3 Matrices

In this chapter we describe the resins used for the manufacture of composite artefacts.
The concept of curing is discussed with respect to the chemistry of typical polymer
matrices. The advantages and disadvantages of thermosets and thermoplastics are
also discussed.
In the case of thermosets, the importance of thermoplastic and rubber toughening is
considered. While we concentrate on polymer matrix materials, ceramic and metal
matrices are referred to for completeness.

3.1 Introduction

Most reinforcing fibres have moduli that are at least 10 times the moduli of glassy
polymers, so the role of the matrix is to maintain their orientation and transfer loads to
the fibres. The rate of transfer of load between the fibres and matrix is a critical aspect,
especially for discontinuous fibre systems and at fibre and/or matrix breaks in continu-
ous fibre composites.

3.1.1 Mechanisms of Viscoelasticity in Polymers

The modulus of a polymer is strongly dependent on the mechanism of deformation
available [1]. Depending on the molecular structure of the polymer, it can exist at
room temperature as either
a. glass
b. retarded high elastic material (‘leathery material’)
c. rubber
d. viscoelastic fluid.
All polymers can exist in either state, depending on the temperature. Thus, a glassy
polymer will go through four stages of viscoelasticity on heating. The glassy and
rubbery states can be characterized by unique values of modulus, whereas within the
transition regions (b) and (d) the properties are time-dependent. Of these, the glass
transition temperature (Tg) is a critical parameter as it defines the service temperature
of a rigid structural composite. Figure 3.1 provides a schematic showing the change in
relaxation modulus with temperature. The modulus changes over several decades as

[Link] Published online by Cambridge University Press

 3.1 Introduction 77

10

9 1 GPa
Glass
Crystalline
8
Log ER/Pa

Leather-like
7

6 Crosslinked 1 MPa

5 Rubber

Viscous
4
flow

3 1 kPa
Tg Temperature Tm

Figure 3.1 The relaxation modulus (ER) of a polymer as a function of morphology (crystallization
and crosslinking) and temperature. An illustration of the mechanical properties at different
temperatures relative to the glass transition temperature.

the glassy material becomes rubbery. Tg is defined conventionally as the maximum in

tanδ, which occurs at the temperature for maximum damping and is shown in
Figure 3.1 at the inflection point of the storage modulus versus temperature curve.
In the transition region the material exhibits ‘leather-like’ properties. The presence of
crystallites increases the modulus in an analogous way to the incorporation of
particulate fillers. Crosslinking of the polymer chains also enhances the modulus
above Tg. Light crosslinking also extends the rubbery plateau (temperature range)
and is often referred to as vulcanization. In aerospace engineering the onset of the loss
of a glassy modulus is often used to define the value of Tg, which in Figure 3.1 occurs
at approximately 5–10
C lower.

3.1.2 Magnitude of the Modulus (E) of a Glassy Polymer

In a polymer glass the only forces resisting deformation are the intermolecular
forces between molecular segments. These range from van der Waals through to
hydrogen bonding. Crosslinking introduces irreversible covalent bonds in addition
to the intermolecular forces. To have a major effect on modulus, crosslinking
needs to be extensive. With linear polymers of regular structure, such as polypro-
pylene or polyethylene, crystallization enhances the modulus because of the higher
density. In the latter, the modulus can be modelled using composite mechanics. As
discussed in Chapter 2, polymer molecules are randomly coiled through rotation
about skeletal bonds so that under an applied load the forces of interaction
between the molecules carry the load. Therefore, a simple estimate of modulus
can be obtained from consideration of a simple frozen hydrocarbon gas, such as
methane (CH4) glass.

[Link] Published online by Cambridge University Press

 78 Matrices

The intermolecular energy ðϕp Þ between molecules as a function of distance, R is

described by the Lennard–Jones potential:
ϕp ¼ BRn  AR6 , (3.1)

where A and B are constants and n ¼ 12 for the simple methane molecule.
As the molecules move together there is an attractive force, which has a maximum
value (negative by convention) of εo at a fixed distance Ro. When pushed closer a
repulsive force develops and the value of ϕp becomes progressively more positive.
Therefore eq. (3.1) can be normalized to provide the potential energy between two
methane molecules:

12
6
ϕp R R
¼ 2 : (3.2)
εo Ro Ro

For a close-packed structure in which the molecule has

3 12 neighbours, eq. (3.2) can be
simplified in terms of molar volume, V, since Rα V1 :

4
2
ϕp Vo Vo
¼ 2 , (3.3)
ϕo V V
where ϕo ¼ 12N A εo .
Under compression, the change in molar volume of the glassy methane is given by
dE ¼ PdV, assuming the change in entropy (dS) is zero. Differentiation gives
dP d2 E i
¼ ,
dV dV 2
where E i is the interaction energy. Therefore, the bulk modulus, K, is given by

2 
dP d Ei
K ¼ V ¼ V : (3.4)
dV dV 2
Thus, eq. (3.3) becomes

"
6
4 #
ϕ Vo Vo
K¼4 o 5 3 , (3.5)
Vo V V

where ϕo ¼ 12N A εo .
Since V  V o , eq. (3.5) becomes
ϕo εo N A
K¼8 ¼ 96 : (3.6)
Vo Vo
For methane, εo ¼ 0:0127 eV.
The molar volume, Vo, is given by molecular weight/density. For CH4, this is:

16  103 kg mol1
Vo ¼ ¼ 29  106 m3 mol1 :
547 kg m3
Therefore, K  4 GPa.

[Link] Published online by Cambridge University Press

 3.1 Introduction 79

The bulk modulus, K, is related to Young’s modulus, E, according to eq. (3.7):

E ¼ 3Kð1  2 n Þ, (3.7)

where n is Poisson’s ratio. For a polymer, n ¼ 0:3  0:4. Therefore,

E  2:44:8 GPa.
The estimate of Young’s modulus for glassy methane is 4 GPa, and the typical
maximum value for a glassy polymer will be similar when other deformation mech-
anisms are absent. Covalent crosslinking between polymer molecules can add resist-
ance to deformation, but the maximum achievable is 7 GPa. This value has only
been approached in highly crosslinked phenolic resins, but these have very low failure
strains because ductile mechanisms, which reduce stress concentrations at a crack tip,
are not available. Matrix resins such as epoxies have moduli of 2–3.5 GPa and failure
strains of 1.5–6% because cooperative skeletal rotation within the bridging network
chains provides the mechanism of ductility.

3.1.3 Magnitude of the Modulus of a Rubber

The thermodynamic and statistical analysis of the deformation of a rubber by Treloar
[2] provides the magnitude of the modulus. Deformation of a rubber or elastomer
involves flow, so a shear modulus, G, is given by:
ρRT
G¼ , (3.8)
Mc

where ρ is the density, R is the gas constant, T is the temperature, and Mc is the
number-average molecular weight of a network chain.
Unvulcanized or non-crosslinked rubbers deform by a reptation mechanism involv-
ing rotation about skeletal bonds leading to non-recoverable flow. To stabilize the
flow of rubbers they are vulcanized or crosslinked. Natural and many synthetic
rubbers employ sulphur for crosslinking, hence the term vulcanization. The degree
of crosslinking determines the shear modulus of a rubber, as shown by eq. (3.8). As
the degree of crosslinking increases, the length (defined by Mc) of the network chains
decreases and G increases.
Real vulcanized rubbers are not perfectly crosslinked, so network defects will exist.
These include ineffective network chains such as loops, entanglements, and incom-
plete network chains or network-ends. Equation (3.8) therefore needs to be corrected:

2 

<r c > ρRT Mc
G¼ 12 , (3.9)
<r 20 > M c M

where <r 2c > is the average square end-to-end distance of a network chain and <r 20 >
is the unperturbed value for a free chain. This correction represents the influence of the
molecular structure of the network chain. M is the number-average molecular weight
of the polymer (before crosslinking) and (1–2 Mc/M) represents the effects of
network defects.

[Link] Published online by Cambridge University Press

 80 Matrices

A typical value of G for a rubber is 0.4 MPa, and a low strain elastic modulus
(i.e. tensile) would be 1 MPa. Poly(butadiene) rubber exhibits ideal high elastic
behaviour at room temperature, as indicated by zero hysteresis in the load–extension
curve. Natural rubber (poly(isoprene)), however, exhibits significant hysteresis. This
arises from the energy restriction imposed on the skeletal rotations by the pendant
methyl group, but also at high extensions by strain-induced crystallization of the
chain. The tensile modulus then has a contribution from loading of the molecular
chain bonds.

