2.

4 Classical lattice with two identical atoms

Patterson chpt 2.2
Ibach chpt 4.2

We continue to look into a 1D lattice. We will study the dynamics (i.e. the motion) of atoms in the
lattice. First, we will look into two atoms. We have:
• One dimensional 2 atomic chain
• Bonds are approximated by springs (Chpt 2.3)
• The displacement 𝑢𝑖 and position 𝑥 is measured in x-direction (Chpt 2.1)
• We will employ the harmonic approximation (Chpt 2.3)
• And a nearest neighbor approximation (Chpt 1.4)
2.4 Classical lattice with two identical atoms

We consider a force acting on atom A1 along bond direction

i) When atom A1 moves on

𝐹1 = 2𝛾 𝑥2 − 𝑥1 = 2𝛾 𝑥2 − 2𝛾𝑥1 = 𝛾𝑥2 − 𝛾𝑥1 − 𝛾𝑥1 + 𝛾𝑥2

2.4 Classical lattice with two identical atoms
We take a moment to identify all terms in the given expression

𝐹1 = 2𝛾 𝑥2 − 𝑥1 = 2𝛾 𝑥2 − 2𝛾𝑥1 = 𝛾𝑥2 − 𝛾𝑥1 − 𝛾𝑥1 + 𝛾𝑥2

Keep in mind that we only consider nearest neighbour interactions here. This means we want to
identify a series of springs (Chpt 2.1).
2.4 Classical lattice with two identical atoms
𝐹1 = 2𝛾 𝑥2 − 𝑥1 = 2𝛾 𝑥2 − 2𝛾𝑥1 = 𝛾𝑥2 − 𝛾𝑥1 − 𝛾𝑥1 + 𝛾𝑥2

𝐹1 = 2𝛾 𝑥2 − 𝑥1 = 2𝛾 𝑥2 − 2𝛾𝑥1 = 𝛾𝑥2 − 𝛾𝑥1 − 𝛾𝑥1 + 𝛾𝑥2

2.4 Classical lattice with two identical atoms
𝐹1 = 2𝛾 𝑥2 − 𝑥1 = 2𝛾 𝑥2 − 2𝛾𝑥1 = 𝛾𝑥2 − 𝛾𝑥1 − 𝛾𝑥1 + 𝛾𝑥2

𝐹1 = 2𝛾 𝑥2 − 𝑥1 = 2𝛾 𝑥2 − 2𝛾𝑥1 = 𝛾𝑥2 − 𝛾𝑥1 − 𝛾𝑥1 + 𝛾𝑥2

2.4 Classical lattice with two identical atoms
Similarly, we can formulate the force on atom A2

𝐹2 = 2𝛾𝑥1 − 2𝛾𝑥2
𝛾𝑥1 − 𝛾𝑥2 − 𝛾𝑥2 + 𝛾𝑥1
= 𝛾 𝑥1 − 𝑥2 − 𝛾 𝑥2 − 𝑥1

We use the complex ansatz (chpt 2.2)

𝑥𝑛 = 𝑥𝑛 (𝑡) = 𝑢𝑛 exp(𝑖𝜔𝑡)
where 𝑢𝑛 is the displacement vector (amplitude) around the equilibrium position 𝑥𝑛,0 of atom 𝑛.

𝑢𝑛 = 𝑥𝑛 − 𝑥𝑛,0
2.4 Classical lattice with two identical atoms

And rewrite the force equilibrium according to Newton’s law actio = reactio

𝐹𝑎 = 𝐹𝑟

𝑚𝑥ሷ 𝑛 = 𝛾 𝑥𝑚 − 𝑥𝑛 − 𝛾(𝑥𝑛 − 𝑥𝑚 )
𝑚𝑥ሷ 𝑚 = 𝛾 𝑥𝑛 − 𝑥𝑚 − 𝛾 𝑥𝑚 − 𝑥𝑛

= −𝛾 𝑥𝑚 − 𝑥𝑛 + 𝛾 𝑥𝑛 − 𝑥𝑚

Keep in mind that the position 𝑥𝑖 is a function of time! We insert now

𝑥𝑛 = 𝑥𝑛 (𝑡) = 𝑢𝑛 exp(𝑖𝜔𝑡)
And
𝑑2
𝑥ሷ 𝑛 = 2 𝑥𝑛 (𝑡) = 𝑢𝑛 𝜔 exp(𝑖𝜔𝑡)
𝑑𝑡
2.4 Classical lattice with two identical atoms
Applying the 2 atomic model to the equations, we can choose
𝑛=1
𝑚=2

which leads to the system of equations of motion for the one-dimensional atomic chain with two identical
atoms

