Swansea University

School
of
ENGINEERING

Ceramic Materials
EG – 383

Ceramics

Prof Neil McMurray

 EG383 - Ceramic Materials

Ceramics

Description

Section 1 Lectures 1 & 2 Introduction

Section 2 Lectures 3 & 4 Single Crystals and Sintering

Section 3 Lectures 5 & 6 Conventional Ceramics

Section 4 Lectures 7 & 8 Advanced Powder Technology

Section 5 Lectures 9 & 10 Glass

Section 6 Lectures 11 & 12 Glass Ceramics

Section 7 Lectures 13 & 14 Zirconia and Transformation Toughening

Section 8 Lectures 15 & 16 Advanced Ceramic Applications

Zirconia Electro-ceramics

Section 19 Lecture 17 & 18 Advanced Ceramic Applications

High Temperature Engineering Ceramics

Section 10 Lectures 19 & 20 Fibre and Whisker Reinforced Ceramics

2
 Section 1 - Introduction
(Lectures 1 & 2)

The word ceramic derives from the Greek ‘Keramos’ which means ‘burnt-stuff’ or pottery. The advent of
ceramics was in even earlier times than that of the ancient Greeks and as such they represent the oldest
man-made materials. House bricks, earthenware pots and porcelain cups are everyday examples of the
use of ceramic materials. However, nowadays the word ceramic has come to be applied to a much wider
range of materials than those used to make these commonplace items. We will classify as ceramics all
non-metallic and inorganic solids and the following materials fall under this classification: the traditional
clay ceramics , the new ‘purer’ ceramics, eg. alumina and silicon nitride, glasses, glass-ceramics, and
cement.

Ceramic composites consist of a mixture of two distinct materials; one of the materials is known as the
reinforcing phase and the other as the matrix. The reinforcing phase, which is normally in the form of
fibres although particulates are also used, is embedded in the matrix. The reason for employing such a
mixture is to optimise the mechanical properties, as far as possible, by utilizing the best features of the
reinforcing and matrix materials. Even though there are some examples of the use of composites in
ancient civilisations, eg. straw-reinforced mud bricks, ceramic composites have only been developed to
the point where they are used in both technological and everyday applications in the last few decades.
The most commonplace composites consist of ceramic fibres, which are strong, in a low-density polymer
matrix, although other combinations such as ceramic fibre-metal matrix and ceramic fibre-ceramic matrix
are also encountered.

Interatomic Bonds in Ceramics

With the exception of carbon, ceramics are made up of a at least two types of atoms. In fact, most
ceramics are a combination of one or more metals with a non-metallic element, eg. NaCl, MgO, FeAl2O4.
The class of greatest interest are oxides since they are generally stable in air at high temperatures.
However, carbides, nitrides and halides are also used. Ceramic crystals are primarily ionic or covalent in
character, although features of both types of bond are almost always present. Thus, the ideal conditions
for the formation of an ionic bond occur only for very special pairs of atoms, and the perfectly covalent
bond is similarly rare. In practice, the electrons will be shared unequally between the atoms involved. It
is therefore necessary to have some criterion for deciding what the nature of any particular bond will be.
Such a criterion is supplied by the concept of electronegativity, which is a measure of the power of an
atom in a molecule to attract electrons. The electro-negativities of elements frequently encountered in the
field of ceramics are given in Table 1.1.

Table 1.1 – Electronegativities of Elements

Li Be B C N O F
1.0 1.5 2.0 2.5 3.0 3.5 4.0
Na Mg Al Si P S Cl
0.9 1.2 1.5 1.8 2.1 2.5 3.0
K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Gc As Sc Br
0.8 1.0 1.3 1.5 1.6 1.6 1.5 1.8 1.8 1.8 1.9 1.6 1.6 1.8 2.0 2.4 2.8
Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Tc I
0.8 1.0 1.2 1.4 1.6 1.8 1.9 2.2 2.2 2.2 1.9 1.7 1.7 1.8 1.9 2.1 2.5
Cs Ba La-Lu Hf Ta W Re Os Ir Pt Au Hg TI Pb Bi Po At
0.7 0.9 1.1-1.2 1.3 1.5 1.7 1.9 2.2 2.2 2.2 2.4 1.9 1.8 1.8 1.9 2.0 2.2
Fr Ra Ac Th Pa U
0.7 0.9 1.1 1.3 1.5 1.7

The difference in electronegativity between the two atoms constituting the ceramic materials can then be
evaluated. The greater this difference, the more likely it is that one atom will take one or more electrons
from the other ie. the greater the amount of ionic character in the resulting bond. The relation between the
electronegativity difference and the percentage of ionic character in the bond is given in Table 1.2.

3
Example - CaO

Ca = 1.0 and 0 = 3.5

∴ difference in electronegativity = 2.5
∴ percentage ionic character ≅ 79%

Table 1.2
Relation between electronegativity difference and percentage of Ionic Character
Electronegativity difference % ionic character
0.2 1
0.4 4
0.6 9
0.8 15
1.0 22
1.2 30
1.4 39
1.6 47
1.8 55
2.0 63
2.2 70
2.4 76
2.6 82
2.8 86
3.0 89
3.2 92

Table 1.3
Bonding in ceramics
(where a single bond type is given it comprises over 70% of the bonding)
Material Bonding
Si Covalent
SiC Covalent
Si3N4 Covalent
NaCl Ionic
MgO Ionic
Mica Ionic
Al2O3 Covalent-ionic
SiO2 (quartz) Covalent-ionic
Soda-lime glass Covalent-ionis

The ionic and covalent bonds are therefore the limiting conditions of a continuous series. The proportion
of ionic bonding in a compound then indicates the position of the compound in this series, Table 1.3
shows the predominant bonding type in some common ceramics.

Mechanical Properties

Strength – The high compressive strength of ceramics has long been known to man and was well
established by the time Pheidias constructed his Pathenon in Athens in about 450 B.C. In more recent
times Messrs Josiah Wedgwood and Sons Ltd published a photograph showing an eleven-tonne double-
decker bus supported by six of their fine bone china coffee cups.

Although ceramics are hard and reasonably strong they lack ductility and this, in turn, prevents any stress
concentrations which may be present from being relieved by the onset of plastic flow. Whereas metals are
malleable and ductile most ceramics, particularly where covalent bonds are present, are brittle. This is
due largely to the overall width of dislocations present in the structure and this in turn is dependent on the
nature of interatomic bonds. In a metal the ions are held in position by their mutual attraction for the

4
‘flexible’ electron cloud, whilst in ceramics such as diamond the strong covalent bonds are strictly
directional. These factors in turn affect the equilibrium width of any dislocations present., the width of a
dislocation is defined as the distance over which the displacement of atoms or ions is greater than b/4
where b is the Burgers vector of the material.

In Figure 1.1 it will be seen that the dislocation width, w, of the dislocation in the case of a metal (i) is
much greater than in a covalently bonded material (ii). In fact it is estimated that with malleable metals
such as gold, copper and aluminium dislocation widths of about ten ion spacings are present. When the
width is great the force necessary to move such a dislocation is small because the highly distorted region
at the centre of the dislocation is not critically localised on a single particular crystal site. In figure 1(ii),
however, a large force will be necessary to move the dislocation away from its strongly locallised site. At
the same time there is a comparative lack of slip systems in a ceramic material (see overleaf) and these
influences will combine to cause the interatomic forces of attraction in the region of the dislocation to be
exceeded so that the fracture occurs. However, it is the difference in the width of dislocation as between
metals and covalently- or ionically- bonded materials which are primarily responsible for the difference in
their mechanical proportion.

Figure 1.1 - Illustrating the width of a dislocation: (i) a wide dislocation (metal):
(ii) a narrow dislocation (covalently-bonded material).

(b) movement of an edge dislocation

Because many ceramics exhibit a high percentage of ionic bonding not all the slip planes in the crystal
structure will actually be available. This is because of the electrostatic repulsion between ions of similar
charge, as shown in figure 1.2(ii).

5
 Figure 1.2 - Deformation in crystals. (i) ‘Slip’ occurs in a metallic crystal. (ii) ‘Cleavage’ occurs in ionic crystals.

The force required to move a dislocation through the crystal lattice is called the Peierls-Nabarro force.
The Peierls (ττp) stress is the shear stress required to move a dislocation through a crystal lattice in a
particular direction

τ p ~ G exp(− 2πw / b ) (1.1)

Where G is the modulus and b is the Burgers vector, ie distance between atoms in the slip direction. Note
that the dislocation width appears in the exponential term in equation (1.1), so that the Peierls stress will
be very sensitive to the value of w.

The Peierls equation (1.1) shows that metals, with wide dislocations, will require a low stress to move the
dislocations. Physically, this is so because when the dislocation is wide, the highly distorted region at the
core of the dislocation is not localized on any particular atom in the crystal lattice. However, in ceramic
materials with directional covalent bonds, the interfacial energy is high and the dislocations are narrow
and relatively immobile. This fact combined with the restriction on slip systems imposed by the
requirements of electrostatic forces results in the low degree of plasticity of ceramic materials. Ceramics
become more ductile at high temperatures because thermal activation helps the dislocations overcome the
energy barrier.

Microcracks and Flaws since ceramics tend to suffer from the presence of microcracks which act as
stress raisers it follows that tensile stresses must generally be kept low if sudden failure is to be avoided.
Moreover the incidence of such faults in a ceramic is much more variable than in a metal where the
quality of the structure is relatively easy to control. Consequently a bigger ‘scatter’ of measured strengths
is to be expected in ceramics. Their compressive strengths, however, are high compared with their tensile
strengths. Retention of strength at high temperatures is one of the useful features of such materials. For
example, titanium diboride will maintain a compressive strength of 250 MPa at 2000°C making it one of
the strongest materials known at that temperature.

Low Ductility in most ceramics ductility is, for all practical considerations, zero. This lack of ductility
arises principally from the low dislocation width found in ceramics exhibiting covalent or mixed covalent
– ionic bonding, as discussed in the preceding section. The lack of available slip systems is also
important, particularly in the case of polycrystalline ceramics where slip systems are unlikely to be co-

6
planar (ie to line up) between adjacent crystals. In practice of course such internal imperfections as
porosity and microcracks further reduce strength and ductility. Such faults are more difficult to control
during manufacture than they are in metals since most engineering ceramics are compacted from powders
which suggests that some porosity is inevitable. A few ceramics such as MgO in which bonding is
primarily ionic exhibit limited ductility when in the form of a single crystal but in general ceramics can
only be classed as ‘brittle’.

Creep This takes place in crystalline ceramic materials at relatively high temperatures since crystalline
ceramics are stable solids with high melting points. Glasses on the other hand are non-crystalline and
have fairly low softening temperatures so that considerable creep will occur at moderate temperatures.
With these materials steady-state creep occurs according to an Arrhenius-type expression (1.2):

dε (1.2)
= Aσ n exp( −Q / RT )
dt

where A is a constant, Q is the activation energy (also a constant for the condition), R is the universal gas
constant, T the absolute temperature, σ the stress and n the time exponent which for many ceramics and
glasses is in the region of 10.

Hardness Ceramic materials generally are hard when compared with most other engineering materials.
This hardness makes certain ceramics useful as abrasives or, suitably bonded, as cutting tool materials.
Thus, cubic boron nitride is almost as hard as diamond whilst materials like silicon carbide, alumina
(aluminium oxide) and beryllium oxide are already well known. Some hard ceramics are listed in Table
1.4, hardened tool steel being included for comparison.

Table 1.4 – Hardness of Some Ceramics

Material Knoop hardness number
Diamond 7000
Cubic boron nitride 6900
Titanium diboride 3300
Boron carbide 2900
Silicon carbide 2600
Silicon nitride 2600
Alumina 2000
Beryllium oxide 1220
Hardened steel 700

The hardest of these substances depend for their properties on the very strong stable nature of their
directional covalent bonds.

Refractoriness This generally defined at the ability of a material to withstand high temperatures without
appreciable deformation under service conditions. Some refractories like fireclay and high-alumina brick
soften gradually over a range of temperature and may ultimately collapse under load at a temperature well
below the normal fusion temperature. Materials most widely used as refractories in furnaces are clays
containing various proportions of silica (SiO2) and (Al2O3).

Certain ceramic materials have very high melting points indeed and for special purposes these may be
used, assuming that their relative high cost is justified. A few of these high temperature ceramics are
included in Table 1.5.

7
 Table 1.5
Melting Points of some High-temperature Refractory Materials
Material Melting point (°C)
Hafnium carbide 3900
Tantalum carbide 3890
Thorium oxide 3315
Magnesium oxide 2800
Zirconium oxide 2600
Beryllium oxide 2550
Aluminium oxide 2050
Silicon nitride 1900
Fused silica glass 1680(approx)
Soda glass 340-430(approx)

Again, as with hardness, this property can be thought of as a function of chemical bond strength. Strongly
bonded materials require a large input of thermal energy to disrupt the atomic bonds.

Electrical Properties Until a short while ago the use of ceramics in the electrical industries was restricted
to insulation. Today ceramic materials include many substances varying in properties from low-loss high-
frequency insulators to semi-conductors, conductors and ferro-magnets. Consequently they are used as
thermistors, thermoelectrics, transistors and storage cells in memory systems. Indeed the current race to
develop a really high-temperature superconductor is based at present on sophisticated ceramics.

8
 Section 2 - Single Crystals and sintering
(Lectures 3 & 4)

Single Crystal Synthesis

Physically large single crystals of ceramic material are technically important in a number of applications.
The principal difficulties to be overcome are the high melting point of many ceramics and the slow rate of
crystallisaiton from viscous ceramic melts.

