 橋 伐 な岻態
怠 誕呑 戴
which integrates to 岾 峇 and therefore the heat capacity in the high temperature limit
戴 脹

系蝶 蝦 ぬ軽倦喋
consistent with the Dulong & Petit law.
In the low-temperature limit, as 劇 蝦 ど, 捲 蝦 タ and we can replace 降帖 in the upper limit of the
integration with タ. The integral then becomes a standard integral
著
捲替結 掴 4講 替
豹 穴捲 =
待 (結 掴 伐 な岻態 15
and
12 劇 戴
系蝶 蝦 "軽"倦喋 講 替 磐 卑
5 帖

This expression gives good agreement with experiment.

4-6

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Density of states as a function of frequency for (a) the Debye solid and (b) an actual
crystal structure. The spectrum for the crystal starts as 降態 for small 降, but
discontinuities develop at singular points. Reproduced from Kittel’s book.

(a) Dispersion relation for a diatomic crystal. (b) Dispersion relation within the Debye
approximation.

(a) Heat capacity of solid argon at low temperature plotted against 劇 戴 . (b) Comparison
of Debye and Einstein models. Inset shows the difference in behaviour at low
temperatures. Experimental data for silver. Reproduced from Kittel.

4-7

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Anharmonicity
So far, we have considered our interatomic potential to be harmonic, i.e., our model springs
exert the same force whether extended or compressed and the spring constant is
independent of the magnitude of the displacement. The Harmonic approximation works well
for small deviations from the equilibrium positions but for large displacements (high 劇) it
becomes bad.

Core
repulsion

Harmonic approx.
(symmetric)

Coulomb/van der
Waals attraction

For a more accurate description, we need to go beyond the harmonic terms in the expansion
(堅 伐 欠)態 項 態 戟 (堅 伐 欠)戴 項 戴 戟
戟(堅) = 戟(欠) + 峭 態嶌 + 峭 戴嶌 髪橋
2 項堅 追退銚 婀娶娶娶娶娶娶娶婬娶娶娶娶娶娶娶婉
婀娶娶娶娶娶娶娶娶婬娶娶娶娶娶娶娶娶婉 6 項堅 追退銚
張銚追陳墜津沈頂"銚椎椎追墜掴沈陳銚痛沈墜津 銚津朕銚追陳墜津沈頂"痛勅追陳鎚

The consequences of including the anharmonic terms include:

1. Thermal expansion—the average interatomic distance 極堅玉 > 欠 (it costs less energy to
expand than to contract)
2. Phonons interact with each other—they are no longer pure normal modes. In an
anharmonic crystal, a phonon (lattice wave) causes momentary local
contraction/expansion of the lattice. This causes a second phonon to ‘see’ a different
spring constant and the modes are therefore not independent; phonons scatter off
each other and this determines the thermal conductivity of the solid.
We will not discuss thermal expansion here as this was covered in PHAS1228.

4-8

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Conservation laws for phonon-phonon collisions

Phonon-phonon collisions obey the following conservation laws

ŒæŒ̋ºß­Œ """""""屋圭 層 髪 "屋圭 匝 噺 屋圭 惣

"""""""""̋º̋ø̌ """""""屋 怠 髪屋 態 噺屋 戴

but we must not forget that phonon momentum is only defined modulo a reciprocal lattice
vector and therefore one distinguishes between two types of scattering events:

1. The resulting wavevector 圭 惣 lies within the 1 BZ. This is known as a ‘Normal’ event
st

(N). In an N event, two phonons travelling to the right combine to produce a phonon
travelling to the right.
2. The resulting wavevector 圭 惣 lies outside the 1 BZ but can be translated into the 1
st st

BZ by a reciprocal lattice vector 屑. Such events are called ‘Umklapp’ events (U) . In a
2

U event, two phonons travelling to the right combine to produce a phonon travelling
to the left!
Heat flow carried by phonons is unaffected by N events but is impeded by Umklapp
scattering and therefore U events contribute to thermal resistivity 岾1斑thermal conductivity峇

Thermal conductivity
Thermal conductivity 腔 relates the steady-state flow of heat across a solid to the temperature
gradient, i.e.
穴劇
倹通 噺 伐腔
穴捲
thermal current density [Wm貸態 ] 噺 " 伐ß̶̋øŒ̇Ø"̊溸­̊ßÆß "[Wm貸怠 K 貸怠 ] × temp. gradient [Km貸怠 ]

2
‘Umklapp’ means ‘flip-over’ in German. Umklapp events were discovered by Peierls.

4-9

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

From kinetic theory of gases:

1
腔 = 懸違 "健"系寞蝶
3
where 懸違 is the average particle velocity, 健 is the mean free path and 系寞蝶 is the specific heat
per unit volume.
Importantly, particle conservation is not used in the derivation of this result and therefore it
3
can be applied to phonons, which are bosons and hence do not conserve particle number.

hot cold
end end

phonons created phonons destroyed

(more phonons) (less phonons)

In metals, electrons contribute to 腔 (we will evaluate this contribution later); in insulators, the
only contribution is from phonons.

Kinetic theory of gases

Imagine that the two ends of a solid are held at
different temperatures. Moving by 捲 from 劇 髪 劇 to 劇,
the particle (phonon) gives up energy 潔 劇, where 潔 is
the specific heat per particle (if 券 is the particle density
per unit volume, 潔 = 系寞蝶 /券).
If the time between collisions (scattering time) is 酵
穴劇 穴劇
劇= 捲= 懸 酵
穴捲 穴捲 掴
where 懸掴 is the component of the velocity along 捲. The thermal current density is then
倹通 = 蔽
券 婉 × 極懸
婀娶婬娶劇)
× (伐潔 陛 掴玉
鳥勅津鎚沈痛槻 朕勅銚痛 塚勅鎮墜頂沈痛槻

穴劇
噺 伐券潔極懸掴 玉態 酵
穴捲
Therefore 腔 = 券潔極懸掴 玉態 酵 = 系寞蝶 極懸掴 玉態 酵. For an isotropic distribution of velocities, 極懸掴 玉態 = 極懸槻 玉態 =
怠 怠 怠
極懸佃 玉態 = 極懸玉態, so that 腔 = 系寞蝶 極懸玉態 酵 = 系寞蝶 極懸玉健, or
戴 戴 戴
1
腔 = 系寞蝶 懸違 健
3
Only Umklapp collisions contribute.
(Remember, 系寞蝶 is the heat capacity per unit volume)

3
In fact, it works even better for phonons because the speed of sound 懸聴 is a constant for acoustic
phonons not too close to the BZB, unlike 懸 for ideal gas particles, which follows the Maxwell-
Boltzmann distribution.

4-10

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Temperature dependence of thermal conductivity

The temperature dependence of the thermal conductivity is determined mostly by 系寞蝶 and 健 (懸違
is approximately constant). The mean free path 健 is inversely proportional to the number of
phonons.
High T (劇! 帖 ): all phonon modes up to 屋降帖 are excited; there are many phonons with large
enough |倦| to produce U-events. The number of phonons is proportional to the thermal
energy (倦喋 劇), hence 健 苅 劇 貸怠 . The heat capacity is approximately constant (classical value).
Intermediate T (劇ｂ 帖 ): U-processes start to freeze out as average energy of phonons
decreases. The number of phonons ｂ ̋®ı岫伐 帖 /劇"), hence 健 苅 ̋®ı岫 帖 /劇).
Low T (劇 企 帖 ): all U-processes switch off. 健 is limited by the size and shape of the crystal.
The temperature dependence comes entirely from 系寞蝶 ~(劇 帖 )戴.

crystal
size

limited by limited by

4-11

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Thermal conductivity of natural abundance

(1.1% 13C) diamond (lower squares),
isotopically enriched (0.1% 13C) diamond
(upper squares), together with the low
temperature data of Slack (circles) and the
high temperature data of Vandersande et al.
(plusses). Reproduced from Wei et al. Physical
Review Letters 70, 3764 (1993).