−𝜔2 𝑚 𝑢1 = 𝛾 𝑢2 − 𝑢1 − 𝛾 𝑢1 − 𝑢2
−𝜔2 𝑚 𝑢2 = 𝛾 𝑢1 − 𝑢2 − 𝛾 𝑢2 − 𝑢1

Reordering gives
0 = 𝛾 𝑢2 − 𝑢1 − 𝛾 𝑢1 − 𝑢2 + 𝜔2 𝑚 𝑢1
0 = 𝛾 𝑢1 − 𝑢2 − 𝛾 𝑢2 − 𝑢1 + 𝜔2 𝑚 𝑢2
and in matrix form
2𝛾 − 𝜔2 𝑚 −2𝛾 𝑢1
⋅ 𝑢 =0
−2𝛾 2𝛾 − 𝜔2 𝑚 2
2.4 Classical lattice with two identical atoms
Because
2𝛾 − 𝜔2 𝑚 −2𝛾 𝑢1
⋅ 𝑢2 = 0
−2𝛾 2𝛾 − 𝜔2 𝑚
Takes the form
𝑀 ⋅ 𝑢ത = 0
we can find non-trivial solutions (i.e. those that are not 𝑢ത = 0) with

det 𝑀 = 0

Hence,
2𝛾 − 𝜔2 𝑚 −2𝛾
det 𝑀 = det
−2𝛾 2𝛾 − 𝜔2 𝑚

We rewrite this in form of an eigenvalue problem.

= det 𝐷 − 𝟙 𝜔2 𝑚
2.4 Classical lattice with two identical atoms
Reminder Eigenvalue Problem
2.4 Classical lattice with two identical atoms

= det 𝐷 − 𝟙 𝜔2 𝑚 • These are characteristic oscillation frequencies.

0 = 2𝛾 − 𝜔2 𝑚 2 − −2𝛾 −2𝛾
= 2𝛾 − 𝜔2 𝑚 2 − 4𝛾 2
which is a quadratic equation
• They are determined by the spring constant and the
mass of the atoms.
• Larger mass will slow down the oscillation and vice versa
• A smaller spring constant will slow down the oscillation and
vice versa

4𝛾 2 = 2𝛾 − 𝜔 2 𝑚 2
±2𝛾 = 2𝛾 − 𝜔2 𝑚 • Because we formulated an eigenvalue problem, the
frequencies are eigen values of the system, the so-
which has two solutions called eigen-frequencies.
1)
0 = 𝜔2 𝑚
𝜔1 = 0
2)
4𝛾
𝜔22 =
𝑚
2.4 Classical lattice with two identical atoms
Next, we put the solution 𝜔1 back into (I)

−𝜔2 𝑚 𝑢1 = 𝛾 𝑢2 − 𝑢1 − 𝛾 𝑢1 − 𝑢2
−𝜔2 𝑚 𝑢2 = 𝛾 𝑢1 − 𝑢2 − 𝛾 𝑢2 − 𝑢1

−02 𝑚 𝑢1 = 𝛾 𝑢2 − 𝑢1 − 𝛾 𝑢1 − 𝑢2
−02 𝑚 𝑢2 = 𝛾 𝑢1 − 𝑢2 − 𝛾 𝑢2 − 𝑢1

0 = 𝛾 𝑢2 − 𝑢1 − 𝛾 𝑢1 − 𝑢2
0 = 𝛾 𝑢1 − 𝑢2 − 𝛾 𝑢2 − 𝑢1

𝛾 𝑢1 − 𝑢2 = 𝛾 𝑢2 − 𝑢1
𝛾 𝑢2 − 𝑢1 = 𝛾 𝑢1 − 𝑢2
where we can read the solution
𝑢1 = 𝑢2
1
𝑢ത =
1
2.4 Classical lattice with two identical atoms
1
The result 𝑢ത = indicates that there is no relative motion between atom A1 and A2, because the
1
displacement 𝑢1 and 𝑢2 are equal. Also remember that this is the trivial solution 𝜔1 = 0.
2.4 Classical lattice with two identical atoms

