Crystal Structure and Dynamics

Paolo G. Radaelli, Michaelmas Term 2015

Part 1: Symmetry in the solid state

Lectures 1-5
Web Site:
http://www2.physics.ox.ac.uk/students/course-materials/c3-condensed-matter-major-option

Bibliography
◦ C. Hammond The Basics of Crystallography and Diffraction, Oxford University Press (from Black-
wells). A well-established textbook for crystallography and diffraction.

◦ D.S. Sivia, Elementary Scattering Theory for X-ray and Neutron Users, Oxford University Press,
2011. An excellent introduction to elastic and inelastic scattering theory and the Fourier trans-
form.

◦ Martin T. Dove, Structure and Dynamics: An Atomic View of Materials, Oxford Master Series in
Physics, 2003. Covers much the same materials as Lectures 1-11.

◦ T. Hahn, ed., International tables for crystallography, vol. A (Kluver Academic Publisher, Dodrecht:
Holland/Boston: USA/ London: UK, 2002), 5th ed. The International Tables for Crystallography
are an indispensable text for any condensed-matter physicist. It currently consists of 8 volumes.
A selection of pages is provided on the web site. The full content of Vol A is accessible (at
Oxford) on http://it.iucr.org/Ab/contents/.

◦ C. Giacovazzo, H.L. Monaco, D. Viterbo, F. Scordari, G. Gilli, G. Zanotti and M. Catti, Fundamen-
tals of crystallography (International Union of Crystallography, Oxford University Press Inc.,
New York)

◦ Paolo G. Radaelli, Symmetry in Crystallography: Understanding the International Tables , Oxford

University Press (2011). Contains a discussion of crystallographic symmetry using the more
abstract group-theoretical approach, as described in lectures 1-3, but in an extended form.

◦ ”Visions of Symmetry: Notebooks, Periodic Drawings, and Related Work of M. C. Escher”,

W.H.Freeman and Company, 1990. A collection of symmetry drawings by M.C. Escher, also to
be found here:
http://www.mcescher.com/Gallery/gallery-symmetry.htm

◦ Neil W. Ashcroft and N. David Mermin, Solid State Physics, HRW International Editions, CBS
Publishing Asia Ltd (1976) is now a rather old book, but, sadly, it is probably still the best solid-
state physics book around. Graduate-level but accessible.

◦ Zbigniew Dautera and Mariusz Jaskolskib, How to read (and understand) Volume A of Interna-
tional Tables for Crystallography: an introduction for nonspecialists, Journal of Applied Crystal-
lography 43, 11501171, (2010). A very useful paper explaining how to use the ITC Vol A. Also
linked on the course web site.

1
Contents

1 Lecture 1 — The translational and rotational symmetry of crystal

in “real” space. 6

1.1 Introduction: crystals and their symmetry . . . . . . . . . . . . 6

1.2 Translational symmetry of crystals . . . . . . . . . . . . . . . . 7

1.2.1 Bravais Lattices . . . . . . . . . . . . . . . . . . . . . . . 7

1.2.2 Crystallographic coordinates . . . . . . . . . . . . . . . 9

1.2.3 More on fractional coordinates . . . . . . . . . . . . . . 11

1.2.4 Different choices of basis vectors and coordinate systems 12

1.2.5 Measuring distances and angles between atoms . . . . 12

1.2.6 The dot product between two position vectors: the met-
ric tensor . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.3 Rotational symmetry of crystals . . . . . . . . . . . . . . . . . . 14

1.3.1 Crystal symmetry: an infinite “grid” of rotations/reflections 14

1.3.2 Crystallographic rotations (proper and improper) in 3D

and their symbols . . . . . . . . . . . . . . . . . . . . . 15

1.3.3 Proper and improper rotations expressed in terms of

matrices and vectors . . . . . . . . . . . . . . . . . . . . 19

1.3.4 Rotations about points other than the origin . . . . . . . 19

1.3.5 Normal form of the symmetry operators . . . . . . . . . 20

1.3.6 The Seitz and ITC notations . . . . . . . . . . . . . . . . 21

2 Lecture 2 — An introduction to crystallographic groups 22

2.1 More on composition of operators . . . . . . . . . . . . . . . . 22

2.1.1 Compositions of operators in normal form . . . . . . . . 22

2.1.2 The inverse operator . . . . . . . . . . . . . . . . . . . . 22

2.1.3 Rotations around different origins vs Roto-translations . 23

2
 2.2 Invariant points, lines and planes: special points and symmetry
elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.2.1 Graphical representation of symmetry elements . . . . 26

2.2.2 “Companions” of general and special points . . . . . . . 26

2.3 Some elements of group theory . . . . . . . . . . . . . . . . . . 27

2.3.1 Groups as mathematical sets . . . . . . . . . . . . . . . 27

2.3.2 Full crystallographic groups (including translations) . . . 28

2.3.3 Finite groups in crystallography: crystal classes and

point groups . . . . . . . . . . . . . . . . . . . . . . . . 28

2.4 Point Groups in 2 and 3 dimensions . . . . . . . . . . . . . . . 30

2.4.1 Graphical representations of point groups through sym-

metry elements . . . . . . . . . . . . . . . . . . . . . . . 31

2.4.2 Notation of point groups . . . . . . . . . . . . . . . . . . 31

3 Lecture 3 —The symmetry of periodic crystals: space groups 34

3.1 Representation and notation of space groups . . . . . . . . . . 34

3.1.1 Graphical representation and symmetry elements . . . 34

3.1.2 Notation for space groups . . . . . . . . . . . . . . . . . 35

3.2 “Companions” revisited: site multiplicities . . . . . . . . . . . . 36

3.2.1 General and special multiplicities . . . . . . . . . . . . . 36

3.2.2 Companions of special points of symmetry vs points of

equal symmetry . . . . . . . . . . . . . . . . . . . . . . 37

3.2.3 Notation for general and special positions: multiplicity,

Wyckoff letter and site symmetry . . . . . . . . . . . . . 38

3.3 Summary: Understanding the International Tables of Crystal-

lography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4 Lecture 4 — Symmetry in 2 dimensions: wallpaper groups. 43

4.1 Wallpaper vs space group symmetry . . . . . . . . . . . . . . . 43

4.1.1 Bravais lattices in 2D . . . . . . . . . . . . . . . . . . . . 43

3
 4.1.2 Rotations in 2D . . . . . . . . . . . . . . . . . . . . . . . 43

4.1.3 Roto-translations in 2D . . . . . . . . . . . . . . . . . . 44

4.2 Construction and identification of the 17 wallpaper groups . . . 44

4.2.1 Point groups in 2D . . . . . . . . . . . . . . . . . . . . . 44

4.2.2 Wallpaper groups . . . . . . . . . . . . . . . . . . . . . 45

4.2.3 Identification of wallpaper groups . . . . . . . . . . . . . 45

5 Lecture 5 — Reciprocal lattice and reciprocal space 47

5.1 Basic definitions . . . . . . . . . . . . . . . . . . . . . . . . . . 47

5.2 The reciprocal lattice . . . . . . . . . . . . . . . . . . . . . . . . 47

5.2.1 Dual basis in 3D . . . . . . . . . . . . . . . . . . . . . . 47

5.2.2 Centring extinctions . . . . . . . . . . . . . . . . . . . . 48

5.2.3 Dual basis — alternative definition without cross product 51

5.2.4 Dot products in reciprocal space . . . . . . . . . . . . . 52

5.2.5 A very useful example: the hexagonal system in 2D . . 52

5.3 Fourier transform of lattice functions . . . . . . . . . . . . . . . 53

5.4 Atomic-like functions . . . . . . . . . . . . . . . . . . . . . . . . 53

5.5 Centring extinctions . . . . . . . . . . . . . . . . . . . . . . . . 54

5.6 Extinctions due to roto-translations . . . . . . . . . . . . . . . . 54

5.7 The symmetry of |F (q)|2 and the Laue classes . . . . . . . . . 55

6 Appendix I: Fourier transform of lattice functions 56

7 Appendix II: “Real” crystal structures 57

7.1 Cohesive forces in crystals — atomic radii . . . . . . . . . . . . 58

7.2 Close-packed structures . . . . . . . . . . . . . . . . . . . . . . 59

7.3 Packing spheres of different radii . . . . . . . . . . . . . . . . . 60

7.4 Framework structures . . . . . . . . . . . . . . . . . . . . . . . 61

4
7.5 Layered structures . . . . . . . . . . . . . . . . . . . . . . . . . 62

7.6 Molecular structures . . . . . . . . . . . . . . . . . . . . . . . . 62

5
1 Lecture 1 — The translational and rotational sym-
metry of crystal in “real” space.

1.1 Introduction: crystals and their symmetry

In 1928, Swiss physicist Felix Bloch obtained his PhD at University of Leipzig,
under the supervision of Werner Heisenberg. In his doctoral dissertation, he
presented what we now call the quantum theory of crystals, and what a mo-
mentous occasion that was! Much of the technology around us — from digital
photography to information processing and storage, lighting, communication,
medical imaging and much more is underpinned by our understanding of the
behavior of electrons in metals (like copper) and semiconductors (like silicon).
Today’s contemporary physics also focuses on crystals, in particular on prop-
erties that cannot be described in terms of simple one-electron physics, such
as high-temperature superconductivity.

One may ask what is special about a crystal as compared, for example, to
a piece of glass. From previous introductory courses, you already know part
of the answer: crystals have a lattice, i.e., atoms are found regularly on the
corners of a 3-dimensional grid. You also know that some crystals have a
basis, comprising several atoms within the unit cell, which is replicated by
placing copies at all nodes of the lattice.

At a more fundamental level, crystals differ from glasses and liquids because
they have fewer symmetries. In a liquid or glass, properties look identical on
average if we move from any point A to any point B (translation) and if we
look in any direction (rotation). By contrast in a crystal, properties vary in
a periodic manner by translation, and are only invariant by certain discrete
rotations. The theory of symmetries in crystals underpin all of the crystal
physics we have mentioned and much more. Here, we will start by describing
translational symmetry in a more general way that you might have seen so far,
suitable for application to crystals with non-orthogonal lattices. We will then
look at ways of classifying rotational symmetries systematically, and we will
learn how to use the International Tables of Crystallography (ITC hereafter),
an essential resource for condensed matter physics research.

Symmetry can be defined as the invariance of a system upon certain transfor-

mations. Intuitively, we understand for instance that a cube remains invariant
by rotations through 90◦ around a face normal or through 120◦ around a cube
diagonal. What we are doing here is to apply a global transformation, in which
each point of the cube is made to rotate around an axis so that it moves from

6
the original position to a new position. Only for certain values of the rotation
angle is the result the same as the starting configuration.

Unlike a solid polyhedron, which is generally invariant by a discrete set of

rotations/reflections, crystals are invariant by an infinite set of discrete global
transformations — translations, rotations, inversion and combinations thereof.
These will be explored in the remainder of this lecture.

1.2 Translational symmetry of crystals

1.2.1 Bravais Lattices

Translational invariance, periodicity and the presence of a lattice are equiv-

alent and interchangeable properties that characterise all crystals. The fact
that a crystal can be thought as built up by unit cells means that the crystal
is invariant by translation by any of the cell edges and their (infinite) combi-
nations. Although this symmetry is exact only for an infinite crystal, a typical
1mm crystal of a metal is made up of ∼ 10−22 unit cell, making exact transla-
tional lattice symmetry a very good approximation, at least for the purpose of
calculating scattering properties.

As you already know, in 3 dimensions there are 14 Bravais lattices (from 19-
century French physicist Auguste Bravais) — see fig. 1. The lattices have
distinct rotational symmetries, belonging to one of the 7 lattice systems (see
below), and distinct topologies, since some lattice systems admit both primi-
tive and centred lattices1 .

We recall that primitive lattices have nodes only at the corners of the unit
cells, whereas centred lattices (or lattices with a basis) have nodes at other
positions of the unit cell as well. The conventional notation is as follows:

Body-centred lattices (symbol I) have a 2-node basis at the corners and at

the centre of the unit cell (positions 0, 0, 0 and 12 , 12 , 12 ). An example of
this is the BBC lattice. The fractional vector2 [ 12 , 12 , 12 ] (and equally in the
examples below) is known as centring vector.