Thermal conductivity of isotopically pure LiF

showing the effect of boundaries for
sandblasted crystals. Mean crystal widths: (A)
7.25 mm, (B) 4.00 mm, (C) 2.14 mm, (D) 1.06
mm. Reproduced from Thatcher Physical
Review 156, 975 (1967).

4-12

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Free electron theory of metals

Electronic properties
Solids are often classified according to their electrical conductivity as metals, insulators and
semiconductors. The electrical conductivity 購 ( 貸怠 m貸怠 ) relates the current density 倹 (A m貸態)
and the electric field 継 (V m貸怠)
倹 = 購継.
For a solid of length 詣, cross section 畦 and resistance 迎
詣
購=
迎畦
怠 眺凋
and the resistivity 貢 = = (with units of Œ).
蹄 挑

As a rough guideline:
Metals: 10貸腿 < 貢 < 10貸泰 Œ 貢 increases with 劇
Semiconductors: 10貸泰 < 貢 < 10 Œ 貢 decreases with 劇
Insulators: 貢 > 10 Œ
Different electronic properties arise from differences in the energy states available for the
electrons. When atoms form a solid, the sharp atomic energy levels broaden to form bands.
Electrons are fermions and at 劇 = 0 fill up the available energy levels according to the Pauli
Exclusion Principle up to some energy 香庁 (Fermi energy). In metals, there is no gap between
the highest occupied and lowest empty state. The electrons are easily excited into nearby
empty states and therefore can give rise to a current, as we shall see. In semiconductors
there is an energy gap of order 倦喋 劇 between occupied and unoccupied states and therefore
thermal energy is required for conductivity. For insulators, the gap is much larger than 倦喋 劇.
The electronic structure also explains the optical properties of materials, e.g. why metals are
shiny/opaque whereas insulators are transparent or coloured.

5-1

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Density of states, Atom Metal

Density of states,
filled empty

Energy, Energy,
Atoms – discrete energy levels; Metals – transitions possible at
optical transitions at single all
energy (wavelength, )

Semiconductor/insulator
Density of states,

Semiconductor if
gap
Insulator if

filled empty Gap, is related to colour

Energy,

Metals
Soon after the discovery of the electron (J. J. Thompson, 1897), Drude proposed a classical
theory of conductivity in metals. He considered the conduction electrons as a gas of classical
particles in a box and applied the kinetic theory of gases. The Drude model could explain
Ohm’s law for electrical conductivity limited by collisions and the relationship between
thermal and electrical conductivities of metals but also had some notable failures, e.g. it
grossly overestimated the heat capacity. To improve on Drude’s model requires quantum
mechanics. The first attempt to apply quantum statistics to explain metals was by
Sommerfeld.

5-2

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Free-electron theory of metals

The free-electron model of metals is built on the idea that electrons occupy electronic states
in accordance with the Pauli Exclusion Principle, i.e. obeying Fermi-Dirac statistics. We
ignore (for now) the crystalline lattice and assume:
- A fixed background of static1 positive charges due to the nuclei (with core electrons)
- The valence electrons are completely delocalised and propagate freely without
interacting with the ion cores or other electrons
This is the simplest model of metals but it is found to work well for the simple alkali earth
metals (Na, K, Rb, Cs).
We consider the conduction (valence) electrons to be confined in a box of dimension 詣
corresponding to the size of the crystal. The potential energy inside the box is assumed
constant and arbitrarily set to zero. Essential predictions do not depend on how deep the
well is. The key point is that the behaviour of the electrons is governed by the Pauli
Exclusion Principle.
In 1D, Schroedinger’s equation with 撃(捲) = 0 is
屋態 項 態 閤
伐 = 香閤.
2兼 項捲 態
The eigenstates are plane waves
1
閤(捲) = 結 沈賃掴
ヂ詣
with energy eigenvalues
屋態 倦 態
香= .
2兼
We will assume periodic boundary conditions such that 閤(捲 + 詣) = 閤(捲), or 結 沈賃(掴袋挑) = 結 沈賃掴 ,
態訂
which gives the allowed 倦 values as 倦 = 喧, where 喧 is an integer.
挑

In 3D, we have
屋態
伐 椛閤(慶) = 香閤(慶)
2兼
With eigenstates and eigenvalues given by
1 1 屋態 倦 態
閤(慶) = 結 沈圭糾慶 = 結 沈(賃猫 掴袋賃熱 槻袋賃年佃) ,"""香 = .
ヂ撃 ヂ撃 2兼
態訂
Applying the periodic boundary conditions, gives the allowed 圭 = (喧, 圏, 堅), with integer
挑
喧, 圏, 堅.

1
This is known as the “adiabatic approximation”

5-3

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Density of states (DOS)

Just as for phonons we can define density of states functions 訣(倦) and 訣(香) in terms of 倦
and 香 (rather than 降 for phonons). In 3D, the number of states with 倦 between 倦 and 倦 + 穴倦
is

態
1 撃 態
訣(倦)穴倦 = 婀娶婬娶婉
4講倦 穴倦 糾 戴 = 倦 穴倦
旦誰狸探鱈奪"誰脱"坦竪奪狸狸 2講 2講 態
岾婀婬婉 峇
詣
#"坦担叩担奪坦"丹奪嘆"
探樽辿担"旦誰狸探鱈奪

According to the Pauli Exclusion Principle, each state can be occupied by at most two
electrons with opposite spin (i.e. can’t have all the electrons in the ground state). The DOS
as a function of energy is therefore
撃 態
訣(香)穴香 = 2訣(倦)穴倦 = 倦 穴倦"
講態
or
撃 態 穴倦
訣(香) = 倦
講 態 穴香
屋鉄 賃 鉄
Using 香 = , we can eliminate 倦 to give
態陳

撃 2兼 戴/態 怠/態
訣(香) = 磐 卑 香
2講 態 屋態
i.e., 訣(香) 苅 香 怠/態 .
At 劇 = 0, all the states up to the Fermi energy 香 = 香庁 are filled and the total number of
electrons in the crystal is
敵鈍 敵鈍
撃 2兼 戴/態 怠/態 撃 2兼 戴/態 2 戴/態
軽 = 豹 訣(香)穴香 = 豹 磐 卑 香 穴香 = 磐 卑 ""香
待 待 2講 態 屋態 2講 態 屋態 3 庁
Rearranging for 香庁 gives
屋態 軽 態/戴
香庁 = 磐3講 態 卑
2兼 撃
which increases with increasing electron density (due to Pauli repulsion between the
electrons). In 倦-space, this corresponds to filling of all the 倦-states within a sphere of radius
態陳 怠/態
倦庁 where 倦庁 = 岾 香 庁 峇
屋鉄

軽 怠/戴
倦庁 = 磐3講 態 卑 Fermi
撃
surface
which depends only of the electron density. The
surface of this sphere is known as the Fermi
surface. The density of states on the Fermi surface
is
3軽
訣(香庁 ) = .
2 香庁
filled empty
The Fermi surface determines the electrical and states states

5-4

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

thermal properties of metals.

