Solid State Chemistry

Topic 15A Crystal structure
➤ Why do you need to know this material?

Crystalline solids are important in many technologies, and
to be able to account for their mechanical, electrical, opti-
cal, and magnetic properties you need to understand their Lattice point
microscopic structures.
Structural motif
➤ What is the key idea?
The regular arrangement of the atoms in periodic crystals
can be described in terms of unit cells.
Figure 15A.1 Each lattice point specifies the location of a
➤ What do you need to know already? structural motif (e.g. a molecule or a group of molecules). The
space lattice is the entire array of lattice points; the crystal
Light use is made of some of the language used to describe
structure is the collection of structural motifs arranged according
symmetry (Topic 10A).
to the lattice.
The internal structure of a crystal is a regular array of its atoms,

ions, or molecules. The features of this regular array, such as the
details of the stacking pattern and its characteristic dimensions,
are a crucial aspect of the link between the structure and prop-
erties of the solid.
15A.1 Periodic crystal lattices

A periodic crystal is built up from regularly repeating ‘struc- Figure 15A.2 A unit cell is a parallel-sided (but not necessarily
tural motifs’, which may be atoms, molecules, or groups of rectangular) figure from which the entire space lattice can be
constructed by using only translations (not reflections, rotations,
atoms, molecules, or ions. A space lattice is the pattern formed
or inversions). In the two-dimensional case shown here, each
by lattice points representing the locations of these motifs
lattice point is shared by four neighbouring cells.
(Fig. 15A.1). More formally, a space lattice is a three-dimen-
sional, infinite array of points, each of which is surrounded in
an identical way by its neighbours. In effect, a space lattice is straight lines, and such unit cells are described as primitive
an abstract scaffolding for the crystal structure. In some cases (Fig. 15A.3). If each of the four points of a two-dimensional
there may be a structural motif centred on each lattice point, unit cell in Fig. 15A.2 is regarded as shared with its four neigh-
but that is not necessary: all that is necessary is that an iden- bours, then the cell has only one lattice point overall; this is a
tical structural motif is associated with each lattice point. The requirement for a unit cell to be described as primitive. The
solids known as quasicrystals are ‘aperiodic’, in the sense that same definition applies in three dimensions, where each of the
the space lattice, though still filling space, does not have trans- eight points of a primitive unit cell is shared by eight neigh-
lational symmetry. This Topic deals only with periodic crystals. bours, giving one lattice point overall. It is often more con-
A unit cell is an imaginary parallelepiped (parallel-sided venient to draw larger non-primitive unit cells that also have
figure) from which the entire space lattice can be constructed lattice points at their centres or on pairs of opposite faces. An
by purely translational displacements (Fig. 15A.2). The cell is infinite number of different unit cells can describe the same
commonly formed by joining neighbouring lattice points by lattice, but the one with sides that have the shortest lengths and
658 15 Solids
• A monoclinic unit cell has one twofold rotation axis

(Fig. 15A.6).
• A triclinic unit cell has no rotational symmetry, and typi-
cally all three sides and angles are different (Fig. 15A.7).
Table 15A.1 lists the essential symmetries, the elements that must
be present for the unit cell to belong to a particular crystal system.
In three dimensions there are only 14 distinct space lat-
tices. The unit cells of these Bravais lattices are illustrated in
Fig. 15A.8. It is conventional to portray the lattices by primitive
unit cells in some cases and by non-primitive unit cells in oth-
Figure 15A.3 A primitive unit cell, an example of which is shown
ers. The following notation is used:
by the shaded volume, has lattice points only at its vertices. If each
of the eight points is regarded as shared with its eight neighbours, • A primitive unit cell (P) has lattice points only at the corners.
the unit cell has only one lattice point overall.
• A body-centred unit cell (I) also has a lattice point at its
centre.
a
b c β
α
γ a a
c b c
b γ β C2
α
Figure 15A.4 The notation for the sides and angles of a unit cell.
Note that the angle α lies in the plane (b,c). Figure 15A.6 A unit cell belonging to the monoclinic system has
a twofold rotation axis (denoted C2 and shown in more detail in
the insert).
that are most nearly perpendicular to one another is normally
chosen. The lengths of the sides of a unit cell are denoted a, b,
and c, and the angles between them are denoted α, β, and γ
(Fig. 15A.4).
Unit cells are classified into seven crystal systems by noting
the rotational symmetry elements they possess. For example:
• A cubic unit cell has four threefold rotation axes pointing
to the corners of a tetrahedron, and passing through the
centre of the cube (Fig. 15A.5).
C3 Figure 15A.7 A triclinic unit cell has no axes of rotational symmetry.

C3
C3 C3 Table 15A.1 The seven crystal systems⋆
System Essential symmetries

Triclinic None
Monoclinic One C2 axis
Orthorhombic Three perpendicular C2 axes
Rhombohedral One C3 axis
Tetragonal One C4 axis
Hexagonal One C6 axis
Figure 15A.5 A unit cell belonging to the cubic system has four Cubic Four C3 axes in a tetrahedral arrangement
threefold rotation axes, denoted C3, arranged tetrahedrally. The ⋆
Cn denotes an n-fold rotation axis, in which identical structures are obtained after rota-
insert shows the threefold symmetry. tion by 360°/n about that axis.
15A Crystal structure 659
a a C2
Cubic P Cubic I Cubic F C2
a b
a a
β
c c
Figure 15A.9 The two-dimensional lattice used in Brief
illustration 15A.1; a (non-primitive) unit cell is indicated by the
Tetragonal P Tetragonal I Monoclinic P Monoclinic C shaded area. The lattice points at the corners of the unit cell are
related by rotations about the twofold rotation axes of symmetry
a b
shown; the lattice point at the centre is unaffected by these
operations.
c
Orthorhombic P Orthorhombic C Orthorhombic I Orthorhombic F

a a a
a
b a
Exercise
γ 120° a ≠90° E15A.1 The orthorhombic unit cell of NiSO4 has the dimensions a = 634 pm,
c α β c
b = 784 pm, and c = 516 pm, and the mass density of the solid is estimated as
3.9 g cm−3. Identify the number of formula units in a unit cell and calculate a
more precise value of the mass density.
Triclinic Hexagonal Trigonal R
Figure 15A.8 The 14 Bravais lattices. The points are lattice points,
and are not necessarily occupied by atoms. P denotes a primitive
unit cell (R is used for a trigonal lattice), I a body-centred unit
The identification of lattice
15A.2
cell, F a face-centred unit cell, and C (or A or B) a cell with lattice planes
points on two opposite faces. Trigonal lattices may belong to the
rhombohedral or hexagonal systems (Table 15A.1). The interpretation of the diffraction techniques that are used
to measure the size of unit cells and the arrangement of mol-
ecules within them makes use of the orientation and sepa-
• A face-centred unit cell (F) has lattice points at its corners
ration of planes that pass through the crystal (Topic 15B).
and also at the centres of its six faces.
Two-dimensional lattices are easier to visualize than three-
• A side-centred unit cell (A, B, or C) has lattice points at dimensional lattices, so in this discussion the concepts involved
its corners and at the centres of two opposite faces. in identifying lattice planes are introduced for two dimensions
For simple structures, it is often convenient to choose an atom be- initially, and then the results are extended by analogy to three
longing to the structural motif, or the centre of a molecule, as the dimensions. Note that lattice planes do not necessarily pass
location of a lattice point or the vertex of a unit cell, but that is not through lattice points.
a necessary requirement. Symmetry-related lattice points within
the unit cell of a Bravais lattice have identical surroundings.
15A.2(a) The Miller indices
Consider a two-dimensional rectangular lattice formed from a
Brief illustration 15A.1 unit cell of sides a and b (Fig. 15A.10). Each panel in the illus-
tration shows a set of evenly spaced planes that can be identified
The two-dimensional lattice shown in Fig. 15A.9 consists
of a rectangular array of lattice points, with one additional by considering, for each set, the plane lying closest to the origin
point at the centre of each rectangle; a (non-primitive) unit (but not passing through it) and then quoting the distances at
cell is indicated by the shading. This cell has twofold rotation which this plane intersects the a and b axes. These distances are:
axes passing through the mid-points of opposite sides of the (a) (1a,1b), (b) 12 a, 13 b , (c) (−1a,1b), and (d) (∞a,1b), with ∞
rectangle. Rotations about these axes interchange the lattice indicating that the plane is parallel to an axis and intersects it
points at the corners of the rectangle, but the lattice point at (notionally) at infinity. If it is agreed to quote distances along
the centre is not affected. It therefore follows that the lattice the axes as multiples of the corresponding dimensions of the
points at the corners are equivalent, but the lattice point at the unit cell, then these intersections can be expressed more simply
centre is distinct. as (1,1), 12 , 13 , (−1,1), and (∞,1), respectively. If the lattice in
660 15 Solids
a b
b
{110} {111}
a
(a) (b) a
c
{010} {111}
Figure 15A.11 Some representative planes in three dimensions

(c) (d)
and their Miller indices; the origin is indicated by the open circle.
Figure 15A.10 Some of the sets of equally spaced planes that can Note that the index 0 indicates that a plane is parallel to the
be drawn in a rectangular space lattice; the origin is indicated by corresponding axis, and that the indexing may also be used for
the open lattice point. The Miller indices {hkl} of each set of planes unit cells with non-orthogonal axes.
are: (a) {110}, (b) {230}, (c) { 110} , and (d) {010}.
Figure 15A.11 shows a three-dimensional representation
of a selection of planes, including one in a lattice with non-
orthogonal axes.
Fig. 15A.10 is the top view of a three-dimensional orthorhom-
bic lattice, all four sets of planes intersect the c axis at infinity.
Therefore, in the three-dimensional case the labels are (1,1,∞), 15A.2(b) The separation of neighbouring
12 , 13 , , (−1,1,∞), and (∞,1,∞). planes
The inconvenience of fractions and infinity in these labels The Miller indices are very useful for expressing the separation
can be avoided by specifying a plane by using its Miller indices, of neighbouring planes.
(hkl), where h, k, and l are the reciprocals of the intersection
distances along the a, b, and c axes, respectively. For example, How is that done? 15A.1 Deriving an expression for the
the plane 12 , 13 , has Miller indices (230). As will be seen, this
notation brings with it additional advantages. The Miller nota- separation of planes
tion has the following features: Consider the {hk0} planes of a square lattice built from a
unit cell with sides of length a (Fig. 15A.12). The separation
• Negative indices are written with a bar over the number,
between the lattice planes is equal to the perpendicular dis-
as in ( 110).
tance dhk0 from the (hk0) plane to the origin.
• The notation (hkl) denotes an individual plane. A set of Expressions for the sine and cosine of the angle φ are found
parallel planes with identical spacing is denoted {hkl}. by considering the sides of the two right-angle triangles shown
For example, in the figure:
dhk 0 hdhk 0 dhk 0 kdhk 0

sin cos
(a/h) a (a/k) a
Intersect axes at (a,b,∞c) 12 a, 13 b,c (−a,b,∞c) (∞a,b,∞c)
Remove cell dimensions (1,1,∞) 12 , 13 , (−1,1,∞) (∞,1,∞) a
Take reciprocals (1,1,0) (2,3,0) (−1,1,0) (0,1,0)
Express as indices (110) (230) ( 110) (010) (hk0)
Sets of parallel planes {110} {230} {110} {010}

a
a/k
A helpful feature to remember is that the smaller the abso- dhk0
lute value of h in {hkl}, the more nearly parallel the set of planes ϕ
ϕ
is to the a axis (the {h00} planes are an exception). The same is
true of k and the b axis, and l and the c axis. When h = 0, the
a/h
planes intersect the a axis at infinity, so the {0kl} planes are par-
allel to the a axis. Similarly, the {h0l} planes are parallel to the b Figure 15A.12 The construction used to find the spacing of the
axis and the {hk0} planes are parallel to the c axis. (hk0) plane in a square unit cell.
15A Crystal structure 661
The length of the hypotenuse of the lower triangle is a/h Collect your thoughts For the first part, all you need to do is
because a Miller index h indicates that the plane intersects the a substitute the given values into eqn 15A.1b. You could do the
axis at a distance a/h from the origin. Likewise, the hypotenuse same for part (b), but note that the Miller indices for the sec-
of the upper triangle is a/k. Then, because sin2φ + cos2φ = 1, it ond set of planes are just twice those of the first part. By refer-
follows that ring to eqn 15A.1b you can see that multiplying the values of
h, k, and l by n gives the following expression for the spacing of
2 2
hdhk 0 kdhk 0 the {nh nk nl} planes
1
a a 1/dhkl 2

2
which can be rearranged by dividing both sides by d hk0 into 1 2 2
(nh) (nk) (nl )2
2h
2
k 2 l 2 n2
2
2 2 2 n 2 2 2 2
dnh a b c a b c dhkl
1 h2 k 2 a ,nk ,nl
2
or dhk 0
d hk 0 a2 (h2 k 2 )1/ 2 which implies that
By extension to three dimensions, the separation of the {hkl} dhkl
dnh ,nk ,nl =
planes, dhkl, of a cubic lattice is given by n
The solution Substituting the indices into eqn 15A.1b gives
1 h2 + k 2 + l 2
2 =
dhkl a2 (15A.1a) 1 l2 22 32
a Separation of planes
22.0nm 2
dhkl = 2 2 2 1/2 d123 (0.82 nm) (0.94 nm) (0.75 nm)2
2 2 2
(h + k + l ) [cubic lattice]
Hence, d123 = 0.21 nm. It then follows immediately that d246 is

one-half this value, or 0.11 nm.
The corresponding expression for a general orthorhombic lat-
tice (one in which the axes are mutually perpendicular, but not Self-test 15A.1 Calculate the separation of (a) the {133} planes
equal in length) is the generalization of this expression: and (b) the {399} planes in the same lattice.
1 h2 k 2 l 2
Answer: 0.19 nm, 0.063 nm
Separation of planes
2
[orthorhombic latttice]
(15A.1b)
dhkl a 2 b2 c 2
Example 15A.1 Exercises

Using the Miller indices
E15A.2 State the Miller indices of the planes that intersect the crystallographic
Calculate the separation of (a) the {123} planes and (b) the axes at the distances (2a, 3b, 2c) and (2a, 2b, ∞c).
{246} planes of an orthorhombic unit cell with a = 0.82 nm, E15A.3 The unit cells of SbCl3 are orthorhombic with dimensions a = 812 pm,
b = 0.94 nm, and c = 0.75 nm. b = 947 pm, and c = 637 pm. Calculate the spacing, d, of the {321} planes.
Checklist of concepts
☐ 1. A periodic crystal is built up from regularly repeating ☐ 5. Unit cells are classified into seven crystal systems
structural motifs. according to their rotational symmetries: unit cells are
☐ 2. A space lattice is the pattern formed by points (the classified as cubic, monoclinic, or triclinic according to
lattice points) representing the locations of structural the essential symmetries they possess.
motifs (atoms, molecules, or groups of atoms, mol- ☐ 6. The Bravais lattices are the 14 distinct space lattices in
ecules, or ions). three dimensions (Fig. 15A.8).
☐ 3. A unit cell is an imaginary parallel-sided figure from ☐ 7. The unit cells of the Bravais lattices are classed as primi-
which the entire space lattice can be constructed by tive (P), body-centred (I), face-centred (F), and side-
purely translational displacements. centred (A, B, or C).
☐ 4. A primitive unit cell has lattice points only at its vertices ☐ 8. A lattice plane is specified by a set of Miller indices
and only one lattice point overall; non-primitive unit (hkl); sets of planes are denoted {hkl}.
cells also have lattice points at their centres or on pairs
of opposite faces.
662 15 Solids
Checklist of equations
Property Equation Comment Equation number

Separation of planes in a cubic lattice 1/d 2
hkl
2 2
(h k l )/a 2 2
h, k, and l are Miller indices 15A.1a
2 2 2 2 2 2 2
Separation of planes in an orthorhombic lattice 1/d hkl h /a k /b l /c 15A.1b
TOPIC 15B Diffraction techniques
for structural studies of solid materials. The actual process of

➤ Why do you need to know this material? going from the observed diffraction pattern to a structure is
To account for the properties of solids it is necessary to rather involved, but such is the degree of integration of com-
understand their detailed structures and how they are puters into the experimental apparatus that the technique is
determined by a variety of diffraction techniques. almost fully automated, even for large molecules and com-
plex solids. The analysis is aided by molecular modelling tech-
➤ What is the key idea? niques, which can guide the investigation towards a plausible
The regular arrangement of the atoms in periodic crystals structure.
can be determined by techniques based on diffraction. X-rays are electromagnetic radiation with wavelengths of the
order of 10−10 m. They are typically generated by bombarding a
➤ What do you need to know already? metal with high-energy electrons (Fig. 15B.1). The electrons de-
You need to be familiar with the description of crystal celerate as they plunge into the metal and generate radiation with a
structures and the use of Miller indices to identify lattice continuous range of wavelengths called Bremsstrahlung (Bremse
planes (Topic 15A). You also need to be familiar with the is German for deceleration, Strahlung for ray). Superimposed on
wave description of electromagnetic radiation (The chem- the continuum are a few high-intensity, sharp peaks (Fig. 15B.2).
ist’s toolkit 7A.1 and The chemist’s toolkit 7A.2), and the basic These peaks arise from collisions of the incoming electrons with
properties of the Fourier transform (The chemist’s toolkit the electrons in the inner shells of the atoms. The collision expels
12C.1). Light use is made of the de Broglie relation (Topic an electron from an inner shell, and an electron of higher energy
7A) and the equipartition theorem (Energy: A first look). drops into the vacancy, emitting the excess energy as an X-ray
photon (Fig. 15B.3). If the electron falls into a K shell (a shell with
n = 1), the X-rays are classified as ‘K-radiation’, and similarly for
transitions into the L (n = 2) and M (n = 3) shells. Strong, distinct
lines are labelled Kα, Kβ, and so on. Synchrotrons (Topic 11A) gen-
Diffraction techniques can be used to determine the details of erate high-intensity X-ray radiation which is increasingly used in
the arrangement of ions, atoms, and molecules in a crystalline diffraction experiments on account of the resulting greater inten-
solid to high precision. Such techniques are now so well de- sity in the diffraction pattern, and hence higher sensitivity.
veloped that both the collection of the diffraction data and its An early method of observing diffraction consisted of pass-
interpretation in terms of a structure are automated to a high ing a beam containing X-rays with a range of wavelengths into
degree. a single crystal, and recording the diffraction pattern pho-
tographically. The idea behind this approach is that a crystal
might not be suitably orientated to cause diffraction for a sin-
gle wavelength but, whatever its orientation, diffraction would
15B.1 X-ray crystallography be achieved for at least one of the wavelengths present in the
As explained in The chemist’s toolkit 7A.2, a characteristic prop-

erty of waves is that when they are present in the same region of X-rays
space they interfere with one another. A greater displacement
is obtained where peaks or troughs of the waves coincide and
a smaller displacement where peaks coincide with troughs. Cooling water Electron
Diffraction is the interference caused by an object in the path of beam
waves; it occurs when the dimensions of the diffracting object
Metal
are comparable to the wavelength of the radiation. target
Beryllium X-rays
window
15B.1(a) X-ray diffraction
Figure 15B.1 X-rays are generated by directing an electron beam
X-rays diffract when passed through a crystal because their on to a cooled metal target. Beryllium is transparent to X-rays (on
wavelengths are comparable to the separation of lattice planes. account of the small number of electrons in each atom) and is
Consequently, X-ray diffraction is a very powerful technique used for the windows.
664 15 Solids
Kα
Calcite
Kβ
Intensity
Bremsstrahlung
Aragonite
Wavelength, λ 20 30 40 50 60 70 80
Glancing angle, 2θ/°
Figure 15B.2 The X-ray emission from a metal consists of a
Figure 15B.4 X-ray powder diffraction patterns of two
broad, featureless Bremsstrahlung background, with sharp peaks
polymorphs of CaCO3, calcite and aragonite. The patterns are
superimposed on it. The label K indicates that the radiation comes
distinctive and can be used to identify the polymorph present in
from a transition in which an electron falls into a vacancy in the K
an unknown sample.
shell of the atom.
Ejected X-ray
electron Ω
Ionization
χ
Electron
ϕ
beam
Sample
Energy
X-ray
L beam 2θ
K To
detector
Figure 15B.3 The processes that contribute to the generation