3.2 Polymeric Matrices for Composites

The fabrication of a composite artefact involves the impregnation of fibres with either a
thermosetting resin or a thermoplastic. Both thermoplastics and thermosets are used, but
thermosets are preferred because they are liquid and ‘monomeric’ and have low viscosity
so that fibre impregnation is easy and possible at room temperature. They are ‘cured’ into
crosslinked polymers, either thermally or with hardeners and/or catalysts. Thermoplastics
are linear polymers that on heating can become viscoelastic fluids for infiltration of
continuous fibres or mats, or mixed with discontinuous fibres under pressure. The critical
aspect of the latter is the high temperatures required for impregnation, and as a conse-
quence the reproducibility of morphology of the polymer on cooling.
The advantages of a thermoset (relative to thermoplastics) are:

 low viscosity for impregnation and wetting of fibres;

 availability of a large number of fabrication processes; and
 the length of the fibres employed in the process is maximized.
The disadvantages of thermosets are also worth considering:

 material and artefact are formed simultaneously during curing;

 cast resins have low failure strain (εmu) and fracture toughness (K1c, G1c);
 curing times can be extremely long; and
 the critical length (Lc) of a reinforcing fibre tends to be longer when embedded in
thermosets (see Chapter 5).
The advantages of thermoplastic matrices (relative to thermosets) are:

 higher ductility;
 no chemistry is involved in the process;
 more rapid composite processing arising from the need for only heating and
cooling;
 Lc tends to be lower and short fibre composites can be manufactured by standard
thermoplastic moulding techniques such as injection moulding;
 low moisture absorption; and
 good hot/wet props.

[Link] Published online by Cambridge University Press

 3.2 Polymeric Matrices for Composites 81

These benefits are offset by the following disadvantages:

 high melt viscosity for fibre impregnation;

 high impregnation temperatures;
 crystalline morphology may vary;
 continuous fibre composites present complications to fabrication processes;
 uses mainly short-fibre-reinforced material, which has low mechanical performance
arising from the reduced reinforcement efficiency of load transfer to discontinuous
and randomly aligned fibres;
 the polymer has a low Tg, which is often offset by the presence of crystallites of
high melting point (Tm);
 service temperatures are generally lower for thermoplastics-based structural
composites because a small loss of rigidity will still occur at Tg; and
 thermoplastics have a lower modulus (Em).

3.2.1 Requirements of a Resin Matrix

The following aspects of a resin intended for composite manufacture are desirable:

 low viscosity for impregnation;

 high tensile modulus, Em (a maximum of 4 GPa for a cured thermoset resin matrix);
 high failure strain, εmu;
 controllable curing for a thermoset; and
 controllable microstructure for a thermoplastic.
Typical thermosets resins employed as matrices are phenolics; unsaturated polyester
and related resins (vinyl esters, urethane acrylates), which are mainly used in indus-
trial applications; epoxy resins, which are used for structural and high-performance
applications; advanced resins, including polyimides, bismaleimides, cyanate esters,
and other specialist resins used in high-temperature applications.
Typical thermoplastics include most injection mouldable plastics, providing the
reinforcements are compatible. The limitation is that the fibres need to be discontinu-
ous in order to flow around the mould (i.e. short fibre composites). More advanced
thermoplastics include polyether ether ketone (PEEK), polyether sulphone (PES), and
polyetherimide (PEI), which can be used with continuous fibres as well for discontinu-
ous fibre mouldings. The continuous fibres have dominated applications because only
high performance can justify their high cost. Polypropene is an interesting matrix
polymer; because of its low cost it can be used in mouldings but also as a matrix for
continuous glass fibres (continuous and long glass fibres). Its limitations are deter-
mined by its compatibility with the reinforcements, but mainly by its low glass
transition temperature.
The fabrication of a composite involves the impregnation and wet-out of the fibres,
so thermosetting resins are preferred because a large number of manufacturing
processes are available for ‘long’ and continuous fibres.

[Link] Published online by Cambridge University Press

 82 Matrices

3.3 Curing of Thermoset Resins

The formation of most polymer composites involves the ‘curing’ of the resin after the
fibres have been impregnated. Curing either occurs by 3D step-growth polymeriza-
tion; addition copolymerization of unsaturated resin with a reactive diluent or mono-
mer; addition polymerization of cyclic monomers; or addition copolymerization of
cyclic monomers.

3.3.1 3D Step-Growth Polymerization

As the name implies, the molecular weight of the matrix builds in a step-wise manner
until it becomes infinite as the resin gels into a network. To achieve the rigidity of the
matrix for load bearing, further crosslinking needs to occur by diffusion of oligomers
to reaction sites, as shown in Box 3.1.
From the above it can be seen that the functionality (the number of reactive groups)
of the monomers will determine the extent of reaction of these groups for gelation.
With highly functional monomers, gelation occurs at low extents of reaction but the
potential degree of crosslinking is higher. To obtain matrices with high moduli and
glass transition temperatures, large degrees of crosslinking are required. From the
perspective of composite manufacture the above discussion explains the need for post-
curing.
The Carothers theory [3] predicts the molecular weight of a polymer obtained by
this mechanism as a function of the extent of reaction. For bi-functional monomers,
linear polymers are formed and the number-average degree of polymerization of the
chain, Pn, is given by:
Pn ¼ 1=ð1  ρÞ, (3.10)

where ρ is the extent of reaction of functional groups and Pn ¼ Mn/Mo, where Mn is the
number-average molecular weight of the polymer from a monomer of molecular
weight Mo.

Box 3.1
Monomers
#
Dimers, trimers
#
Highly branched oligomers (broad distribution)
#
Network formation (gelation)
#
Network crosslinking (by diffusion and reaction of oligomers)

[Link] Published online by Cambridge University Press

 3.3 Curing of Thermoset Resins 83

For the purposes of this discussion, we are interested in the gelation point for curing
of multifunctional monomers. By applying eq. (3.10) we get
Pn ¼ 2=ð2  ρf av Þ, (3.11)

where fav is the average functionality of the mixture of monomers. For linear poly-
merization with bi-functional monomers, fav ¼ 2 and eq. (3.11) reduces to eq. (3.10).
At gelation, Mn approaches ∞; thus,
ρc ¼ 2=f av , (3.12)

where ρc is the critical extent of reaction for gelation or gel point.

From eq. (3.12) we see that the gelation occurs at lower extents of cure for systems
with high functionality.
The Carothers theory can be extended to include an imbalance in the concen-
trations of reactive groups. The Flory statistical theory of gelation considers the
probability of reaction of functional groups to predict the gel point. This is beyond
the scope of this book, but is of fundamental interest to process engineers modelling
matrix curing.
Step-growth polymerization often involves the loss of a small molecule (often
water), in which case it is referred to as condensation polymerization. Phenolic resins
are an example of condensation or step-growth curing, and demonstrate how curing
can be interrupted. It is used here to illustrate the fundamentals or prepreg technology.

3.3.2 Phenolic Resins

Phenolic resins represent the first truly synthetic polymer, which became available at
the turn of the twentieth century. Before then the early plastics were modified natural
products. The starting reactants are formaldehyde (H2CO) and phenol (C6H6OH).
There are two synthetic routes: resole and novolak.

3.3.3 Resole Resins

When an excess of formaldehyde is used, resoles are formed; the chemistry is
illustrated Figure 3.2.
Under alkaline conditions, trihydroxy methyl phenol forms initially, which has a
functionality of 3. The phenolic hydroxyl is not involved in the reaction. This product
is referred to as resole or A-stage resin. Since the resin is monomeric, it has a low
viscosity and can be used to readily impregnate the reinforcement. In early applica-
tions the reinforcement was paper, which is a discontinuous cellulose fibrous matt. As
shown in Figure 3.2, on further heating the trihydroxy methyl phenol monomers react
with the loss of water and build up into a crosslinked polymer. The reaction can be
interrupted at any stage to provide a resitol or B-stage resin. The resitol properties can
be adjusted in this way, but it retains its reactivity and the volatile condensation water
is removed in a controlled manner. Thus, a pre-impregnated reinforced layer (or ply)

[Link] Published online by Cambridge University Press

 84 Matrices

OH

HOCH2 CH2OH Resole (A-stage)

CH2OH

Gradual increase f=3

in melt and Gradual Resitol (B-stage)
solution viscosities increase Heating (soluble, fusible)
prior to gelation degree of
-H2O
branching

OH

CH2 CH2 Resit (C-stage)

(insoluble, infusible)
CH2

Figure 3.2 Schematic of the chemistry of phenol formaldehyde resole resins (alkali catalysed).

Figure 3.3 Schematic of ply lamination of resole impregnated sheets for decorative laminate
manufacture: an early example of ‘prepreg’ technology.

with residual reactivity can be prepared. A laminate can be prepared by stacking the
pre-impregnated plies for consolidation under pressure and heat. Curing occurs by a
continuation of the reaction when an insoluble or infusible resit or C-stage resin is
formed. This technique was developed to ensure that the volatile condensate could be
controlled and voidage limited. Paper laminates were used in the electrical applica-
tions and subsequently as a means of manufacturing decorative laminates. Phenolic
resins tend to yellow and brown to differing degrees, so water white resins were
needed for the latter. These are usually the analogous urea–formaldehyde (UF) or
melamine–formaldehyde (MF) resins. Figure 3.3 illustrates the manufacture of a
decorative laminate.

[Link] Published online by Cambridge University Press

 3.3 Curing of Thermoset Resins 85

Figure 3.4 Bisphenol-A-based benzoxazine resin (Araldite MT 35600).

Resole resins are used for high-performance laminates from glass and carbon fibres,
mainly as textile reinforcements where fire resistance is the main requirement. The main
difficulties for structural composites revolves around the need for low voidage, which is
not easy to achieve due to the water released during cure. To achieve low voidage, high
pressures are employed to prevent bubble formation by keeping the water solubilized.
This has the knock-on effect of reducing the matrix Tg by plasticization.
Because of the potential high-temperature and fire resistance of phenolic matrix
composites, other related systems have been evaluated. For example bisoxazoline and
benzoxazine resins (Figure 3.4) have been developed to meet these disadvantages.
Ishida and Agag [4] have recently reviewed the range of resins available. Figure 3.4
shows the structure of a commercial resin for homo- and copolymerization with epoxy
resins (see Section 3.4) into a cured resin with a highly aromatic molecular structure
akin to the phenolics.
Resole-type phenolics are also used as a precursor for a carbon or graphite matrix
for so-called carbon (fibre)–carbon ceramic composites.

