Review - Methods For Improving The Mechanical Properties of Oxide Glasses

J O U R N A L OF M A T E R I A L S S C I E N C E 24 (1989) 4177 4208
Review
Methods for improving the mechanical
properties of oxide glasses
I. W. D O N A L D
Atomic Weapons Establishment, Aldermaston, Berkshire, UK

Methods are reviewed for improving the mechanical properties of oxide glasses. These are
divided into surface and bulk techniques and include thermal and chemical strengthening,
controlled crystallization, and particle, fibre and whisker reinforcement. The merits and
limitations of individual techniques are compared and discussed. In conclusion, a number of
applications for these materials are briefly outlined.
1. I n t r o d u c t i o n
Man-made oxide-based glasses, initially in the form of
decorative glazes, appeared around 6000 years ago in
Mesopotamia and Egypt. Around the beginning of the
first century AD, the developing art of glass-making
spread throughout the countries of the Roman
Empire. In the seventeenth century, man's perception
of the universe was transformed by the advent of the
optical telescope. At the present time, glasses are used
in diverse and far-ranging applications, including the
architectural, transportation, engineering, electronic,
telecommunications and aerospace industries. Glasses
in the form of optical fibres are today in the process of
revolutionizing the telecommunications and electronic
fields, and glasses are recognized as one of the most
useful and important classes of material known. For
many applications, however, the intrinsic brittle
behaviour of glasses, with their high susceptibility
to catastrophic and often impressive failure, is a
serious disadvantage. Over the years, therefore, many
attempts have been made at improving the mechanical
properties of these materials, either by making them
less prone to failure at low applied stresses, or by
preventing the catastrophic disintegration of materials
during failure.
Some of the methods that have been used for
improving the mechanical properties of glasses are
considered in this review, and their individual merits
and limitations are discussed. Methods are classified
under two broad headings, namely surface or bulk
techniques. Surface methods are used generally for
increasing the strength of glass or improving its
susceptibility to the influence of surface defects, whilst
bulk methods have been employed successfully for
increasing both the strength and fracture toughness of
these materials. Many of the methods to be described,
for example particle and fibre reinforcement, can also
be applied to other ceramic materials in addition to
oxide glasses; however, emphasis is given throughout
this review only to glassy oxide materials. In conclusion, a number of applications and potential
0022-2461/89 $03.00 + .12
9 1989 Chapman and Hall Ltd.
applications for strengthened and toughened glasses

are outlined.
2. Mechanical properties of glasses

The mechanical properties of oxide glasses, in which
covalent-ionic bonding predominates, are dominated
by brittle behaviour. For all practical purposes, these
materials exhibit completely elastic behaviour up to
their breaking points, with no indication of significant
macroscopic ductility, at least at temperatures less
than the glass transition temperature, Tg, and at
ambient hydrostatic pressures. The theoretical strength
of glass has been estimated to be of the order of E/10,
where E is Young's modulus [1]. This suggests that
oxide glasses should exhibit strengths of the order of
~7000MPa. In practice, however, useful strengths
rarely exceed 100 MPa. This large discrepancy between
theoretical and practical strengths has been explained
on the basis of defects in glass, particularly defects at
the surface, which act as stress concentration sites,
enabling the theoretical strength limit to be exceeded
locally for very small applied stresses. Griffith [2, 3],
based on earlier work by Inglis [4] and Kolosoff [5]
emphasized this weakening influence of defects on
existing materials, although the significance of his early
work went largely unappreciated for many years [6].
It had been shown by lnglis [4] for example, that the
maximum stress at a crack tip, crm, could be given by
Gm
2(Y ( C / ~ ) 1/2
(])
where cr is the applied stress, c the crack length, and e

the radius of curvature of the crack tip.
Griffith followed on from this work by developing
a theory, based on energy considerations, for the
propagation of pre-existing cracks, in which he
showed, for conditions of plane stress, that the fracture stress, ac, of a brittle material, could be given by
the relationship
r
(2ET/Trc) '/2
(2)
where 7 is the fracture surface energy.
4177
For conditions of plane strain, the relationship

becomes
ar =
[2E7/7c (1 - vZ)c]'/z
(3)
where, v is Poisson's ratio.

It follows from these expressions (Equations 1 to 3)
that sharp cracks act as mechanical levers by concentrating the stress at the crack tip. The smaller the
value of ~, or the higher the value of c, the greater this
stress concentration effect will be and, consequently,
the lower the strength of the material. The low
strengths of oxide glasses and related materials are
thereby a direct consequence of the presence of stress
concentrating defects, and this is coupled, for these
brittle materials, with a lack of significant operative
stress relieving mechanisms. Cracks once initiated
will, therefore, propagate unimpeded, under the
influence of a critical applied load, and hence failure is
catastrophic. This is in contrast to ductile metals
where dislocation motion can relieve stresses at crack
tips, effectively blunting the cracks and therefore
making these materials less susceptible to catastrophic
failure.
Extension of Griffith's earlier work has led to the
field of fracture mechanics and the concept of a critical
stress intensity factor, Kc, (Klc for conditions of plane
strain), which is directly related to the stress at
which a crack can propagate continuously, i.e., catastrophically. The critical stress intensity factor is a
measure of the "toughness" of a material, or its
resistance to crack propagation and, unlike strength,
it is usually regarded as a materials constant. The
strength of a brittle material is given by
~Tf = KiclYC 1/2
(4)
where Y is a dimensionless constant which takes into

account sample geometry and loading characteristics.
The strength is, therefore, dependent on the intrinsic
fracture toughness of the material and the severity of
defects. Specific details of fracture mechanics analyses
and experimental techniques, with particular attention to ceramic materials, are described more comprehensively elsewhere (e.g. [7]).
Another parameter used frequently to measure the
toughness of a material is the work of fracture [8].
This is defined as the work done per unit area in
propagating a crack. This technique measures the
mean fracture energy for the total process, unlike the
fracture mechanics treatment which gives information
on the fracture initiation energy. In general, the crack
initiation energy is not equal to the energy required to
propagate a crack, but may be greater in the case of a
brittle material, or less in the case of a ductile or
fibre-reinforced material. It should be noted, however,
that unlike fracture toughness, K~, work of fracture
has been found to be strongly dependent on loading
geometry (e.g. [9]) and therefore, like mechanical
strength, it is not a materials constant.
In addition to low mechanical strength and fracture toughness, oxide glasses suffer from another
serious disadvantage that further limits their usefulness, particularly as structural load-bearing materials.
This is the phenomenon of static fatigue whereby,
4178
depending on the environment, small, pre-existing

cracks or defects grow under the influence of an
applied stress which is considerably lower than the
normal fracture stress of the material. When a critical
flaw size is reached under these conditions, flaw
propagation becomes catastrophic. The overall mechanical behaviour of glass, therefore, depends strongly
on the environment, and is also a function of time.
Static fatigue data for oxide glasses have been reviewed
in detail by Adams and McMillan [10]. Static fatigue
can be likened to stress corrosion of crystalline
metallic materials where existing surface defects grow
under the influence of an external stress which is below
the normal failure stress of the material. For oxide
glasses, the corrosive medium is usually water, which
is normally present in the environment, although
other agents can invoke similar, but generally lesser,
effects [11].
Some typical mechanical property data comparing
glasses and other materials are given in Table I.
3. Methods for improving the

mechanical properties of glasses
There are a number of methods by which the mechanical properties of oxide glasses can be improved, as
summarized in Table II. Many of these methods can
lead to very substantial increases in fracture strength
of glasses, although toughness, and in particular resistance to catastrophic failure, "may not be improved
upon significantly. Other methods can lead to improvements in toughness at the expense of strength, whilst
others are effective at improving strength and toughness simultaneously. Methods can be very broadly
classified into surface or bulk techniques. Many of the
techniques to be described, particularly surface compressive methods, are also quite effective at limiting
the deleterious effects of static fatigue.
3.1. Surface methods for improvement in

strength
Fracture of a brittle solid is almost invariably initiated
from the surface due, as already described, to the
presence of stress intensifying defects; thus, it is
feasible to increase the strength of these materials by
removing or minimizing such defects, or by placing
the surface in a state of compression. Methods for
accomplishing this are outlined below.
3. 1.1. E t c h i n g a n d r e l a t e d t e c h n i q u e s
Surface damage can be removed very successfully by,
for example, etching using a dilute aqueous solution
of H F (or, ideally, a mixture of H F and another acid,
e.g. HzSO4 or HC1, that will dissolve water-insoluble
fluoride compounds formed by reaction between H F
and the glass). Bulk glasses exhibiting very high
mechanical strengths, in excess of 1000MPa, have
been produced by this method [12-14], in which cracks
are either removed almost completely or crack tip
radii increased significantly [15]. Reaction with other
liquid reagents, even water [16], can lead to strength
enhancement, and etching with anhydrous molten
salts, for example NaBF 4 in NaNO3 or KBF 4 in KNO3
[17], can enhance strength. Similarly, the related
T A B LE I Some typical values for the mechanical properties of various materials (in dense bulk form)
Material
Epoxy resin
Fused silica
Oxide glass
Glass-ceramic
Thermally
strengthened glass
Chemically
strengthened glass
Chemically strengthened
glass-ceramic
M gO
Alumina
Si3N 4
SiC
B4C
Diamond
Y203 partially
stabilized zirconia
16 vol % ZrO 2
dispersion-strengthened alumina
Cast iron*
Aluminium
alloy*
Ductile high
strength steel*
High strength
superalloy*
High strength
titanium alloy*
Maximum
flexural strength
(MPa)
Work of
fracture
(kJ m 2)
100
50
100
400
500
0.2
0.002
0.003
0.005
-
900
--
1500
140
500
800
> 400
400
650
1200
320
580
1800
0.3
0.05
< 0.17
< 0.05
0.01
-
Fracture
toughness, K~
(MPa m 1'2)
E
(GPa)
Hardness
(VPN)
0.8
0.6
0.6
0.8-2.6
-
2-4
75
140
150
140
640
700
950
700
--
140
> 700
150
950
< 3.0
<4
< 7
< 5
< 5
3.4
6.4
690
1400-1900
1600-1800
2400-2800
2800-3200
m 8000
1300
15.0
> 7
~10
20-100
80
I40
5-130
30-250
200-550
200
200-450
1400
[420
250
400
320
440
450
960
200
55
125
*Tensile data
technique of "flame polishing" in which the surface of

glass is softened by heating, leading to removal or
healing of flaws, is a standard method for improving
the mechanical properties of glass articles. Simple
annealing can also increase the strength of glass
by blunting crack tips [18], although heating at
temperatures less than the glass transition temperature can have the reverse effect and lead to a decrease
in strength [19]. Unfortunately, the increases in
strength effected by these processes are very transient
in nature, and normal handling rapidly reduces the
strength of glass articles to values approaching their
former low values. After etching or subjecting to other
surface treatments, however, the surface may be
protected to some degree by application of a suitable
protective coating; this will be described in more detail
later.
3. 1.2. Formation of a surface compressive
layer
A number of methods have been devised for generating surface compressive stresses. Strength increases
are achieved because the compressive stress at the
surface must be overcome before defects are subjected
to tensile forces. As a very rough guide, the strength of
a glass article with a surface compressive stress is
equal to the magnitude of the compressive stress plus
the normal fracture strength of the untreated material.
In order to impart useful strength the depth of a
compressive layer must generally be greater than the
size of typical flaws, i.e. greater than 50 #m.
3.1.2.1. Thermal treatments. One method by which a

state of surface compression can be achieved is to cool
the glass rapidly from a temperature above the glass
transition temperature. Rapid cooling, in this sense,
normally involves quenching by air jets or, more
recently, by jets of liquid, or even gas fluidized particulate matter [20], directed at the glass surface. This
process gives rise to the familiar "toughened", or
"tempered" glass, used for many years in the automobile industry. In this respect, the common phrases
employed, that is to say, "toughened" or "tempered",
can be misleading since it is the strength of the glass
and its strain to failure that are increased, and strictly
speaking toughness is only increased very marginally;
and when fracture does occur, it is still catastrophic.
The surface of the glass cools more rapidly than the
interior under the quenching conditions; thus, the
temperature of the surface rapidly falls below ~ , the
surface then behaving as a rigid, elastic Hookean
solid. Visco-elastic deformation of the interior of the
glass is still possible, however, during the stage when
the glass surface becomes rigid and contracts. Stresses
in the interior are therefore relaxed initially until a
stage is reached when the interior also becomes rigid.
As, at this stage, the temperature of the interior is still
higher than that of the surface, contraction of
the interior is opposed by the surface; consequently
the surface is placed in a state of compression as the
temperature differential equilibrates. A balancing
tensile stress is developed in the interior of the glass.
The stress distribution obtained by this method is
4179
TA B LE I I Summary of methods for improving the mechanical properties of glasses

Method
Applicable to
Comments/Mechanism
Surface modification
(a) Etching and related
techniques
most glasses
removal of surface damage
(b) Thermal
strengthening
limited to glasses of
t >/ 2mm, a n d e /> 4.5 x 10 6~
surface "solidifies" before interior:

this gives surface compression
(c) Chemical
strengthening
many alkali-containing
glasses, particularly
alkali alumino-silicates
exchange of small ions in glass surface

by larger ions to give higher density
surface: this gives surface compression
(d) Surface
crystallization
limited to a number of
specific compositions
formation of lower expansion surface

crystallized layer: this gives surface
compression
(e) Thin coating
most glasses
protection of pristine surface
(f) Cladding
glasses of c~ > cladding
coating of surface with compatible

lower expansion material: this gives
surface compression
Bulk method~
(a) Bulk crystallization
many silicate glasses
internal nucleation and growth of

crystals to give fine-grained,
porosity-free polycrystalline
glass-ceramic material: grain
boundaries impede crack propagation
(b) Particle
reinforcement
many glasses
dispersion-strengthened or age-hardened
systems: this gives limited load transfer
and particle deformation; particles can
impede crack propagation
(c) Transformation
toughening
glasses containing Zr02
special case of dispersion-strengthening

involves stress-induced structural
transformation in particulate phase
during crack propagation
(d) Fibre reinforcement
most glasses
crack impedement, load transfer, fibre

deformation/pull-out, matrix
microcracking
(e) Whisker
reinforcement
most glasses
crack impedement, load transfer,

whisker pull-out for large aspect ratios,
matrix microcracking
many glasses
alternate layers of glass and

thermoplastic material to give
composite sandwich
(f) Laminated
structures
'
approximately parabolic, as illustrated in Fig. 1. As a

rough guide, the compressive stress at the surface, ac,
is approximately twice the value of the tensile stress,
a,, in the interior, and the thickness of the compressive
layer, to, is around 20 to 25% that of the glass itself.
The magnitude of the surface compressive stress

attainable by thermal methods is limited by a number
of factors. These include limitations imposed by
the maximum practical cooling rate achievable and
the thermal characteristics of the glass, in addition
II C
0"t
Glass Surface
41 80
Glass Surface
Figure 1 Typical stress profile of a thermally

str mgthened glass.
to geometrical considerations. These tend to limit

thermal methods to glass articles of relatively simple
shape, moderately high thermal expansion, and thicknesses greater than about 1.5 to 2 ram. The maximum
strength realized in practice is of the order of 350 to
400 MPa, which is approximately three to five times
stronger than the untreated glass.
As a consequence of the high strengthening stresses,
a significant amount of strain energy can be stored
in a thermally strengthened glass. When fracture
occurs, by penetration of a crack through the compressive zone into the tensile region, this strain energy
is released and appears largely in the form of the
surface energy of the fragments produced. Consequently, depending on the precise magnitude of the
internal tension, a large surface area may be created,
and this can give rise to a considerable number of
small, approximately equi-sized glass fragments.
Because the interior of the glass is in a state of tension,
it is important that the parent glass is of very high
quality. If, for example, internal defects are present,
they may lead to premature failure of the article, either
during the quenching process or, more deleteriously,
at a later date whilst the article is in service.
Thermal methods were first employed in an attempt
to improve the strength of glass articles in the 1870s,
but it was not until the 1920s that practical methods
of achieving the cooling rates required, without
damaging the glass, were devised, as reviewed recently
by Gardon [21]. Further more detailed information on
thermal strengthening, including a comprehensive
description of the theoretical concepts, is also provided by Gardon.
3.1.2.2. S u r f a c e crystallization. A number of reports
have been given of glasses strengthened by a simple

heat treatment in air to give a surface crystallized layer
which has a lower thermal expansion than that of the
bulk glass. On cooling these materials from the crystallization temperature, the interior attempts to contract more than the surface, and so places the surface
in compression. As for the thermally strengthened
glasses, a balancing tensile stress is created in the
interior. Unlike the parabolic stress distribution noted
for thermally strengthened glasses, however, the stress
distribution obtained by surface crystallization is
usually very different in character. In particular, the
transition from compression to tension may be very
sharp. This sharp transition can lead to degradation of
the glass article due to spalling of the surface layers,
either during the surface crystallization process, or
later due to the influence of static fatigue.
Glasses most suitable for strengthening by surface
crystallization include lithium alumino-silicates,
where a crystalline surface layer of low expansion
/%spodumene or //-eucryptite can be produced, as
reported by Stookey et al. [22], Petticrew et al. [23],
and Keifer et al. [24], and lithium silicates, as noted by
McMillan et al. [25]. Other compositions include zinc
alumino-silicates, where low expansion willemite
or stuffed keatite can be formed, as reported by
McMillan et al. [26], and McMillan and Partridge
[27-29].
Strengthening has also been noted by Adams and