Hydrothermal growth of materials is typically carried out in the pressure range of 50-500 MPa (ca 500-
5000 atm). By such processes many beautiful crystals are produced in nature. Hydrothermal growth
refers to a process in which nutrient is dissolved in one part of the system, and is transported to another
region because of thermal gradients. There it is deposited as a single crystal. In most modern processes, a
mineralizer is added to the water to increase the solubility of the nutrient. Crystal growth under higher
pressure conditions includes the prevention of volatilisation, increased solubility, and factors relating to
the deposition of the material onto preferred sites on the crystal faces. Hydrothermal growth avoids the
need to melt the ceramic altogether.

The most important example is the high pressure synthesis of quartz (SiO2) crystals which are used as
oscillators in watches and other instruments. As shown in Figure 2.1, the nutrient is at the bottom of the
vertical pressure vessel at 400°C and 170 MPa (ca 1700 atm). The aqueous, alkaline solution transports
the SiO2 to a region at 340°C where seed crystals are held in a cage. Grown crystals are typically 100-
200mm long and many are grown in one operation in autoclaves which may be 10m high.

Figure 2.1 – Cross section of an autoclave for growing quartz crystals. To convert MPa to atm, divide by 0.101

9
Catalyst Solvent Process Diamond is a polymorph of carbon found naturally in certain locations,
particularly the Kimberly area of South Africa. At normal pressure, it is a metastable phase, graphite
being the stable polymorph. The equilibrium line P(T) separating the two phases is sketched in Figure
2.2. At room temperature, a pressure of only 2 MPa (ca 20 atm) is required to reach the stability region of
diamond, but much higher pressures are required to effect the transformation, because of kinetic factors.

Figure2.2 – The equilibrium phase line for diamond-graphite. Diamond growth from Ni-C solution is carried out in the
cross-hatched region (14). To convert GPa to atm, divide by 1.013 x 10-4

Industrial diamonds are produced from metal/carbon solutions (catalyst-solvent process), which in many
ways resembles the hydrothermal growth processes outlined above with the liquid metal taking the place
of the aqueous solution. Diamond growth only occurs in the shaded pressure-temperature region in Figure
2.2. Temperature must be sufficiently high to produce a melt, and pressure sufficiently high to precipitate
the diamond phase. Typical conditions are 6 GPa (ca 60,000 atm) and 1800 K.

Diamond was synthesized first in the laboratory in 1955 (General Electric Laboratory in the United States
and ASEA in Sweden). The process was commercialised in 1957. Today it is an important source of
industrial, abrasive diamond material. Industrial diamonds are black due to nitrogen (N) in the crystal
structure. Large, transparent single crystals can also be grown, but the process in uneconomical. Tools
and wire-drawing dies are also manufactured with polycrystalline compacts of diamond.

The belt apparatus for diamond synthesis is depicted in Figure 2.3. For pressures above 2 GPa (ca 20,000
atm), tapered pistons are employed which are advanced to compress the enclosed sample directly. In the
belt apparatus, two opposed pistons compress the sample held in a shaped cylinder. The gasket deforms
in such a way that the sample cannot extrude. Laterial forces are applied to the pistons and belt which
increase the internal pressure that can be contained without yielding. Critical parts of the apparatus are
constructed of cemented tungsten carbide. The gaskets also act as insulators with the electrical current
passing through the pistons and sample producing the required temperature.

10
 Figure 2.3 (a) Schematic of the belt apparatus used for diamond synthesis; (b) detail of the cell (14)

Cubic boron nitride, the second hardest substance known, is also produced industrially. It is used as an
abrasive, particularly for shaping steels to which diamond abrasives tend to stick (see Abrasives; Boron
compounds).

Crystallization from melt many ceramic compositions are utilized as single crystals that are carefully
crystallized from a melt. A variety of techniques have been developed for growing single crystals from a
melt. Important examples include the:

1 Verneuil process
2 Czochralski technique
3 EFG technique
4 Gradient furnace technique

The Verneuil process (figure 2.4) is a flame fusion technique developed between 1886 and 1902 by
Auguste Victor Louis Verneuil. A controlled flow of fine Al2O3 powder is gravity-fed through a
hydrogen-oxygen torch or an oxygen-gas torch at temperatures above 2050°C (3720°F). The Al2O3
particles fuse and deposit on an Al2O3 pedestal. The position of the pedestal and temperature of the torch
are carefully controlled to nucleate and sustain growth of a single crystal. As the Al2O3 accumulates and
the crystal grows, the pedestal is slowly lowered until a rounded, elongated single-crystal “boule” has
formed. Boules as large as 9 cm (3.54 in) in diameter reportedly have been produced and the total world
production in one year was approximately 200,000 kg. Most were machined into “jewel” bearings for
watches and instruments. However, this market has decreased since the commercial introduction of
electronic watches.

11
 Al2O3 POWDER

H2 GAS

O2 GAS

H2/O2 FLAME

SEED CRYSTAL

PEDESTAL

Figure 2.4 – Verneuil Process

A variety of different coloured synthetic gemstones have been produced by the Verneuil process by
doping the Al2O3 with different impurities. A ruby red colour results from Cr additions and a sapphire
blue colour from Ti plus Fe. Orange is achieved with Ni + Cr + Fe, yellow with Ni, yellow-green with Ni
+ Fe + Ti, green with Co + V + Ni, and purple with Cr + Ti + Fe.

The Verneuil process does not produce high-optical-quality crystals. The crystals have inclusions, voids,
growth rings, distortions, and residual stress and are not suitable for precision optical applications.

The Czochralski process produces much-higher-optical-quality single-crystal Al2O3, which is relatively

free of inclusions, growth rings, and dislocations. This material is suitable for laser and other optical
applications. Al2O3 is melted in an iridium crucible. A seed crystal is touched to the surface of the melt
and slowly withdrawn as illustrated in Figure 2.5. Large sizes are feasible. Crystals reportedly have been
grown from a 15-cm (5.9 in) diameter crucible.

12
 Figure 2.5 Apparatus for the Czochralski method for growth of single crystals

EFG Technique stands for ‘edge-defined film-fed growth’ technique. The EFG technique is a
modification of the Czochralski process and is illustrated in Figure 8. A molybdenum die is immersed in
the molten alumina. A capillary hole allows molten alumina to be wicked by surface tension to the top of
the die. A seed crystal is touched to the surface of the melt at the top of the die and slowly pulled away to
start single-crystal growth. A film of the melt spreads over the surface of the die and the growing crystal
assumes the cross-sectional shape defined by the sharp edges of the die. The EFG technique can produce
shapes similar to those that are produced by extrusion.

Figure 2.6 Schematic illustrating single crystal growth by the edge-defined film-fed (EFG) technique.

13
The EFG process was developed as an outgrowth of studies to grow single-crystal filaments. Filaments as
fine as 0.005 cm (0.002 in) in diameter have been grown. Continuous filaments 0.025 cm (0.009 in) in
diameter were grown at a rate of 60 m/hr (197 ft/hr). Twenty-five filaments were grown simultaneously.
These had a tensile strength of 2758 MPa (400 ksi). Single-hole and multihole single-crystal tubing over
1.5-m (5-ft) long also have been produced, as well as plates up to 1.25-cm (0.5-in) thick, 30-cm (11.8-in)
wide, and 30-cm (11.8-in) long. The need for large, high-quality single crystals for lasers, laser windows,
substrates, transparent armor, and other applications led to the development of the gradient furnace or
heat-exhanger technique. A seed crystal is placed in the bottom of a molybdenum crucible and covered
with alumina chips. Helium gas is impinged through a tungsten tube onto the bottom of the crucible
directly under the seed crystal (figure 2.7). The crucible is heated to melt all of the alumina except the
seed crystal, which is kept cooler by the helium flow. The temperature of the melt and seed crystal are
then controlled by adjusting the furnace and helium flow to initiate crystallisation of the melt on the seed
crystal. Continued control of the temperature gradient of the melt and growing crystal results in
crystallisation of the complete contents of the crucible into a single crystal. Sapphire crystals free of light
scatter have been grown in the 30.5-cm (12-in) diameter, 12.7-cm (5-in) thick size range. An example is
illustrated in Figure 2.8. Average dislocation densities less than 1 x 103/cm2 can be achieved.

Furnace
Al2O3 Chips

Seed
Crystal

←→ He Gas

Figure 2.7 – Gradient Furnace

14
Figure 2.8 Single-crystal sapphire grown by the heat-exchanger method

15
Fabrication by Sintering Route

Overview
Owing to the generally high melting points of the raw materials involved, the fabrication of ceramic
materials commonly includes a heat treatment in which a powder, already formed into a required shape, is
converted into a dense solid by sintering.

Raw material
Powder Preparation

Ceramic Powder
Water
and Consolidation
Binder
“Green Body”
Firing
(Sintering)
Dense Product

Powder preparation is carried out by one of two routes:

1 Conminution ie particle size-reduction grinding or pulverisation or pre-existing material. This is
conventional process route for clays, silicates.
2 Precipitation – material is synthesised and solid particles precipitate from gas or liquid phase.

There are three main categories of processes by which sintering can take place, these being dependent on
the composition being fired and, in particular, on the extent to which a liquid phase is formed during the
heat treatment. These three are: viscous-flow sintering, liquid phase sintering and solid state sintering

Viscous-Flow Sintering
On the basis that random packing of a powder consisting of uniform spheres is capable of reducing the
pore volume (residual gas space) to some 36% of the total, it can be seen that the conversion of some 20%
or more of the original solid mixture to liquid at the firing temperature will be sufficient (see Figure 2.9)
to provide liquid for the filling of the remaining pores. Thus, consolidation of the material can occur by
formation of the liquid, flow of the liquid under the capillary forces exerted by the powder, and either
recrystallisation or vitrification (glass formation) of the liquid on cooling. The result is a dense product.
This process of viscous-flow sintering or vitrification is especially important for products formed form
naturally available raw materials, such as the clays. It is the basis, for example, of the consolidation of
porcelains.

Figure 2.9 Viscous-flow sintering or (where the liquid on cooling is present as a glass) vitrification: the shaded spheres
represent a low-melting component in the mixture.

16
Liquid-phase Sintering

In this category, the composition and firing temperature are selected such that some liquid is formed, but
in a quantity insufficient to fill the residual gas phase simply by viscous flow. Under these conditions, the
attainment of complete densification requires some change in the shape of the solid grains so that the
residual gas volume is eventually reduced to the point where the liquid can fill it (see Figure 2.10). This
process of liquid-phase sintering involves flattening where the solid particles make contact and, hence,
requires a degree of solubility of the solid phase in the liquid. As in viscous-flow sintering, the driving
force for densification lies in the reduction of the surface energy resulting from the reduction in the
interfacial area between liquid and gas phases. Where other driving forces are present (eg those arising
from chemical forces), other microstructural changes will occur simultaneously with densification. This
process is important for many ceramic products such as basic refractories, alumina insulators and barium
titanate dielectrics.

Figure 2.10 – Liquid-phase sintering: the driving force for densification and contact flattening arises from the surface
energy of the small pores present in the liquid in the centre state

Solid-State Sintering

In solid-state sintering, the composition and firing temperature are such that no liquid is formed, all
densification being achieved by reshaping of the powder. This reshaping, which is most commonly
achieved by solid-state diffusion of atoms, is driven by the energy reduction achieved by elimination of
the solid-gas interface and its replacement by a solid-solid interface (see Figure 2.11). This process is
employed for high-quality technical ceramics such as alumina lamp envelopes and ferrite magnetic
ceramics. It requires the use of fine powders and high temperatures in order to allow sufficient atom
diffusion to bring about consolidation in reasonable times.

Figure 2.11 Solid-state sintering: grain reshaping results in the replacement of the solid-gas interface in(a) by the solid-
state grain boundaries in (b)

17
 Solid State Sintering Sequence

2 x Separate 1 x Grain
Particle Surface Boundary

Figure 2.12 - For a powder compact, microstructural changes that occur during solid state sintering
(a)Powder particles after pressing.
(b) Particle coalescence and pore formation as sintering begins.
(c) As sintering proceeds, the pores change size and shape.

Figure 2.12 shows how initially the particles sinter together giving a continuous pore phase. At this stage,
the individual particles can still be distinguished. As firing continues, the pores decrease in size and grain
boundaries develop between particles.

As the grain boundaries begin to migrate at the firing temperature, the original particles cannot be
distinguished. Later, the continuous pore channels (along the line where three grains meet) begin to pinch
off leaving isolated pores. This pinching off occurs when the channels can reduce their surface area by
changing from a cylinder to a series of spherical pores. Densification in sintering is a diffusion process
and therefore depends on the time and temperature. Moreover, diffusion in the solid state is most rapid
along grain boundaries. Sintering of voids is therefore enormously slow if the voids are not located on
grain boundaries. In fact, pores effectively continue to shrink only when on boundaries so that grain
boundary migration, which can move the boundaries away from the voids, may inhibit sintering by
isolating the voids inside the grains. Control of migration is therefore important if low porosities are
needed. One form of grain growth is particularly damaging but frequently observed during solid state
sintering. This is called exaggerated grain growth (or discontinuous grain growth or secondary
recrystallisation) and results in some grains growing markedly.

18
Exaggerated growth occurs when normal grain growth is inhibited by impurities, inclusions, pores etc.
Thus, when a few grain boundaries break away from their pinning forces and these grains can grow very
rapidly. Usually, densification continues by diffusion until the porosity is reduced to ~10%. (N.B at high
porosities, virtually all the porosity is ‘open’ or interlinked, but pinching off of the continuous pores
increases gradually giving ‘closed’ porosity such that ‘open’ pores essentially disappear at ~ 4 – 5% total
porosity). At this stage, the pores have shrunk to a size and separation which allow migration to begin on
certain ‘favourable’ boundaries giving exaggerated grain growth ie the smaller the pores and the greater
their separation, the less they stabilise the grain size. Thus, at ~ 10% porosity, the distribution of void
sizes and spacing is such that the voids impede grain growth for most grains, but occasional boundaries
can break away. Impurities can act in a similar way to pores by hindering migration. However, adding
impurities may not improve sintering since again, certain boundaries may break away from their impurity
atmosphere giving exaggerated grain growth. This phenomenon is also favoured by use of initially
uneven sized powders. To achieve zero porosity (which is possible) migration is best prevented by using
high-purity, uniformly-sized powder particles.