3.3.4 Novolak Resins

The paper laminates described above were the basis of the electrical industry when it
developed in the late nineteenth and early twentieth centuries, but the technology was
limited to simple shapes. There was a need to develop moulding compounds for
moulding into more complex shapes. Baekeland is credited with the development of
the technology in 1909 (the first patent was in 1907), which was invented in Belgium
and commercialized in the USA. Swinburne, who had developed the resole approach,
carried out similar research in the UK, but missed out on the patent by days.
In this approach the phenol is reacted with a deficiency of formaldehyde so that a
material with only short linear molecules is formed. Acidic catalysis is employed to
control the chemistry. The product is referred to as a novolak (Figure 3.5).
The novolaks have 5–7 phenolic groups per chain as a result of the excess phenol
used in the reaction. The deficiency in formaldehyde content is addressed by adding
hexamethylenetetramine (HMTA), the condensation product of formaldehyde and
ammonia. The compound can be thermally shaped in a mould. The degree of flow
in the tooling and hence complexity of the shape can be adjusted by pre-reacting the
novolak with HMTA in the compounding stage, usually on a hot two-roll mill, to
produce a moulding powder. The pressures employed in the moulding process are

[Link] Published online by Cambridge University Press

 86 Matrices

Figure 3.5 Simplified structure of a novolak phenolic resin showing ortho- or 2-substitution.
The –CH2– groups can also be attached at para- or 4-position of the phenol ring. n has a value
of 2–5.

determined by the extent to which the ‘cure’ volatiles are lost in the compounding
stage. The volatiles released are water and ammonia. Some HMTA residues are also
retained in the molecular structure of the resin, which explains the colour that
develops during curing. The other critical additive is the reinforcement, which can
be either fibrous (e.g. cotton flock, wood flour, or glass fibre) or particulate (e.g.
ground minerals or glass beads). The early synthetic plastics utilized composite
concepts, as discussed in Chapter 1.

3.3.5 Resins for Composite Manufacture

The phenolic resins have much potential for use for fibre composites, but there are
severe disadvantages: (1) curing chemistry is a condensation reaction involving the
loss of H2O, which is difficult to control; (2) moulding under high pressure is used to
keep the water in solution and prevent bubble formation – therefore, the manufacture
of void-free laminates is difficult; and (3) the resin is brittle and not easily toughened.
Two resin systems that cure without the loss of water or other volatiles have developed
alongside the application of fibre-composite materials. These are the unsaturated
polyester and related resins and epoxy resins. The latter can be cured by a variety of
mechanisms, without volatile formation, using step-growth or chain-growth polymer-
ization. The unsaturated polyester resins were responsible for the development of
‘glass fibre’ or ‘fibreglass’ reinforced plastics because they could be cold cured using
addition polymerization.

3.3.6 Addition Polymerization

The curing of unsaturated polyester, vinyl ester, and urethane acrylate/methacrylate
resins, as well as some epoxies employing catalytic systems, occurs by addition
polymerization. Addition polymerization is a chain reaction with the following steps:
Initiation : I ¼ 2R
Propagation : R þ M ¼ P1 
P1  þ M ¼ P2  up to Pn  (3.13)
Transfer : Pn  þ SH ¼ Pn H þ S
Termination : Pn  þ Pm  ¼ Pnþm or Pn þ Pm ,

where I is an initiator (often referred to incorrectly as a catalyst); R* is an active

species (free radical ðR Þ, cation ðRþ Þ, anion ðR Þ, or complex) formed from the

[Link] Published online by Cambridge University Press

 3.4 Epoxy Resins 87

dissociation (or reaction) of the initiator; P is the propagating active polymer; Pn , Pm ,

and Pnþm are unactive polymers of degree of polymerization n, m, and n + m,
respectively; SH is a transfer agent, which could be a monomer, polymer, initiator,
or additive; S is an active species formed from the transfer agent after the transfer of
the H atom to the growing active polymer to form an non-active polymer, Pn H. Added
transfer agents are used to control the molecular weight of the resulting polymer.
The reason for including this chemistry in the book is to ensure that composite
engineers understand how the nature of the curing of resins used as matrices deter-
mines the structure and properties. For example, the step-growth systems usually need
a post-cure stage to achieve the degree of crosslinking. On the other hand, with
addition polymerization each initiated chain grows until it is terminated so that
unreacted monomer is always present, which is in contrast to step-growth polymer-
ization. Post-cure is required to ensure that the free monomer concentration is
minimized. Furthermore, the differing reactivity of the chain active centres towards
co-monomers leads to a range of molecular structures from alternating to homopoly-
mer blocks. The differing fracture properties of unsaturated polyesters and vinyl esters
can in part be attributed to this phenomenon.

3.4 Epoxy Resins

Epoxy resins is a generic term for resins that contain the epoxy or epoxide groups that
can react with a ‘hardener’ in a step-growth manner or polymerize catalytically in an
addition polymerization.
There are two principal types of epoxy resin, glycidyl ethers or glycidyl amines,
represented by the common examples shown in Figure 3.6.
The glycidyl ether DGEBA is available with differing degrees of polymerization
and with n ¼ 0–34. The mechanical properties of these resins vary significantly
across the range because the molecular length between the reactive epoxy groups
increases with n. Thus, the crosslink density is reduced, as is the resin modulus, but
the failure strain increases. Since DGEBA is difunctional in epoxy groups, the
thermomechanical properties Tg, Em, and Gm are low in magnitude, while the linear
thermal expansion coefficient, αm, will be high because of the low crosslink dens-
ities. As a result, these resins are used mainly in industrial composites. The presence
of high concentrations of –OH groups in DGEBA resins means they find applica-
tions as adhesives.
The cured glycidyl amines such as TGDDM have higher crosslink densities and
therefore superior thermomechanical performance. As a result, they find application in
aerospace structural composites.
Higher crosslink density can be achieved with multifunctional glycidyl ethers such
as the novolak epoxy resin (which is a reaction product of epichlorohydrin and a
novolak phenolic resin; epichlorohydrin is the standard reactant for synthesizing
epoxy resins). There are also a range of mixed glycidyl ether and glycidyl amines,
of which the most common is triglycidyl p-aminophenol (TGAP).

[Link] Published online by Cambridge University Press

 88 Matrices

O O O
O-CH2-CH-CH2 O-CH2-CH-CH2 O -CH2-CH-CH2

CH2 CH2
n

Novolak epoxy resin

o
o CH2CH CH2
CH2 CHCH2O N
CH2CH CH2
o

Triglycidyl p-aminophenol (TGAP)

Figure 3.6 The structures of five common epoxy resins: diglycidyl ether of bisphenol A
(DGEBA), where n ¼ 0–34; diglycidyl ether of bisphenol F (DGEBF); tetraglycidyl 4,40 -
diaminodiphenylmethane (TGDDM); novolak epoxy resin; triglycidyl p-aminophenol (TGAP).

3.4.1 Curing of Epoxy Resins

The epoxy group can be polymerized either by a chain polymerization initiated by a
curing agent or by step-growth copolymerization with a hardener.

[Link] Curing Agents

Often referred to as catalytic curing agents, these molecules open the epoxy group in
the initiation step and chain growth occurs by either anionic (negative) or cationic
(positive) addition polymerization, as described in Section 3.3.6. The product has a
relatively non-polar molecular structure, which is a ‘polyether’. A classic curing agent
is boron trifluoride, as an etherate or amine adduct, which initiates cationic chain
propagation, as shown in Figure 3.7.
Tertiary amines initiate anionic polymerization of the epoxy groups, but this is
often employed in combination with a co-curing agent (Figure 3.8). A typical example
is the tertiary amine-initiated anhydride systems, such as nadicmethylene tetrahy-
drophthalic anhydride (NMA). Please note that anhydride curing agents alone require
trace concentrations of water in the initiation stage, whereby the anhydride reacts to
form the analogous acid which initiates an anionic chain reaction. The cured structure
of the network is one of a ‘polyester’ type. A common tertiary amine used with NMA
is benzyl dimethylamine (BDMA).
The cationic homopolymerization of the epoxy groups using a tertiary amine
creates a polyether crosslinked structure (Figure 3.9). Addition polymerization tends
to provide relatively low-polar cured structures so that intermolecular cohesive forces

[Link] Published online by Cambridge University Press

 3.4 Epoxy Resins 89

BF3:NH2Et H+ + [BF3NR2]–

H O OH
O O+ +
H+ + O CH2 CH

OH
+
O CH2 CH O CH2 CH

Figure 3.7 The catalytic cure of an epoxy resin using cationic initiator, BF3 amine adduct, to form
a polyether chain crosslink structure.