Single-face-centered lattices (symbols A, B, C) have a 2-node basis at the

corners and at the centre of a pair of faces intersecting the a, b and c
axes, respectively. So, for instance, the nodes of a C-centred lattice are
at 0, 0, 0 and 12 , 21 , 0 etc.
1
The distinction between lattice systems and crystal systems is explained in sec. 2.4.2
below.
2
See below for the square bracket notation.

7
 Lattice
System

Rhombohedral

Figure 1: The 14 Bravais lattices in 3D. The R-centred hexagonal cell can be used as a
conventional cell for the rhombohedral lattice.

8
All-face-centred lattices (symbol F ) have a 4-node basis at the centre of all
faces, i.e., at 0, 0, 0 , 21 , 21 , 0, 0, 21 , 21 2 and 12 , 0, 21 . An example of this is
the FCC lattice.

Rhombohedrally-centred lattices (symbol R) are hexagonal lattices with

a = b 6= c, α = β = 90◦ , γ = 120◦ . R-centred lattices have a 3-node
2 1 1
basis at 0, 0, 0 , 3, 3, 3, and 13 , 32 , 23 (or, alternatively, at 0, 0, 0 , 1 2 1
3, 3, 3,
and 23 , 13 , 23 ; the two alternatives are called obverse and reverse setting
and are equivalent). R-centred lattices are a bit of an anomaly in that
they are completely equivalent to rhombohedral (primitive) lattices (this
is why they are not listed separately in fig. 1). However, the R-centred
conventional unit cell is often more convenient to use than the trigonal
primitive unit cell, which has α 6= β 6= γ.

It is important to emphasise the fact that it is always possible to choose a

primitive cell to describe any lattice. So, why does one bother with centred
cells in the first place, and why do they occur only in certain lattice systems?
The answer is that conventional unit cells are chosen to have the full
rotational symmetry of the lattice when primitive cells do not.

1.2.2 Crystallographic coordinates

The symmetry of the unit cell affects the choice of the coordinate system that
is used to define the position of atoms in the crystal. The logic is that if an
atom exists at positions x, y, z, than the “companion atoms” of that atom in dif-
ferent unit cells will have coordinates x+nx , y +ny , z +nz , where [nx , ny , nz ] is
a vector of integers or simple fractions (see below). We don’t use cartesian
coordinates in crystallography. More formally, these companion atoms are
called symmetry-equivalent atoms by translation.

Let us start by looking at the cubic, tetragonal and orthorhombic lattices you
should already be familiar with. The position of the nodes (i.e., unit cell origins
and centring positions, when present) w.r.t. a chosen origin are expressed by
means of position vectors3 of the form:

rn = a î nx + b ĵ ny + c k̂ nz (1)
3
Here, we will call vector an object of the form shown in eq. 2, containing both components
and basis vectors. In this form, a vector is invariant by any coordinate transformation, which
affect both components and basis vectors. Both basis vectors and components (coordinates)
are written as arrays — see below. For a concise introduction to this kind of notation, see ITC
volume B, chapter 1.

9
where a, b, c are the lattice constants (a = b for the tetragonal lattice, a = b = c
for the cubic lattice) and î, ĵ, k̂ are unit vectors along the x, y and z axes,
while nx , ny and nz are arbitrary integers (primitive lattices) or integers and
half-integers (centred lattices). We can interpret these vectors as invariant
translation symmetry operators — if points A and B are separated by one
such vector, the properties of the crystal are identical in A and B. in particular,
if there is an atom in A, there will be an identical “companion” atom in B.

By symmetry operator we mean a transformation of the coordinates

(e.g., of an atom), which generates the coordinates of a companion
atom.

The position vector of points in the crystal that are not on lattice nodes can
be expressed as

r = a î x + b ĵ y + c k̂ z (2)

One can see the relation between x, y and z (which are dimensionless) and
the usual Cartesian coordinates, which have dimensions. We can define the
basis vectors (not to be confused with the unit cell basis) a = a î, b = b ĵ
and c = c k̂. In the cubic, tetragonal and orthorhombic lattices, the basis
vectors are orthogonal, but this is not so in other most important cases —
for example, that of hexagonal crystal like graphite, where the angle between
a = b 6= c, α = β = 90◦ and γ = 120◦ . We can, however, generalise the
notion of basis vectors so that it applies to cases such as this and even the
most general (triclinic) lattice. In summary:

• In crystallography, we do not usually employ Cartesian coordinates. In-

stead, we employ a dimensionless coordinate systems associated with
appropriate basis vectors.

• Basis vectors have the dimension of a length, and coordinates (position

vector components) are dimensionless.

• We can denote the basis vectors as ai , where the correspondence with

the usual crystallographic notation is

a1 = a; a2 = b; a3 = c (3)

• We will sometimes employ explicit array and matrix multiplication for clarity.

10
 In this case, the array of basis vectors is written as a row 4 , as in [a] =
[a1 a2 a3 ].

• With this notation, a generic position within the crystal will be defined by
means of a position vector, written as

X
r= ai xi = ax + by + cz (4)
i

Points within the unit cell at the origin of the lattice have 0 ≤ x ≤ 1,
0 ≤ y ≤ 1, 0 ≤ z ≤ 1 — the so-called fractional coordinates.

From eq. 4, we can re-write the position of the lattice points in the most
general case as:

X
rn = ai ni = anx + bny + cnz (5)
i

NOTATION: Real-space coordinates of an atom are usually written without

any brackets. When the need for clarity arises, components of position vec-
tors and in general of any vector defined in real space are surrounded by
square brackets, so [x, y, z] is the position vector w.r.t. the origin of an atom
with coordinates x, y, z 5 .

1.2.3 More on fractional coordinates

Fractional coordinates form the basis of the atomic-scale description of crys-

tals. In fact, the position of all the atoms in the crystal in completely defined
in terms of the fractional coordinates of the atoms in just one unit cell. Let an
atom be at fractional coordinates x, y, z, with 0 ≤ x ≤ 1, 0 ≤ y ≤ 1, 0 ≤ z ≤ 1.
Then, all other “companion” atoms (i.e., atoms equivalent by translation) will
be found at coordinates x + nx , y + ny , z + nz , where [nx , ny , nz ] is either
and integer vector with positive or negative integral components (for primitive
unit cells) or an integer vector plus a centring vector of simple fractions (for
centred unit cells).
4
By writing basis vectors as rows and coordinates or components as columns, vectors
become the usual product of a row and a column array. Moreover, upon coordinate trans-
formations, the transformation matrices for basis vectors and components are inverse of one
another. The subscript and superscript notation also emphasises this.
5
Although this is the most commonly used convention, the ITC Vol A does not always follow
it, and uses, for example, t( 12 , 21 , 12 ) to indicate the translation vector [ 12 , 12 , 12 ] .

11
For example, in the case of the I-centred unit cell, companion atoms to x, y, z
will be found at x + nx , y + ny , z + nz and x + nx + 21 , y + ny + 12 , z + nz + 12 , with
[nx , ny , nz ] being an integer vector and [ 12 , 21 , 12 ] being the I-centring vector.

1.2.4 Different choices of basis vectors and coordinate systems

There are many possible choices of primitive unit cells/basis vectors, as well
as many possible choice of the origin of the coordinate system, and conse-
quently there exist many equivalent coordinate systems. In fact, coordinate
transformations are the bane of crystallography, and a lot of effort has been
devoted, to the extent that this is possible, to define universal conventions.
This subject will not be discussed further, except for the following general
points.

• In order to describe the same lattice, the two transformation matrices con-
verting one basis set to the other and vice versa must be integer matri-
ces. This is obvious, since the new basis vectors connect lattice points,
and must be expressible with integral coordinates on the old basis.

• This implies that the determinant of these matrices is exactly 1 or -1 (de-

pending on the relative handedness of the basis sets), because the
determinant of an integer matrix is an integer and the determinant of
the inverse matrix is the inverse of the original determinant.

• One can show that the inverse of an integer matrix M is again an integer
matrix if and only if the determinant of M is exactly 1 or -1. Therefore,
all integer unimodular matrices define valid coordinate transformations
for primitive cells.

1.2.5 Measuring distances and angles between atoms

Measuring distances between nearby atoms is clearly very important, for ex-
ample, to determine whether such atoms are chemically bonded and the
strength of the bond. Likewise, angles between bonds are very important,
since they define the coordination of each atom. We know, for example, that
in diamond each C atom is surrounded by a tetrahedron of C atoms, which
means that the angle between bonds linking to the same atom is the tetrahe-
dral angle — 109.5◦ . Therefore, once we know the fractional coordinates of
all the atoms, we want to be able to measure distances and angles. This is
clearly best done by calculating dot products of the relative position vectors

12
between two atoms, the components of which are simply the difference of the
fractional coordinates. This can be straightforwardly extended to atoms in dif-
ferent unit cells by using their coordinates (fractional plus appropriate integer
vectors) as explained above.

1.2.6 The dot product between two position vectors: the metric tensor

• The dot product between two position vectors is given explicitly by

r1 · r2 = a · a x1 x2 + b · b x1 x2 + c · c z1 z2 +
+a · b [x1 y2 + y1 x2 ] + a · c [x1 z2 + z1 x2 ] + b · c [y1 z2 + z1 y2 ] =
= a2 x1 x2 + b2 x1 x2 + c2 z1 z2 + ab cos γ [x1 y2 + y1 x2 ]
+ac cos β [x1 z2 + z1 x2 ] + bc cos α [y1 z2 + z1 y2 ] (6)

• The quantities in square bracket represent the elements of a symmetric

matrix, known as the metric tensor. The metric tensor elements have
the dimensions of length square.

Gij = ai · aj (7)

Calculating lengths and angles using the metric tensor

• You are generally given the lattice parameters a, b, c, α, β and γ. In terms of these, the
metric tensor can be written as

a2
 
ab cos γ ac cos β
G =  ab cos γ b2 bc cos α  (8)
ac cos β bc cos α c2

• To measure the length v of a vector v:

v1
 

v 2 = |v|2 = [ v 1 v 2 v 3 ]G  v 2  (9)
v3

• To measure the angle θ between two vectors v and u:

v1
 
1
cos θ = [ u1 u2 u3 ]G  v 2  (10)
uv
v3

13
1.3 Rotational symmetry of crystals

We have seen how in all crystals “companion atoms” to an atom in a given unit
cell are generated by translation, i.e., by adding appropriate integer and cen-
tring vectors to their coordinates. In addition to these symmetry-equivalent
atoms by translations, most crystals also include symmetry-equivalent
atoms by rotation, inversion and and reflection, as well as combinations
of these with non-lattice translations (see below). This is very convenient,
because it often enables us to define the crystal fully with a much smaller
number of atoms than those contained in one unit cell.

However, it is worth emphasising from the outset the profound difference be-
tween the rotation/reflection symmetry of a simple polyhedron (say, a cube
or a dodecahedron) and that of a crystal. The symmetry of the polyhedra
consists of a finite set of rotations/reflections that always leave the centre of
the polyhedron invariant. This is called a point-group symmetry, and as
we shall see is also very important in crystallography. By contrast, the full
symmetry of a crystal always contains an infinite number of elements.
Notably, if there is even a single symmetry axis of rotation or a single mirror
plane, then there is always an infinite number of them. Intuitively, this must be
the case because of the lattice translational symmetry: if there is a rotation
axis in one unit cell, there also must be one in all the other unit cells. How-
ever, there are actually more rotations that one would expect purely based on
this argument.

1.3.1 Crystal symmetry: an infinite “grid” of rotations/reflections

Let us see how this comes about. Consider a simple two-fold rotation axis
around the origin, parallel to the b axis. This means that an atom at posi-
tion x, y, z will have a symmetry-equivalent companion atom at −x, y, −z (or
x̄, y, z̄ to use standard crystallographic notation). Both atoms will also have
an infinite set of companions by translation, as explained above. So, for in-
stance, the second atom wil have a companion at 1 − x, y, z̄, which logically
will be a companion of the original atom at x, y, z. This is an example of
composition of symmetries, i.e., of the sequential application of symmetry
operators. The position vectors of atoms 1 and 3 will be related as follows:

      
1−x −1 0 0 x 1
 y  =  0 1 0  y  +  0  (11)
−z 0 0 −1 z 0

14
This can be re-written:

1 1
      
1−x −1 0 0 x− 2 2
 y−  =  0 1 0   y + 0  (12)
z 0 0 −1 z 0

One can see immediately that eq. 12 represent a rotation around a two-
fold axis parallel to b shifted so that it passes through position 12 , 0, 0
rather than through the origin. It is worth noting that this axis is not a com-
panion by translation to the original rotational axis (such companions occur
at 1, 0, 0 etc.).