of X-rays. An incoming electron collides with an electron (in the Figure 15B.5 A four-circle diffractometer. The settings of the
K shell), and ejects it. Another electron (from the L shell in this orientations (φ, χ, θ, and Ω) of the components are controlled by
illustration) falls into the vacancy and emits its excess energy as an computer; each (hkl) reflection is monitored in turn, and their
X-ray photon. intensities are recorded.
beam. There is currently a resurgence of interest in this ap- The method developed by the Braggs (William and his son
proach because synchrotron radiation spans a range of X-ray Lawrence) is the foundation of almost all modern work in
wavelengths. X-ray crystallography. They used a single crystal and a mono-
A more common technique uses monochromatic radiation chromatic beam of X-rays, and rotated the crystal until a reflec-
and a powdered sample, which consists of many tiny crystal- tion was detected. There are many different sets of planes in a
lites, oriented at random. At least some of the crystallites will crystal, so there are many angles at which a reflection occurs.
be appropriately orientated to give diffraction. In modern The complete set of data consists of the list of angles at which
‘powder diffractometers’ the intensities of the reflections are reflections are observed together with their intensities.
monitored electronically as the detector is rotated around the Single-crystal diffraction patterns are measured by using a
sample in a plane containing the incident ray. Powder diffrac- ‘four-circle diffractometer’ (Fig. 15B.5). Once the dimensions
tion techniques are used to identify the composition of a sam- and symmetry of the unit cell of the crystal in question have
ple of a solid substance by comparison of the positions of the been identified, the angular setting of the detector on the four
diffraction lines and their intensities with previously recorded circles is adjusted so that the precise position and intensity of
patterns of known structures (Fig. 15B.4); large databases of each peak in the diffraction pattern can be measured. Modern
such information are available. This approach can be used to instruments use area detectors and image plates, which sam-
determine the composition of mixed phases, and hence to con- ple whole regions of diffraction patterns simultaneously, rather
struct a phase diagram. The technique is also used for the initial than peak by peak, thus increasing the speed with which data is
determination of the dimensions and symmetries of unit cells. collected.
15B Diffraction techniques 665
15B.1(b) Bragg’s law 2d sin Bragg’s law [alternative form] (15B.1b)
An early approach to the analysis of diffraction patterns pro- and to regard the nth-order reflection as arising from the {nh
duced by crystals was to regard a lattice plane as a semi-trans- nk nl} planes. As established in Example 15A.1 in Topic 15A,
parent mirror and to model a crystal as a stack of reflecting the spacing of the {nh nk nl} planes is dhkl/n, where dhkl is the
lattice planes of separation d (Fig. 15B.6). The model makes it spacing of the {hkl} planes. The primary use of Bragg’s law is in
easy to calculate the angle the crystal must make to the incom- the determination of the spacing between the layers in the lat-
ing beam of X-rays for constructive interference to occur. It tice because d may readily be calculated from a measured value
has also given rise to the name reflection to denote an intense of the angle θ.
beam arising from constructive interference.
Consider the reflection of two parallel rays of the same wave-
length and phase by two adjacent planes of a lattice, as shown Brief illustration 15B.1
in Fig. 15B.6. One ray strikes point D on the upper plane and
the phase of the parallel ray at point A is the same as that of the A first-order reflection from the {111} planes of a cubic crystal
first ray at point D. The parallel ray must travel an additional was observed at a glancing angle, 2θ, of 22.4° when X-rays
of wavelength 154 pm were used. According to eqn 15B.1b,
distance AB before striking the plane immediately below. The
the {111} planes responsible for the diffraction have separation
reflected rays also differ in path length by the distance BC. As
d111 = λ/(2 sin θ), therefore
is evident from the inset in Fig. 15B.6, both the lengths AB and
BC are d sin θ; the total path length difference of the two rays 154 pm
is then d111 396 pm
2 sin 2 sin11.2
AB BC 2d sin
The separation of the {111} planes of a cubic lattice of side a is
where 2θ is the glancing angle (2θ rather than θ, because the given by eqn 15A.1a in Topic 15A as d111 = a/31/2. It therefore
beam is deflected through 2θ from its initial direction). For follows that a =31/2d111 = 687 pm.
many glancing angles the path-length difference is not an in-
teger number of wavelengths, and the waves interfere largely
destructively. However, when the path-length difference is an
integer number of wavelengths (AB + BC = nλ), the reflected Some types of unit cell give characteristic patterns of lines.
waves are in phase and interfere constructively. It follows that a In a cubic lattice with dimension a, the spacing of the {hkl}
reflection should be observed when θ satisfies Bragg’s law: planes, dhkl, is given by eqn 15A.1a (dhkl = a/(h2 + k2 + l2)1/2); the
angles at which the {hkl} planes give first-order reflections are
n 2d sin Bragg’s law (15B.1a)
given by
Reflections with n = 2, 3,… are called second order, third order,
and so on; they correspond to path-length differences of 2, 3,… dhkl = a/(h 2 + k 2 + l 2)1/2
wavelengths. In modern work it is normal to absorb the n into λ λ
d, to write Bragg’s law as sinθ = = (h 2 + k 2 + l 2 )1/2
2dhkl 2a
Not all integral values of h2 + k2 + l2 are obtained when integer

values of the indices are substituted:
θ θ
d D {hkl} {100} {110} {111} {200} {210}

A θ C
h2 + k2 + l2 1 2 3 4 5
d
{hkl} {211} {220} {300} {221} {310}
2 2 2
θ h +k +l 6 8 9 9 10
B
Notice that h2 + k2 + l2 = 7 (and 15,…) does not appear. As a
Figure 15B.6 The conventional derivation of Bragg’s law treats result, in the pattern of diffraction lines there is a larger gap
each lattice plane as a plane reflecting the incident radiation. between the {211} and {220} reflections than between nearby
The path lengths for reflection from adjacent planes differ by lines, and likewise between {321} (for which h2 + k2 + l2 = 14)
AB + BC, which depends on the angle θ. Constructive interference and {400} (for which h2 + k2 + l2 = 16). The absence of these lines
(a ‘reflection’) occurs when AB + BC is equal to an integer number leads to a characteristic pattern which helps in identifying the
of wavelengths. type of unit cell.
666 15 Solids
40 corresponds to sin θ = 0. That is, the scattering factor has its

maximum value in the forward direction. In this limit eqn
Br–
15B.2 simplifies to
Scattering factor, f(θ)
30
1
Fe2+
sin kr 2
20 f (0) lim 4 (r ) r dr
Ca2+ kr 0 0 kr
volume
element

10
Cl– 4 (r )r dr (r ) 4 r 2dr
2
O 0 0
C
H The factor 4πr2dr is the volume of a spherical shell of radius r
0
0 0.2 0.4 0.6 0.8 1 1.2 and thickness dr. Multiplication of this volume by the electron
(sin θ)/λ density at r gives ρ(r)4πr2dr which is the number of electrons in
Figure 15B.7 The variation of the scattering factor of atoms and
this shell. Integration over the full range of r then gives the total
ions with atomic number and angle. The scattering factor in the number of electrons in the atom. Hence f (0) = Ne. For example,
forward direction (at θ = 0, and hence at (sin θ)/λ = 0) is equal to the scattering factors of Na+, K+, and Cl− in the forward direc-
the number of electrons present in the species. tion are 8, 18, and 18, respectively.
15B.1(c) Scattering factors 15B.1(d) The electron density

X-ray scattering is caused by the oscillations an incoming elec- The structure factor, Fhkl, is the net amplitude of a given {hkl}
tromagnetic wave generates in the electrons of atoms. Heavy, reflection that takes into account the positions and types of all
electron-rich atoms give rise to stronger scattering than light the atoms in the unit cell. It can be expressed in terms of the
atoms. This dependence on the number of electrons is ex- locations and scattering factors of the atoms.
pressed in terms of the scattering factor, f, of the element. If the
scattering factor is large, then the atoms scatter X-rays strongly.
It turns out that, in a spherically symmetrical atom, the scatter- How is that done? 15B.2 Relating the structure factor to
ing factor for the atom is related to the electron density, ρ(r), the location of the atoms and their scattering factors
and the glancing angle, 2θ, by
Suppose that a unit cell contains several atoms with scattering
sin kr 2 4 factors fj and coordinates (xja,yjb,zjc), where xj is the coordinate
f ( ) 4 (r ) r dr k sin of the atom j in the a direction, expressed as a fraction of the
0 kr
length a, and likewise for the other coordinates.
Scattering factor (15B.2)
Step 1 Consider the (h00) reflection with h = 1
The scattering factor is greatest in the forward direction (θ = 0, The reflection shown in Fig. 15B.8 corresponds to two waves
Fig. 15B.7), and it can be shown that in that direction it is equal from adjacent A planes; the phase difference of the waves is 2π.
to the total number of electrons in the atom, Ne. If there is a B atom at a fraction x of the distance between the
two A planes, then it gives rise to a wave with a phase differ-
ence 2πx relative to an A reflection. To understand this point,
How is that done? 15B.1 Evaluating the scattering factor note that, if x = 0, there is no phase difference; if x = 12 the phase
difference is π; if x = 1, the atom in the B plane (a ‘B atom’) lies
in the forward direction
where the atom in the lower A plane (an ‘A atom’) is and the
The first step is to examine the function (sin kr)/kr in the limit phase difference is 2π.
that kr → 0. Use sin x x 16 x 3  and write
Step 2 Consider the (h00) reflection with h = 2
sin kr 3
kr (kr ) 
1
In this case, there is a 2 × 2π phase difference between the
lim
kr 0 kr
lim
kr 0
6
kr kr 0

lim 1 16 (kr )2  1 waves from the two A layers, and if B were to lie at x = 12 it
would give rise to a wave that differed in phase by 2π from
The point kr = 0 is a maximum which you can see by inspect- the wave from the lower A layer. Thus, for a general fractional
ing the first two terms in the expansion, 1 − 16 (kr )2. As kr moves position x, the phase difference for a (200) reflection is 2 × 2πx.
away from zero the value decreases, so kr = 0 is a maximum.
In fact it turns out to be the global maximum, which is most Step 3 Generalize these conclusions
easily verified simply by plotting the function. It therefore fol- For a general (h00) reflection, the phase difference is h × 2πx. For
lows that the scattering factor is a maximum for kr = 0, which three dimensions, this result generalizes to φhkl = 2π(hx + ky + lz).
Phase Phase Collect your thoughts You need to evaluate the sum in eqn
difference = 2πx difference = 2 × 2πx 15B.3. The sum is over all atoms in the unit cell, so you need
to know the location of each atom, expressed as a fraction of
the unit cell parameters. Several of the atomic coordinates are
A A already marked in the figure. Note that:
xa xa
a a
• The atoms at the corners of the cube are shared between
B B eight adjacent unit cells, so in the calculation of the struc-
ture factor these atoms are given a weight of 18 and their
A A scattering factors taken to be 18 f .
Phase Phase • The atoms on the faces are shared between two cells
(a) difference = 2π (b) difference = 2 × 2π and so have a weight of 12 ; those on the edges are shared
Figure 15B.8 Diffraction from a crystal containing two kinds of between four cells, and so have a weight of 14 .
atoms. (a) For a (100) reflection from the A planes, there is a phase Then follow these steps:
difference of 2π between waves reflected by neighbouring planes.
(b) For a (200) reflection, the phase difference is 4π. The reflection • Write f + for the Na+ scattering factor and f −
for the Cl−
from a B plane at a fractional distance xa from an A plane has a scattering factor.
phase that is x times these phase differences. • For simplicity, ignore the fact that scattering occurs in
non-forward directions and suppose that all the Na+ have
Step 4 Formulate the total amplitude of the scattered waves the same scattering factor, and likewise for the Cl− ions.
If the amplitude of the waves scattered from A is fA at the detec- • The best way to proceed is to draw up a table showing the
tor, that of the waves scattered from B with phase difference φhkl weights, positions, and phases.
is f Beiφhkl . The total amplitude at the detector is therefore
• The phase factors eiφ are evaluated by noting that h, k,
hkl
Fhkl f A f Beihkl and l are integers. A useful identity is that einπ is +1 for
even n, and −1 for odd n, more succinctly expressed as
When there are several atoms present, each with scattering fac- ein (1)n .
tor fj and phase φhkl(j) = 2π(hxj + kyj + lzj), the total amplitude
of the (hkl) reflection, the structure factor, is The solution The table for the Na+ ions is
Fhkl = ∑ f j e
iϕ hkl ( j )
(15B.3) Atom Weight x y z ϕhkl
j Structure factor
1 1
8
0 0 0 0
2 1
8
1 0 0 2πh
3 1
8
0 1 0 2πk
Example 15B.1 Calculating a structure factor 4 1
0 0 1 2πl
8
5 1
1 1 0 2π(h + k)
Calculate the structure factor for the unit cell of NaCl depicted 8
in Fig. 15B.9. 6 1
8
1 0 1 2π(h + l)
7 1
8
0 1 1 2π(k + l)
8 1
8
1 1 1 2π(h + k + l)
Na+
9 1
2
1
2
1
2
0 2 12 h 12 k
(1,0,1)
(1,1,1) 10 1
2
1
2
0 1
2 2 12 h 12 l
11 1
2
0 1
2
1
2 2 12 k 12 l
Cl
–
12 1
2
1 1
2
1
2 2 h 12 k 12 l
13 1
2
1
2
1 1
2 2 12 h k 12 l
14 1
2
1
2
1
2
1 2 12 h 12 k l
(0,0,0)
(½,0,0)
The phase factors for the first eight Na+ ions in the table are all
(½,½,0) +1, and as they each have a weight of 18 , the total contribution
(1,0,0) (1,1,0) to the structure factor is f +. The remaining six ions all have
weight 12 , and their contribution to the structure factor is
Figure 15B.9 The location of the atoms for the structure factor
calculation in Example 15B.1. The small spheres are Na+, the
1
2
[e i 2 ( h / 2 k / 2 ) e i 2 ( h / 2 l / 2 ) e i 2 ( k / 2 l / 2 ) e i 2 ( h k / 2 l / 2 )
large spheres are Cl−. ei 2 (h / 2 k l / 2) ei 2 (h / 2 k / 2 1) ]
668 15 Solids
The terms ei2πh , ei2πk , and ei2πl are all +1, so the last three terms Comment. If f + = f −, which is the case for identical atoms, the
can be simplified hkl all-odd reflections have zero intensity; such a structure
would be a cubic P lattice with lattice parameter a/2.
1
2
[e i 2 ( h / 2 k / 2 ) e i 2 ( h / 2 l / 2 ) e i 2 ( k / 2 l / 2 ) e i 2 ( k / 2 l / 2 ) Self-test 15B.1 Which reflections cannot be observed for a
ei 2 ( h / 2 l / 2 ) ei 2 ( h / 2 k / 2 ) ] cubic I lattice?
Answer: for h + k + l odd, Fhkl = 0
A further simplification is to use ein (1)n :
1
2
[(1)h k (1)h l (1)k l (1)k l (1)h l (1)h k ]
(1)h k (1)h l (1)k l The intensity of a reflection is proportional to the square
modulus of the amplitude of the wave, which is in turn pro-
The overall contribution of the Na+ ions to the structure factor portional to the structure factor, Fhkl. If the structure factor is
is therefore Fhkl f A f B eihkl , the intensity, Ihkl, is
f [1 (1)h k (1)h l (1)k l ]
* Fhkl = ( f A + fBe
I hkl Fhkl
− iϕ hkl
)( fA + fBeiϕ ) hkl
The table for the Cl− ions is

iϕ hkl
= fA2 + fB2 + fA fB(e + e − iϕ )
hkl
eiϕ + e−iϕ = 2cos ϕ

Atom Weight x y z ϕhkl
1 1
4
1
2
0 0 πh = f A2 + f B2 + 2 fA fB cos ϕhkl
2 1
4
0 1
2
0 πk
3 1
4
1 1
2
0 2 h 12 k The cosine term either adds to or subtracts from f A2 + f B2 de-
4 1
4
1
2
1 0 2 h k
1
2 pending on the value of φhkl, which in turn depends on h, k, and
5 1
4
0 0 1
2
πl l and x, y, and z. Hence, there is a variation in the intensities of
6 1
4
1 0 1
2 2 h 12 l the reflections with different hkl. The A and B reflections in-
7 1
0 1 1
2 k 12 l terfere destructively when the phase difference is π, and in this
4 2
8 1
1 1 1
2 h k 12 l
case the total intensity is zero if the atoms have the same scatter-
4 2
ing power. For example, if the unit cell is cubic I with a B atom
9 1 1 1 1
2 12 h 12 k 12 l
2 2 2
at x= y= z = 12 , then the A,B phase difference is (h + k + l)π.
10 1 1
0 1 2 12 h l
4 2
Therefore, all reflections for odd values of h + k + l vanish if A
11 1
4
0 1
2
1 2 12 k l and B are identical atoms because the waves from A and B are
12 1
4
1 1
2
1 2 h 12 k l displaced in phase by π.
13 1
4
1
2
1 1 2 12 h k l For a cubic P lattice diffraction is possible for all {hkl}, there-
fore the diffraction pattern for a cubic I lattice can be con-
structed from that for cubic P by striking out all reflections
A similar procedure to that used for the Na+ ions gives the fol- with odd values of h + k + l. Similarly, for a cubic F lattice the
lowing contribution of the Cl− ions to the structure factor missing lines are ones with two out of h, k, and l odd, and the
remaining one even, or two even and one odd. Recognition of
f [(1)h k l (1)h (1)k (1)l ] these systematic absences in a powder spectrum can be used to
The structure factor is therefore assign the lattice type (Fig. 15B.10).
The intensity of the {hkl} reflection is proportional to |Fhkl|2,
Fhkl f [1 (1)h k (1)h l (1)k l ] so in principle the structure factors can be determined experi-
f [(1)h k l (1)h (1)k (1)l ] mentally by taking the square root of the corresponding inten-
sities (but see below). Then, once all the structure factors Fhkl
Now note that: are known the electron density distribution, ρ(r), in the unit
cell can be calculated by using
• if h, k, and l are all even, Fhkl = f +{1 + 1 + 1 + 1} +
f −{1 + 1 + 1 + 1} = 4( f + + f −) 1
• if h, k, and l are all odd, Fhkl = 4( f − f ) + −
(r ) Fhkl e2i(hx ky lz )
V hkl
Fourier synthesis (15B.4)
• if one index is odd and two are even, or vice versa, Fhkl = 0 where V is the volume of the unit cell. Equation 15B.4 is called
The hkl all-odd reflections are therefore less intense than the a Fourier synthesis of the electron density. Fourier transforms
hkl all even, and some of the reflections are absent. occur throughout chemistry in a variety of guises, and are
(221)(300)
(410)(322)
(502)(430)
(411) (330)
(a)
(200)
(400)
(220)
(222)
(320)
(321)
(331)
(420)
(421)
(332)
(422)
(100)
(210)
(310)
(211)
(311)
(110)
(111)
(b)
Electron density, ρ(x)