General Considerations

In general, the rate of consolidation of each of the three processes is increased by the use of the fine
powders and high temperatures, this becoming more critical as the quality of liquid is reduced. The
choice of process generally depends on the properties required in the product: where these are limited to
such items as impermeability at room temperature, viscous-flow sintering recommends itself for its
convenience and readily available raw materials; where demands are very specific, such as for the bearing
of mechanical loads at high temperature, the avoidance of liquid phases under these conditions becomes
critical and solid-state processing is favoured. Liquid-phase sintering can be seen as a compromise
position, the requirements for powder quality and high temperature being commonly less stringent than
for solid-state sintering; careful control of the distribution, quantity and eventual nature of the liquid phase
(eg crystallisation on cooling) offers further opportunity for property improvement.

In this discussion, the driving force for consolidation has been considered in terms of the reduction of
surface energy. An additional driving force can be imposed by pressing the powder during the heat
treatment. This technique of hot pressing can be very effective for materials that are otherwise difficult to
densify.

19
 Section 3 - Conventional Ceramics
(Lectures 5 & 6)

Clay-based Ceramics

One of the most widely used ceramic raw materials is clay. This inexpensive ingredient, found naturally
in great abundance, often is used as mined without any upgrading of quality.

Most of the clay-based products fall within two broad classifications: the structural clay products and
the whitewares. Structural clay products include building bricks, tiles, and sewer pipes – applications in
which structural integrity is important. The whiteware ceramics become white after the high-temperature
firing. Included in this group are porcelain, pottery, tableware, china, and plumbing fixtures (sanitary
ware).

The Characteristics of Clay

The clay minerals play two very important roles in ceramic bodies. First, when water is added, they
become very plastic, a condition termed hydroplasticity. This property is very important in forming
operations, as discussed below. In addition, clay fuses or melts over a range of temperatures; thus a dense
and strong ceramic piece may be produced during firing without compete melting such that the desired
shape is maintained. This fusion temperature range, of course, depends on the composition of the clay.

Clays are aluminosilicates, being composed of alumina (Al2O3) and silica (SiO2), that contain chemically
bound water. They have a broad range of physical characteristics, chemical compositions, and structures;
common impurities include compounds (usually oxides) or barium, calcium, sodium, potassium, and iron,
and also some organic matter. Crystal structures for the clay minerals are relatively complicated;
however, one prevailing characteristic is a layered structure. The most common clay minerals that are of
interest have what is called the kaolinite structure. Kaolinite clay [Al2(Si2O5)(OH)4] has the crystal
structure shown in Figure 3.1. When water is added, the water molecules fit in between these layered
sheets and form a thin film around the clay particles. The particles are thus free to move over one
another, which accounts for the resulting plasticity of the water-clay mixture.

Figure 3.1
The Structure of kaolinite clay

20
Compositions of Clay Products
In addition to clay, many of these products (in particular the whitewares) also contain some nonplastic
ingredients; the nonclay minerals include flint, or finely ground quartz, and a flux such as feldspar. The
quartz is used primarily as a filler material, being inexpensive, relatively hard, and chemically unreactive.
It experiences little change during high-temperature heat treatment because it has a high melting
temperature; when melted, however, quartz has the ability to form a glass.

When mixed with clay, a flux forms a glass that has a relative low melting point. The feldspars are some
of the more common fluxing agents; they are a group of aluminsilicate materials that contain K+, Na+, and
Ca2+ ions.

As would be expected, the changes that take place during drying and firing processes, and also the
characteristics of the finished piece, are influenced by the processes, and also the characteristics of the
finished piece, are influenced by the proportions of these three constituents: clay, quartz, and flux. A
typical porcelain might contain approximately 50% clay, 25% quartz, and 25% feldspar.

Fabrication Techniques
The as-mined raw materials usually have to go through a milling or grinding operation in which particle
size is reduced; this is followed by screening or sizing to yield a powdered product having a desired range
of particle sizes. For multicomponent systems, powders must be thoroughly mixed with water and
perhaps other ingredients to give flow characteristics that are compatible with the particular forming
technique. The formed piece must have sufficient mechanical strength to remain intact during
transporting, drying, and firing operations. Two common shaping techniques are utilized for forming,
clay-based compositions: hydroplastic forming and slip casting.

Forming Methods
The forming methods used to produce ceramic bodies are usually classified depending on the amount of
liquid (normally water) which is used e.g.

Forming Process Water (%) Pressure (MN/m2)

Slip casting 25 – 30 0
Plastic forming 12 – 25 0.035 – 0.35
Extrusion 15 – 20 0.35 – 70
Dry pressing 5 – 10 7 – 100
Isostatic pressing 0 – 15 35 – 1000

Slip casting, which is extensively used for clay based ceramics, allows complex shapes to be formed. The
clay body is made into a slurry or slip and cast into a plaster-of-Paris mould. The fine pores in the plaster-
of-Paris absorb water from the clay slip adjacent to the plaster, so that a ‘solid’ clay layer builds up on the
mould wall e.g. Figure 3.2.

Figure 3.2a

21
 Figure 3.2b
The steps in (a) solid and (b) drain slip casting using a plaster of paris mould

When the thickness of clay and build-up on the mould wall is sufficient the mould is drained and the
shaped clay greenware is removed from the split mould.

An alternative to slip casting, also used extensively for clay-base ceramics is plastic forming. This
includes ‘throwing’ on a potters wheel but, more often, involves a ‘jigger’, Figure 3.3.

Figure 3.3 – Operational sequence of the Miller Jigger

Operation of the Ram method of forming ware:
(a) slug of plastic body in mould, (b) pressing the piece, (c) lifting the piece up with the top mould, (d) releasing the
piece by air pressure, (e) blowing moisture from the mould

22
Bricks and tiles are more usually formed by dry pressing. In this case, the powder is poured into a die of
the shape required and a ram is pressed into the die to form the brick, etc, Figure 3.4.

Figure 3.4

In isostatic pressing, Figure 3.5, the powder preform is enclosed in a rubber bag and then subjected to a
high hydrostatic pressure. This process has the advantages that no water need be used, the density is high
and uniform in the green state and the high pressure compaction aids subsequent sintering of the powder
particles. The disadvantages are that the process is costly and the rubber moulds do not allow close
tolerances to be held.

Figure 3.5

Drying and Firing

A ceramic piece that has been formed hydroplastically or by slip casting retains significant porosity and
insufficient strength for most practical applications. In addition, it may still contain some liquid (e.g.
water), which was added to assist in the forming operation. This liquid is removed in a drying process;
density and strength are enhanced as a result of a high-temperature hear treatment or firing procedure. A
body that has been formed and dried but not fired is termed green.

23
Drying

As a clay-based ceramic body dries, it also experiences some shrinkage. In the early stages of drying the
clay particles are virtually surrounded by and separated from one another by a thin film of water. As
drying progresses and water is removed, the interparticle separation decreases, which is manifested as
shrinkage (Figure 3.6).

Figure 3.6
Several stages in the removal of water from between clay particles during the drying process
(a) Wet body. (b) Partially dry body. (c) Completely dry body

Firing

After drying, a body is usually fired at a temperature between 900 and 1400°C; the firing temperature
depends on the composition and desired properties of the finished piece. During the firing operation, the
density is further increased (with an attendant decrease in porosity) and the mechanical strength is
enhanced. Densification during firing occurs through a combination of viscous flow and liquid phase
sintering processes.

When clay-based materials are heated to elevated temperatures, some rather complex and involved
reactions occur. One of these is vitrification, the gradual formation of a liquid glass that flows into and
fills some of the pore volume. The degree of vitrification depends on firing temperature and time, as well
as the composition of the body. The temperature at which the liquid phase forms is lowered by the
addition of fluxing agents such as feldspar. This fused phase flows around the remaining unmelted
particles and fills in the pores as a result of surface tension forces (or capillary action); shrinkage also
accompanies this process. Upon cooling, this fused phase forms a glassy matrix that results in a dense,
strong body. Thus the final microstructure consists of the vitrified phase, any unreacted quartz particles,
and some porosity.

Stress Concentration and Griffith Flaws

The fracture strength of a solid material is a function of the cohesive forces that exist between atoms. On
this basis, the theoretical cohesive strength of a brittle elastic solid has been estimated to be approximately
E/10, where E is the modulus of elasticity. The experimental fracture strengths of most engineering
materials normally lie between 10 and 1000 times below this theoretical value. In the 1920’s, A. A.
Griffith proposed that this discrepancy between theoretical cohesive strength and observed fracture
strength could be explained by the presence of very small, microscopic flaws or cracks that always exist
under normal conditions at the surface and within the interior of a body of material. These flaws are a
detriment to the fracture strength because an applied stress may be amplified or concentrated at t eh tip,
the magnitude of this amplification depending of crack orientation and geometry. This phenomenon is
demonstrated in Figure 3.7, a stress profile across a cross section containing an internal crack. As
indicated by this profile, the magnitude of this localized stress diminishes with distance away from the
crack tip. At positions far removed, the stress is just the nominal stress σ0, or the load divided by the
specimen cross-sectional area. Due to their ability to amplify an applied stress in their locale, these flaws
are sometimes called stress raisers.

24
 Figure 3.7
(a) The geometry of surface and internal cracks. (b) Schematic stress profile along the line X-X′ in (a),
demonstrating stress amplification at crack tip positions.

If it is assumed that a crack has an elliptical shape and is oriented with its long axis perpendicular to the
applied stress, the maximum stress at the crack tip, σm, may be approximated by

12
 a 
σm = 2σ 0   (3.1)
 ρt 

where σ0 is the magnitude of the nominal applied tensile stress, ρt is the radius of curvature of the crack
tip (Figure 3.7a), and a represents the length of a surface crack, or half of the length of an internal one.
Thus for a relatively long microcrack that has a small tip radius of curvature, the factor (a/ρt)1/2 may be
very large. This will yield a value of σm that is many times that of σ0.

Sometimes the ratio σm/σ0 is denoted as the stress concentration factor Kt:

12
σ  a 
K t = m = 2  (3.2)
σ0  ρt 

which is simply a measure of the degree to which an external stress is amplified at the tip of a small crack.

The effect of a stress raiser is more significant in brittle than in ductile materials. For a ductile material,
plastic deformation ensues when the maximum stress exceeds the yield strength. This leads to a more
uniform distribution of stress in the vicinity of the stress raiser and to the development of a maximum
stress concentration factor less than the theoretical value. Such yielding and stress redistribution do not
occur to any appreciable extent around flaws and discontinuities in brittle materials; therefore, essentially
the theoretical stress concentration will result.

Griffith then went on to propose that all brittle materials contain a population of small cracks and flaws
that have a variety of sizes, geometries, and orientations. Fracture will result when, upon application of a
tensile stress, the theoretical cohesive strength of the material is exceeded at the tip of one of these flaws.
This leads to the formation of a crack that then rapidly propagates. If no flaws were present, the fracture
strength would be equal to the cohesive strength of the material. Very small and virtually defect-free
metallic and ceramic whiskers have been grown with fracture strengths that approach their theoretical
values.

25
Griffith Theory of Brittle Fracture

During the propagation of a crack, there is a release of what is termed the elastic strain energy, some of
the energy that is stored in the material as it is elastically deformed. Furthermore, during the crack
extension process, new free surfaces are created at the faces of a crack, which give rise to an increase in
surface energy of the system. Griffith developed a criterion for criterion for crack propagation of an
elliptical crack (Figure 3.7a) by performing an energy balance using these two energies. He demonstrated
that the critical stress σc required for crack propagation in a brittle material is described by

 2 Eγ s 
12 (3.3)
σc = 
 πa 

where E = modulus of elasticity

γs = specific surface energy
a = one half the length of an internal crack.

Worth noting is that this expression does not involve the crack tip radius ρt, as does the stress
concentration equation (Equation 3.1); however, it is assumed that the radius is sufficiently sharp (on the
order of the interatomic spacing) so as to raise the local stress at the tip above the cohesive strength of the
material (Equation 3.3) applies only to completely brittle materials for which there is no plastic
deformation.

26
 Section 4 - Advanced Powder Technologies
(Lectures 5 & 6)

Monodispersed sub-micrometre oxide powders

Since 1980 increasing attention has been paid to the preparation of monodispersed sub-micrometre oxide
powders with a view to obtaining denser packing than can be achieved with non-spherical material and
lower sintering temperature. However, the principles that determine formation of monodispersed particles
were established nearly 40 years ago when La Mer & Dinegar (1950) considered three reaction regions to
describe growth in acidified sodium thiosulphate and in ethanolic sulphur solutions diluted with water
(figure 4.1). The molecular sulphur concentration, Cx, increased above the saturation value, Cs, during an
initial induction period until a critical supersaturation concentration, Css, was reached in time t1 after
mixing reactants. Homogeneous self-nucleation then occurred followed by particle growth involving
diffusion of molecular S to the nuclei up until t3 when Cx = Cs. Monodispersed sulphur sols were
produced when one burst of nucleation occurred in a short time interval, t2-t1, which was achieved by
using low (10-2M or less) reactant concentrations. Uniform growth took place when molecular S was
released slowly because diffusion to nuclei occurred without build-up of solute concentration and further
bursts of nucleation. This qualitative approach to homogeneous nucleation is applicable to sol formation
by cation hydrolysis, hydro-thermal processing of solutions, vapour-phase reactions and the controlled
hydrolysis of alkoxides.