McMillan [30] for calcium aluminate based glasses,
surface crystallized to produce a crystalline phase
with a similar thermal expansion to that of the bulk
glass. In this instance, however, the strength increases
involved were generally less than that obtained by
formation of a lower expansion surface, and strengthening was attributed to the limiting in size of Griffithtype flaws in the glass surface. Attempts have been
made at surface crystallizing other glass compositions,
but with only limited success, as reviewed recently by
Partridge [31].
In order to promote surface crystallization, rather
than crystallization of the bulk material, it is necessary
to provide crystal nucleation sites on the glass surface.
This has been accomplished by various techniques
aimed at providing, for example, an even distribution
of very fine scratches in the surface from which crystals can nucleate and grow [25-27]. One of the most
successful methods has been polishing, followed by
vibration in a bed of sand or silicon carbide grit using
a vibratory table [25]. On subsequent heat treatment
of the glass, this provides a homogeneous distribution
of small evenly spaced crystals at the surface.
Some representative mechanical property data for
surface crystallized glasses are given in Table IlI.
3.1.2.3 Chemical treatments. In addition to the thermal
treatments which yield surface compressive stresses,

chemical methods are also feasible. Chemical methods
have the distinct advantage that almost any geometry
can be employed and, in addition, very thin sections
less than 2ram in thickness can be used, because
the method does not rely on the setting up of thermal
gradients. Furthermore, in contrast to thermal
strengthening, glasses with low or even zero thermal
expansion coefficients can be treated successfully by
chemical methods. Treatment is also carried out at
temperatures lower than Tg so that problems associated with viscous deformation and consequent
permanent distortion of the body, which can occur
during thermal treatments, are alleviated.
Chemical strengthening generally relies on exchanging small ions in a glass surface by larger ions. The
larger ions can be provided by treating the glass article
in a molten salt bath. Substitution of larger for smaller
ions in the glass surface tends to expand the glass
structure, but this expansion is restricted by the glass
interior. Consequently, the surface of the glass is
placed in a state of compression, and a balancing
tensile stress is again generated in the interior. The
stress profile of chemically strengthened glasses is
similar to that of a surface crystallized glass, although
the stress gradients are not normally as steep, and the
transition from compression to tension is less severe;
a typical stress profile for a chemically strengthened
glass is illustrated in Fig. 2.
The successful use of chemical ion-exchange
strengthening was first reported by Kistler in 1962 [32]
for soda-lime-silica glasses treated in potassium
nitrate. The driving force for chemical ion-exchange is
the chemical concentration gradient, and this depends
on the diffusivity of the respective ions.The highest
4181
T A B L E I I I Details of surface crystallized glasses

Glass
number*
Heat-treatment conditions
Temperature
(~
(a)
890
860
890
860
800
+ 850
800
+ 850
800
+ 850
800
+ 850
800
+ 850
800
+ 850
800
+ 850
750
750
750
750
800
800
850
700
800
850
800
850
3
4
5
6
7
8
9
10
11
12
12
13
13
14
14
14
15
15
80
80
I00
IO0
Before
(MPa)
Reference
After
(MPa)
18
48
20
60
600-700
600-700
600-700
600-700
[22]
[22]
[22]
[22]
2
+2
2
+2
2
+2
2
+2
2
+2
2
+2
2
+2
0.5
0.5
0.5
50
16.7
9.4
1.9
I6.7
0.5
0.1
9.9
2.4
willemite
366
[271
willemite
372
[27]
willemite
439
[27]
willemite
414-517
[27]
willemite
524
[27]
willemite
607-634
[27]
willemite
708
[271
1.0
2.5
4.3
1.0
2.5
willemite
willemite
willemite
willemite
willemite
493
[28]
551
[28]
60
60
90
90
100
100
100
160
160
70
70
70
70
70
70
70
70
70
584
300
300
620
620
670
670
670
830
830
[28]
[29]
[29]
[29]
[29]
[29]
[29]
[29]
[29]
[29]
11
40
49
23
54
90
90
90
90
90
145
245
275
180
283
[30]
[30]
[30]
[30]
[30]
20
120
510
[30]
albite
albite
stuffed
stuffed
stuffed
stuffed
stuffed
stuffed
stuffed
keatite
keatite
keatite
keatite
keatite
keatite
keatite
Calcium zmc alummo-silicategNsses

16
16
16
16
16
(d)
fl-eucryptite
fl-eucryptite
fl-eucryptite
fl-eucryptite
Flexural strength
Zinc alumina-silicateglasses
2
(c)
Thickness
of surface
layer
(gm)
Lithium alumina-silicateglasses
1
1
1
1
(b)
Time
(h)
Major crystalline phase

formed
820
820
820
840
840
Calciumaluminateglasses
17
850
*Specific glass compositions are given in Table VI
aC
/
O_
E
8
\\
Iiiii//
~1 \\\\
\\I
/I J
~-
tc2
-I
Glass ~urface
4182
"
/
Glass Surface
Figure2
Typical stress profile of a chemically

strengthened glass.
T A B L E I V Summary of ion-exchange treatments, ionic radii differences and maximum compressive stresses
Basic
glass composition
Salt bath
Mechanism
Ionic radius
difference
(nm)
Approximate
volume change*
(#m 3g- i )
Maximum
induced
compressive stress*
(MPa)
Li20-(A1203)-SiO 2
Na20-(Al203)-SiO 2
Li~O-(AI,O3)-SiO 2
Li,O-(AI=O3)-SiO2
Li20-(AI203)-SiO 2
MgO-(AI~O3)-SiO 2
NaNO 3
KNO 3
KNO~
RbNO 3
CsNO 3
Li2SO4
Na + ~.~ Li +
K ~ ~.~ Na +
K + ~,~ Li +
Rb + ~ Li +
Cs + ~- Li +
2Li + ~ Mg 2+
0.020
0.035
0.055
0.071
0.087
90
220
310
440
665
620
1450
2000
2210
3965
*taken from [58].
diffusivities are generally found for alumino-silicate

compositions, although early work [33, 34] indicated
that alkali zirconia-silicate systems also exhibited
useful diffusivities. These glasses therefore provide the
most useful materials by yielding compressive layers
of practical thicknesses in realistic treatment times.
Using, for example, a lithium alumino-silicate composition, Li + ions near the glass surface are replaced
by larger Na + ions from a sodium salt bath, for
example, sodium nitrate [35-45]. Similarly, for sodium
alumino-silicate glasses, Na ions would be replaced
by larger K ions from a potassium salt bath, for
example, potassium nitrate [46-53]. Numerous other
ion-exchange treatments have been investigated,
including mixed multi-ion exchanges (e.g. [41]), and
exchange of Na + or K + for Rb +, Cs +, Ag +, Cd 2+,
Zn 2+ or Cu +/Cu 2+ [43, 46, 54-57]. Assuming no stress
relaxation or related effects, very high compressive
stresses are theoretically feasible by ion-exchange [58],
as noted in Table IV.
Ion-exchange is a diffusion-controlled process, thus
it is temperature and time dependent. Consequently,
the higher the ion-exchange temperature, the shorter
the time required to develop a compressive layer of
practical thickness. The upper temperature is limited,
however, due to the possibility of stress relaxation.
Hence, if ion-exchange is carried out at temperatures
approaching or exceeding Tg, the stresses generated by
the process are quickly relaxed or eliminated by
viscous flow of the glass. For a given glass composition, the strength increase achievable by ionexchange depends on the treatment conditions, i.e.
temperature and time. A maximum in strength is

normally observed as a function of time, as illustrated
in Fig. 3 for a range of glass compositions and treatment conditions. For a given glass composition, maximum strength is achieved at shorter treatment times as
the temperature is increased; however, as shown in
Fig. 4, due to the result of stress relaxation effects, the
overall strength decreases with increasing treatment
temperature. Similarly, as shown in Fig. 5, a strength
decrease is noted on heating a chemically strengthened
glass in air, the magnitude of the decrease being
dependent on the temperature and time (e.g. [48, 52]).
This is again due to the effects of stress relaxation and
ionic diffusion which lead to a reduction in the surface
compressive stress. Ion-exchange times, t, may vary
from a few minutes to several hundred hours, depending on the precise glass composition, the temperature
of the bath, and the depth and magnitude of compressive layer required; a characteristic t ~/2 dependence
is normally observed. A prerequisite for an ionexchange medium is, of course, that it does not chemically corrode or otherwise degrade the glass surface at
the temperature employed. Materials have also been
p,'oduced by electric field-assisted ion-exchange in
which a d.c. field is used to enhance the ion-exchange
process [59-64].
Very high mechanical strengths can be achieved
using the chemical ion-exchange method, with some
reports of materials exhibiting strengths in flexure in
excess of 800 MPa [33, 45, 62, 64]. This is nearly an
order of magnitude higher than untreated glass, and is
a direct consequence of the need to overcome the very
800
600
//
13.
a
s
c
400
IJ_
200
01
8
t I12 (hi/2}
10
Figure 3 Strength of chemical ion-exchanged

glasses as a function of treatment time: a and c
lithium magnesium alumino-silicate glasses; after
[44] and [45], b sodium silicate glass; after [43], d
sodium calcium silicate glass; after [86], e sodium
calcium silicate glass; after [47].
4183
Figure 4 Strength of an ion-exchanged sodium
800
alumino-silicate glass as a function of treatment

time at different temperatures (a 450 ~ C, b 500 ~ C,
c 550~ d 575~ after [34].
600
Cl
O..
4;
400
kL
20C
2
/.112 (hl/2)
high compressive stresses during external loading of

the glass, before the strength-impairing effects of surface defects become significant.
The thickness of the compressive layer may vary
from a few micrometres to several hundred mi~crometres, depending on the precise glass composition,
treatment conditions and starting glass thickness. In
general, the thickness of the layer will be less than that
obtained during thermal treatments. If the layer is
very thin relative to the overall thickness of the glass
(tcl in Fig. 2), the maximum balancing internal tensile
stress, o-~,, may be very small so that, unlike thermally
strengthened glasses, these materials can be cut or
machined after treatment. On the other hand, for
compressive layers that are thick relative to the overall
thickness of the glass (tc2 in Fig. 2), the internal tensile
stress, ~t2, may be above that required for rapid crack
branching, so that when fracture is initiated, by penetration of a crack through the compressive layer, a
large number of small, approximately equi-sized glass
fragments is produced, almost explosively. This
phenomenon has given rise to the special class of
chemically strengthened materials, the "frangible" or
"command-break" glasses, (e.g. [65]).
In general, the ion-exchanged layer formed during
the chemical strengthening of glass articles remains
amorphous. A number of reports have shown, however, that treatment in a salt bath at temperatures
above the range normally employed can induce crystallization of the ion-exchanged surface layer. This
modified technique may be used successfully under
certain conditions for increasing the strength very
significantly [22, 66, 67]. It is not practised widely,
however, due in part to the hazards associated with
the use of salt baths at the high temperatures necessary
to induce crystallization. In addition, deformation of
the article may occur due to viscous flow, since the
treatment temperature is greater than ~ .
In addition to glasses, some glass-ceramics have also
been treated successfully using chemical ion-exchange
[68-70]. These include Na20-(K20)-A1203-SiO 2 compositions, nucleated with TiO2, to give glass-ceramics
containing nepheline crystals (Na3KA14Si4016). Treatment of these glass-ceramics in a potassium salt bath
results in the partial or complete exchange ofNa ions
in the glass-ceramic surface by K + ions, resulting in a
surface volume increase and the creation of a surface
compressive layer. Ultra-high strengths approaching
1500MPa have been achieved by ion-exchange of
these glass-ceramics. Alternatively, treatment of
MgO-A1203-SiO 2 glass-ceramics, nucleated by ZrO 2
or TiO2 and containing/?-quartz crystals, in a lithium
800
6O0
rl
~s
40C
0)
1.1_
Figure 5 Ambient temperature strength of
2OO
'
2;0
'
Ageing temperature (~
4184
400'
'
660
ion-exchanged glasses as a function of

heating temperature in air-glasses aged for
times indicated.(a lithium alumino-silicate
glass; 4 h, after [48], b unspecified glass;
88
h, c unspecified glass 100 h, d unspecified
glass 500 h, after [52]).
TAB L E V Details of chemically strengthened materials

Glass No.*
Salt
Treatment conditions
Temperature
(o C)
(a)
(b)
(c)
(d)
Mechanism
Flexural strength
Time
(h)
Before
After
(MPa)
(MPa)
Reference
Littlium alumino-silieate glasses

18
19
20
21
22
23
23
NaNO~
NaNO3
NaNO 3
NaNO 3
NaNO 3
NaNO3
NaNO 3
400
400
400
400
400
23
24
25
26
NaNO 3
NaNO 3
45% Na2SO 4 + 55% ZnSO 4
NaNO 3
4
4
4
4
4
385
385
385
385
585
6
49
97.5
49
0.25
404
Na
Na
Na
Na
Na
Na
Na
Na
Na
Na
Na
+
+ ~+
+
+
Li +
Li +
Li +
Li +
Li +
+
Li +
+ ~-~ Li +
+ ~.~ Li +
~.~ Li ~
+
Li ~
+
Li +
74
74
74
731
681
669
703
572
470
383
336
[36]
[36]
[36]
[36]
[36]
[44]
[44]
[44]
69
309
400
669
[44]
[39]
[42]
234
234
234
579
497
607
598
476
338
200
600
469
498
650
700
690
[47]
[47]
Sodium alumino-silicale glasses

27
27
27
28
29
30
31
32
33
34
35
36
37
KNO 3
KNO 3
KNO 3
KNO 3
KNO 3
KNO 3
KNO 3
KNO~
KNO~
KNO~
KNO 3
KNO 3
KNQ
350
350
400
400
400
400
400
400
500
500
482
482
400
0.17
17
0.I7
16
16
I6
I6
I6
24
24
4
4
24
K+~-Na
K + ~ Na
K + ~.~ N a
K+~.~Na
K+~-Na
K + ~Na
K + .~Na
K + ~-Na
K + .~Na
K + ~Na
K + ~.~Na
K + ~Na
K + ~- Na
+
+
+
+
+
+
~
+
+
+
+
+
+
37
37
KNQ
KNO~
430
430
1
16
K + ~- Na +
K + ~ Na +
385
385
385
385
454
496
454
427
49
49
49
49
0.5
0.5
1.5
1.5
Na
Na
Na
Na
Na
Na
K
K
+
+
+
+
+
+
+
+
~
~.~
~_
~
~
~
~
~
Li +
Li +
Li +
Li +
Li +
Li +
Na*
Na +
80-100
80-I00
80 I00
80-100
400
400
400
400
4
16
4
4
Na
K
Na
Na
+
+
+
+
~
~
~.~
~.~
Li +
Na +
Li +
Li +
62
590
8
8
8
96
8
8
4
4
8
4
24
K+
K+
K+
K+
K+
K+
2Li +
2Li +
2Li +
~_
~_
~
~
~
~
~.~
~
.~
Na +
Na +
Na +
Na +
Na +
Na +
M g :+
M g 2+
M g 2+
54
54
55
55
55
65
552
483
[471
[471
[481
[481
[481
[481
[49]
[491
[5~1
[51]
[43]
[43]
[43]
Mixed lithium sodium alumino-silicate glasses

38
39
40
41
42
NaNO
NaNO
NaNO
NaNO
NaNO
3
3
3
3
3
42
42
43
NaNO 3
KNO~
KNO 3
503
516
574
858
414
666
772
786
[44]
[44]
[44]
[44]
[41]
228
235
441
366
[36]
[47]
[431
600
1407
862
1303
1379
1476
207
379
1069
310
448
[68]
[681
[68]
{68]
[411
[411
[411
Aluminaz/?ee alkali silicate glasses

44
45
46
47
NaNO 3
KNO~
NaNO 3
NaNO 3
55-83
[43]
(e) Glass-ceramic compositions
(f)
48
KNQ
48
49
49
49
50
5[
52
53
54
52%
KNO
KNO
52%
52%
90%
90%
90%
52%
55
Li 2 S O 4
KC1 +
3
3
KC1 +
KCI +
Li2SO 4
Li2SO 4
Li2SO 4
KC1 +
48% KaSO 4
48% K2SO 4
48% K2SO 4
+ 10% K2SO 4
+ 10% K2SO 4
Jr- 1 0 % K 2 S O 4
48% K2SO 4
730
590
590
730
730
800
800
850
780
1050
K + .~_ Li +
2Li + ~ M g 2+
55
103
100
[68]
[68]
[69]
[69]
[69]
[69]
[701
Glass compositions surface crystallized by chemical methods"