Some typical values for monovalent metals:
Na Cu
Electron density, 2.65 × 10態腿 m貸戴 8.45 × 10態腿 m貸戴
錆/惨
Fermi wavevector, 暫擦 9.2 × 10苔 m貸怠 13.6 × 10苔 m貸怠
Fermi wavelength, 0.68 nm 0.46 nm
似擦 = 匝慈/暫擦 (~interatomic spacing)
Fermi velocity, 1.07 × 10滞 m s 貸怠 1.57 × 10滞 m s貸怠
士擦 = 屋暫擦 /仕 (large but not relativistic)
Fermi energy, 而擦 3.23 eV 7.0 eV
Fermi temperature, 35,000 K 81,000 K
参擦 = 而擦 /暫刷

Electrons near the Fermi surface have a wavelength comparable to the interatomic spacing
and a kinetic energy comparable to the binding energy.
Thermal excitations
Thermal energy (~0.025 eV at room temperature) cannot affect the deepest electron states,
only those very close to the Fermi surface. Being fermions, electrons obey Fermi-Dirac
statistics, with the probability of occupying a state with energy 香 given by the Fermi-Dirac
distribution function
1
血(香) =
結 岫敵貸禎岻 賃遁脹 +1
where 航 is the chemical potential. Unlike bosons, which can be created or destroyed,
fermions are conserved particles and must be supplied by a reservoir at energy 航. The
chemical potential ensures that the total number of particles 軽 is correct. We can think of it
as a normalisation parameter which adjusts the “zero” of the energy in the Schrödinger
equation.

5-5

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Electronic heat capacity

One of the major successes of the free electron model is its explanation for the additional (to
phonons) heat capacity of metals. Here we derive an approximate expression for 系蝶 due to
electrons.
In thermal equilibrium, the number of electrons per unit energy range is given by the product
of the density of states and the Fermi-Dirac distribution function
券(香, 劇) = 訣(香)血(香, 劇)

(width of transition region greatly exaggerated for clarity)

怠
The shaded areas above can be approximated as triangles with height 訣(香庁 ) and base
態
怠
~2倦喋 劇. Therefore, compared to 劇 = 0, approximately 訣(香庁 )倦喋 劇 electrons increase their
態
energy by about 倦喋 劇, i.e., the change in total energy of all the electrons is
1
継(劇) 伐 継(0) 蛤 訣(香庁 )倦喋態 劇 態 .
2
Differentiating this w.r.t. 劇"gives the electronic heat capacity 系蝶勅鎮 蛤 訣(香庁 )倦喋態 劇, or using the
戴
expressions 訣(香庁 ) = 軽/香庁 and 香庁 = 倦喋 劇庁
態
3軽 態 3 劇
系蝶勅鎮 蛤 倦喋 劇 = 軽倦喋 磐 卑
2 香庁 2 劇庁
An exact calculation gives
講態 劇
系蝶勅鎮 = 軽倦喋 磐 卑.
2 劇庁
We have seen that 劇庁 ~10,000 K and therefore the electronic contribution to 系蝶 is much
smaller than the classical equipartition result of 3軽倦喋 . At room temperature, it is also much
椎朕墜津墜津
smaller than the lattice (phonon) heat capacity (系蝶 ~3軽倦喋 at high temperature). At very
椎朕墜津墜津
low temperatures, however, 系蝶 蝦 ど as 劇 戴 , whereas 系蝶勅鎮 苅 劇, so the electronic
contribution can be observed. For a non-magnetic solid, at low temperature
椎朕墜津墜津
系蝶痛墜痛銚鎮 = 系蝶 + 系蝶勅鎮 = 紅劇 戴 + 紘劇.

5-6

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Experimentally, one usually plots 系蝶 劇 vs 劇 態 which gives a straight line with slope 紅 and
intercept 紘.

Experimental heat capacity values for potassium at low temperature. Reproduced from Kittel;
data from W. H. Lien and N. E. Phillips, Physical Review 133, A1370 (1964).
訂鉄 朝 態
Note that our free electron expression for 紘 contains no unknown quantities 岾紘 噺 倦 峇
態 敵鈍 喋
鉄
屋鉄 賃鈍
and since 香庁 = , 紘 苅 兼. Experimentally measured values for 紘, however, tend to be
態陳
larger than this theoretical value. To account for the difference, we define an “effective mass”
兼茅 as
兼茅 紘 岫墜長鎚勅追塚勅鳥岻
= 岫捗追勅勅"勅鎮勅頂痛追墜津岻
兼勅 紘
where 兼勅 is the free electron mass. Typical values of 兼茅 兼勅 are 1.26 (Na), 1.4 (Cs, Cu), 2.0
(Pb) etc. For some metals, however, the free electron description breaks down
spectacularly, e.g., “heavy fermion” metals can have effective masses of around 1,000兼勅 .

Electrical conduction
In the absence of electric fields or thermal gradients, there is no net flow of electrons in any
direction. The Fermi surface is symmetric around 圭 噺 ど and therefore for any occupied state
圭 with momentum 屋圭 there is an occupied state – 圭 with opposite momentum.
In an electromagnetic field, electrons experience the Lorentz force
d 穴圭
釧 噺 伐結(櫛 髪 携 抜 遇) = (兼携) 噺 屋
dt 穴建
(Although this equation applies to a localised particle, we can think about the electro as a
wavepacket formed from superposition of plane waves).
First, let us neglect the magnetic field (setting 遇 to zero) and consider the effect of 櫛.
穴圭
屋 噺 伐結櫛
穴建
Integrating w.r.t. time gives
結櫛
圭噺伐 建
屋

5-7

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

This tells us that all the electrons are accelerated by the field in the same way, i.e. the Fermi
surface shifts by 圭 in the direction opposite to the field.

If there were no restoring forces, electrons would accelerate indefinitely in the presence of
the field. In reality, electrons are scattered off impurities/defects, phonons, and other
electrons. These collisions reset the electron momentum and limit the velocity to its steady
state value. If the average time since the electron’s last collision is 酵, the steady state value
of 圭 is
結櫛
圭噺伐 酵
屋
and the average drift velocity of the electrons is
屋 圭 結櫛酵
携辰 = 噺伐
兼 兼
We can describe the effect of collisions by adding a damping term to our equation of motion
穴携 携
兼磐 + 卑 噺 伐結(櫛 + 携 × 遇)
穴建 酵
We call 酵 the relaxation time. The relaxation time ensures that the electron has a finite
steady state drift velocity and that in the absence of any fields (i.e. 櫛 = 遇 = 0), the velocity
decays to zero:
穴携 携
噺伐 蝦 """"携 = 携待 結 貸痛/邸
穴建 酵
The electrical current density is then
券結 態 酵
啓 = 券"圏"極携袈 玉 = 櫛
兼
This is just Ohm’s law 啓 = 購櫛 with
券結 態 酵
購=
兼
Example:
Consider sodium, Na, which has a conductivity of 2 × 10胎 貸怠 m貸怠 at 300 K. Na is a
monovalent metal with a bcc structure with lattice parameter 欠 = 4.25 Å. There are two
態
lattice sites per unit cell and hence 2 electrons/unit cell, giving 券 = = 2 × 10態腿 m貸戴.
銚典
We can therefore determine the relaxation time 酵 = 購兼/券結 態 = 2.6 × 10貸怠替 s.

5-8

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Alternatively, we can calculate the mean free path 健 for electrons at the Fermi surface
2

with a Fermi velocity 懸庁 as 健 噺 懸庁 酵 蛤 ぬどど"D, which is approximately 100 lattice

spacings.
At low temperature, the conductivity of sodium is 5.3 × 10怠待 " 貸怠 m貸怠 . The
corresponding relxation time and mean free path are 酵 噺 ば 抜 など貸怠怠 s and 健 蛤 ばばど どどど"D
– very large!
So how much is the Fermi surface shifted by in an applied field? If 酵ｂなど貸怠替 s, then for a field
勅帳邸
of 1,000 V m貸怠 (1 V across a 1 mm sample) 倦噺伐 ~10替 m貸怠 . The typical Fermi
屋
綻賃
wavevector 倦庁 is of order 10怠待 m貸怠 , so that ~10 貸滞
, i.e. the Fermi surface is only weakly
賃鈍
perturbed by the applied field.