Figure 4.1 – Schematic representation of solute concentration versus time before and after nucleation

Sol Formation through Cation Hydrolysis

Many metal ions (M2+) are subjected to hydrolysis because of a high electronic charge (eg Zr4+) or high
charge density (eg Be2+). Initial products of hydrolysis can condense and polymerise to form polyvalent
metal or polynuclear ions, which are themselves colloidal (figure 4.2). For example
(AlO4al12(OH)24(H2O)12)7+ and (AlO4Al12(OH)25(H2O)11)6+ are produced on increasing the pH (eg with
sodium carbonate) of Al(III) salt solutions for mole ratios of OH/Al ≥ 2. They have a structure consisting
of twelve AlO6 octahedra surrounding a single AlO4 tetrahedron.

27
 Figure 4.2 – Schematic representation of hydrolysis and condensation-polymerisation of polyvalent ions in aqueous
solution

Additional examples of polynuclear ions are shown in table 4.1 which illustrates the wide variety of
cations that form hydrolysis products as well as the formation of different polynuclear species form a
cation (table 4.1).

Table 4.1 – Occurrence of polynuclear cationic hydrolysis products (Baes & Mesmer, 1976)

Hydrous metal oxides (also called hydroxides and hydrated metal oxides) are precipitated for values of
OH/M≥Z

Gas-phase Synthesis of Powdered Ceramics

Gas phase reactions are often associated with production of non-oxide ceramics such as silicon nitride.
Reactants can be gaseous, volatile liquids or solids with reaction temperatures being as high as 15,000 K.
A common feature of powder preparations utilising gases, hydrothermal conditions and alkoxide
hydrolysis is homogeneous nucleation, ie nucleation not requiring ‘seeding’. A variety of techniques
involving flame hydrolysis, laser synthesis and plasma routes that have been used to prepare advanced
ceramic powders will be described.

Flame-hydrolysed powders

These materials, for example SiO2, are made by hydrolysis of SiCl4 in a H2/O2 stationary flame (equation
4.1) and particle formation is shown schematically in Figure 4.3. Molten spherical primary particles grow
to larger droplet by coalescence. As particles being to solidify they stick together, on collision, forming
solid aggregates that also collide and form clusters known as agglomerates. The aggregated nature of
Aerosil® silica, an oxide manufactured by using flame hydrolysis, is shown in figure 4.3.

28
 SiCl4 + 2H2 + O2 → SiO2 + 4HCl 4.1

Figure 4.3 – Schematic diagram illustrating the formation of primary particles, aggregates and
agglomerates in a gas-phase reaction (Ulrich, 1984)

Thermal gas-phase reactions

SiCl4 and SiH4 react with NH3 according to the overall equations:

3SiCl4(g) + 4NH3(g) → Si3N4(s) + 12HCl(g) 4.2

3SiH4(g) + 4NH3(g) →Si3N4(s) + 12H2(g) 4.3

Reacting SiH4 with NH3 at atmospheric pressure between 773 and 1173 K gives Si3N4 powders with
particle diameters in the range 30-200 nm, nitrogen surface areas up to 26 m2 g-1, cation impurities less
than 100 p.p.m. and oxygen content less than 2 weight %, although the NH3/SiH4 ratio affected
crystallisation temperature (up to 1753 K).

Laser Synthesis

An alternative to furnace heating has been pioneered by Haggerty & co-workers at the Massachusetts
Institute of Technology who used a 150 W CO2 laser as the heat source. In their work a cross-flow cell
was used whereby reactant gases diluted in a carrier stream (argon) emerged from a nozzle and
orthogonally intersected the laser beam. Silane was used in preference to SiCl4 because of a strong
absorbance band near the wavelength,10.6 µm, of the laser radiation. Si3N4 was prepared by using the
gases shown in equation (4.3) whereas Si and SiC powders were made according to the reactions.

29
 SiH4(g) → Si(s) + 2H2(g) 4.4

2SiH4(g) + C2H4(g) → 2SiC(s) + 6H2 4.5

Fast heating and cooling rates of 106 K s-1, 105 K s-1 respectively, together with reaction times around 10-3
s, are characteristic properties for these laser-driven processes which yield powders at the rate of about 10
g h-1, while gas pressure, reactant-gas ratios, mass-flow rates and laser intensity control particle size,
crystallinity and stoichiometry of the powders. Representative processing conditions are shown in table
4.2, and table 4.3 contains powder properties. Stoichiometric Si3N4 contains 60 weight % Si and 40
weight % N, which indicates that material in table 4.3 is deficient in nitrogen although stoichiometric
silicon nitride was obtained on increasing the laser intensity; nitride powders were theoretically dense and
crystallised in N2 between 1573 and 1673 K. The oxygen content of Si3N4 powders derived from laser-
heated gases (table 4.3) is less than in Si3N4 made by other methods; lower O2 and hence SiO2 levels are
desirable for sintering and strength of components.

Table 4.2 – Representative processing conditions for laser-driven gas-phase reactions

Table 4.3 – Physical properties of powders derived from laser-driven gas-phase reactions

30
 Figure 4.4 – Transmission electron micrograph for silicon powder made by a laser-driven gas-phase

These reactions offer potential advantages for powder preparation, fast reaction times due to high
temperature and rapid cooling rates leading to a high degree of supersaturation and homogeneous
nucleation.

Radio Frequency Plasma Reactions

Canteloup & Mocellin (1975) developed an r.f. argon plasma reactor operating at 8.2 MHz and a 10 kW
power level in which vapours derived from evaporation of Si (particle size less than 35 µm) and Al
powders injected into the tail of the plasma flame reacted with N2 or NH3 at temperatures as high at 6000
K; metal reactants avoided production of corrosive gases while residence time in the flame was about
1.4 x 102s. A low efficiency (15%) for conversion to a mixture of α- and β-Si3N4 containing unreacted Si
occurred in N2, but much higher yields were obtained in NH3; similar efficiencies resulted with Al.
Particle diameters for AIN were 7-8 nm corresponding to surface areas of 260 m2 g-1 although larger sizes,
11-14 nm, were produced for Si3N4 powders that had surface areas around 160 m2g-1.

Vogt and his co-workers have carried out systematic studies on the synthesis, in thermal plasmas, of SiC,
Si3N4, Al2O3, Ni and B4C powders. Their reactor, shown schematically in figure 4.4, consists of a copper
induction tube (7.5 cm diameter) housed in a quartz mantle that was connected to a water-cooled
stainless-steel quenching chamber (19.5 cm diameter); typical operating parameters were 360 kHz, a
power level between 25 and 40 kW and gas pressure of 8 x 104 Pa. Al2O3 was prepared by oxidation of
aluminium trimethyl, (Al(CH3)3). A comparison of chemical purity between plasma derived and
commercial SiC powders is shown in table 4.4. Figure 4.6 shows a micrograph of plasma derived Al2O3
powder.

31
 Figure 4.5 – Schematic diagram of r.f. plasma reactor

Table 4.4
Comparison of chemical purity between plasma-derived and commercial silicon carbide powders

32
 Figure 4.6 – Plasma-synthesised alumina powder

Stabilisation of Aqueous Dispersions

Even advanced ceramic powders are usually mixed with water prior to forming the green body (as in
conventional clay based ceramics). However, it is important that the powder particles remain un-
aggregated. This is because particle aggregates sinter as though the aggregate was a large single particle
negating the advantages of monodispersity and sphericity in the advanced powder. Stabilisation of
aqueous powder dispersions (sols) may be achieved by electrostatic repulsion between particles or steric
repulsion between polymer chains which have been adsorbed onto the particle surface. Typically the
polymer used will also act as a sort of glue binding together particles in the green body. Such polymers
are called stabilising binders.

Electrostatic Stabilisation
Insoluble oxides in aqueous suspension develop surface electrical charges by surface hydroxylation
followed by dissociation of surface hydroxyl groups as represented by the equations

4.6

4.7

The surface charge, ψ0, a function of pH, results in compensating anions forming a diffuse charge layer
around the surface. The region between the surface and undisturbed electrolyte in bulk solution over
which ψ0 decays to the bulk solution value is known as the electrical double layer, which has a reciprocal
thickness κ given by the expression

κ = (2cNe2z2/ε0εkT)½ 4.8

where c is the ionic strength of the solution, N the Avogadro number, e the electronic charge, z the
valency of the ions, ε0 the permittivity of the vacuum, ε the relative permittivity of the medium, k the
Boltzmann constant and T the absolute temperature. In a 1-1 electrolyte such as potassium nitrate, κ-1 has
a magnitude of ca. 100 nm when c = 10-5 mol dm-3 and decreases to ca. 3nm when c = 10-2 mol dm-3.

Electrostatic repulsion occurs when electrical double layers of sol particles overlap. Double layers are at
equilibrium if the particle approach is slow so that the energy of repulsion VR can be equated, by using
equilibrium thermodynamics, to the isothermal work for bringing particles from infinity to a separation of
closest approach, H0. For two flat plates in a z-z electrolyte.

33
 VR = 64ckT(tanh(zeψ0)/4kT)2exp(-κH0)κ 4.9

and a similar equation applies for spheres:

Steric Stabilisation

The stability of many ‘protected’ colloidal dispersions cannot be explained solely on the basis of
electrostatic repulsion and van der Waals attraction; other stabilising mechanisms must be investigated.
Steric stabilisation’ is a name which is used (somewhat loosely) to describe several different possible
stabilising mechanisms involving adsorbed macromolecules. These include the following:

a. An encounter between particles could involve desorption of a polymeric stabilising agent at the
point of contact. Since adsorption is a spontaneous process, ∆Gads. is negative and ∆Gdes. positive.
This positive free energy of desorption corresponds to particle-particle repulsion and enhanced
stability.
b. When particles collide, their adsorbed layers may be compressed without penetrating into one
another. This ‘denting’ mechanism will reduce the configurations available to the adsorbed polymer
molecules; therefore, there will be a decrease in entropy and an increase in free energy, and stability
will be enhanced by an elastic effect.
c. The adsorbed layers between the particles may interpenetrate and so give a local increase in the
concentration of polymer segments. Depending on the balance between polymer-polymer and
polymer-dispersion medium interactions, this may lead to either repulsion or attraction by an
osmotic mechanism. Enthalpic and entropic changes will be involved. If interpenetration takes
place to a significant extent, elastic repulsion will also operate.

Steric stabilisers are usually block copolymer molecules (eg poly (ethylene oxide) surfactants), with a
lyophobic part (the ‘anchor’ group) which attaches strongly to the particle surface, and a lyophilic chain
which trails freely in the dispersion medium. The conditions for stabilisation are similar to those for
polymer solubility outlined in the previous section. If the dispersion medium is a good solvent for the
lyophilic moieties of the adsorbed polymer, interpenetration is not favoured and interparticle repulsion
results; but if, on the other hand, the dispersion medium is a poor solvent, interpenetration of the polymer
chains is favoured and attraction results. In the latter case, the polymer chains will interpenetrate to the
point where further interpenetration is prevented by elastic repulsion.

34
 Section 5 - Glass
(Lectures 9 & 10)

Glass can simply be defined as “an inorganic product of fusion which has cooled to a rigid condition with
crystallising”. Although this definition is not precise, it does illustrate what happens when glass is made.
Whereas most ceramics are wholly or, at least, partly crystalline in structure, glasses are non-crystalline.
Any material which has cooled from the liquid state to a rigid condition without crystallising is therefore
properly called a glass, regardless of chemical composition. Both organic and inorganic materials can
therefore form glass, but only inorganic ones can be classed as ceramics.

Glasses are sometimes called supercooled liquids because they are true liquids in which only minor
rearrangements have taken place as the liquid cools to a rigid condition. The term ‘glassy’ or ‘vitreous’
are then used to describe the non-crystalline condition. If, in any circumstances, crystals may develop in
the glass on cooling, it is said to have ‘devitrified’. The changes in volume occurring on cooling a liquid
depends on whether the liquid crystallises or not, as seen in figure 5.1.

Figure 5.1

Unlike crystalline solids, glasses do not have an exact melting point, yet it is possible to say that a glass is
molten at a certain temperature since, when it is molten it behaves as a true liquid eg it will take up the
shape of any container and it has a definite and measurable viscosity. As the glass cools, its viscosity
increases steadily and rapidly until eventually the viscosity is so high that the material is, to all extents and
purposes, a solid, ie the elastic constants can be measured. Since glasses do not have a definite melting
point, another concept is introduced. This is the glass-transition temperature – the temperature where
the viscosity is 1015 poise ie the glass is unable to interchange atom positions. At slightly higher
temperatures, when the viscosity (η) is 1013 poise, atoms can rearrange. This is defined as the annealing
temperature – the temperature at which the elastic cooling stresses can be annealed out. At 1013 poise,
strains can be annealed out in ~ 15 minutes, whereas at lower temperatures and viscosities, this process
takes longer eg at 1025 poise, the time to relax strains is ~106 years.

35
The Structure of Glasses
The most important glass forming oxide is silica, SiO2. The crystalline forms of silica are made up of
interlinking silica tetrahedral built up in a regular, periodic manner. The transformations of silica (eg
quartz to tridymite) are very slow and the reverse transformations on cooling are even slower. This is why
crystallisation from the melt is so difficult resulting in a glass instead of a crystalline solid. Glasses can
then be treated as a disordered form of the solid possessing short but not long range order, as seen in
Figure 5.2.