Cubic P
Cubic I
0
Cubic F 0 0.5 1
x
Figure 15B.10 The powder diffraction patterns and the systematic Figure 15B.11 The plot of the electron density calculated in
absences of three versions of a cubic cell as a function of angle: Example 15B.2 (thick) and Self-test 15B.2 (thin). Interact with the
cubic F (fcc; h, k, l all even or all odd are present), cubic I (bcc; dynamic version of this graph in the e-book.
h + k + l = odd are absent), cubic P. Comparison of the observed
pattern with patterns like these enables the unit cell to be
identified. The locations of the lines give the cell dimensions. Only 15 values of Fh are given, so ρ(x) will be approximate: the
result (computed using mathematical software) is plotted in
escribed in more detail in The chemist’s toolkit 12C.1. The essence
d Fig. 15B.11 (thick line). There are three clear maxima in this
of the procedure in this case is to express the varying electron den- function, which can be identified as the positions of three atoms.
sity in a unit cell as a superposition of sine and cosine waves.
Comment. The more terms that are included (meaning the
more reflections that are measured), the more accurate is the
Example 15B.2 density plot. Terms corresponding to high values of h (which
Calculating an electron density by Fourier correspond to short-wavelength cosine terms in the sum)
synthesis account for the finer details of the electron density; low values
Consider the {h00} planes of a crystal extending indefinitely in of h account for the broad features.
the x direction. In an X-ray analysis the structure factors were Self-test 15B.2 Use mathematical software to experiment with
found as follows: the result of altering the structure factors in the table: con-
h 0 1 2 3 4 5 6 7 8 9 sider the effect of both changing the signs and amplitudes. For
example, use the same values of Fh as above, but change all the
Fh 16 −10 2 −1 7 −10 8 −3 2 −3
signs for h ≥ 6.
h 10 11 12 13 14 15 Answer: Fig. 15B.11 (thin line)
Fh 6 −5 3 −2 2 −3
It was also found that F−h = Fh. Construct a plot of the electron
density projected on to the x-axis of the unit cell.
15B.1(e) The determination of structure
Collect your thoughts You need to substitute these values into
eqn 15B.4, but because the problem is one-dimensional, the
The structure factors used in eqn 15B.4 to compute the electron
sum is over only the index h and only the terms e 2ihx need be density are in general complex quantities that can be written
considered. |Fhkl|eiα, where |Fhkl| is the amplitude and α the phase. However,
the observed intensity Ihkl is proportional to the square modu-
The solution Because F−h = Fh, the sum, rather than running lus of the structure factor, |Fhkl|2, so from the experiment no in-
from h = −∞ to +∞, can be written as running from 1 to +∞: formation is available about the phase, which may lie anywhere
=F from 0 to 2π. This ambiguity is called the phase problem; its
}h consequences are illustrated by comparing the two plots in
V ρ( x ) = ∑ Fh −2 πihx
= F0 + ∑ h
( F e −2πihx
+ F−h e
2 πihx
)
Fig. 15B.11 in which the phases of the structure factors have
h=− h=1
been changed but the amplitudes kept the same. Some way
= F0 + ∑F (e h
−2 πihx
+ e 2 πihx ) must be found to assign phases to the structure factors, because
h=1
unless these are known ρ cannot be evaluated using eqn 15B.4.
e ix + e−i x = 2cos x
The phase problem is less severe for centrosymmetric unit cells,
= F0 + 2∑Fh cos πhx for then the structure factors are real. It still remains a problem
h=1 though, to decide whether Fhkl is positive or negative.
670 15 Solids
The phase problem can be overcome to some extent by a of each peak, relative to the origin, corresponds to the separa-
variety of methods. One procedure that is widely used for in- tion and relative orientation of a pair of atoms in the cell. Note
organic materials with a reasonably small number of atoms in that the Patterson map is centrosymmetric and has a strong
a unit cell, and for organic molecules with a small number of feature at the origin.
heavy atoms, is the Patterson synthesis. Instead of the structure
factors Fhkl, the values of |Fhkl|2, which can be obtained without
ambiguity from the intensities, are used in an expression that Heavy atoms dominate the scattering because their scatter-
resembles eqn 15B.4: ing factors are large, of the order of their atomic numbers, and
1 their locations may be deduced quite readily. The sign of Fhkl
V
P (r ) | Fhkl |2 e 2 i (hx ky lz ) Patterson synthesis (15B.5) can now be calculated from the known locations of the heavy
hkl
atoms in the unit cell, and to a high probability the phase calcu-
where the r are the vector separations between the atoms in the
lated for them will be the same as the phase for the entire unit
unit cell; that is, the distances and directions between atoms.
cell. To see why, consider a centrosymmetric cell for which each
Whereas the electron density function ρ(r) is the probability
term in the structure factor is either positive or negative. The
density of the positions of atoms, the function P(r) is a map of
structure factor has the form
the probability density of the separations between atoms; P(r)
is often called the Patterson map. In such a map, a peak at a F () f heavy () f light () f light  (15B.6)
position specified by a vector r from the origin arises from pairs
of atoms that are separated by r. Thus, if atom A is at the coor- where fheavy is the scattering factor of the heavy atom and flight
dinates (xA,yA,zA) and atom B is at (xB,yB,zB), then there will be the scattering factors of the light atoms. The flight are all much
a peak at (xA − xB, yA − yB, zA − zB) in the Patterson map. There smaller than fheavy, and their phases are more or less random if
will also be a peak at the negative of these coordinates, because the atoms are distributed throughout the unit cell. Therefore,
there is a separation vector from B to A as well as a separation the net effect of the flight is to change F only slightly from fheavy,
vector from A to B. The height of the peak in the map is propor- and with reasonable confidence F will have the same sign as
tional to the product of the atomic numbers of the two atoms, that calculated from the location of the heavy atoms. This phase
ZAZB. The Patterson map also shows a strong feature at its ori- can then be combined with the observed |F| (from the reflec-
gin arising from the separation between each atom and itself, tion intensity) to perform a Fourier synthesis of the full elec-
which is necessarily zero. tron density in the unit cell, and hence to locate the light atoms
as well as the heavy atoms.
Modern structural analyses make extensive use of direct
methods. Direct methods are based on the possibility of treat-
Brief illustration 15B.2
ing the atoms in a unit cell as being virtually randomly dis-
For the electron density shown in Fig. 15B.12a, the corre- tributed (from the radiation’s point of view), and then to use
sponding Patterson map is shown in Fig. 15B.12b. The location statistical techniques to compute the probabilities that the
phases have a particular value. It is possible to deduce rela-
tions between some structure factors and sums (and sums of
squares) of others, which have the effect of constraining the
phases to particular values (with high probability, so long as the
R1
R4
structure factors are large). For example, the Sayre probability
R4
R3 relation has the form
R2 R1
R2 sign of Fh h,k k ,l l is probably equal to (sign of Fhkl )
R3 –R1 –R2 (sign of Fhk l ) Sayre probability relation (15B.7)
–R3
–R4 For example, if F122 and F232 are both large and negative, then it
is highly likely that F354, provided it is large, will be positive.
In the final stages of the determination of a crystal structure,
(a) (b) the parameters describing the structure (atom positions, for in-
stance) are adjusted systematically to give the best fit between
Figure 15B.12 The Patterson map corresponding to the electron
density in (a) is shown in (b). The distance and orientation of each
the observed intensities and those calculated from the model
spot from the origin gives the orientation and separation of one of the structure deduced from the diffraction pattern. This pro-
atom–atom separation in (a); in addition, there is a large spot at cess is called structure refinement. Not only does the proce-
the origin. Some of the typical distances and their contribution to dure give accurate positions for all the atoms in the unit cell,
(b) are shown as R1, etc. but it also gives an estimate of the errors in those positions and
in the bond lengths and angles derived from them. The proce- h

dure also provides information on the vibrational amplitudes (mkT )1/ 2
of the atoms.
Therefore, at 373 K,
Exercises
6.626 ×10−34 Js
E15B.1 What are the values of the angle θ of the three diffraction lines with λ= −27
{(1.675 ×10 kg) × (1.381×10−23 JK −1 ) × (373 K)}1/2
the smallest θ expected from a cubic I unit cell with lattice parameter 291 pm
when the X-ray wavelength is 72 pm? Hint: Are all reflections possible for
1 J = 1kg m 2s −2
such a unit cell?
6.626 ×10−34 kgm 2 s −1
E15B.2 What is the value of the scattering factor in the forward direction for Br ?
−
=
(1.675 ×10 ×1.381×10 × 373) (kg 2 m 2 s −2 )1/2
−27 −23 1/2
E15B.3 The coordinates, in units of a, of the atoms in a cubic P unit cell are
(0,0,0), (0,1,0), (0,0,1), (0,1,1), (1,0,0), (1,1,0), (1,0,1), and (1,1,1). Calculate = 2.26 ×10−10 m = 226 pm
the structure factor Fhkl when all the atoms are identical. Where possible,
simplify your expression by using einπ = (−1)n, as in Example 15B.1.
Self-test 15B.3 Calculate the temperature needed for the aver-
age wavelength of the neutrons to be 100 pm.
Neutron and electron

Answer: 1900 K
15B.2
diffraction
Neutrons and electrons can also give rise to diffraction on ac- Neutron diffraction differs from X-ray diffraction in two
count of their wave nature; their wavelength is given by the de main respects. First, the scattering of neutrons is a nuclear
Broglie relation (Topic 7A, λ = h/p). Neutrons generated in a phenomenon. Neutrons pass through the extranuclear elec-
nuclear reactor and then slowed to thermal velocities have trons of atoms and interact with the nuclei through the ‘strong
wavelengths similar to those of X-rays and may also be used force’ that is responsible for binding nucleons together. As a
for diffraction studies. For instance, a neutron generated in result, the intensity with which neutrons are scattered is in-
a reactor and slowed to thermal velocities by repeated colli- dependent of the number of electrons and neighbouring el-
sions with a moderator (such as graphite) until it is travelling ements in the periodic table might scatter neutrons with
at about 4 km s−1 has a wavelength of about 100 pm. In practice, markedly different intensities. Neutron diffraction can be used
a range of wavelengths occurs in a neutron beam, but a mono- to distinguish atoms of elements such as Ni and Co that are
chromatic beam can be selected by diffraction from a crystal, present in the same compound and to study order–disorder
such as a single crystal of germanium. phase transitions in FeCo. A second difference is that neutrons
possess a magnetic moment due to their spin. This magnetic
moment can couple to the magnetic fields of atoms or ions in
Example 15B.3 Calculating the typical wavelength of a crystal (if the ions have unpaired electrons) and modify the
thermal neutrons diffraction pattern. One consequence is that neutron diffrac-
tion is well suited to the investigation of magnetically ordered
Calculate the typical wavelength of neutrons after reach-
lattices in which neighbouring atoms may be of the same el-
ing thermal equilibrium with their surroundings at 373 K.
For simplicity, assume that the particles are travelling in one
ement but have different orientations of their electronic spin
dimension. (Fig. 15B.13).
Electrons accelerated through a potential difference of 40 kV
Collect your thoughts In order to use the de Broglie rela- have wavelengths of about 6 pm, and so are also suitable for dif-
tion, you need to know the momentum of the neutrons, and fraction studies of molecules. Consider the scattering of elec-
therefore their velocity. The velocity can be computed from trons from pairs of atoms with centres separated by a distance
the kinetic energy, which you can assume has its equipartition Rij and orientated at a definite angle θ to an incident beam of
value for translation in one dimension, Ek = 12 kT (see Energy: A electrons. When the molecule consists of a number of atoms,
first look). The mass of a neutron is 1.675 × 10−27 kg.
the scattering intensity can be calculated by summing over the
The solution From the equipartition principle, the mean trans- contribution from all pairs. The total intensity I(θ) is given by
lational kinetic energy of a neutron at a temperature T travel- the Wierl equation:
ling in the x-direction is Ek = 12 kT . The kinetic energy is also
equal to p2/2m, where p is the momentum of the neutron and sin sRij 4 1
m is its mass. Hence, p = (mkT)1/2. It follows from the de Broglie I ( ) fi f j s sin 2 Wierl equation (15B.8)
relation, λ = h/p, that the wavelength of the neutron is i ,j sRij
672 15 Solids
where λ is the wavelength of the electrons in the beam, and

f is the electron scattering factor, a measure of the elec-
tron s cattering power of the atom. The main application of
electron diffraction techniques is to the study of surfaces
(Topic 19A).
Exercises
E15B.4 What speed should neutrons have if they are to have a wavelength of
Figure 15B.13 If the spins of atoms at lattice points are orderly, 65 pm?
as in this material, where the spins of one set of atoms are aligned
E15B.5 Calculate the wavelength of neutrons that have reached thermal
antiparallel to those of the other set, neutron diffraction detects
equilibrium by collision with a moderator at 350 K.
two interpenetrating simple cubic lattices on account of the
magnetic interaction of the neutron with the atoms, but X-ray
diffraction would see only a single bcc lattice.
☐ 1. A reflection refers to an intense beam emerging in a ☐ 8. The phase problem arises because it is possible to
particular direction and arising from constructive inter- measure only the intensity of the reflections and not
ference. their phases; as a result Fourier synthesis cannot be
☐ 2. The glancing angle, 2θ, is the angle through which a used in a straightforward way to determine the electron
beam is deflected. density.
☐ 3. Bragg’s law relates the angle of a diffracted beam to the ☐ 9. A Patterson map is a map of the interatomic vectors.
spacing of a given set of lattice planes. ☐ 10. Direct methods use statistical techniques to determine
☐ 4. The scattering factor is a measure of the ability of an the likely phases of the reflections.
atom to scatter electromagnetic radiation. ☐ 11. Structure refinement is the adjustment of structural
☐ 5. The structure factor is the overall amplitude of a wave parameters to give the best fit between the observed
diffracted by the {hkl} planes and atoms distributed intensities and those calculated from the model of the
through the unit cell. structure deduced from the diffraction pattern.
☐ 6. The electron density and the diffraction pattern are ☐ 12. The Wierl equation relates the intensity of electron
related by a Fourier transform. scattering to the distances between pairs of atoms in the
sample.
☐ 7. Fourier synthesis is the construction of the electron
density distribution from structure factors.
Bragg’s law λ = 2d sin θ d is the lattice spacing, 2θ the glancing angle 15B.1b

Scattering factor f 4 0 { (r )sin kr /kr }r dr, k = (4π/λ) sin θ
2
Spherically symmetrical atom 15B.2
ihkl ( j )
Structure factor Fhkl f j e , φhkl(j) = 2π(hxj + kyj + lzj) Definition 15B.3
j
Fourier synthesis (r ) (1/V ) Fhkl e 2 i ( hx ky lz )

V is the volume of the unit cell 15B.4
hkl
Patterson synthesis P (r ) (1/V ) | Fhkl |2 e 2 i (hx ky lz ) 15B.5

hkl
Wierl equation I ( ) fi f j (sin sRij /sRij ), s (4 / )sin 12 15B.8

i ,j
TOPIC 15C Bonding in solids

To understand the properties and structures of solid mate-
rials you need to know about the type of bonding that
holds together the atoms, ions, and molecules.

A
Four characteristic types of bonding result in metals, ionic
solids, covalent solids, and molecular solids.
Figure 15C.1 A layer of close-packed spheres used to build a
➤ What do you need to know already? three-dimensional close-packed structure.
You need to be familiar with molecular interactions (Topic
14B) and the general features of crystal structures (Topic are structures that are identical in two dimensions (the close-
15A). For the discussion of metallic bonding you should be packed layers) but differ in the third dimension.
aware of the principles of Hückel molecular orbital theory In all polytypes, the spheres of the second close-packed layer
(Topic 9E). The discussion of ionic bonding makes use of lie in the depressions of the first layer (Fig. 15C.2). The third
the concept of enthalpy (Topic 2B). layer may be added in either of two ways. In one, the spheres
are placed directly above the first layer to give an ABA pattern
of layers (Fig. 15C.3a). Alternatively, the spheres may be placed
over the depressions in the first layer that are not occupied by
Solids may be classified into four broad types, namely metals, the second layer (these depressions are visible in Fig. 15C.2),
ionic solids, covalent (or network) solids, and molecular sol- so giving an ABC pattern (Fig. 15C.3b). Two polytypes are
ids. Each type is characterized by the nature of the bonding be- formed if the two stacking patterns are repeated in the vertical
tween the constituents. direction:
• Hexagonally close-packed (hcp): the ABA pattern is
repeated, to give the sequence of layers ABABAB.…
• Cubic close-packed (ccp): the ABC pattern is repeated, to
15C.1 Metals give the sequence ABCABC.…
In a metal the electrons are delocalized over arrays of identi- The origins of these names can be seen by referring to
cal cations and bind the whole together into a rigid but duc- Fig. 15C.4. The ccp structure gives rise to a face-centred unit
tile and malleable structure. The crystalline forms of metallic
elements can be discussed in terms of a model in which their
atoms are treated as identical hard spheres. Most metallic ele-
ments crystallize in one of three simple forms, two of which can
be explained in terms of the hard spheres packing together in
the closest possible arrangement.
15C.1(a) Close packing B

A
Figure 15C.1 shows a close-packed layer of identical spheres,
one with maximum utilization of space. A close-packed three- Figure 15C.2 To achieve the greatest packing density, the second
dimensional structure is obtained by stacking such layers layer of close-packed spheres must sit in the depressions of the
on top of one another. However, this stacking can be done in first layer. The two layers are the AB component of the close-
different ways and results in close-packed polytypes, which packed structure.
674 15 Solids
Table 15C.1 The crystal structures of some elements⋆
Structure Element
hcp‡ Be, Cd, Co, He, Mg, Sc, Ti, Zn
fcc‡ (ccp, cubic F) Ag, Al, Ar, Au, Ca, Cu, Kr, Ne, Ni, Pb, Pd, Pt, Rh, Rn,
Sr, Xe
C bcc (cubic I) Ba, Cr, Cs, Fe, K, Li, Mn, Mo, Na, Rb, Ta, V, W
A
cubic P Po
B B
A A ⋆
The notation used to describe unit cells is introduced in Topic 15A. The structures are
for the elements at 298 K and 1 bar; for the elements that are normally gases, the structures
(a) (b) are for their solid form close to the freezing temperature and 1 bar.
‡
Close-packed structures.
Figure 15C.3 (a) The third layer of close-packed spheres might
occupy the depressions lying directly above the spheres in the first
layer, resulting in an ABA structure, which corresponds to hexagonal ccp and hcp structures. Another measure of their compactness
close-packing. (b) Alternatively, the third layer might lie in the
is the packing fraction, the fraction of space occupied by the
depressions that are not above the spheres in the first layer, resulting
spheres, which is 0.740 (see Example 15C.1) for these close-
in an ABC structure, which corresponds to cubic close-packing.
packed structures. That is, in a close-packed solid of identical
hard spheres, only 26.0 per cent of the volume is empty space.
The fact that many metals are close-packed accounts for their
high mass densities.
A B
C
B
Example 15C.1 Calculating a packing fraction
A A Calculate the packing fraction of a ccp structure formed from

hard spheres.
(a) (b)
Collect your thoughts You need to calculate the volume of
Figure 15C.4 Fragments of the structures shown in Fig. 15C.3 the unit cell and the volume of the spheres that are wholly or
revealing the (a) hexagonal (b) cubic symmetry. The labels of the partly contained within the cell, and then take the ratio of the
spheres are the same as for the layers in Fig. 15C.3. two volumes. The key step is to establish a relation between
the radius of the spheres, R, and the cell dimension, a. Figure
cell, so may also be denoted cubic F (or fcc, for face-centred 15C.5 shows that the sphere in the middle of a face just touches
cubic). The hcp structure has a hexagonal unit cell, with a struc- the two spheres at opposite corners of the face. The length of
tural motif consisting of a sphere at one corner of the cell and the face diagonal is therefore 4R. To calculate the volume of the
another at the centre of the cell (1). It is also possible to have spheres in the cell, you need to account for the fraction of each
random sequences of layers; however, the hcp and ccp poly- sphere that is contained within the cell. Spheres at the corners
types are the most important. Table 15C.1 lists the structures contribute 18 of their volume to the cell; those on the faces con-
adopted by a selection of elements. tribute 12 to the cell.
4R
a
a
a = 81/2R
4R
a = 81/2R
The compactness of close-packed structures is indicated by Figure 15C.5 In a ccp unit cell (a fcc unit cell) a sphere located
their coordination number, the number of spheres immedi- on the face of the cube just touches the spheres at opposite
ately surrounding any selected sphere, which is 12 for both the corners of the face.
15C Bonding in solids 675
The solution As can be seen from Fig. 15C.5, the length of

(a) N=1
the face diagonal is 4R. From Pythagoras’ theorem it follows
that a2 + a2 = (4R)2, so 2a2 = 16R2 and therefore a = 81/2R. (b) N=2
The volume of the unit cell is a3, which is therefore 83/2R3.
There are eight spheres at the corners, each contributing 18 (c) N=3
of their volume to the cell (giving a net contribution of 1
sphere), and six spheres on the faces, each contributing 12 (d) N=4
(giving a net contribution of 3 spheres). The total volume
occupied by the spheres is equivalent to 4 complete spheres.
Because the volume of each sphere is 34 πR3 , the total occu- (e) N=∞
pied volume is 163 πR3. The fraction of space occupied is
therefore
Figure 15C.6 The formation of a band of N molecular orbitals by
83/2 = 16 2 successive addition of N atomic orbitals in a line. When N becomes
16
πR 3 π infinite the band covers a finite range of energy and the orbitals
3
= = 0.740
8 R3 3 2
3/2 within it are very closely spaced, but still discrete.
An hcp structure has the same coordination number, so its

packing fraction is the same. of electrical properties of solids discussed in Topic 15E, and is
described here.
Self-test 15C.1 Calculate the packing fraction of the cubic I As a starting point, consider a one-dimensional solid, which
(body-centred cubic, bcc) structure in which one sphere is at consists of a single, infinitely long line of atoms. Suppose that
the centre of a cube formed by eight others. The spheres touch each atom has one s orbital available for forming molecular
along the body diagonal of the cube. orbitals. The LCAO-MOs of the solid are constructed by add-
Answer: 31/2π/8 = 0.680 ing N atoms in succession to a line, and then inferring the elec-
tronic structure by using the building-up principle. One atom
contributes one s orbital at a certain energy (Fig. 15C.6). When
a second atom is brought up it overlaps the first and forms a
As shown in Table 15C.1, a number of common metals bonding and an antibonding orbital. The third atom overlaps
adopt structures that are less than close-packed. The depar- its nearest neighbour (and only slightly the next-nearest), and
ture from close packing suggests that factors, such as specific from these three atomic orbitals, three molecular orbitals are
covalent bonding between neighbouring atoms, are begin- formed: one is fully bonding, one fully antibonding, and the
ning to influence the structure and impose a particular geo- intermediate orbital is nonbonding between neighbours. The
metrical arrangement. One such arrangement results in a fourth atom leads to the formation of a fourth molecular or-
cubic I (bcc, for body-centred cubic) structure, with one bital. At this stage, it can be seen that the effect of bringing up
sphere at the centre of a cube formed by eight others. The co- successive atoms is to spread the range of energies covered by
ordination number of a bcc structure is only 8, but there are the molecular orbitals and to fill in the range of energies with
six more atoms not much further away than the eight near- more and more orbitals (one more for each atom). When N
est neighbours. The packing fraction of 0.680 (Self-test 15C.1) atoms have been added to the line, there are N molecular orbit-
is not much smaller than the value for a close-packed struc- als covering a finite range of energies: this set of orbitals is said
ture (0.740), and shows that about two-thirds of the available to form a band.
space is occupied by the atoms. The energies of the molecular orbitals that form the band
can be found by using the Hückel approximations described in
15C.1(b) Electronic structure of metals Topic 9E. The energies Ek of the molecular orbitals are then
The central aspect of solids that determines their electrical k

properties (Topic 15E) is the distribution of their electrons. Ek 2 cos k 1,2,,N
N 1
There are two models of this distribution. In one, the nearly Energy levels [linear array of s orbitals] (15C.1)
free-electron approximation, the valence electrons are as-
sumed to be trapped in a box with a periodic potential, with where α is the Coulomb integral and β is the (s,s) resonance
low potential energy corresponding to the locations of cations. integral. It is not hard to show that this expression implies that
In the tight-binding approximation, the valence electrons are when N is infinitely large, the separation between neighbouring
assumed to occupy molecular orbitals delocalized throughout levels, Ek+1 − Ek, is infinitely small, but the band still has finite
the solid. The latter model is more in accord with the discussion width overall, with EN − E1 → −4β.
676 15 Solids
Highest level of p band Fully antibonding