4
For the other case, we put 𝜔22 = 𝛾 into (I) Sorting for 𝑢𝑛 and 𝑢𝑚
𝑚

−𝜔2 𝑚 𝑢1 = 𝛾 𝑢2 − 𝑢1 − 𝛾 𝑢1 − 𝑢2 2𝑢𝑚 𝛾 = 2𝑢𝑛 𝛾 − 4𝑢𝑛 𝛾

1
−𝜔2 𝑚 𝑢2 = 𝛾 𝑢1 − 𝑢2 − 𝛾 𝑢2 − 𝑢1 𝑢𝑚 𝛾 = 𝛾 2𝑢𝑛 − 4𝑢𝑛
2
𝑢𝑚 = −𝑢𝑛
4
− 𝛾 𝑚 𝑢1 = 𝛾 𝑢2 − 𝑢1 − 𝛾 𝑢1 − 𝑢2
𝑚 1
𝑢ത =
−1
4
− 𝛾 𝑚 𝑢2 = 𝛾 𝑢1 − 𝑢2 − 𝛾 𝑢2 − 𝑢1
𝑚
In this solution, the atoms oscillate in opposite directions.
They either move towards each other from their
Changing to a general notation respective equilibrium position, or they move away from
each other.
4𝛾
𝑚 𝑢𝑛 = 𝛾 𝑢𝑚 − 𝑢𝑛 − 𝛾(𝑢𝑛 − 𝑢𝑚 )
𝑚
4𝛾𝑢𝑛 = −2𝑢𝑛 𝛾 + 2 𝑢𝑚 𝛾
2.4 Classical lattice with two identical atoms
Normal mode solutions

Both atoms vibrate with the same frequency

Note
• in 𝑛-dimensions with 𝑚 atoms, there are 𝑛 ⋅ 𝑚 independent solutions to the equations of motions
• Here we had 2 atom in one dimension, so 𝑛 = 1 and 𝑚 = 2 which gives two solutions, 𝜔1 and 𝜔2
• In three dimensions, we would have 6 solutions, and a FCC structure has 12 solutions, etc.
• Each eigenfrequency, i.e. normal mode describes an independent vibrations with a single frequency
• In a normal mode, all atoms vibrate with the same frequency
• Any vibration in a crystal is a superposition of normal modes
2.4.1 Extension to periodic systems
The system seen so far used only two atoms and the harmonic approximation. We now continue to
include

- Anharmonic terms (chpt 2.3)

- Interactions beyond nearest neighbors (chpt 1.4)
- Periodic boundary conditions (chpt 1.3)

This increases the physical accuracy of our model. For the latter boundary condition, we even have a
simple physical explanation. Because we assume perfect crystals, we can safely assume that effect on
the outside of the crystal are not very important to understand the lattice dynamics within the crystal.
This is of course different for interfaces.

We want to consider a crystal with periodic boundary conditions and N atoms.

2.4.1 Extension to periodic systems
Note that the periodic boundary conditions here are cyclic or Born van Karman boundary conditions.
We require a large number of atom 𝑁. The atoms are lined up in one dimension. They are space by a
distance a. Because we work with translational symmetry, the separation of lattice points in
2𝜋𝑚
momentum space is given by Δ𝑘 = 𝑘 − 𝑘 ′ = 𝑞 = .
𝑎
Regarding the boundary conditions, we apply 𝑢0 = 𝑢𝑁 and we can depict the system as follows
2.4.1 Extension to periodic systems
We start to consider a potential of an atom inside a solid. From Chpt 2.3, we re-call the spring model
of atomic bonds

Now we have a solid in 3D with N-nearest neighbors. In addition, we want to go beyond nearest
neighbor interactions. As we saw in Chpt. 1.6, bonds in crystals can have very different nature. Even
in homo-nuclear materials, i.e. there is only one kind of element, there several kinds of bond type.
For example, in graphene are covalent bonds and van der Waals bonds.

To work out the lattice dynamics, we want to introduce a general potential 𝑉(𝑥)ҧ that expresses the
bonds in the crystal.
2.4.1 Extension to periodic systems
Note that in many cases, the formation of real bonds, i.e. the redistribution and sharing of electronic density, goes
beyond nearest neighbor interactions. In metallic bonds (chpt 1.6), even the classical model states that valence
electrons and lose can propagate (i.e. travel) through the entire solid.