This is completely general to any translation vector and also to any kind of ro-
tation, reflection or inversion (with appropriate modifications). The presence
of one such rotation/inversion/reflection operators, together with the compo-
sition of symmetries with the lattice translation, generates an infinite grid of
rotations/inversions/reflections around different points of the crystal.

The process is illustrated in fig. 2. The whole construction has the periodicity
of the lattice, so all the unit cells are identical, as one would expect, but there
are 2-fold axes generated in this way also in the centre of the unit cell and at
the mid-points of the edges, as made clear from eq. 12.

1.3.2 Crystallographic rotations (proper and improper) in 3D and their

symbols

We have just seen an example of crystallographic symmetry consisting of lat-

tice translations and an infinite set of two-fold rotations around different points
of the unit cell. Rotations other than two-fold are of course possible, but there
is a limited number of them, because the lattice is required to be symmet-
rical (invariant) by all the symmetry rotations. We recall that improper
rotations arise from the composition of an ordinary (proper) rotation with the
inversion. Each proper rotations has a corresponding improper rotation. The
proper rotation corresponding to the inversion is the identity. In 3 dimensions,
the following proper and improper rotations are compatible with some crys-
tallographic lattices (see fig. 3 for the graphic symbols used in the ITC — see
the next lectures for further explanations of these symbols):

Proper rotations

• Identity (letter symbol 1) (all lattices). This is a “null” operators that leaves
things unaltered.

15
Figure 2: b-axis projection of the infinite grid of 2-fold axes in Space Group P 2 (number 3
in the ITC). The thick black lines define the boundaries of the unit cell. The shaded area is the
asymmetric unit cell (see text). The circles are symmetry-equivalent “companion” atoms.

16
• Two-fold rotation (letter symbol 2) (monoclinic, orthorhombic, tetragonal,
hexagonal and cubic lattices).

• Three-fold rotation (letter symbol 3) (trigonal, hexagonal and cubic lattices).

• Four-fold rotation (letter symbol 4) (tetragonal and cubic lattices).

• Six-fold rotation (letter symbol 6) (hexagonal lattice).

Improper rotations

• Inversion (letter symbol 1̄) (all lattices). The inversion by the origin corre-
sponds to a change in sign of all coordinates in all coordinate systems.

• Mirror plane (letter symbol m) (monoclinic, orthorhombic, tetragonal, hexag-

onal and cubic lattices). It is a composition of a two-fold rotation with
the inversion, and could therefore be written as 2̄.

• “Three-bar” operator (letter symbol 3̄) (trigonal, hexagonal and cubic lat-
tices). It is a composition of a three-fold rotation with the inversion

• “Four-bar” operator (letter symbol 4̄) (tetragonal and cubic lattices). It is a

composition of a four-fold rotation with the inversion

• “Six-bar” operator (letter symbol 6̄) (hexagonal lattice).

The effect of the “three-bar” ( 3̄), “four-bar”( 4̄) and “six-bar” ( 6̄) improper
rotations may be somewhat unfamiliar, and is therefore illustrated in fig. 4.

17
 2 3 4 6 m g

1 3 4 6

m a,b or c n e d

3
8 1
8

2 21 31
41 43 62 64
61 65
32 42
63

Figure 3: The most important symbols for symmetry elements (see sec. 2.2.1 for a defini-
tion) employed in the ITC to describe 3D space groups. Fraction next to the symmetry element
indicate the height (z coordinate) with respect to the origin.

18
 -­‐  
+   -­‐   +/-­‐  

-­‐   +   +   +   +/-­‐  

+   -­‐   +/-­‐  
-­‐  

Figure 4: Action of the 3̄ , 4̄ and 6̄ operators and their powers. The set of equivalent
points forms a trigonal antiprism, a tetragonally-distorted tetrahedron and a trigonal prism,
respectively. Points marked with ”+” and ”-” are above or below the projection plane, respec-
tively. Positions marked with ”+/-” correspond to pairs of equivalent points above and below
the plane. Starting from any of these points, multiple applications of these improper rotations
generate all the other points both above and below the plane.

1.3.3 Proper and improper rotations expressed in terms of matrices

and vectors

Rotations about the origin

In Cartesian coordinates, a rotation about the origin is expressed as:

   
x(2) x(1)
 y(2)  = R  y(1)  (13)
z(2) z(1)

where x(1), y(1), z(1) are the coordinates of the original point, x(2), y(2), z(2)
are the coordinates of the companion point and R in an orthogonal matrix
with det(R) = +1. Matrices with det(R) = −1 represent improper rotations.

In crystallographic (non-Cartesian) coordinates, eq. 13 is modified as:

   
x(2) x(1)
 y(2)  = D  y(1)  (14)
z(2) z(1)

where in general D is a non-orthogonal matrix, which, however, still has

det(D) = ±1. In any coordinate systems, the inversion through the origin is
represented by minus the identity matrix (change of sign of all coordinates).

1.3.4 Rotations about points other than the origin

As we have seen, the internal rotational symmetry of crystals always includes

rotations around an infinite number of points that are not the origin. In Carte-

19
sian coordinates, the expression for a proper of improper rotation through the
point x0 , y0 , z0 , is

     
x(2) x(1) − x0 x0
 y(2)  = R  y(1) − y0  +  y0 
z(2) z(1) − z0 z0
   
x(1) tx
= R  y(1)  +  ty  (15)
z(1) tz

where

     
tx x0 x0
 ty  =  y0  − R  y0  (16)
tz z0 z0

for example, a mirror plane perpendicular to the x axis and located at x = 1/4
will produce the following transformation:

        
x(2) −x(1) + 1/2 −1 0 0 x(1) 1/2
 y(2)  =  y(1)  =  0 1 0   y(1)  +  0  (17)
z(2) z(1) 0 0 1 z(1) 0

Note that when using non-Cartesian coordinates the form of symmetry oper-
ators about a point other than the origin remains the same:

     
x(2) x(1) tx
 y(2)  = D  y(1)  +  ty  (18)
z(2) z(1) tz

In general, D is not orthogonal, but its determinant is still ±1. 6

1.3.5 Normal form of the symmetry operators

Symmetry operators written in the form of eq. 15 are said to be in normal

form. They are written as the composition (application of two operators in
succession) of a rotation (proper or improper) — the rotational part, followed
by a translation — the translational part. Every crystallographic symmetry
6
We will use R in the remainder to indicate rotation matrices in any coordinate system.

20
operator can be written in normal form, including proper and improper
rotations about any point and other compositions of rotations and translations,
as described in more detail in the next lecture.

1.3.6 The Seitz and ITC notations

The Seitz notation is a shorthand notation used in crystallography to denote

operators in normal form. The generic notation of eq. 18 is {R | t}, where R
is the rotational part and t is the translational part (let us not forget that the
rotation is always applied first!). The rotational part R is usually represented
by its letter symbol (see sec. 1.3.2) with a subscript to indicate its orientation
rather than by the corresponding matrix. For example, the Seitz notation of
1
the operator in eq. 17 is {mx | 2 00}. The Seitz notation of a pure (lattice)
translation is {1 | tx ty tz }, where the identity operator (proper rotation of 0
degrees) 1 is represented by the identity matrix 1 in all coordinate systems.

Although the Seitz notation for operators is the most useful to describe the
theory of crystallographic groups, the ITC use a different notation that shows
explicitly the point(s) about which the operator acts with respect to the origin
of the coordinate systems. For example, the ITC notation of the same oper-
ator in eq. 17 is m 14 , y, z, meaning that the points of the mirror plane about
which we are reflecting have coordinates 14 , y, z .

21
2 Lecture 2 — An introduction to crystallographic groups

2.1 More on composition of operators

2.1.1 Compositions of operators in normal form

As we have seen at the end of the previous lecture, all crystallographic sym-
metry operators (i.e., transformation of coordinates to generate companion
atoms) can be expressed in normal form. We have also seen that operator
composition corresponds to the application of two symmetry operators, one
after another. In the remainder, composition will be indicated with the symbol
◦, so that O2 ◦O1 corresponds to the application of O1 first, followed by the ap-
plication of O2 . Let’s ask the following question: assume that O1 = {R1 | t1 }
and O2 = {R2 | t2 } are both in normal form. What is the normal form of
O2 ◦ O1 ? This is easily deduced from eq. 15:

     
x(2) x(1) t1x
 y(2)  = R 1  y(1)  +  t1y 
z(2) z(1) t1z
     
x(3) x(2) t2x
 y(3)  = R 2  y(2)  +  t2y 
z(3) z(2) t2z
     
x(1) t1x t2x
= R 2 R 1  y(1)  + R 2  t1y  +  t2y  (19)
z(1) t1z t2z

whence we deduce the very important relation:

{R2 | t2 } ◦ {R1 | t1 } = {R2 R1 | R2 t1 + t2 } (20)

2.1.2 The inverse operator

Suppose that operator O = {R | t} (in normal/Seitz form) connect the co-

ordinates of an atom at x1 , y1 , z1 with those of its companion point x2 , y2 , z2 .
What is the Seitz form of the inverse operator O−1 that connects x2 , y2 , z2 to
x1 , y1 , z1 ? Since O−1 ◦ O = O ◦ O−1 must be the identity operator, and since
the Seitz notation of the identity operator clearly is is {1 | 000} , from eq. 20 it
must be:

22
 R2 R1 = 1
R2 t1 + t2 = [0 0 0] (21)

whence

O−1 = {R−1 | −R−1 t} (22)

Test your understanding: using this result, prove that the operator {mx | 12 00}
(see sec. 1.3.6) is the inverse of itself. This should be obvious, since applying
a mirror symmetry operator twice, no matter where the plane of the mirror is,
is equivalent to doing nothing at all.

2.1.3 Rotations around different origins vs Roto-translations

All proper and improper rotations about the origin have Seitz notation {R |
000}. As we have seen, proper and improper rotations about a point other
than the origin have Seitz notation {R | t}, where t is an appropriate transla-
tion vector (usually not a lattice vector). However, not all symmetry operators
of the form {R | t} are proper/improper rotations about a point other than the
origin. If we apply such an operator n times, were n is the order of the axis or
of the improper rotation, we must necessarily obtain the identity. By looking
at eq. 20 we can see that, in order for this to be the case, t must not have any
component that is left invariant by R.

Consider for instance the composition of a lattice translation along the a axis
with a rotation around the a axis (rotation applied first):

{1 | 100} ◦ {2x | 000} = {2x | 100} (23)

This is clearly not a rotation about an axis, since no point is left invariant by
it. Another example of this is the composition of a lattice translation along the
[110] direction with a mirror plane perpendicular to the c axis (mirror applied
first):

{1 | 110} ◦ {mz | 000} = {mz | 110} (24)

Operators of this kind are called proper and improper roto-translation. As

we have just seen, an infinite set of roto-translations are generated “automat-

23
ically” by composition of rotations and lattice translations. However, a special
class of symmetries in crystallography is represented by proper or improper
rotations followed by a non-lattice translation.

There are strong restrictions on both the rotational and the translational
parts of proper and improper roto-translation operators, as follows:

• The rotational part of a roto-translation operator must leave either a line

or a plane invariant. Compositions of an improper rotation leaving a
single point invariant with a translation are always improper rotations
about a point other than the origin. Therefore, only 2, 3, 4, 6 (in-
variant lines) and m (invariant plane) can form roto-translations.
Roto-translations where the rotational part is a proper rotation are called
screw operators. Roto-translations where the rotational part is a mirror
plane are called glide operators.

• The translational part of a roto-translation operator must be a vector in

the invariant line or a plane, i.e., Rt = t. Adding a translation vector
perpendicular to the line or plane results in a shift of origin of the roto-
translation (see below for more details).