Figure 5.2

In the liquid state, all or nearly all the corners of the tetrahedral remain shared. The very high viscosity of
liquid silica is then a direct result of the tetrahedral being lightly linked in all directions. The thermal
characteristics of silicate glasses are summarised below and illustrated in Figure 5.3.
1 Above the melting point glass is fluid enough to be considered a liquid.
2 The melting point corresponds to the temperature at which the viscosity is 10 Pa-s (100 P); the The
working point represents the temperature at which the viscosity is 103 Pa-s (104 P); the glass is
easily deformed at this viscosity.
3 The softening point, the temperature at which the viscosity is 4 x 106 Pa-s (4 x 107 P), is the
maximum temperature at which a glass piece may be handled without causing significant
dimensional alterations.
4 The annealing point is the temperature at which the viscosity is 1012 Pa-s (1013 P); at this
temperature atomic diffusion is sufficiently rapid that any residual stresses may be removed within
about 15 minutes.
5 The strain point corresponds to the temperature at which the viscosity becomes 3 x 1013 Pa-s (3 x
1014 P); for temperatures below the strain point, fracture will occur before the onset of plastic
deformation. The glass transition temperature will be above the strain point.

Figure 5.3 Logarithm of viscosity versus temperature for fused silica and several silica glasses

36
The sluggishness of the crystal transformations ensures that, on cooling liquid silica, the randomness of
the liquid structure is maintained into the solid, glassy state. The lack of long range order accounts for the
fact that bonds between adjacent tetrahedral are established in the glassy state on cooling – and broken on
heating – progressively rather than simultaneously as in crystals. The glassy state of silica then possesses
a range of temperatures, from ~ 1700K to ~ 2000K, in which the viscosity varies from ~104 to 108 poise
(ie the consistency of molasses). In this range, silica glass can be worked and shaped. However, this
range is too high and too narrow (and therefore too costly) for most practical applications. Fused silica,
called silica glass or vitreous silica, is therefore only used for a few specialised products such as windows
in high temperature environments, some laboratory ware, etc.

The working range can be reduced by adding soda (Na2O) and other metal oxides, as seen in Figure 5.4.

Figure 5.4

As the amount of metal oxide is increased, the viscosity drops considerably. This can be explained in
terms of simple geometry. As the metal oxide enters the silica melt, it dissociates as

MO = M++ + O-- (5.1)

The oxygen ion then enters the silica network, each oxygen ion causing one break in the network of
tetrahedral as shown in Figure 5.5.

 Si – O – Si  + M++ + O-- =  Si – O- + M++ + -O – Si 

Figure 5.5

37
Each break in the network carries a negative charge so that, to maintain electrical neutrality, the metal
cation is located near the break in the holes in the network. As progressively larger quantities of metal
oxide are added, the network is increasingly broken down, as shown in Figure 5.6, and the viscosity falls
accordingly. Eventually, discrete tetrahedral exist when two metal oxide molecules are added for every
SiO2 molecule (ie at the formula M2SiO4).

Figure 5.6

Further information on the structure of silicate melts can be obtained from viscosity measurements. The
viscosity of ionic melts is largely an anion effect because the cations are generally too small to affect the
flow properties. Viscosity measurements therefore give an idea of the size and shape of the silicate
groups present. The viscosity of silicatemelts decreases as the concentration of metal oxides increases.
For each composition, the viscosity can be measured over a range of temperatures allowing the activation
energy for viscous flow to be determined from the Arhenius equation.

η = Aη exp -Eη/RT (5.2)

where η = viscosity, Aη is a constant and Eη is the activation energy for viscous flow. The activation
energy/composition curves for various metal oxide/silica binary systems can then be derived, as in Figure
5.7

Figure 5.7

Figure 5.7 shows that the activation energy decreases very rapidly with metal oxide additions up to ~10 –
15% and then decreases more gradually with further additions. This behaviour is interpreted as follows.
Up to ~10 – 15% metal oxide, the picture derived earlier is held to be correct. Then, at ~20% MO, the
silica network is thought to be completely broken down into discrete groups of tetrahedral. With
increasing concentration, the size of these groups decreases giving only a gradual fall in the magnitude of
the activation energy.

38
Compositions and Properties of Technical Glasses

Silica is not the only network-forming oxide, other network formers include B2O3 and P2O5. Metal
oxides, which do break up the network are called network modifiers. In contrast, most trivalent and
heavier tetravalent oxides (Al2O3, TiO2, ZrO2) can act in a dual capacity and are called intermediate
oxides. These may act either as modifiers or as network formers depending on the composition. Glasses
frequently contain other minor but important constituents. – Colouring agents, which are usually
transition metal oxides. – Opalizing agents such as CaF2, SnO2, TiO2 which make the glass opaque. –
Fining agents which aid removal of gas bubbles from the melt (ie As2O3, Sb2O3 + NaNO3, NaCl, etc). –
Dissolved gases as accidental contaminants which include water vapour, CO2, SO2, etc.

Among the more common glasses are:

High Silica Glasses: This class includes pure fused silica (above 99.5% SiO2). Although pure silica is
difficult to melt, the glass has certain valuable properties, eg a very low expansion coefficient 5 x 10-7/K.
These glasses can be used at very high temperatures and it has excellent optical properties.

Soda-Lime-Silica Glasses: This class account for 90% of glass tonnage. They are low in cost, being
made from sand, dolomite (CaCo3-MgCO3) feldspars ( (NaKCa) 0.Al2O3.6SiO2) and soda ash (Na2CO3).
They are relatively easily melted and have expansion coefficient of 80-900 x 107/K. Although the
chemical durability of soda-lime glasses are suitable for most applications, it is not sufficiently acid
resistant for laboratory ware. The composition of window glass, for example, is about 72% SiO2, 1%
Al2O3, 9% CaO, 4% MgO, 14% Na2O.

Borosilicate Glass: The composition of a typical borosilicate glass is 80.5% SiO2, 2.2% Al2O3, 3.8%
Na2O, 0.5% K2O and 12.9% B2O3. This type of glass, eg Pyrex, combines some of the best properties of
soda-lime glass and high silica glass – moderate cost and meltability, low expansion coefficient (~30 x 10-
7
/K) and good chemical durability.

Lead Glasses: These have a wide variety of compositions for special applications. Lead acts as a flux
and imparts four important properties to the glass – high refractive index, high electrical resistivity, high
density and low softening temperature. These glasses are used for optical and electrical applications,
radiation shielding windows and fine art glassware. A typical composition of a high refractive index glass
is 54% SiO2, 1% Na2O, 8% K2O, 37% PbO.

Aluminosilicate Glasses: These are similar to high silica glasses but slightly easier to melt. They have
high softening temperatures, high electrical resistivities and fairly good chemical durabilities, eg for low
expansion 57% SiO2, 20.5% Al2O3, 5.5% CaO, 12% MgO, 1.0% Na2O, 4% B2O3.

Mechanical Properties of Glass

A characteristic of most glasses is their brittleness. Alkali silicate glasses are Hookean solids, exhibiting
a linear stress/strain curve to fracture. The tensile modulus of glass is ~70GN/m2 (~107 psi), a value
which isn’t much affected by compositional changes. This low modulus (~⅓ that for steel) limits the use
of glass as a stress-bearing material, although it is to some extent offset by a lower density which partly
compensates on a specific modulus basis. Early calculations of the tensile strength of glass suggested a
theoretical strength of ~7-21 Gn/m2 (1-3 x 106 psi) whereas, in practice, ordinary window glass has a
strength of only 42MN/m2 with a scatter in individual values of ± 30%. Similarly, assuming Hookes’
Law to be obeyed to fracture, strains of ~20% are theoretically possible compared with observed strains at
fracture of usually ~0.1%.

39
a) Tensile Fracture of Glass

It is generally agreed that the low strength of glass is due to the presence of microcracks on the glass
surfaces which cause local stress concentrations sufficient to cause fracture. This view is supported by the
fact that glass is strong in compression but not in tension, ie fracture occurs when a stress is applied which
is great enough to propagate the largest flaw. Also, the importance of surface flaws is indicated by the
observation that larger samples are generally weaker than smaller ones. These flaws are important in
glass as

i. There is no macroscopic plastic flow to relax stress concentrations at the crack tip
ii. The bonding in glasses has a large ionic component which is characteristically brittle at low
temperatures
iii. There are no grain boundaries, etc in glass to deflect or stop the crack.

As the crack grows under an applied tensile stress, the rate of release of elastic energy increases. The
fracture surface becomes progressively rougher until, eventually the crack branches and often isolates a
small wedge of material from the main fragments. This roughening and branching produces the
characteristic markings on a tensile fracture surface, called the mirror, mist and hackle zones.

Figure 5.8

b) Improving the Strength of Glass

The strength of glass can be improved by treatments which affect the surface and, hence, the surface flaws
eg heat-treatment at high temperatures increases the strength by rounding off the tips of cracks and so
reducing the stress concentrations. With these ‘fire-polished’ surfaces, tested without mechanically
damaging the surface, strengths of 100-1000 MN/m2 are obtained. Although these values are far higher
than ‘off-the-shelf’ samples, they will show a large variability in strength, a size effect and are subject to
static fatigue. These facts suggest failure still originates from surface flaws. It was thought that this
variability indicated that inherent flaws existed in the glass. However, recent work has established that
fracture starts at surface inclusions which become bonded to the surface at high temperatures. These can
act as stress raisers by having elastic moduli different to the glass or by causing cracking on cooling by
having different thermal contractions.

Specimens free from abrasion and inclusions can now be produced by drawing fibres from a pool of glass
at high temperatures or by carefully dissolving the surface in hydrofluoric acid. These ‘flaw-free’
materials have a high and reproducible strength which does depend on composition eg with
Na2O/CaO/SiO2 glass, the breaking strength is 3.5 GN/m2 whilst silica glass is 6 Gn/m2. Yet, even these
materials are subject to static fatigue suggesting fracture still originates at the surface. If these ‘flaw-free’
samples are heated to ~200°C, the strength decreases and the scatter reappears. This is a result of fresh
surface inclusions forming from airborne dust. Similarly even gentle handling of the surfaces markedly
reduces the strengths.

40
Two techniques are generally used commercially to improve glass strength.

i. Remove surface flaws and protect the surface. In the glass-fibre industry, fibres are drawn directly
from molten glass and a plastic coating given to protect the surfaces from damage.

ii. Accept the flaws and make then less effective weakeners by introducing a surface compressive
stress. This method is widely used in the safety glass industry. These surfaces compressive stresses
may be introduced either by thermal toughening or by chemical ion exchange.

Thermal toughening introduces surface compressive stresses by rapid cooling of the outside surface of the
glass. In contrast, chemical strengthening is achieved by replacing alkali ions at the glass surface with
ions of larger radius. This ion exchange is performed at temperatures just below the glass transition
temperature.

Figure 5.9

Conventional compression layers formed by chill tempering can only increase the tensile strength to ~100
MN/m2 whereas strengths of 700 MN/m2 can be obtained by changing the chemical compositions of the
surface. Chemical strengthening has the further advantage that it can be applied to complex shapes.

41
 Section 6 - Refractories
(Lectures 11 & 12)

There is no general definition of a refractory. Essentially it is a material of “high melting point”, but this
is a relative term and melting point is not the only criterion of usefulness.

Most refractories are ceramic materials made from high melting-point oxides, particularly SiO2, Al2O3
and MgO. Carbon is an important refractory, however, and carbides, borides and nitrides are also being
developed for high-temperature work. Metals like Mo and W are refractory metals and find uses in
research apparatus. Even these may be melted in heavily water-cooled copper containers which, if not
classified as refractories, certainly replace them.

Finally there are materials like asbestos which do not have a high melting point, but which are used as
medium or low-temperature insulation and should not be excluded from consideration.

Table 6.1
Mpt °C
Refractory Oxides - SiO2 Silica 1610 Quartz
1713 Cristobalite
1702 Tridymite
Al2O3 Alumina 2140 Corundum
MgO Magnesia 2900 Periclase
Be Berylia 2530
ZrO2 Zirconia 2715

Refractory Metals - Tungsten 1557

Molybdenum 2610

Carbon - Graphite Sublimes at 3652

Properties and Testing

Failing a simple definition for a refractory material, a discussion of refractory qualities is essential. The
most important requirements of the user are:

1. Rigidity and maintenance of size, shape and strength at the operating temperature, which will
presumably be “high”.

2. An ability to withstand thermal shock such as is met in heating up and cooling down of furnaces,
or in fluctuations which occur during charging or during normal operation

3. Resistance to chemical attack by whatever gas, slag or metal is likely to be encountered.

As there is no supreme material capable of standing up to every possible condition, the choice must be
made to meet the requirements of the job to be done. Compromise is usually necessary. Other properties
such as cold crushing strength, and thermal conductivity are usually of secondary importance but
occasionally, as in insulating bricks, one of these may become of prime importance.

42
Reversible thermal expansion is determined versus fused silica (whose coefficient of expansion is very
small) in any standard type of dilatometer. Here one is determining the coefficient of thermal expansion
along with any volume changes due to polymorphic transformations. The latter may give evidence of
imperfect firing (eg in silica). The expansion is a necessary design datum as it must be accommodated as
the furnace heats and cools during operation. It is also related to appalling resistance, and a low value if
desirable on all accounts, but every type of brick has its own characteristic behaviour in this respect, and
in some cases, eg magnesite, an otherwise admirable brick may be difficult to use because the coefficient
of expansion is inherently high (see Figure 6.1).

Figure 6.1 - Typical expansion curves for various refractories

Refractoriness. It is not generally practicable to measure the melting point of a refractory. Usually
melting extends over a range of temperature – perhaps over several hundreds of centigrade degrees. The
determination of solidus and liquidus temperatures would not be easy and might not be very useful. A
material may be partly liquid yet appear solid. At a higher temperature where the proportion of liquid
exceeds a critical value which may depend on its viscosity, the material will either collapse and appear
liquid or pasty, or will exude liquid and collapse slowly later. The apparent temperature of collapse may
well depend on the rate of heating, particularly if the liquid is very viscous.