Therefore, we need to expand our model from and interaction potential of

𝑉 𝑥1 , 𝑥2

Following the argument above, the new potential for the one-dimensional case with N atoms in the solid is

𝑉(𝑥1 , 𝑥2 , … , 𝑥𝑁 )
In three dimensions, we would have

𝑉(𝑥1 , 𝑥2 , … , 𝑥𝑁 , 𝑦1 , 𝑦2 , … , 𝑦𝑁 , 𝑧1 , 𝑧2 , … , 𝑧𝑁 )
= 𝑉(𝑟1ҧ . 𝑟2ҧ , … , 𝑟𝑛ҧ )

which is an increasingly difficult problem.

2.4.1 Extension to periodic systems
Working in one dimension we start with

𝑉(𝑥1 , 𝑥2 , … , 𝑥𝑁 )
Using the multivariant Taylor expansion

𝑉 𝑥1 , 𝑥2 , … , 𝑥𝑁
1 𝜕2
= 𝑉 0, … , 0 + ෍ 𝑉 อ 𝑥𝑛 − 0 𝑥𝑛′ − 0
2 𝜕𝑥𝑛 𝜕𝑥𝑛′
𝑛,𝑛′ ҧ
𝑥=0
1 𝜕3
+ ෍ 𝑉 อ 𝑥𝑛 𝑥𝑛′ 𝑥𝑛"
6 ′ 𝜕𝑥𝑛 𝜕𝑥𝑛′ 𝜕𝑥𝑛"
𝑛,𝑛 ,𝑛" ҧ
𝑥=0
3
1 𝜕
+ ෍ 𝑉 อ 𝑥𝑛 𝑥𝑛′ 𝑥𝑛" 𝑥𝑛′′′ + 𝒪(𝑥 5 )
24 𝜕𝑥𝑛 𝜕𝑥𝑛′ 𝜕𝑥𝑛" 𝜕𝑥𝑛′′′
𝑛,𝑛′ ,𝑛",𝑛′′′ ҧ
𝑥=0
2.4.1 Extension to periodic systems
This is a very complicated problem. The solution of the dynamic problem of 𝑉 𝑥1 , 𝑥2 , … , 𝑥𝑁 can be solved within the
harmonic approximation. However, this means that effect caused by anharmonic contributions will not by obtainable
from any solution that is based on the harmonic approximation. Anharmonic terms introduce interaction between
normal modes. Example of effect that are linked to the anharmonic terms are thermal expansion, specific hear
‘correction’ of insulators at high temperature, and thermal resistivity of insulators at high temperature.

In the harmonic approximation 𝑉 𝑥1 , 𝑥2 , … , 𝑥𝑁 simplifies to

1 𝜕2
𝑉 𝑥1 , 𝑥2 , … , 𝑥𝑁 ≅ 𝑉 0, … , 0 + ෍ 𝑉 อ 𝑥𝑛 − 0 𝑥𝑛′ − 0
2 𝜕𝑥𝑛 𝜕𝑥𝑛′
𝑛,𝑛′ ҧ
𝑥=0

We introduce the short-hand notation

𝜕2
𝑉อ = 𝑉𝑛,𝑛′
𝜕𝑥𝑛 𝜕𝑥𝑛′
ҧ
𝑥=0
Due to symmetry, we know that
𝑉𝑛,𝑛′ = 𝑉𝑛′,𝑛
2.4.1 Extension to periodic systems
We use Newtons 2nd law 𝐹 = 𝑚𝑎 and

𝐹𝐴 = 𝐹𝑅
in addition, we use the result of chpt 2.1 and formulate the force equilibrium as

𝐹 = 𝑚 𝑥𝑛ሷ = − ෍ 𝑉𝑛,𝑛′ 𝑥𝑛′

𝑛

When we only consider atom 𝑘, the force is expressed as

𝐹 = 𝑚 𝑥𝑛ሷ = −𝑉𝑘,𝑛 𝑥𝑘
And with periodic boundary conditions

𝑉𝑛,𝑘 = 𝑉(𝑛 − 𝑘)

which states that the potential shall depend on the distance between 𝑛 and 𝑘.
2.4.1 Extension to periodic systems
We employ the normal mode solution with eigenmodes 𝜔 = 𝜔𝑖