• The translational part must be equal to a lattice translation divided by

the order of the rotation. This is easy to see by applying the roto-
translation multiple times:

{R | t} ◦ {R | t} = {RR | Rt + t} = {RR | 2t}

{R | t} ◦ {R | t} ◦ {R | t} = {RRR | 3t}
... (25)

where we have exploited the fact that Rt = t. Applying a roto-translation

a number of times equal to the order of the rotation results in a pure
translation. So, for instance, {3 | t} ◦ {3 | t} ◦ {3 | t} = {1 | 3t}, and
this must necessarily be a lattice translation. Symbols are used to
represent roto-translations with the same rotational part and different
translations parts to within a lattice translation (see below for further
clarifications).

Screw axes

Screw axes are proper roto-translation operators, the rotational part of which
is a proper rotation, i.e., 2, 3, 4, 6. The translational part, called screw vector,

24
is parallel to the direction of the rotation axis, and, as stated before, must be
equal to a lattice translation divided by n, where n is the order of the rotation.
Having decided that we will assign unique symbols to roto-translations with
screw vectors within a primitive unit cell, for a rotation of order n one needs
n − 1 distinct symbols, with screw vectors tl /n, 2tl /n, . . . (n − 1)tl /n, where
tl is a simple lattice translation — usually [100], [010], [001], [110] etc. The
corresponding symbol are 21 for two-fold screws, 31 and 32 for three-fold, 41 ,
42 , 43 for four-fold etc., as illustrated in fig. 3.

Glide planes

Glide planes are improper roto-translation operators, the rotational part of

which is a mirror plane m, while the translational part t — the glide vector —
is a non-lattice vector in the mirror plane. Since m ◦ m = 1, the glide vector is
always half of a lattice translation, i.e., t = tl /2. Glide planes have different
names depending on the orientation of the glide vector — see tab. 1. The
graphical representation of glides planes depends on whether they are shown
flat or edge-on in the projection — see fig. 3.

Table 1: Nomenclature for glide planes. The glide vector is tl /2, where t is a
lattice vector.

symbol tl (glide vector is tl /2) Notes

g [10] or [01] along the glide line In two dimensions only
a [100]
b [010]
c [001]
n [110] or [101] or [011]
e [100] and [010] Indicates two symmetry-
related families of parallel
planes. Occurs only in face
centred orthorhombic groups
39, 41, 64, 67, 68.
d [ 12 12 0] and equivalent or [ 21 12 12 ] F and I centred lattices only,
where these tl are lattice vec-
tors.

2.2 Invariant points, lines and planes: special points and sym-
metry elements

From the previous paragraphs, it should already be clear that some points
within the unit cell are special points, by virtue of being left invariant by cer-
tain symmetry operators. By contrast, general points are only left invariant
by the identity operator. A set of special points left invariant by the same

25
operator (or set of operators) is called a symmetry element. Symmetry
elements can be:

• A single point, for operators 1̄, 3̄, 4̄, 6̄.

• A line of points, for operators 2, 3, 4, 6

• A plane of points, for operator m.

Roto-translations do not give rise to special points, because no point is

left invariant by them. However, points on a glide plane or a screw axis are
transformed into other points on the same glide plane or screw axis, so in this
weaker sense they are distinct from all other points. Such points are also said
to be the symmetry element associated with the glide plane/screw axis.

If a point/line/plane is the symmetry element associated with a certain proper/improper

rotation/roto-translation {R | t}, it is necessarily also the symmetry element
of {R | t} ◦ {R | t}, {R | t} ◦ {R | t} ◦ {R | t} and so on for all the powers of
that operator.

2.2.1 Graphical representation of symmetry elements

Looking back at fig. 2 and at a few example pages of the ITC, we can see
many examples of symmetry elements drawn within the unit cell with the spe-
cial symbols shown in fig. 3. The symbols give an intuitive representation of
how the operator and, in some cases its powers (repeated applications), act.
For example, the symbol for 4̄ contains the symbol of a 2-fold axis, because
4̄ ◦ 4̄ = 2.

2.2.2 “Companions” of general and special points

Special points of symmetry, i.e., point lying on symmetry elements other than
roto-translations, have fewer companions by rotation than general points.
This follows from the fact that some symmetry operators transform special
points into themselves, and therefore do not generate companions. The fol-
lowing question arise naturally from this consideration: How many compan-
ion points does each general or special point have in the unit cell in
each type of symmetry? In order to answer this question rigorously, we
have to introduce a few element of the mathematical theory of groups.

26
2.3 Some elements of group theory

2.3.1 Groups as mathematical sets

As we have seen, if two symmetry operators are known to exist in a crystal,

say O1 and O2 , then we can construct a third symmetry operator by applying
O1 and O2 one after the other. This means that the set of all symmetry op-
erators is closed by composition. Likewise, we can always create the inverse
of an operator, e.g., by reversing the sense of rotation and the direction of
translations etc. These considerations indicate that the set of all symmetry
operators of a crystal has the mathematical structure of a group. More rigor-
ously, a group is defined as a set of elements with a defined binary operation
known as composition, which obeys the following rules.

• A binary operation (usually called composition or multiplication) must be

defined. As before, we indicate this with the symbol “◦”.

• Composition must be associative: for every three elements f , g and h of

the set

f ◦ (g ◦ h) = (f ◦ g) ◦ h (26)

• The “neutral element” (i.e., the identity, usually indicated with E) must exist,
so that for every element g:

g◦E =E◦g =g (27)

• Each element g has an inverse element g −1 so that

g ◦ g −1 = g −1 ◦ g = E (28)

• A subgroup is a subset of a group that is also a group. We state the

following without proof: for finite groups, the number of elements
(known as the order) of a subgroup is always a divisor of the order
of the main group.

• A set of generators is a subset of the group (not usually a subgroup) that

can generate the whole group by composition. Infinite groups (e.g., the
set of all lattice translations) can have a finite set of generators (in this
example, the primitive translations).

27
• Composition is in general not commutative, i.e., g ◦ f 6= f ◦ g for some
pairs of elements. Groups for which all compositions commute are
known as Abelian groups. Although translations always commute, a
few examples of applications of eq. 20 should clarify that many oper-
ators in Seitz form do not commute. Therefore, most crystallographic
groups are non-Abelian.

2.3.2 Full crystallographic groups (including translations)

By applying eq. 20 and 22, one can easily become convinced that all the
symmetry operators of a given crystal must form a group as defined
above. By employing the Seitz notation, one can explicitly test associativity
(eq. 26) and the fact that right and left inverse are the same (eq. 28), with the
neutral element being {1 | 000}. The set of pure lattice translations, which
are expressed as {1 | t}, is clearly a group and is always a subgroup of any
crystal symmetry group. Therefore, crystal symmetry groups are infinite
groups.

• In 3 dimensions, crystal symmetry groups are known as space groups.

As we shall see, there are 230 such groups.

• In 2 dimensions, crystal symmetry groups are known as planar (or wall-

paper) groups, the second name referring to the fact that these groups
describe all possible symmetries of a wallpaper (excluding colour sym-
metry). There are 17 wallpaper groups.

• The seven frieze groups describe all the possible symmetries of 1-dimensional
patterns.

2.3.3 Finite groups in crystallography: crystal classes and point groups

Although the full symmetry of a crystal is always described by an infinite

group, finite groups have a very important place in crystallography. These are
groups of proper and improper rotations around the origin, and are completely
equivalent and isomorphic to groups of matrices used to describe these rota-
tions, e.g., in Cartesian coordinates. There are two important uses of these
finite groups in crystallography:

Crystal classes

Crystal classes are groups formed by all rotational parts of all symmetry
operators of a given space or wallpaper group. As one can see from eq.

28
20, the rotational parts of symmetry operators in Seitz form are simply mul-
tiplied by each other upon composition, in the usual sense of rotation matrix
multiplication. Also, the rotational part of the inverse operator is the inverse
rotational matrix (eq. 22). It is therefore evident that all the rotational parts of
all operators in a given symmetry, considered in isolation from the associated
translations, form a finite group, known as the crystal class of that space or
wallpaper group.

Site symmetry groups

Let us consider a generic point in a crystal, shift the origin of the coordinate
system onto that point and consider all operators of Seitz form {R | 000}
in that coordinate system, i.e., all operators describing proper and improper
rotations around that point. There is always at least one such operators,
i.e., {1 | 000}. For points of special symmetry, the site symmetry group7
is the set of operators that leave that point invariant. Remembering the dis-
cussion in sec. 2.2, we conclude that the symmetry elements (excluding
roto-translations) of the site symmetry group operators at a certain point all
pass through that point. Roto-translations are never associated with any
site symmetry group, because they can never be written as {R | 000}, re-
gardless of the choice of origin.

The equivalence between site symmetry groups and crystal classes

From the definitions above, it should not be difficult to convince oneself that
site symmetry groups at any point in the crystal are subgroups of the
crystal class of the space/wallpaper group. However, in many space and
wallpaper groups, there is no point in the crystal with a point group
symmetry equal to the crystal class. Groups in which this is the case are
called symmorphic groups. It turns out that there is at least one symmorphic
group for each crystal class8 , so that, for classification purposes, the set of
all possible crystal classes is equal to the set of all possible site symmetry
groups. In the remainder, we will discuss the crystallographic point groups
(which are 10 in 2 dimensions, 32 in 3 dimensions), keeping in mind that this
refers both to possible site symmetry groups and to crystal classes.
7
I call “site symmetry group” the symmetry around a fixed point in the crystal, since “point
group” is used as a more general term that also denotes crystal classes — see below.
8
in fact, there are many more symmorphic groups (73 in 3 dimensions) than crystal classes
(32 in 3 dimensions) For example, P mmm, Cmmm, F mmm and Immm are all symmorphic
groups of the crystal class mmm.

29
2.4 Point Groups in 2 and 3 dimensions

Point groups are groups of symmetry operators that leave one point invari-
ant, and correspond to the symmetries of simple polygons and polyhedra.
Although there is an infinite number of point groups, the number of crystallo-
graphic point groups is rather small, since only axes of order 2, 3, 4 and 6 are
allowed (other axes are not compatible with a crystal lattice)9 . There are 10
crystallographic point groups in 2 dimensions and 23 in 3 dimensions.
Crystallographic point groups are listd in the ITC vol A from page 763 (green
edition). The relevant pages are linked through from the course web site. The
entry for point group 6mm is shown in fig. 5 .

Equivalent  points  for  

Primary  symmetry  direc-on   general  posi-on  (here  12  c)   Symmetry  elements  representa-on  
Secondary  symmetry  direc-on  
Ter-ary  symmetry  direc-on  

Site  symmetry  
Wychoff  le7er  
Site  mul-plicity  

Figure 5: An explanation of the most important symbol in the Point Groups subsection of the
International Tables for Crystallography. All the 10 2D point groups (11 if one counts 31m and
3m1 as two separate groups) are reproduced in the ITC (see lecture web site). The graphical
symbols are described in fig. 3. Note that primary, secondary and tertiary symmetries are
never equivalent by symmetry (see text).

9
For example, the well-known icosahedral point groups 235 and m3̄5̄ are not crystallo-
graphic point groups because they have 5-fold axes.

30
2.4.1 Graphical representations of point groups through symmetry el-
ements

The most intuitive way to represent a point group is through its symmetry ele-
ments, which all intersect at the point that is left invariant by that point group.
Allowed symmetry elements of a point group are all pure proper and improper
rotations shown fig. 3. Roto-translations are obviously not allowed, because
they leave no point invariant. An example of the ITC point group entries is
shown in 5, and the symmetry elements representation is shown in more de-
rails in 6 . One point worth emphasising is that some symmetry elements are
related to each other by symmetry (e.g., the group of mirror planes marked
with “1” are related to each other by the 6-fold rotations), whereas others are
not (the mirror planes marked with “1” and “2” are not related to each other
by any symmetry)10 . This has important consequences in understanding the
letter notation of point groups (sec. 2.4.2).

Figure 6: Left. A showflake by by Vermont scientist-artist Wilson Bentley, c. 1902. Right

The symmetry group of the snowflake, 6mm in the ITC notation. The number identifies groups
of operators that are equivalent by symmetry.

2.4.2 Notation of point groups

In the notation employed by the ITC, known as Hermann–Mauguin notation11 ,

point groups are also described using up to 4 symmetry operator symbols for
10
This is formalised in the theory of conjugation classes of group elements, which is not
discussed here.
11
There is an alternative notation for point groups, known as the Schoenflies notation, which
is very popular in older physics articles and books. Schoenflies symbols are listed in the ITC
together with the Hermann–Mauguin notation.