How is that done? 15C.1 Evaluating the separation p Band
+ −+−+−+−
between neighbouring levels and the width of a band
p
This calculation requires looking at a specific limiting case of
eqn 15C.1. Lowest level of p band Fully bonding
− + + − − + +
Step 1 Write an expression for the energy difference between two
Band
neighbouring levels Fully antibonding
gap Highest level of s band
From eqn 15C.1 it follows that the energy separation between + − + − +
neighbouring energy levels k and k +1 is s
+ + + + +
(k +1)π kπ s Band
Ek+1 − Ek = α + β cos − α + 2β cos
N +1 N +1 Lowest level of s band Fully bonding
(k +1)π kπ Figure 15C.7 The overlap of s orbitals gives rise to an s band and
= 2β cos − cos
N +1 N +1 the overlap of p orbitals gives rise to a p band. In this case, the s
and p orbitals of the atoms are so widely spaced in energy that
Step 2 Find the limit as N → ∞ there is a band gap; it is also possible that the separation will be
First use the trigonometric identity cos(A + B) = cos A cos B − less, resulting in the bands overlapping.
sin A sin B to express the boxed term as
→1 as N→ → 0 as N→ The band can be thought of as consisting of N different molecu-

(k +1)π kπ π kπ π
lar orbitals, the lowest-energy orbital (k = 1) being fully bonding,
cos = cos cos − sin sin and the highest-energy orbital (k = N) being fully antibonding
N +1 N +1 N +1 N +1 N +1
between adjacent atoms (Fig. 15C.7). The molecular orbitals of
intermediate energy have k − 1 nodes distributed along the chain
As N → ∞, /(N 1) 0 and therefore cos{/(N 1)} 1
of atoms. Similar bands form in three-dimensional solids (see
and sin{/(N 1)} 0. In this limit the boxed term tends to
Problem 15C.5).
cos{k/(N 1}) and therefore the energy difference becomes
k k Brief illustration 15C.1

Ek 1 Ek 2 cos cos 0
N 1 N 1
To illustrate the dependence of Ek+1 − Ek on N, note that
It follows that when N is infinitely large, the difference between
2
neighbouring energy levels is infinitely small. for N 3: E2 E1 2 cos cos 1.414
4 4
Step 3 Write an expression for the width of the band for N → ∞ 2
for N 30: E2 E1 2 cos cos 0.0307
The width of the band is simply EN − E1. Each of the energies 31 31
can be approximated as follows for the case that N → ∞. 2
for N 300: E2 E1 2 cos cos 0.000 327
301 301
E1 2 cos
N 1
The energy difference decreases with increasing N, as expected.
As N → ∞, the term π/(N + 1) tends to zero so the cosine tends
to 1; therefore in this limit The band formed from overlap of s orbitals is called the s
E1 2
band. If the atoms have p orbitals available, the same procedure
leads to a p band (as shown in the upper half of Fig. 15C.7).
When k has its maximum value of N, If the atomic p orbitals lie higher in energy than the s orbit-
als, then the p band lies higher than the s band, and there may
N be a band gap, a range of energies to which no orbital corre-
EN 2 cos
N 1 sponds. However, it is also possible for the bands to touch, with
As N → ∞, the 1 in the denominator can be ignored, so the the highest orbital of the s band coincident with the lowest level
cosine term becomes cos π = −1. Therefore, in this limit EN = of the p band, or even overlap (as is the case for the 3s and 3p
α − 2β, and so the band width is EN − E1 → −4β. Recall that β is bands in magnesium).
negative, so that the band width, −4β, is positive. Now consider the electronic structure of a solid formed from
N atoms each able to contribute one electron (for example, the
Cl–
Unoccupied levels
Cs+
Energy
Fermi level
Occupied levels
Figure 15C.8 When N electrons occupy a band of N orbitals, only Figure 15C.9 The caesium chloride structure consists of two
half of the orbitals are occupied (at T = 0) because two electrons interpenetrating simple cubic arrays of ions, one of cations and
will occupy each orbital. The highest occupied level is known as the other of anions, so that each cube of ions of one kind has a
the Fermi level. counter-ion at its centre.
alkali metals). The N atomic orbitals give rise to a band consist- number of nearest neighbours of opposite charge; the structure
ing of N molecular orbitals. Each of these orbitals can accom- itself is characterized as having (N+,N−) coordination, where N+
modate two spin-paired electrons, so at T = 0 only the lowest is the coordination number of the cation and N− that of the
1
2
N molecular orbitals are occupied (Fig. 15C.8). The HOMO anion.
is called the Fermi level. Only the small number of electrons Even if, by chance, the ions have the same size, the re-
close to the Fermi level can undergo thermal excitation, so only quirement that the unit cells are electrically neutral make
these electrons contribute to the heat capacity of the metal. It it impossible to achieve 12-coordinate close-packed ionic
is for this reason that Dulong and Petit’s law for heat capacities structures. As a result, ionic solids are generally less dense
(Topic 7A) gives reasonable agreement with experiment at nor- than metals. The best packing that can be achieved is the
mal temperatures by considering only the atoms in a sample, (8,8)-coordinate caesium chloride structure in which each
not the atoms plus the ‘free’ electrons. The presence of an in- cation is surrounded by eight anions and each anion is sur-
completely filled band is responsible for electrical conductivity, rounded by eight cations (Fig. 15C.9). In this structure, an
as explained in Topic 15E. ion of one charge occupies the centre of a cubic unit cell with
eight counter-ions at its corners. The cell is electrically neu-
tral because the ions at the corners of the cell are shared be-
Exercises tween eight cells and so contribute one eighth of their charge
E15C.1 Calculate the packing fraction for close-packed cylinders; you need to each. The structure shown in Fig. 15C.9 is adopted by CsCl
only consider one layer. Hint: Start by identifying a suitable unit cell.
itself and also by CaS.
E15C.2 Calculate the packing fractions of (i) a primitive cubic unit cell, (ii) a When the radii of the ions differ more than they do in CsCl,
bcc unit cell, (iii) an fcc unit cell, where each cell is composed of identical hard
even eight-coordinate packing cannot be achieved. One com-
spheres. Hint: Start by identifying the unit cell and working out which atoms
are in contact. mon structure adopted is the (6,6)-coordinate rock salt struc-
ture typified by rock salt itself, NaCl (Fig. 15C.10). In this
structure, each cation is surrounded by six anions and each
anion is surrounded by six cations. The rock salt structure can
15C.2 Ionic solids be pictured as consisting of two interpenetrating slightly ex-
panded cubic F (fcc) arrays, one composed of cations and the
An ionic solid consists of cations and anions held together by
electrostatic interactions. Two key questions arise in consider-
ing such solids: the relative locations adopted by the ions and Cl–
the energetics of the resulting structure.
Na+
15C.2(a) Structure
When crystals of compounds of monatomic ions (such as NaCl
and MgO) are modelled by stacks of hard spheres it is neces-
sary to allow for the possibility that the ions have different radii Figure 15C.10 The rock salt (NaCl) structure consists of two
(typically with the cations smaller than the anions) and differ- mutually interpenetrating slightly expanded face-centred cubic
ent electrical charges. The coordination number of an ion is the arrays of ions. The assembly shown here is a unit cell.
678 15 Solids
S2– Other scales are also available (such as one based on F− for dis-
cussing halides), and it is essential not to mix values from differ-
Zn2+ ent scales. Ionic radii are so arbitrary that predictions based on
them must be viewed cautiously.
Brief illustration 15C.2
From the values of ionic radii in the Resource section, the radius
ratio for MgO is
Figure 15C.11 The structure of the sphalerite form of ZnS
showing the location of the Zn ions in half the tetrahedral holes radius of Mg 2

formed by the fcc array of S ions. 72 pm
0.51
other of anions. This structure is adopted by NaCl itself and 140 pm

also by several other MX compounds, including KBr, AgCl, radius of O2
MgO, and ScN. which is consistent with the observed rock salt structure of
The switch from the caesium chloride structure to the rock MgO crystals.
salt structure is related to the value of the radius ratio, γ:
rsmaller
Radius ratio [definition] (15C.2) 15C.2(b) Energetics
rlarger
The lattice energy of a solid is the change in potential energy
where rsmaller is the radius of the smaller ions in the crystal and when the ions go from being packed together in a solid to being
rlarger that of the larger ions. The radius-ratio rule, which is de- widely separated as a gas. All lattice energies are positive; a high
rived by considering the geometrical problem of packing the lattice energy indicates that the ions interact strongly with one
maximum number of hard spheres of one radius around a hard another to give a tightly bonded solid. The lattice enthalpy,
sphere of a different radius, can be summarized as follows: ∆HL, is the change in standard molar enthalpy for the process
MX(s) → M+(g) + X−(g) and its equivalent for other charge
Radius ratio Structural type types and stoichiometries. At T = 0 the lattice enthalpy is equal
γ < 21/2 − 1 = 0.414 sphalerite (Fig. 15C.11) to the lattice energy; at normal temperatures they differ by only
0.414 < γ < 3 1/2
− 1 = 0.732 rock salt (Fig. 15C.10) a few kilojoules per mole, an amount so small compared to
γ > 0.732 caesium chloride (Fig. 15C.9) typical lattice energies that the difference is normally neglected.
Each ion in a solid experiences favourable (energy-lowering)
The deviation of a structure from that expected on the basis of electrostatic interactions from all the other oppositely charged
this rule is often taken to be an indication of a shift from ionic ions and unfavourable (energy-raising) interactions from all
towards covalent bonding. A major source of unreliability, the other like-charged ions. The total Coulomb potential en-
though, is the arbitrariness of ionic radii (as explained in a mo- ergy is the sum of all the electrostatic contributions. Each cat-
ment) and their variation with coordination number. ion is surrounded by anions, and so there is a large negative
Experimental measurements give the distances between the contribution to the potential energy from the interaction of op-
centres of two ions, and a decision has to be made about how to posite charges. Beyond those nearest neighbours, there are cati-
apportion this difference between the two ions. One approach is ons that contribute a positive term to the total potential energy
simply to assign a value to the radius of one ion and then use this of the central cation. There is also a negative contribution from
value to infer the radii of other ions. A scale based on the value the anions beyond those cations, a positive contribution from
140 pm for the radius of the O2− ion is widely used (Table 15C.2). the cations beyond them, and so on, to the edge of the solid.
These favourable and unfavourable interactions become pro-
Table 15C.2 Ionic radii, r/pm⋆ gressively weaker as the distance from the central ion increases,
but the net outcome of all these contributions is dominated by
Na+ 102(6‡), 116(8) the interaction between nearest neighbours and is therefore a
K+ 138(6), 151(8) lowering of the potential energy.
F− 128(2), 131(4) First, consider a simple one-dimensional model of an ionic
Cl −
181 (close packing) solid that consists of a long line of uniformly spaced alternating
⋆
cations and anions; the distance between neighbouring centres
This scale is based on a value 140 pm for the radius of the O2− ion. More values are given
in the Resource section.
is d, the sum of the ionic radii (Fig. 15C.12). If the charge num-
‡
Coordination number. bers of the ions have the same absolute value (+1 and −1, or +2
+z –z +z –z +z –z Table 15C.3 Madelung constants
d Structural type A
Caesium chloride 1.763
Figure 15C.12 A one-dimensional model of an ionic solid
Fluorite 2.519
consisting of a line of alternating cations and anions.
Rock salt 1.748
2
and −2, for instance), then z1 = +z, z2 = −z, and z1z2 = −z . The Rutile 2.408
potential energy of the central ion is calculated by summing all Sphalerite (zinc blende) 1.638
the terms, with negative terms representing favourable inter- Wurtzite 1.641
actions with oppositely charged ions and positive terms repre-
senting unfavourable interactions with like-charged ions. For
the interaction with ions extending in a line to the right of the
central ion, the contribution of the Coulomb interaction to the
lattice energy is Overlap contribution
Potential energy, Ep
1 ze 2 2
ze ze 2 2
ze2 2
2 2
Ep Lattice parameter, d
4 0 d 2d 3d 4d 0
ln 2
2 2
ze
1 12 13 14
Total

4 0 d
Coulombic contribution
z 2e2
ln 2
4 0 d
To complete the calculation Ep is multiplied by 2 to obtain the Figure 15C.13 The contributions to the total potential energy of
total energy arising from interactions on both sides of the ion, an ionic crystal.
and then by Avogadro’s constant, NA, to obtain an expression
for the Coulomb contribution to the (molar) lattice energy. The
outcome is where d is the distance between the atoms, and C′ and d * are
2
z N Ae 2 constants. It turns out that the value of C′ is not needed (it can-
Ep 2 ln 2 cels in expressions that make use of this formula; see below);
4 0 d
d * is commonly taken to be 34.5 pm. The total potential energy
with d = rcation + ranion. This energy is negative, corresponding to is the sum of Ep and Ep*, and passes through a minimum when
a net favourable interaction. The calculation can be extended to d(Ep Ep* )/dd 0 (Fig. 15C.13).
three-dimensional arrays of ions with different charge numbers A short calculation leads to the Born–Mayer equation for
zA and zB: the minimum total potential energy (see Problem P15C.9):
| zA zB | N Ae2
Ep A (15C.3) N | z z | e2  d*  Born–Mayer
4 0 d Ep,min =
− A A B 1 −  A (15C.5)
4 πε 0d  d  equation
The factor A is a positive numerical constant called the Madelung
constant; its value depends on how the ions are arranged in the Provided zero-point contributions to the energy are ignored,
crystal. For a rock salt structure, A = 1.748; Table 15C.3 lists the negative of this potential energy can be identified with the
Madelung constants for other common structures. lattice energy. The important features of this equation are:
The Coulomb interaction is not the only contribution to the
lattice energy. When atomic orbitals overlap to form bonding • Because Ep,min ∝ |zAzB|, the potential energy decreases
and antibonding molecular orbitals and both kinds of orbitals (becomes more negative) with increasing charge number
are full, there is an increase in energy because the antibond- of the ions.
ing orbital is raised in energy more than the bonding orbital is • Because the electrostatic (and dominant) contribution to
lowered (Topic 9D). This positive contribution to the potential Ep,min is proportional to 1/d, the potential energy decreases
energy depends on the overlap of the atomic orbitals, and, be- (becomes more negative) with decreasing ionic radius.
cause orbitals decay exponentially with distance, at large dis-
The second conclusion follows from the fact that the smaller
tances from the nucleus it is often modelled by writing
the ionic radii, the smaller is the value of d. High lattice ener-

Ep N A C e d /d (15C.4) gies are expected when the ions are highly charged (so |zAzB| is
large) and small (so d is small).
680 15 Solids
Brief illustration 15C.3 cycle to be useful, experimental values for the enthalpy
changes for all the steps need to be available (for example,
To estimate Ep,min for MgO, which has a rock salt structure from tabulated data), apart, of course, from the step involv-
(A = 1.748), use the following values: d = r(Mg2+) + r(O2−) = 72 + ing the formation of the lattice from the ions. For the cycle in
140 pm = 212 pm. Note that Fig. 15C.14 the enthalpy changes are (for convenience, starting
at the elements):
23 1 19 2
N Ae 2 (6.022 14 10 mol ) (1.602 176 10 C)

4 0 4 (8.854 19 1012 J 1 C 2 m 1 )
∆H/(kJ mol−1)
4 1
1.38935 10 Jm mol 1. Sublimation of K(s) +89 [dissociation enthalpy of
K(s)]
Then 2. Dissociation of 12 Cl 2 (g ) +122 [ 12 × dissociation enthalpy
of Cl2(g)]
|z z |
O Mg 2 2
N A e2/ 4 0 3. Ionization of K(g) +418 [ionization enthalpy of

4 4 K(g)]
Ep, min (1 . 38935 10 J m mol 1 )
2.12 1010 m 4. Electron attachment to −349 [electron-gain enthalpy of
Cl(g) Cl(g)]
d

1
d /
d 5. Formation of solid from −∆HL/(kJ mol−1) [value to be determined]
A gaseous ions
34.5 pm
1 1.748 6. Decomposition of +437 [negative of enthalpy of
212 pm
compound to its formation of KCl(s)]
elements in their
3.84 103 kJmol 1 reference states
The cycle is closed, so the sum of these enthalpy changes is

It is not possible to measure the lattice enthalpy directly, but equal to zero, and the lattice enthalpy can be inferred from the
values can be obtained by combining experimental values of resulting equation.
other enthalpy changes by using a Born–Haber cycle. Such a
The solution The sum of contributions around the cycle is
cycle is a closed path of transformations starting and ending at
the same point, one step of which is the formation of the solid 89 122 418 349 H L /(kJmol 1 ) 437 0
compound from a gas of widely separated ions.
It follows that ∆HL = +717 kJ mol−1.
Self-test 15C.2 Calculate the lattice enthalpy of CaO from the
Example 15C.2 Using the Born–Haber cycle following data:
Calculate the lattice enthalpy of KCl.

∆H/(kJ mol−1)
Collect your thoughts You need to construct a suitable Born–
Sublimation of Ca(s) +178
Haber cycle, such as the one shown in Fig. 15C.14. For the 2+
Ionization of Ca(g) to Ca (g) +1735
K (g) + e (g) + Cl(g)
+ – Dissociation of O2 (g )
1
2 +249
Electron attachment to O(g) −141
–349 4 Electron attachment to O (g) −
+844
3 +418
K+(g) + Cl–(g) Formation of CaO(s) from Ca(s) and 12 O2 (g ) −635
K(g) + Cl(g)
2 +122 Answer: +3500 kJ mol−1
K(g) + ½Cl2(g)
1 +89 –ΔHL
K(s) + ½Cl2(g) 5
6 +437 Some lattice enthalpies obtained by the Born–Haber cycle

KCl(s) are listed in Table 15C.4. As can be seen from the data, the
trends in values are in general accord with the predictions
Figure 15C.14 The Born–Haber cycle for KCl at 298 K. Enthalpy of the Born–Mayer equation. Agreement is typically taken
changes are in kilojoules per mole. to imply that the ionic model of bonding is valid for the
Table 15C.4 Lattice enthalpies at 298 K, ∆HL/(kJ mol−1)⋆ Brief illustration 15C.4
NaF 926
Diamond and graphite are two allotropes of carbon. In dia-
NaBr 752
mond each sp3-hybridized carbon is bonded tetrahedrally to
MgO 3850 its four neighbours (Fig. 15C.15). The network of strong C−C
MgS 3406 bonds is repeated throughout the crystal and, as a result, dia-
⋆
More values are given in the Resource section. mond is very hard (in fact, the hardest known substance). In
graphite, σ bonds between sp2-hybridized carbon atoms form
hexagonal rings which, when repeated throughout a plane,
s ubstance; disagreement implies that there is a covalent contri- give rise to ‘graphene’ sheets (Fig. 15C.16). The sheets can slide
bution to the bonding. It is important, though, to be cautious, against each other when impurities are present; as a result,
because numerical agreement might be coincidental and, as impure graphite is used widely as a lubricant.
noted above, the values for ionic radii are subject to significant
uncertainty.
Exercises
E15C.3 Determine the radius of the smallest cation that can have (i) sixfold
and (ii) eightfold coordination with the Cl− ion (radius 181 pm).
E15C.4 Does titanium expand or contract as it transforms from hcp to bcc?
The atomic radius of titanium is 145.8 pm in hcp but 142.5 pm in bcc. Hint:
Consider the change in packing fraction.
E15C.5 Calculate the lattice enthalpy of MgO from the following data:
Figure 15C.15 A fragment of the structure of diamond. Each C

ΔH/(kJ mol−1) atom is tetrahedrally bonded to four neighbours. This framework-
Sublimation of Mg(s) +15 like structure results in a rigid crystal.
2+
Ionization of Mg(g) to Mg (g) +2188
Dissociation of O2(g) +249
Electron attachment to O(g) −141
Electron attachment to O−(g) +844
Formation of MgO(s) from Mg(s) and −602
2 O2 (g )
1
in their reference states.
15C.3 Covalent and molecular solids

X-ray diffraction studies of solids reveal a huge amount of in-
formation, including interatomic distances, bond angles, ste-
reochemistry, and vibrational parameters. This section can do (a) (b)
no more than hint at the diversity of types of solids found when Figure 15C.16 Graphite consists of flat planes of hexagons of
molecules pack together or atoms link together in extended carbon atoms lying above one another. (a) The arrangement of
networks. carbon atoms in a ‘graphene’ sheet; (b) the relative arrangement of
In covalent solids (or covalent network solids), covalent neighbouring sheets. The planes can slide over one another easily
bonds in a definite spatial orientation link the atoms together when impurities are present.
into a network that extends through the crystal; effectively the
crystal is a giant molecule. The demands of directional bond-
ing, which have only a small effect on the structures of many
metals, now override the geometrical problem of packing Molecular solids, which are the subject of the over-
spheres together, resulting in a wide variety of often quite elab- whelming majority of modern structural determinations,
orate structures. are held together by van der Waals interactions between
682 15 Solids
the individual molecular components (Topic 14B). The ob-

served crystal structure is nature’s solution to the problem
of condensing objects of various shapes into an aggregate
of minimum energy (actually, for T > 0, of minimum Gibbs
energy). The prediction of the structure is difficult, but soft-
ware specifically designed to explore interaction energies
can now make reasonably reliable predictions. The problem
is made more complicated by the role of hydrogen bonds,
which in some cases dominate the crystal structure, as in ice
(Fig. 15C.17), but in others (for example, in solid phenol)
distort a structure that is determined largely by the van der
Figure 15C.17 A fragment of the crystal structure of ice (ice-I). Waals interactions.
Each O atom (shown as coloured spheres) is at the centre of a
tetrahedron of four O atoms at a distance of 276 pm. The central
O atom is attached by two short O−H bonds to two H atoms
(shown as white spheres) and by two long hydrogen bonds to
the H atoms of two of the neighbouring molecules. For each O−O
separation both of the alternative H atom locations are shown.
Overall, the structure consists of planes of hexagonal puckered
rings of H2O molecules (like the chair form of cyclohexane).
☐ 1. A close-packed layer is a layer of spheres arranged so ☐ 8. A band gap is a range of energies to which no orbital
there is maximum utilization of space. corresponds.
☐ 2. A hexagonally close-packed structure is one where the ☐ 9. The Fermi level is the highest occupied molecular
sequence of close-packed layers is ABABAB …. orbital at T = 0.
☐ 3. A cubic close-packed structure is one where the ☐ 10. The coordination number of an ionic lattice is denoted
sequence of close-packed layers is ABCABC …. (N+,N−), where N+ is the number of nearest neighbour
☐ 4. The coordination number is the number of spheres anions around a cation and N− the number of nearest
immediately surrounding any selected sphere. neighbour cations around an anion.
☐ 5. In the nearly free-electron approximation the valence ☐ 11. The lattice energy of a solid is the change in potential
electrons are assumed to be trapped in a box with a peri- energy when the ions go from being packed together in
odic potential energy, with low energy corresponding to a solid to being widely separated as a gas.
the locations of cations. ☐ 12. A Born–Haber cycle is a closed path of transformations
☐ 6. In the tight-binding approximation the valence elec- starting and ending at the same point, one step of which
trons are assumed to occupy molecular orbitals delocal- is the formation of the solid compound from a gas of
ized throughout the solid. widely separated ions.
☐ 7. In metals atomic orbitals overlap to form a band, which ☐ 13. A molecular solid is a solid consisting of discrete mol-
is a set of molecular orbitals that are closely spaced and ecules held together by van der Waals interactions, and
cover a finite range of energy; electrons occupy the possibly hydrogen bonds.
orbitals within the band.
Energy levels of a linear array of orbitals Ek = α + 2β cos{kπ/(N+1)}, k = 1,2,…, N Hückel approximation 15C.1
Band width EN − E1 → −4β as N → ∞ Hückel approximation
Radius ratio γ = rsmaller/rlarger For criteria, see Section 15C.2 15C.2
Born–Mayer equation Ep,min = −{NA|zAzB|e2/4πε0d}(1−d⋆/d)A 15C.5
TOPIC 15D The mechanical properties
of solids

An understanding of the mechanical properties of
solid materials is crucial to the development of modern
(a) (b)
materials.

