Tutorial 4: Crystalline Structures (Lattices)

Contents
1 Introduction and Definitions 1
1.1 Lattice Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Unit Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Face Centered Cubic Lattice 3

3 Packing Ratio 4

4 Symmetries 5

5 Interatomic Forces 6
5.1 Valence Electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5.2 Noble Gas Lattice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1 Introduction and Definitions

In this tutorial we begin our discussion on lattices. We forget about the electrons and
consider only the atoms (or ions) that make up the material.
We first formulate a few definitions and illustrate them on two simple examples.

1.1 Lattice Vectors

Consider the two crystalline structures in Fig. 1.

Origin First of all we need an origin, i.e. a place from which to start the description. To
this end choose a random point in the lattice and denote its position d~0 which is identified
with the zero-vector: d~0 = ~0.

Primitive Vectors We then look for vectors (as many as the number of dimensions of
the structure; here 2) with which we can get to all points on the lattice. If we succeed,
we are dealing with a so-called Bravais lattice.
In Fig. 1(a) we may choose

~a1 = ax̂ and ~a2 = aŷ (1)

~ can be reached via
with whom all points R
~ = n1~a1 + n2~a2 ,
R (2)

1
 a 𝑎 b 𝑎
𝑎 3𝑎
𝑎
60∘ 2𝑎

𝑎
𝒂𝟐
𝑑Ԧ1
𝒂𝟏

𝒂𝟐 𝑑Ԧ2
𝒂𝟐 𝑑Ԧ3

𝒂𝟏 𝒂𝟏
𝑑Ԧ1

Figure 1: Lattices. The two lattices considered. a. Square lattice. Green: possible
unit cell, red: the primitive unit cell and the Wigner-Seitz cell coincide. b. Honeycomb
lattice. Two possible sets of base and primitive vectors are indicated. Red: Primitive
unit cell. Since this is not a Bravais lattice there is no Wigner-Seitz cell. If only the red
(hexagonal) lattice is considered the Wigner-Seitz cell is indicated in yellow.

where ni (i = 1, 2) are integers. In other words, with each point R ~ in the lattice a set of
numbers {n1 , n2 } can be associated. The vectors ~a1 and ~a2 are the primitive vectors of
the square lattice.
On the other hand, the points of the structure in Fig. 1(b) are not reachable by just
two vectors. If we choose, e.g., the origin d~0 on a red point and
√ √
3a 3a 3a 3a
~a1 = x̂ − ŷ and ~a2 = x̂ + ŷ (3)
2 2 2 2
all red points can be reached but none of the blue. We are thus not dealing with a Bravais
lattice.

Base Vectors In order to reach all points in Fig. 1(b) we must introduce a basis. This
is equivalent to having multiple origins. While the first one is at d~0 = ~0 we choose a
second one at d~1 = ax̂. Together with the already defined primitive vectors, all points
(red and blue) can now be reached. The vectors d~0 and d~1 thus provide origins for two
equivalent but mutually displaced Bravais lattices.
Note that the choice presented above is not unique. Alternatively one might want to
use the following base and primitive vectors:
√ √
~ ~ ~ 5a 3a ~ 3a 3a
d0 = ~0, d1 = ax̂, d2 = x̂ + ŷ, d3 = x̂ + ŷ (4)
2 2 2 2
√
~a1 = 3ax̂, ~a2 = 3aŷ. (5)
In general though one prefers working with a minimal number of base vectors – in this
case two. It is impossible to find a collection of basis and primitive vectors with only
one base vector. If it where possible, so only d~0 = ~0, it would be a Bravais lattice (by
definition).
We have shown two different ways to describe a honeycomb lattice. This illustrates
that a non-Bravais lattice can be thought of as mutually displaced Bravais lattices in
various ways. Here, either as two triangular (aka hexagonal) or as four rectangular lattices.

2
1.2 Unit Cells
A (small) structure which repeats itself throughout the crystal is called a unit cell. The
shape of a unit cell must be such that the entire area (or volume in 3D) can be covered
with replicas of one and the same unit cell without leaving any holes – kind of like kitchen
tiles. Some examples are shown in Fig. 1.
Although unit cells come in various forms, shapes and sizes (like kitchen tiles), two
types are of particular importance: primitive unit cells and the Wigner-Seitz cell.