𝑥𝑛 = 𝑢𝑛 𝑒 𝑖𝜔𝑡
with
𝜕𝑢𝑛
=0
𝜕𝑥𝑡

𝑚𝜕 2 2 𝑖𝜔𝑡
𝐹 = 𝑚𝑎 ↔ 𝑥𝑛 = 𝑚𝜔 𝑢 𝑛 𝑒
𝜕𝑡 2

𝐹 = 𝑉𝑛,𝑘 𝑥 ↔ 𝑉𝑛,𝑘 𝑥𝑘 = 𝑉𝑛,𝑘 𝑢𝑛 𝑒 𝑖𝜔t

Collecting the terms, we see that

0 = 𝑚𝜔2 𝑢𝑛 𝑒 𝑖𝜔𝑡 − 𝑉𝑛,𝑘 𝑢𝑛 𝑒 𝑖𝜔t

2.4.1 Extension to periodic systems
We want to have a short-hand notation

0 = 𝑚𝜔2 𝑢𝑛 𝑒 𝑖𝜔𝑡 − 𝑉𝑛,𝑘 𝑢𝑛 𝑒 𝑖𝜔t

0 = 𝑚𝜔2 − 𝑉𝑛,𝑘 𝑢𝑛 𝑒 𝑖𝜔𝑡

so we introduce an operator 𝑃, that when applied to 𝑢𝑛 constructs the above equation. We define

𝑃 𝑢𝑛 = 𝑚𝜔2 𝑒 𝑖𝜔𝑡 − ෍ 𝑉𝑛,𝑛′ 𝑢𝑛 = 0

𝑛′

This includes the summation of interaction terms for all n’ atoms.

2.4.1 Extension to periodic systems
Furthermore, we use translational symmetry

𝑚𝜔2 𝑢𝑛+1 − ෍ 𝑉 𝑛′ − 𝑛 + 1 𝑢𝑛′ = 0

𝑛′
For 𝑛 → 𝑛 + 1

𝑚𝜔2 𝑢𝑛+1 − ෍ 𝑉 𝑛′ − 𝑛 − 1 𝑢𝑛′ = 0

𝑛′
And for 𝑛′ → 𝑛′ + 1
𝑚𝜔2 𝑢𝑛+1 − ෍ 𝑉 𝑛′ − 𝑛 𝑢𝑛′ +1 = 0
𝑛′
2.4.1 Extension to periodic systems
Check some ‘extrema’
i)
𝑝 ⋅ 𝑢𝑛 = 0 ↔ 𝑝 ⋅ 𝑢𝑛+1 = 0
ii) For any function 𝑓
𝑝 ⋅ 𝑓 = 0 ↔ 𝑢𝑛+1 = 𝑒 𝑖𝑞𝑎 𝑢𝑛

We note in
𝑢𝑛+1 = 𝑒 𝑖𝑞𝑎 𝑢𝑛
the connection of neighbouring atoms, i.e. 𝑛 and 𝑛 + 1, is given through an oscillation 𝑒 𝑖𝑞𝑎 = cos 𝑞𝑎 + 𝑖 sin(𝑞𝑎)
where 𝑎 relates to the distance between atoms. From previous chapters, we know that 𝑞 must relate to the momentum.
We will later recognize 𝑞 as momentum transfer vector 𝑞 = 𝑘 − 𝑘′.

A relation similar to 𝑢𝑛+1 = 𝑒 𝑖𝑞𝑎 𝑢𝑛 is known as Bloch theorem. The Bloch theorem connects translation symmetry to
the modulation of a wave function within a lattice. The Bloch theorem is given as

ത
Ψ 𝑟ҧ = 𝑒 𝑖𝑘⋅𝑟ҧ 𝑢(𝑟)ҧ
Which will be derived in the third chapter of this lecture.
2.4.1 Extension to periodic systems
The expression
𝑢𝑛+1 = 𝑒 𝑖𝑞𝑎 𝑢𝑛
Has 𝑁 linearly independent solutions because we consider N atoms in one dimension.

For the case 𝑛 = 0

𝑢𝑛 = 𝑒 𝑖𝑞𝑛 𝑎 𝑢0
𝑢1 = 𝑒 𝑖𝑞𝑎 𝑢0
And due to periodic boundaries with N atoms

𝑢𝑛 = 𝑢0
Which can be seen in
𝑢𝑛 = 𝑒 𝑖𝑞𝑁𝑎 𝑢0 = 𝑢0
𝑒 𝑖𝑞𝑁𝑎 = 1

𝑞𝑁𝑎 = 2𝜋𝑚 with 𝑚 ∈ ℤ

2.4.1 Extension to periodic systems
We limit 𝑚 ≤ 𝑁 (note 𝑁 = 2𝑁’) and consider the Born-von-
Karman boundary conditions again.