For a perfect (pure) crystal, the dominant collisions are with phonons giving a temperature
dependent scattering time 酵椎朕 (劇). At hight temperature (劇 伎 帖 ), the number of phonons is
roughly proportional to 劇, therefore 酵椎朕 苅 な 劇; as 劇 蝦 ど 酵椎朕 蝦 タ.
Real metals have defects (impurities, vacancies, structural defects) which scatter electrons
giving a finite scattering time 酵待 even at 劇 噺 ど. Assuming the different scattering
mechanisms are independent, the scattering
rates add
1 1 1
= +
酵 酵椎朕 酵待
怠 陳 怠
Since resistivity 貢 噺 = 苅 , the
蹄 津勅 鉄 邸 邸
contributions to resistivity from phonon and
defect scattering add
兼 1 1
貢噺 態 峭 + 嶌 噺 貢沈 (劇) 髪 貢待
券結 酵椎朕 酵待
This is known as MattheisenÕs rule. 貢沈 is the
ideal resistivity, which is proportional to 劇 at
high temperature due to phonon scattering,
whereas 貢待 is the residual resistivity
determined by purity and structural perfection
of the sample. The ratio of the total resistivity Temperature-variation of resistance
at room temperature over that at low of potassium below 20 K for two
temperature 貢(300K) 貢待 is known as the samples of different purity.
residual resistivity ratio. Reproduced from Macdonald and
Mendelssohn, Proc. Royal Soc.
London, Series A 202, 103 (1950).

2
Because all the collisions involve only electrons near the Fermi surface, as that’s where the only accessible
empty states for scattering into are.

5-9

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Thermal conductivity
In insulators, heat transport is only due to phonons. In metals, conduction electrons also
contribute and thus metals usually have higher thermal conductivities. We can again use the
result from kinetic theory of gases
1
腔 = 系寞蝶 懸庁 健
3
where we take 懸庁 as the mean velocity since only electrons within ~倦喋 劇 of 香庁 can change
寵
their occupation as 劇 varies. The mean free path is then 健 = 懸庁 酵. Using 系寞蝶 = 楠 =
蝶
訂鉄 朝 脹 怠
倦 岾 峇 and 香庁 = 兼懸庁態 we arrive at
態 蝶 喋 脹鈍 態

講 態 券倦喋態 劇酵
腔=
3 兼
This is generally much larger than the phonon contribution. For example, at room
temperature:
Quartz (isulator): 腔 蛤 なぬ Wm貸怠 計 貸怠
Copper (metal): 腔 蛤 ねどど Wm貸怠 計 貸怠
But
Diamond (insulator): 腔 蛤 に どどど Wm貸怠 計 貸怠
(Diamond has a very high Debye temperature ( 帖 > 2,000 K) related to its stiff bonding and
therefore low phonon scattering).

Wiedemann-Franz law
For many simple metals, the ratio of thermal to electrical conductivities is proportional to
temperature with a constant of proportionality that is independent of the material. This is
known as the Wiedemann-Franz law
腔
=詣
購劇
The constant 詣 is known as the Lorenz number ad has a value of approximately 2 ×
10貸腿 W K 貸態.
The Lorenz number can be obtained from free electron theory
講 態 態 券酵
腔 磐 倦 劇 卑 講態 倦 態
3 喋 兼 喋
= 態 = 磐 卑 = 2.45 × 10貸腿 W K 貸態
購劇 結 酵券 3 結
磐 卑劇
兼
and is one of the triumphs of the free electron model.
However, 詣 is not always found to be universal. The Wiedemann-Franz law breaks down
because of our incorrect assumption that electrical and thermal transport are limited by the
same scattering processes, i.e. that 酵痛朕勅追陳銚鎮 = 酵勅鎮勅頂痛追沈頂銚鎮 . More accurately,
腔 酵痛朕勅追陳銚鎮
=詣
購劇 酵勅鎮勅頂痛追沈頂銚鎮

5-10

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Electrical conductivity is due to a uniform shift in 圭 of the

Fermi sphere, therefore the conductivity will be limited by
processes that restore the Fermi surface to the original
position (i.e. its position at zero electric field). This
typically requires a large change in momentum, i.e.
phonons with large |圭|, which are only available at high
temperatures.

Typical relaxation processes for

electrical conductivity`
In thermal transport, there is no net flow of electrons but
electrons travelling in one direction have a broader
distribution of wavevectors around 倦庁 than those going in
the other direction. The thermal conductivity 腔 is therefore
limited by processes that even out the difference in
distributions. This does not necessarity required large |圭|
phonons.

In both cases, only electrons close to the Fermi energy

can be scattered since the maximum phonon energy
Typical relaxation processes for
倦喋 帖 is much smaller than 香庁 = 倦喋 劇庁 and therefore is thermal conductivity`
insufficient to promote electrons from deep inside the
Fermi sphere to available states.

The Hall effect

So far we have ignored the effect of magnetic fields. When a metal carrying a current is
placed in a magnetic field 遇, a transverse electric field 櫛滝 develops perpendicular to the
current density 啓 and to 遇 and is given by
櫛滝 = 迎滝 "遇 × 啓
This is known as the Hall effect and 迎滝 is the Hall coefficient.
Consider a metal block in the shape of a parallelepiped carrying a current in a magnetic field
as shown. If the electron velocity is 携, the Lorentz force on the electron 伐結携 × 遇 will deflect
the electrons downwards and negative charge density will accumulate on the lower surface
until the electric field it produces 櫛滝 counteracts the Lorentz force.

5-11

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

+ + + + + +
+ + + + + +

– – – – – –
– – – – – –

If 櫛 is the total field (i.e. 櫛滝 plus the applied field required to drive the current), the equation
of motion for the electron is
穴携 携
兼 磐 + 卑 噺 伐結櫛 伐 結携 × 遇
穴建 酵
鳥携
In steady state, = 0 and
鳥痛
兼携
噺 伐結櫛 伐 結携 × 遇
酵
Since in steady state 懸槻 = 懸佃 = 0, 携 × 遇 is purely along 検, its 捲-component is zero and
therefore along 捲 we have
兼懸掴
噺 伐結継掴
酵
which is just the equation of motion in the absence of a magnetic field.
Along 検, the cross product gives 伐懸掴 稽佃 and since 懸槻 = 0,
ど 噺 伐結継槻 伐 結(伐懸掴 稽佃 )
倹掴 稽
継槻 = 懸掴 稽佃 = 岩 迎滝 倹掴 稽
伐券結
So the Hall coefficient is
1
迎滝 噺 伐
券結
This theoretical Hall coefficient is negative and is a direct measure of the electron
concentration.
So how does this compare with experiment?
Metal Li Na K Rb Cu Be Cd Zn
Valence 1 1 1 1 1 2 2 3
嗣酸蚕伺司姿 蚕姉使 0.8 1.2 1.1 1.0 1.5 伐0.2 伐1.2 伐0.8
三屈 /三屈

While the prediction works relatively well for the simple alkali metals, for others not only the
magnitude but even the sign is not predicted correctly. This is one of the big failures of the
free electron model. To improve on this, we need to take the lattice into account.

5-12

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Successes and failures of the free electron model

The free electron model successfully predicts/explains
- the magnitude and temperature dependence of the heat capacity for simple metals
- the order of magnitude of the thermal conductivity
- Lorenz number for simple metals
- the Hall effect in some simple metals
- the magnetic susceptibility of simple metals (later)
However it fails to explain
- why some materials are not metals (i.e. insulators, semiconductors)
- why the mean free path of electrons is much larger than the interatomic distances
- the positive Hall coefficients of some metals
- the large effective masses obtained from heat capacity measurements

5-13

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Electron bands
So far we have ignored the periodic potential due to the ion cores which is ultimately
responsible for the failures of the free electron model. The most important effect of the
crystalline potential is the creation of energy bands, which we will now approach from two
(opposite) starting points:
1. The nearly free electron model, which treats the crystalline potential as a perturbation
on the free electron states.
2. The tight binding model, where electron bands emerge due to overlap of the
wavefunctions of isolated atoms as they are brought closer together to form a solid.