31
pure proper and improper rotations (fig. 3). The following bullet points provide
a key to understand point group symbols.

• When the first symbol is 2, 4 or 6 and the second symbols is m, they can
be joined by a slash, as in 2/m and (4/m)mm (pronounced “two over
m” and “four over m-mm”), forming a compound symbol. This has the
meaning of a 2-fold, 4-fold or 6-fold axis with a mirror plane perpendic-
ular to it.

• Each symbol or compound symbol is oriented along a particular symmetry

direction in the crystal. A symmetry direction is simply the direction of
a symmetry element; remembering that m ≡ 2̄, the symmetry direction
of a mirror plane is perpendicular to the plane. In 3 dimensions, there
are up to 3 inequivalent (see here below) symmetry directions, known
as primary, secondary and tertiary.

• Symmetry directions listed in point group symbols are never equivalent by

symmetry. In other words, the symbols never lists symmetry-equivalent
symmetry elements (see discussion in sec. 2.4.1).

Tab. 2 lists the symmetry directions and possible point groups for each of the
7 crystal systems12 .

I
II III

Figure 7: The primary, secondary and tertiary symmetry directions for cubic groups.

12
Roughly speaking, the 7 crystal systems are set of point groups/crystal classes sharing
the same axis of highest order and the same symmetry directions. They map onto the 7
lattice systems (fig. 1 with one exception: the trigonal crystal system, which is characterised
by the presence of a 3-fold axis of highest order, can give rise to crystals with rhombohedral
(R-centred hexagonal) or primitive hexagonal lattices

32
Table 2: Crystal systems, symmetry directions and point groups in 3 dimen-
sions. Although the notion of basis vectors is not intrinsic to point groups,
secondary and tertiary symmetry directions in the tetragonal, trigonal and
hexagonal systems are referred to a basis , with the secondary direction
(second symbol) always chosen along the a axis. This creates pairs of point
groups which are geometrically the same but are rotated with respect to each
other in a fixed basis frame.

Crystal system Symmetry direc- Point groups notes

tions
Triclinic None 1, 1̄
Monoclinic Unique symmetry 2, m, 2/m Conventionally along b
direction or c axis
Orthorhombic 3 orthogonal 222, 2mm, mmm Various axis choices
symmetry direc-
tions
Tetragonal 4 or 4̄ or 4/m axis 4, 4̄, 4/m 4-fold axis along c
,, There can be 422, 4mm, 4̄m2, Reference to basis vec-
two additional 4/mmm tors creates a distinc-
2-fold symmetry tion between 4̄m2 and
directions in the 4̄2m, which are rotated
orthogonal plane, by 45◦ .
at 45◦ to each
other
Trigonal 3 or 3̄ axis 3, 3̄ 3-fold axis along c
,, There can be 32, 3m, 3̄m Reference to basis vec-
one additional tors creates the pairs
2-fold symmetry 32 → 312/321, 3m →
direction in the 31m/3m1 and 3̄m →
orthogonal plane 3̄1m/3̄m1, which are
rotated by 30◦ .
Hexagonal 6 or 6̄ or 6/m axis 6, 6̄, 6/m 6-fold axis along c
,, There can be 622, 6mm, 6̄m2, Reference to basis vec-
two additional 6/mmm tors creates a distinc-
2-fold symmetry tion between 6̄m2 and
directions in the 6̄2m, which are rotated
orthogonal plane, by 30◦ .
at 30◦ to each
other)
Cubic 2-fold and 3-fold 23, m3̄ See fig. 7 for symmetry
symmetry direc- directions
tion along cube
faces and diago-
nals
,, 2-fold along cube 432, 4̄3m, m3̄m
faces becomes
4-fold, additional
2-fold symmetry
direction along
cube edges

33
 f edges !hk" !!hk"
egment (c)
f parallel edges3 Lecture 3 —The symmetry of periodic !crystals:
!10" !10" space
egment through origin
groups
e edge (01) or !0!1"
point (a)
3.1 Representation and notation of space groups

3.1.1 Graphical representation and symmetry elements

ational Tables
As wefor Crystallography
have (2006).
seen in sec. 2.4.1, point Vol.
groups can A, Space group
be represented 108, pp. 400–401.
by a collec-
tion of symmetry elements, each oriented along one of up to 3 inequivalent
b !!h !k"that!!hk"
!hk"2.3.3)
symmetry directions. We have also seen (sec. !h!k"
the rotational parts

cm C 10
ngle (i)

4mm
of the symmetry operators of a space group for a group, which is one of the
f parallel edges !0!1"
32 (3D) point
egment through origin (g) groups 4v
13 . It follows logically !01"
that the symmetry elements of
a space group are proper or improper rotations and roto-translations,
f parallel edges !10" !!10"
108
egment through with rotational
origin (e) I 4cm
parts corresponding to the elements of the crystal class
and different origins and/or glide/screw vectors. These operators are
Patterson sym
oriented in the same way as the elements of the crystal class.

e !hk" !!h !k" !!kh" !k !h"

e (d)

agon !hk" !h! !k" !!kh" !k !h"

ated square (g) !!hk" !h!k" !kh" !!k !h"
e
!11" !!1!1" !!11" !1!1"
e (f)
e !10" !!10" !01" !0!1"
e (d)
Figure 8: Symmetry elements in the unit cell of space group I4cm (left) and of its crystal
class 4mm (right).
768
An example of this for space group I4cm (no. 108) is illustrated in fig. 8. The
on 4c e
crystal class of I4cm is 4mm (right panel), which has 8 group elements: 7
metric unitare shown
0 ≤ xas≤symmetry
2;
1
0 ≤elements
y ≤ 12 ; —0 remember
≤ z ≤ 12 ; that
y≤ −x
the12 symmetry element
for 4+ (rotation by 90◦ is the same as that for 4+ ◦ 4+ = 2 and 4+ ◦ 4+ ◦
etry operations 13
Here, we are specifically discussing space groups, although, as we shall see in the next
0, 0)+ set lecture, the concepts and notations are very similar also for wallpaper groups (2D).

(2) 2 0, 0, z (3) 4+ 0, 0, z (4) 4− 0, 0, z

x, 0, z (6) c 0, y, z (7) c x, x̄, z
34 (8) c x, x, z
1
2, 12 )+ set
, 12 , 12 ) (2) 2(0, 0, 12 ) 1
4 , 14 , z (3) 4+ (0, 0, 12 ) 0, 12 , z (4) 4− (0, 0, 12 ) 1
2 , 0, z
1 1 1 1 1
4+ = 4− ), to which one has to add the identity. The symmetry elements of
I4cm all correspond to those of 4mm in their rotational part, but have different
translational parts, which is reflected by their different origin and in some
cases by the glide/screw vectors. For example, along the c axis there are both
4-fold rotations and 4-fold screws (42 ), as well as 21 screws; perpendicular to
the horizontal axis there are two types of glide planes (c and b); perpendicular
to the [110] and [11̄0] direction there are both proper mirrors (m) and glides
(e).

3.1.2 Notation for space groups

The notation for space groups may appear to be daunting, but is actually
rather simple if one has understood the point group notation in sec. 2.4.2.

• The first symbol is always the lattice symbol, one of the 14 Bravais lattice.

• There is a precise correspondence between the symbol of the crystal

class (e.g., 4mm in the example above) and the remaining symbols if
the space groups (e.g., I4cm (no. 108)), whereby symbols in the crys-
tal are either kept as such or replaced by a roto-translation symmetry
operator with the same rotational part. Therefore, the symbol 2 in the
crystal class can remain 2 or replaced by 21 ; m can be replaced by g
(2D wallpaper groups only), a, b, c, d, n and e; 3 can be replaced by 31
or 32 etc. It is therefore extremely easy to deduce the crystal class from
the space group symbol: for example P 4̄c2 (No. 166) → 4̄m2.

• As explained above, in each space group there are usually several opera-
tors with the same rotational part but different translational parts, often
a mixture of rotations and roto-translations. Therefore, an order of pri-
ority must be established to decide which of these operators to list in
the space group symbol. As a rule, pure proper/improper rotations
are listed if they exist, in preference to roto-translations.

• The symbol of a symmorphic group is always the lattice symbol followed by

the crystal class symbol, e.g., I4/mmm.

35
3.2 “Companions” revisited: site multiplicities

3.2.1 General and special multiplicities

Armed with some knowledge of group theory and of the crystallographic point
groups, we can now revisit and answer the question posed in sec. 2.2.2: How
many companion points does each general or special point have in the
unit cell in each space or wallpaper group symmetry? We shall first state
an important theorem (the proof is given in the footnote14 ).

Theorem: if companions p1 0 and p2 0 to a generic point p are generated by two

operators having the same rotational part, then either p1 0 = p2 0 or p1 0 and p2 0
are in different primitive unit cells.

General multiplicity The consequence of this theorem is that each element

of the crystal class generates to each generic point exactly one companion
point in the primitive cell. Therefore, the number of equivalent points to a
generic point in the primitive unit cell is the order (number of elements) of
the crystal class. The general multiplicity (or multiplicity of the general po-
sitions) is defined as the number of equivalent points to a generic point in the
conventional unit cell. The general multiplicity of a space (or wallpaper)
group is the order of its crystal class times the number of nodes in the
basis (see sec. 1.2.1).

Example (see later for the interpretation of these symbols), for space group
Ia3̄ (no. 206 in the ITC, page 632 in the ”green edition”) has crystal class m3̄
(page 787 in the ITC), which has order 24. The general multiplicity is the order
of the crystal class times the number of nodes in the basis, in these case 2,
since this is an I-centred lattice. The multiplicity (number of equivalent sites
in the conventional unit cell) of a generic point x, y, z for Ia3̄ is therefore 48
(see top of page 633).

Multiplicity of special positions

As we have discussed in sec. 2.2.2, special points of symmetry have fewer

companions than general positions, because certain operators bring that point
into itself. But how many companions will remain? This can be determined
rigorously using more advanced group theory (the so-called cosets and quo-
tient groups). The answer is stated below, and an intuitive explanation is
14
In fact, let p1 0 = {R | t1 } p and p2 0 = {R | t2 } p. We can write p = {R−1 | −Rt1 } p1 0
and p2 0 = {R | t2 } ◦ {R−1 | −Rt1 } p1 0 = {1 | −R2 t1 + t2 } p1 0. Therefore p1 0 and p2 0 are
related by a translation, which is necessarily a lattice translation if these are generic points.
We conclude that they are either the same point (the translation being [000]) or are in different
primitive cells.

36
given in the footnote15 .

The multiplicity of a special point (≡ number of companions in the con-

ventional unit cell) is the general multiplicity of that space (or wallpaper)
group divided by the order (≡ number of elements) of the point group at
that point. Note that, as stated in sec. 2.3.1, the order of the point group is
a divisor of the order of the crystal class, because site symmetry groups are
subgroups of the crystal class (sec. 2.3.3).

Example Continuing on the example above, let us consider the special posi-
tion marked 8 b on page 633 of the ITC, which has a point group symmetry 3̄
(page 777). The order of this point group is 6. Consequently, the multiplicity
of the 8 b special position is 48/6 = 8, which is indicated by the number 8 in
8 b.

3.2.2 Companions of special points of symmetry vs points of equal

symmetry

Let’s look again carefully at fig. 2 to illustrate a very important but rather
subtle point. Fig. 2 shows a typical grid of symmetry elements for one of
the simplest space groups (P 2). One can see that there are four 2-fold axes
in the unit cell, one at the origin, two on the centre of the edges and one
in the centre of the cell (it is easy to see that there are 4 axes by shifting a
little bit the origin of the unit cell). Points on these 2-fold axes have the same
point group symmetry, i.e., 2. Could these points be companions of each
other in the sense illustrated above? The answer is clearly no. In fact, the
crystal class is also 2 and the number of nodes in the basis is 1 (primitive
lattice), so the multiplicity of points on 2-fold axes is 1 — these points have
no companions.