The test for refractoriness is to compare the sagging of a “cone” of the material (either cut from the solid
or made up from powder and a bond of dextrin) with that of standard Seger cones, when they are heated
together, as a standard rate in an oxidizing atmosphere, until the test cone bends over. The number of the
best matching Seger cone is quoted as the refractoriness of the brick; or its nominal melting point (which
can be checked by pyrometer) is referred to as the Pyrometric Cone Equivalent (P.C.E.) Temperature.
This is usually much higher than the working temperature of the brick. Refractoriness is a guide to
quality rather than a measure of usefulness.

43
Hot Strength

Whilst refractoriness tests are useful with alumino-silicate materials, the PCE assessment is of more
limited use with higher strength materials. A strength measurement may then be obtained by use of
refractoriness-under-load methods (RUL) carried out at elevated temperatures. A sample (~5cm diameter
and ~7.5cm high) is loaded at 175 MN/m2 (25 psi) at room temperature and the temperature of collapse
noted when the sample is heated at a fixed rate.

The specimen is heated in a carbon granule furnace at a standard rate (10°/min) and a record of its height
made, preferably by automatic plotting, until the test piece collapses or sinks at 90 per cent of its original
length. The Refractoriness under Load may be quoted as the range of three temperatures:

Initial softening (at which the curve is horizontal) (see Figure 6.2)
Rapid collapse (contraction rate 0.05 in/min)
Total collapse (beyond 10 per cent shortening)

Sometimes the temperature at which shortening is 5 per cent is noted.

Figure 6.2 – Typical R.U.L. curves for various bricks. The behaviours of silica and chrome-magnesite under sustained
load at 1600°°C are compared in the inset

A more satisfactory alternative is to stop heating at a suitable temperature, related to the working
temperature, and to measure the creep rate during the next few hours. The object is of course to determine
the maximum temperature at which the brick could be used under compressive load. It will be
appreciated that usually not all of a brick would be a the hot-face temperature. The “cold” end would
normally remain rigid even after the hot end had exceeded initial softening. The extension of the period
of these severe conditions from hours to weeks is not readily predicted and leads to difficulty in
correlating test results with practical performance.

However, the thermal conductivity of refractories is normally low so that RUL tests develop a temperature
gradient within the sample. The RUL method is therefore being replaced steadily by the modulus-of-
rupture method (MOR). This test measures the breaking strength of a bar (150 x 25 x 25 mm) in 3 point
bend. This test may be carried out at room temperature but a more useful assessment of refractory
performance is derived from tests at high temperature. The MOR value is calculated as:

MOR (MN/m2) = 3wl/2bd2

44
where w is the load at failure, l the separation of the lower bearing edges of the 3-point bend (normally
125 mm) and b and d are the width and depth of the samples. This allows strength to be obtained for
samples maintained at an uniform temperature. However, even this test is now being increasingly
replaced by creep testing methods, usually in compression, bend or torsion. A creep experiment simply
involves holding the specimen under load at a fixed temperature and measuring how the strain varies with
time as a function of stress and temperature. The disadvantage of MOR results is that information derived
from tests lasting a relatively short time cannot be realistically employed to predict the long-term
behaviour of refractories under stress at elevated temperatures. Moreover, since strains can be measured
with greater precision than stress values, creep methods enable small differences in strength to be
determined with increased accuracy.

Resistance to Spalling

Spalling is usually defined as the fracture of a refractory brick or block and can originate in several ways.
Thermal spalling is caused by the fact that refractories rarely operate at constant temperature and may fail
as a result of rapid changes in operating temperature. Many different tests have been employed to
measure the effects of temperature changes, eg testpieces may be heated from one end and the rate of
heating required to cause cracking can be determined. It is, however, noteworthy that the response to
heating and cooling may be different and may also be dependent on the temperature ranges considered.
Thermal spalling is caused by the low thermal conductivity of refractories. Rapid heating or cooling then
changes only the surface temperature and the associated thermal expansion or contraction can set up large
stress concentrations between the different regions of the brick. These stresses may be sufficient to cause
failure. The resistance to thermal spalling may be estimated from the expression:

Spalling resistance = Strength x Conductivity

Expansion coefficient x Heating rate

Mechanical spalling can arise through compression of a refractory structure due to the expansion of the
whole for a rise in temperature causing shear fracture. Structural spalling is due to variations in the
expansion coefficient of the surface of a refractory and the bulk of the material because of slag penetration
or structural changes in service which enable stresses to be set up which are sufficient to shear off the
surface layer.

Production of Alumino-silicate Refractories

Alumino-silicate refractories with up to ~42% Al2O3 are made from selected clays of low alkali and iron
oxide. Higher quality clays have ~94% Al2O3 + SiO2 and only ~0.5% Na2O + K2O. Calcined pure
kaolinite (Al2 (Si2O5) (OH)4) contains ~43% Al2O3. The proportion of Al2O3 differs in different deposits
because of variations in the clay type present and in the types of associated minerals. High-alumina
refractories are made from rocks such as kyanite and sillimanite (Al2O3 SiO2) with additions, where
necessary, of alumina prepared from bauxite (Al2O2 2 H2O). Products containing over 85% Al2O3 are
produced using bauxite only.

The Aluminor-silica System

The Al2O3-SiO2 equilibrium diagram is shown in Figure 6.3, with the composition of the various
refractories marked on it. Up to 1545°C firebrick consists of mullite and silica. Mullite is present as
needle shaped crystals which form an interlocking matrix which can retain its strength and rigidity to high
temperatures towards 1810°C at which it melts incongruently. The strength depends, of course, on the
proportion of mullite present, ie on the Al2O3 content. Silica is present as tridymite or cristobalite (or
both), but the effect of fluxes (when present) is to transform some of it to glassy silicates –
thermodynamically very viscous liquids.

45
 Figure 6.4 – The Al2O3 – SiO2 equilibrium diagram (after Bowen and Greig) indicating the compositions of various
kinds of brick based on the system, and their approximate maximum working temperatures which depend also on other
aspects of composition

Above 1545°C liquid forms – again a very viscous liquid but viscosity falls off rapidly with temperature –
and the structure becomes a matrix of mullite needles with liquid in the interstices. Considering siliceous
firebrick (25% Al2O3) we find that the proportion of liquid is about 70% and it is unlikely that the
remaining 30% in the form of mullite will be able to maintain the shape of the brick for very long. The
refractoriness of the brick has been exceeded at 1545°C and indeed probably at a lower temperature in the
presence of flames. A brick containing 45% Al2O3 on the other hand contains only 40% liquid at 1545°C,
the remaining 60% being mullite, which remains solid. Temperature increases beyond 1545°C cause the
amount of liquid to rise in all cases, but this rise is slow. At 1700°C the high-alumina brick has still only
45% of liquid (although the other has 80% and has undoubtedly slumped). It is clear that refractoriness in
advance of 1700°C is quite possible with firebrick of suitable quality but R.U.L. values better than
1545°C will be difficult to attain.

Considering even higher, Al2O3 content, at 72%, at the mullite composition, the proportion of liquid at
1545°C has fallen practically to zero and remains negligible up to 1810°C. The advantages of very high
alumino are obvious.

The compositions of various types of commercial firebrick are given in Table 6.2 and their respective
RUL and spalling resistance data is given in Table 6.3.

46
 Table 6.2 – Composition of Silica and Alumino-silicate Bricks

Table 6.3 – Typical Properties of Silica and Alumino-silicate Bricks

47
 Section 7 – Glass Ceramics
(Lectures 13 & 14)

Defective glass can result from unplanned devitrification but if crystallisation is controlled new types of
material, called glass-ceramics, can be produced. Most ‘oven-to-table’ ware and ceramic hobs found in
the modern kitchen are manufactured from a particular range of glass-ceramics which have low
coefficients of thermal expansion. Glass-ceramics are polycrystalline materials formed by the controlled
devitrification of glass of carefully chosen compositions which give a larger number of nuclei for
crystallisation. After shaping the component whilst the material is still in the glassy state, a two-stage
heat treatment is followed to transform the glass to a glass-ceramic, as seen in Figure 7.1.

Figure 7.1 – Micrograph of ‘Silceram’ glass-ceramic showing a larger volume fraction of fern-like crystals with a small
amount of residual glass

Theories of glass structure assume that the structure and compositions do not change from point to point
in the glass. However, electron microscope studies have shown that many glasses have a definite
structure. Essentially, certain components in the glass aggregate into droplets in a matrix of different
compositions. The size, shape and number/unit volume of droplets can be varied by controlling the
composition of the glass, and the time of holding at temperatures near the liquidus. The structure can vary
from individual droplets to interpenetrating skeletons. One aspect of this ‘phase-separation’ has been
receiving increasing attention is the ‘controlled-crystallisation’ of glasses which has led to the
development of glass-ceramics. Two factors are important in the crystallisation of glass: nucleation and
growth of the new crystalline phase. In the formation of nuclei, discrete regions of the new phase must be
formed which are above a critical size. The growth of the new phase crystals then depends on the free
energy of crystallisation and the rate of transport of material to the growing crystal. The nucleation rate
and the rate of crystal growth depend on the temperature. With non-viscous materials, both rates increase
rapidly as the temperature decreased. For example, for undercooled copper, as illustrated in Figure 7.2.

48
 Figure 7.2 – Temperature dependent nucleation and growth rates for undercooled copper

With the type of behaviour, crystallisation can not be avoided. However, nucleation and growth depend
on the transfer of material to the developing nucleus. With a viscous liquid, the viscosity increases
rapidly as the temperature falls which drastically impedes molecular transport and rearrangement. Thus,
with a glass, the nucleation and crystal growth rates pass through a maximum and then decreases to zero,
as illustrated in Figure 7.3.

Figure 7.3 – Temperature dependent nucleation and growth rates in glass

In a typical glass, spontaneous crystallisation occurs if the temperature is held in the region where the
nucleation and growth curves overlap. This crystallisation can be aided by incorporating a material in the
glass which is readily precipitated as a new phase, eg TiO2, as a nucleating agent. This enables nuclei to
be formed at lower temperatures.

49
 Figure 7.4 – Temperature dependent nucleation and growth in glass ceramic

The process of preparing glass ceramics is to melt, cool and form the glass in the conventional manner.
Then, heat in the temperature range where nuclei form (ie where nucleus formation and growth rate
curves overlap), the holding time controlling the number of nuclei. The temperature is then raised to
allow the bulk crystalline phase to grow on the nuclei. With sufficient nuclei, a fine grained ceramic can
be formed. In this way, ceramics can be formed by conventional processes, free from pores, with a
uniform crystal size, randomly oriented and interlocked in a glass bond. Although many systems can be
treated to give glass ceramics, two have been particularly developed (both TiO2 nucleated). These are:

i. Lithium alumino-silicates namely β-spodumene (Li2O.Al2O3.4SiO2) and β-eucryptite

(Li2O.Al2O3.2SiO2) which have low thermal expansions with excellent thermal shock resistance,
and

ii. Magnesium alumino-silicates which have corderite (2.MgO.2Al2O3.5SiO2) as precipitates which

have excellent microwave transmission and good thermal shock resistance.

Glass ceramics can vary from fine (<10nm) interlocking crystals in transparent glass to several micron
crystals in materials with very low residual glass contents. By controlling the amount and type of
precipitates, many combinations of mechanical, thermal, electrical, chemical and physical properties can
be obtained. Glass-ceramics therefore offer an exciting new range of ceramic materials.

50
 Section 8 – Zirconia and Transformation Toughening
(Lectures 15 & 16)

Zirconia (ZrO2) is a hard refractory material which has been used for many years as a high temperature
furnace and crucible refractory at temperatures up to 2500°C. It is a polymorphic compound existing in
three different crystalline phases. The cubic form, stable above 2380°C changes below that temperature
to a tetragonal form and on further cooling to below 1170°C a monoclinic form becomes stable:

2380°C 1170°C
ZrO2
Tetragonal Monoclinic ZrO2


ZrO2
However, the final transformation (below 1170°C) is martensitic in nature and in fact shares some of the
features of the austenite → martensite transformation in steels. In particular the polymorphic change is
accompanied by a sudden volume change of some 3%. Since zirconia is a brittle substance this sudden
expansion causes severe cracking, making it useless to produce components from pure ZrO2. However,
additions of other metallic oxides will suppress the martensitic transformation to below ambient
temperatures and so stabilise the tetragonal or even cubic phase. Varying amounts of lime (CaO) and
magnesia (MgO) were added for this purpose when zirconia was used as a furnace refractory but yttria
(yttrium oxide. Y2O3) is now used as a stabiliser in modern zirconia ceramics.

The yttria-rich section of the ZrO2 phase diagram of Figure 8.1 shows that yttria additions in excess of
about 18wt% completely stabilise the c-phase to room temperature. A zirconia with the cubic structure
stabilised by a large yttria addition is called a fully stabilised zirconia (FSZ). The main uses of FSZ’s are
in electrical applications, such as oxygen sensors, fuel cells, etc.