O O O
+ +
C C NR3 C NR3
O
+ NR3
O

C C O– C–O–CH2–CH–CH2

O O O O–

O O
+ C
C NR3
O
C
O O
C CH2–CH–CH2
O O
O
C

O–
C
O

Figure 3.8 Tertiary amine-initiated anionic copolymerization of anhydride ring and epoxy groups
during anhydride curing. A polyester chain is formed.

tend to ensure that the thermomechanics of the cured resins have low values unless
high degrees of crosslinking are introduced by post-curing at elevated temperatures.
We see in the next section that hardeners in general introduce polar groups into the
molecular structures of cured epoxy resins.

[Link] Published online by Cambridge University Press

 90 Matrices

R3N: O
CH2–CH

R3N⊕
CH2–CH
OΘ O
CH2–CH
CH2–CH
O–CH2–CH
OΘ A ‘polyether’ structure

Figure 3.9 Catalytic curing of epoxy resin using tertiary amine-initiated anionic polymerization.
A polyether structure is formed.

H H O
N – CH2CH2 – N CH2 – CH–
H DIAMINE H O

CH2 – CH–

OH OH

CH – CH2 CH2 – CH

OH N – CH2CH2 – N OH

CH – CH2 CH2 – CH

Figure 3.10 Typical cure reaction of epoxy groups with a diamine hardener. The first stage
involves a primary amine and subsequently with the remaining secondary amine groups. The
latter are less reactive so complete reaction (i.e. full crosslinking) as shown is unlikely.

[Link] Hardeners
Curing with hardeners occurs by step-growth addition reaction with amines, carbox-
ylic acids, thiols, and cyanate esters. Typical hardeners are either aliphatic multifunc-
tional amines or aromatic diamines. Dicyandiamide (DICY) is a multifunctional
hardener that is commonly used as a latent ‘curing’ agent for epoxy resins and
especially for prepreg applications.
A common curing agent for epoxies often used for adhesives is a diamine.
Figure 3.10 gives the simplest, ethylene diamine, which is a typical representative
of the range of available and common hardeners. A hardener differs from a catalytic
curing agent because the mechanism involves step-growth addition reactions, as
shown in Figure 3.10.
Cure of epoxy resin using step-growth addition of ethylene diamine (an example
of an aliphatic diamine hardener) involves initial reaction with a primary amine,
–NH2, followed by a subsequent reaction with the secondary amine, –NH–CH2–,

[Link] Published online by Cambridge University Press

 3.4 Epoxy Resins 91

Figure 3.11 1,4 Diamino diphenylsulphone (1,4 DDS): an example of an aromatic diamine
hardener.

which is formed. Reaction of the secondary amines with epoxy groups occurs at a
lower rate so that post-curing at higher temperatures can increase the degree of
crosslinking. This is also accompanied by the formation of β-hydroxyl groups
(–CH2CH(OH)–) so that the molecular structure of the crosslinked resin is polar,
providing hydrogen bonding sites for absorbed water. Therefore, these resins are
susceptible to moisture absorption.
Aliphatic diamines such as that shown in Figure 3.10 are highly reactive, enabling
curing at room temperature. Extending the aliphatic chain or introducing aromatic
groups into the structure can modify the rate of cure. Diamino diphenylsulphone
(DDS) is often used as a hardener for high-performance epoxy resins and can be used
in prepreg systems for high-temperature autoclave processing. The reactivity of 1,3
DDS and 1,4 DDS (Figure 3.11) also differ, which increases the opportunities for
designing epoxy resins with controlled curing.

[Link] Latent Curing Systems

For prepreg systems, a latent curing system is required so that the pre-impregnated
fibre sheet can be stored and the curing of the resin can be thermally initiated. Early
systems employed mixed curing agents such as the BF3 adducts in combination with
DDS, which becomes active at about 160
C. The viscosity of the resin and hence the
tack of the prepreg could be adjusted by advancing the DDS/ epoxy reaction in situ.
Dicyandiamide (DICY) is also used to provide latency of cure in a prepreg.
Micronized DICY is dispersed into the liquid resin. Since the hardener is not soluble
at room temperature, prepreg with improved shelf life can be obtained. The viscosity
and tack of the prepreg is adjusted by dissolving a ‘thermoplastic’ such as polyether
sulphone (PES) at temperatures below that of the DICY. In this way the fibres can be
impregnated with a low-viscosity resin to maintain their alignment.
The cure temperature is determined by the melting point of DICY, which occurs at
165
C. High- and low-temperature curing systems are available: ‘diuron’ or ‘mon-
uron’ amine accelerators can be added to reduce the curing temperature to 120
C.
Low-temperature systems have clear advantages, such as control of the exotherm
during curing. Careful control of the cure schedule is required for the production of
good-quality void-free laminates. Thus, a dwell at the temperature of minimum resin
viscosity should be included at typically 90
C. A disadvantage of low-temperature
cure of DICY-based systems is the increased possibility of residual DICY. There is
also a potential for inconsistent resin microstructures, which can be improved by
ensuring that the DICY is micronized into a uniform fine particle size. DICY is used at

[Link] Published online by Cambridge University Press

 92 Matrices

O
+ H2N C NHCN
3
NH

CH OH
OH
CH2
CH CH2
N CH N CN
CH CH2
NH
OH

C N

CH OH
OH
CH2
CH CH2
N CH N CN
CH CH2
NH
OH
C NH

Figure 3.12 The reactions involved in the cure of epoxy resins with dicyandiamide (DICY).

concentrations of 4–6 phr, indicating that the mechanism of cure involves its multi-
functionality. Dicyandiamide exists in two molecular tautomers:

H2N C N C N NH2 C NHCN

(3.14)
NH2 NH
Reaction with an epoxy group is therefore complex (Figure 3.12) and involves the
nitrile as well as the primary amine groups. It is also likely to involve other
crosslinking mechanisms.

[Link] Co-curing and Other Curing Agents

There is a large range of curing agents for epoxy resins, many of which are employed
for adhesives and/or coatings. It is beyond the scope of this book to provide complete
details. Relevant examples used for composite matrices only are included. Full details
of epoxy resin systems are given elsewhere [5,6]. Here, we list additional curing
agents that are commonly employed for composite manufacture.

Tertiary Amines
For industrial composites, BDMA and tris(dimethylaminomethyl) phenol (Figure 3.13)
are often used in combination with other curing agents, such as NMA. BDMA is also
used as a co-catalyst with polysulphides and polymercaptans, which tend to have
relatively long chain lengths and enable the crosslink density to be controlled.

[Link] Published online by Cambridge University Press

 3.4 Epoxy Resins 93

(a) (b)

Figure 3.13 Commonly used tertiary amines employed as co-curing agents: (a) BDMA; (b) tris
(dimethylaminomethyl) phenol.

Imidazoles
These activators introduce high reactivity and promote curing at lower temperatures.
On the other hand, trisubstituted ureas or urons, which are reaction products of
isocyanates with dimethylamine, exhibit outstanding latency of cure and find applica-
tion in one-pack adhesives.
Carboxylic acids and anhydrides and related carboxylic functional polyester resins
are also employed to cure epoxy resins as discussed above.
Sulphur-containing curing agents such as mercaptan-terminated polysulphides can
be used as co-curing agents with other long-chain polyamides and cycloaliphatic
amines, and introduce high degrees of flexibility because of the low crosslink density.
In this way, the extensibility and toughness of the resin can be improved.
Quaternary phosphonium salts such as alkyl and triphenylphosphonium halides are
effective latent accelerators for anhydride and phenolic resin-cured epoxy resins. The
latter systems are effective at advancing the cure.
Cationic salts such as triarylsulphonium are used as photoinitiators.

[Link] Reactive Diluents

Diluents are added to epoxy resins as process aids to reduce the viscosity for fibre
impregnation. They can be either unreactive or reactive. The former have limited
value in composites technology since the properties of the cured resin will be reduced
by the presence of a non-bound diluent. Reactive diluents are small molecules that
enable the viscosity to be reduced but on curing they are involved in the reaction with
the epoxy resin. While the reactive functionality can be other than epoxy, the majority
are mono-functional epoxies that can be reacted into the network crosslinks. Examples
include glycidyl ethers such as butyl glycidyl ether and phenyl glycidyl ether, or
oxiraine compounds such as styrene oxide or octylene oxide.