Primitive Unit Cell A unit cell that contains only one Bravais-lattice point is called
a primitive unit cell. The number of atoms in a primitive cell is thus equal to the number
of base vectors. Note though that the primitive cell must not necessarily have the same
symmetry as the crystalline structure.
The easiest way to find the primitive unit cell is to construct the parallelogram of the
primitive vectors. Its volume (in 3D) is thus given by
VP = |(~a1 × ~a2 ) · ~a3 | (6)
which can be associated with the inverse atom density: n = 1/VP . In the 2D case just
replace ~a3 = ẑ.

Wigner-Seitz Cell If the primitive unit cell has the same symmetry as the structure
we call it the Wigner-Seitz cell. Note that this is only possible for a Bravais lattice. The
honeycomb lattice, e.g., does not have a Wigner-Seitz cell, but its sub-lattices (red or blue
points only) do.
To construct the Wigner-Seitz cell one must find the area which is closest to a given
lattice point. To this end, connect the origin to its closest neighbors and draw a line
through the middle. Then connect the origin to its next-nearest neighbors and draw a
line through their middle. After repeating this procedure, the thus enclosed area is the
desired cell.

2 Face Centered Cubic Lattice

Statement of the Problem Consider the two two-atomic FCC lattices sketched in
Fig. 2. While in Fig. 2(a) all corner atoms are of type blue and all face atoms are of
type orange, Fig. 2(b) illustrates a FCC lattice in which the corners and two faces are
blue and only the four remaining faces are orange. For both of them find an appropriate
description in terms of lattice and base vectors. How many atoms are in a primitive cell?

a) The blue points form a simple cubic lattice. The three primitive vectors are thus
~a1 = ax̂, ~a2 = aŷ and ~a3 = aẑ. (7)
The orange points form an additional three cubic lattices. Hence, besides the origin
d~0 = ~0, which is the first base vector, we need another three base vectors:
ax̂ + aŷ ax̂ + aẑ aŷ + aẑ
d~1 = , d~2 = and d~3 = . (8)
2 2 2
Since there are a total of four base vectors the primitive cell must contain four atoms,
namely one blue and three orange.

3
 a b

𝑎Ԧ1
𝑎Ԧ3

𝑑Ԧ3 𝑎Ԧ3
𝑑Ԧ2

𝑑Ԧ1
𝑎Ԧ2
𝑎Ԧ2

𝑎Ԧ1 𝑑Ԧ1

Figure 2: Two-atomic FCC lattices. The primitive vectors are indicated in red and
the base vectors green dashed.

b) The blue points come in planes parallel to the x − z plane and distanced by a in the
ŷ direction. We can thus use two vectors (~a1 and ~a2 ) to describe the plane and ~a3 = aŷ
to take us from plane to plane:

ax̂ + aẑ ax̂ − aẑ

~a1 = , ~a2 = and ~a3 = aŷ. (9)
2 2

This structure is also formed by the orange points, only displaced. In addition to d~0 = ~0,
the base vector
ax̂ + aŷ
d~1 = (10)
2
takes care of this. A primitive cell contains two atoms: one blue and one orange.

3 Packing Ratio
Definition Imagine a unit cell where all lattice points are replaced by balls. The size of
the balls are increased (all together) until they touch each other. The ratio between the
volume of the balls (V ) and the volume of the unit cell (V ) is called the packing ratio:

V
R= . (11)
V

a b

Figure 3: Packing rario. a. Cubic lattice. b. FCC lattice.

4
Cubic Lattice [Fig. 3(a)] Lets choose the standard primitive unit cell, a cube with a
ball in each corner, as our unit cell. The balls increase in size until their radii reach a/2.
There are 8 balls (one for each corner) but only 1/8 of each is actually in the unit cell.
The total volume taken up by the balls is thus equal to the volume of one whole ball:

4π  a 3
V = . (12)
3 2
The unit cell on the other hand covers the volume V = a3 and hence the packing ratio is

4π 1 π
R= 3
= ≈ 0.52 (13)
3 2 6

FCC Lattice [Fig. 3(b)] In the unit cell with volume V = a3 there are 8 times 1/8 balls
in each corner and 6 times 1/2 balls on each face. As the radii are increased they touch
sooner than in the simple cubic √
case. Looking at√just one plane, the maximal radius must
be 1/4 of the diagonal (length 2a), namely: 2a/4. Hence, the total volume enclosed
by the balls is
√ !3
πa3
 
1 1 4π 2a
V = 8 · + 6 · = √ (14)
8 2 3 4 3 2
and the packing ratio is found to be
π
R = √ ≈ 0.74 (15)
3 2

The FCC lattice is denser than the cubic lattice.