Nearly free electron model

We have seen that for monovalent sodium, 倦庁 corresponds to 膏庁 of the order of the lattice
spacing. For higher valences, 倦庁 will be even higher. At some point, 倦庁 will reach the
Brillouin zone boundary (BZB). As we saw with X-rays and lattice waves, when 圭 lies on the
BZB, it satisfies the Bragg condition and therefore just like any other wave, the electron
wave will be elastically scattered to produce a wave with 圭 嫗 related to 圭 by a reciprocal
lattice vector (圭 嫗 伐 圭 = 屑). The effect of this is to open an energy gap 香直 at the BZB.

訂
To see why the gap appears, let’s consider a 1D crystal. In 1D, an electron wave with 倦 =
銚
(i.e. 膏 = 2欠) satisfies the Bragg equation 2穴 sin 肯 = 券膏 for 穴 = 欠 and 券 = 1 if 2肯 = 180°, i.e.
訂
the electron wave will be backscattered to produce a wave with 倦 嫗 噺 伐 travelling in the
銚
訂
opposite direction. This wave also satisfies the Bragg condition and will be scattered to 倦 =
銚
and so on. The two waves will interfere to produce a standing wave. There are therefore two
possibilities for the electron wavefunction 閤:
1 沈訂 沈訂
閤± = 磐結 銚 掴 ± 結 貸 銚 掴 卑
ヂ2詣
or, equivalently,

6-1

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

2 講捲 2 講捲
閤 袋 = 俵 cos 岾 峇 """""""""and 貸
= 件 俵 sin 岾 峇
詣 欠 詣 欠

The probability density for a particle is 貢 = 閤 茅 閤 = |閤|態 . Whereas for a travelling wave
怠 怠
貢 = 結 沈賃掴 結 貸沈賃掴 = (i.e. uniform everywhere), for the two standing waves
挑 挑
2 講捲 2 講捲
|閤 袋 |態 = cos 態 岾 峇 """""""""and |閤 貸 |態 = sin態 岾 峇"
詣 欠 詣 欠
If the positive ion cores are located at 捲 = 0, 欠, 2欠 …, we can see that |閤 袋 |態 piles up
electron density on the positive ions whereas |閤 貸 |態 piles up electron density in between the
ions. We would therefore expect the state described by the wavefunction 閤 袋 to have a lower
energy than 閤 貸 , i.e. there will be an energy gap between these two states at the BZB.

positive
ion core
periodic crystalline
potential,

So how big is the gap?

Since the potential 撃(捲) has the periodicity of the crystalline lattice, we can always express it
as a Fourier series
2講券捲
撃(捲) 噺 伐 布 撃津 cos 磐 卑
欠
津

Where 撃津 are the Fourier coefficients and the minus sign is chosen for convenience. Note,
態訂
券 are just the reciprocal lattice vectors 屑 for the 1D case. If we treat this potential as a
銚
perturbation on the free electron states, we can use perturbation theory to calculate the
change in energy of these states due to 撃(捲). In first order perturbation theory, if the
Hamiltonian 茎 with eigenstates 閤 is perturbed by 茎旺 (i.e. 茎 蝦 茎 + 茎旺), the change in energy
of the states 香 is given by
香 噺 完 閤 茅 茎 嫗 閤"穴慶 戴
The energy gap 香直 between the two states 閤 袋 and 閤 貸 at the BZB is therefore
香直 噺 香 貸 伐 香 袋
挑
2講券捲 2 講捲 講捲
香直 = 豹 伐 布 撃津 cos 磐 卑 峽sin態 岾 峇 伐 ̊æœ態 岾 峇峺 穴捲
待 欠 詣 欠 欠
津

6-2

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

2 挑 2講券捲 2講捲
= 豹 布 撃津 cos 磐 卑 cos 磐 卑 穴捲
詣 待 欠 欠
津

Interchanging the order of the sum and the integral and using the property of orthogonal
integrals
態訂
豹 cos(兼捲) cos(券捲) 穴捲 = 講絞陳津
待

where 絞陳津 = 1 if 兼 = 券 and 0 if 兼 塙 券, we find that only the 券 = 1 term in the sum gives a
non-zero integral and therefore arrive at
挑
2 2講捲
香直 = 撃怠 豹 cos態 磐 卑 穴捲 = 撃怠
詣 待 欠
st st
i.e. the magntitude of the gap at the 1 BZB is simply the 1 Fourier coefficient of the crystal
態訂
potential. The same calculation shows that the gap at 倦 = ± is given by the second
銚
Fourier coefficient 撃態 , etc. More generally, in 3D, we can write down the Fourier series
expansion as
1
撃(慶) = 布 撃屑 結 沈屑糾慶
2
屑

where the sum runs over all reciprocal lattice vectors 屑, and 撃屑 are the corresponding
屑
Fourier coefficients. The gap at the BZB at 圭 = is then 香直 = |撃屑 |.
態

Metals, insulators and semiconductors

The effect of the periodic potential is thus to open energy gaps and split the allowed electron
states into energy bands. We saw earlier that in order to produce an electric current, the
applied field needs to shift the Fermi surface by 圭. This requires that empty states are
available within an energy range that is accessible with an applied electric field. For the 1D
case, we can see that if an electron band is partially full, states are available close in energy
for an electron to be accelerated into by the applied field.

Metal Insulator

6-3

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

A partially filled band therefore corresponds to a metal. If, however, the band is full, the
nearest available states are a band gap away in energy and unless the band gap is
vanishingly small, this leads to insulating behaviour. In cases where the gap is comparable
to 倦喋 劇, electrons may be thermally excited across the gap and give rise to conductivity. Such
materials are called semiconductors (and they become insulators at low temperature).

So what is the requirement to have a full band? We need to know how many states are in a
Brillouin zone.

In 1D, application of the periodic boundary conditions showed that the allowed states are
spaced by 倦 = 2講/詣. If our 1D crystal is composed on 軽 unit cells with lattice parameter 欠
(and therefore has length 詣 = 軽欠), the number of states within the 1 Brillouin zone (between
st

訂 訂
倦 噺 伐 and ) is
銚 銚

2講
岾 峇 詣
欠 氾
2講 = 欠 = 軽
岾 峇
詣

i.e., just as in our discussion of phonons, the number of allowed 倦 states in the 1 Brillouin
st

zone is equal to the number of primitive cells in the crystal.

Since each 倦-state can be occupied by two electrons (with spin 菓 and 課) each Brillouin zone
contains 2軽 allowed electron states. Let’s now consider the following two possibilities.
One electron per primitive cell
If each primitive cell contains only one atom and each atom is monovalent (i.e. has only one
st
valence electron), then only half of the electron states in the 1 Brillouin zone will be filled
and such a crystal will be a metal.
Two electrons per primitive cell
If each primitive cell contains two valence electrons (i.e. if each primitive cell contains one
divalent atom or two monovalent atoms), then all 2軽 states in the first Brillouin zone will be
occupied and such a 1D crystal will be an insulator.

nd rd
Since higher Brilouin zones (2 , 3 , etc.) also contain 2軽 electron states each, a 1D crystal
will be metallic if it has an odd number of electrons per primitive cell, and insulating if it has
an even number of electrons per primitive cell.

In 2D, the states now form a grid with 倦掴 噺 倦槻 = 2講/詣, each state occupying an area
(2講/詣)態 . The area of the Brillouin zone is (2講/欠)態 so the number of states in the first (and
any other) Brillouin zone is
2講 態
岾 峇
欠 氾 詣態
= =軽
2講 態 欠態
岾 峇
詣

6-4

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

where again 軽 is the number of primitive unit cells in the crystal. As in the 1D case, the total
number of electron states in 2D (and by the same argument in 3D) is therefore 2軽 per
Brillouin zone. However, the classification into metals and insulators is more subtle. Let’s
look again at the cases with one and two electrons per primitive cell; if we assume one atom
per primitive cell, these correspond to the cases where the atoms are monovalent and
divalent respectively.