Then, he interesting range of m is

𝑁 𝑁
− +1≤𝑚≤
2 2

2𝜋𝑚
Or in momentum space with 𝑞 =
𝑎

𝜋 𝜋
− <𝑞≤
𝑎 𝑎

Which defines the first Brillouin zone in 1D in reciprocal space.

Note that the definition of the Brillouin zone differs from the
construct using the Voronoi construct by using momentum 𝑞 here
and not lattice points.
2.4.1 Extension to periodic systems

Working with q dependence ∞ ∞

𝑚𝜔𝑞2 ′
= ෍𝑉 𝑛 − 𝑛 𝑒 𝑖𝑞𝑛(𝑛′ −𝑛) 1
= ෍ 𝑉 𝑙 𝑒 𝑖𝑞𝑎𝑙 + ෍ 𝑉 𝑙 𝑒 −𝑖𝑞𝑎𝑙
𝑚
𝑙=0 𝑙=0
We introduce the shorthand notation 𝑙 = 𝑛′ − 𝑛 and increasing
the systems size to the one of a perfect crystal ∞
1
𝑚𝜔𝑞2 = ෍ 𝑉 𝑙 𝑒 𝑖𝑞𝑛𝑙 ෍ 𝑉 𝑙 𝑒 −𝑖𝑞𝑎𝑙 + 𝑒 𝑖𝑞𝑎𝑙
𝑚
∞ 𝑙=0
1 1
𝜔𝑞2 = ෍ 𝑉 𝑙 𝑒 𝑖𝑞𝑎𝑙 We identify the identify cos 𝑥 = 𝑒 𝑖𝑥 + 𝑒 −𝑖𝑥 and arrive at
2
𝑚
𝑙=−∞
∞
2
∞ −∞ 𝜔𝑞2 = ෍ 𝑉 𝑙 cos 𝑞𝑙𝑎
1 𝑚
= ෍ 𝑉 𝑙 𝑒 𝑖𝑞𝑎𝑙 + ෍ 𝑉 𝑙 𝑒 𝑖𝑞𝑎𝑙 𝑙=0
𝑚
𝑙=0 𝑙=0
∞
1
∞ ∞ 𝜔𝑞2 = ෍ 𝑉 𝑙 cos 𝑞𝑙𝑎
1 𝑚
= ෍ 𝑉 𝑙 𝑒 𝑖𝑞𝑎𝑙 + ෍ 𝑉 −𝑙 𝑒 𝑖𝑞𝑎(−𝑙) 𝑙=−∞
𝑚
𝑙=0 𝑙=0

Now, we find a q dependence on the solution of 𝝎𝟐 .

2.4.1 Extension to periodic systems
There are a few things that can be noted for our solution of the eigenmodes for a periodic one-dimensional system of
atoms

In ∞
2
1
𝜔𝑞 = ෍ 𝑉 𝑙 cos 𝑞𝑙𝑎
𝑚
𝑙=−∞

we see an even function that reflects the inversion symmetry of the system. It follows

𝜔2 𝑞 = 𝜔2 −𝑞

At zero momentum transfer q=0, the derivative vanishes

′
𝜔2 𝑞 ቚ =0
𝑞=0
This is very important when discussion dispersion of waves within a system.
2.4.1 Extension to periodic systems
For 𝑞 ≪ 1

𝜔2 𝑞 ≅ 𝑐 𝑞 2
and
𝜔 𝑞 ≈ 𝑐′𝑞
where 𝑐 and 𝑐′ are constants. This allows to identify q as wave vector for the long wavelength limit
with
𝑞 = 2𝜋/𝜆

Note: there is no unique q to describe a vibration because we work with discrete points, but the wave
is continuous
2.4.1 Extension to periodic systems
For only nearest neighbor interactions
𝑉 𝑙 = 0 𝑖𝑓 |𝑙| > 1
Which means we set the potential between atoms beyond nearest neighbour interactions to 0. We
only consider a displacement of one atom in the potential of another atom

Based on symmetry, we also know that

𝑉 +𝑙 = 𝑉 −𝑙
2.4.1 Extension to periodic systems
And for a simultaneous displacement of all atoms (a constant shift of the crystal)
0 = ෍ 𝑉𝑛,𝑛′
𝑛′
Now we can characterize the dependence of V on l in more detail .It follows for the nearest neighbours