[Link] Toughening Techniques

Cured epoxy resins can exhibit high fracture toughness, but this is generally at the
expense of the achievable Young’s modulus. For composite materials we generally
require the matrix to have the highest stiffness, which is obtained from high degrees of

[Link] Published online by Cambridge University Press

 94 Matrices

crosslinking, but these systems will have relatively low fracture toughness. The
fracture toughness of these resins can be improved by adding either mineral fillers,
rubber particles, or thermoplastic particles, or introducing a soluble thermoplastic or
linear polymer which phase-separates during curing to provide the morphology with
the highest fracture toughness.
Mineral fillers are particles that increase the fracture toughness of epoxy resins; the
mechanism is considered to arise from (1) crack pinning, (2) bowing of the crack
front, and (3) break-away. Thus, K1c increases with particle volume fraction, Vp,
reaching a plateau then decreasing; i.e. an optimum Vp can be expected. The effect
of particle size is unclear because as the diameter decreases the surface area becomes
more significant and influences the structure of the glassy polymer in the
interphase region.
For composite materials, recent research has concentrated on increasing the modu-
lus of the matrix resin within a laminate in order to improve the compression strength
of the material by preventing fibre buckling. A reduction in the thermal strains arises
from the lower thermal expansion coefficient of the ‘filled matrix’.
Rubber particles are typically carboxyl-terminated butadiene–acrylonitrile rubber
(CTBN), and are employed at <20% by weight. The terminal carboxylic acid groups
provide a mechanism of interfacial bonding to the rubber particles. Butadiene rubber
has a Tg of –75
C and therefore exhibits high resilience at room temperature.
Butadiene–acrylonitrile rubber, generally referred to as nitrile rubber, has a typical
Tg of –38
C (for a copolymer with 18% acrylonitrile) and is much less resilient at
room temperature.
Efficient toughening is achieved with the correct two-phase morphology. Simple
dispersion does not provide the required microstructure, so it is necessary to dissolve
the rubber in the epoxy for dispersion at the molecular level. Phase separation occurs
during curing as the molecular weight of the epoxy increases through crosslinking.
The rubber phase is finely dispersed and bonded to the epoxy continuous phase.
In thermoplastic modification epoxy resins used as matrices for composites can be
toughened using appropriate linear polymers, which phase-separate into a co-
continuous morphology. Three differing morphologies are possible, depending on
the concentration of the modifier: (1) epoxy continuous phase with dispersed thermo-
plastic ‘particles’ (2) co-continuous morphology; and (3) thermoplastic continuous
phase with dispersed epoxy ‘particles’. To achieve a co-continuous morphology a
precise concentration of thermoplastic of defined molecular weight and structure is
required. Typically, end-capped linear PES or polyether ether sulphone (PEES) meet
these requirements. To understand this we will consider the thermodynamics of
polymer solubility.

3.4.2 Compatibility of Polymers

The compatibility of a resin with a linear polymer can be understood using the concept
of solubility parameters based on the thermodynamics of mixing described in the
Gibbs equation:

[Link] Published online by Cambridge University Press

 3.4 Epoxy Resins 95

ΔGm ¼ ΔH m  TΔSm , (3.15)

where ΔGm , ΔH m , and ΔSm are the changes in free energy, enthalpy, and entropy of
mixing, respectively.
For complete mixing, ΔGm has to be negative, while ΔSm will tend to be positive.
For small molecules, ΔSm will be large and positive so that –TΔSm will readily ensure
mixing when there is no interaction between the components and ΔHm ¼ 0. With
polymer–polymer mixing the change in entropy is small and close to zero so that for
polymer–polymer compatibility, ΔHm should be negative. This arises when the two
components interact, which can be defined by eq. (3.16):

ΔH m ¼ n 1 n 2 ðδ1  δ2 Þ2 V, (3.16)
where δ1 and δ2 are the solubility parameters of components 1 and 2, n 1, n 2 are the
volume fractions of the two components, V is the total volume of the system, and δ is
defined as the square root of cohesive energy density representing the energy of
interaction between like molecules. Polar molecules will have high values of δ. For
mixing, the interaction between components 1 and 2 must equal that between com-
ponents (1 and 1) and (2 and 2), otherwise the molecules will prefer themselves. Thus,
for mixing ðδ1  δ2 Þ and hence ΔH m will be zero. This simply describes the maxim
‘like dissolves like’.
The solubility parameter can be readily calculated from its molecular structure
according to the simple equation
XG
δ¼ρ , (3.17)
M
where ρ is the density, M is the molecular weight, and G is the molar attraction
constant for an individual ‘molecular’ element of the molecule. The sum of the
contributions from the structural elements provides the attraction energy density for
that molecule.
Equation (3.17) needs a correction factor for polar molecules and a more accurate
value needs to include all the molecular interaction components of cohesive energy:
dipole, polar, and hydrogen bonding. The total solubility parameter δt is given by

δ1 ¼ ðδ2d þ δ2p þ δ2h Þ1=2 , (3.18)

where δd, δp, and δh are the dipole, polar, and hydrogen bonding contributions to the
solubility parameter, respectively.
Calculation of solubility parameters according to Hoy [7] and van Krevelen and Te
Nijenhuis [8] enables the individual contributions to the total solubility parameter to
be obtained. The total solubility parameter, δt, of DGEBA is in the range
20.6–21.24 MPa½, depending on the molecular weight. The dipole component, δd,
decreases from 16.47 to 16.22 MPa½ with molecular weight, while the polar compon-
ent increases because of the higher number of polar OH groups. Table 3.1 shows that
after curing the value of δt tends to be higher, at 22–24 MPa½. The hydrogen bonding
component is fairly constant at 8–10 MPa½.

[Link] Published online by Cambridge University Press

 96 Matrices

Table 3.1. Calculated solubility parameters (δt) and the hydrogen bonding components (δh) of selected epoxy
monomers, hardener (DDS), and typical cured epoxy networks [9] and comparative polymers

δt (MPa½) δt (MPa½) δh
Epoxy monomer segment/polymer Function (calculated) (measured) (MPa½)

DGEBF Monomer segment 22.36 – 9.75

DGEBA Monomer segment 21.26 – 8.13
TGAP Monomer segment 22.92 – 10.34
TGDDM Monomer segment 21.84 – 8.88
4,4’, DDS Monomer segment 27.18 – 9.18
TGAP/DDS (2:1) Cured epoxy resin 24.06 – 10.25
TGDDM/DDS (5:1) Cured epoxy resin 22.30 – 8.95
TGDDM/ TGAP /DDS (19:13:10) Cured epoxy resin 22.89 – 9.4
DEGBA/NMA (1:2) Cured epoxy resin 24.0 – –
PES Thermoplastic flow control 24.7 23.0 –
and toughening agent
PEE) Thermoplastic flow control 23.5 – –
and toughening agent
Poly bisphenol A sulphone Comparative thermoplastic 21.7 21.2 –
Polycarbonate Comparative thermoplastic 20.3 20.4 –
Polybutadiene Component of nitrile rubber 16.8 17.0 –
toughening agent (CTBN)
Polyacrylonitrile Component of nitrile rubber 26.3 26.2 –
toughening agent (CTBN)

The general rule of thumb for identifying compatible pairs is given by:

δ1 ¼ δ2  2 MPa1=2 : (3.19)

Therefore, for the rubber to dissolve in the epoxy resin it needs a value δt in the range
20–26 MPa½. As shown in Table 3.1, butadiene rubber has a value of δt outside of this
range. However, we see that polyacrylonitrile has a value of 26.3 MPa½, so that nitrile
rubber, which is an acrylonitrile–butadiene copolymer, will have a closer solubility
parameter and compatible with a typical epoxy resin. We see that PES has a similar
value to advanced epoxy resins.
Solubility parameters have additive characteristics so that compatibility can be
achieved using mixed systems, as indicated by eq. (3.20):

δ m ¼ x1 δ 1 þ x2 δ 2 , (3.20)

where δm is the solubility parameter of a mixture of components 1 and 2 at mole

fractions of x1 and x2.

[Link] Phase Separation in Polymer Blends

Flory and Huggins’ polymer solution theory [3] is a statistical model in which a
polymer segment is placed, statistically, into a lattice of solvent molecules. The

[Link] Published online by Cambridge University Press

 3.4 Epoxy Resins 97

assumption is that the segment of a polymer chain has a similar molecular size to the
solvent molecule. Equation (3.21) describes the free energy of mixing, ΔGm:

ΔGm ¼ RT½n1 lnϕ1 þ n2 lnϕ2 þ n1 ϕ2 χ , (3.21)

where n1 is the number of solvent molecules, n2 is the number of segments in the

polymer, and ϕ1 and ϕ2 are the mole fractions of sites 1 and 2. χ is the Flory interaction
parameter, which has both enthalpic (χH) and entropic (χS) components:

χ ¼ χH þ χS: (3.22)

The enthalpic component is related to the solubility parameter according to eq. (3.23):

V 1 ðδ1  δ2 Þ2
χH ¼ , (3.23)
RT
where V1 is the molar volume of the solvent.
The development of eq. (3.21) to understand the criteria for solubility of a polymer
is beyond the scope of this book. However, for solubility it is understood that χ has a
value of ½ when the entropic and enthalpic components are in balance and the
solution can be considered to be a mixture at the molecular level. We also see in eq.
(3.23) that χH is temperature-dependent. As a result, a polymer–solvent blend will
exhibit a temperature above which it has one phase (i.e. a solution), which is referred
to as the theta temperature, θ. Below this temperature the polymer will not be soluble.
Theta conditions refer to a dilute solution at the theta temperature, Tθ and χ ¼ ½, and
are considered to be ideal.
In the context of polymers employed for composite matrices it is relevant to
highlight the effect of molecular weight of components on phase separation in
polymer blends. The Flory–Huggins theory can be applied to binary polymer–polymer
systems, as discussed by Scott [10], who estimated the partial free energies of mixing:

 
m1
ΔF m1 ¼ RT lnn 1 þ 1  n 2 þ m1 χn 2 ,
2
(3.24)
m2

 
m2
ΔF m2 ¼ RT lnn 2 þ 1  n 2 þ m2 χn 21 , (3.25)
m1

where ΔFm1 and ΔFm2 are the partial molar free energies of mixing for a polymer-1–
polymer-2 binary system. m1 and m2 are essentially the degrees of polymerization of
the two components at volume fractions of v1 and v2.
The critical value of the Flory interaction parameter for solubility/phase separation,
χc, can be obtained by differentiation and setting ΔFm ¼ 0:


2
1 1 1
χc ¼ ½
þ : (3.26)
2 m1 m½2

For polymer–solvent systems χc ¼ 0.5, but for polymer–polymer systems χc  0.01.