Figure 8.1 – The zirconia-rich part of the ZrO2-Y2O3 phase diagram

An intermediate amount of yttria, eg 5wt%, gives a partially stabilised zirconia (PSZ), which has a
microstructure containing the three main crystal forms c, t and m. A PSZ is normally fired at an elevated
temperature, which depends on the composition but is of the order of 1,750°C and in the cubic phase
field; it is then heat treated at a lower temperature (1,300 – 1,400°C) in the cubic plus tetragonal phase
field to produce a fine dispersion of particles of the t-phase. On cooling to room temperature some, but
not all, of the t-particles transform martensitically and become monoclinic. Those particles that remain in
the tetragonal state do so because the elastic constraint of the cubic matrix opposes the volume change
associated with the martensitic transformation. The t-particles are therefore in a metastable state. It has
been found that the inhibition of the martensitic transformation to the monoclinic depends on the size of

51
the particles; the smaller the particle the more likely it will remain in the metastable tetragonal state.
PSZ’s have a low thermal conductivity, which is effectively invariant with temperature, and a high
coefficient of thermal expansion. These characteristics make them suitable for thermal barrier coatings on
metal substrates. The low conductivity results in a large temperature difference across the PSZ coating,
thus protecting the metal substrate in high temperature environments, and the high thermal expansion
coefficient reduces the thermal expansion mismatch between the ceramic coating and the substrate.

Finally we come to low yttria additional as typified by ZrO2-2wt% Y2O3. These zirconias have a fine-
grained microstructure (less than 1µm) and are fully tetragonal; the metastable tetragonal state being a
consequence of the fine grain size. These zirconias are known as tetragonal zirconia polycrystals, TZP’s.

Both TZP’s and PSZ’s have good mechanical properties, especially the TZP’s which have Kic values in
the range 10-15 MN/m3/2. The good toughness of these two types of zirconia is associated with the
martensitic t→m transformation which increases the toughness by two distinct mechanisms. Firstly, if a
restricted number of particles undergo the transformation during cooling from the fabrication temperature,
a fine distribution of microcracks is produced. The microcracks increase the toughness by interacting
with a propagating crack, causing deflection and blunting of the crack (Figure 8.2a), However, this
toughening mechanism, known as microcrack toughening, occurs at the expense of strength, which is
reduced by the microcracks acting as flaws. The stress field at a crack tip can induce a metastable t-
particle to transform to the monoclinic. This is the basis of the second toughening mechanism,
transformation toughening, where the propagation of a crack is hindered by both the transforming
particles at the crack tip and by the compressive back-stress due to the transformed particles in the crack
wake (Figure 8.2b). Transformation toughening, unlike microcrack toughening, does not have a
detrimental effect on strength.

PSZ’s have a low thermal conductivity at all temperatures but a high coefficient of thermal expansion.
Both of these properties make PSZ’s suitable as protective coatings on metals working at high
temperatures such as superalloy rotor blades in jet turbines. The high coefficient of thermal expansion
reduces the thermal expansion mismatch between PSZ coating and the metal substrate. A typical PSZ is
‘Cerafine’ containing 5% wt Y2O3.

The ‘environmentally friendly’ properties of zirconia when used inside the human body have led to its use
in implantology. This biocompatibility, combined with a bending strength and toughness which are
double that of the currently used alumina, have led to its adoption for hip joint replacements in both North
America and France. An yttria-stabilised zirconia with the trade name Prozyr is used.

Figure 8.2 – Toughening mechanisms associated with the tetragonal-to-monoclinic transformation in zirconias:
(a) microcrack toughening; (b) transformation toughening

52
Electrical Conductivity in Zirconia

In the late nineteenth century that Nernst used yttria-zirconia to examine the electrolytic behaviour of
oxides. In the early 1970s, the ZrO2 – Y2O3 electrolyte was demonstrated in a sensor device used to
measure the oxygen level in a gas. The most commonly used material is the cubic phase with additions of
yttria in the range 10-16wt%. Investigators have also shown that the tetragonal phase also demonstrates
high oxygen ion conductivity.

The cubic phase has the fluorite structure in the stabilised systems based on (CaO, MgO, Y2O3)-ZrO2
within specific compositional ranges. The void space generated by the presence of an oxygen ion vacancy
is considered to be larger than for the corresponding Zr4+ vacancy and since the solute cations are of lower
vacancy than zirconia, then there are many more of them.

The lattice substitution of the lower-valent dopants (Y3+ and Ca2+) on Zr4+ sites to form the defects YZr' or
Ca"Zr , causes doubly charged oxygen vacancies (Vδ) to be formed to maintain charge neutrality. This
results in the generation of >1021 cm-3 mobile ionic species in the ZrO2 lattice. This combined with the
relatively open oxygen lattice in cubic ZrO2, results in exceptionally high oxygen ion conductivities. For
acceptable oxygen sensor manufacture, the composition and fabrication process must be chosen to
optimise a number of properties. Oxygen ion conductivity, density (to ensure gas imperviousness),
microstructure, grain boundary phases, impurities, mechanical strength, and thermal shock resistance must
be controlled to obtain the optimum performance. Y2O3 is always chosen as the stabilizer due to the
notably higher oxygen ion conductivity (approximately an order of magnitude greater) which it induces in
ZrO2 relative to the CaO stabilised material. This becomes very important particularly for the lower
temperature operation (~350°C) of the device. Small percentages of other additives, such as silicates, are
often added to provide a useful degree of densification at temperatures below those conventionally
required for pure materials. Addition of a maximum of around 0.5mol% silicate flux results in the best
microstructure, density, and ionic conductivity.

Figure 8.3 – Ionic conductivity of some solid electrolytes with concentrated H2SO4 for comparison

53
 Section 9 – Advanced Ceramic Applications
Zirconia Electro-ceramics
(Lectures 17 & 18)

9.1 Early Applications

One of the earliest applications of oxygen-ion-conducting solid electrolytes was a radiation source.
Nernst (1900) was working on an alternative and a more efficient means of producing visible light than
the carbon filament lamp. He experimented with mixed oxides and discovered that some of them showed
very good conductivities at high temperatures. He tested mixed oxides such as 15% Y2O3-85% ZrO2 and
observed that O2 was evolved at the anode when a DC voltage was applied. He suggested that a nearly
pure oxygen ion conduction was taking place. Today, Nernst ‘glow bars’ are still in use as infrared
sources in some spectrophotometers.

Interest in zirconia – as a ceramic – was great, due to its chemical inertness and high melting point. Its
use in laboratory equipment was inhibited due to the change in the crystal structure, from monoclinic to
tetragonal, around 1050°C, which is accompanied by a change in volume resulting in fracture. However,
intensive research over many years led to the discovery that the crystal structure could be stabilised in a
cubic from by the proper mix of oxides with much more acceptable thermal expansion characteristics
(Figure 9.1). The conductivity is due mainly to oxygen ions in a wide range of metal oxide solid solutions
(mixtures of MgO, ZrO2, CaO, Y2O3, CeO2 etc). Haber and co-workers experimented with oxygen-ion-
conducting solid electrolytes and utilised them as fuel cells and concentration cells and demonstrated also
that they obey Faraday’s law.

Figure 9.1 – Thermal expansion curves for (a) pure zirconia, (b) calcia fully stabilised and
(c) calcia partially stabilised zirconia

54
9.2 Basic Structure of an Oxygen-Ion-Conducting Solid Electrolyte Cell

The basic structure of an oxygen-ion-conducting solid electrolyte cell is shown in Figure 9.2. The solid
electrolyte is a disc at the centre of a ceramic tube which is surrounded by a heater that brings the cell to it
operating temperature. The disc faces a chamber on either side, these being hermetically sealed from each
other. One chamber contains a reference oxygen pressure, which could be a gas such as air or a sealed
metal/metal oxide in equilibrium at the operating temperature. The other chamber is the sample chamber.
The disc is covered by a porous and electronically conducting electrode on either side (e.g. Pt) and
connections are brought out. A temperature sensor is usually included so as to measure and control the
operating temperature.

Figure 9.2 – Basic structure of an oxygen-ion-conducting solid electro-lyte cell

Three main modes of operating the cell are possible.

1. The Concentration Cell Mode

If a sample gas containing a partial pressure of oxygen, p’O2, is admitted to the sample chamber, the
reference chamber contains an oxygen reference having a partial pressure of p”O2, and the operating
temperature of the cell is TK, then the EMF across the electrodes will be given by the Nernst equation,

E = (RT / 4 F ) ln( p" O2 / p' O 2 ) 9.1

where R is the universal gas constant and F is the Faraday constant. The term p”O2 is greater than p’O2.

2. The Fuel Cell Mode

A fuel cell is an electrochemical device in which the chemical energy of a fuel is converted directly and
efficiently into low-voltage direct-current electrical energy. As the conversion is supposed to be carried
out isothermally, the Carnot limitation on efficiency does not apply. The fuel admitted to the anode
chamber is oxidised, releasing electrons to the load circuit, whilst the oxidiser at the cathode chamber is
reduced by the electrons that have come from the anode, via the load. If we consider CO as a fuel, then
the reactions would be:
CO + O2- → CO2 + 2e at anode

O2 + 4e → 2O2- at cathode

and an output EMF of 1.33 V would be expected.

55
3. The Oxygen Ion Pump Cell Mode

In this mode a DC current (I mA) is allowed to flow between the cell electrodes. A gas mixture
containing oxygen such as air could be in the reference chamber and if an inert gas such as nitrogen flows
through the sample chamber at the rate of L litres per minute, the flow of the current results in pumping
oxygen molecules from one side to the other (four electrons per O2). The cell obeys Faraday’s laws and
hence the O2 pumped into the inert gas will be given by

PPM O2 = (3500 x I)/L 9.2

9.3 Miscellaneous Applications

Oxygen-ion-conducting solid electrolytes have been proposed for: the determination of high fixed-point
temperatures on the thermodynamic scale (Lindsay and Ruka 1968); high-temperature electrodes for
magnetohydrodynamic (MHD) generators (Casselton and Watson 1968); controllable oxygen getters
(Fouletier and Kleitz 1975); and the measurement of pH (e.g. Hettiarachchi and MacDonald 1984).

The measurement of dissolved oxygen in liquids, including molten metals (e.g. Pluschkell and Engell
(1965) and Fischer and Ackerman (1965)) is a particularly well-developed application. Partially stabilised
(with MgO) zirconia sensors are now extensively used to monitor deoxidation in copper refining and in
the modern steel industry (Nagata and Goto 1983). At temperatures of 1500°C and above such probes can
only be used once but the probe lifetime is sufficient to obtain valid readings in a few tens of seconds. As
with all other oxygen sensing applications involving zirconia, appropriate developments in the ceramic
technology of the material have been essential in underpinning the development of successful devices for
molten steel application. Over one million such sensors are now used each year in modern steel-making
plants.

9.4 Monitoring O2 In Car Exhaust Gases – Lambda Sensors

Throughout the world there is increasing concern about the minimisation of the emission of noxious gases
(carbon monoxide and oxides of nitrogen) and unburned hydrocarbons from automobile exhausts. A
widely adopted approach to this problem is the incorporation of a three-way catalyst in the exhaust system
so that all the combustion reactions in the exhaust mixture can be brought to equilibrium when the
products will be carbon dioxide, nitrogen and water. The use of such catalysts is most effective when the
air/fuel ratio taken in by the engine is close to the stoichiometric value (around 15:1), as shown in Figure
9.3. This method of reducing undesirable emissions results in poor fuel economy, but is straightforward
to automate in a closed loop control system using a zirconia sensor because such devices offer a huge
change in signal in the vicinity of the stoichiometric (the so-called ‘lambda’) point (Figure 9.4). Figure
9.5 shows a typical sensor, whilst Figure 9.6 shows one of the latest designs which uses thin and thick
film technologies.

Zirconia-based oxygen sensors are now also being utilised in lean-burn engines (with air/fuel ratio greater
than 15:1) where greater fuel economy may be achieved. In order for a lean-burn engine to operate
effectively, timing, spark duration and other engine parameters must be under computer control. The
air/fuel ratio must be known and controlled and this can best be done by closed-loop measurement
utilising an oxygen exhaust sensor.

While the three-way catalyst cannot function in the lean-burn environment, the amount of unburned
hydrocarbons and carbon monoxide will be small (see Figure 9.3). However, in contrast to the situation at
the lambda point, the change in oxygen level across the lean region is small (see Figure 9.4). This poses a
much more severe sensor design problem as small errors in temperature measurement, or sensor non-
ideality, can clearly give rise to large errors in the measured air/fuel ratio. The preferred solution now
emerging seems to be for an oxygen pumping device where the oxygen reduction current at an electrode
exposed to the test gas is proportional, in a known manner, to the partial oxygen pressure in the gas.

56
 Figure 9.3 – Products of combustion in internal combustion engine exhaust gases as a function of the air/fuel ratio

Figure 9.4 – Sensor output as a function of air/fuel ratio at different temperatures. The lambda point occurs at a ratio
of 14.5 to 1

57
 Figure 9.5 – Auto exhaust lambda sensor (Courtesy of General Motors Corporation)

Figure 9.6 – Thick/thin film lambda sensor (CSF Thomson)

9.5 Potentiometric Analysis Of Hot Gases

The molten steel and vehicle exhaust applications of oxygen-ion-conducting solid electrolytes have
developed very rapidly in the last few years in response to market pressures and the availability of high-
quality zirconia ceramics. The other major market, which was first suggested by Haber, is the analysis of
oxygen in hot gases, particularly for combustion control of large boilers to maximise fuel economy and in
the analysis of gas mixtures, notably in metal-hardening carburising atmospheres. Early designs of
oxygen sensors were proposed by Mobius and by Weissbart and Ruka. There are a great variety of
designs in the market place today, the following are the most common ones.