4 Symmetries
Introduction Consider a 2D square lattice and start rotating it around the origin.
When the rotated lattice overlaps the original lattice, stop and record the angle. Repeat
this procedure until a full rotation (2π) is completed. The angles that you find are:

π 3π
θ = 0, , π, . (16)
2 2

Statement of the Problem Find all possible angles that are allowed in a generic
lattice.

𝑎′

𝑎 𝑎
𝜃 𝜃
A’ A B B’
𝑎

Figure 4: Symmetries.

5
Solution We begin by considering two adjacent lattice points (A and B; see Fig. 4)
distanced by a. The lattice (whose structure is not given) is rotated around A in the
clockwise direction, as indicated by the dashed arrow, until atom A0 overlaps with a point
on the not-rotated lattice. The same is done for B 0 with a counter-clockwise rotation
around B. This results in the trapezoid shown in Fig. 4, where the length of a0 is given
by a0 = a + 2a cos θ. But since the rotated atoms A0 and B 0 overlap with original lattice
points, the distance between them must be an integer multiple of a and hence

a + 2a cos θ = na. (17)

This can be rephrased as

n−1
cos θ = . (18)
2
From the condition | cos θ| ≤ 1 we find that only the integers n = 3, 2, 1, 0, −1, which
correspond to the angles
π π 2π
θ = 0, , , , π (19)
3 2 3
are allowed. A rotation around any other angle is never a symmetry of a lattice.

5 Interatomic Forces
A crystal is a large collection of atoms bound to each other. But how are they bound?
How come they do not fall apart?

5.1 Valence Electrons

The underlying principle comes from chemistry: An atom in a molecule or crystalline
structure will always strive towards having as many valence electrons1 as possible (usually
8, but 2 in the case of H and He). To this end an atom with only a few valence electrons
can give them to another atom with an almost full outer shell. Alternatively atoms can
share valence electrons so that both become full. This interaction is called a covalent
bond. Lets discuss an example for both options.

Ionic Lattice Consider a crystal made of sodium (Na, 1 valence electron) and chlorine
(Cl, 7 valence electrons) atoms. The Na gives an electron to the Cl turning them both
into ions: Na+ and Cl− . Since the atoms are now charged the coulomb force holds them
together and they form a lattice (each ion forms an FCC lattice by itself, one is offset
with respect to the other). The NaCl crystal is commonly known as table salt.
Why is salt soluble in water? Water molecules (H2 O) have a large electric dipole
moment due to the large charge imbalance between the oxygen and hydrogen atoms. They
stick with their negative (positive) end to the Na+ (Cl− ) atoms, effectively screening the
strong Coulomb forces acting between the Na+ and Cl− . The crystal falls apart.
1
The electrons in the outer most shell (in Bohr’s shell model) are called valence electrons. These are
the primary electrons that participate in ineratomic interactions because they are the furthest from the
nucleus and hence the closest to other atoms. The atoms in the periodic table are arranged such that
the column represents the natural number of valence electrons.

6
Covalent Lattice A lattice made up of atoms bound through covalent bonds is called
a covalent lattice. Arguably the most prominent example is the diamond whose lattice
structure is called the diamond cubic structure. It may be imagined as a regular FCC
lattice but with two carbon (C) atoms associated with each FCC lattice point. Since a C
atom has 4 valence electrons it will bind to 4 other C atoms and thus become stable.

5.2 Noble Gas Lattice

A noble gas2 is an atom which naturally already has 8 valence electrons. The above
mentioned chemical principle can thus not be responsible for the binding of such atoms. In
fact, the force holding a noble gas crystal together is based on dipole-dipole interactions.
Each atom is made of a positively charged nucleus surrounded by a negative electron
“cloud”. When two atoms come close to each other they mutually induce dipole moments
which interact via the Coulomb force. This interaction is known as the van der Waals3
interaction.
The induced dipole-dipole interaction is attractive (repulsive) at long (short) distances.
Starting from a gaseous phase, such interactions together with dissipation can be shown
to form a crystalline structure4 .