The mechanical properties of solids are expressed in terms
of various ‘moduli’ that are related to the intermolecular
(c)
potential energy of the constituents.
Figure 15D.1 Types of stress applied to a body. (a) Uniaxial stress,
➤ What do you need to know already?
(b) pure shear stress, (c) hydrostatic pressure.
You need to be familiar with the Lennard-Jones potential
energy (Topic 14B).
Yield point
Plastic
deformation
The fundamental concepts needed for the discussion of the me-
Stress
chanical properties of solids are stress and strain. The stress on

an object is the applied force divided by the area to which the
force is applied. For instance, if a mass m hangs from a wire Elastic deformation
of radius r, and therefore cross-section πr2, the mass exerts a
gravitational force mg on the wire. The uniaxial stress along the
length of the wire would be reported as mg/πr2. The strain is Strain
the resulting fractional distortion of the object. The study of the
relation between stress and strain is called rheology from the Figure 15D.2 At small strains, a body obeys Hooke’s law (stress
proportional to strain) and is elastic (recovers its shape when the
Greek word for ‘flow’.
stress is removed). At high strains, the body is no longer elastic,
Stress may be applied in different ways (Fig. 15D.1):
may become plastic, and finally yield.
• Uniaxial stress is a stress, in this case a compression or
extension, applied in one direction.
threshold, the strain becomes plastic in the sense that recovery
• Hydrostatic stress is a stress applied simultaneously in all does not occur when the stress is removed. Plastic deformation
directions, as in a body immersed in a fluid. occurs when bond breaking takes place and, in pure metals,
• Pure shear is a stress that tends to push opposite faces of typically takes place through the agency of dislocations. Brittle
the sample in opposite directions. materials, such as ionic solids, exhibit sudden fracture as the
stress focused by cracks causes them to spread catastrophically.
A sample subjected to a low stress typically undergoes elastic
The response of a solid to an applied stress is commonly
deformation in the sense that it recovers its original shape
summarized by a number of coefficients of proportionality
when the stress is removed. For low stresses, the strain is lin-
known as moduli:
early proportional to the stress, and the stress–strain relation is
a Hooke’s law of force (Fig. 15D.2). The response becomes non- uniaxial stress
Young’s modulus: E = (15D.1a)
linear at high stresses but may remain elastic. Above a certain uniaxial strain
684 15 Solids
Uniaxial The Young’s modulus of iron at room temperature is 215 GPa.

(normal) Therefore
Transverse strain
strain Shear strain 1.24 1010 kg m 1 s 2
uniaxial strain 0.0581
2.15 1011 kg m 1 s 2

Pa
which corresponds to elongation of the wire by 5.81 per cent.

(a) (b)
Figure 15D.3 (a) Uniaxial stress and the resulting uniaxial and
transverse strain; Poisson’s ratio indicates the extent to which a
body changes shape when subjected to a uniaxial stress. (b) Shear If neighbouring molecules interact by a Lennard-Jones po-
stress and the resulting strain. tential energy (Topic 14B), then the bulk modulus and the
compressibility of the solid are related to the Lennard-Jones pa-
rameter ε (the depth of the potential well) by
pressure
Bulk modulus: K = (15D.1b)
fractional change in volume 8N A 1 V
K T m (15D.4)
shear stress Vm K 8N A
Shear modulus: G = (15D.1c)
shear strain
For the derivation of these relations, see A deeper look 15D.1,
‘Uniaxial strain’ refers to stretching and compression of the
available to read in the e-book accompanying this text. The
material in one direction, as shown in Fig. 15D.3a, and ‘shear
bulk modulus is large and the compressibility low (the solid
strain’ refers to the distortion arising from a shear stress, as de-
stiff) if the potential well is deep and the solid is dense (its
picted in Fig. 15D.3b. The fractional change in volume is δV/V,
molar volume small).
where δV is the change in volume of a sample of volume V;
The differing rheological characteristics of metals can be
similarly, the uniaxial strain and the shear strain are (dimen-
traced to the presence of slip planes, which are planes of
sionless) fractional changes in dimensions. The bulk modulus
atoms that, when under stress, may slip or slide relative to
is the inverse of the isothermal compressibility, κT, discussed in
one another. The slip planes of a ccp structure are the close-
Topic 2D (eqn 2D.7, κT = −(∂V/∂p)T/V).
packed planes, and careful inspection of a unit cell shows
A third ratio, called Poisson’s ratio, indicates how the sample
that there are eight sets of slip planes in different directions.
changes its shape:
As a result, metals with ccp structures, like copper, are mal-
transverse strain leable, meaning they can easily be bent, flattened, or ham-
P on]
Poisson’s ratio [definitio (15D.2)
normal strain mered into shape. In contrast, a hexagonal close-packed
structure has only one set of slip planes so that metals with
Transverse and normal strains are illustrated in Fig. 15D.3a:
hexagonal close packing, such as zinc or cadmium, tend to
they are the mutually perpendicular uniaxial distortions aris-
be more brittle.
ing from the ‘normal’ uniaxial stress. The three moduli intro-
duced in eqn 15D.1 are interrelated in the following way (see
Problem P15D.1):
E E
Exercises
Relations between
G K moduli
(15D.3) E15D.1 The bulk modulus of polystyrene is 3.43 GPa. Calculate the hydrostatic
2(1 P ) 3(1 2 P )
pressure (stress) needed for a sample of this material to change volume by
1.0 per cent.
E15D.2 The Young’s modulus of polystyrene is 4.42 GPa. A polystyrene rod of
Brief illustration 15D.1
diameter 2.0 mm is subject to a force of 500 N along its length. Calculate the
stress and hence the percentage increase in the length of the rod when the
The uniaxial stress when a mass of m = 10.0 kg is suspended stress is applied.
from an iron wire of radius r = 0.050 mm is
E15D.3 Poisson’s ratio for polyethene is 0.45. What change in volume takes
(10.0 kg ) (9.81 ms 2 ) place when a cube of polyethene of volume 1.0 cm3 is subjected to a uniaxial
mg
uniaxial stress stress that produces a strain of 1.0 per cent?
r 2 (5.0 105 m)2
1.24 1010 kg m 1 s 2
15D The mechanical properties of solids 685
☐ 1. Uniaxial stress is a simple stress (compression or exten- ☐ 5. The response of a solid to an applied stress is summa-
sion) applied to a sample in one direction. rized by the Young’s modulus, the bulk modulus, the
☐ 2. Hydrostatic stress is a stress applied simultaneously in shear modulus, and Poisson’s ratio.
all directions, as in a body immersed in a fluid. ☐ 6. The differing rheological characteristics of metals can be
☐ 3. Pure shear is a stress that tends to push opposite faces of traced to the presence of slip planes.
the sample in opposite directions.
☐ 4. A sample subjected to a small stress typically undergoes
elastic deformation; as the stress increases the sample
becomes plastic.
Young’s modulus E = uniaxial stress/uniaxial strain Definition 15D.1a
Bulk modulus K = pressure/fractional change in volume Definition 15D.1b
Shear modulus G = shear stress/shear strain Definition 15D.1c
Poisson’s ratio νP = transverse strain/normal strain Definition 15D.2
TOPIC 15E The electrical properties
of solids
• A metallic conductor is a substance with an electrical

➤ Why do you need to know this material? conductivity that decreases as the temperature is raised.
The electrical properties of solids underlie numerous tech- • A semiconductor is a substance with an electrical conduc-
nological applications on which the infrastructure of the tivity that increases as the temperature is raised.
modern world depends. • A superconductor is a solid that, below a critical tempera-
ture, conducts electricity without resistance.
Electrons in solids occupy bands that determine the elec- A semiconductor generally has a lower conductivity than that
trical conductivities of various types of solid. typical of metallic conductors, but the magnitude of the con-
ductivity is not the criterion for distinguishing between them.
➤ What do you need to know already? It is conventional to classify semiconductors with very low
You need to be familiar with the formation of bands in electrical conductivities, such as most synthetic polymers, as
solids (Topic 15C). insulators. The term is one of convenience rather than one of
fundamental significance.
The electrical conductivity of common materials arises from 15E.1 Metallic conductors
the motion of electrons, but some ionic solids display ionic
conductivity in which whole ions migrate through the lattice. To understand the origins of the electric conductivity in
Three types of solid are distinguished by the temperature de- conductors and semiconductors, it is necessary to explore the
pendence of their electrical conductivity (Fig. 15E.1): consequences of the formation of bands (Topic 15C). The start-
ing point is Fig. 15C.8, which is repeated here for convenience
as Fig. 15E.2. It shows the electronic structure of a solid formed
from a line of N atoms, each of which contributes one electron
108
Metallic conductor (such as the alkali metals). At T = 0, only the lowest 12 N molec-
ular orbitals are occupied, up to the Fermi level. The levels are
Superconductor
very closely spaced, so there are unoccupied molecular orbitals
Conductivity/(S cm–1)
104
just above the Fermi level. A solid with a partially filled band is
1
Semiconductor
10–4
Unoccupied levels
Energy
10–8
0 10 100 1000 Fermi level
Temperature, T/K
Occupied levels
Figure 15E.1 The variation of the electrical conductivity of a
substance with temperature is the basis of its classification as
a metallic conductor, a semiconductor, or a superconductor.
Conductivity is expressed in siemens per metre (S m−1 or, as here, Figure 15E.2 (A reproduction of Fig. 15C.8.) When N electrons
S cm−1), where 1 S = 1 Ω−1 (the resistance is expressed in ohms, Ω); occupy a band of N orbitals at T = 0, it is only half full. The highest
note the log scale. occupied level is the Fermi level.
15E The electrical properties of solids 687
Conduction
Energy
Fermi level band
Band
Energy
Thermal
gap, Eg excitation
(a) (b) (c)

Valence
Figure 15E.3 Each part of this illustration separates out the band
waves travelling in opposite directions. (a) The two sets of waves
have the same energy, are equally occupied, and there is no net (a) T = 0 (b) T > 0
motion. (b) When a potential difference is applied (positive on the
right) the two sets no longer have the same energy. There are now Figure 15E.4 (a) Typical band structure for a semiconductor:
more electrons moving to the right than to the left and therefore at T = 0 the valence band is full and the conduction band is
there is a net current. (c) If the band is full, the two sets remain empty. (b) At higher temperatures electrons populate the levels
equally populated and there is no net flow even when a potential of the conduction band leading to electrical conductivity which
difference is applied. increases with temperature.
at a higher energy, separated from the top of the lower band by

expected to be a metallic conductor, an observation which can an energy Eg, known as the band gap. At T = 0 this material is
be understood in the following way. an insulator (formally, a low conductivity semiconductor) be-
The key point is that each molecular orbital in a band can cause there are no partially filled bands. If the temperature is
be regarded as the superposition of two waves travelling in high enough electrons are excited out of the lower band into
opposite directions (in the same sense that cos x ∝ eix + e−ix). the upper. There are now incomplete bands and conduction can
Figure 15E.3a is an adaptation of Fig. 15E.2 in which the waves occur.
are separated into two components according to their direction The question that now arises is how the populations of the
of travel. In the absence of any applied field, electrons occupy two bands, and therefore the conductivity of the semiconduct-
both components equally and there is no net motion through ing material, depend on the temperature. The discussion starts
the solid. This absence of net motion is true whether the band by introducing the density of states, ρ(E), defined such that the
is full or incomplete. When a potential difference is applied the number of states between E and E + dE is ρ(E)dE. Note that
energies of the components differ, as it is energetically favour- the ‘state’ of an electron includes its spin, so each spatial orbital
able for electrons to travel towards regions of positive poten- counts as two states. To obtain the number of electrons dN(E)
tial. Now the two components are no longer equally occupied that occupy states between E and E + dE, ρ(E)dE is multiplied
(Fig. 15E.3b) and provided the band is not full there are more by the probability f(E) of occupation of the state with energy E.
electrons travelling in one direction than the other and electri- That is,
cal conduction occurs. If the band is full, however, the popu-
Number of
lations of the two components remain equal (Fig. 15E.3c) and states
Probability of
occupation of
there is no net motion in either direction. The material does not between a state with
E and E dE
conduct: it is an insulator.
energy E
The conductivity of a metallic conductor decreases as the dN (E ) (E )dE f (E ) (15E.1)

temperature is raised. This decrease is due to collisions between
the moving electrons and the atoms. The higher the temper- The function f(E) is the Fermi–Dirac distribution, a version of
ature, the more vigorous is the thermal motion of the atoms, the Boltzmann distribution that takes into account the Pauli ex-
so collisions between the moving electrons and an atom are clusion principle, that each orbital can be occupied by no more
more likely. That is, the electrons are scattered out of their paths than two electrons:
through the solid, and are less efficient at transporting charge.
1
f (E) = ( E − μ )/kT
Fermi–Dirac distribution (15E.2a)
e +1
In this expression μ is a temperature-dependent parameter
15E.2 Insulators and semiconductors known as the ‘chemical potential’ (it has a subtle relation to the
familiar chemical potential of thermodynamics); provided T > 0,
Now consider a solid which has an arrangement of bands as μ is the energy of the state for which f = 12 . At T = 0, only states
shown in Fig. 15E.4. At T = 0 the lower band is full, and the up to a certain energy, known as the Fermi energy, EF, are
Fermi level lies at the top of the band. A second empty band lies occupied. Provided the temperature is not so high that many
688 15 Solids
electrons are excited to states above the Fermi energy, the the conductivity increases as more electrons are promoted
chemical potential can be identified with EF, in which case the across the band gap, so the material is a semiconductor.
Fermi–Dirac distribution becomes The lower band, which is full at T = 0, is called the valence
1 band and the upper band, which is empty at T = 0 and to which
f (E) = Fermi–Dirac distribution (15E.2b) electrons are thermally excited, is called the conduction band.
e( E − EF )/kT + 1
When electrons leave the valence band they can be thought
This expression implies that f (EF ) = 12 . For energies well above of as creating positively charged ‘holes’ in that band (i.e. the
EF, the exponential term is so large that the 1 in the denomina- absence of an electron), and the electrical conductivity arises
tor can be neglected, and then from the movement of these holes and the promoted electrons.
Fermi–Dirac distribution Figure 15E.4 depicts an intrinsic semiconductor, in which
f (E ) ≈ e −( E − EF )/kT (15E.2c)
[approximate form for E > EF ] semiconduction is a property of the band structure of the pure
material. Examples of intrinsic semiconductors include silicon
The function now resembles a Boltzmann distribution, decay-
and germanium. A compound semiconductor is an intrinsic
ing exponentially with increasing energy; the higher the tem-
semiconductor formed from a combination of different ele-
perature, the longer is the exponential tail.
ments, such as GaN, CdS, and many d-metal oxides.
There is a distinction between the Fermi energy and the
An extrinsic semiconductor is one in which charge carriers
Fermi level:
(electrons or holes) are present as a result of the replacement of
• The Fermi level is the uppermost occupied level at T = 0. some atoms (to the extent of about 1 in 109) by dopant atoms, the
• The Fermi energy is the energy level at which f (E ) = 12 at atoms of another element. If, for example, pure silicon (a Group
any temperature. 14 element) is doped with atoms of indium (a Group 13 element)
an electron can be transferred from a Si atom to a neighbouring
The Fermi energy coincides with the Fermi level as T → 0. In atom, thereby creating a hole in the valence band and increas-
Figure 15E.5 shows the form of f(E) at different temperatures. ing the conductivity. The resulting semiconductor is described
At T = 0 the probability distribution is a step function, equal to as p-type, the p indicating that the positive holes are responsible
1 for E < EF, and 0 at higher energies, as in Fig. 15E.2. At higher for conduction. Figure 15E.6a shows the band structure of such
temperatures the probability of occupation of levels above EF a semiconductor. The dopant atoms result in a set of empty lev-
increases at the expense of those below EF, with the greatest els, called acceptor levels, which lie just above the top of the va-
changes occurring in the energies close to EF. As the tempera- lence band. Electrons from the valance band are transferred into
ture is raised, electrons are promoted from the lower band to the these levels, thereby generating holes in the band.
upper. This promotion is represented by the tail of the Fermi– If the dopant atoms are from a Group 15 element (e.g. phos-
Dirac distribution extending across the band gap and is signifi- phorus), an electron can be transferred from a P atom into the
cant only when kT is comparable to or greater than the band gap. otherwise empty conduction band, thereby increasing the con-
The material, an insulator at T = 0, is now a conductor, because ductivity. This type of doping results in an n-type semiconductor,
both bands are partially filled. As the temperature is increased,
1
T= 0
Probability of occupation, f
0.8 Acceptor
Energy
Donor
levels
levels
0.6
1/10
0.4
1/3
0.2
1 (a) (b)
3
10
0 Figure 15E.6 (a) A dopant with fewer electrons than its host
–6 –4 –2 0 2 4 6
(E – EF)/ EF contributes levels that accept electrons from the valence band.
The resulting holes in the band give rise to electrical conductivity;
Figure 15E.5 The Fermi–Dirac distribution, which gives the the doped semiconductor is classified as p-type. (b) A dopant
probability of occupation of a state with energy E and at a with more electrons than its host contributes occupied levels that
temperature T. At higher energies the probability decays can supply electrons to the conduction band, thus giving rise
exponentially towards zero. The curves are labelled with the value to electrical conductivity; the substance is classed as an n-type
of EF/kT. semiconductor.
15E The electrical properties of solids 689
where n refers to the negative charge of the carriers. The band recombination stimulates lattice vibrations. This is the case for
structure is shown in Fig. 15E.6b. The dopant atoms create a silicon semiconductors, and is one reason why computers need
set of occupied levels, called donor levels, just below the bot- efficient cooling systems.
tom of the conduction band, and electrons from these levels are Another electronic device, a ‘transistor’, consists of a p-type
transferred into the conduction band. In practical cases the level semiconductor sandwiched between two n-type semiconduc-
of doping is such that the charge carriers created by the dopant tors, and as such has two p–n junctions. Under the correct condi-
atoms are greatly in excess of those arising from thermal exci- tions it is possible to control the current flowing between the two
tation across the band gap: electrical conductivity is therefore n-type semiconductors by varying the current flowing into the
dominated by the type and extent of the doping. p-type semiconductor. Most significantly, the change in the current
Doped semiconductors are of great technological impor- between the n-type semiconductors can be larger than the change
tance because they are the materials from which the active com- in the current in the p-type semiconductor; in other words the de-
ponents of electronic circuits are made. The simplest example of vice can act as an amplifier. It is the exploitation of this property
an electronic device constructed from doped semiconductors that has led to the development of modern solid-state electronics.
is the ‘p–n diode’ which consists of a p-type semiconductor in
contact with an n-type semiconductor, thereby creating a p–n
junction. A p–n junction conducts electricity only in one di-
Exercises
rection. To understand this property consider first the arrange- E15E.1 Calculate f(E), the probability predicted by the Fermi–Dirac
distribution, for the case where the energy E is the thermal energy, kT, above
ment shown in Fig. 15E.7a where the p-type semiconductor is the Fermi energy: E = EF + kT.
attached to the negative electrode and the n-type is attached to
E15E.2 Is arsenic-doped germanium a p-type or n-type semiconductor?
the positive electrode; this arrangement is known as ‘reverse
bias’. The positively charged holes in the p-type semiconduc-
tor are attracted to the negative electrode, and the negatively
charged electrons in the n-type semiconductor are attracted
to the positive electrode. As a consequence, charge does not 15E.3 Superconductors
flow across the junction so the device does not conduct. Now
consider what happens when the electrodes are connected the The resistance to flow of electrical current of a normal metallic
other way round, as shown in Fig. 15E.7b, an arrangement conductor decreases smoothly with decreasing temperature but
known as ‘forward bias’. Electrons in the n-type semiconduc- never vanishes. However, a superconductor conducts electric-
tor move towards the positive electrode, and holes move in the ity without resistance once the temperature is below the critical
opposite direction: as a result charge flows across the junction. temperature, Tc. Following the discovery in 1911 that mercury
The p–n junction therefore conducts only under forward bias. is a superconductor below 4.2 K, the normal boiling point of
As electrons and holes move across a p–n junction under liquid helium, physicists and chemists made slow but steady
forward bias, they recombine and release energy. However, as progress in the discovery of superconductors with higher val-
long as the forward bias persists, the flow of charge from the ues of Tc. Metals, such as tungsten, mercury, and lead, have Tc
electrodes to the junction replenishes them with electrons and values below about 10 K. Intermetallic compounds, such as
holes. In some solids, the energy of electron–hole recombi- Nb3X (X = Sn, Al, or Ge), and alloys, such as Nb/Ti and Nb/Zr,
nation is released as heat and the device becomes warm. The have intermediate Tc values ranging between 10 K and 23 K. In
reason lies in the fact that the return of the electron to a hole in- 1986, high-temperature superconductors (HTSCs) were dis-
volves a change in the electron’s linear momentum, which the covered. Several ceramics, inorganic powders that have been
atoms of the lattice must absorb, and therefore electron–hole fused and hardened by heating to a high temperature, contain-
ing oxocuprate motifs, CumOn, are now known with Tc values
n p n p well above 77 K, the boiling point of the inexpensive refrigerant
liquid nitrogen. For example, HgBa2Ca2Cu3O8 has Tc = 153 K.
Electron Hole The elements that exhibit superconductivity cluster in cer-
tain parts of the periodic table. The metals iron, cobalt, nickel,
+ – – + copper, silver, and gold do not display superconductivity; nor
do the alkali metals. One of the most widely studied oxocu-
prate superconductors YBa2Cu3O7 (informally known as ‘123’
on account of the proportions of the metal atoms in the com-
pound) has the structure shown in Fig. 15E.8. The square-
(a) (b)
pyramidal CuO5 units arranged as two-dimensional layers and
Figure 15E.7 A p–n junction under (a) reverse bias, the square planar CuO4 units arranged in sheets are common
(b) forward bias. structural features of oxocuprate HTSCs.
690 15 Solids
Cu O
Ba
e–
Y
(a) (b)
Figure 15E.9 The formation of a Cooper pair. One electron
Figure 15E.8 Structure of the YBa2Cu3O7 superconductor. (a) Metal distorts the crystal lattice and the second electron has a lower
atom positions. (b) The polyhedra show the positions of oxygen energy if it goes to that region. These electron–lattice interactions
atoms and indicate that the Cu ions are either in square-planar or effectively bind the two electrons into a pair.
square-pyramidal coordination environments.
The mechanism of superconduction is well-understood for e lectron as it travels through the solid because the distortion
low-temperature materials, and is based on the properties of a caused by one electron can attract back the other electron
Cooper pair, a pair of electrons that exists on account of the should it be scattered out of its path in a collision. Because the
indirect electron–electron interactions mediated by the nuclei Cooper pair is stable against scattering, it can carry charge freely
of the atoms in the lattice. Thus, if one electron is in a particu- through the solid, and hence give rise to superconduction.
lar region of a solid, the nuclei there move towards it to give The Cooper pairs responsible for low-temperature supercon-
a distorted local structure (Fig. 15E.9). That local distortion is ductivity are likely to be important in HTSCs, but the mecha-
rich in positive charge, so it is favourable for a second electron nism for pairing is hotly debated. There is evidence implicating
to join the first. Hence, there is a virtual attraction between the the arrangement of CuO5 layers and CuO4 sheets in the mecha-
two electrons and they move together as a pair. The local dis- nism. It is believed that movement of electrons along the linked
tortion is disrupted by thermal motion of the ions in the solid, CuO4 units accounts for superconductivity, whereas the linked
so the virtual attraction occurs only at very low temperatures. CuO5 units act as ‘charge reservoirs’ that maintain an appropri-
A Cooper pair undergoes less scattering than an individual ate number of electrons in the superconducting layers.
☐ 1. Electronic conductors are classified as metallic conduc- ☐ 5. In a semiconductor at T = 0 there is a full valence band
tors or semiconductors according to the temperature and, at higher energy, an empty conduction band.
dependence of their conductivities; an insulator is a ☐ 6. In an intrinsic semiconductor electrical conductivity is
semiconductor with very low conductivity. due to electrons thermally promoted from the valence
☐ 2. Superconductors conduct electricity without resistance band to the conduction band.
below a critical temperature Tc. ☐ 7. In an extrinsic semiconductor electrical conductivity is
☐ 3. The Fermi–Dirac distribution gives the probability due to electrons or holes generated by the inclusion of
that a state with a particular energy is occupied by an dopant atoms.
electron. ☐ 8. Semiconductors are classified as p-type or n-type
☐ 4. The Fermi energy is the energy of the level for which the according to whether conduction is due to holes in the
probability of occupation is 12 . valence band or electrons in the conduction band.
( E EF ) /kT
Fermi–Dirac distribution f (E ) 1/{e 1} EF is the Fermi energy 15E.2b
TOPIC 15F The magnetic properties
of solids
where χ is the dimensionless volume magnetic susceptibility.