56
95% Li2SO 4 + 5% Na2SO 4
860
0.08
57
9 5 % Li2SO 4 + 5 % N a 2 S O 4
860
0.25
fl-eucryptite
surface layer
fi-eucryptite
surface layer
638
[22]
700
[22]
*Specific glass c o m p o s i t i o n s a r e given in T a b l e VI.
4185
A summary of mechanical property data for a

variety of chemically strengthened materials is given
in Table V. Details of the specific glass compositions
reported in both surface crystallization and chemical
ion-exchange work are given in Table VI. Further
information on chemical strengthening may be found
in the paper by Bartholomew and Garfinkel [33].
salt bath leads to exchange of Li ions for Mg 2+ ions

(2 Li + ~ Mg 2+). An increase in the surface density is
again achieved, leading to the formation of a surface
compressive layer. Strengths up to around 1100 MPa
have been achieved using these compositions.
Glass fibres have also been treated by ion-exchange
[71]. In this instance, however, an increase in strength
was found only on treating unetched fibres, a decrease
in strength being noted on treatment of etched
material.
TAB
LE
Glass
number
1
3.1.2.4. Other surface treatments. Methods of removing surface damage, for example by etching, can
Specific glass-compositions (tool % )
V I
Li20
Na20
11.34
K20
0.31
CaO
0.07
TiO 2
MgO
70.96
0.16
AszO 3
1.08
1.98
PbO
39.69
9.88
49.34
1.09
4.97
34.55
9.96
49.41
1.11
37.50
9.88
49.45
1.08
1.99
37.55
9.99
49.39
1.08
1.96
38.05
10.03
49.96
31.64
11.22
57.14
6.74
54.88
10.13
50.73
4.60
33.78
I0
4.94
34.20
11
4.64
12
1.17
13
14
4.73
15
16
17
.
.
-
2.00
6.17
Others
49.43
P205
9.88
SiOz
16.43
32.08
9.51
47.60
40.00
10.00
48.83
40.05
1.05
10.01
48.89
37.05
9.69
48.53
0.91
40.10
10.02
48.97
24.00
20.30
9.60
46.10
63.16
.
.
0.09
A1203
--
B203
37.63
0.64
ZnO
36.84
18
18.00
22.00
60.00
19
18.00
26.00
56.00
20
22.00
18.00
60.00
21
22.00
22.00
56.00
22
16.34
23
29.60
24
30.30
25
16.67
26
11.03
0.47
0.06
.
.
9.07
20.01
54.58
9.96
9.90
49.39
1.15
8.12
10.11
50.28
1.19
16.67
66.66
16.19
71.08
0.19
As203
t.31
72.65
0.67
BaO
11.69
72.31
0.62
.
-
0.36
27
13.71
0.97
6.17
4.52
28
11.33
2.53
0.24
1.90
29
15.8l
5.30
5.16
0.58
73.15
30
--
14.40
9.55
3.69
0.58
71.78
31
--
3.98
10.63
1.21
84.18
32
17.29
1.42
2.39
3.77
--
1.92
12.48
60.73
33
13.35
2.47
0.23
0.65
5.93
10.78
66.59
34
--
13.53
2.47
0.23
5.92
10.76
67.09
35
--
21.83
2.39
--
4.62
11.06
49.41
2.75
ZrO 2
36
--
20.95
2.30
3.52
4.49
12.44
46.68
7.62
2.00
ZrO 2
37
--
15.00
2.00
6.50
4.50
1.00
70.00
1.00
PbO
38
20.76
4.82
10.35
2.43
10.29
51.35
10.92
3.79
11.51
50.81
1.03
11.03
3.84
11.63
51.33
19.42
1.20
10.42
47.00
0.98
39
17.68
4.26
40
17.86
4.31
41
18.72
2.26
42
11.61
12.22
2.54
17.96
50.84
4.83
43
11.03
7.38
1.61
4.70
12.57
61.35
1.36
44
25.00
75.00
45
76.40
72.10
9.93
ZrO 2 +
0.15As203
76.46
7.31
ZrO 2 +
0.15As203
.
14.81
46
17.82
47
16.08
48
49
50
51
5.15
52
4.08
13.87
5.85
11.42
6.86
.
-
8.79
.
.
6.68
22.16
51.53
0.26
As203
6.80
22.48
50.74
0.26
As203
6.30
20.82
54.36
0.24
As203
6.12
1.51
12.70
72.36
2.00
ZrO 2 +
9.09
13.I8
71.17
2.48
ZrO 2
19.23
18.72
54.59
11.69
75.26
.
-
53
54
4.08
6.83
55
3.42
4.73
19.24
69.51
56
2.49
12.00
5.25
--
15.72
64.37
57
7.09
7.62
--
4.89
14.06
66.18
41 86
.
2.37
17.00
7.46
0.16
AszO ~
1.99
ZrO 2 4- 0.15
3.10
ZrO2
0.17
As203
0.16
As203
As203
promote very large, although transient, increases in

mechanical strength of brittle materials. More practical methods include cladding the material with a more
resiliant coating, or cladding with a higher thermal
expansion material to promote the formation of a
surface compressive layer. These methods are not in
general, however, as versatile as the methods of
thermal or chemical strengthening. Cladding of glass
articles with a glass coating of lower thermal expansion has been practised for a number of years [72].
More recent work by James [73] has concentrated on
coating glass rods of various compositions with low
thermal expansion borate glass. Using this low viscosity glass, consistent, even coatings ~ 2 0 0 # m thick
were readily achieved. Strength increases were noted
for abraded rods from around 110 to 350 MPa. Glass
fibres have also been strengthened by cladding.
For example, Krohn and Cooper [74] produced
borosilicate glass fibres coated with silica by the
drawing down of a silica coated glass rod. For a fibre
of outside diameter about 180/~m with a core to cladding ratio of 2/3, a strength increase for abraded fibres
from about 155 to 290 MPa was noted. Krohn [75]
also examined the influence of thermal expansion
mismatch, A~, between different glass cores and claddings; a maximum surface compression of about
320MPa was found for a system of A~ ,~ 8.5 x
10-6~ ~. Enhanced strength attributed to surface
compression has also been noted for glass articles,
including fibres, coated with a thin layer of titania
[76-78]. Similarly, coating with tin oxide improves the
mechanical properties [79]. Finally, coating with polymeric materials has been used for many years in the
glass fibre industry for improving the resistance of
glass fibres to surface damage during handling and
storage [80].
centre tension, ac, can be derived using the following

expression
b
~v/~cc
(5)
where 2 is the wavelength of light employed.

The glass is then etched in an aqueous solution of
H F + H2SO 4 to remove some of the surface. After
removal of a thin surface layer, the new, reduced
centre tension is measured. This process is then
repeated a convenient number of times until zero or
near-zero centre tension is achieved. An excellent
example of the variation in fringe separation with
thickness change, taken from the work of Bradshaw
[84], is shown in Fig. 6. The stress profile is subsequently derived from the relationship between the
thickness change and centre tension change. This may
be accomplished by plotting a graph of thickness
change against centre tension and measuring the
gradient at successive points to calculate the stress
profile using the relationship
a
G(t/2 -
L) + Actc
(6)
where a is the stress in the layer removed, G the

gradient of the curve, t the original thickness of the
sample, L the thickness loss per side, and Acrethe total
decrease in centre tension taken from the start of the
etching treatment.
In general, the stress profiles measured experimentally differ somewhat from the idealised profile for
a chemically strengthened glass shown in Fig. 2. Very
often, the maximum compressive stress is found not at
the surface, but at some distance within the glass
3. 1.3. S t r e s s p r o f i l e d e t e r m i n a t i o n
The stress profile of surface modified glasses is an

important property which controls not only the ultimate strength of the material, but also the way in
which it fractures. Stress profiles in optically transparent materials may be measured by conventional
photoelastic methods [81], by the scattered light
technique [81, 82], or by progressively removing
surface layers by etching and then measuring the
resultant dimensional changes [83]. This latter method
is also suitable for measuring the stress profiles of
optically opaque materials. An alternative method for
measuring transparent materials, which is more suited
to the measurement of stress profiles in thin, chemically strengthened glasses where the stress gradients
are relatively high, has been developed by Bradshaw
[84]. This method combines the scattered light technique with etching. It involves the scattering of light
from a finely collimated laser beam passed through the
central plane of the sample and plane polarized at 45 ~
to the plane of the sample. Due to the birefringence of
the glass when stressed, this beam splits into two
components and gives rise to interference fringes
when viewed at 45 ~ to the surface of the glass. From
a knowledge of the fringe separation, b, and the stress
optical coefficient, C, of the glass, a value for the
Figure 6 Variation in fringe separation with decrease in thickness

for Corning 0319 chemically strengthened glass; after [84]. (I is
as-received material with t = 1.3 mm; 2 to 18 are after progressively
etching away some of the surface with t = 0 . 7 m m for I8.)
4187
Figure 7 "Relaxed" stress profiles of chemically

strengthened glasses: (a) lithium magnesium
alumino-silicate glass; after [44], (b) sodium
alumino-silicate glass (Coming 0319); after [84].
"~ 200
2~
go
N
;200
Thickness (#m ) ~ ~ ~ _
'
3(~0
~. 200
(a)
Glass Surface
400
Eo200
O
n
:s
i
09
ess (/zm)
zoo (b)
Glass Surface
interior, as shown in Fig. 7a for a Li20-Na20-MgOA1203-SiO 2 glass treated in NaNO3 [44, 45], and in
Fig. 7b for a commercially available Na20-AI203SiO2 glass [84]. This type of effect may be related
to the influence of thermal stress relaxation of the
glass during treatment, particularly when treatment
is carried out at temperatures approaching Tg.
Alternatively, it may be related to differences in
thermal expansion between the bulk glass and the ionexchanged layer, the latter possessing a higher thermal
expansion than that of the bulk material. In the
extreme case, this expansion mismatch may be sufficiently large to create tension in the outermost layers
of the glass, rather than compression, and this may
lead to crazing and spalling of the glass surface, as has

been observed by Donald and Hill [44, 45] for a LizOMgO-Al203-SiO 2 glass. This effect is shown in Fig. 8.
Further stress profile data for a variety of thermally
strengthened and chemically strengthened glasses
have also been reported by a number of authors,
employing some of the techniques outlined above [38,
42, 48, 51, 84-86].
3. 1.4. Fracture behaviour and life prediction

studies of surface modified g/asses
All these surface methods just described increase the
strength of brittle oxide glasses, and their strain to
failure. In all instances, however, ultimate failure
200
Eo
2
0
83
g
200
[a)
Glass Surface
Figure 8 Chemically strengthened lithium magnesium alumino-silicate glass: (a) anomalous stress profile exhibiting a thin surface tensile
zone, (b) surface crazing effect caused as a result of this tensile layer; after [44].
4188
is still catastrophic, the materials behaving as fully

elastic solids, and fracture toughness is not improved
upon significantly.
The actual fracture behaviour of surface modified
glasses has been studied extensively in terms of examining the short time failure characteristics of these
materials under dynamic loading conditions. Typical
short time fracture behaviour, as a function of treatment time, for a Li20-MgO-A1203-SiO 2 glass of
thickness of about 1.3 mm treated in NaNO3 at 385 ~
is illustrated in Fig. 9, after the work of Donald
and Hill [44]. The decrease in flanged particle size with
increasing treatment time is a consequence of the
increasing magnitude of the internal tensile stress and
correspondingly higher internal strain energy. There
has been much less work in the important areas of
impact damage and life assessment studies. This is
partly because the relatively high stress gradients
associated with these materials makes analysis by conventional fracture mechanics techniques very difficult.
In one early study, however, Barsom [87] discussed
fracture behaviour in terms of the stored elastic strain
energy and the elastic energy release rate during crack
extension. Useful fracture mechanics analyses have
also been carried out by Lawn and Marshall [88] both
for thermally and chemically strengthened glasses.
The spontaneous failure or franging of surface modified glasses has been noted in some cases [87-91].
Bakioglu et al. [89] have shown, via a fracture mechanics analysis, that this effect could be caused by the
slow growth of subcritical surface cracks, but only if
the original critical flaw extends almost into the
tensile zone. Hsiao [90] noted that NiS inclusions in
thermally strengthened glasses could also lead to
spontaneous failure due to a phase transformation
involving a volume change occuring within the NiS as
Figure 9 F r a c t u r e b e h a v i o u r of
chemically s t r e n g t h e n e d l i t h i u m
m a g n e s i u m a l u m i n o - s i l i c a t e glass
discs ~ 1 . 2 m m thick as a function of the t r e a t m e n t time in
s o d i u m nitrate at 3 8 5 ~
after
[44]. F o r (a) t = 6 h , (b) t =
17h, (c) t = 2 5 h , (d) t = 4 9 h ,
(e) t = 6 5 h , (f) t = 97.5h.
41 89
40o
Figure 10 Modulus of rupture in four-
-'."
=
300
point bending of Corning 0319 chemically

strenghthened glass as a function of
scratch depth; after [96].
u_
200
0'
'
2'0
'
40
'
d0
'
8'o
'
100
'
Maximum scratch depth (tim)
a function of time. Chang and Chou [91] have also

examined the failure of thermally strengthened glass
at stresses less than their normal failure stress. In
addition, impact fracture behaviour has been studied
to some degree for these materials, particularly in
light of their role in aircraft transparencies [92-95].
Fracture studies of damaged chemically strengthened
glasses have not been widely reported. In one study,
however, Donald and Metcalfe [96] found that the
short-term strength of Corning 0319 chemically
strengthened glass containing surface scratches
decreased with increasing scratch depth, as illustrated
in Fig. 10, and Swain et al. [97] have also attempted
to predict the resistance of thermally strengthened
glasses to surface damage.
3.2. Bulk methods of improving strength and

toughness
In addition to surface methods for improving mechanical strength, a number of bulk methods have also
been utilized in attempts to improve the strength and
other properties of oxide glasses. A number of the bulk
methods are particularly aimed at improving the fracture toughness of brittle materials. General methods
for improving strength and/or toughness are outlined
below.
3.2. 1. Bulk crystallization to produce glassceramics

Glass-ceramics are generally significantly stronger
than their glassy counterparts, and exhibit many other
useful properties (e.g. [98-102]). The increased
strength is attributable to a combination of factors,
including microstructural limitation of the maximum
size of Griffith-type flaws present in the material. The
strength may also be high compared to many conventional ceramics, and this too is ascribed to microstructural features, including very small grain size
(significantly smaller than most ceramics produced
by conventional techniques, including sintering and
hot-pressing), and to the complete absence of internal
defects such as grain-boundary porosity and inclusions.
The strength of a glass-ceramic material normally
increases with increasing crystallization temperature,
reaching a maximum over a particular temperature
range, as illustrated in Fig. 11. On increasing the
crystallization temperature further, strength may then
start to decrease; this is due to a combination of
mechanisms, including grain growth or the formation
of crystalline phases of thermal expansion markedly
different to that of the matrix, which results in the
creation of high internal stresses [103-107]. Microhardness has also been noted to undergo similar
400
s
c
o
200
Figure 11 Strength of glass-ceramic materials as a
600
41 90
800
Crystallization temperature (~
1000
function of crystallization heat treatment temperature (a magnesium alumino-silicate glass-ceramic

nucleated with ZrO2; after [103], b lithium silicate
glass-ceramic nucleated with P205; after [104],
c lithium alumino-silicate glass-ceramic nucleated
with P205, after [105], d magnesium aluminosilicate glass-ceramic nucleated with P205; after
[106], e lithium zinc silicate glass-ceramic nucleated
with P205; after [107]).
behaviour (e.g. [108-11 l]. The toughness of a glassceramic is also greater than that of the parent glass,
due to the crack impeding properties of grain boundaries, but it is still very considerably lower than that of
a ductile metal. Hence, glass-ceramics, although generally mechanically stronger than their glassy counterparts, are still brittle solids, exhibiting classical brittle
behaviour and catastrophic failure characteristics.
3.2.2. Particle reinforcement
Strengthening by incorporation of a fine dispersion of
second-phase particles has long been utilized for
metallic systems. The particles may be produced in situ
by precipitation of a secondary phase (precipitation or
age hardening), or may be added directly by powder
metallurgy routes (dispersion strengthening). Dispersion strengthening and precipitation hardening
techniques may also be utilized for ceramic materials.
Since dislocation mobility is already extremely low in
ceramics (at ambient temperatures), and dislocations
as conventionally defined do not exist in glasses,
particle dispersions would not be expected to contribute very significantly to an increase in strength.
Significant strengthening has, nevertheless, been
observed for certain particle reinforced ceramic
systems, including glass reinforced with oxidized
nickel microspheres less than or equal to 25#m in
diameter [112], alumina containing molybdenum
dispersions of submicrometre particle size [113],
various materials containing zirconia dispersions
[114], and a number of glasses containing aluminium
particles [115, 116]. Later work on zirconia and
zirconia-containing ceramics has led to a new class
of strong, tough materials, the "transformationtoughened" ceramics, and this is discussed more fully
later. Increases in strength have also been reported for
glasses containing alumina and tungsten dispersions
[117, 118]. Conversely, decreases in strength have been
noted for many systems, depending on such factors as
dispersoid size and shape, and process parameters
[119-121].
The increase in strength noted for some of the
particle reinforced ceramic systems has been attributed to a number of different mechanisms. For
example, it has been suggested that strengthening may
be due to the fine dispersions limiting the size of
Griffith-type flaws, thereby raising the stress required
to initiate or propagate a crack. Strengthening may
also result from line-tension effects due to particles
initially pinning a propagating crack front and causing it to bow out between the obstacles [122, 123], in
a similar manner to that observed on an atomic scale
for dislocations in crystalline materials. In addition, if
particles of higher elastic modulus than that of the
matrix are employed, and strong particle-matrix
bonding exists, stress transfer from the matrix to the
reinforcing phase will occur, and some of the load will
be shared by the higher modulus particles. This means
that a higher stress can be achieved before the failure
strain of the matrix is reached, resulting in a stronger
material. In ductile particle reinforced systems ligament formation, in which cracks do not circumvent
the particles but evoke plastic deformation and rup-
ture of the particles instead, may lead to enhanced