Monovalent
The total area of the Brillouin zone is (2講/欠")態 . If we have one electron per primitive cell, half
st
of the states in the 1 BZ are occupied,
怠 態訂 態
corresponding to an area of 岾 峇 . These
態 銚
occupied states correspond to all the states
within a circle of radius 倦庁 and area 講倦庁態 (the
area of the 2D Fermi surface). Equating these
two areas gives
態 訂 訂
倦庁 = 謬 "岾 峇 = 0.79
訂 銚 銚

Therefore, for a monovalent crystal, the Fermi

surface is completely contained within the first
Brillouin zone (the closest it gets to the BZB is
79% of the way along the 極10玉 directions). Such
a crystal will be a metal.

Divalent
The Fermi surface area is now twice as large,
i.e. 講倦庁態 = (2講/欠)態 , giving

4 講 講
倦庁 = 俵 岾 峇 = 1.12" 岾 峇
講 欠 欠

Along the 極10玉 directions the Fermi surface

nd
now spills over into the 2 Brillouin zone,
whereas along 極11玉 it is still completely
st
contained in the 1 Brillouin zone.

If we now consider the effect of the crystalline potential (i.e. the energy gaps at the BZB) on
the states with wavevectors 圭 parallel to these two directions we can see that if the energy
gaps |撃怠待 | and |撃怠怠 | are small enough, the electrons will start to fill states in the second band
for 圭 along [10] while for 圭 along [11] the band is only partially full. In both directions, nearby
empty states are available for the electrons to be accelerated into and the crystal remains

6-5

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

metallic. Ca, Sr, Ba, for example, are all divalent metals. If the gaps |撃怠待 | and |撃怠怠 | are large
enough so that the bands no longer overlap, the material will be an insulator.

BZB BZB
along along
[11] [10]

So how does the nearly free electron Fermi surface look for a 2D crystal? Since at the BZB
the electron waves form standing waves, their group velocity is zero,
項降 1 項香
懸直 = = = 0 at BZB.
項倦 屋 項倦
i.e., the gradient of energy perpendicular to the BZB tends to zero as we approach the BZB.
擢敵
Since the gradient is perpendicular to the contours of constant energy, the contours of
擢賃
constant energy must be perpendicular to the BZB. Therefore the Fermi surface (which is a
surface of constant energy, 香庁 ) must be perpendicular to the BZB, i.e., contours of the Fermi
surface bend towards the BZB.

Compared to the free electron surface, there are now more filled states in the first Brillouin
zone and fewer in the second. As the crystalline potential increases, the distortion of the
Fermi surface becomes stronger until for large enough 撃, all the electrons become confined
to the first Brillouin zone and the material becomes an insulator. The evolution of the Fermi

6-6

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

surface with increasing crystalline potential for a 2D crystal with divalent atoms is sketched
below.

6-7

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Bloch’s theorem
While trying to understand “how the electrons could sneak by all the ions in a metal so as to
avoid a mean free path of the order of atomic distances”, Felix Bloch realised through a
simple Fourier analysis that the wavefunctions in a periodic potential “differed from a plane
wave of free electrons only by a periodic modulation”; i.e. if 撃(慶) = 撃(慶 + 栗) where 栗 is any
lattice vector, then irrespective of the form of 撃(慶), the solutions of the Schroedinger
equation are of the form

閤圭 (慶) = 憲圭 (慶)結 沈圭糾慶

where 憲圭 (慶) is a function that has the periodicity of the lattice, i.e. 憲圭 (慶) = 憲圭 (慶 + 栗). This is
one statement of BlochÕs theorem. Note, the free electron wavefunction satisfies Bloch’s
theorem with 憲圭 (慶) = 1 (a constant potential can be thought of as a periodic potential with
any periodicity). In general, 閤圭 (慶) is neither strictly periodic nor simply a plane wave. We use
圭 corresponding to the plane wave modulation simply to label the wavefunctions.

Consider a Bloch wave at two positions separated by a lattice vector

閤圭 (慶 + 栗) = 憲圭 (慶 + 栗)結 沈圭糾(慶袋栗) = 憲圭 (慶)結 沈圭糾慶 結 沈圭糾栗 = 閤圭 (慶)結 沈圭糾栗

This gives us an alternative statement of Bloch’s theorem:

閤圭 (慶 + 栗) = 閤圭 (慶)結 沈圭糾栗

That is, for any two equivalent points in the crystal (points separated by 栗), the wavefunction
can only differby a phase factor 結 沈圭糾栗. This second statement of Bloch’s theorem is quite
intuitive if we consider that the electron density |閤|態 = 閤 茅 閤 should have the periodicity of the
crystal, which is indeed the case for a Bloch function:
態
|"閤圭 (慶 + 栗)|態 = 弁閤圭 (慶)結 沈圭糾栗 弁 = |"閤圭 (慶)|態

Simple proof of Bloch’s theorem

For a simple proof of Bloch’s theorem in its second form, consider a 1D crystal as a chain of
軽 atoms with spacing 欠 under periodic boundary conditions such that
閤(捲 + 軽欠) = 閤(捲)
The symmetry of this arrangement implies we should look for solutions such that
閤(捲 + 欠) = 系閤(捲)

6-8

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

where 系 is a (complex) constant. Then, from applying the periodic boundary conditions we
1

find
閤(捲 + 軽欠) = 系 朝 閤(捲) = 閤(捲)
沈態訂
椎
系 朝 = 1 or""""系 = 結 朝

態訂 態訂
Where 喧 is an integer. Since the allowed wavevectors 倦 are 喧 = 喧, we see that 系 = 結 沈賃銚
挑 朝銚
and
閤(捲 + 欠) = 結 沈賃銚 閤(捲)
In 1D, the lattice vectors are just 迎 = 券欠, where 券 is an integer and therefore
閤(捲 + 迎) = 結 沈賃眺 閤(捲)
which is just the second statement of Bloch’s theorem in 1D.

Reciprocal space periodicities and electron bands

When labelling Bloch functions as 閤圭 , we assign the label 圭 based on the wavevector of the
corresponding plane wave modulation 結 沈圭糾慶 in 憲圭 (慶)結 沈圭糾慶. Since 憲圭 (慶) is not a plane wave, the
圭 in 憲圭 is simply a label. However, this decomposition into 憲圭 (慶) and 結 沈圭糾慶 is not unique; we
can see this in two ways:
1. Let 圭 = 圭 嫗 + 屑, where 屑 is a reciprocal lattice vector. Then
閤圭 (慶 + 栗) = 閤圭 (慶)結 沈圭糾栗
嫦
= 閤圭 (慶)結 沈圭 糾栗 結米
沈屑糾栗

怠
嫦
= 閤圭 (慶)結 沈圭 糾栗
i.e. the phase modulation of the wavefunction from one cell to another can be as well
described by 圭 as by any 圭 嫗 related"圭 to by 屑. Thus we could equally have labelled
閤圭 as 閤圭 嫦 .

2. Alternatively, with the same definitions, we can write

嫦 糾慶
閤圭 (慶) = 憲圭 (慶)結 沈圭糾慶 = 婀娶
憲圭娶婬娶 沈屑糾慶 沈圭
(慶)結娶婉 結
通圭 嫦 (慶)
沈屑糾慶
Since both 憲圭 and 結 have the periodicity of the lattice then so does their product,
which we can therefore re-label as 憲圭 嫦 .
Thus our choice of 圭 is arbitrary to within a reciprocal lattice vector 屑. We can therefore
choose between several different representations of the electron wavefunctions and the
corresponding energy bands. For example, consider the nearly free electron dispersion
relation.