➤ Why do you need to know this material? A closely related quantity is the molar magnetic susceptibil-
The magnetic properties of solids give an indication of the ity, χm:
electronic structures of individual molecules and many m Vm Molar magnetic susceptibility [definition] (15F.2)
modern information storage devices make use of the
additional properties that arise when the spins on different where Vm is the molar volume of the substance.
centres interact. The magnetization can be thought of as contributing to the
density of lines of force in the material (Fig. 15F.1). Materials
➤ What is the key idea? for which χ > 0 are called paramagnetic; they tend to move into
The principal magnetic properties of solids arise from the a magnetic field and the density of lines of force within them,
spins of unpaired electrons and their interactions. formally the ‘magnetic flux density’, is greater than in a vacuum.
Those for which χ < 0 are called diamagnetic and tend to move
➤ What do you need to know already? out of a magnetic field; the density of lines of force within them
You need to be aware of the properties of electron angu- is lower than in a vacuum. A paramagnetic material consists
lar momentum (Topic 8B) and the relation of magnetic of ions or molecules with unpaired electrons, such as radicals
moments to angular momenta (Topic 8C). and many d-metal complexes; a diamagnetic substance (which
is far more common) is one with no unpaired electrons.
The magnetic susceptibility is traditionally measured with a
‘Gouy balance’. This instrument consists of a sensitive balance
from which the sample, contained in a narrow tube, hangs be-
The magnetic properties of metallic solids and semiconductors
tween the poles of a magnet. If the sample is paramagnetic, it
depend strongly on the band structures of the material. In this
is drawn into the field and its apparent weight is greater when
section, attention is confined largely to the much simpler mag-
the field is turned on. A diamagnetic sample tends to be ex-
netic properties of collections of individual molecules or ions,
pelled from the field and appears to weigh less when the field is
such as d-metal complexes. Much of the discussion therefore
turned on. The balance is normally calibrated against a sample
applies to liquid- and gas-phase samples, as well as to solids.
of known susceptibility.
The modern way of measuring magnetic susceptibility
makes use of a ‘superconducting quantum interference de-
vice’ (a SQUID, Fig. 15F.2). A SQUID makes use of the quan-
15F.1 Magnetic susceptibility tization of magnetic flux and the property of current loops in
Some molecules possess permanent magnetic dipole mo-
ments. In the absence of an external magnetic field, the ori-
entation of the dipole is random and the material has no
net magnetic moment. However, certain orientations are fa-
voured when a magnetic field is applied. The magnetization,
M , the net dipole-moment density, is the resulting average
molecular magnetic dipole moment multiplied by the num-
ber density of molecules in the sample. The magnetization in- (a) (b) (c)
duced by a magnetic field of strength H is proportional to H , Figure 15F.1 (a) In a vacuum, the strength of a magnetic field
and is written can be represented by the density of lines of force; (b) in a
diamagnetic material, the density is reduced; (c) in a paramagnetic
M H Volume magnetic susceptibility [definition] (15F.1) material, the density is increased.
692 15 Solids
of the electron spins fluctuate. Some o rientations have lower

SQUID Magnetic energy than others, and the magnetization depends on the
field
randomizing influence of thermal motion. Thermal averag-
ing of the permanent magnetic moments in the presence
Superconducting of an applied magnetic field results in a contribution to the
wire Current Sample magnetic susceptibility that is proportional to m2/3kT. It fol-
lows that the spin contribution to the molar magnetic sus-
ceptibility is
N A g e2 0 B2 S(S 1) Molar magnetic susceptibility

m [spin contribution]
(15F.4a)
3kT
Figure 15F.2 The arrangement used to measure magnetic
susceptibility with a SQUID. The sample is moved upwards in where μ0 is the magnetic constant (formerly, and still widely,
small increments and the potential difference across the SQUID is the ‘vacuum permeability’). This susceptibility is positive, so
measured. the spin magnetic moments contribute to the paramagnetic
susceptibilities of materials. Equation 15F.4a is commonly writ-
ten as the Curie law:
Table 15F.1 Magnetic susceptibilities at 298 K⋆
C N g 2 2 S(S 1)
χ/10−6 χm/(10−10 m3 mol−1) m , C A e 0 B Curie law (15F.4b)
T 3k
H2O(l) −9.02 −1.63
The spin contribution to the susceptibility decreases with in-
NaCl(s) −16 −3.8
creasing temperature because the thermal motion randomizes
Cu(s) −9.7 −0.69
the spin orientations.
CuSO4·5H2O(s) +167 +183
⋆
More values are given in the Resource section.
Brief illustration 15F.1
superconductors that, as part of the circuit, include a weakly
conducting link through which electrons must tunnel. The cur- Consider a complex salt with three unpaired electrons per
complex cation and molar volume 61.7 cm3 mol−1; its molar
rent that flows in the loop in a magnetic field depends on the
magnetic susceptibility can be calculated using eqn 15F.4b.
value of the magnetic flux, and a SQUID can be exploited as a
First, note that
very sensitive magnetometer. Table 15F.1 lists some experimen-
tal values of the magnetic susceptibility. N A g e2 0 B2
6.3001 106 m3 K mol 1
3k
Then with S = 23 eqn 15F.4b gives
Permanent and induced
15F.2 1
3 3
magnetic moments m (6.3001 106 m3 K mol 1 ) 2 2

298 K
7.92 108 m3 mol 1
The permanent magnetic moment of a molecule arises from
any unpaired electron spins in the molecule. The magnitude, m, and from eqn 15F.2
of the magnetic moment of an electron is proportional to the 8 3 1
magnitude of the spin angular momentum, {s(s +1)}1/ 2 : m 7.92 10 m mol
5 3 1
1.28 103
Vm 6.17 10 m mol
e Magnetic
m g e {s(s 1)}1/ 2 B B moment (15F.3)
2me [magnitude]
At low temperatures, some paramagnetic solids make a phase
where ge = 2.0023 and μB, the Bohr magneton, has the value transition to a state in which large domains of spins align with
9.274 × 10−24 J T−1. If there are several electron spins in each parallel orientations. This cooperative alignment gives rise
molecule, they combine to give a total spin S, and then s(s + 1) to a very strong magnetization and is called ferromagnetism
is replaced by S(S + 1). (Fig. 15F.3). In other cases, exchange interactions lead to al-
The magnetization, and consequently the magnetic suscep- ternating spin orientations: the spins are locked into a low-
tibility, depends on the temperature because the o rientations magnetization arrangement to give an antiferromagnetic phase.
15F The magnetic properties of solids 693
(a) Magnetic properties of

15F.3
superconductors
(b)
Superconductors have unique magnetic properties. Supercon

(c)
ductors classed as Type I show abrupt loss of superconductivity
when an applied magnetic field exceeds a critical value H c char-
Figure 15F.3 (a) In a paramagnetic material, the electron spins are acteristic of the material. An empirical relation between the
aligned at random in the absence of an applied magnetic field. value of H c , the temperature T, and the critical temperature Tc is
(b) In a ferromagnetic material, the electron spins are locked into a
parallel alignment over large domains. (c) In an antiferromagnetic
T2
material, the electron spins are locked into an antiparallel H c (T ) H c (0) 1 2 Dependence H c on T (15F.5)
arrangement. The latter two arrangements survive even in the Tc
absence of an applied field.
provided T ≤ Tc. Note that the critical field falls as T rises from
0 towards Tc. Therefore, to maintain superconductivity in the
The ferromagnetic phase has a non-zero magnetization in the
presence of a magnetic field, it is best to keep T well below Tc
absence of an applied field, but the antiferromagnetic phase has
and to select a material with a high H c (0).
zero magnetization because the spin magnetic moments cancel.
The ferromagnetic transition occurs at the Curie temperature,
and the antiferromagnetic transition occurs at the Néel tempera-
Brief illustration 15F.2
ture. Which type of cooperative behaviour occurs depends on
the details of the band structure of the solid. Lead has Tc = 7.19 K and H c (0) 63.9 kA m 1. At T = 6.0 K
Magnetic moments can also be induced in molecules. To the magnetic field that would quench its superconductivity
see how this effect arises, it is necessary to note that the circu- would be
lation of electronic currents induced by an applied field gives
rise to a magnetic field which usually opposes the applied field 0.304
and makes the substance diamagnetic. In these cases, the in- (6 . 0 K )2

H c (6.0 K ) (63.9 kA m 1 ) 1 19 kAm 1
duced electronic currents occur within the molecular orbit- (7.19 K )2

als that are occupied in its ground state. There are a few cases
in which molecules are paramagnetic despite having no un-
The lead remains superconducting at 6.0 K for this and weaker
paired electrons. In these materials the induced electron cur- applied field strengths. If the temperature is lowered to 5.0 K
rents flow in the opposite direction because they can make the corresponding calculation gives H c (5.0 K ) 33 kA m 1, and
use of unoccupied orbitals that lie close to the HOMO in en- superconductivity survives at higher field strengths. In each
ergy (a similar effect is the paramagnetic contribution to the case, superconductivity would survive to higher field strengths
chemical shift, Topic 12B). This orbital paramagnetism is dis- for a material with a higher H c (0) .
tinguished from spin paramagnetism by being independent of
temperature; it is called temperature-independent paramag-
netism (TIP).
These remarks can be summarized as follows. All molecules A magnetic field for which H > H c does not penetrate into a
have a diamagnetic component to their susceptibility, but this Type I superconductor at temperatures below Tc. This complete
contribution is dominated by spin paramagnetism if the mol- exclusion of a magnetic field from a material is known as the
ecules have unpaired electrons. In a few cases (where there are Meissner effect, which can be demonstrated by the levitation
low-lying excited states) TIP is strong enough to make the mol- of a superconductor above a magnet. Type II superconductors,
ecules paramagnetic even though all their electrons are paired. which include the HTSCs, show a gradual loss of diamagnetism
with increasing magnetic field.
Exercises
E15F.1 The magnitude of the magnetic moment of CrCl3 is 3.81μB. How many Exercise
unpaired electrons does the Cr possess? 1
E15F.3 Niobium, Nb, has Tc = 9.5 K and H c (0) 158 kA m . Calculate the
E15F.2 Estimate the spin-only molar susceptibility of CuSO4·5H2O at 25 °C. highest magnetic field at which superconductivity can be maintained at 6 K.
694 15 Solids
☐ 1. The magnetization of a material is the average molecu- ☐ 6. Ferromagnetism is the cooperative alignment of elec-
lar magnetic dipole moment multiplied by the number tron spins in a material and gives rise to strong perma-
density of the molecules. nent magnetization.
☐ 2. The magnetic susceptibility expresses the relation ☐ 7. Antiferromagnetism results from alternating spin ori-
between the magnetization and the applied magnetic entations in a material and leads to weak magnetization.
field strength. ☐ 8. Temperature-independent paramagnetism arises from
☐ 3. Diamagnetic materials have negative magnetic suscep- induced electron currents.
tibilities and tend to move out of a magnetic field. ☐ 9. The Meissner effect is the exclusion of a magnetic field
☐ 4. Paramagnetic materials have positive magnetic suscep- from a Type I superconductor.
tibilities and tend to move into a magnetic field.
☐ 5. The Curie law describes the temperature dependence of
the magnetic susceptibility.
Magnetization M H Definition 15F.1
Molar magnetic susceptibility χm = χVm Definition 15F.2
Magnetic moment m = ge{s(s + 1)}1/2μB B e/2me 15F.3
Curie law m C /T , C N A g e2 0 B2 S(S 1)/3k Paramagnetism 15F.4b
Dependence of H c on T H c (T ) H c (0)(1 T 2 /Tc2 ) Empirical 15F.5
TOPIC 15G The optical properties
of solids
The optical properties of solids are of ever-increasing
importance in modern technology, not only for the gen-
eration of light but for the propagation and manipulation
of information.
➤ What is the key idea? Figure 15G.1 The electron–hole pair shown on the left can
The optical properties of molecules in solids differ from migrate through a solid lattice as the excitation hops from
molecule to molecule. The mobile excitation is called an exciton.
those of isolated molecules as a result of the interaction of
their transition dipoles.
on one molecule could not move to another. This interaction
➤ What do you need to know already? affects the energy levels of the system. The strength of the inter-
action also governs the rate at which an exciton moves through
You need to be familiar with the concept of transition
the crystal: a strong interaction results in fast migration and a
dipoles (Topics 8C and 11A) and of the band theory of
vanishingly small interaction leaves the exciton localized on its
solids (Topic 15C).
original molecule. The specific mechanism of interaction that
leads to exciton migration is the interaction between the tran-
sition dipoles of the excitation (Topic 11A). Thus, an electric
dipole transition in a molecule is accompanied by a shift of
Topic 11A explains the factors that affect the energy and inten- charge, and this transient dipole exerts a force on an adjacent
sity of light absorbed by isolated atoms and molecules in the molecule. The latter responds by shifting its charge. This pro-
gas phase and in solution. Significant differences arise when the cess continues and the excitation migrates through the crystal.
molecules are neighbours in a solid. The energy shift arising from the interaction between tran-
sition dipoles can be explained as follows. The potential en-
ergy, Ep, of interaction between two parallel electric dipole
moments μ1 and μ2 separated by a distance r is V = μ1μ2(1 −
15G.1 Excitons 3 cos2θ)/4πε0r3, where the angle θ is defined in (1). A head-to-tail
alignment corresponds to θ = 0, and a parallel alignment cor-
Consider an electronic excitation of a molecule (or an ion) in a responds to θ = 90°. From the expression for V it follows that
crystal. If the excitation corresponds to the removal of an elec- V < 0 (a favourable, energy-lowering interaction) for 0 ≤ θ < 54.7°,
tron from one orbital of a molecule and its elevation to an or- V = 0 when θ = 54.7° (at this angle 1 − 3 cos2θ = 0), and V > 0 (an
bital of higher energy, then the excited state of the molecule can unfavourable, energy-raising interaction) for 54.7° < θ ≤ 90°.
be envisaged as the coexistence of an electron and a hole. This
electron–hole pair, which behaves as a particle-like exciton, μ1
migrates from molecule to molecule in the crystal (Fig. 15G.1).
A migrating excitation of this kind is called a Frenkel exciton,
r
and is commonly found in molecular solids. The electron and
μ2 θ
hole can also be on different molecules, but in each other’s vi-
cinity. A migrating excitation of this kind, which is now spread
over several molecules (more usually ions), is called a Wannier 1
exciton. Exciton formation causes spectral lines to shift, split,
and change intensity. In a head-to-tail arrangement, there is a favourable interac-
The migration of a Frenkel exciton (the only type considered tion between the region of partial positive charge in one mol-
here) implies that there is an interaction between the molecules ecule and the region of partial negative charge in the other
that constitute the crystal: if this were not the case the excitation molecule. In contrast, in a parallel arrangement, the molecular
696 15 Solids
that are in-phase between neighbouring unit cells. Within

each unit cell the transition dipoles may be arrayed in the two
different ways shown in the illustration. The two orientations
correspond to different interaction energies, with interaction
(a) being unfavourable in one and favourable in the other, so the
two transitions appear in the spectrum as two bands of differ-
ent frequencies. The magnitude of the Davydov splitting is de-
termined by the energy of interaction between the transition
dipoles within the unit cell.
(b)
Figure 15G.2 (a) The alignment of transition dipoles (the arrows) 15G.2 Metals and semiconductors
shown here is energetically unfavourable, and the exciton
absorption is shifted to higher energy (higher frequency). (b) The Figure 15C.8 shows the band structure in an idealized metallic
alignment shown here is energetically favourable for a transition conductor at T = 0. The absorption of a photon can excite elec-
in this orientation, and the exciton band occurs at lower frequency trons from the occupied levels to the unoccupied levels. There
than in the isolated molecules. is a near continuum of unoccupied energy levels above the
Fermi level, so absorption occurs over a wide range of frequen-
cies. In metals, the bands are sufficiently wide that radiation is
interaction is unfavourable because of the close approach of re-
absorbed from the radiofrequency to the ultraviolet region of
gions of partial charge with the same sign. It follows that an all-
the electromagnetic spectrum but not to very high-frequency
parallel arrangement of the transition dipoles (Fig. 15G.2a) is
electromagnetic radiation, such as X-rays and γ-rays, so met-
energetically unfavourable, so the absorption occurs at a higher
als are transparent at these frequencies. This range of absorbed
frequency than in the isolated molecule. Conversely, a head-to-
frequencies includes the entire visible spectrum, so it might be
tail alignment of transition dipoles (Fig. 15G.2b) is energeti-
expected that all metals should be black. However, metals are in
cally favourable, and the transition occurs at a lower frequency
fact lustrous (that is, they reflect light) and some are coloured
than in the isolated molecules.
(that is, they absorb light of certain wavelengths), so the model
If there are N molecules per unit cell, there are N exciton
clearly needs some improvement.
bands in the spectrum (if all of them are allowed). The splitting
between the bands is the Davydov splitting. To understand the
origin of the splitting, consider the case N = 2 with the mol- 15G.2(a) Light absorption
ecules arranged as in Fig. 15G.3 and suppose that the transi-
To explain the lustrous appearance of a smooth metal sur-
tion dipoles are along the length of the molecules. The radiation
face, it is important to realize that the absorbed energy can be
stimulates the collective excitation of the transition dipoles
reemitted very efficiently as light, with only a small fraction
of the energy being released into the bulk as heat. Because the
atoms near the surface of the material absorb most of the ra-
diation, emission also occurs primarily from the surface. In
essence, if the sample is excited with visible light, then elec-
trons near the surface are driven into oscillation at the same
(a) frequency, and visible light is emitted from the surface, so ac-
counting for the lustre of the material.
The perceived colour of a metal depends on the frequency
range of reflected light. That in turn depends on the frequency
(b) (a)
range of light that can be absorbed and, by extension, on the
band structure. Silver reflects light with nearly equal efficiency
(b)
across the visible spectrum because its band structure has
Davydov splitting many unoccupied energy levels that can be populated by ab-
sorption of, and depopulated by emission of, visible light. On
Figure 15G.3 When the transition moments within a unit cell
lie in different relative directions, as depicted in (a) and (b), the the other hand, copper has its characteristic colour because it
energies of the transitions are shifted and give rise to the two has relatively fewer unoccupied energy levels that can be ex-
bands labelled (a) and (b) in the spectrum. The separation of the cited with violet, blue, and green light. The material reflects at
bands is the Davydov splitting. all wavelengths, but more light is emitted at lower frequencies
15G The optical properties of solids 697
kind are widely used in electronic displays. The wavelength of