Statement of the Problem Here we consider a simplified, 1D, two atomic model for
the van der Waals interaction. Two identical atoms are positioned at a distance R from
each other. The coordinate of the electronic “cloud” of atom i, ri  R, is measured with
respect to its nucleus and confined to 1D. This “cloud” is bound to the nucleus via a
“spring” and hence described by a harmonic potential (frequency ω0 ).
Calculate the energy of the system without interatomic interactions: E0 , and with
interatomic interactions: E. By showing that E < E0 prove that the bound state is
stable.

Natural Space The energy of atom i is p2i /2m + mω02 ri2 /2 and hence the unperturbed
Hamiltonian is given by

p21 1 p2 1
H0 = + mω02 r12 + 2 + mω02 r22 . (20)
2m 2 2m 2
The interaction V has four contributions:

• nucleus-nucleus: e2 /R

• electron-electron: e2 /(R + r2 − r1 )

• nucleus 1 - electron 2: −e2 /(R + r2 )

• nucleus 2 - electron 1: −e2 /(R − r1 )

2
The atoms found in the last column of the periodic table are called noble gases. At room temperature
they are in the gaseous phase, hence the name.
3
Johannes Diderik van der Waals is best known for his work on the equation of states for which he
received the 1910 Nobel prize.
4
See this recent seminar by Jonathan Simon from the University of Chicago. Starting from around
minute 2.20 to about 3.20 you can witness a simulation illustrating this concept.

7
and hence
e2
 
1 1 1
V = 1+ r2 −r1 − r2 − . (21)
R 1+ R
1+ R 1 − rR1
For the latter three contributions, in virtue of ri  R, we use the Taylor approximation
1/(1 + x) ≈ 1 − x + x2 :
" 2 #
2

e r2 − r1 r2 − r1 r2  r2  2 r1  r1  2
V = 1+1− + −1+ − −1− −
R R R R R R R
" #
2
e2

r2 − r1  r 2  r 2
2 1
= − − (22)
R R R R
2e2 r1 r2
= − .
R3
The total Hamiltonian, which is given by H = H0 + V , describes a system of two coupled
oscillators

Diagonal Space In order to find the energy of the system the Hamiltonian H must be
diagonalized (the coupled oscillators must be decoupled). To this end we introduce the
new coordinates and momenta:
r1 + r2 r1 − r2
rs = √ , ra = √ , (23)
2 2
p1 + p 2 p1 − p2
√ps = and pa = √ . (24)
2 2
Plugging them into the expression for H0 in Eq. (20) we get
2 2
(p2s + p2a ) mω02 2 (p2s − p2a ) mω02
H0 = + (rs + ra ) + + (rs − ra )2
4m 4 4m 4
p2s mω02 2 pa2 mω02 2
= + r + + r . (25)
2m 2 s 2m 2 a
This is the Hamiltonian of two independent harmonic oscillators. The ground state energy
of H0 is thus
~ω0 ~ω0
E0 = + = ~ω0 . (26)
2 2
Of course, this could also be seen directly in Eq. (20).
We now turn our attention to the interaction term. Plugging Eq. (23) into Eq. (22)
one obtains
e2 e2 2
V = − 3 (rs + ra ) (rs − ra ) = − 3 rs − ra2


(27)
R R
and the total Hamiltonian H = H0 + V is found to be
p2s 2e2 pa2 2e2
   
m 2 2 m 2
H= + ω0 − 3
rs + + ω0 + 3
ra2 . (28)
2m 2 mR 2m 2 mR
This Hamiltonian describes two independent oscillators with eigen frequencies ωs and ωa
(ground state energy: E = ~ωs /2 + ~ωa /2), where
s
2e2
ωa/s = ω0 1 ± 2 . (29)
ω0 mR3

8
 √
Using the Taylor approximation 1 + x ≈ 1 + x/2 − x2 /8 we may write

e2 e4
 
ωa/s = ω0 1 ± 2 − . (30)
ω0 mR3 2ω04 m2 R6

The ground state energy of H is thus given by

~ ~e4
E= (ωs + ωa ) = ~ω0 − 4 2 6 . (31)
2 2ω0 m R

Comparing Eq. (26) with Eq. (31) shows that E < E0 and hence the van der Waals
interaction bound state is energetically preferred and stable.