58
 Section 10 – Advanced Ceramic Applications
High Temperature Engineering Ceramics
(Lecture 20)

Structural and Refractory Applications of Engineering Ceramics

Structural components derived from engineering ceramics are used as monoliths, coatings and composites
in conjunction with or as replacements for metals when applications rely on mechanical behaviour of the
ceramics and their refractory properties, that is chemical resistance to the working environment. Physical
properties of ceramics and metals are compared in table 10.1 some of their magnitudes are shown in
tables 10.2 and 10.3. Nickel superalloys are currently the main high-temperature materials for
components such as combustors in gas turbine engines. They have melting points around 1573 K and a
maximum working temperature near 1300 K. Cast iron parts in reciprocating (ie petrol and diesel)
engines have properties shown in table 10.3. Compared with metals, ceramics are generally more
resistant to oxidation, corrosion, creep and wear in addition to being better thermal insulators. They have
higher melting points (table 10.4) and greater strengths than superalloys at elevated temperature so that a
major potential application, particularly for silicon nitride, is in gas turbine and reciprocating engines
where operating temperatures higher than attainable with metals can result in greater efficiencies. This
enhanced strength is shown in figure 10.1 for hot-pressed silicon nitride (HPSN), hot-pressed silicon
carbide (HPSC), hot isostatically pressed silicon nitride (HIPSN), sintered silicon nitride (SSN), sintered
silicon carbide (SSC), reaction-bonded silicon nitride (RBSN) and reaction-bonded silicon carbide
(RBSC). Although ceramics offer improvements in engine efficiency, incorporation of silicon nitride
over the past three decades has been slow, mainly because of the difficulty in reproducible fabrication of
dense components to close dimensional tolerances.

Table 10.1 – Relative properties of ceramics and metals

Table 10.2 – Physical properties for alloys, oxide and non-oxide ceramics

59
Table 10.3 – Mechanical properties of cast irons, oxide and non-oxide ceramics

Table 10.4 – Melting point and maximum working temperatures of ceramics

Figure 10.1 – Variation of strength with temperature for non-oxide ceramics

60
Silicon nitride occurs in two phases, the α and the β forms. The β form, whose structure is shown in
figure 10.2, consists of SiN4 tetrahedra jointed together by sharing corners in a three-dimensional
network. It is possible to replace silicon by aluminium and maintain charge neutrality in the crystal lattice
by substitution of nitrogen with oxygen. The resulting solid solutions in the Si-Al-O-N system are known
as β′-sialons (K.H. Jack, 1986) whose structures are identical with β-Si3N4 over the composition range
Si6-bAlbObN8-b (0<b<4). They exhibit mechanical behaviour similar to β-Si3N4 and have some features of
aluminium oxide. However, in contrast with Al2O3 which consists of six-coordinated Al, β′-sialon
contains Al that is four-coordinated by oxygen and this results in an enhanced Al-O bond strength
compared with the oxide. Unlike Si3N4, β’-sialons can be densified readily by pressureless sintering and
they have been put into commercial production by Lucas Cookson Syalon Limited. Syalon components
shown in Figure 10.3 include automotive parts such as valves, valve guides and seats, tappets, rocker
inserts and precombustion chambers in addition to weld shrouds, location pins, extrusion dies, tube
drawing dies and plugs.

Figure 10.2 – Crystal structure of β -Si3N4 and β ′-(Si,Al)3(O,N)4.

•, metal atom, o, non-metal atom

Figure 10.3 Examples of Syalon Components

61
Aluminium titanate is used as port liners in some automobile engines because its low thermal conductivity
(2 W m-1K-1) reduces heat flow to the cylinder block and hence the amount of cooling required. Glass
ceramics have applications in cooking utensils, tableware, heat exchangers, vacuum tube components and
missile radomes. Partially stabilised zirconia was developed in 1975 at the Australian Commonwealth
Scientific & Industrial Research Organisation (CSIRO) and is nowadays manufactured by Nilcra-PSZ
Limited. This materials is particularly suited for withstanding mechanical and thermal shock because of
its high fracture toughness. Examples are dies for extrusion of copper and aluminium tubes, diesel engine
cam follower faces, valve guides, cylinder liners and piston caps, wear and corrosion-resistant nozzles in
papermaking equipment, wear resistant inserts such as tabletting dies as well as scissors and knives.

Not all ceramic components require high-temperature strength. The high Young’s modulus (550 GPa) of
titanium diboride, TiB2, makes it useful for armour plating (Knoch, 1987) whereas ceramics are suitable
materials in seals because of their chemical resistance (table 1.5). Hence sintered silicon carbide is used
for mechanical seals and sliding bearings whereas boron nitride, which is not wetted by glass and liquid
metals, constitutes break rings in the horizontal continuous casting process for steels. Boron carbide, a
harder ceramic than SiC, is suited to wear-resistant applications such as grit blasting nozzles whereas
Si3N4 is also used as ball bearings. An established industrial use for engineering ceramics is as cutting
tools for steels where high-temperature hardness of sialons and zirconia toughened alumina together with
their low reactivity towards metals are desirable properties; this application has recently been reviewed by
D.H. Jack (1986).

Worldwide markets for ceramic coatings in 1985 have been estimated at around 1100 US$M. About 45%
of this market is in optical coatings, the remainder for thermal, wear and corrosion resistance. For
example the thermal barrier coating in Figure 10.4 contains an outer ZrO2 layer on the combustion
chamber from the RB211 gas turbine engine (Meetham, 1986) and produced an increase in combustor
lifetime because of a reduction in substrate temperature by 50 K.

Figure 10.4 Thermal barrier-coated RB211 combustion chamber

62
 Section 11 - Fibre and Whisker
Reinforced Ceramics
(Lectures 21 & 22)

Generally ceramics have sufficient stiffness and strength and hence the prime objective in producing
ceramic matrix composites is not to enhance these properties. The major problem with ceramics is their
brittleness and sensitivity to flaws. Thus the main aim of reinforcing a ceramic, whether with fibres or
particles, is to improve the fracture toughness. This improvement is illustrated by the stress-strain curves
of Figure 11.1. The failure of a ceramic is catastrophic and the small area under the stress-strain curve is
indicative of the low toughness. The failure of a particulate-reinforced, or short random fibre-reinforced,
ceramic composite is similar to that of the ceramic matrix in that it is catastrophic; note however that the
toughness has increased as demonstrated by the larger area under the stress-strain curve. In contrast the
failure of a continuous fibre composite is not catastrophic and the composite maintains a substantial load-
carrying capacity well after failure has commenced. Therefore not only has the fracture toughness
increased but the failure mode is more desirable. Thus the attraction of both particulate- and fibre-
reinforced ceramic composites is the improved toughness which makes them less sensitive to flaws and
more reliable during assembly and service.

Figure 11.1 Comparison of the stress-strain behaviour of (a) ceramic, (b) particulate, or short random fibre, composite
and (c) a continuous fibre composite

Ceramic matrix composite production is the least established composite production technology. Due to
the high melting point of most crystalline ceramics, techniques involving a melt are unsuitable. Melt
infiltration is even difficult with precursor melts for glasses and glass-ceramics as these are more viscous
and more reactive than molten metals. The routes which at this stage of development look most
promising are:

i) conventional mixing of the components followed by cold pressing and firing or hot pressing,
ii) mixing using a slurry followed by sintering;
iii) mix a ceramic-producing polymer and the toughening phase then pyrolyse the polymer; and
iv) chemical vapour deposition of the matrix, i.e., impregnate a pre-form with a gas which deposits the
ceramic.

63
Increasing Toughness

The various toughening mechanisms which may be operative in a composite are illustrated in Figure 11.2.
We have already discussed transformation toughening and microcrack toughening and will not consider
them any further except to point out that microcracks may be produced by differences in thermal
expansion coefficients of the components as well as by phase transformations.

Figure 11.2(a) shows the crack bowing mechanism. Resistance of the reinforcing phase to fracture
hinders the progress of the crack and causes it to bow into a non-linear crack front. Bowing reduces the
stress intensity, K, on the matrix whilst producing an increase in K on the reinforcing phase. As the extent
of the bowing increases so K on the toughening phase rises until fracture of the toughening phase occurs
and the crack advances. Crack deflection (Fig. 11.2(b) is similar to crack bowing in that the reinforcing
phase perturbs the crack front. Deflection results in a non-planar crack which requires an increase in the
applied stress to maintain a sufficient stress intensity at the crack tip for crack propagation. The
effectiveness of the crack bowing and crack deflection mechanisms depends on the morphology of the
toughening phase; the greatest improvement in toughness is obtained with constituents, such as fibres,
with high aspect ratios.

When a crack causes debonding (Fig. 11.2(c)) extra energy is required due to the creation of new
interfaces. When pullout occurs (Figure 11.2(d)) further energy is also needed to overcome frictional
forces.

If some debonding has taken place, but the fibre is strong and has not fractured, then the fibre will bridge
with faces in the wake of the propagating crack (Fig. 11.2(e)). For the crack to continue to grow the faces
must open but, in simple terms, crack opening is hindered by the bridging fibre which has to deform
elastically. This toughening mechanism, which is known as wake toughening or fibre bridging, is most
effective when strong fibres are used in as high a proportion as possible, but even then the contribution to
toughness is thought to be rather limited.

Figure 11.2 Toughening mechanisms in composites: (a) crack bowing; (b) crack deflection; (c) debonding; (d) pull-out
(e) wake toughening (fibre bridging)

64
Carbon-carbon composites consist of fibrous carbon in a carbonaceous matrix. Although both
constituents are the same element their morphology and structure differ greatly from composite to
composite. The carbon in these composites may vary from highly crystalline graphite through a wide
range of quasicrystalline forms such as turbostratic carbon, to amorphous, glassy carbon. By control of
the proportions, morphology and crystallography of the constituents, together with the porosity content, it
is possible to produce a series of composites ranging from a soft, porous material used for furnace
insulation to a hard, high temperature, structural material.

The composite is produced by impregnating the fibrous reinforcement, which is available in many forms
from unidirectional lay-ups to multi-directional weaves. Impregnation is achieved by two basic methods:
either

i) carbonisation of an organic solid or liquid, or

ii) chemical vapour deposition (CVD) of carbon from a hydrocarbon.

Examples of the organic precursor materials for method i) are pitch and phenolic resin. A typical
manufacturing procedure using a phenolic resin would be to cure after impregnation of the fibrous
reinforcement and then to pyrolyse at a higher temperature. Shrinkage occurs as the resin decomposes
during pyrolysis, and therefore, in order to obtain high densities, a number of impregnation, cure and
pyrolysis cycles are required. In the CVD process carbon is deposited by passing a hydrocarbon gas
through the fibrous reinforcement which is held at a temperature of about 1,100°C. NB A wide range of
ceramic fibres and matrices are possible (similarly for whisker reinforced composites) see table 11.1.

Figure 11.3 Scanning electron micrographs of the fracture surfaces of two glass-ceramic matrix composites: (a) tough
composite exhibiting pull-out; (b) less tough composite with planar fracture and negligible pull-out

65
The main forte of structural carbon-carbon composites is that they maintain their strength and toughness
to high temperatures of the order of 2,000°C. The superior high temperature mechanical properties,
especially when assessed in terms of specific strength, of these composites compared to alumina, silicon
nitride and to a high temperature nickel alloy (superalloy) is illustrated in Figure 11.4.

Figure 11.4 High temperature mechanical properties of carbon-carbon composite and some other high temperature
materials

The problem with carbon-carbon composites is that they oxidise and sublime in the presence of oxygen at
temperatures in excess of 600°C. Therefore for high temperature applications we must protect the
composite by either using an inert atmosphere or a coating. For example, the leading edges and the nose
cap of the space shuttle are sealed by a coat of silicon carbide. However, this does not give complete
protection as, due to the thermal mismatch between silicon carbide and the underlying composite,
microcracks develop in the coating. Consequently additional protection is supplied by impregnating the
surface with tetraethylorthosilicate which enhances oxidation resistance by leaving silica in the
microcracks.

Whisker-Reinforced Ceramics

Whiskers are usually single crystals that have grown preferentially along a specific crystal axis. Whiskers
typically range in size from about 0.5 to 10µm in diameter and a few microns to a few centimetres in
length. A note of caution is in order. The smaller whiskers can become lodged in the lungs and represent
a health hazard. Thus, they should be handled with proper precautions such as a hood, respirator, and
careful cleaning of work areas.

Whiskers generally grow under vapour-solid or vapour-liquid-solid conditions that result in a small defect
size in the whiskers. For example, defect size in some SiC whiskers was estimated to range from 0.1 to
0.4µm.

66
Defects identified were voids, inclusions, and surface roughness. Such small defect size results in very
high strength. Calculated strengths are in the range of 9 to 17 GPa for various SiC whiskers. Figure 11.5
illustrates various SiC whiskers as viewed by SEM.

Figure 11.5 β -Silicon carbide whiskers from pyrolysis of a phenylmethylpolysilane

Whiskers have been added to a wide variety of ceramics in an effort to achieve increased toughness. In
most cases, hot pressing has been required to achieve near-theoretical density of the matrix. The whiskers
are fully dense and rigid and thus do not shrink during sintering. As a result, they inhibit the material
transport of the matrix powder necessary for sintering to reach completion. Hot pressing provides the
force to reorient whiskers and to force the matrix material to move large enough distances to permit full
densification of the matrix around the whiskers. The reorientation of the whiskers during hot pressing
produces a microstructure with many of the whiskers aligned perpendicular to the direction of hot
pressing. This microstructure results in anisotropy in the strength and toughness. The strength and
toughness are strongly enhanced perpendicular to the length of the fibres, but decrease to values not much
better than the matrix parallel to the whiskers. Increased randomness of whisker orientation and
decreased anisotropy of strength and toughness have been demonstrated with conventional sintering with
<15% whisker addition, with hot isostatic pressing, and with liquid phase sintering.

Major efforts have been conducted with SiC whickers in Al2O3, Si3N4 and MoSi2. Table 11.1 summarizes
some of the results.

67
Table 11.1 Examples of Whisker-Reinforced Ceramics

68