This example illustrates an important point: there can be different special

points of symmetry in the unit cell having the same point-group symmetry,
without them being necessarily companions of each other We will illustrate
this later on the Escher symmetry drawings.
15
Consider a general point very close to our special position. The elements of the point
group of the special position leave it invariant, but will generate a number of general points
around it, equal to the order of the point group, say m. If the order of the crystal class is n
and the number of nodes in the basis is b, there will be a total of n × b such general points in
the unit cell, which merge in groups of m as the generic point is brought close to the special
position. We can conclude that the number of such groups is n/m × b, and this is the number
of the companions of the special point itself.

37
3.2.3 Notation for general and special positions: multiplicity, Wyckoff
letter and site symmetry

In the ITC Volume A, each space/wallpaper group entry lists the general and
all the special positions in the unit cell. Positions that are symmetry equivalent
to each other are listed in on the same line (or group of lines if there are many
such companions). We therefore often encounter non-equivalent sites with
the same multiplicity and the same site symmetry, each listed on a separate
line. In order to distinguish them from each other, each entry is given a unique
letter, known as Wyckoff letter. The letter a is assigned to the site with the
highest point-group symmetry and so on, until the general position (which
has always site symmetry 1) is assigned the highest letter. In the ITC Vol A,
positions are listed listed in order of decreasing multiplicity. The structure of
the positions entries is as follows:

• Multiplicity (see sec. 3.2.1).

• Wyckoff letter (see here above). When referring to a site, e.g., in a paper,
one always uses both the multiplicity and the Wyckoff letter, as in 2 b or
8 j.

• Site symmetry (see sec. 2.3.3). The notation for the site symmetry is
slightly modified from that of standard point group notation16 .

• Coordinates of the equivalent positions. Note that these coordinates will not
necessarily fall within the unit cell (primitive or conventional). To shorten
notation, the minus sign is replaced by a bar above the coordinate (e.g.,
x̄, ȳ, z ≡ −x, −y, z).

• In the case of non-primitive lattices, the centring vectors are listed at the
top of the list of coordinates. For example, I-centred space groups
would list (0, 0, 0)+ and ( 21 , 12 , 21 )+, meaning that the number of equiv-
alent positions in the conventional unit cell is to be doubled by adding
the two vectors to each position.
16
The main difference is that the site symmetry symbol often contains dots (.) to indicate
missing symmetry operators. As you will recall, the site symmetry group is a subgroup of
the crystal class, so some operators in the crystal class may be missing in the site symmetry
group. These are indicated with dots. Sometimes there are additional symbols that would
have been implied in the crystal class but are no longer so because some other operator is
missing. For example, in space group P 4̄c2 (No. 116 — crystal class 4̄m2), 2 a and 2 b sites
have both site symmetry 2.22 ≡ 222. The two symmetry operators after the dots, at 90◦ from
each other, would have been equivalent (and therefore are not listed) in 4̄m2 but are no longer
equivalent in 222

38
3.3 Summary: Understanding the International Tables of Crys-
tallography

The ITC have been designed to enable scientists with minimal understanding
of group theory to work with crystal structures and diffraction patterns. Most
of the ITC entries, except for the graphical symbols that we have already
introduced, are self-explanatory (see fig. 9 and 10). In the lecture, we will
discuss the case of Ag2 O2 , a simple oxide that crystallise in space group
P 21 /c (no. 14) — see fig. 11 .

By using the ITC, one can (amongst other things):

• Given the coordinates of one atom, determine the number and coordinates
of all companion atoms. This greatly simplifies the description of crystal
structures.

• Determine the local point-group symmetry around each atom, which is the
same as the symmetry of the electron density and is connected to the
symmetry of the atomic wavefunction. This is particularly important for
certain types of spectroscopy.

• Determine the crystal class. Crystal classes are associated with the pres-
ence or absence of certain macroscopic properties, such as optical ac-
tivity, piezoelectricity, ferroelectricity etc. (see later part of this course).

39
 IT  numbering  
SG  symbols  (Hermann-­‐Mauguin  nota6on)   Crystal  class  (point  group)  

SG  symbol  (Schoenflies  nota6on)  

Crystal  system  

SG  diagrams  
projected  along  the  3  
direc6ons  
b  is  orthogonal  
to  a  and  c.  
 Cell  choce  1  
here  means  that  
the  origin  is  on  
an  inversion  
centre.  

A  generic  posi6on  
and  its  equivalent.    
Dots  with  and  
without  commas  are  
related  by  an  
improper  operator  
(molecules  located  
here  would  have  
opposite  chirality)  

Asymmetric  unit  cell:  

the  smallest  part  of  
the  paGern  that  
generates  the  whole  
paGern  by  applying  
all  symmetry  
A  symmetry  operator  notated  as  follows:    a  2-­‐fold  axis  with  an   operators.  
associated  transla6on  of  ½  along  the  b=axis,  i.e.,  a  screw  axis.    This  
is  located  at  (0,y,¼)  (see  SG  diagrams,  especially  top  leO).    Only  a  
subset  of  the  symmetry  operators  are  indicated.  All  others  can  be  
obtained  by  composi6on  with  transla6on  operators.  

Figure 9: Explanation of the most important symbols and notations in the

Space Group entries in the International Tables. The example illustrated here
is SG 14: P 21 /c .

40
 Refers  to  numbering  of  operators  on  previous  page  

Group  generators  

Primi4ve  transla4ons  

Equivalent  posi4ons,  obtained  by  

applying  the  operators  on  the  previous  
page  (with  numbering  indicated).    
Reflects  normal  form  of  operators.  

4  pairs  of  equivalent  inversion  

centres,  not  equivalent  to  
centres  in  other  pairs  (not  in  the  
same  class).    They  have  their  
own  dis4nct  Wickoff  leEer.  

Reflec4on  condi4ons  for  each  class  of  reflec4ons.    

General:    valid  for  all  atoms  (i.e.,  no  h0l  reflec4on  
will  be  observed  unless  l=2n;  this  is  due  to  the  c  
glide).    Special:    valid  only  for  atoms  located  at  
corresponding  posi4ons  (leI  side),  i.e.,  atoms  on  
inversion  centres  do  not  contribute  unless  k+l=2n  

Figure 10: IT entry for SG 14: P 21 /c (page 2).

41
Figure 11: Crystal structure data for silver (I,III) oxide (Ag2 O2 ), space group P 21 /c (no. 14).
Data are from the Inorganic Crystal Structures Database (ICSD), accessible free of charge
from UK academic institutions at http://icsd.cds.rsc.org/icsd/.

42
4 Lecture 4 — Symmetry in 2 dimensions: wallpaper
groups.

It is extremely useful to apply what we have learned about space groups to

the much simpler case of wallpaper groups in 2D. First of all, compared with
the 3D case, in 2D there is a much smaller number of point groups (10 vs 32)
and wallpaper groups (17 vs 230), and the latter can essentially all be built
by hand (see below). Secondly, wallpaper groups can be employed to anal-
yse the symmetry of real 2D periodic patterns, including the beautiful Escher
symmetry drawings (see course web site), in a way that is much simpler and
more intuitive than the 3D analogues.

4.1 Wallpaper vs space group symmetry

4.1.1 Bravais lattices in 2D

In 2D, there are 5 types of lattices: oblique, p-rectangular, c-rectangular,

square and hexagonal (see Fig. 12). In the c-rectangular lattice, no primi-
tive cell has the full symmetry of the lattice, as in the 3D centred lattices. It is
therefore convenient to adopt a non-primitive centred cell having double the
area of the primitive cell. As in the 3D case, the origin of the unit cell is to a
large extent arbitrary. It is convenient to choose it to coincide with a symmetry
element.

4.1.2 Rotations in 2D

Each rotation in 2D may correspond to either a proper rotation or an improper

rotation in the “embedding” 3D space. For example, the 2D inversion, techni-
cally a proper rotation because the determinant of the corresponding matrix
is +1, may correspond to either the 3D inversion (improper) or the 3D rotation
about z (proper). The convention is to consider rotations about z as proper
operators, and reflections about lines in the planes as improper operators
(mirrors). In summary:

• Rotations about z, symbols 2, 3, 4, 6

• Reflections in the plane, symbol m.

43
 Oblique  

p c

Rectangular  

Square   Hexagonal  

Figure 12: The 5 Bravails lattices in 2 dimensions.

4.1.3 Roto-translations in 2D

There is a single roto-translation operator in 2D, the glide line (symbol g),
which is the roto-translation corresponding to the mirror line m. As in the 3D
case, twice a glide translation is always a lattice translation: in fact, if one ap-
plies the glide operator twice as in g ◦ g, one obtains a pure translation (since
the two mirrors cancel out), which therefore must be a lattice translation.

4.2 Construction and identification of the 17 wallpaper groups

4.2.1 Point groups in 2D

The 10 point groups in 2D, which correspond to both 2D crystal classes and
2D site symmetries, are listed in the middle column of tab. 3. They are a
subset of the 32 3D point groups, and can be thought as those 3D point

44
groups that do not change the value of the z coordinate.

4.2.2 Wallpaper groups

The 17 Plane (or Wallpaper) groups describing the symmetry of all periodic
2D patterns are listed in tab. 3, right column, next to their crystal class. As
ih the 3D case, their symbols are obtained from the crystal class symbol by
replacing systematically rotations with roto-translations. However, since there
is a single roto-translation operation in 2D (g, which can only replace m), the
procedure is straightforwards and is shown in tab. 3. One often obtains
the same wallpaper group twice with a different symbol, corresponding to an
exchange of axes or by a different priority convention. Can you explain why,
for example, cg is not distinct from cm? (hint: look at set of parallel symmetry
element in the wallpaper group representation — see web site).

Table 3: The 17 wallpaper groups. The symbols are obtained by combining

the 5 Bravais lattices with the 10 2D point groups, and replacing g with m
systematically. Strikeout symbols are duplicate of other symbols.

crystal system crystal class wallpaper groups

1 p1
oblique
2 p2
m pm, cm,pg, cg
rectangular
2mm p2mm, p2mg (=p2gm), p2gg, c2mm, c2mg, c2gg
4 p4
square
4mm p4mm, p4gm, p4mg
3 p3
hexagonal
3m1-31m p3m1, p3g1, p31m, p31g
6 p6
6mm p6mm, p6mg, p6gm, p6gg

4.2.3 Identification of wallpaper groups

When observing a 2D periodic pattern, the decision three in fig. 13 can be

used to identify the corresponding wallpaper group. It is instructive (and quite
fun) to apply the decision tree to the Escher symmetry drawings (see web
site).

45
 c2mm
p2mm
pm
p2 p2gg
cm
Has rotations
p4gm p4mm p31m off mirros?
p2mg pg
p3m1
p4 p1
p3 Has glides?

respectively. Diamonds are branching points.

Axes on Has glides?

46
p6 p6mm mirrors? Has orthogonal
All axes
mirrors?
on mirrors? Has glides?

Has mirrors? Has mirrors? Has mirrors? Has mirrors? Has mirrors?

6 4 3 2 1
Axis of highest order

identify the axis of highest order. Continuous and dotted lines are ”Yes” and ”No” branches,
Figure 13: Decision-making tree to identify wallpaper patterns. The first step (bottom) is to
5 Lecture 5 — Reciprocal lattice and reciprocal space

5.1 Basic definitions

• The reciprocal lattice is a constructed from an infinite set of vectors

chosen in such a way that the dot product of any of these vectors
with any of the “real” lattice vectors is an integer multiple of 2π. We
recall that the real lattice vectors are the position vectors of the nodes of
the real lattice with respect to an arbitrary “origin” node. Note that since
the dimensions of vectors of the real lattice is (length), the dimension of
the reciprocal lattice vectors is (length−1 )

• The reciprocal lattice vectors are constructed as a linear combination

of reciprocal basis vectors (a.k.a. dual basis vectors) with integer
coefficients. As you have already seen in previous years, reciprocal
lattice vectors are orthogonal to families of nodal planes in real space,
and in this context, their integer coefficients are known as Miller indices
(usually h, k and l).

• The reciprocal space is a space of vectors defined by linear combina-

tions of reciprocal basis vectors with dimensionless components.