fracture toughness [124]. In the case of crystalline
ceramics, including glass-ceramics, fine dispersions
may also inhibit grain growth during fabrication or
use, thereby giving a grain size strengthening effect.
Under suitable conditions, the effect of thermal
expansion mismatch, A~, between matrix and dispersoid may be used to advantage, and promote strengthening of the matrix: If the value of Ac~ is negative
(A0~
=
0%atrix - - 0%spersoid) , and the particle-matrix
interface can support the resulting radial tensile
stresses on cooling from the fabrication temperature,
the surrounding matrix will be subjected to tangential
compressive stresses, and the matrix may be strengthened. If, on the other hand, Ae is positive, induced
tangential tensile stresses may weaken the matrix and,
in the limiting case, localized failure of the matrix may
occur giving rise to a microcracked system. The resulting radial and tangential stresses er and at, respectively,
may be assessed using Selsing's formula [125]
- ar =
=
2cq
Ao~AT/{([1
Jr-
vml/2Em) -t-
([1
--
2Vp]/Ep)}
(7)
where, AT is the difference between fabrication and

ambient temperature, vmPoisson's ratio of the matrix,
Vp Poisson's ratio of the dispersed phase, Em Young's
modulus of the matrix, and Ep Young's modulus of
the dispersoid.
If Ac~ is positive, matrix microcracking is expected
when the value of cr exceeds the tensile strength of the
matrix, and microcracking has indeed been observed
in a number of cases. For example, in a glass-alumina
system of Ac~ ~ + 7 x 10 6~ I [126], and in glassthoria [120], MgO-W [127], and glass-glass composites of positive Ac~ [128]. On the other hand,
when Ac~ is negative, strengthening is predicted for
suitable bonded systems, and this has been observed in
practice for some glass-nickel systems, cited earlier
[112], and for a number of glass-alumina and glasszirconia composites [126]. If the value of Ac~ is too
negative, however, the particle-matrix interface or
the surrounding matrix may not be strong enough to
withstand the induced radial tensile stresses, and
decohesion will occur leading to a loss in strength, as
has been observed, for example, for a glass-nickel
system with Ae ~ -- 12 x 10-6~ 1 [121], and for a
glass-thoria composite of particle size greater than a
critical value [120].
In general, none of the above methods of particle
reinforcement lead to really practical increases in fracture toughness, although certain particle reinforced
aluminium-glass [115] and Fe-Ni-Co alloy-glass
[129] systems have been found to exhibit Ktc values in
the range 5 to 7 M P a m ~'2. These materials remain
prone to catastrophic failure, however, and fail at
relatively low fracture strains. The expanding field
of transformation strengthening or toughening is,
on the other hand, providing some useful materials,
exhibiting both high mechanical strength and
moderate toughness.
The phenomenon of transformation toughening has
4191
arisen due to the utilization of rapid stress-induced

structural transformations that involve molar volume
and/or shape changes. In principle, the concept should
be applicable to a number of structural transformations in various ceramic materials, but has so far only
been demonstrated effectively for zirconia. On cooling
zirconia to room temperature from around 1200 ~
the tetragonal to monoclinic phase transformation
would normally occur. If, however, the zirconia is
finely divided, or is incorporated into a matrix where
it is subjected to a constraining pressure (due to thermal expansion mismatch between matrix and particle,
with O~matrix > O~particle ) the zirconia can be maintained in
a metastable tetragonal form. When a dispersion of
such particles is present in a ceramic matrix it can act
to toughen the matrix, by impeding the propagation
of cracks through the matrix. The toughening mechanism is due to the volume change accompanying the
stress-induced transformation of the metastable
tetragonal particles to the monoclinic form in the
presence of a propagating crack. Tensile stresses a r e
generated in the vicinity of a propagating crack. If
these stresses are high enough, the constraint on the
adjacent particles is removed, and the zirconia transforms spontaneously by a diffusionless martensitictype reaction to the stable monoclinic form. The
accompanying volume change creates a compressive
stress field in the vicinity of the propagating crack, and
this acts to close up the crack tip, so hindering its
continued propagation.
The zirconia particle size is important in determining the resultant properties of a zirconia dispersion strengthened material. If the particles are too
small, they will not transform, whilst if they are too
large, they will transform spontaneously. There is,
therefore, a critical particle size range for a given
system, above or below which significant toughening
is not achieved. This critical size depends on a number
of factors, including the magnitude of the constraint
imposed by the matrix on the particles, and the chemical composition of the zirconia. It is possible, for
example, to alloy zirconia with other oxides that will
stabilize the cubic and tetragonal phases. For these
materials larger particle sizes can be tolerated without
spontaneous transformation to the monoclinic form
occurring.
The method of transformation toughening was first
successfully applied by Garvie et al. [130] in 1975, and
since then considerable effort has been devoted by
various research groups (e.g. Porter and Heuer [131,
132], Claussen [133, 134], Lange [135]) to finding new,
improved formulations. Some of the detailed work
has been reviewed recently by Heuer [132]. One of
the most widely studied systems has been alumina
containing tetragonal zirconia particles. The most
successful systems have been found to contain ~ 15%
zirconia of 1 to 2 #m particle size, and exhibit fracture
toughness values, K~c, of the order of 1 0 M P a m 1/2,
coupled with a flexural strength in excess of 400 MPa
[136]. Other matrices have also been utilized with
limited success, including zinc oxide, silicon carbide
and silicon nitride [137].
Investigations have also centred around the forma41 92
tion of partially stabilized zirconia.It has been found,

as mentioned earlier, that alloying zirconia with
certain other oxides, in particular MgO, CaO and
Y 2 0 3 , c a n stabilize the cubic or tetragonal zirconia
phase. If suitable compositions are quenched from a
high temperature, a single cubic phase material
results. These materials can then subsequently be aged
to "precipitation harden" the zirconia matrix by the
development of a fine dispersion of tetragonal zirconia
crystals in the cubic zirconia matrix. These particles
then act to toughen the matrix by the transformation
toughening mechanism. If, however, the materials are
overaged, the particles transform into the monoclinic
phase and little or no toughening results.
Although most of the work to date has centred on
transformation toughening of "conventional" crystalline ceramic materials, e.g. alumina and zirconia, a
limited amount of work has been reported on the
toughening of glasses and glass-ceramics, and it would
appear that there is considerable scope for further work
in this area. For example, Fagherazzi et al. [138]
reported the preparation of a Na20-A1203-ZrO2-SiO 2
glass-ceramic containing tetragonal and monoclinic
zirconia crystals, although the full potential of the
transformation toughening process does not seem to
have been exploited. Keefer and Michalske [139] have
reported the preparation of silicate glass-ceramics
containing up to around 30% zirconia, in which small
crystals of tetragonal Zirconia can be precipitated out
on careful heat treatment. Glass-ceramics based on
the 3 ZrO2-2 S i O 2 system have also been prepared
by Nogami and Tomozawa [140] using a sol-gel
technique. A Klc fracture toughness value of around
5 MPa m t/2 was observed for a sample containing a
fine dispersion of tetragonal zirconia crystals ~ 40 nm
in size. Mechanical property data for some of these
materials are summarized in Fig. 12 and Table VII.
3.2.3. Fibre reinforcement
The use of high modulus fibres in a lower modulus

matrix is a well known and routinely practised technique for increasing the strength of, for example,
organic polymeric materials. The general theory of
fibre reinforcement indicates that significant strengthening will only occur, however, if the elastic modulus
of the fibres is considerably greater than tl4at of the
matrix, and if tensile stresses can be transmitted to the
fibres. If fibres of lower modulus are employed, the
ultimate failure stress will be reduced because the
matrix, rather than the fibres, will carry a greater
proportion of the applied load.
Stresses may be transmitted to the fibres by plastic
or elastic deformation of the matrix. In the absence of
internal stresses, the strength, ac of a continuous fibre
reinforced composite may be estimated by the simple
rule of mixtures criterion, assuming the strains in each
component are equal, as
(9"c
~-
O'fu g f -~- o m V m
(8)
or as
O"c =
O'f g f -~- a r n u r m
(9)
where afu is the ultimate failure strength of the fibre, Vf
the volume fraction of fibre, (3m the matrix stress
Figurel2 Normalized
strength of particlereinforced glass and glass-ceramic matrix composites as a function of dispersoid concentration:
(a sodium borosilicate glass + 3 to I3 ~m diameter
tungsten microspheres; after [118], b sodium
calcium silicate glass + ~<40#m aluminiumalloy microspheres; after [116], c sodium calcium
silicate glass + 30/~m diameter oxidized nickel
microspheres; after [112], d sodium borosilicate
glass + 15/~m diameter alumina particles; after
[117], e as a but containing 50#m diameter microspheres, f as d but containing 60/~m diameter
particles, g glass + zirconia particles; after [119],
h glass + 74 to 105#m diameter nickel microspheres; after [121].)
j J
o
Z
2'o
Volume % dispersoid
~o
4'o
at the fibre failure strain, Vm the volume fraction

of the matrix phase, ~rr the fibre stress corresponding
to the matrix failure strain, and Crmuthe ultimate
failure strength of the matrix, depending on which
component fails first, and assuming that composite
failure occurs immediately following failure of one of
the components.
For discontinuous reinforcement, the ultimate
strength of a given fibre can only be utilized if it lies
parallel to the tensile axis, and if its length exceeds a
critical length, defined as the minimum fibre length in
the composite which can just be loaded to its failure
stress. If randomly orientated fibres are employed, the
proportion of fibres capable of being loaded to their
fracture stress will be reduced, and hence the ultimate
strength of the composite will be lower than that of
an equivalent aligned system. Suitable "efficiency"
factors have been derived to take into account the
effect of fibre orientation and length, although a
knowledge of the critical length is required in order to

predict the length efficiency factor (e.g. Krenchel [141],
Laws [142]).
On the basis of the above general theory of fibre
reinforce-ment, there would seem to be only very
limited scope for strengthening a high modulus brittle
ceramic material, which exhibits neither plastic flow
nor extensive elastic deformation, by the fibre reinforcement mechanism.
The incorporation of fibres into relatively high
modulus ceramic materials was, however, originally
carried out specifically in order to increase the fracture
toughness, rather than the strength, of these brittle
materials. The early aim was, therefore, to produce a
composite material which combined the advantages of
a ceramic, e.g. high strength at elevated temperatures,
low density, high thermal stability and oxidation
resistance, high mechanical hardness and abrasion
resistance, and good electrical insulating properties
T A B L E V I 1 Properties of particle-reinforced glass and glass-ceramic matrix composites

Matrix
SBS glass
SBS glass
SBS glass
SBS glass
SBS glass
LSAS glass
SBS glass
SAS glass
C7052
ZS glass-ceramic
SCS glass
SCS glass
Bioactive
SCPS glass
Type
A1203
AI2Q
AI203
W
W
AI
Ni
Ni
Fe-Ni Co
alloy (kovar)
ZrO 2
A1
A1 alloy
AI alloy
Particle
Diameter
(gm)
Volume
fraction
Composite flexural strength

Absolute
(MPa)
Normalized
60
21
15
20
3-13
100 130
25
25
44 75
0.475
0.475
0.423
0.50
0.50
0.20
0.20
0.20
0.68
117
163
174
57
155
1.16
1.40
[ .49
1.16
2.72
83
106
1.56
1.56
0.04
53
42
40
0.40
0.40
0.40
Reference
Fracture
toughness
(MPa m 1'2)
7.0
5.5
1 I8
161
86
1.69
2.18
1.51
4.8
-
[117]
[117]
[117]
[118]
[118]
[115]
[I 12]
[112]
[129]
[140]
[I 16]
[116]
[116]
SBS--sodium borosilicate;
LSAS lithium sodium alumino-silicate;
SAS--sodium alumino-silicate;
C7052--Corning Code 7052 borosilicate sealing glass;
ZS--zirconium silicate; SCS--sodium calcium silicate; SCPS--sodium calcium phospho-silicate
41 93
etc, with a high resistance to catastrophic failure. It

was the lack of toughness that early work tried to
improve upon, by addition of fibres to act as barriers
to crack propagation, thereby preventing the familiar
catastrophic failure behaviour associated with these
materials.
Fibres are intuitively effective at providing barriers
to crack propagation and thereby increasing the fracture toughness of a material, although their precise
role is complex and difficult to describe or quantify
fully. The concept of fibre reinforcement is not new.
Straw, for example, has been used to produce reinforced building bricks for several thousand years, and
is still in use today in many parts of the world. It was
only in the 1960s, on the other hand, that serious
attempts were made to produce high quality, high
performance fibre reinforced ceramics, driven by the
immense potential of these materials in aerospacetype applications, and in particular including gas
turbine components.
There are a number of mechanisms by which the
total work of fracture may be raised. One is to increase
the effective crack propagation area by providing
planes of weakness within the material, in a direction
parallel to the tensile axis, along which a propagating
crack may be deflected, thereby effectively blunting
crack tips. Aveston has shown [143] that very high
values for the work of fracture are possible by this
mechanism. If, on the other hand, the fibre-matrix
bond is sufficiently strong, a crack may propagate
relatively unimpeded through the composite, and the
work of fracture will be low. If discontinuous fibres
are employed of length, l, less than the critical length,
lc, the fibres will not be loaded to their fracture stress
and they must be withdrawn from the matrix as the
fracture planes separate. It has been shown by Cottrell
[144], that the contribution due to pull-out can be very
significant, and the maximum value 7P. . . . is obtained
f o r l = lc, when
YPmax =
r f "c l~/lZr
example, due to chemical reaction with the matrix. In

these cases high values for fracture toughness may still
be achieved, nevertheless, by the pull-out mechanism.
If the fibres are randomly orientated, many fibres
will cross the fracture plane obliquely and will be
subjected to a bending moment. In the case of brittle
fibres, this will lower the applied tensile stress required
to fracture the fibres. For ductile fibres, on the other
hand, plastic bending may accommodate the extra
strain at the convex portion of the fibre during withdrawal from the matrix, and may contribute significantly to the total work of fracture. If fibres make a
small enough angle to the fracture plane, they will be
unable to pull-out, at least initially, and will either fail
in shear, or will break through the matrix. Fragmentation of the matrix during failure by such a mechanism further increases the work required, although this
contributes only a very small increase relative to the
other mechanisms. The fracture surface of a glassceramic/discontinuous nickel filament composite is
shown in Fig. 13, illustrating the mechanisms of fibre
pull-out, plastic bending and rupture of fibres, and
matrix fragmentation [145].
The thermal expansion difference between fibre and
matrix is a very important parameter in any composite
system, since it determines the residual stress-strain
(10)
where ~ is the interfacial shear stress resisting pull-out,

and r the fibre radius. If the fibre length l > lc, some
pull-out is still observed, since many of the fibres will
intersect the crack plane within a distance lc/2, and
hence a fraction lc/l of the fibres will not be loaded to
their fracture stress, and must pull-out as the planes
separate. The fact that pull-out effects have also been
observed for many continuous brittle-fibre systems has
been attributed to the statistical distribution of strength
along these fibres, allowing fracture of many fibres to
occur at positions away from the fracture plane.
In the case of ductile reinforcing fibres, a large
contribution to the fracture toughness of the composite may be provided by plastic flow and rupture of the
fibres. For fibres of length, l > lc, the contribution, 7r,
due to fibre rupture will be
7r =
V~ ~,- d -- tc/1)
(11)
where 7r is the fibre rupture energy. Unfortunately, the

high work of fracture of metals cannot always be
utilized to full advantage because wire embrittlement
may occur for some metals during fabrication; for
41 9 4
Figure 13 Fracture surfaces of a glass-ceramic composite containing 40 vol % discontinuous nickel filaments 125 #m diameter showing the effects of fibre pull-out, fibre deformation-plastic failure,
and matrix fragmentation; after [145].
distributions after fabrication, and can significantly