Thus, polymers of infinite molecular weight will tend to be incompatible unless there

[Link] Published online by Cambridge University Press

 98 Matrices

is a specific interaction (i.e. ΔHm is negative). In the context of matrices for fibre
composites, which are mainly crosslinked network polymers such as epoxy resins,
curing causes the molecular weight to increase and eventually become infinite. As a
result, any dissolved linear polymer will tend to phase-separate during cure. This is
referred to as reaction-induced phase separation and is the basis for creating the
correct morphology for high fracture toughness when thermoplastics are blended into
thermosetting resins. For example, when PES is employed as the flow control and/or
toughening agent at a concentration of about 20%, a co-continuous microstructure
forms during cure, while below that the PES will phase-separate with a continuous
epoxy resin phase. At higher concentrations the epoxy component will be discontinu-
ous. Examination of the thermodynamics above demonstrates that the morphology
achieved can be dependent on the cure schedule and the temperatures employed, as
well as the molecular weight of the toughening agent.

3.4.3 Cure Property Diagrams

The glass transition temperature, Tg, of a thermosetting resin is not so easy to measure
because two major parameters contribute to the formation of a glass: the rates of
crosslinking and cooling. As shown in Figure 3.14, a linear polymer will exhibit a
linear change in volume with temperature on cooling until the glass transition tem-
perature is reached, when the rate of contraction decreases. On rapid cooling, a glass
will form at a higher temperature. A polymeric material consists of randomly coiled
molecular chains so that the reorganization of segments in the chain occurs with
cooling. To achieve the lowest statistically possible structural arrangement an infini-
tesimally slow cooling rate is required. Some polymers have a regular structural

Liquid

Fast cool
Volume

Glass

Slow cool
Tg1 Tg2

Temperature

Figure 3.14 Schematic showing the effect of cooling rate on the glass transition
temperature, Tg, of a linear polymer.

[Link] Published online by Cambridge University Press

 3.4 Epoxy Resins 99

arrangement at the segmental dimension and will crystallize. Irregularly structured

polymers cannot crystallize and form a glass.
Since most polymers do not normally crystallize completely (i.e. 100%), they
exhibit melting and glass transition temperatures. Typically, Tg occurs in the range
0.5–0.8 Tm. Thus, a glass with the lowest possible Tg will exhibit the smallest
‘unoccupied’ volume, which is referred to as free volume and defined as the statistical
volume for chain–chain crossover, unperturbed by interactions [11]. Therefore, the
unoccupied volume in a polymer will consist of free volume and nano-voids [12].
However, a thermosetting polymer such as an epoxy resin differs from a linear
polymer in that they are generally non-polymeric, becoming polymeric only during
curing. There is an additional shrinkage mechanism, resulting from the crosslinking of
the molecules during curing. This results in a different behaviour on cooling, as shown
in Figure 3.15. A glass forms at a higher transition temperature, Tg2, with a larger
unoccupied volume when a resin with a higher extent of crosslinking (degree of cure)
is cooled. The cured resin can have a lower density despite a higher glass transition
temperature. Thus, the resin in the as-cured state will be in a non-equilibrium state. We
see in Chapter 9 that this is responsible for the anomalous moisture absorption of some
resin systems, which occurs under thermal cycling.
Gillham and coworkers [13–15] have studied in detail the properties of matrix
resins during representative cure schedules using thermal analysis. In particular,
torsional braid analysis enables the conversion of a liquid resin into a solid during
curing and subsequent cooling. From these studies, time transformation diagrams and
cure–temperature property diagrams could be derived.
As discussed in Section 3.3.1, the cure of an epoxy resin involves the gradual
increase in average molecular weight as the relatively low molecular weight

High
Liquid
Low
Degree of cure
Volume

Glass

Tg1 Tg2 Tc
Temperature

Figure 3.15 Schematic showing the effect of cooling and degree of cure on the glass transition
temperature, Tg, of a thermosetting polymer. Tc is the cure temperature.

[Link] Published online by Cambridge University Press

 100 Matrices

monomers or oligomers react with the hardeners. The following events occur during a
cure schedule with rising temperature:
1. gelation
2. phase separation
3. vitrification (glass formation)
4. devitrification (occurs when the temperature exceeds the lowest Tg0, the lowest Tg
of the system)
5. degradation (at high temperatures thermal chain bond scission is possible: resin cure is
exothermic and control of cure and hence the actual temperature of the part is essential).
Since cure reactions are exothermic, the temperatures of the part, especially in thick
sections and at different positions in the moulding, can vary and be significantly
elevated above the cure temperature, Tc. As a result, to avoid excessive rises in
temperature from exothermic processes, the initial cure temperature is often kept
relatively ‘low’, then followed by a post-cure at an elevated temperature to achieve
full cure. The temperature schedule will depend on the resin system employed.
A time–temperature–transformation (TTT) diagram can be obtained from the experi-
mental determination of the times to gelation, tgel, and vitrification, tvit, under isothermal
conditions. Full details are beyond the scope of this text, but a complete description is
given elsewhere [5,16–18]. Figure 3.16 gives an example of a TTT diagram.
Fu
l

Char
lc

Gel Vit
ur

rubber rifi
e

cat
Sol/gel ion
rubber
Devitri Tg∞
ficatio
G

Tc n
el

Gel gla
at

ss
io
Vi
n
tr

Ph
ifi
ca

as
tio

e
se
n

pa Sol/gel glass
ra
tio
Liquid n
gel Tg

Tg0
Sol glass

log (tc, gel)

log (tc)

Figure 3.16 Time–temperature–transformation (TTT) diagram. Tc and tc are the cure temperature
and time, respectively. Tg, Tg0, Tg∞ are the glass transition temperatures at times tc, t0 (the resin/
curing agent mixture) and the fully cured resin (tc ¼ t∞). gel Tg is the Tg at tc,gel ¼ tc,vit, the cure
times for gelation and vitrification, respectively (the gelation and vitrification lines). Sol glass is
soluble, whereas gel glass is crosslinked and only swells in a solvent. The liquid region is
represented by iso-viscosity contours (differing by 10) for a homogeneous system. Char
represents the degradation region [16].

[Link] Published online by Cambridge University Press

 3.4 Epoxy Resins 101

A TTT diagram describes the complexity of the cure of the resin. Initially the cure
temperature needs to exceed sol Tg (Tg0) for the reactions to proceed. As the complex-
ity of the molecular structure increases with cure, so does the Tg (tc) (tc is the cure time
at a constant temperature, Tc) until it reaches the cure temperature Tc, when it vitrifies.
As Tc approaches Tg (tc) the molecular mobility decreases and the rate of cure
decreases according to a ‘second’ or higher-order kinetic law, eventually becoming
diffusion-controlled. Subsequently, the cure reactions are quenched, despite the pres-
ence of reactive functional groups. These occluded reactive sites provide a latent cure
mechanism because, as discussed in Section 3.3.3, the cure mechanisms can be
reactivated on raising the temperature.
The TTT curve shows the relationship between gelation and glass formation during
curing and provides the time and temperature dependence and demonstrates how full
cure can be achieved. Thus, it is common practice to employ a post-cure at an elevated
temperature. Phase separation of a rubber or thermoplastic modifier occurs as the cure
progresses, as shown by the increased viscosity identified by a series of iso-
viscous lines.
Gillham and coworkers [13–15] have discussed the TTT diagrams for epoxy resins.
However, Prime [18] has reviewed the application of TTT concepts to the cure of
other thermosets, phenolic, amino, allyl, and unsaturated polyester resins.
Wang and Gillham [19] have extended the concept to the cure–temperature prop-
erty diagrams to represent structural changes (Figure 3.17). In particular, they include
β-transitions, which strongly influence the fracture toughness of the cast resin and the
elastic modulus. The cure parameter, C, is given by
T g  T g0
C¼ : (3.27)
T g∞  T g0

Sol/gel rubber region

Tg∞

Sol fluid region Glass transition

ion
ficat region
Vitri
Temperature

Tg0 Gelled glass region I

cTg

Ungelled glass ition Tβ∞

β-trans
Gel Tβ region

Gelled glass region II

Gelation
Tβ0 gel Tg
0 1
Cure parameter, C

Figure 3.17 Cure–temperature property (CTP) diagram illustrating the influence of the
temperature range for the glass–rubber transition and the β transition. cTg is the onset
temperature [16].

[Link] Published online by Cambridge University Press

 102 Matrices

0.5

0.45 Glass
0.1 /s
0.4 0.5 /s
MY721 at 5
1 /s
0.35 strain rates
5 /s
Loss tangent

0.3 10 /s
0.25

0.2
Post-curing
0.15 peak
1 Beta

0.05

0
-200.0 -100.0 0.0 100.0 200.0 300.0
Temperature (qC)

Figure 3.18 Dynamic mechanical thermal analysis (DMTA) spectrum of a DDS-cured

TGDDM (MY721). The post-curing peak is present after 180
C cure but absent after 200
C
cure. The beta peak at Tβ is also demonstrated (from J. P. Foreman).