𝑉 +𝑙 = 𝑉 0 = 𝑉 −𝑙 = 0

−𝑉 0 = 2 𝑉 𝑙 = 2 𝑉 −𝑙
1
𝑉 𝑙 = 𝑉 −𝑙 = − 𝑉 0
2

Then, we continue working with the eigenmodes solution of a system of a one-dimensional atomic chain
∞
1
𝜔𝑞2 = ෍ 𝑉 𝑙 cos 𝑞𝑙𝑎
𝑚
𝑙=−∞

Where we now use what we learned about V(l) in the nearest neighbor approximation.
2.4.1 Extension to periodic systems
It follows that ∞
1
𝜔𝑞2 = ෍ 𝑉 𝑙 cos 𝑞𝑙𝑎
𝑚
𝑙=−∞

Note that terms with 𝑙 > 1 drop out, then the terms left are

1
= 𝑉(0) cos 0 + 𝑉 −1 cos −𝑞𝑎 + 𝑉 1 cos 𝑞𝑎
𝑚
1 1 1
= 𝑉 0 − 𝑉 0 cos 𝑞𝑎 − 𝑉 0 cos 𝑞𝑎
𝑚 2 2
1
= 𝑉 0 1 − cos 𝑞𝑎
𝑚
Using the half-angle formula
2
𝜃 1 − cos 𝜃
sin =
2 2
2.4.1 Extension to periodic systems
We arrive at
2𝑉 0 𝑞𝑎
𝜔𝑞 = sin
𝑚 2
2.4.1 Extension to periodic systems
We can now give the equation of motion of atom n in the mode m as

𝑎
2𝜋𝑚 2𝜋𝑚
(𝑚) 𝑖𝑛𝑎( 𝑁𝑎 ) 𝑖𝑡 𝜔𝑐 sin 2 𝑁𝑎
𝑥𝑛 = 𝑢0 𝑒 𝑒

𝑞𝑎
𝑖𝑛𝑎𝑞 𝑖𝑡 𝜔𝑐 sin 2
= 𝑢0 𝑒 𝑒
where we used

𝑥𝑛 = 𝑢𝑛 𝑒 𝑖𝜔𝑡
𝑞𝑁𝑎 = 2𝜋𝑚
𝑢𝑛 = 𝑢0 𝑒 𝑖𝑞𝑛𝑎

2𝑉 0 𝑞𝑎
𝜔𝑞 = sin
𝑚 2
2.4.1 Extension to periodic systems
Comprehension questions
• How can we justify Born-von-Karman boundary conditions and when are they unjustified?
• Why do we not consider lattice dynamics of one atom here?
• Which model potential is used to construct 𝐹1 = 2𝛾 𝑥2 − 𝑥1 in the one-dimensional atomic chain
model with two identical atoms? When is this assumption justified?
• Why is the linear term missing in the Taylor expansion of 𝑉 𝑥1 , 𝑥2 , … , 𝑥𝑁 ?
• Under which assumption(s) can the Taylor expansion of 𝑉 𝑥1 , 𝑥2 , … , 𝑥𝑁 lead to the equation of
motions in Chpt 2.1, for instance 𝐹1 = 2𝛾 𝑥2 − 𝑥1 ?
• Can the exact arrangement of atoms in a perfect lattice be described in terms of vibrations?
• What is the reason leading to the q-dependence of 𝜔2 = 𝜔2 𝑞 ?
• What is the assumption causing the q-dependence of 𝜔2 = 𝜔2 (𝑞)?
2.4.1 Extension to periodic systems
Exercises

1. Formulate the system of equation of motion for the one-dimensional atomic chain with three
identical atoms.
2. Formulate the system of equation of motion for the one-dimensional atomic chain with two
identical atoms but anisotropic bonds. This means (looking from left to right) the spring constant
between A2 and A1 is 𝛾21 and between A1 and A2 is 𝛾12 with 𝛾21 ≠ 𝛾12 .
3. Verify the Taylor expansion of 𝑉 𝑥1 , 𝑥2 , … , 𝑥𝑁 given in 2.4.1.
4. Proof 𝜔2 𝑞 ≅ 𝑐 𝑞2 for 𝑞 ≪ 1
5. Proof 𝜔 𝑞 ≅ 𝑐 ′ 𝑞 for 𝑞 ≪ 1