5.2 The reciprocal lattice

5.2.1 Dual basis in 3D

In 3 dimensions, you have already seen the following to calculate the dual
basis vectors, which makes use of the properties of the vector product:

a2 × a3
b1 = 2π (29)
a1 · (a2 × a3 )
a3 × a1
b2 = 2π
a1 · (a2 × a3 )
a1 × a2
b3 = 2π
a1 · (a2 × a3 )

With this definition of the dual basis, we have

ai · bj = 2πδij (30)

47
Note that

1/2
v = a1 · (a2 × a3 ) = abc 1 − cos2 α − cos2 β − cos2 γ + 2 cos α cos β cos γ
(31)

is the unit cell volume.

In crystallographic textbooks, the dual basis vectors are often written as a∗ ,

b∗ and c∗ .

Using this definition of the dual basis, the dot product between vectors in real
space (expressed on the real-space basis), and vectors in reciprocal space
(expressed on the dual basis) is:

X
q · v = 2π qi v i (32)
i

For this definition of dual basis to conform to our original definition of recipro-
cal lattice, we must have for every h, k, l integers:

(n1 a1 + n2 a2 + n3 a3 ) · (hb1 + kb2 + lb3 ) =

X
= 2π(h n1 + k n2 + l n3 ) = 2π hi ni = 2πm (33)
i

where m is an integer. We can see that this is always true if n1 , n2 , n3 are also
integers, that is, if a1 , a2 , a3 is a primitive basis. It can be easily shown that
all equivalent primitive bases in real space define equivalent primitive dual
bases and reciprocal space lattices 17 .

On the contrary, if a1 , a2 , a3 is a conventional basis, some vectors expressed

as (hb1 + kb2 + lb3 ) will not belong to the reciprocal lattice.

5.2.2 Centring extinctions

If a conventional real-space basis is used to construct the dual basis, only

certain reciprocal-lattice vectors will yield a 2πm dot product with all
real-lattice vectors — a necessary (but not sufficient, see below) condition to
17
This is due to the fact that the transformation matrix of the dual basis is the inverse matrix
of the transformation matrix of the real-space basis. In this course, I will shy away from a full
treatment of crystallographic coordinate transformations — a huge subject in itself, but one
that at this level does not add much to the comprehension of the subject.

48
observe Bragg scattering at these points. These reciprocal-lattice vectors are
exactly those generated by one of the corresponding primitive bases.
All other Miller indices will not generate reciprocal-lattice vectors. There will
never be scattering intensities at these points, which are therefore said to be
extinct by centring. This is illustrated with an example in fig. 14 (in 2D). One
can see that the conventional reciprocal-space unit cell (associated with the
dual basis a∗c and b∗c ) is smaller than the primitive reciprocal-space unit cell
(a∗p and b∗p ), and therefore generates more reciprocal lattice points. Bragg
scattering is never observed at these points, for the reasons explained here
below — we will say that they are extinct by centering.

• Each non-primitive lattice type has centring extinctions, which can be ex-
pressed in terms of the Miller indices hkl (see table 4).

Table 4: Centering extinction and scattering conditions for the centered lat-
tices. The “Extinction” columns lists the Miller indices of reflections that are
extinct by centering, i.e., are “extra” RLV generated as a result of using a
conventional basis instead of a primitive one. The complementary “Scatter-
ing” column corresponds to the listing in the International Tables vol. A, and
lists the Miller indices of “allowed” reflections. “n” is any integer (positive or
negative).

Lattice type Extinction Scattering

P none all

A k + l = 2n + 1 k + l = 2n
B h + l = 2n + 1 h + l = 2n
C h + k = 2n + 1 h + k = 2n

F k + l = 2n + 1 or k + l = 2n and
h + l = 2n + 1 or h + l = 2n and
h + k = 2n + 1 h + k = 2n

I h + k + l = 2n + 1 h + k + l = 2n

R −h + k + l = 3n + 1 or −h + k + l = 3n
−h + k + l = 3n + 2

• The non-exctinct reciprocal lattice points also form a lattice, which is natu-
rally one of the 14 Bravais lattices. For each real-space Bravais lattice,
tab. 5 lists the corresponding RL type.

49
 Real  Space  
Conven,onal  u.c.   Primi,ve  u.c.  

bp*  

bc   bp  
bc*   ap  
ac*  
ac   ap*  

Reciprocal  Space  

bp*  

bc*  
ac*  

ap*  

Figure 14: C-centred cells: construction of reciprocal basis vectors on the real-space lattice
(top) and corresponding reciprocal lattice (bottom). One can see that the primitive reciprocal
basis vectors generate fewer points (closed circles) than the conventional reciprocal basis
vectors (open an closed points). Only closed points correspond to observed Bragg peaks, the
others are extinct by centring.

50
Table 5: Reciprocal-lattice Bravais lattice for any given real-space Bravais
lattice (BL).

Crystal system Real-space BL Reciprocal-space BL

Triclinic P P

Monoclinic C C

Orthorhombic P P
A or B or C A or B or C
I F
F I

Tetragonal P P
I I

Trigonal P P
R R

Hexagonal P P

Cubic P P
I F
F I

5.2.3 Dual basis — alternative definition without cross product

Let us assume a basis vector set ai for our vector space as before, and let us
consider the following set of new vectors.

X
bi = 2π ak (G−1 )ki (34)
k

From Eq. 7 follows:

Gik (G−1 )kj = 2πδij

X X
ai · bj = ai · 2π ak (G−1 )ki = 2π (35)
k k

This is exactly the same as eq. 30, so this new definition of the dual basis
must be equivalent to the one given in eq. 29.

51
 Recap of the key formulas for the dual basis

• From direct to dual bases (eq. 34)

X
bi = 2π ak (G−1 )ki
k

• Dot product relation between the two bases (eq. 35)

ai · bj = 2πδij

• Dot product between vectors expressed on the two different bases (eq. 32):
X
q · v = 2π qi v i
i

5.2.4 Dot products in reciprocal space

As we shall see later, it is very useful to calculate the dot product between two
vectors in reciprocal space. This can be tricky in non-Cartesian coordinate
systems. A quick way to do this is to determine the reciprocal-space metric
tensor, which is related to the real-space one. The reciprocal-space metric
tensor is G̃ = (2π)2 G −1 , so that, for two reciprocal-space vectors q and r:
X
q·r= G̃ij qi rj (36)
i,j

5.2.5 A very useful example: the hexagonal system in 2D

• The hexagonal system in 2D has a number of important applications in con-

temporary solid-state physics problems, particularly for carbon-based
materials such as graphene and carbon nanotubes.

• By crystallographic convention, the real-space basis vectors form an angle

a ∧ b of 120◦ (this is also true in 3D).

• The real-space metric tensor is therefore:

 
2 1 −1/2
G=a (37)
−1/2 1

• From eq. 36, we can find the reciprocal-space metric tensor :

52
 (2π)2 4
 
1 1/2
G̃ = (38)
a2 3 1/2 1

• It follows, for example, that the length of a vector in reciprocal space is

given by:

r
2π 4 2
q= (h + hk + k 2 ) (39)
a 3

5.3 Fourier transform of lattice functions

As you have already seen in previous years (see also Appendix I), the Fourier
transform of a function f (r) (real or complex) with the periodicity of the lattice
can be written as:

1
Z
i
X P
−2πi
F (q) = 3 e i qi n dxf (x)e−iq·x
(2π) 2 ni u.c.

v0
Z
i i
X P P
−2πi
= 3 e i qi n dxi f (xi )e−2πi i qi x (40)
(2π) 2
ni u.c.

where v0 is the volume of the unit cell

The triple infinite summation in ni = nx , ny , nz is over all positive and negative

integers. The qi = qx , qy , qz are reciprocal space coordinates on the dual
basis. The xi = x, y, z are real-space crystallographic coordinates (this is
essential to obtain the qi ni term in the exponent). The integral is over one
unit cell. F (q) is non-zero only for q belonging to the primitive RL. In fact, if
q belongs to the primitive reciprocal lattice, then by definition its dot product
to the symmetry lattice translation is a multiple of 2π, the exponential factor is
1 and the finite summation yields N (i.e., the number of unit cells). If q does
not belong to the primitive reciprocal lattice, the exponential factor will vary
over the unit circle in complex number space and will always average to zero.

It is the periodic nature of f (r) that is responsible for the discrete nature of
F (q).

5.4 Atomic-like functions

Although the result in eq. 51 is completely general, in the case of scattering

experiments the function f (x) represents either the density of nuclear scat-

53
tering centres (neutrons) or the electron density (x-rays, electrons). These
functions are sharply pointed around the atomic positions, and it is therefore
convenient to write:

X
f (x) = fj (x) (41)
j

where the summation if over all the atoms in the unit cell, and fj (x) is (some-
what arbitrarily) the density of scatterers assigned to a particular atom (see
next lectures for a fully explanation). With this, eq. 51 becomes:

v0
Z
i i i −xi )
P P
−2πi
dxi fj (xi )e−2πi
X P X
−2πi i qi n i qi xj i qi (x
F (q) = 3 e e j =
(2π) 2
ni j u.c.

v0 i
P i
e−2πi
X P X
= 3 e−2πi i qi n i qi xj fj (q) (42)
(2π) 2
ni j

When evaluated on reciprocal lattice points, the rightmost term is now the
structure factor Fhkl .

5.5 Centring extinctions

In particular, we can show that those “‘conventional”’ RLV that we called ex-
tinct by centring are indeed extinct, that is, F (q) = 0. For example, in the
case of an I-centrel lattice, we have:

X  h k l

Fhkl = e−2πi(hx+ky+lz) 1 + e−2πi( 2 + 2 + 2 fj (q) (43)
j

and this is = 0 for h + k + l 6= 2m, which precisely identifies the non-primitive

RL nodes.

5.6 Extinctions due to roto-translations

Similarly, the presence of roto-translation operators (glide planes and screw

axes) produces extinction conditions, which, however, do not affect whole RL
sub-lattices but only certain sections of it. In the case of glides, there are
planes of hkl’s in which some of the reflections are extinct. In the case of
screws, there are lines of hkl’s in which some of the reflections are extinct.

54
Here, I will show only one example, that of a glide plane ⊥ to the x axis and
with glide vector along the y direction, which generates the two companions
x, y, z and −x, y + 12 , z. We have:

X  k

Fhkl = e−2πi(ky+lz) e−2πihx + e−2πi(−hx+ 2 ) fj (q) (44)
j

We can see that, within the plane of reflections having h = 0, all reflections
with k 6= 2m are extinct.

5.7 The symmetry of |F (q)|2 and the Laue classes

• It is very useful to consider the symmetry of the RL when |F (x)|2 is asso-

ciated with the RL nodes. In fact, this corresponds to the symmetry of
the diffraction experiment, and tells us how many unique reflections we
need to measure.

• Translational invariance is lost once |F (x)|2 is associated with the RL nodes.

• Let R be the rotational part and t the translational part of a generic symme-
try operators. One can prove that

N
Z
−1
F (q) = 3 d(x)f (x)e−i(R q)·x −iq·t
e = F (R−1 q)e−iq·t (45)
(2π) 2 u.c.

Eq. 45 shows that the reciprocal lattice weighed with |F (q)|2 has the full
point-group symmetry of the crystal class.

• This is because the phase factor e−iq·t clearly disappears when taking the
modulus squared. In fact, there is more to this symmetry when f(x) is
real, i.e., f (x) = f ∗ (x): in this case

N
Z
F ∗ (q) = 3 dxf ∗ (x)eiq·x (46)
(2π) 2 u.c.

N
Z
= 3 dxf (x)eiq·x = F (−q)
(2π) 2 u.c.

• Consequently, |F (q)|2 = F (q) F (−q) = |F (−q)|2 is centrosymmetric. As

we shall shortly see, the lattice function used to calculate non-resonant
scattering cross-sections is real. Consequently, the |F (q)|2 -weighed RL

55
 (proportional to the Bragg peak intensity) has the symmetry of the crys-
tal class augumented by the center of symmetry. This is necessarily
one of the 11 centrosymmetryc point groups, and is known as the Laue
class of the crystal.

Fridel’s law

For normal (non-anomalous) scattering, the reciprocal lattice weighed with |F (q)|2 has the full
point-group symmetry of the crystal class supplemented by the inversion. This symmetry is
known as the Laue class of the space group.
In particular, for normal (non-anomalous) scattering, Fridel’s law holds:

|F (hkl)|2 = |F (h̄k̄¯l)|2 (47)

Fridel’s law is violated for non-centrosymmetric crystals in anomalous conditions.
Anomalous scattering enables one, for example, to determine the orientation of a polar crystal
or the chirality of a chiral crystal in an absolute way.