Figure 3.18 shows a typical dynamic mechanical thermal analysis (DMTA) spectrum
for a DDS-cured epoxy resin showing the β-transition, which occurs at –50
C. The Tg
appears to have two components after cure at 180
C. The so-called post-curing peak
is not present after cure at 200
C. The origin of the low-temperature peak is not clear,
but could arise from the presence of ‘incompletely’ reacted components of the MY721
epoxy [20].
The low-temperature beta peak provides a molecular energy absorption mechanism
and therefore introduces a degree of fracture toughness to the resin casting.
Furthermore, the modulus of the glass is also affected. Group interaction modelling
extends eq. (3.1) into a methodology for predicting the thermal and mechanical
properties of polymers [21] from the molecular structure. The model has been applied
to aerospace epoxy resins [22].

3.5 Unsaturated Polyesters Resins for Lamination

Unsaturated polyester resins (UPE) represent a milestone in the development of the

composites industry. These resins are solutions of an unsaturated polyester dissolved
in a reactive diluent, which is mainly styrene. The curing of early resins was subject to
air inhibition, which limited their introduction. This issue (air inhibition) was solved
by ground-breaking work at Scott Bader and Co. Ltd by incorporating paraffin wax in
ppm (parts per million) quantities to provide an oxygen barrier at the moulding surface
[23]. There are two main classes of resin, casting and laminating, and the latter are
most relevant to this discussion.

[Link] Published online by Cambridge University Press

 3.5 Unsaturated Polyesters Resins for Lamination 103

Figure 3.19 Schematic description of a cured unsaturated polyester resin; –A–B– is the polyester,
S is the reactive diluent (styrene) built into a copolymer network.

Polyester laminating resins are viscous, pale-yellow liquids. These resins are
solutions of unsaturated ‘oligomeric’ polyester of low degree of polymerization,
8–10 (i.e. MW  2,000) in an unsaturated reactive diluent, which is commonly
styrene. These resins cure into a crosslinked network polymer by rapid addition
copolymerization of the unsaturated groups in the polyester with those of the reactive
diluent. To tune the thermomechanical properties of the cured resin, the crosslink
density is controlled in two ways:
1. the spacing of the unsaturated groups (–C¼C–) in the oligomeric polyester by
incorporating saturated monomeric units via copolymerization with a
saturated acid;
2. varying the reactivity of the unsaturated groups to styrene during curing by
changing the glycol component in the synthesis, which provides an important
means of controlling the rate of cure during composite manufacture [24–27].
The structure of a cured UPE is often represented schematically, as shown in
Figure 3.19.
While the cured structure given in Figure 3.19 is commonly used, Figure 3.20 is a
better representation because of the average length of the polystyrene bridges
(n  2–3). This is a consequence of the kinetics of the copolymerization, which are
discussed below.

3.5.1 The Synthesis of Unsaturated Polyester

Figure 3.21 provides the molecular structures of the common monomers employed for
the synthesis of polyester.
Maleic anhydride and propylene glycol together with phthalic anhydride are the
most common starting materials. The choice of anhydrides, as opposed to diacids, is
determined by the more rapid esterification and lower number of moles of H2O that
evolve during the polycondensation, as illustrated in Figure 3.22.
The esterification reaction is a slow process involving the distillation of the water of
condensation over 12–20 hours at temperatures up to 200
C in a nitrogen atmosphere.
In batch reactors this is achieved with a nitrogen purge that means the diol often distils
out with the condensate. Esterification is acid-catalysed, but sulphuric acid cannot be
used because of side reactions that result in coloured products. The other common

[Link] Published online by Cambridge University Press

 104 Matrices

(a)
I–I

CH2=CH CH2=CH
B
CH2=CH

CH2=CH
A
CH2=CH
CH2=CH

CH2=CH

I–I
C

(b)
(CH2–CH)n (CH2–CH)n

(CH2–CH)n

POLYESTER POLYESTER
POLYESTER
(CH2–CH)n
(CH2–CH)n

CH2–CH – I

(CH2–CH)n

(CH2–CH)n

POLYESTER

Figure 3.20 Schematics of (a) the structure of uncured unsaturated polyester resin
containing (A) unsaturated PE oligomer, (B) reactive diluent styrene, and (C)
‘catalyst’ system-initiator fragment ‘I’; and (b) the structure of the cured resin
illustrating a ‘simple’ model of the network (n  2–3) [24,28].

[Link] Published online by Cambridge University Press

 3.5 Unsaturated Polyesters Resins for Lamination 105

OH

OH
HO
O
Ethylene glycol
HO OH
OH
Maleic anhydride
Propylene glycol
O
OH
O OH
HO
HO
Diethylene glycol
O
Fumaric acid
HO OH
O

Neopentyl glycol
O

O
HO O O o-Phthalic anhydride
OH

O OH

Dihydroxy dipropoxy bisphenol A

OH
O
Isophthalic acid

Cl O
Cl
O
Cl
Cl O
Tetrachlorophthalic
anhydride

Figure 3.21 Chemical structures of the acids, anhydrides, and glycols used in the synthesis of a
UPE resin [26].

esterification catalyst is butyl titanate, but unfortunately it hydrolyses to titania, which

results in a white-pigmented resin that is generally unsuitable for the synthesis of
laminating resins. Therefore, self-catalysed conditions are employed. The reaction is
continued until an acid value of 30–40 mg KOH/g (of resin) is achieved. From 1 mol
of diol and 1 mol of diacid this is equivalent to an extent of reaction of p ¼ 0.95,
which equates to Pn  20 (see eq. 3.10), but in reality Pn  10 (Mn ¼ 1,400–1,850)
because a 5–10% excess of diol is used. Therefore, a typical recipe might be:

 1 mol phthalic anhydride

 1 mol maleic anhydride
 2.2 mol propylene glycol (1,2 propane diol).
The chemistry of the synthesis is shown in Figure 3.22.

[Link] Published online by Cambridge University Press

 106 Matrices

O
C
CH=CH + +
O H3C CH CH2 OH
C C
O C OH
O O
O

160–190 °C
–H2O

O C CH = CH C O CH CH2 O C C O CH CH2 O
O O CH3 O O CH3

Figure 3.22 Synthesis of unsaturated polyester from phthalic anhydride, maleic anhydride, and
propylene glycol.

The excess diol can be lost during synthesis in the following ways:
1. the diol distils as an azeotrope with the water;
2. self-condensation of the diol into volatile cyclic ethers; and
3. self-condensation into higher ether diols (e.g. diethylene glycol from ethylene
glycol), which can be incorporated into the resin structure.
The acidity of the mixture and the temperature reached by the reactants determine the
structure of the polyester because of the extent of cis–trans isomerization of maleic
anhydride.

[Link] Cis–Trans Isomerization of Maleate to Fumarate

During Polycondensation
The most significant aspect of the synthesis that impacts on the reactivity of the resin
and the thermomechanical properties of the cured resin is the extent of the cis–trans
isomerization of the maleic anhydride into fumaric acid (Figure 3.23), which is
incorporated as fumarate unsaturated groups in the polyester chain. The trans
–C¼C– bonds of the fumarate groups in the UPE chain have a higher reactivity to
the styrene diluent in free radical copolymerization. Fumaric acid is also used to
synthesize resins with higher reactivity during curing, but the additional costs associ-
ated with removing the extra mole of water is not always justified by the benefits.
Therefore, controlling the isomerization of maleate to fumarate is an important factor
in resin production. Figure 3.24 shows how the degree of isomerization differs with
the glycol selected. Propylene glycol is most effective. The cis–trans isomerization is
acid-catalysed so the acidity of the component diacid is an important consideration. To
promote isomerization in diethylene glycol resins it is essential to include phthalic
anhydride in the recipe because ortho-phthalic is the strongest acid (other than maleic).

[Link] Published online by Cambridge University Press

 3.5 Unsaturated Polyesters Resins for Lamination 107

O O HO

OH O

OH O

OH
O O

Maleic anhydride Maleic acid Fumaric acid

Figure 3.23 Maleic anhydride and subsequent acid-catalysed cis–trans isomerization of maleic
acid to fumaric acid.

100 Propylene glycol

75
Trans - structure (%)

Ethylene glycol

50

Diethylene glycol
25

0 500 1,000 1,500 2,000

Average molecular weight (g mol-1)

Figure 3.24 Degree of cis–trans isomerization in glycol maleate polyesters [25,26].

Table 3.2 gives the values of pKa for the phthalic acids with comparatively strong and
weak acids. A strong acid has a low pKa, so we see that the alternative phthalic acids
are less able to promote isomerization.

3.5.2 Formulation of Unsaturated Polyester Ester Resins

The unsaturated polyester with the correct acid value is blended with a reactive
diluent, which is mainly styrene (at 30% w/w) in the presence of an inhibitor or free
radical scavenger, usually hydroquinone. The latter stabilizes the resin, providing a
good shelf life and control over the induction period or ‘working time’ before
gelation, after addition of catalysts and accelerators. In some applications a more
complex inhibitor system is employed. Therefore, a typical recipe is given in
Table 3.3 [28].