Conduction
band
emitted light depends on the band gap of the semiconductor.
Gallium arsenide itself emits infrared light, but its band gap is
Energy
Band
gap, Eg widened by incorporating phosphorus, and a material of com-
Valence
position approximately GaAs0.6P0.4 emits light in the red region
band of the spectrum.
A light-emitting diode is not a laser (Topic 11G) because
stimulated emission is not involved. In diode lasers, light
Figure 15G.4 In some materials, the band gap Eg is very large
and electron promotion can occur only by excitation with emission due to electron–hole recombination is employed
electromagnetic radiation. as the basis of laser action, and the population inversion can
be sustained by sweeping away the electrons that fall into the
holes of the p-type semiconductor. High-power diode lasers
are also used to pump other lasers. One example is the pump-
(corresponding to yellow, orange, and red). Similar arguments
ing of Nd:YAG lasers (Topic 11G) by Ga0.91Al0.09As/Ga0.7Al0.3As
account for the colours of other metals, such as the yellow of
diode lasers. Diode lasers are used widely in laser pointers,
gold. It is interesting to note that the colour of gold can be ac-
laser printers, medicine (instead of a scalpel in some surgical
counted for only by including relativistic effects in the calcula-
procedures), and telecommunications (where information en-
tion of its band structure.
coded in a modulated laser beam travels through fibre optic
Now consider semiconductors. If the band gap Eg is compa-
cables).
rable to kT the promotion of electrons from the conduction to
the valence band of a semiconductor can be the result of ther-
mal excitation. In some materials, the band gap is very large Exercise
and electron promotion can occur only by excitation with elec- E15G.1 The promotion of an electron from the valence band into the
tromagnetic radiation. However, as is seen from Fig. 15G.4, conduction band in pure TiO2 by light absorption requires a wavelength
there is a frequency νmin = Eg/h below which light absorption of less than 350 nm. Calculate the energy gap in electronvolts between the
cannot occur. Above this frequency threshold, a wide range of valence and conduction bands.
frequencies can be absorbed by the material, as in a metal.
Brief illustration 15G.1 15G.3 Nonlinear optical phenomena

The energy of the band gap in the semiconductor cadmium Nonlinear optical phenomena arise from changes in the op-
sulfide (CdS) is 2.4 eV (equivalent to 3.8 × 10−19 J). It follows tical properties of a material in the presence of intense elec-
that the minimum electronic absorption frequency is tromagnetic radiation. In frequency doubling (or ‘second
harmonic generation’), an intense laser beam is converted
3.8 1019 J to radiation with twice (and in general a multiple) of its ini-
min 5.8 1014 s 1
6.626 1034 Js tial frequency as it passes through a suitable material. It fol-
lows that frequency doubling and tripling of an Nd:YAG
This frequency, of 580 THz, corresponds to a wavelength of laser, which emits radiation at 1064 nm (Topic 11G), produce
520 nm (green light). Lower frequencies, corresponding to yel- green light at 532 nm and ultraviolet radiation at 355 nm, re-
low, orange, and red, are not absorbed and consequently CdS spectively. Common materials that can be used for frequency
appears yellow-orange. doubling in laser systems include crystals of potassium dihy-
drogenphosphate (KH2PO4), lithium niobate (LiNbO3), and
β-barium borate (β-BaB2O4).
15G.2(b)Light-emitting diodes and Frequency doubling can be explained by examining how a
substance responds nonlinearly to incident radiation of fre-
diode lasers quency ω = 2πν. Radiation of a particular frequency arises
The unique electrical properties of p–n junctions between sem- from oscillations of an electric dipole at that frequency and
iconductors (which are described in Topic 15E) can be put to the incident electric field E of the radiation induces an elec-
good use in optical devices. In some materials, most notably tric dipole moment of magnitude μ, in the substance. At low
gallium arsenide, GaAs, energy from electron–hole recombi- light intensity, most materials respond linearly, in the sense that
nation is released not as heat but is carried away by photons E, where α is the polarizability (Topic 14A). At high light
as electrons move across the junction driven by the appropri- intensity, the hyperpolarizability β of the material becomes im-
ate potential difference. Practical light-emitting diodes of this portant (Topic 14A) and the induced dipole becomes
698 15 Solids
Induced dipole E 2 E02 cos 2 t 12 E02 (1 cos 2t ) (15G.2)

E 12 E 2  moment in terms (15G.1)
of the hyperpolaarizability
Hence, the nonlinear term contributes an induced electric di-
The nonlinear term β E can be expanded as follows if it is sup-
2 pole that includes a component that oscillates at the frequency
posed that the incident electric field is E0 cos ωt: 2ω and that can act as a source of radiation of that frequency.
☐ 1. An exciton is an electron–hole pair caused by optical ☐ 3. Nonlinear optical phenomena arise from changes in
excitation in a solid; Frenkel excitons are localized on the optical properties of a material in the presence of
a single molecule, whereas Wannier excitons are spread intense electromagnetic radiation; they can give rise to
over several molecules. frequency doubling.
☐ 2. If the unit cell contains N molecules, there are N exciton
bands in the spectrum separated by the Davydov
splitting.
Exercises and problems 699
FOCUS 15 Solids
To test your understanding of this material, work through Selected solutions can be found at the end of this Focus in
the Exercises, Additional exercises, Discussion questions, and the e-book. Solutions to even-numbered questions are available
Problems found throughout this Focus. online only to lecturers.
TOPIC 15A Crystal structure

Discussion questions
D15A.1 Describe the relationship between space lattice and unit cell. D15A.3 Draw unit cells representative of the three possible cubic lattices. State
how many lattice points are in each of your cells and identify whether or not
D15A.2 Explain how planes in a lattice are labelled.
the cells you have drawn are primitive.
Additional exercises
−1
E15A.4 An orthorhombic unit cell of a compound of molar mass 135.01 g mol E15A.6 Calculate the separations of the planes {112}, {110}, and {224} in a
has the dimensions a = 589 pm, b = 822 pm, and c = 798 pm. The mass density crystal in which the cubic unit cell has side 562 pm.
of the solid is estimated as 2.9 g cm−3. Identify the number of formula units in a
E15A.7 Calculate the separations of the planes {123}, {222}, and {246} in a
unit cell and calculate a more precise value of the mass density.
crystal in which the cubic unit cell has side 712 pm.
E15A.5 State the Miller indices of the planes that intersect the crystallographic
E15A.8 An orthorhombic unit cell has dimensions a = 769 pm, b = 891 pm,
axes at the distances (−a, 2b, −c) and (a, 4b, −4c).
and c = 690 pm. Calculate the spacing, d, of the {312} planes.
Problems
P15A.1 Although the crystallization of large biological molecules may not be pyridine-2,6-diamidoxime (1, C7H9N5O2). The compound they isolated
as readily accomplished as that of small molecules, their crystal lattices are no from the reaction of the ligand with CuSO4(aq) did not contain a
different. The protein tobacco seed globulin forms face-centred cubic crystals [Cu(C7H9N5O2)2]2+ complex cation as expected. Instead, X-ray diffraction
with unit cell dimension of 12.3 nm and a mass density of 1.287 g cm−3. analysis revealed a linear polymer of formula [{Cu(C7H9N5O2)(SO4)}·2H2O]n,
Determine its molar mass (assume there is one molecule associated with each which features bridging sulfate groups. The unit cell was primitive monoclinic
lattice point). with a = 1.0427 nm, b = 0.8876 nm, c = 1.3777 nm, and β = 93.254°. The mass
density of the crystals is 2.024 g cm−3. How many monomer units are there in
P15A.2 Show that the volume of a monoclinic unit cell is V = abc sin β.
the unit cell?
P15A.3 Derive an expression for the volume of a hexagonal unit cell.
P15A.4 Show that the volume of a triclinic unit cell of sides a, b, and c and
angles α, β, and γ is O O
V abc(1 cos cos cos 2 cos cos cos )
2 2 2 1/ 2
HO N OH
N N
Use this expression to derive expressions for monoclinic and orthorhombic H H
unit cells. For the derivation, it may be helpful to use the result from vector
analysis that V = a·b×c and to calculate V2 initially. The compound Rb3TlF6
has a tetragonal unit cell with dimensions a = 651 pm and c = 934 pm. 1 Pyridine-2,6-diamidoxime
Calculate the volume of the unit cell.
P15A.5 The volume of a monoclinic unit cell is abc sin β. Naphthalene has
a monoclinic unit cell with two molecules in each cell and sides in the ratio ‡
P15A.8 D. Sellmann et al. (Inorg. Chem. 36, 1397 (1997)) describe
1.377:1:1.436. The angle β is 122.82° and the mass density of the solid is the synthesis and reactivity of the ruthenium nitrido compound
1.152 g cm−3. Calculate the dimensions of the cell. [N(C4H9)4][Ru(N)(S2C6H4)2]. The ruthenium complex anion has the two
1,2-benzenedithiolate ligands (2) at the base of a rectangular pyramid and
P15A.6 Fully crystalline polyethene has its chains aligned in an
the nitrido ligand at the apex. Compute the mass density of the compound
orthorhombic unit cell of dimensions 740 pm × 493 pm × 253 pm. There
given that it crystallizes with an orthorhombic unit cell with a = 3.6881 nm,
are two repeating CH2CH2 units in each unit cell. Calculate the theoretical
b = 0.9402 nm, and c = 1.7652 nm and eight formula units in each cell. The
mass density of fully crystalline polyethene. The actual mass density ranges
replacement of the ruthenium with osmium results in a compound with the
from 0.92 to 0.95 g cm−3.
‡
P15A.7 B.A. Bovenzi and G.A. Pearse, Jr (J. Chem. Soc. Dalton Trans. 2793
‡
(1997)) synthesized coordination compounds of the tridentate ligand These problems were supplied by Charles Trapp and Carmen Giunta.
700 15 Solids
same crystal structure and a unit cell with a volume less than 1 per cent larger. P15A.9 Show that the separation of the {hkl} planes in an orthorhombic crystal
Estimate the mass density of the osmium analogue. with sides a, b, and c is given by eqn 15A.1b.
S–
S–
2 1,2-Benzenedithiolate ion
TOPIC 15B Diffraction techniques

D15B.1 What is meant by a systematic absence? How do they arise and how D15B.3 Describe the consequences of the phase problem in determining
can they be helpful in identifying the type of unit cell? structure factors and how the problem is overcome.
D15B.2 Discuss what is meant by ‘scattering factor’. How is it related to the
number of electrons in the atoms scattering X-rays?
E15B.6 The angle of a Bragg reflection from a set of crystal planes separated by E15B.16 Calculate the structure factors for a body-centred cubic (cubic I)
99.3 pm is 20.85°. Calculate the wavelength of the X-rays. structure in which the scattering factor of the central ion is twice that of the
ions at the corners of the cube.
E15B.7 The angle of a Bragg reflection from a set of crystal planes separated by
128.2 pm is 19.76°. Calculate the wavelength of the X-rays. E15B.17 In an X-ray investigation, the following structure factors were
determined (with F−h00 = Fh00):
E15B.8 What are the values of the angle θ of the three diffraction lines with
the smallest θ expected from a cubic F unit cell with lattice parameter 407 pm h 0 1 2 3 4 5 6 7 8 9
when the X-ray wavelength is 129 pm?
Fh00 10 −10 8 −8 6 −6 4 −4 2 −2
E15B.9 Potassium nitrate crystals have orthorhombic unit cells of dimensions
Construct the electron density along the corresponding direction.
a = 542 pm, b = 917 pm, and c = 645 pm. Calculate the values of θ for the
(100), (010), and (111) reflections using radiation of wavelength 154 pm. E15B.18 In an X-ray investigation, the following structure factors were
determined (with F−h00 = Fh00):
E15B.10 Calcium carbonate crystals in the form of aragonite have
orthorhombic unit cells of dimensions a = 574.1 pm, b = 796.8 pm, and h 0 1 2 3 4 5 6 7 8 9
c = 495.9 pm. Calculate the values of θ for the (100), (010), and (111)
Fh00 10 10 4 4 6 6 8 8 10 10
reflections using radiation of wavelength 83.42 pm.
Construct the electron density along the corresponding direction.
E15B.11 Radiation from an X-ray source consists of two components of
wavelengths 154.433 pm and 154.051 pm. Calculate the difference in glancing E15B.19 Construct the Patterson map from the information in Exercise
angles (2θ) of the diffraction lines arising from the two components in a E15B.17.
diffraction pattern from planes of separation 77.8 pm.
E15B.20 Construct the Patterson map from the information in Exercise
E15B.12 Consider a source that emits X-radiation at a range of wavelengths, E15B.18.
with two components of wavelengths 93.222 and 95.123 pm. Calculate the
E15B.21 In a Patterson map, the spots correspond to the lengths and directions
separation of the glancing angles (2θ) arising from the two components in a
of the vectors joining the atoms in a unit cell. Sketch the pattern that would be
diffraction pattern from planes of separation 82.3 pm.
obtained for a planar, triangular isolated BF3 molecule.
E15B.13 What is the value of the scattering factor in the forward direction
E15B.22 In a Patterson map, the spots correspond to the lengths and directions
for Mg2+?
of the vectors joining the atoms in a unit cell. Sketch the pattern that would be
E15B.14 The coordinates, in units of a, of the atoms in a cubic I unit cell are (0,0,0), obtained from the carbon atoms in an isolated benzene molecule.
(0,1,0), (0,0,1), (0,1,1), (1,0,0), (1,1,0), (1,0,1), (1,1,1), and 12 , 12 , 12 . Calculate the
E15B.23 What speed should electrons have if they are to have a wavelength of
structure factor Fhkl when all the atoms are identical. Where possible, simplify your
105 pm?
expression by using einπ = (−1)n, as in Example 15B.1.
E15B.24 Calculate the wavelength of electrons that have reached thermal
E15B.15 Calculate the structure factors for an orthorhombic C structure in which
equilibrium by collision with a moderator at 380 K.
the scattering factors of the two ions on the faces are twice that of the ions at the
corners of the cube. Assume that a = b = c, that is the unit cell is a cube.
Problems
P15B.1 In the early days of X-ray crystallography there was an urgent need to angle from a mechanically ruled grating. Another method was to estimate the
know the wavelengths of X-rays. One technique was to measure the diffraction separation of lattice planes from the measured density of a crystal. The mass
density of NaCl is 2.17 g cm−3 and the (100) reflection using radiation of a for 0 ≤ r ≤ R and ρ(r) = 0 for r > R, with R a parameter that represents the
certain wavelength occurred at 6.0°. Calculate the wavelength of the X-rays. radius of the atom. Explore how f varies with Z and R.
P15B.2 The element polonium crystallizes in a cubic system. Bragg reflections, P15B.6 The coordinates of the four I atoms in the unit cell of KIO4 are (0,0,0),
with X-rays of wavelength 154 pm, occur at sin θ = 0.225, 0.316, and 0.388 0, 12 , 12 , 12 , 12 , 12 , 12 ,0, 34 . By calculating the phase of the I reflection in the
from the {100}, {110}, and {111} sets of planes. The separation between the structure factor, show that the I atoms contribute no net intensity to the (114)
sixth and seventh lines observed in the diffraction pattern is larger than reflection.
between the fifth and sixth lines. Is the unit cell cubic P, I, or F? Calculate the
P15B.7 The coordinates, as multiples of a, of the A atoms, with scattering
unit cell dimension.
factor fA, in a cubic lattice are (0,0,0), (0,1,0), (0,0,1), (0,1,1), (1,0,0),
P15B.3 Elemental silver reflects X-rays of wavelength 154.18 pm at angles of (1,1,0), (1,0,1), and (1,1,1). There is also a B atom, with scattering factor fB,
at 2 , 2 , 2 . Calculate the structure factors Fhkl and predict the form of the
1 1 1
19.076°, 22.171°, and 32.256°. However, there are no other reflections at angles
of less than 33°. Assuming a cubic unit cell, determine its type and dimension. diffraction pattern when (a) fA = f, fB = 0, (b) f B = 12 f A, and (c) fA = fB = f.
Calculate the mass density of silver. Hint: Calculate the expected reflections
P15B.8 Here we explore electron diffraction patterns. (a) Predict from the
from different types of cubic unit cell and compare those with the data given.
Wierl equation, eqn 15B.8, the positions of the first maximum and first
P15B.4 In their book X-rays and crystal structures (which begins ‘It is now minimum in the neutron and electron diffraction patterns of a Br2 molecule
two years since Dr. Laue conceived the idea …’) the Braggs give a number of obtained with neutrons of wavelength 78 pm wavelength and electrons of
simple examples of X-ray analysis. For instance, they report that the reflection wavelength 4.0 pm. (b) Use the Wierl equation to predict the appearance of
from {100} planes in KCl occurs at 5.38°, but for NaCl it occurs at 6.00° for the electron diffraction pattern of CCl4 with an (as yet) undetermined C–Cl
X-rays of the same wavelength. If the side of the NaCl unit cell is 564 pm, bond length but of known tetrahedral symmetry; assume the electron energy
what is the side of the KCl unit cell? The mass densities of KCl and NaCl are to be 10 keV. Take fCl = 17f and fC = 6f and note that R(Cl,Cl) = (8/3)1/2R(C,Cl).
1.99 g cm−3 and 2.17 g cm−3 respectively. Do these values support the X-ray Plot I/f 2 against positions of the maxima occurred at 3.17°, 5.37°, and 7.90°
analysis? and minima occurred at 1.77°, 4.10°, 6.67°, and 9.17°. What is the C–Cl bond
length in CCl4?
P15B.5 Use mathematical software to draw a graph of the scattering factor f
against (sin θ)/λ for an atom of atomic number Z for which ρ(r) = 3Z/4πR3
TOPIC 15C Bonding in solids