influence the resultant mechanical properties. The
residual thermal strain, Sin, experienced by the matrix
when unidirectionally aligned fibres of different thermal expansions are employed may be assessed semiquantitatively, assuming interfacial decohesion does
not occur in the case of negative Ac~,using the following relationship
e~ = ErVrA~AT/Eo
(12)
where E r is Young's modulus of the fibre, AT the

difference between the fabrication temperature and
the ambient temperature, and Ec Young's modulus of
the composite.
If the value of em is positive, the matrix will be in a
state of tension; this will give rise to a microcracked
system ifs m > era, where smis the normal failure strain
of the unreinforced matrix. Conversely, if Sm is negative, the matrix will be in compression.
For a given system, the failure stress of the composite, oc, may be given by
O'c =
Ec(~;m -- ~;m)
(13)
Assuming that the simple rule of mixtures criterion is

obeyed
Ec
E m Vm +
EfV~
(14)
The corresponding strain experienced by the fibres is

given by
~f
-- E m g m A ~ x A r / E c
(15)
The optimum strength for a given system corresponds

to the situation where the failure stress of the fibre is
reached simultaneously with the failure strain of the
matrix. More detailed fibre reinforcement theories
and concepts, applied to general composite materials,
can be found elsewhere, e.g. Hale and Kelly [146].
A summary of the contrasting factors contributing
to the strength and toughness of glass and glassceramic matrix composites is given in Table VIII.
Reinforcement of ceramic materials by fibres produced some very encouraging results in the 1960s and
1970s in terms of providing composites with high
values for work of fracture. Materials included, for
example, a lithium zinc silicate glass-ceramic reinforced by nickel with a work of fracture approaching
17 kJ m 2 [145]; a cordierite glass-ceramic containing
SiC filaments with a work of fracture of around
20kJm -2 [143]; and a number of carbon fibre reinforced glass and glass-ceramic composites exhibiting
works of fracture of up to around 10 kJ m 2 [ 147-151].
Much of this early work, up until around the mid
1970s, has been reviewed in detail by Donald and
McMillan [152].
In general, early work concerned with the reinforcement of ceramics by metal or ceramic filaments,
although providing substantial improvements in fracture toughness, particularly when using metal reinforcement, unfortunately led to little improvement, or
even a reduction, in the mechanical strength [152].
This was because the relatively large diameter fibres
employed, normally greater than 50/~m and often as
large as 200#m, acted as stress concentration sites
T A B L E V I I I Factors promoting strength a n d toughness in

glass and glass-ceramic matrix composites
(a) FO R H I G H S T R E N G T H
I. Modulus of fibre > modulus of matrix
2. Continuous, aligned fibres for maximum strength in one
direction
3. Use small diameter fibres (e.g. 0 < 25/1m)
4. Controlled bonding between fibre and matrix
(b) FO R H I G H T O U G H N E S S
1. Allow local microcracking of the matrix to reduce stressconcentration effects, e.g. by providing weak fibre-matrix
bonding
2. Allow extensive fibre pull-out
3. Use ductile fibres
which weakened the matrix; toughness was therefore

increased at the expense of strength. Notable exceptions where improvements in strength were achieved
included some glasses reinforced by tungsten filaments
[153, 154]. Simultaneous increases in toughness and
strength were also observed by Phillips and co-workers
[147-151], for unidirectionally aligned carbon fibre
reinforced glasses and glass-ceramics, and by Levitt
[155] for carbon fibre reinforced lithium aluminosilicate ceramics containing a glassy phase. Carbon
fibres were the only high quality, small diameter
(~8/~m) fibres available at that time. Maximum
strengths up to around 900MPa were achieved for
these systems, with works of fracture approaching
10 kJ m 2. Significant strengthening was possible due
to a combination of factors, including the influence of
small fibre size and the large difference in elastic
modulus between the carbon fibres and the relatively
low modulus glasses and glass-ceramics (relative to
many other ceramic materials; for example, alumina
and silicon nitride). For systems of this type, where
high modulus, small diameter fibres are incorporated
into a low modulus matrix, Aveston et al. [156] have
shown that the cracking strain of the matrix, era, (and
hence fracture strength) can be increased, according to
the relationship
~m
(12zTmEfVZ/EcE2mrVm)El3
(16)
where r is the frictional bond strength, 7m the work of

fracture of the matrix, and r the fibre radius.
Although the carbon fibre composites were quite
impressive with respect to high strength and toughness
at ambient temperatures, they were not suitable for
use at elevated temperatures above 400 ~ in oxidizing
environments. This was because carbon is readily
oxidized, and protection by the matrix is limited due
to the phenomenon of matrix microcracking that
occurs under stress, this allowing ingress of oxygen to
the fibres.
Another notable exception, where high toughness
coupled with high strength was achieved, was for a
cordierite glass-ceramic reinforced by relatively large
diameter ( ~ 100 ktm) SiC fibres [143]. This system was
unique in that the strengthening effect was attributable
to the influence of the higher thermal expansion fibres,
which were strongly bonded to the matrix, placing the
matrix in compression on cooling from the fabrication
temperature. As the high strength relied on a thermal
4195
expansion difference, however, the strength decreased

with increasing temperature. In addition, the fibre was
expensive, being produced by chemical vapour deposition of SiC onto a 12 #m diameter tungsten filament
substrate.
After the initial great enthusiasm shown in fibre
reinforced ceramic composites in the 1960s and early
1970s, interest declined sharply by the mid-to-late
1970s, because of the difficulties outlined above, which
were coupled with the general lack of suitable inexpensive small diameter, high strength and high modulus
fibres capable of promoting simultaneous increases in
both strength and toughness.
Over the last few years, however, a remarkable
revival in interest has taken place in the field of
ceramic matrix composites due to the advent of a
whole new generation of high-performance, small
diameter fibres, available commercially at realistic
prices. There is now available, for example, a diverse
range of ceramic fibres with diameters down to a few
micrometres, including silicon carbide and alumina, in
addition to coated carbon fibres with improved
oxidation resistance. A selection of commercially

available fibres is summarized in Table IX.
Reinforcement o f glasses and glass-ceramics still
offers the best scope for accomplishing simultaneous
increases in both strength and toughness, because of
the relatively low elastic modulus of these materials.
For example, lithium alumino-silicate glasses and
glass-ceramics reinforced by silicon carbide fibres
with strengths in the range ~600 to 1550MPa and
fracture toughnesses up to K~c ,-~ 27.5MPam ~/2 have
been reported [157-160]. The value of strength of
1550MPa coupled with a fracture toughness of
27.5 MPa m 1/2is, in particular, an astonishing achievement for a ceramic based material [157]. Other work
on silicate glasses and glass-ceramics reinforced with
SiC fibres has been reported [161-169]. In particular,
recent work by Dawson et al. [169] has indicated
that pyrex glass reinforced by SiC fibres can yield
composites which exhibit strengths up to ~ 1250 MPa,
coupled with values for work of fracture of the order
of 50 kJ m 2. Oxynitride glass-ceramics have also been
investigated, including a Ba-Si-AI-O-N system
T A B LE I X Fibre availability and properties

Fibre
(a) Ceramic fibres

Alumina
Alumina
Alumino-silicate
Aluminoborosilicate
Zirconia-silicate
Silicon carbide
Diameter
(~tm)
Trade name
Manufacturer
20
2-4
2-4
10
FP-Alumina
RF-Saffil
Fibermax
Nextel 440
DuPont
ICI
Carborundum
3M
14
10-15
Nextel Zl 1
Nicalon
3M
Nippon
Carbon Co.
Carborundum
Boron nitride
Thermal
expansion
(10 6C-1)
8.5
8.5
-
3.1
-
Maximum use
Temperature
(~
1370
1400
1650
1425
1 0 0 0
1000
2480
(760 in oxidizing
environment)
< 1100
< 400 in oxidizing
environment
< 1000
Tensile
strength
(MPa)
Modulus
(GPa)
1380
1035
1725
380
300
1310
2750
75
200
1990
420
205-240
Fused silica
Carbon
1-40
7-8
Various
Various
Various
Metal-coated
carbon
Silicon carbide
Boron
Cycom
Cyanamid
SCS
AVCO
AVCO
4.9
4.5
< 1000
-
Various
Experimental
only
Bekaert
5-20
5-20
< 1000
< 1500
400-4000
<4400
< 400
< 400
~9-20
< 1000
< 1500
< 220
(b) Metal filaments

Die-drawn metals
Taylor-wire
Wollaston metal
filament tows
(316 stainless
steel,
Inconel 601,
Hastelloy X, Ni,
Ti)
(c) Whiskers
SiC whisker
SiC whisker
SiC whisker
SiC whisker
Si3 N4 whisker
4196
140
100-140
> 25
1-I00
4-22
Bekinox
0.6
SC-9
(length, 10-80/~m)
0.05-0.2
SCW
(length, 10-40/xm)
0.1-0.5
Tokamax
(length, 50-200 #m)
3-11
VLS
(length, up to 100mm)
0.5
0 (axial)
8 (radial)
-
1500-3000
70
250-400
3000
235
3800
3600
400
400
ARCO
--
6890
689
Tateho
--
2100
480
< 1600
< 14000
< 700
< 1500
1370
380
Tokai Carbon
LANL
Tateho
Figure 14 Lithium zinc silicate glass-ceramic/stainless steel filament reinforced composites, after [187]. (a) micrograph of sectioned composite
containing 40 vol % continuous, aligned 22 #m diameter filaments; (b) fracture surface of similar 20 vol %, 22/~m composite; (c) fracture
surface of similar 20 vol %, 8 #m composite.
reinforced with SiC fibres, with a bending strength of

275MPa [170]. Various glass-alumina fibre systems
have also been examined [171], with strengths in
bending to around 300MPa and fracture toughness
values up to 4 . 0 M P a m 1/2. It was noted that the
highest strengths were obtained for systems of matched
thermal expansion. (That optimum strengths are
obtained for matched thermal expansion systems has
been emphasized previously by Donald and McMillan
[152]. They compared the range of strengths reported
for various ceramic-matrix systems with the theor-
etical values, and showed that the best results were

generally obtained when A~ tends towards zero.) A
revival of interest in carbon fibre reinforced glasses
and glass-ceramics is also underway [172-177], driven
by potential intermediate temperature applications for
these materials. A number of reviews are available
which report in more detail some of the work up to the
early 1980s on this new generation of materials [178186].
The majority of the recent work has concentrated
on using the new ceramic fibres. Metal reinforcement,
41 97
on the other hand, does offer a number of distinct

advantages over ceramic fibres, including a lower
susceptibility to damage and degradation during
composite fabrication. At the present time, unfortunately, conventional die-drawn metal filaments are
either of too large a diameter to promote significant
strength increases, due to the reasons outlined earlier,
or materials that are available in smaller sizes are
prohibitively expensive. Metal filament tows,
prepared most probably by the Wollaston route, with
individual filament diameters down to around 4 #m
are, on the other hand, available commercially at the
present time, and these do offer scope for composite
fabrication, although they are only available in a very
limited number of materials at the present time. A
number of these materials have been used for reinforcing glass-ceramic matrices by Donald and co-workers
[187]. Some micrographs are shown in Fig. 14. In
addition, a revival of interest is taking place in the
production of small diameter metal filaments in a very
wide range of metals and alloys directly from the
melt, using the Taylor-wire route, as reviewed recently
by Donald [188]. In this intrinsically inexpensive
process, a fine glass-encapsulated metal filament
is produced by the drawing down of a glass tube
containing the molten metal or alloy. With suitable
control of the process parameters, it is feasible to
produce metal filaments with diameters in the range
1 to 100#m, with glass coating thicknesses of
around 2 to 30 ~tm. The fact that the metal filament
is produced directly with a glass coating offers the
possibility of preparing glass matrix composites directly
from suitable Taylor-wires; and the viability of this
route for preparing composites has indeed been shown
recently [189]. A typical micrograph of a sectioned
glass-matrix composite reinforced by copper filaments
of average diameter of about 6 #m, which has been
prepared directly from Taylor-wire, is shown in
Fig. 15. In principle, it should be possible to prepare
more practical small diameter, high-strength, oxidation

resistant filaments of the superalloy variety using the
Taylor-wire route, although little work has been
reported in this area. It should also be feasible to
employ a glass-ceramic precursor coating, from which
a more practical and more refractory glass-ceramic
composite could be prepared directly, by suitable
choice of heat-treatment schedule after consolidating
the glass matrix. It is possible that very useful composites could be prepared using such materials.
The requirements of a relatively strong bond
between fibre and matrix to promote strength and, for
brittle fibre systems, a weak interface to promote
toughness, can be reconciled to some degree in duplex
fibre systems [190-192]. In these materials, a duplex
fibre element is employed consisting of an outer sheath
which is bonded strongly to the matrix, and an inner
core element which is less strongly, e.g. frictionally,
bonded. If failure of the sheath occurs, pull-out effects
are still observed between the sheath and inner core
element.
The failure behaviour of a ceramic composite is
interesting. At low volume fractions of fibre, the failure
mode is very similar to that obtained for a monolithic
homogeneous ceramic. At higher concentrations,
however, a load-displacement curve is obtained
which, in many respects, models that of a ductile
metal, as illustrated in Fig. 16. A linear region is noted
initially at low loads, but at some point deviation from
linearity occurs due to the onset of significant matrix
microcracking. Above this "proportional limit" a
non-linear regime is then encountered as progressive
and multiple microcracking of the matrix continues,
and fibres undergo the effects of pull-out and/or
deformation. Finally, on further deformation, the
toad decreases, although total strains of several %
may be achieved before the load finally approaches the
zero level.
Specific data for mechanical strength and toughness
Figure 15 Glass-matrix composite prepared from Taylor-wire; after [189]. (a) micrograph of sectioned sample, (b) fracture surface of
composite.
41 98
for a variety of fibre reinforced glass and glassceramic matrix systems are summarized in Table X
and Figs 17 and 18.
3.2.4. Whisker reinforcement

Whiskers are small, single crystal filaments, normally
about 0.1 to 10/tin in diameter with aspect ratios of
the order of 10 to 100, although higher aspect ratios
have been achieved, with some reports of whiskers
up to 100ram in length (see e.g. [193]). In their
as-produced state, whiskers can exhibit extremely high
strengths, with values approaching the theoretical
limit. Their properties are, however, very susceptible
to mechanical damage and therefore, in practice,
maximum useful strengths are generally limited to
less than 1500 MPa.
A summary of work on ceramic-matrix composites
reinforced by ceramic whiskers has been given by
Bracke et al. [194]. More recent work involving SiC
whiskers includes that of Wei and Becher [195] using
a range of ceramic matrices, Chokshi and Porter [196]
using an alumina matrix, and Claussen et al. [197] and
Becher and Tiegs [198] using zirconia. There has been
little reported work specifically on whisker reinforcement of oxide glasses and glass-ceramics, although
Gadkaree and Chyung [199] have reported some work
in this area. Glasses studied in their investigation
included an alumino-silicate composition, together
with soda-lime-silica and borosilicate glasses. Bariumstuffed cordierite and barium-osumilite glass-ceramics
were also used. The resultant composite strength
was found to be dependent on a number of factors,
including whisker-matrix thermal expansion mismatch and matrix chemical composition. It was noted
that poor composite performance was obtained for
either a positive or negative expansion mismatch.
Work on glass and glass-ceramic materials reinforced
by SiC whiskers has also been reported by Layden and
Prewo [200]. Data are summarized in Table X.
3.2.5. Directionally sofidified or crystallized
structures
A number of fibre reinforced composite systems have
been prepared by unidirectional solidification. Some
of the materials that have been produced by this route
have been reviewed by Donald and McMillan [152].