1
Formally, this is justified by considering the symmetry of the Hamiltonian. Let’s define a translation
operator 劇栗 for each lattice vector 栗 which shifts its argument by 栗, so that for any function 血(慶),
劇栗 血(慶) = 血(慶 + 栗). Then, since the Hamiltonian has the periodicity of the crystal, 茎(慶 + 栗) = 茎(慶),
劇栗 茎(慶)閤(慶) = 茎(慶 + 栗)閤(慶 + 栗) = 茎(慶)劇栗 閤(慶),
which implies that 劇栗 茎(慶) = 茎(慶)劇栗 , i.e., 茎 and 劇栗 commute and therefore the eigenstates of 茎 can
always be chosen to also be eigenstates of 劇栗 so that 劇栗 閤(慶) = 系栗 閤(慶) where 系栗 is the eigenvalue;
hence 閤(慶 + 栗) = 系栗 閤(慶).

6-9

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

So far we have plotted it using the

Extended zone scheme
extended zone scheme.

Repeated zone scheme

Since any Bloch state 閤圭 can also

be represented as 閤圭袋屑, the energy
香(圭 + 屑) = 香(圭) and we can
choose any reciprocal lattice vector
屑 as the origin of the electron
bands. This leads to the periodic
(or repeated) zone scheme.

Reduced zone scheme

Alternatively, we can define 圭 to be
always within the first Brillouin
zone, as is most commonly done.
This allows the reduced zone
scheme representation. The energy
is a multivalued function of 圭 and
usually given an additional label 券
(the band index), e.g. 香津圭, 閤津圭 to
distinguish the different states.

6-10

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

6-11

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Crystal momentum
A Bloch wave is not generally an eigenstate of momentum. To see this let’s apply the
momentum operator 伐件屋繕 to the Bloch wave
伐件屋繕閤圭 噺 伐件屋繕盤憲圭 (慶)結 沈圭糾慶 匪 噺 屋圭閤圭 伐 件屋結 沈圭糾慶 繕憲圭 (慶)
Therefore, unlike for a plane wave, 屋圭 does not represent a true momentum of the Bloch
electron. This should already be obvious from the fact that for a Boch wave, 圭 is only defined
to within 屑 or by considering 圭 at the BZB, where the standing waves have zero momentum
but a finite 圭. Nevertheless, as in our discussion of phonons, 屋圭—known as the crystal
momentum—plays a similar role to momentum when considering the response of Bloch
electrons to electromagnetic fields.
It can be shown quite generally though that an electron in an energy level 香津圭 has a non-
vanishing mean velocity given by
1 項香津 (倦)
""""""""""懸津 (倦) = in 1D
屋 項倦
1
or"""""懸津 (倦) = 繕圭 香津圭 in 3D
屋
擢摘 怠 擢敵
Note that this is just the group velocity 懸直 = = .
擢賃 屋 擢賃

The above equation tells us that the electron travels through a periodic potential due to the
ion cores without any degradation of its mean velocity! The effect of the periodic lattice is to
modify the slope 項香/項倦 and therefore the value of 懸 but the electrons still propagate freely
without scattering off the perfectly periodic array of ions.

6-12

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Tight-binding model
The nearly free electron model was based on the assumption that the periodic potential only
weakly perturbs the free electron states; only electrons with 圭 close to the BZB are
significantly affected by the interaction with the ions. Implicit in this model is the sharp
distinction between the core electrons (assumed to be fully localised on the ions) and the
(free) valence/conduction electrons. This distinction is quite well justified for “simple” alkali
metals like sodium whose outer 3嫌 electrons are very delocalised, while the inner electrons
are almost entirely atomic-like. The Fermi surface of sodium (which is monovalent and 倦庁 is
not too close to the BZB) is therefore very free-electron-like. In transition metals, on the other
hand, the outer electrons are in 穴-orbitals, which are more tightly localised on the atom.
Therefore the free-electron approximation is inappropriate and instead it is better to
approach the problem from the opposite starting point, that is by looking at a periodic array
of atomic orbitals where the electrons are fairly well localised. This is known as the tight-
binding model.
Consider first what happens when two neutral atoms are brought close together so their
electron wavefunctions begin to overlap (e.g., when two hydrogen atoms with 1嫌 electrons
form an H2 molecule).

Isolated atoms
(far apart)

As their wavefunctions overlap, there are two possible combinations

閤 ± = 閤凋 ± 閤喋
The total electron energy is the sum of kinetic (KE) and potential (PE) energies. For the
symmetric state 閤 袋 , both KE and PE are lower as compared to the energy of the two
isolated atoms since KE is related to the curvature of the wavefunction 弁繕 匝 閤弁 and the
curvature is now lower in the region between the two atoms, while PE is also lower as there
is more negative electron density in the region between the two positive ions. The 閤 袋 state
is therefore called the bonding orbital. Similarly, the antisymmetric state 閤 貸 , known as the
antibonding orbital, has a higher energy than that of the isolated atoms.

6-13

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

(antibonding)
Electron energy

Total energy
(bonding)

Atomic separation, R Atomic separation, R

Total energy = electron energy + nuclear

repulsion ( )

The linear combination of two atomic orbitals from two atoms therefore leads to two
molecular orbitals. Similarly, for 軽 atoms, 軽 atomic orbitals give rise to 軽 new electron
orbitals with different energies. For large 軽, these orbitals form a band. The same is true of
other atomic orbitals (嫌 喧 穴, etc.) that all form bands, which may overlap in energy.

Consider an array of atoms at positions 慶陳 , each described by an individual atomic

wavefunction 剛.

6-14

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

If the wavefunction overlap is not too strong, we can consider the wavefunctions of the
crystal to be a linear combination of the atomic orbitals (LCAO)

閤圭 (慶) = 布 系圭陳 剛"(慶 伐 慶陳 )

陳
怠
To satisfy Bloch’s theorem, an appropriate choice for the coefficients 系圭陳 is 結 沈圭糾慶尿 . Let’s
ヂ朝
check that this choice does indeed satisfy Bloch’s theorem:
1
閤圭 (慶 + 栗) = 結 沈圭糾慶尿 剛(慶 伐 慶陳 + 栗)
ヂ軽
1
= 結 沈圭糾栗 班 布 結 沈圭糾岫慶尿貸栗) 剛盤慶 伐 (慶陳 伐 栗)匪藩
ヂ軽 陳

If 慶陳 is the position of an atom, then so is (慶陳 伐 栗), and since the sum runs over all atoms
(i.e. all 栗), the term in { } is simply 閤圭 (慶). Hence,
閤圭 (慶 + 栗) = 結 沈圭糾栗 閤圭 (慶)
and 閤圭 (慶) is a valid Bloch function.
If 茎 is the Hamiltonian for the system, the energy is given by

香圭 = 豹 閤圭茅 (慶)茎閤圭 (慶)穴撃""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""

1
= 豹 蕃布 結 貸沈圭糾慶尿 剛 茅 (慶 伐 慶陳 )否 茎 蕃布 結 沈圭糾慶韮 剛(慶 伐 慶津 )否 穴撃"
軽
陳 津
1
= 布 布 結 沈圭糾岫慶韮 貸慶尿) 豹 剛 茅 (慶 伐 慶陳 )茎剛(慶 伐 慶津 )穴撃
軽
陳 津

or if we write the last integral, known as the overlap integral, in the more compact Dirac
notation as 極剛陳 |茎|剛津 玉, with 剛陳 岩 剛(慶 伐 慶陳 ) etc.,
1
香圭 = 布 布 結 沈圭糾岫慶韮 貸慶尿) 極剛陳 |茎|剛津 玉
軽
陳 津

Since 剛 is the same on each site and 茎 has the periodicity of the lattice, the overlap integral
depends only on 慶陳 伐 慶津 and not on the individual 慶津 and 慶陳 . Thus if we define 素陳 = " 慶陳 伐 慶津