D15C.1 In what respects is the hard-sphere model of metallic solids D15C.2 Describe the caesium-chloride and rock-salt structures in terms of the
deficient? occupation of holes in expanded close-packed lattices.
E15C.6 Calculate the packing fraction for equilateral triangular rods stacked as E15C.8 Determine the radius of the smallest anion that can have (i) sixfold and
shown in (3). (ii) eightfold coordination with the Rb+ ion (radius 149 pm).
E15C.9 Does iron expand or contract as it transforms from hcp to bcc? The
atomic radius of iron is 126 pm in hcp but 122 pm in bcc.
E15C.10 Calculate the lattice enthalpy of MgBr2 from the following data:
ΔH/(kJ mol−1)
Sublimation of Mg(s) +148
Ionization of Mg(g) to Mg2+(g) +2187
Vaporization of Br2(l) +31
Dissociation of Br2(g) +193
3
Electron attachment to Br(g) −331
E15C.7 Calculate the packing fraction for an orthorhombic C cell in which all
Formation of MgBr2(s) from Mg(s) −524
three sides are the same (assume that the spheres touch along one of the face
and Br2(l) in their reference states.
diagonals which includes an atom on the face).
Problems
P15C.1 Calculate the atomic packing factor for diamond (refer to Fig. 15C.15); P15C.2 Rods of elliptical cross-section with semi-minor and semi-major
assume that the atoms touch along the body diagonal. axes a and b are close-packed as shown in (4). What is the packing
702 15 Solids
fraction? Draw a graph of the packing fraction against the eccentricity ε P15C.6 The energy levels of N atoms in the tight-binding Hückel
of the ellipse. For an ellipse with semi-minor axis a and semi-major axis b, approximation are the roots of a tridiagonal determinant (eqn 15C.1):
ε = (1 − b2/a2)1/2.
k
Ek 2 cos k 1,2, ,N
N 1
a If the atoms are arranged in a ring, the solutions are the roots of a ‘cyclic’
b determinant:
2k
Ek 2 cos k 0, 1, 2, , 12 N
N
(for N even). Discuss the consequences, if any, of joining the ends of an
initially straight length of material.
P15C.7 Verify that the lowest value of the radius ratio for (a) sixfold
4 coordination is 0.414, and (b) for eightfold coordination is 0.732.
P15C.3 (a) Calculate the mass density of diamond assuming that it is a close-
P15C.8 (a) Use the Born–Mayer equation for the lattice enthalpy and a Born–
packed structure of hard spheres with radii equal to half the carbon–carbon Haber cycle to estimate the enthalpy of formation of CaCl(s). The sublimation
bond length of 154.45 pm. (b) The diamond lattice is in fact based on a enthalpy of Ca(s) is 176 kJ mol−1 and it can be assumed that the ionic radius
face-centred cubic lattice but with two atoms per lattice point, such that the of Ca+ is close to that of K+; other necessary data are to be found in Example
structure consists of two interpenetrating fcc lattices, one with its origin 15C.2 or in the tables in the Resource section. (b) Show that an explanation
at (0,0,0) and the other with its origin at (1/4,1/4,1/4). The experimentally for the nonexistence of CaCl(s) can be found in the reaction enthalpy for the
determined mass density is 3.516 g cm−3: can you explain the difference disproportionation reaction 2 CaCl(s) → Ca(s) + CaCl2(s).
between this value and that in (a)?
P15C.9 Derive the Born–Mayer equation (eqn 15C.5) by calculating the
P15C.4 When energy levels in a band form a continuum, the density of states
energy at which d(Ep Ep )/dd 0 , with Ep and Ep given by eqns 15C.3 and
ρ(E), the number of levels in an energy range divided by the width of the 15C.4, respectively.
range, may be written as ρ(E) = dk/dE, where dk is the change in the quantum
number k and dE is the energy change. (a) Use eqn 15C.1 to show that P15C.10 Suppose that ions are arranged in a (somewhat artificial) two-
dimensional lattice like the fragment shown in the figure below. Calculate the
(N 1)/2 Madelung constant for the central ion in this array.
(E) 1/ 2
E 2
1
2
where k, N, α, and β have the meanings described in Topic 15C. (b) Use this
expression to show that ρ(E) becomes infinite as E approaches α ± 2β. That is, etc.
show that the density of states increases towards the edges of the bands in a
one-dimensional metallic conductor.
P15C.5 In three dimensions the variation of density of states is like that shown
in the figure below. Account for the fact that in a three-dimensional solid the
greatest density of states is near the centre of the band and the lowest density
is at the edges.
+
p Band –
Energy
s Band
Density of states,
TOPIC 15D The mechanical properties of solids

Discussion question
D15D.1 Distinguish between the behaviour of a solid which undergoes elastic
deformation when a stress is applied to one which undergoes plastic deformation.
3
E15D.4 A sample of polystyrene of volume 1.0 cm is subjected to a hydrostatic E15D.6 Poisson’s ratio for lead is 0.41. What change in volume takes place
pressure (stress) of 1000 bar. Calculate the volume of the sample after the when a cube of lead of volume 1.0 dm3 is subjected to a uniaxial stress that
pressure has been applied. The bulk modulus of polystyrene is 3.43 GPa. produces a strain of 2.0 per cent?
E15D.5 Calculate the force which needs to be applied to a polystyrene rod
of diameter 1.0 mm to increase its length from 10.00 cm to 10.05 cm. The
Young’s modulus of polystyrene is 4.42 GPa.
Problems
P15D.1 For an isotropic substance, the moduli and Poisson’s ratio may be P15D.2 The bulk modulus for liquid water at 298 K is 3.4 GPa. If it is assumed
expressed in terms of two parameters λ and μ called the Lamé constants: that the molecules are interacting with a Lennard-Jones potential energy,
estimate the well depth ε (give your answer in kJ mol−1).
(3 2 ) 3 2
E K G p
3 2( )
Use the Lamé constants to confirm the relations between G, K, and E given in
eqn 15D.3.
TOPIC 15E The electrical properties of solids

Discussion question
D15E.1 Describe the characteristics of the Fermi–Dirac distribution; contrast
it with the Boltzmann distribution.
E15E.3 Calculate f(E), the probability predicted by the Fermi–Dirac E15E.5 Suppose that the Fermi energy is 2.00 eV. At 298 K, calculate the energy
distribution, for the case where the energy E is the thermal energy, kT, below above the Fermi energy at which the probability has fallen to 0.10; express
the Fermi energy: E = EF − kT. Comment on the value you obtain in relation your answer in eV.
to that from Exercise E15E.1.
E15E.6 Is gallium-doped germanium a p-type or n-type semiconductor?
E15E.4 A typical value for the Fermi energy is 1.00 eV. At 298 K, calculate the
energy above the Fermi energy at which the probability has fallen to 0.25;
express your answer in eV.
Problems
P15E.1 Refer to eqn 15E.2b and express f(E) as a function of the variable P15E.4 In an intrinsic semiconductor, the band gap is so small that the
(E − EF)/EF and EF/kT. Then, using mathematical software, display the set of Fermi–Dirac distribution results in some electrons populating the conduction
curves shown in Fig. 15E.5 as a single surface. band. It follows from the exponential form of the Fermi–Dirac distribution
that the conductance G, the inverse of the resistance (with units of siemens,
P15E.2 In this and the following problem you are invited to explore further
1 S = 1 Ω−1), of an intrinsic semiconductor should have an Arrhenius-like
some of the properties of the Fermi–Dirac distribution, f(E), eqn 15E.2a. E / 2 kT
temperature dependence, shown in practice to have the form G G0 e g ,
(a) Show that at T = 0, f(E) = 1 for E < μ, and f(E) = 0 for E > μ. (b) For
where Eg is the band gap. The conductance of a sample of germanium varied
a three-dimensional solid of volume V, it turns out that in eqn 15E.1
with temperature as indicated below. Estimate the value of Eg.
ρ(E) = CE1/2, with C = 4πV(2me/h2)3/2. If the number of electrons in the solid of
volume V is N, then this number must be equal to the result of integrating eqn
T/K 312 354 420
15E.1 over the full range of energy: N 0 dN (E ) 0 ( E ) f ( E )dE. Evaluate
the integral at T = 0. (Hint: It can be split into two integrals, one G/S 0.0847 0.429 2.86
between E = 0 and μ, and one between E = μ and ∞.) (c) Equate the
expression obtained by evaluating the integral with N, and hence show P15E.5 A sample of an n-type semiconductor is found to be an insulator at
that (3N /8)2 / 3 (h2 /2me ), where N = N /V , the number density of very low temperatures. As the temperature is raised, there comes a point
electrons in the solid. (d) Evaluate μ for sodium, which has mass density at which the conductivity increases markedly, but after this point the
0.97 g cm−3; assume that each atom contributes one electron. conductivity remains pretty much constant as the temperature is raised
further. At much higher temperatures, the conductivity starts to increase
P15E.3 By inspection of eqn 15E.2a and the expression for dN(E) in eqn 15E.1
steadily, with no sign of it reaching a plateau. Explain these observations.
(and without attempting to evaluate integrals explicitly), show that in order
‡
for N to remain constant as the temperature is raised, the chemical potential P15E.6 P.G. Radaelli et al. (Science 265, 380 (1994)) reported the synthesis and
must decrease from its value at T = 0. structure of a material that becomes superconducting at temperatures below
704 15 Solids
45 K. The compound is based on a layered compound Hg2Ba2YCu2O8−δ, which replacing Y by Ca, accompanied by a change in unit cell volume by less than
has a tetragonal unit cell with a = 0.38606 nm and c = 2.8915 nm; each unit cell 1 per cent. Estimate the Ca content x in superconducting Hg2Ba2Y1−xCaxCu2O7.55
contains two formula units. The compound is made superconducting by partially given that the mass density of the compound is 7.651 g cm−3.
TOPIC 15F The magnetic properties of solids

Discussion question
D15F.1 Compare and contrast the polarization (Topic 14A) with the
magnetization.
2+ −8 3 −1
E15F.4 The magnitude of the magnetic moment of Mn in its complexes is E15F.8 Data on a single crystal of NiSO4·7H2O give χm = 6.00 × 10 m mol at
typically 5.3μB. How many unpaired electrons does the ion possess? 298 K. Identify the effective number of unpaired electrons in this compound
and compare your result with the theoretical value.
E15F.5 Calculate the molar susceptibility of benzene given that its volume
susceptibility is −7.2 × 10−7 and its mass density is 0.879 g cm−3 at 25 °C. E15F.9 Estimate the spin-only molar susceptibility of MnSO4·4H2O at 298 K.
E15F.6 Calculate the molar susceptibility of cyclohexane given that its volume E15F.10 To what temperature must Nb be cooled for it to remain
susceptibility is −7.9 × 10−7 and its mass density is 811 kg m−3 at 25 °C. superconducting in a magnetic field of 150 kA m−1? The necessary data are
3 −1 given in E15F.3.
E15F.7 Data on a single crystal of MnF2 give χm = 0.1463 cm mol at 294.53 K.
Identify the effective number of unpaired electrons in this compound and
compare your result with the theoretical value.
Problems
‡
P15F.1 J.J. Dannenberg et al. (J. Phys. Chem. 100, 9631 (1996)) carried out P15F.2 An NO molecule has thermally accessible electronically excited states.
theoretical studies of organic molecules consisting of chains of unsaturated It also has an unpaired electron, and so may be expected to be paramagnetic.
four-membered rings. The calculations suggest that such compounds have However, its ground state is not paramagnetic because the magnetic moment
large numbers of unpaired spins, and that they should therefore have unusual of the orbital motion of the unpaired electron almost exactly cancels the
magnetic properties. For example, the lowest-energy state of the compound spin magnetic moment. The first excited state (at 121 cm−1) is paramagnetic
shown as (5) is computed to have S = 3, but the energies of S = 2 and S = 4 because the orbital magnetic moment adds to, rather than cancels, the spin
structures are each predicted to be 50 kJ mol−1 higher in energy. Compute the magnetic moment. The upper state has a magnetic moment of magnitude
molar magnetic susceptibility of these three low-lying levels at 298 K. Estimate 2μB. Because the upper state is thermally accessible, the paramagnetic
the molar susceptibility at 298 K if each level is present in proportion to its susceptibility of NO shows a pronounced temperature dependence even near
Boltzmann factor (effectively assuming that the degeneracy is the same for all room temperature. Calculate the molar paramagnetic susceptibility of NO and
three of these levels). plot it as a function of temperature.
TOPIC 15G The optical properties of solids

D15G.1 Explain the origin of Davydov splitting in the exciton bands of a D15G.2 Explain how the nonlinear response of a material to an electric field
crystal. may give rise to frequency doubling. Why is frequency doubling typically seen
only when using an intense beam from a laser as the light source?
Additional exercise
E15G.2 The band gap in silicon is 1.12 eV. Calculate the maximum wavelength
of electromagnetic radiation that results in promotion of electrons from the
valence to the conduction band.
Problems
P15G.1 This and the following problem explore quantitatively the spectra S b (1) b (2)d ; do the same for the second eigenvector, which corresponds
of molecular solids. First consider a dimer formed from two identical to Ψ− = ψb(1) − ψb(2). (c) The monomer transition dipole moment is
monomers. For the first monomer, the normalized ground state wavefunction μmon = ∫ψ b (1)μˆψ a (1)dτ =∫ψ b (2)μˆψ a (2)dτ . For the dimer the transition
is ψa(1) and the normalized excited state wavefunction is ψb(1); for the second moment is μdim = ∫Ψ ± μˆΨ 0 dτ , where Ψ0 is the wavefunction of the dimer
monomer the wavefunctions are ψa(2) and ψb(2)—the label in parenthesis ground state. Because it is assumed that there is no interaction between the
identifies to which monomer the wavefunction refers, but otherwise ψa and ground-state wavefunctions of the dimer, Ψ0 can be written as (1/21/2)(ψa(1) +
ψb are the same for each monomer. In each monomer there is a transition ψa(2)). Find expressions for μdim , for the two excited state wavefunctions, Ψ±.
between ψa and ψb with transition dipole moment μmon and wavenumber In solving this problem it is helpful to realize that it is closely analogous to the
νmon . For convenience the energy of the ground state is taken as zero, so the overlap of two atomic orbitals to give molecular orbitals (Topic 9E).
energy of the excited state, expressed as a wavenumber, is νmon . It is assumed
P15G.2 Continues from the previous problem. (a) Consider a dimer formed of
that dimerization does not affect the ground state wavefunctions, but the
monomers which have μmon = 4.00 D, mon 25000 cm 1, and r = 0.50 nm. Plot
excited state wavefunctions become mixed so the excited state of the dimer
a graph to show how the energies (expressed as a wavenumber) of the excited
has wavefunctions Ψ± = c1,±Ψb(1) + c2,±Ψb(2); the mixing of the two monomer
states,  vary with the angle θ. (b) Now expand the treatment given above to a
wavefunctions gives two dimer wavefunctions, denoted , the coefficients
chain of N monomers with μmon = 4.00 D, mon 25000 cm 1, and r = 0.50 nm.
ci,± are to be determined. In the basis ψb(1), ψb(2) the hamiltonian matrix has
For simplicity, assume that θ = 0 and that only nearest neighbours interact
the form
with interaction energy V (expressed here as a wavenumber). For example the
 νmon β  hamiltonian matrix for the case N = 4 is
Ĥ =  
 β ν 
 mon   νmon V 0 0 
 
 V νmon V 0 
The diagonal elements are the energies (as a wavenumber) of the excited Hˆ = 
state of the monomer. The off-diagonal elements correspond to the energy of  0 V νmon V 
interaction between the transition dipoles. Using the arrangement illustrated  0 0 V νmon 

in (1) of Topic 15G, this interaction energy (as a wavenumber) is:
This matrix is analogous to the one that characterizes a band in a solid
mon
2
 (1 3 cos 2 ) (15C.1b) and so the eigenvalues (which in this case are the wavenumber of
4 0hcr 3 the transitions) can be written down by analogy with eqn 15C.1. Calculate the
The eigenvectors of the hamiltonian matrix are the wavefunctions for the wavenumber of the lowest energy transition for N = 5, 10, and 15, and then
c
excited state of the dimer, and these can be written (c12,,±± ). The eigenvalues are generalize your result for large N. (c) How does the transition dipole moment
the energies corresponding to these wavefunctions, and because it has been of the lowest energy transition vary with the size of the chain?
assumed that the ground states of the dimer are the same as for the monomer, P15G.3 Show that if a substance responds nonlinearly to two sources of
these energies will correspond to the transitions in the dimer. radiation, one of frequency ω1 and the other of frequency ω2, then it may
1 1
(a) Show that (1 ) and (−1 ) are eigenvectors of the hamiltonian matrix give rise to radiation of the sum and difference of the two frequencies. This
and that the corresponding eigenvalues are  mon  . (b) The first nonlinear optical phenomenon is known as frequency mixing and is used
eigenvector corresponds to writing the wavefunction as Ψ+ = c1,+ψb(1) + to expand the wavelength range of lasers in laboratory applications, such as
c2,+ψb(2) = ψb(1) + ψb(2). Normalize the wavefunction, assuming that spectroscopy and photochemistry.
FOCUS 15 Solids
Integrated activities
I15.1 Calculate the thermal expansion coefficient, α = (∂V/∂T)p/V, of diamond f as a function of (sin θ)/λ. Hint: Interpret 4πρ(r)r2 as the radial distribution
given that the (111) reflection shifts from 22.0403° to 21.9664° on heating a function P(r); use mathematical software to evaluate the necessary integrals.
crystal from 100 K to 300 K and 154.0562 pm X-rays are used. (c) Explore how the scattering factor changes when the actual 1s wavefunction
of a hydrogenic atom is replaced by a Gaussian function. Use mathematical
I15.2 Calculate the scattering factor for a hydrogenic atom of atomic number Z
software to evaluate the necessary integrals.
in which the single electron occupies (a) the 1s orbital, (b) the 2s orbital. Plot

Solid State Chemistry

Uploaded by

Copyright:

Available Formats

Solid State Chemistry

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Solid State Chemistry

Uploaded by

Copyright:

Available Formats

Topic 15A Crystal structure

➤ Why do you need to know this material?

The internal structure of a crystal is a regular array of its atoms,

15A.1 Periodic crystal lattices

• A monoclinic unit cell has one twofold rotation axis

C3 Figure 15A.7 A triclinic unit cell has no axes of rotational symmetry.

System Essential symmetries

Cubic P Cubic I Cubic F C2

Orthorhombic P Orthorhombic C Orthorhombic I Orthorhombic F

Figure 15A.11 Some representative planes in three dimensions

dhk 0 hdhk 0 dhk 0 kdhk 0

Sets of parallel planes {110} {230} {110} {010}

Hence, d123 = 0.21 nm. It then follows immediately that d246 is

Example 15A.1 Exercises

Property Equation Comment Equation number

for structural studies of solid materials. The actual process of

As explained in The chemist’s toolkit 7A.2, a characteristic prop-

Figure 15B.3 The processes that contribute to the generation

15B.1(b) Bragg’s law   2d sin Bragg’s law [alternative form] (15B.1b)

Not all integral values of h2 + k2 + l2 are obtained when integer

d D {hkl} {100} {110} {111} {200} {210}

40 corresponds to sin θ = 0. That is, the scattering factor has its

15B.1(c) Scattering factors 15B.1(d) The electron density

The table for the Cl− ions is

eiϕ + e−iϕ = 2cos ϕ

Electron density, ρ(x)

Neutron and electron

where λ is the wavelength of the electrons in the beam, and

Fourier synthesis  (r )  (1/V )  Fhkl e 2 i ( hx  ky lz )

Patterson synthesis P (r )  (1/V )  | Fhkl |2 e 2 i (hx ky lz ) 15B.5

Wierl equation I ( )   fi f j (sin sRij /sRij ), s  (4 / )sin 12  15B.8

➤ Why do you need to know this material?

➤ What is the key idea?

15C.1(a) Close packing B

Table 15C.1 The crystal structures of some elements⋆

A A Calculate the packing fraction of a ccp structure formed from

The solution As can be seen from Fig. 15C.5, the length of

An hcp structure has the same coordination number, so its

The central aspect of solids that determines their electrical k

Highest level of p band Fully antibonding

→1 as N→ → 0 as N→ The band can be thought of as consisting of N different molecu-

 k k  Brief illustration 15C.1

Brief illustration 15C.2

+z –z +z –z +z –z Table 15C.3 Madelung constants

The cycle is closed, so the sum of these enthalpy changes is

Calculate the lattice enthalpy of KCl.

6 +437 Some lattice enthalpies obtained by the Born–Haber cycle

Figure 15C.15 A fragment of the structure of diamond. Each C

15C.3 Covalent and molecular solids

the individual molecular components (Topic 14B). The ob-

➤ Why do you need to know this material?

➤ What is the key idea?

chanical properties of solids are stress and strain. The stress on

Uniaxial The Young’s modulus of iron at room temperature is 215 GPa.

which corresponds to elongation of the wire by 5.81 per cent.

• A metallic conductor is a substance with an electrical

(a) (b) (c)

at a higher energy, separated from the top of the lower band by

The conductivity of a metallic conductor decreases as the dN (E )   (E )dE  f (E )  (15E.1)

where χ is the dimensionless volume magnetic susceptibility.

15B.1(b) Bragg’s law 2d sin Bragg’s law [alternative form] (15B.1b)

Fourier synthesis (r ) (1/V ) Fhkl e 2 i ( hx ky lz )

Patterson synthesis P (r ) (1/V ) | Fhkl |2 e 2 i (hx ky lz ) 15B.5

Wierl equation I ( ) fi f j (sin sRij /sRij ), s (4 / )sin 12 15B.8

The central aspect of solids that determines their electrical k

k k Brief illustration 15C.1

The conductivity of a metallic conductor decreases as the dN (E ) (E )dE f (E ) (15E.1)

of the electron spins fluctuate. Some o rientations have lower

N A g e2 0 B2 S(S 1) Molar magnetic susceptibility

magnetic moments m (6.3001 106 m3 K mol 1 ) 2 2

Superconductors have unique magnetic properties. Supercon

Induced dipole E 2 E02 cos 2 t 12 E02 (1 cos 2t ) (15G.2)