Related unidirectional crystallization techniques have
also been applied to a limited number of glassy
materials in order to produce aligned fibre reinforced
glass-ceramic systems. For example, Atkinson and
McMillan [201] produced preferred orientation of
needle-like lithium disilicate crystals in Li20-SiO 2P205 glasses by extruding partially crystallized material
through a die at 800 to 880 ~ Abe et al. [202] have
also reported the formation of calcium phosphate
glass-ceramics reinforced with fi-Ca(PO3)2 fibres by
unidirectional crystallization of a glass employing a
temperature gradient furnace.
3.2.6. Combined methods

A combination of particle and fibre or whisker reinforcement has been utilized in a limited number of
cases; for example, by Becher and Tiegs [198] for a
mullite matrix. The simultaneous use of more than
one reinforcing mechanism is applicable in principle
to most systems, including glasses and glass-ceramics,
although little work has been reported in this area.
3.2.7. Glass laminates
Oxide glasses are employed in a number of important

composite transparencies, including automobile and
aircraft windshields, and bullet-proof glass [203,204].
Bullet-resistant glass, for example, consists of alternate layers of a transparent thermoplastic and an
oxide glass to form a composite sandwich, and is
normally composed of three discrete sections; these
include a forward facing impact section, a middle
transition region, and a final energy absorbing layer.
The first section usually consists of a polyvinyl acetal
material, for example polyvinyl butyral, sandwiched
between two relatively thick layers of glass. The
transition region may consist of a relatively thick
polyurethane section, whilst the final energy absorbing layer consists of a sheet of polycarbonate material.
The total number and thicknesses of laminations can
be varied, depending on the precise application in
question. For applications requiring very strong
but lightweight structures, thermal or chemically
strengthened glasses may be used in the laminate.
600
400
g
o
200
L
4
Crosshead displacement (mm)
Figure 16 Load-displacement curve illustrating

the fracture behaviour of a 40vol%, 22/2m
diameter metal filament reinforced lithium zinc
silicate glass-ceramic; after [187].
4199
0
0
SiC
SiC
0.41 0.47
0.40
0.42
0.55
0.40
0.40
0.49
0.6
0.3 (wt)
0.6
0.30 (wt)
(10 80llm long)
0.6
0.30 (wt)
22
AI203
steel
SiC
10 15
10-15
10-15
7-8
7-8
SiC
SiC
SiC
C
C
0.50
0.50
10-15
10-15
0.40
0.46
0.43
0.30
7 8
7 8
7 8
100
7 8
SiC
*Tested in air at 1000~

LZS--lithium zinc silicate
LAS--lithJum alumino-silicate
MAS--magnesium alumino-silicate
SBS--sodium borosilicate
(c) Whisker re#~/orcement

Corning
1723 glass
Barium osumilite
glass-ceramic
Barium stuffed
cordierite glass
ceramic
Aluminasilicate glass
MAS glass-ceramic
LAS glass-ceramic
Pyrex glass
Pyrex glass
SBS glass
+ I% Nb205
Various glasses
LZS glass-ceramic
stainless
(b) Continuous, aligned reinJorcement

Pyrex glass
C
LAS glass-ceramic
C
Pyrex glass
C
Cordierite
SiC
glass ceramic
LAS glass-ceramic
SiC
Pyrex glass
0.30
0.15
0.08
0.40
0.20
400
*300
358
"310
338
200 311
500
~3
3.5
5.4
~12
~12
480
930
1250
826
971
~4
700
~7
~3
~8
-~6
5.7
1.2
0.6
' 0.8
0.6
*830
1380
680
575
785
680
275
85
130
180
60
Normalized
Absolute
(MPa)
Volume
fraction
Type
Diameter
(Ira1)
Composite flexural strength
Fibre
(a) Discontinuous, randomly aligned re#1fi)rcement

Fused silica
W
50
Pyrex glass
W
50
LZS glass-ceramic
Ni
125
Matrix
T A B L E X Properlies of fibre- and whisker-reinforced glass and glass-ceramic matrix
50
50
20.0
3.4
10.3
5.0
15.7
0.4
--
Work of
fracture
(kJm 2)
4.5
7.2
4.5
3.5
3.4
2.6-4.0
26.3
17.0
25.0
27.5
Fracture
toughness
( M P a m I 3)
186
156
142
124
120
193
173
100
118
169
Elastic
modulus
(GPa)
[199]
[199]
[I99J
[199]
{1991
[172]
[187]
[I 59]
[159]
[169]
[1771
[I77]
[1581
[1581
[1571
[147]
[1791
[172]
[143]
[I54]
[1531
[1.451
[1451
[I48]
Reference
a/
12
Figure 17 Normalized strength of fibre-reinforced

glass and glass-ceramic composites as a function
of fibre content. (a lithium alumino-silicate glassceramic + silicon carbide fibre; after [157],
b pyrex borosilicate glass + carbon fibre; after
[177], c silica + tungsten filaments; after [154],
d pyrex borosilicate glass + carbon fibre; after
[149], e cordierite glass-ceramic + 100#m diameter silicon carbide fibre; after [143], f pyrex
borosilicate glass + carbon fibre; after [147],
g lithium zinc silicate glass-ceramic + 125/~m
diameter nickel filaments, after [145], h pyrex
borosilicate glass + carbon fibre; after [148].)
~I
o
Z
g
20'
'
Volume % fibre
4"0
6'0
'
3.3. Preparation of ceramic composites

Ceramic matrix composites can be prepared by a
number of routes. Probably the most widespread is
hot-pressing. Randomly orientated fibre and whisker
reinforced systems may be prepared by tumbling or
high speed blending of fibre-matrix powder mixtures
followed by loading into a suitable die and pressing at
elevated temperature, usually in v a c u o or inert atmosphere. Densities approaching 100% of theoretical
can readily be achieved by this route, particularly
when employing glass or precursor glass-ceramic
matrix materials. Graphite dies and punches are often
employed; this is due to the ease of machining this
material; coupled with its low thermal expansion and
chemical inertness, although pressing must be carried
out in an inert atmosphere because of its poor oxidation resistance. Dies fabricated from stronger and
more abrasion resistant materials, including stainless
steel, silicon carbide and TZM (Ti-Zr-Mo) alloy,
have been employed, but are less versatile than
graphite. During hot pressing, some alignment of the
fibres generally occurs so that they tend to lie in
the plane of pressing, but within this plane they are
randomly orientated. Unidirectionally aligned fibre
systems have been prepared by several techniques,
'E
including passing continuous fibres through a slurry

of matrix powder and binder [205], and extruding
fibre-matrix slurries containing ammonium alginate
into an acid fixing bath which yields a tape of aligned
fibres [206]. In both cases, the tapes can be cut into
desired lengths or shapes and hot pressed after a suitable heat treatment to remove the binder. Sol-gel
techniques may also be employed to produce particle
or fibre reinforced composites (e.g. [140, 207]). Meltinfiltration has been employed with some success (e.g.
[208]). Further details covering fabrication of ceramic
composites in general are given elsewhere [179, 209,
210].
4. Comparison of methods for

improving the mechanical properties
of oxide glasses and their
limitations
Surface modification of glass, in the form of application of a thin protective coating to pristine or etched
glass surfaces, has found widescale use in the glass
industry for both bulk materials including containers,
and for materials in the form of fibres [76-81]. Any
improvement in properties obtained using many of
these techniques is, however, easily lost through the
a/
/
"6
dz
lO
/
Figure 18 Work of fracture of fibre-reinforced
2'0
Volume % fibre
4O
6b
glass and glass-ceramic matrix composites as a

function of fibre content. (a cordierite glassceramic + 100#m diameter silicon carbide fibre;
after [143], b silica + carbon fibre; after [143],
c lithium zinc silicate glassZceramic + 125#m
diameter nickel filaments; after [145], d lithium
alumino-silicate glass-ceramic + carbon fibre;
after [179], e and f pyrex borosilicate glass +
carbon fibre; after [147, 149], g pyrex borosilicate
glass + discontinuous carbon fibre; after [148].)
4201
introduction of relatively minor damage, particularly

surface scratches more than a few micrometres in
depth - although more recently thicker polymer
coatings have been applied to glass surfaces, and
these can provide quite significant improvement to
accidental damage.
Very large improvements in strength and resistance
to quite serious damage can, on the other hand, be
imparted through thermal or chemical strengthening
treatments that rely on producing a relatively thick
compressive layer on the surface of glass articles.
Thermal strengthening is, due to the nature of the
process, limited to glass articles of relatively simple
shape and of thickness greater than 1.5 to 2 ram. In
addition, the magnitude of the compressive stress
obtainable is restricted by such factors as the thermal
expansion characteristics of the glass and the maximum cooling rate achievable in practice. As the glass is
quenched from a temperature higher than ~ , viscous
deformation and distortion of the glass may also be
difficult to avoid completely, an important constraint
when designs requiring very close dimensional tolerances are required. Due to the parabolic nature of
the stress distribution, once a glass article has been
thermally treated to produce a compressive layer of
practical magnitude, it cannot be further machined by
cutting or drilling, due to the high internal tension.
This is in contrast to chemically strengthened glasses
where a much flatter stress profile is found and where,
depending on the compressive layer thickness in relation to the total thickness of the treated glass article,
a high compressive stress can be generated without
producing a high internal tension. If the internal
tension is low enough it is then quite feasible to
machine a treated glass article by cutting and drilling.
At the other extreme it is also possible, using chemical
treatments, to induce very high internal tensile stresses,
particularly in relatively thin samples; and this is the
basis for the so-called frangible or command-break
glasses that can be fractured predictably and precisely
by application of a small amount of local energy just
sufficient to trigger the break. Chemically strengthened glasses are unfortunately more expensive than
their thermally strengthened equivalents. This is due
in part to the high cost involved in maintaining a
molten salt bath facility; for example, salts have to be
changed at regular intervals due to a build up of
contamination products which reduce the efficiency of
the process, and very careful and precise temperature
control is required. Degradation in the properties of
thermally and chemically strengthened glasses occurs
at elevated temperatures due to stress relaxation
effects. In addition, in the case of chemical strengthening, ionic diffusion also leads to a reduction in
strength, particularly at temperatures approachi~ag or
exceeding the treatment temperature, although, under
prolonged heating, glass degradation may also occur
at temperatures significantly less than this, as illustrated
in Fig. 5.
Other methods for imparting surface compressive
stresses, including cladding with a lower thermal
expansion coating and surface crystallization, are generally less versatile than thermal or chemical treat4202
ments. In addition, they may also suffer from surface

spalling effects due to the very sharp transition
obtained between compression and tension; this is
particularly severe in the case of surface crystallization.
In the case of glass-ceramic materials, control or
tailoring of the thermal expansion characteristics is
one of their major attributes. This control can be
achieved by careful selection of the precursor glass
composition and the heat-treatment schedule. By
control of the expansion, it is possible to tailor
materials for specific applications, including ceramicto-metal and ceramic-to-ceramic seal devices, and
composite materials. Many other properties, including
those of an electrical, chemical or physical nature,
can also be tailored according to the initial glass
composition and heat-treatment schedule adopted. In
addition, conventional, relatively low temperature
glass-forming techniques can be employed to produce
articles of complex design. Controlled heat treatment
may then be carried out in order to provide superior
glass-ceramic components possessing higher strength
and toughness, higher temperature stability and better
abrasion resistance than the equivalent glassy counterpart. A glass-ceramic, although exhibiting higher
fracture toughness than the precursor glass still
behaves, however, as a highly brittle material, and
in particular is prone to Catastrophic failure. Only
composite systems can provide really significant practical improvements in the fracture toughness of glass
and glass-ceramic materials. Increased fracture
toughness can be achieved using particle dispersions,
in particular zirconia [138-140], but fibre reinforcement offers the most satisfactory means of substantially increasing fracture toughness, and in particular
preventing or minimizing the effects of catastrophic
failure, especially at elevated temperatures. In the
early work on fibre systems it was noted that toughness was generally increased at the expense of ultimate
strength, but later work using the new family of fibres
has shown that it is now possible to achieve simultaneous improvements both in strength and toughness. In terms of strength enhancement, the highest
strengths are found for unidirectionally aligned fibre
systems, tested with the fibres parallel to the tensile
axis. Fig. 19 shows the effect on strength of fibre
orientation for glass matrix carbon fibre systems. It is
clear that strength falls off dramatically for only small
variations in fibre orientation. As for all fibre reinforced materials, this factor has serious implications in
the design of suitable materials for practical applications where the distribution of stresses may vary
significantly from an idealistic unidirectional case.
For any composite system, a major area of concern
involves compatibility between the matrix and reinforcing phases. This is a particularly severe problem in
ceramic-based composites where high fabrication and
potential operating temperatures may be involved,
and where differences in thermal expansion characteristics between the constituent phases can lead to
severe problems of degradation, either during fabrication or in service, particularly when temperature
cycling is involved. Glass-ceramic materials offer an
80o[
Figure 19 Composite strength as a function of

fibre orientation. The broken curve represents
range noted for flexural strength of pyrex borosilicate glass-carbon fibre system; after [150]; the
full curve is for tensile strength of a similar
system; after [172].
~=~"
600It
I
400
cz
E
o
o
200
2J0
4()
610
Fibreorientation(deg)
important advantage over other matrix materials in

this respect because, in principle, the expansion
characteristics of the matrix can be tailored to match
or at least approximate those of the reinforcing phase.
In addition, less severe fabrication conditions, including lower hot-pressing temperatures and pressures,
and shorter pressing durations, are in general required
to produce fully dense components. This helps to
minimize deleterious chemical reactions between the
matrix and reinforcement, and leads to less damage of
the reinforcing phase during fabrication. A glassceramic is particularly advantageous because fully
dense composites can be made from the 'precursor
glass at relatively low temperatures. Further heat
treatment to crystallize the glass then produces a
mechanically stronger and more refractory matrix
phase with thermal expansion matched to that of the
reinforcement. Some materials, for example carbon
fibre reinforced glass and glass-ceramic composites,
suffer from the serious disadvantage that they cannot
be employed for prolonged periods under stress in an
oxidizing atmosphere at temperatures greater than
around 400~ due to degradation of the fibres, as
noted in the early work of Phillips and co-workers
[147, 205], and re-emphasized recently by Prewo and
Batt [176]. These materials are, however, perfectly
viable at lower operating temperatures [147, 175],
and may fill an important gap between the low
temperature polymeric composite systems and the
high temperature refractory metal and ceramic systems. In the case of SiC fibre reinforced materials, it
has been shown by Chaim and Heuer [163] for a
lithium aluminosilicate glass-ceramic matrix, that
fibre-matrix reaction, which may lead to a reduction
in ultimate strength, can occur during fabrication.
This is attributed to the formation of a reaction zone
between fibre and matrix which contains amorphous
carbon together with NbC microcrystals through
which cracks can propagate more readily. Degradation in the strength of similar materials during service
at elevated temperatures [160, 166] may be a direct
consequence of further matrix-fibre reaction. The
search for more thermally stable, high strength composite systems is u n f o r t u n a t e ~ f f i c u l t
by the
8'0
fact that some reaction between fibre and matrix is

usually required in order to promote the degree of
bonding required for high strength.
The high temperature properties of a selection of
glass and glass-ceramic matrix composites are summarized in Fig. 20. In general, these are short duration
tests, and holding at a given temperature for prolonged
periods of time may lead to degradation of the properties with time.
5. Applications and Future

Many of the materials described in this review have
found applications in a widescale and diverse range of
subject areas. On the other hand, current applications
for some of the materials, in particular the composites,
are very limited at the present time, and potential
applications must remain highly speculative. A number
of applications, both current and potential, are outlined below.
For example, surface modified glasses, particularly
thermally strengthened glasses, have found widescale
applications as transparencies in the automobile, aircraft and architectural fields. Other applications
include spectacle lenses, pressure or vacuum vessel
port windows, and containers and pipework in the
chemical industry. Chemically strengthened glass, on
the other hand, has found far less high production
usage, mainly as a result of the higher cost of these
materials in relation to their thermally strengthened
counterparts. It has been used in automobile, aircraft,
helicopter and spacecraft transparencies, spectacle
lenses, optical recording discs and as a frangible,
command-break material in some military applications, including air-to-ground missile launch tube
protective covers. In addition, Beauchamp has
proposed [211] that it may be possible to couple a
transducer to a frangible glass or glass-ceramic plate,
and use the fracture and consequent release of strain
energy to trigger an appropriate reaction. A frangible
glass-ceramic would be more suitable for applications
where a higher energy release is required. (Stephens
and Beauchamp have shown experimentally [212] that
a chemically strengthened glass disc 25.4mm in
diameter by 3.175 mm thick, yields an energy release
4203
1600
Figure 20
Short duration composite

strength as a function of testing temperature in air for a number of systems.
(a alumino-silicate glass + 50% silicon
carbide fibre; after [157], b pyrex borosilicate glass + 65% silicon carbide fibre;
after [180], c lithium alumino-silicate
glass-ceramic + 40% carbon fibre (heated
in air for 50 h at the temperature indicated
and then tested at ambient temperature);
after [205], d lithium alumino-silicate
glass-ceramic + 50%
silicon carbide
yarn; after [158], e barium osumilite
glass-ceramic + 25%
silicon carbide
whiskers; after [199].)
1200
12..
.Ic
~9
800
LL
400
0
0
200
400
600
Temperature (~
800
of 0.154J on fracture, whilst a similar glass-ceramic

disc delivers a markedly higher value of 1.29 J.) It has
also been suggested [213] that, because a change in
refractive index of the surface layers of glass is
obtained during chemical ion-exchange, such a
material could be utilized to prepare tailored lightguiding materials, including graded optical fibres.
Glass=ceramic materials have found applications
in Such diverse areas as cooking ware and missile
radomes. Low thermal expansion glass-ceramics
with excellent thermal shock resistance have proved
very successful in oven and cooking ware since the
introduction of Pyroceram and Corning Ware products by Corning Glass Works nearly thirty years ago.
Ceramic cooking hobs have also been constructed
from translucent low expansion glass-ceramics, and
transparent materials have been used m furnace windows and related areas. Matched thermal expansion
glass-ceramic-to-metal seals were first reported
by McMillan and co-workers [214-216] in the 1960s.
Since then, glass-ceramic materials have been developed with tailored expansion characteristics for
sealing to a wide variety of metals, alloys and other
ceramics, particularly for electrical and electronics
applications [217]. Other uses for glass-ceramic
materials include low expansion telescope mirrors,
heat exchangers for gas turbines, architectural and
nuclear waste disposal materials, microelectronic
substrates, and wear resistant bearings and related
products [98, 100, 101,218-226]. Glass-ceramics are
also beginning to find important applications in the
field of biomedical materials where high strength
and hardness combined with chemical inertness and
abrasion resistance are important assets [226-229]. In
addition, glass-ceramics are now available which can
be machined using conventional metal working tools
[230, 231]. These materials contain flaky crystals of a
fluorophlogopite mica phase (KMg3A1Si30~0F2) and
can be machined readily due to a combination of easy
cleavage of the mica flakes and the crack deflecting
properties of the microstructure. They are, therefore,
ideally suited for producing complex shapes and are
currently used, for example, in the US space shuttle,
where over twohundred differently shaped parts are
4204
1000
1200
employed, including door hinges and retaining rings.