極剛陳 |茎|剛津 玉 = 豹 剛 茅 (慶 伐 慶陳 )茎剛(慶 伐 慶津 )穴撃""""""""""""""""""""""""""

= 豹 剛 茅 盤慶 伐 (慶陳 伐 慶津 )匪茎剛(慶)穴撃

= 豹 剛 茅 (慶 伐 素陳 )茎剛(慶)穴撃"""""""""""""

For each 素陳 , there are 軽 possible combinations of 慶津 and 慶陳 and therefore

朝

香圭 = 布 結 貸沈圭糾素尿 豹 剛 茅 (慶 伐 素陳 )茎剛(慶)穴撃
陳

6-15

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

We expect the overlap integral to be largest when 券 = 兼 (i.e. when 慶津 and 慶陳 refer to the
same site) and progressively smaller as |慶陳 伐 慶津 | becomes larger. We will therefore restrict
ourselves to nearest neighbour (n.n.) overlap only, i.e. let
伐糠 塙 ど for"券 = 兼"(i. e. same site)
極剛陳 |茎|剛津 玉 = 班 伐紘 塙 ど when"券"and"兼"refer to n. n.
0 otherwise

The energy then becomes

津.津.

香圭 噺 伐糠 伐 紘 布 結 貸沈圭糾素尿
陳

Examples
1. For a 1D chain of atoms with spacing 欠, the two nearest neighbours are at 貢陳 = ±欠.
Therefore
香賃 噺 伐糠 伐 紘盤結 貸沈賃銚 + 結 沈賃銚 匪
噺 伐糠 伐 に紘 cos 倦欠

bandwidth

The allowed states form a band of width 4紘; the bandwidth increases with increasing
wavefunction overlap.
For small 紘, electrons are more localised (more atomic-like).
For large 紘, electrons are more delocalised (more “free”).
2. For a 2D square lattice, nearest neighbours are at (欠, 0), (伐欠, 0), (0, 欠) and (ど 伐欠).
Therefore
香賃 噺 伐糠 伐 紘盤結 貸沈賃猫 銚 + 結 沈賃猫 銚 + 結 貸沈賃熱 銚 + 結 沈賃熱 銚 匪
噺 伐糠 伐 に紘盤cos 倦掴 欠 + cos 倦槻 欠匪

6-16

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Energy dispersion and constant energy contours for a 2D square lattice

tight-binding band with 継暫 = 伐糠 伐 紘盤cos 倦掴 欠 + cos 倦槻 欠匪.

3. Similarly, in 3D one obtains 噺 伐糠 伐 に紘盤̊æœ 倦掴 欠 髪 ̊æœ 倦槻 欠 髪 ̊æœ 倦佃 欠匪

Constant energy surface 継 噺 伐糠 for a 3D simple cubic lattice with the

above dispersion relation. The filled volume corresponds to one electron
per primitive cell. Image reproduced from Kittel’s book.

6-17

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

6-18

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Physics'3225'2017'.'Part'II'
Prof.'S.'T.'Bramwell''
!Magne&sm)
!Superconduc&vity))
!Semiconduc&vity)
Aim:)to)understand)the)basic)physics)of)magne&sm,)
superconduc&vity)and)semiconduc&vity)in)solids.))

! )Magne&sm)and)superconduc&vity)are)mainly)a)break)from)part)I,)whereas)
semiconduc&vity)leads)on)naturally)from)the)theory)from)band)solids)given)in)part)I.)

!'3'homeworks,'2'marked,'first'given'out'this'Thursday.))
Downloaded by Roy Vesey (rdv0044@googlemail.com)
 lOMoARcPSD|3449400

What'do'we'mean'by'magnets,'
superconductors,'semiconductors?'
¥  The)magne&c)proper&es)of)materials)describe)how)they)
respond)to,)and)create)magne&c)fields.))
¥  Superconduc&vity)is))a)specific)low)temperature)property)of)
some)materials)where)they)lose)all)electrical)resistance)–)we)
will)give)a)proper)defini&on)later.))
¥  Semiconduc&vity)is)the)property)of)thermally)excited)
electrical)conduc&on)across)a)band)gap.))))

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Why'study'magnets,'superconductors,'
semiconductors?''

¥  Technological'importance'

¥  Laboratory'for'key'concepts'in'physics'
Phase)transi&ons,)symmetry)
breaking,)macroscopic)quantum)
coherence,)emergent)excita&ons...))

")Historic)consequences)for)
cosmology,)par&cle)physics)etc.))

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Main'Textbook'for'part'II'
¥  Solid)State)Physics)(Manchester)Physics)
Series))J.)R.)Hook,)H.)E.)Hall)

However)ANY)standard)solid)state)physics)
textbook)contains)essen&ally)the)same)
material.))

Other)classic)textbooks)for)this)level)include)))

)!KiPel,)Solid)State)Physics)
)!AshcroQ)&)Mermin,)Solid)State)Physics)

WebTbased)material)can)occasionally)be)useful)
but)be)cau&ous,)as)it)is)oQen)simply)wrong.)))

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Revision:'the'nature'of'maOer'
MaOer)has)two)essen&al)proper&es)

)1.)Microscopically)it)is)parPculate)(consists)of)atoms)etc.)))
)2.)Macroscopically)it)has)a)phase'structure'(solid,)liquid,)gas)etc.))
Solids)may)be)broadly)classified)into)the'four'solid'types'

aluminium,)copper,)steel) 1)Metals)
silicon,)graphite,)gallium)arsenide)) 2)Band)insulators)(“giant)covalent”))
ice,)iodine,)organic)solids) 3)Molecular)
NaCl,)NiO,)Fe3O4,)K2FeF4,)GdCl3) 4)Ionic)

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Where'do'we'find'magnePsm,'super.'and'semi.conducPvity?'

!)All)materials)are)magnePc)in)some)way;)Ionic)and)metallic)compounds)of)dT)
and)fTblock)elements)tend)to)be)the)most)strongly)magne&c.))

dTblock)

fTblock)

!SuperconducPvity)is)a)common)lowTtemperature)property)of)metals.))))

!'SemiconducPvity)is)a)universal)property)of)band)insulators)with)a)rela&vely)
small)band)gap.))

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Materials'versus'Models'
!)Some)of)the)most)interes&ng)substances)can)only)be)classified)as)being)on)
the)boundaries)between)the)tradi&onal)solid)types.))

!)Examples:)YBa2Cu3O7)(ionic,)metallic)and)superconduc&ng),)I2)(molecular)
solid,)band)semiconductor),)graphite)(semimetal)–)band)insulator)with)
vanishing)band)gap)))

!)The)approach)of)physics)is)to)sidestep)all)this)chemical)complexity)and)
consider)idealised)models:)ideal)ionic)solid,)ideal)band)solid,)ideal)molecular)
solid,)free)electron)metal)and)so)on.))

!)By)making)these)models)gradually)more)realis&c)we)can)understand)general)
behaviour)and)eventually)capture)the)wealth)of)material)behaviour.))))

!)The)principle)object)of)a)physicist’s)study)is)usually)the)idealised)model,)
rather)than)the)real)material.)I’ll)note)several)idealisa&ons)as)we)go)along.))

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Brief'introducPon'to'the'course'✓'

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

MagnePsm'

Downloaded by Roy Vesey (rdv0044@googlemail.com)

 lOMoARcPSD|3449400

Lecture'1'plan'

Introduc&on)to)the)course)✓))
Introduc&on))
)–)magne&c)moment,)sketch)of)magne&sm)course))(ppt))
Paramagne&sm)and)diamagne&sm)(on)the)board))
Paramagne&c)moment)of)isolated)ions)(on)board)+)ppt)))

Downloaded by Roy Vesey (rdv0044@googlemail.com)