6 Appendix I: Fourier transform of lattice functions

In this section, we will consider a generic real or complex function f (r) de-
fined over the real space, r being a position vector from an appropriately
defined origin). We assume that f (r) has the symmetry properties defined by
one of the 230 space groups. We will calculate the Fourier transform of this
function, F (q), over the whole space. As we shall see in the next lectures,
F (q) corresponds to the diffraction structure factor. We have:

1
Z
F (q) = 3 drf (r)e−iq·r (48)
(2π) 2

where the integral extends to the whole space. We now exploit the lattice
periodicity of the function f (r), which we can express by writing r = r0 + x
and
f (r0 + x) = f (x) (49)

The r0 are the symmetry translation vectors, and x is a position vector within
the first unit cell, i.e., x, y, z < 1. We can also decompose the integral in Eq.
48 in integrals over the unit cells:

56
 1 XZ
F (q) = 3 dxf (x)e−iq·(r0 +x) (50)
(2π) 2 r0 u.c.

1 X Z
−iq·r0
= 3 e dxf (x)e−iq·x
(2π) 2 r0 u.c.

where the integral is now over a single unit cell. We now introduce a set of
coordinates that are appropriate for the symmetry18 and recall that in these
coordinates the symmetry translation vectors are expressed as [ni ], i.e., a set
of three integers. By using the relation dx = dxdydz(a × b) · c = v0 dxdydz, v0
being the volume of the unit cell, eq. 50 becomes:

v0
Z
i i
X P P
F (q) = 3 e−2πi i qi n dxi f (xi )e−2πi i qi x (51)
(2π) 2
ni u.c.

The sum is now over all the symmetry translations, i.e., over all the positive
and negative values of the [ni ]. The [ni ] are integers or simple fractions (for
centering translations). We will perform the infinite summation by summing
over a finite number N real-lattice vectors first , and then letting N → ∞. The
following statements is now clear by inspecting Eq. 51:

F (q) is non-zero only for q belonging to the primitive RL.

In fact, if q belongs to the primitive reciprocal lattice, then by definition its

dot product to the symmetry lattice translation is a multiple of 2π, the expo-
nential factor is 1 and the finite summation yields N (i.e., the number of unit
cells). Conversely, if q does not belong to the primitive reciprocal lattice, the
exponential factor will vary over the unit circle in complex number space and
will always average to zero. In particular, F (q) is zero for the conventional
RLV that are extinct by centering (as we anticipated — this explains the
terminology ”extinction” we just introduced). For non-extinct RL vectors, the
infinite summation yields ∞.

7 Appendix II: “Real” crystal structures

Having discussed at length the symmetries of periodic “patterns” in 2 and

3 dimensions, we will devote the last part of this lecture to looking at “real”
18
In this section, it should become absolutely clear why we do not use Cartesian coordi-
nates.

57
crystal structures. This is in itself a vast subject that cannot be exhausted
in such a short space. An interesting set of lectures devoted to the subject
can be found in [?]. It is also worth pointing out to the interested student
the existence of several very useful Crystal Structure Databases. The Inor-
ganic Crystal Structures Database (ICSD), freely accessible on-line from the
UK [?], can be searched for names, chemical formulas, crystallographic data
and more, to display the resulting crystal structures in 3D and even to plot
their powder diffraction patterns. The Cambridge Structural Database is the
corresponding source for small-molecule structures. Here, we will outline a
few basic principles that should provide a starting point to understand “real”
crystal structures.

7.1 Cohesive forces in crystals — atomic radii

A number of different forces contribute to the cohesion of crystals, including:

• The Coulomb interaction between charged ions.

• Chemical bonding and metallic bonding.

• The Van der Waals (dipole-dipole) interaction.

• Hydrogen bonding.

These forces, which often coexist within the same crystal structure, are of
very different strength. Another crucial difference is the directionality of
these forces. Chemical bonding (both ionic and covalent) is usually strongly
directional, and leads to the formation of specific coordination polyhedra
(e.g., octahedra, tetrahedra) within the crystals. Conversely, most other inter-
actions are poorly directional.

One useful way to understand many crystal structures, particularly those of

inorganic compounds of greater interest for physicists, is that of considering
them as packings of spheres of different sizes. Within this very simplistic
picture, each ion is characterised by a radius. Atomic radii are not completely
unique to each species, but vary depending on several factors:

• The valence state of the ion.

• The spin state of the ion.

• The number of neighbours (coordination number).

58
• Whether the bonding is ionic or covalent.

The standard reference for covalent and ionic radii was compiled by R.D.
Shannon and can be found in19 . Several versions of this table can be found
on line.

Over most of the periodic tables, ionic and covalent radii vary between 0.5 Åand 2 Å. Typ-
ical interatomic distances are therefore of the order of 1.5–2.5 Å. This sets the length-
scale of the probes (X-rays, neutrons, electrons) that can be most profitably used to
study these structures.

7.2 Close-packed structures

When all the “spheres” are of equal size and the interactions between them
are not strongly directional, the most common arrangement is one of the
close packed structures (fig. 15):

CCP i.e., Cubic Close-Packed, which has a face-centered cubic (FCC) lat-
tice (space group F m3̄m). Many metals, including all those of the Cu
and Ni groups, adopt this structure.

HCP Hexagonal Close-Packed, which has a hexagonal lattice with two atoms
per unit cell (space group P 63 /mmc). Metals such as Co, Zn, Cd, Hg,
Mg and others adopt this structure.

Several metals, including Fe, Cr and

its group, V and its group and all
the alkaline metals adopt the BCC
(Body Centered Cubic) structure
— space group Im3̄m, which is not
close-packed.

Close-packed and BCC structures

are also adopted by much more
complex systems — for instance C60
Figure 15: The close-packed struc-
(“Buckyballs”) and even viruses (fig. tures of rigid spheres: HCP (left) and
16) — clearly with much larger inter- CCP (right).
sphere distances. Here we have
roughly spherical objects with strong internal bonding, which are weakly
bonded among themselves.
19
R. D. Shannon, Acta Cryst. A32 751-767(1976)

59
7.3 Packing spheres of different radii

Many simple binary or ternary com-

pounds are made of ionic species
with different radii. In these cases,
their crystal structures can often be
thought of as being close-packed ar-
rangements of the larger spheres,
with the smaller spheres located in
the “interstices” or “vacancies” be-
tween the larger spheres. Both Figure 16: Simple arrangements of
CCP and HCP structures have va- complex objects: (left) the CCP struc-
cancies of this type, surrounded by ture of C60 (“Buckyballs”) and (right)
the BCC structure of the foot-and-
four spheres (tetrahedral vacan-
mouth virus.
cies) or six spheres (octahedral va-
cancies). Because of the geometry of the vacancies, this structural arrange-
ment is suitable for ions with strongly directional bonding. When strong di-
rectional bonding is present, compounds with ions with similar radii and even
mono-atomic compounds can adopt these structure. Among the structures
that can be described in this way are:

The NaCl structure (space group F m3̄m) where Na fills all the octahedral
holes of the CCP structure.

The fluorite structure (prototype compound CaF2 , space group F m3̄m), where
the F atoms fill all the tetrahedral holes of the CCP structure (fig. 17).

The zinc blende structure (prototype compound ZnS, space group F 43̄m),
where the Zn ions fill half of the tetrahedral holes of the CCP structure.

The perovskite structure (prototype compound CaTiO3 , space group P m3̄m).

In this interesting ternary example, the CCP array is formed by both
Ca2+ (positively charged) and O2− (negatively charged). The smaller
Ti ion fills a quarter of the octahedral vacancies.

The corundum structure (Al2 O3 , space group R3̄c). Here, the oxygen ions
form a HCP structure, and the much smaller Al ions fit into 2/3 of the the
octahedral vacancies (1/3 of the vacancies are empty). The ilmenite
(FeTiO3 , space group R3̄) is a variant with two metal ions instead of
one.

The diamond structure is adopted, among others, by C and Si. It is identi-

cal to the zinc blende structure but with two identical, strongly-bonded

60
 atoms.

Figure 17: Three cubic structures obtaining by inserting ions in the “inter-
stices” of the CCP structure. Left the fluorite structure ; center the zinc
blende structure; right the perovskite structure.

7.4 Framework structures

Many crystal structures cannot be

simply thought of in terms of close
packing. One notable example is
given by framework structures —
structures built out of very rigid poly-
hedra (most often tetrahedra) with
rather “flexible” connections to each
other. Framework structures are
low-density structures, and can of-
ten collapse rather easily to higher-
density forms upon application of
pressure.
Figure 18: The structure of zeo-
The structure of quartz (SiO2 ) con- lite β (space group P 42 /mmc). Note
sists of corner-sharing SiO4 tetrahe- the complex framework of SiO4 tetra-
hedra, defining a large cavity in the
dra so that each Si is bonded to
middle of the unit cell. Many sili-
four oxygens, and each oxygen is cate and alumino-silicate zeolites ex-
bonded to two silicon atoms. The ist, both natural and synthetic.
resulting structure forms an open
three-dimensional framework, and it is quite flexible, so that different crys-
tal variants exist (α- and β-quartz, crystobalite, trydimite etc.) When cooled
rapidly, the quartz structure is unable to “choose” between these variants
and forms a glass. An even more extreme example of silicate framework
structure is provided by zeolite (SiO2 , fig. 18), where the tetrahedral frame-

61
work encompasses large cavities. Zeolite is the prototype of a large family
of silicates and alumino-silicates, collectively known as “zeolites”, which have
wide-ranging applications in catalysis.

7.5 Layered structures

Many crystal structures have a pronounced 2-dimensional character, with

strong covalent or ionic bonding in 2 dimensions and weaker (typically Van
der Waals) bonding in the third. A well known example of this is graphite
(space group P 63 /mmc with two atoms per unit cell). Due to the weak inter-
layer forces, the layers can “slip” onto each other, so that structures of this
types are often employed as lubricants. Other examples of this kind are pro-
vided by the clays, such as vermiculite (fig. 19), and by the delafossite family
(prototype CuFeO2 , space group R3̄m or P 63 /mmc). Less extreme examples
of 2D structures are provided by the high- Tc superconducting cuprates (fig
20).

Figure 19: The structure of the vermiculite clay (chemical formula A3 B4 O10 ·
(H2 O)n , with A=Mg, Fe, Al, B=Al, Si; space group C2/m) is highly 2-
dimensional. The A site forms triangular layers with formula AO2 , connected
to “rings” of BO2 tetrahedra. These layers are widely separated and weakly
interacting, and, as typical of clays, can accommodate large amounts of
rather disordered water molecules.

7.6 Molecular structures

All the structures we have defined up to this point are built of infinite “net-
works” of atoms, either in 3D or in 2D. By contrast, molecular structures are

62
Figure 20: The structure of the 90 K-high-Tc superconductor YBa2 Cu3 O7
or YBCO is also 2-dimensional, but with a less pronounced 2D structural
character compared to clays. The central ion, Y, is ionically bonded to oxygen,
so the structure does not exfoliate like that of graphite. Nevertheless, the
electronic structure is highly 2D. Note that Cu exists both in square-pyramidal
(“planes”) and in square-planar (“chains”) coordinations.

Figure 21: The molecular crystal structure of aspirin. The individual

molecules are easily identifiable, and are linked to each other by hydrogen
bonds (dotted lines).

built out of well defined “molecules”, with strong internal covalent bonding but
weakly interacting with each other. A simple example is the structure of ice,
with covalent bonding within the H2 O molecule and weak hydrogen bonding
between molecules. Orinary ice is known as “ice 1h”, and has space group
P 63 /mmc. However, due to the particular geometry of the molecules, ice
is highly polymorphic as a function of temperature and pressure, with 15
known different crystallographic structures being known to date. Molecular
structures are adopted by most small molecules (such as drugs) and macro-
molecules (such as proteins). The molecule itself has rigid components (such
as benzene rings) connected to each other by “joints” having some degree
of flexibility. Therefore, the same molecule can often adopt different crys-
tal structures (polymorphism), having different molecular configurations and
packing of different molecules within the unit cell.

63