Glass and glass-ceramic composites, in common
with other ceramic-based composite materials, have
not yet found widescale applications. This is due in
part to the problems and limitations outlined earlier.
In addition, ceramic composites have mainly been
aimed at providing improvements over existing
ceramic and related materials, and providing possible
direct replacements for existing technologies. As for
all new materials, an alternative approach is to make
use of new engineering concepts in which the emerging
materials can be tailored to meet the new demands.
This, of course, requires close cooperation between
materials scientists and design engineers, a union
which has not always been totally successful in the past.
Potential applications include thermal and ballistic
protection shrouds, electromagnetic window materials,
turbine and engine components, and biomedical
materials. Of these potential application areas, a
number show particular promise. For example,
carbon and silicon carbide fibre reinforced glass and
glass-ceramic composites for biomedical applications,
and silicon carbide reinforced cordierite and related
glass-ceramic materials for moderately high temperature (~IO00~
uses. The area of metal filament
reinforced glasses, could also provide a useful
extension to organic-based composites, for applishows promise for applications requiring high toughness and spalling resistance; for example, ballistic protection and related materials. In addition, glass and
glass-ceramic composites, including the carbon fibre
reinforced glasses, could also prove a useful
extension to organic based composites, for applications in which operating temperatures higher than
can currently be met by these materials are likely
to be encountered; they could therefore fill a useful
intermediate temperature gap in the materials field.
A number of questions remain to be answered,
however, before glass and glass-ceramic matrix composites can hope to find more widescale applications
and gain greater acceptance as viable engineering
materials. In particular, for elevated temperature
applications, matrix-reinforcement stability is a
major area of concern. More generally, the effect
of thermal and mechanical fatigue are also important,

and major difficulties remain in the joining or bonding of strengthened components. Due to many of
the problems outlined above, in particular matrixreinforcement compatibility and stability, glass and
glass-ceramic composites will be limited to ambient
and intermediate temperature applications in the
foreseeable future, rather than for prolonged use at
elevated temperatures.
The most promising areas Will continue to include
the reinforcement of materials using the new family of
ceramic fibres, particularly as more advanced and
more versatile fibres are developed with improved
oxidation resistance and a greater resistance to fibre
damage during handling and composite fabrication.
In addition, for certain applications, use of small
diameter metal filaments shows promise and, as outlined earlier, metal reinforcement already offers
the major advantage that it is much less readily
damaged during composite fabrication. Areas of
specific interest include the further development of
small diameter Wollaston filaments prepared from
more refractory alloys, including the high strength
superalloy series, together with the development of
more practical Taylor-wire systems; for example, with
superalloy cores and precursor glass-ceramic coatings
to yield, after consolidation and heat treatment,
matched thermal expansion systems. In addition, due
to the highly anisotropic properties of unidirectionally
aligned fibre systems, which is not a desirable feature
for all applications, future studies should include more
detailed analyses of multidirectional fibre systems.
Such materials have been widely exploited in the
polymer composites field, but only to a very limited
degree for ceramic-matrix systems. Other potentially
promising areas include the simultaneous use of two or
more strengthening and toughening techniques. For
example, combined use of particle and fibre reinforcement, or use of a surface method in conjunction
with a bulk technique; for example, fibre or particle
reinforcement combined with chemical strengthening
to produce a thin compressive surface layer which is
more resistant to the influence of surface defects.
Acknowledgements
The author is indebted to many of his colleagues,
including Messrs W. Bradshaw, B. L. Metcalfe and
M. J. C. Hill, and Dr C. R. Thomas, for useful
discussions. He is also grateful to Mrs. E. A. Prior for
carrying out some of the microscopy, and to AWE(A)
Technical Services for supplying the figures.
References
I. A. KELLY, -Strong Solids", 2nd edn (Clarendon Press,
Oxford, 1973).
2. A . A . G R I E F I T H , Phil. Trans. R. Soe., A221 ([920), 163.
3. ldem., 1st Int. Cong. AppI. Mechanics, Delft (1924) p. 55.
4. C. E. I N G L I S . Trans. hlsl. Naval Arch. 55 (1913) 219.
5. G. K O L O S O F F , Z. Math. Phys, 62 (1914) 26.
6. B. E. MOODY. G[a,s's Tecllnol. 29 (1988) 73.
7. R. W. D A V I D G E , "Mechanical Behaviour of Ceramics'"
(Cambridge University Press, Cambridge, 1979).
8. H. G. T A T T E R S A L L and G. TAPP1N, J. Mater. Sei. I
(1966) 296.
9. D. B. M A R S H A L L and A. G. EVANS, J. Amer. Ceram.

Soe. 68 (1985) 225.
10. R. ADAMS and P. W. M c M I L L A N , J. Mater. Sci. 12
(1977) 643.
II. T. A. M I C H A L S K E and S. w. F R E I M A N , J. Amer.
Ceram. Soe. 66 (1983) 284.
12. C. S Y M M E R S ,
J. B. WA R D
and
B. S U G A R M A N ,
Phys. Chem. Glasses 3 (1962) 76.
I3. w . B R E A R L E Y and D. G. H O L L O W A Y , ibid. 4 (1963)
69.
14. C. K. SAHA and A. R. COOPER, J. Amer. Ceram. Soc.
67 (I984) C158.
15. B. R. P R O C TO R , App/. Mater. Res. 3 (1964) 28.
16. R. E. MOULD, J. Amer. Ceram. Soc. 43 (1960) 160.
[7. N. H. RAY and M. H. STACEY, J. Muter. Sci. 4 (1969)
73.
18. K. H I R A O and M. T O M O Z A W A , J. Amer. Ceram. Soe.
70 (1987) 43.
I9. B. S U G A R M A N , J. Mater. Sci, 2 (1967) 275.
20. P. WARD and G. M. B A L L A R D , US Patent 4205 976
(1980).
21. R. G A R D O N , in "'Glass Science and Technology", Vol. 5,
edited by D. R. Uhhnann and N. J. Kreidl (Academic Press,
New York, 1980) pp 145-216.
22. S. D. STOOKEY, J. S. OLCOTT, H. M G A R F I N K E L
and D. k. R O T H E R M E L , in "Advances In Glass Technology", VI International Conference, (Plenum, New York,
1962) pp. 397-411.
23. R. W. P E T T I C R E W , L. E. SCHOTT and W. E. SMITH.
US Patent 3490 984 (1970).
24. W. K E I F E R , W, SACK and D. K R A U S E , US Patent
3907 577 (1975).
25. P. W. M c M I L L A N . B. P. H O D G S O N and R. E.
BOOTH, J. Mater. Sei. 4 (1969) 1029.
26. P. W. M c M I L L A N , G. P A R T R I D G E and J. G. DARRANT, Phys. Chem. Glasses I0 (1969) 153.
27. P. W. M c M I L L A N and G. P A R T R I D G E , UK Patent
1[08 473 (1968).
28. ldem., UK Patent 1322 796 (1973).
29. G. P A R T R I D G E and P. W. M c M I L L A N , Glass Teehnol.
15 (1974) 127.
30. R. ADAMS and P. W. M c M I L L A N , J. Mazer. Sei. 17
(1982) 2727.
31. G. P A R T R I D G E . Glass Teehnol. 28 (1987) 9.
32. S. S. K I S T L E R , J. Amer. Ceram. Soc. 45 (1962) 59.
33. R. F. B A R T H O L O M E W and H. M. G A R F I N K E L , in
"Glass Science and Technology", Vol. 5, edited by D. R.
Uhhnann and N. J. Kreidl (Academic Press, New York, 1980)
pp. 217-270.
34. M. E. N O R D B E R G , E. L. M O C H E L . H. M. G A R F I N KEL and J. S. OLCOTT, J. Amer. Ceram. Soe. 47 (1964)
215.
35. Coming Glass Works, UK Patent 966 732 (1964).
36. ldem., UK Patent 966 731 (1964).
37. F. J. M A R U S A K , US Patent 3301 649 (1967).
38. PPG Co. UK Patent 107I 351 (I967).
39. A. L. Z I J L S T R A and A. J. B U R G R A A F , J. Non-Cryst.
Solidv 1 (1968) 49.
40. ldem., ibid. 1 (1969) 163.
4I. A. E. S A U N D E R S and R. E. K U B I C H A N , US Patent
3433 611 (1969).
42. H. M. G A R F I N K E L and C. B. KING, J. Amer. Ceram.
Soe. 53 (1970) 686.
43. K. K O B A Y A S H I and K. SATO, Glass Teehnol. 19 (1978)
66.
44. I. W. D O N A L D and M. J. C. HILL, J. Mater. Sei., 23
(1988) 2797.
45. M. J. C. H ILL and 1. w . D O N A L D , Glass Teehnol., 30
(I989) 123.
46. J. C O R N E L I S S E N and A. J. B U R G R A A F , in Proceed-
ings 7th International Conference on Glass (Gordon and

Breach, New York, 1965) pp 319.1-319.4.
47. J. B. WARD.
B. S U G A R M A N
and
C. SYMMERS.
Glass Teehnol. 6 (1965) 90.

48. H. G A R F I N K E L , Glass 1nd. 50(1969) 28.
4205
49. C. w . DOYLE and F. J. MARUSAK, US Patent 3524

737 (1970).
50. A. K. VARSHNEYA and R. J. PETTI, J. Amer. Ceram.
Soc. 59 (1976) 42.
5I. D. W. R I N E H A R T , US Patent 4055 703 (1977).
52. M. J. KERPER and T. G. S C U D E R I , J. Amer. Ceram.
Soe. 49 (1966) 613.
53. J. S. STROUD, Glass Technol. 29 (1988) 108.
54. V. V. MOISEEV, T. V. PERMYAKOVA and M . N .
PLOTNIKOVA, ibid. 11 (1970) 6.
55. Glaverbel, U K Patent 1217 660 (1970).
56. T. YOKO, K. KAMIYA, Y. I S H I N O and S. SAAKA, J,
Amer. Ceram. Soc. 69 (1984) C154.
57. J. C. CAPP and J. E. SHELBY, ibid. 69 (1986) C110.
58. S. D. STOOKEY, "High Strength Materials", edited by
V. F. Zackay (Wiley, London, I965) pp. 669-68l.
59. N. WEBER, US Patent 3218 220 (1965).
60. S. SURNES, J. Amer. Ceram. Soc. 56 (1973) 514.
61. H. OHTA, J. Non-Cryst. Solids 24 (1977) 61.
62. M. ABOU-EL-LEIL and A. R. COOPER, Glass Technol.
21 (1980) 57.
63. Idem., J. Amer. Ceram. Soc. 64 (1981) 141.
64. C. K. SAHA a n d A . R COOPER, ibid 67 (1984) C160.
65. Anon, Bull. Amer. Ceram. Soc. 84 (1975) 19.
66. Corning Glass Works, U K Patent 786 951 (1957).
67. Philips Electronic and Associated Industries, UK Patent, 1215
729, 1970.
68. D. A. DUKE, J. F. MacDOWELL and B. R. KARSTETTER, J. Amer. Ceram. Soc. 50 (1967) 67.
69. G. H. BEALL, B. R. KARSTETTER and H. L. RITTLER, ibid, 50 (1967) 181.
70. Corning Glass Works, UK Patent 1219 193 (1971).
7I. A. R. COOPER and D. A. K R O H N , J. Amer. Ceram.
Soc. 52 (1969) 665.
72. O. R I C H M O N D , W. C. LESLIE and H. A. W R E I D T ,
Trans. A S M 57 (1964) 294.
73. P. F. JAMES, paper presented at Symposium on Glass
Strength, Society Glass Technology, Harrogate, May 1988.
74. D. A. K R O H N and A. R. COOPER, J. Amer. Ceram.
Soc. 52 (1969) 661.
75. D. A. KROHN, ibid. 53 (1970) 505.
76. A. J. DEYRUP, US Patent 2831 780 (1958).
77. P. E H R B U R G E R , A. MUSIELAK and J. LAHAYE,
Bull. Amer. Ceram. Soc. 59 (1980) 844.
78. D. K. NATH, ibid. 65 (1986) 752.
79. C. M. CARLESS, Glass Technol. 28 (1987) 175.
80. K. L. LOEWENSTEIN, "The Manufacturing Technology
of Continuous Glass Fibres, Glass Science and Technology",
Vol. 6, 2nd edn (Elsevier, Amsterdam, 1983).
81. A. KUSKE and G. ROBERTSON, "Photoelastic stress
analysis", (Wiley, London, 1974).
82. R. WELLER, J. Appl. Phys. 10 (1939) 266.
83. E. K. BEAUCHAMP and R. H. A L T H E R R , J. Amer.
Ceram. Soc. 54 (1971) 103.
84. W. BRADSHAW, J. Mater. Sci. 14 (1979) 2981.
85. A. Y. SANE and A. R. COOPER, J. Amer. Ceram. Soc.
61 (1978) 359.
86. Idem., ibid. 70 (1987) 86.
87. J. M. BARSOM, ibid. 51 (1968) 75.
88. B. R. LAWN and D. B. MARSHALL, Phys. Chem. Glasses, 18 (1977) 7.
89. M. BAKIOGLU, F. ERDOGAN and D. P. H. HASSELMAN, J. Mater. Sei. 11 (1976) 1826.
90. C. C. HS1AO, "'Fracture 1977", Vol. 3, ICF4, Waterloo,
Canada, (1977) pp. 985-989.
91. J. CHANG and J. CHOU, J. Non-Cryst. Solids 52 (1982)
395.
92. M. M. ABOU-EL-LEIL,
F . A . CAMARATTA
and
R. R. D1GENOVA, J. Amer. Ceram. Soc. 68 (1985) C18.
93. M. M. C H A U D H A R I and L. L I A N G Y I , Nature 320
(1986) 713.
94. M. M. ABOU-EL-LEIL,
F . A . CAMARATTA
and
R. R. D1GENOVA, J. Amer. Ceram. Soc. 69 (1986) 713.
95. D. R. OAKLEY and M. F. GREEN, paper presented at
Symposium on Glass Strength, Society Glass Technology,
Harrogate, May 1988.
4206
96. I. W. D O N A L D and B. L. METCALFE, AWE, unpublished work (1983).

97. M, V. SWAIN, J. T. HAGAN and J. E. FIELD, J.
Mater. Sci. 12 (1977) 1914.
98. P. W. McMILLAN, "Glass ceramics, 2nd edn (Academic
Press, London, 1979).
99. z. STRNAD, "Glass-Ceramic Materials (Elsevier, Amsterdam, 1986).
100. P. w. McMILLAN and G. P A R T R I D G E , Proc. Brit.
Ceram. Soc. 3 (1965) 241.
101. G. H. BEALL, in "Glass Science and Technology",
Vol. I, "Glass Forming Systems", edited by D. R. Uhlmann
and N. J. Kreidl (Academic, New York, 1983) pp. 403-443.
102. J. HLAVAC, "The Technology of Glass and Ceramics,"
(Elsevier, Amsterdam, 1983) pp. 228-243.
103. A. DE, A. K. MOITRA and P. K. GUPTA, Trans.
Indian Ceram. Soc. 43 (1984) 105.
104. P. W. McMILLAN,
S . V . PHILLIPS
and G. PART R I D G E , J. Mater. Sci. 1 (i966) 269.
105. M. P. BOROM. A. M. T U R K A L O and R. H.
DOREMUS, J. Amer. Ceram. Soc. 58 (1975) 385.
106. D. R. BRIDGE, D. H O L L A N D and P. W. McM1LLAN,
Glass Teehnol. 26 (1985) 286.
107. I. W. D O N A L D , B. L. METCALFE, D. J. WOOD and
J. R. COPLEY, J. Mater. Sci. in press.
108. I. W. D O N A L D
and R. A. McCURRIE,
J. Amer.
Ceram. Soc. 58 (1972) 289.
109. Z-X. CHEN and P. w. McMILLAN, J. Mater. Sci. 20
(1985) 3428.
110. A. J. STRYJAK and P. W. McMILLAN, ibid. 13 (1978)
1794.
I l l . R. W. RICE, ibid. 14 (1979) 2768.
112. M. A. STETT and R. M, F U L R A T H , J. Amer. Ceram.
Soc. 53 (1970) 5.
113. C. O. McHUGH, T. J. WHALEN and M. H U M E N I K ,
ibid. 49 (1966) 486.
114. R. M. M c M E E K I N G and A. G. EVANS, ibid. 68 (1982)
242.
115. V. V. KRSTIC, P. S. NICHOLSON and R. G. HOAGLAND, ibid. 64 (1981) 499.
116. T. B. T R O C Z Y N S K I , P. S. N I C H O L S O N and C. E.
RUCKER, ibid. 71 (1988) C276.
117. D. P. H. HASSELMAN and R. M. FULRATH, ibid. 49
(1966) 68.
I18. Y. NIVAS and R. M. F U L R A T H , ibid. 53 (1970) 188.
119. R. R. TUMMALA and A. L. F R E I E D B E R G , ibid. 52
(1969) 228.
[20. R. W. D A V I D G E and T. J. GREEN, J. Mater. Sei 3
(1968) 629.
121. R. I. BERTOLOTTI and R. M. F U L R A T H , J. Amer.
Ceram. Soc. 50 (1967) 558.
I22. F. F. LANGE, Phil. Mag. 22 (1970) 983.
123. A. G. EVANS. ibid. 26 (1972) 1327.
124. V. D. KRSTIC, ibid. A, 48 (1983) 695,
I25. J. SELSING, J. Amer. Ceram. Soc. 44 (1961) 419.
126. w. J. FREY and J. D. MACKENZIE, J. Mater. Sci. 2
(1967) 124.
127. R. C. ROSSI, J. Amer. Ceram. Soc. 82 (1969) 290.
I28. N. M1YATA and H. JINNO, J. Mater. Sci. 17 (1982) 547.
129. T. L. JESSEN, J. J. MECHOLSKY and R. H. MOORE,
Bull. Amer. Ceram. Soc. 65 (1986) 377.
130. R. C. GARV1E, R. H. H A N N I N K and R. T. PASCOE,
Nature 258 (1975) 703.
131. D. L. PORTER and A. H. HEUER, J. Amer. Ceram.
Soc. 60 (1977) 543.
132. A. H. HEUER, ibid. 70 (1987) 689.
133. N. CLAUSSEN, Z. Werksto/fteeh. 13 (1982) 138.
134. N. CLA USSEN, in "'European Coll. Ceramics in Advanced
Energy Technologies" (Petten, The Netherlands, 1982)
pp. 51-86.
135. F. F. LANGE, J. Mater. Sci. 17 (1982) 225.
136. N. CLAUSSEN, J. Amer. Ceram. Soc. 59 (1976) 49.
137. R. STEPHENS, An Introduction to Zirconia, Magnesium
Elecktron Ltd., Publication No. 13, (1983).
138. S. F A G H E R A Z Z I ,
S. ENZO,
V. G O T T A R D I
and
G. SCARINC1, J. Mater. Sei. 15 (1980) 2693.
[39. K. KEEFER and T. M1CHALSKE. US Patent 4587 224

([986),
140. M. NOGAM1 and M. TOMOZAWA, J. Amer. Ceram.
Soc. 69 (1986) 99.
I41. H. KRENCHEL. "Fibre Reinforcement",
(Akademisk
Forlag, Copenhagen, 1964).
142, v. LAWS, J. Phys. D 4 ( 1 9 7 1 ) I737.
143. J A. AVESTON. in "The Properties of Fibre Composites", Conference Proceedings NPL, 1971 (IPC Science and
Technology Press, Guildford, 1972) pp. 63-74.
144. A. H. COTTRELL, Proc. R. Soc. A282 (1974) I.
145. I. W. DONALD and P. W. McM1LLAN, J, Mater. Sci.
12 (1977) 290.
146. D. K. HALE and A. KELLY, Anti_ Rev. Mater. Sci. 2
(1972) 405.
147, R. A. J. SAMBELL,
A, BR1GGS,
D . C . PH[LL1PS
and D. H. BOWEN, J. Mater. Sei. 7 (1972) 676.
148, R. A. J. SAMBELL, D. H. BOWEN and D. C. PHILLIPS, ibid. 7 (1972) 663.
149. D. C. PHILLIPS, ibid. 7(1972)1175.
150. D. C. PHILLIPS, R. A. J. SAMBELL and D. H.
BOWEN, ibid. 7 (1972) 1454.
151. D. C. PHILLIPS. J. Compos. Mater. 8 (1974) 130.
152. l. W. D O N A L D and P. W. McMILLAN, J. Mate;-. Sci.
11 (1976) 949.
153. G. E I N M A H L , U C R L Technicat Report 16844, May
1966.
154. R. H. D U N G A N , J. A. G I L B E R T and J. C. SMITH,
J. Amer. Ceram. Soe. 56 (1973) 345.
I55. S. R. LEVITT, J. Mate;'. Sci. 8 (1973) 793.
I56. J AVESTON, G. A. COOPER and A. KELLY, in "'The
Properties of Fibre Composites", NPL Conference Proceedings (IPC Science and Technology Press, Guildford, 1972)
pp. 15-24.
157. J. J. BRENNAN and K. W. PREWO, UK Patent Appl.
GB 2075 490A, 1981.
158. tdem,, J, Mater. Sei., 17 (t982) 237t.
159. T. MAH, M. G. M E N D I R A T T A , A. P. KATZ. R. RUH
and K. S. MAZD1YASN1, J. Amer. Ceram. Soc. 68 (1985)
C27.
160. ldem., ibid. 68 (1985) C248.
161. S. L A N K F O R D , Composites 18 (1987) 145.
162. K. M. PREWO, J, Mater. Sci. 22 (1987) 2695.
I63. R. CHAIM and A. H. HEUER, Advanced Ceram. Mater.
2 (1987) 154.
164. K. M. PREWO and J. J. BRENNAN, J. Mater. Sei. 15
(1980) 463.
165. tdem., ibid. 17 (1982) 1201.
166. E. Y. LUH and A. G. EVANS, Ceram. Engng Sei. Proc.
6 (1985) 608.
167. M. P. BOROM and C. A. JOHNSON, J. Amer. Ceram.
Soc. 70 (1987) 1.
168. R. F. COOPER and K. C H Y U N G , J. Mater. Sci. 22
(1987) 3148.
169. D. M. DAWSON, R. F. PRESTON and A. PURSER.
Silicates hMustriels 53 (1988) 129.
170. M. A. HERRON
and S. H. RISBUD,
Bull. Amer.
Ceram. Soe. 65 (1986) 342,
17l. T. A. M I C H A L S K E and J. R. HELLMAN, J. Amer.
Ceram. Soe. 71 (1988) 725.
172. v. c~ N A R D O N E and K. M. PREWO, ibid. 23 (t988)
168.
173. K~ C H Y U N G , J. Mate?'. Sci. 22 (1987) 701.
174. L. F. J O H N S O N , D. P. H. HASSELMAN and E. M1NFORD, ibid. 22 (I987) 3111.
175. R. A. ALLAIRE, V. F. JANAS, S. STUCHLY
and
M. P. TAYLOR, S A M P E Quarterly 19 (1987) 25.
176. K. W. PREWO and J. A. BATT, J. Mater. Sci. 23 (1988)
523.
[77. K. M. PREWO, ibid. 23 (1988) 2745.
t78. R. W. RICE, in Proceedings 5tb Annual Conference Composites and Advanced Ceramic Materials, (Amer. Ceram.
Soc., Columbus, Ohio, 1981) pp. 661-70i.
179. D. C. PHILLIPS, in "Handbook of Composites", Vol. 4,
edited by A. Kelly and S. T. Mileiko (Elsevier, Amsterdam,
1983) pp. 373 428.
[80. K. M, PREWO, J. J. BRENNAN and G. K. LAYDEN,

Bull, Amer. Ceram. Sac. 65 (1986) 305.
181. L. J. SCHIOLER and J. J STIGL1CH, ibid, 65 (1986)
289.
I82. J. J. MECHOLSKY, ibid. 65 (1986) 315.
I83. R. W. DAVIDGE, Composites 18 (1987)92.
[84. T-I. MAH, M. G, M E N D I R A T T A , A. P. KATZ and
K. S. M A Z D I Y A S N I , Bull. Amer. Ceram. Soc. 66 (1987)
304.
185. D. B, MARSHALL and J. E. RITTER. ibid. 66 (1987)
309.
186. G. FISHER, ibid. 63 (1984) 360.
I87. I, W. DONALD, B. L, METCALFE and A. D. BYE, J.
Mater. Sei. Io be submitted.
I88. I. w. DONALD, ibid. 22 (1987) 266t.
I89. 1. w. DONALD, B. L. METCALFE and A. D. BYE, J.
Mater. Sei. Lelt. 7 (1988) 964.
190. A. G. EVANS, J. Mater. Sci, 9 ([974) 1145.
191. M. J. CHAPPEL and R. S. MILLMAN, ibid. 12 (1974)
1933.
192. I. R. McCOLL and J. G. MORLEY, ibid. 12 (1977)
1165.
[93. E. A. FEEST, Mater. Design 7 (1986) 58.
194. P. BRACKE, H. SCHURMANS and J. VORHOEST, in
"'Inorganic Fibres and Composite Materials", European
Patent Office (Pergamon Press, Oxford, 1984) pp. 97-12t.
195. G. C. WEI and P. F. BECHER, Bull. Amer. Ceram. Soe.
64 (I985) 298.
196. A. H. CHOKSHI and J. R. PORTER. J. Amer. Ceram.
Soc. 68 (1985) C144.
197. N. CLAUSSEN, K-L. WEISSKOPF and M. RUHE,
ibid. 69 (1986) 288.
198. P. F. BECHER, T. N. TIEGS, ibid. 70 (1987) 651.
199. K. P. G A D K A R E E and K. C H Y U N G , Bull. Amer,
Ceram. Soc. 65 (1986) 370.
200. G. K. LAYDEN and K M. PREWO, UTRC Final
Report R85-916943-I (1985).
201. D. I. H. A T K I N S O N and P. W. McMILLAN, J. Mate;'.
Sci. I0 (1975) 2012.
202. Y. ABE, M. HOSOE, T. KASUGA, H. ISH1KAWA,
N. S H I U K A I , Y. SUZUKI and J. NAKAYAMA, J.
Amer. Ceram. Soe. 65 (1982) C189.
203. G. F. F R E E G U A R D and D. MARSHALL. Composites
II (1980) 25.
204. R. D. KING and R. W. W R I G H T , Glass Technol. 28
(1987) 73.
205. R A. J. SAMBELL,
D . C . PHILLIPS
and
D. H. BOWEN, in "'Carbon Fibres, Their Place in
Modern Technology: Proceeding 2nd International Conference Carbon Fibres" (Plastics Institute, London, 1974).
206. C. A. CALOW and R. J WAKELIN, J. Inst. Metals 96
(1968) 147.
207. S. DATTA, S. S. MITRA, D. C H A K R A V O R T Y , S.
RAM and D. BAHADUR, J. Mater. Sci. Lett. 5 (1986) 89,
208. W. B. H1LLIG, J. Amer. Ceram. Soc. 71 (1988) C96.
209. P. BRACKE, H. SCHURMANS and J. VORHOEST, in
"Inorganic Fibres and Composite Materials, European Patent
Office" (Pergamon Press, Oxford. 1984) pp. 123-135.
210. J. A. C O R N I E , Y-M. C H I A N G ,
D. R. U H L M A N N ,
A. MORTENSEN and J M. COLLINS, Butt. Amer.
Ceram. Soc. 65 (1986) 293.
21I. E. K. BEAUCHAMP, in "Fracture Mechanics of Ceramics", Vol. 2, edited by R. C. Bradt, D. P. H. Hasselman and
F. F. Lange (Plenum, New York, 1974) pp. 817-827.
212. H. P. STEPHENS and E. K. BEAUCHAMP, Bull. Amer.
Ceram. Soc. 53 (1974) 159.
213. w. G. FRENCH and A. D. PEARSON, ibid. 49 (1970)
974.
214. P. W. McMILLAN and g. P. HODGSON, Engineering
196 (1963) 366.
B.P.HODGSON
and
G. PART R I D G E , Glass Technol. 7 (1966) 121.
G. P A R T R I D G E ,
B . P . HODGSON and H. R. HEAP, ibid7(1966) 128.
217. G. P A R T R I D G E and C A. ELYARD, Brit. Ceram.
Procs. 34 (1984) 219.
4207
218. T. R. NASH, E. M. PASHBY and R. L. C O L L E T T ,

Glass Technol. 24 (1983) 298.
219. G. P A R T R I D G E , ibid. 24 (1983) 293.
220. M. W A D A and S. K A W A M U R A , Bull. Inst. Chem. Res.
Kyoto Univ. 59 (1981) 256.
221. D. G. G R O S S M A N , in "'Advances in Ceramics 4"', edited
by J. H. Simmons, D. R. U h l m a n n and G. H. Beall (Amer.
Ceram. Soc., Columbus, Ohio, 1982) pp. 249-259.
222. G. H. BEALL and D. A. D U K E , in "Glass Science and
Technology I", edited by D. R. U h l m a n n and N. J. Kreidl
(Academic Press, New York, 1983) pp. 241 257.
223. W. R. P R I N D L E , Ceramica 29 (1983) 273.
224. I. W. D O N A L D , A W R E Report No. 0 19/84, H M S O
(London), 1984.
225. G. H. BEALL, J. Non-Cryst. Solids 73 (1985) 413.
4208
226. w. T. M a c C U L L O C H , Brit. Dental. J. 124 (1968) 361.

227. S. F. H U L B E R T ,
L. L. H E N C H ,
D. FORBES
and
L. S. B O W M A N , in "Advances in Surgery", edited by
P. Vincenzini (Elsevier, Amsterdam, 1983) pp. 3-29.
228. G. F I S H E R , Bull. Amer. Ceram. Soc. 62 (1983) 1237.
229. J. W. BORETOS, Adv. Ceram. Mater. 2 (1987) 15.
230. G. H. BEALL, in "Advances in Nucleation and Crystallization in Glasses", edited by L. L. Hench and S. W. Freiman (Amer. Ceram. Soc., Columbus, Ohio, 1971) pp. 251-261.
231. D . G . G R O S S M A N , Glass Technol. 24 (1983) 11.
Received 18 November 1988

and accepted 20 January 1989

Review - Methods For Improving The Mechanical Properties of Oxide Glasses

Uploaded by

Copyright:

Available Formats

Review - Methods For Improving The Mechanical Properties of Oxide Glasses

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Review - Methods For Improving The Mechanical Properties of Oxide Glasses

Uploaded by

Copyright:

Available Formats

J O U R N A L OF M A T E R I A L S S C I E N C E 24 (1989) 4177 4208

Atomic Weapons Establishment, Aldermaston, Berkshire, UK

9 1989 Chapman and Hall Ltd.

applications for strengthened and toughened glasses

2. Mechanical properties of glasses

where cr is the applied stress, c the crack length, and e

where 7 is the fracture surface energy.

For conditions of plane strain, the relationship

where, v is Poisson's ratio.

where Y is a dimensionless constant which takes into

depending on the environment, small, pre-existing

3. Methods for improving the

3.1. Surface methods for improvement in

technique of "flame polishing" in which the surface of

3.1.2.1. Thermal treatments. One method by which a

TA B LE I I Summary of methods for improving the mechanical properties of glasses

removal of surface damage

surface "solidifies" before interior:

exchange of small ions in glass surface

formation of lower expansion surface

(e) Thin coating

protection of pristine surface

glasses of c~ > cladding

coating of surface with compatible

many silicate glasses

internal nucleation and growth of

glasses containing Zr02

special case of dispersion-strengthening

(d) Fibre reinforcement

crack impedement, load transfer, fibre

crack impedement, load transfer,

alternate layers of glass and

approximately parabolic, as illustrated in Fig. 1. As a

The magnitude of the surface compressive stress

Figure 1 Typical stress profile of a thermally

to geometrical considerations. These tend to limit

have been given of glasses strengthened by a simple

Strengthening has also been noted by Adams and

treatments which yield surface compressive stresses,

T A B L E I I I Details of surface crystallized glasses

Calcium zmc alummo-silicategNsses

Major crystalline phase

*Specific glass compositions are given in Table VI

Typical stress profile of a chemically

*taken from [58].

diffusivities are generally found for alumino-silicate

temperature and time. A maximum in strength is

Figure 3 Strength of chemical ion-exchanged

Figure 4 Strength of an ion-exchanged sodium

alumino-silicate glass as a function of treatment

high compressive stresses during external loading of

Figure 5 Ambient temperature strength of

ion-exchanged glasses as a function of

TAB L E V Details of chemically strengthened materials

Littlium alumino-silieate glasses

Sodium alumino-silicale glasses

Mixed lithium sodium alumino-silicate glasses

Aluminaz/?ee alkali silicate glasses