[Link] Published online by Cambridge University Press

 108 Matrices

Table 3.2. The acidity of maleic and phthalic acids

Acid pKa Strength

HCl –7 High
Maleic 1.92
Ortho-phthalic 2.98
Isophthalic 3.46
Terephthalic 3.51
Formic 3.77
Acetic 4.76 Low

Table 3.3. Typical composition of an unsaturated polyester resin

Resin
composition
Polyester Resin (parts by
Component Function structure composition weight)

Polyester 2.1 mol 170 parts

Propylene glycol Diol 1.1 mol
Maleic anhydride Unsaturated acid 0.67 mol
Phthalic anhydride Saturated acid 0.33 mol
Styrene Reactive diluent 0.92 mol 100 parts
Benzyl trimethylammonium chloride Inhibitor 1.4 mmol 0.26 parts
Hydroquinone Inhibitor 0.3 mmol 0.034 parts
Quinone Inhibitor 0.03 mmol 0.0034 parts
Paraffin wax Oxygen barrier for cure 100 ppm 0.027 parts

The paraffin wax is added as an air inhibitor, which migrates to the free surface to
act as an oxygen barrier and prevent reaction with the propagating species involved in
the cure of the resin. It prevents the formation of tacky surfaces.

3.5.3 Polyester Formulation

The formulation of the polyester has a number of variants: the nature of the diol or
glycol; the choice of unsaturated or saturated acids; and the degree of isomerization of
maleate to fumarate esters, which is a function of the choice of glycol (Figure 3.24).
Thus, the crosslink density of the cured resin is strongly dependent on the choice of
components, synthesis conditions, and the reactivity and concentration of the
unsaturated groups.
Maleic anhydride is preferred for general-purpose resins, while fumaric acid-
based resins have superior reactivity during cure and hence the achievable thermo-
mechanical properties. Resins with high fumarate content achieve high crosslink
densities after curing and therefore good temperature and environmental resistance,
but tend to be brittle. Fracture toughness and impact strength can be improved by

[Link] Published online by Cambridge University Press

 3.5 Unsaturated Polyesters Resins for Lamination 109

(a) (b)
Cl O Cl Cl

Cl Cl Cl
Cl OH
O
Cl OH
Cl Cl Cl O
Cl O
O

Figure 3.25 (a) Chlorendic or HET acid and (b) chlorendic anhydride used to introduce fire
retardancy into unsaturated polyester resins.

optimizing the crosslink density using ‘flexible’ chain glycols and the ratio of
saturated to unsaturated acids in the synthesis. High-performance systems often
include polymeric toughening agents. The beauty of some fabrication techniques
is that their compatibility can be used to tailor the fracture toughness of the exposed
surface in, for example, a pipe.
Isophthalates generally have superior properties to the orthophalates, especially in
environments exposed to water, chemicals, and weathering. Residual phthalic anhyd-
ride can result from in-service hydrolysis of mono-phthalate chain ends.
Higher temperature resistance, durability, and chemical resistance akin to epoxy
resins can be obtained by introducing ethoxylated bisphenol A into the formulation.
Fire retardancy is another important property, which is achieved by incorporating
halogenated compounds such as HET acid (chlorendic acid) (Figure 3.25) and/or
dibromoneopentyl glycol. HET acid is commonly used to introduce high chlorine
content for self-extinguishing properties. However, the thermochemical properties of
these resins are inferior to isophthalic resins because the –C¼C– is unreactive and
fumaric acid is often used to provide the crosslinking mechanism, so careful molecular
design is employed for fire-retardant resins.
Fire retardancy can also be achieved or enhanced by adding inorganic fillers such as
antimony oxide and/or alumina trihydrate, but at the expense of resin transparency.

3.5.4 Reactive Diluents

Generally, the reactive diluent is styrene, but for translucent sheeting a fraction can be
replaced with methyl methacrylate. It cannot be used alone because it does not readily
copolymerize with the unsaturated polyester, but provides higher light transmission.
Alternative diluents such as vinyl toluene, α-methyl styrene, and diallyl phthalate can
be employed in specialist applications. The limitation to the use of styrene is its
volatility, and care is needed for safe use. The technological solution is to ensure a low
partial pressure of styrene above a moulding during manufacture. This can be
achieved in an analogous methodology to air inhibition during curing, whereby ppm
quantities of wax or equivalent are added, which migrates to the surface of the
moulding.

[Link] Published online by Cambridge University Press

 110 Matrices

3.6 Curing of Polyester Resins

These liquid resins can be converted into mechanically stable solid materials without
the release of volatiles, and are often referred to as contact resins because the
composite and the material are formed simultaneously. Curing is achieved by free
radical addition copolymerization of the unsaturated polyester with the reactive
diluent, which is commonly styrene. As shown in Figure 3.26, the copolymerization
of the maleate groups (and/or fumarate groups) in the polyester with styrene is
initiated by a free radical R•, which is generated either by thermal dissociation of a
peroxide ‘catalyst’ (it is not a true catalyst but an initiator) in hot curing systems, or by
reaction with an accelerator in cold curing systems.
Typical peroxide catalysts are shown in Table 3.4. Methyl ethyl ketone peroxide
(MEKP) is commonly used for cold curing, while benzoyl peroxide (BP) is employed
in hot curing systems. MEKP is available in differing reactivities achieved by altering
its concentration in the phthalate ester solution. Hydroperoxide is an impurity arising
from the presence of adventitious water, which also affects the reactivity. The Scott
Bader handbook [29] differentiates reactivity, which controls the gelation and cure
times, from activity, which is responsible for under-cure of an artefact. Storage of
MEKP in dry conditions reduces the loss of its activity. MEKP is commonly used in
cold cure formulations with a cobalt accelerator. Free radical formation can also be
accelerated for cold curing with dimethyl aniline (DMA). BP is dispersed in a
phthalate ester for use as a paste. It is thermally activated above 70
C. Table 3.4
summarizes the principles for selection of the catalyst.

Figure 3.26 Copolymerization of maleate groups in unsaturated polyesters with styrene during
curing. The crosslinks involve on average two styrene molecules. R• is a free radical generated
from the peroxide initiator.

[Link] Published online by Cambridge University Press

[Link] Published online by Cambridge University Press

Table 3.4. Peroxides used as ‘catalysts’ in polyester resin cure

Critical Dosing
temperature Concentration concentration
Peroxide ‘catalyst’ Form Reactivity (
C) Accelerator (%) (%)

Benzoyl peroxide (BP) Paste in a phthalate* Thermal cure 70 Diethyl 50 2–4

plasticizer aniline
Cyclohexanone peroxide (CHP) Paste in a phthalate** Slow gel + 90 Cobalt or 50 0.5–4
plasticizer rapid hardening vanadium
rates
Methyl ethyl ketone peroxide (MEKP) Solution in a Medium + rapid gel 80 Cobalt or 60 1–2
(various reactivity grades may contain phthalate plasticizer + medium manganese
hydroperoxide impurity) hardening rates
Acetyl acetone peroxide (AAP) Rapid reactivity for 65 50 (2–4 active O) 1–3
RTM and cold press
Tertiary butyl peroctoate (TBPO) Hot-press 70 Cobalt >50 1–3
moulding/pultrusion
Tertiary butyl perbenzoate (TBPB) Hot-press 90 Cobalt 50 1–3
moulding/pultrusion
* Dimethyl phthalate; ** dimethyl phthalate or dibutyl phthalate.
 112 Matrices

Table 3.5. Accelerators for use in the cure of polyester resins

Cobalt Dosing
concentration concentration
Accelerator Solvent (%) (%) Application

Cobalt naphthenate Styrene 0.5–6.0 0.5–4.0 Glass fibre contact

(100 ppm) moulding
Cobalt octoate Dimethyl phthalate 0.5–1.0 0.5–2.0
Vanadium Styrene (or 2.8–3.2 (by 0.5–2.0
naphthenate/ octoate naphthenic acid) volume)
Dimethyl aniline DMP/styrene 0.5–2.0 Rapid curing with BP for
‘filler’ compounds

Table 3.5 gives some typical accelerators. Cobalt naphthenate and dimethyl aniline
as solutions in styrene are commonly used. Mixing the catalyst and accelerator can be
explosive, so for safety reasons pre-accelerated resins, which have the accelerator
added, are available. This is often indicated by the term ‘PA’. Gel times should be
adjusted by varying the accelerator concentration and not the peroxide catalyst, so it is
important to heed manufacturers’ recommendations.

3.6.1 Curing Chemistry

It is essential to understand cure mechanisms because the material and artefact form at
the same time. Therefore, the manufacturing process can be optimized to achieve
consistent properties. Curing involves the initiation of a copolymerization of styrene
with the UPE. The rate of reaction is a function of the molecular location of the –
C¼C–, which strongly influences microstructures of cured UPE resin and related vinyl
ester resins, and hence their mechanical properties.
Curing involves three stages:
Stage 1 Production of Free Radicals
1. Generation of free radicals from the peroxide (R–O–O–R):
a. Hot cure:

ROOR ¼ ROo þ o O R or ROo þ o OR:

b. Cold cure:

ROOR þ Co2þ ¼ ROo þ ROo  þ Co3þ :

2. Reaction with inhibitor (hydroquinone, HQ):

a. HQ controls the shelf life and working time of the catalysed resin:

ROo þ HQ ¼ ROH þ Qo ,

where Qo is a stable radical unable to initiate curing.

[Link] Published online by Cambridge University Press

 3.6 Curing of Polyester Resins 113

3. Polymer chain growth:

a. Initiation:
ROo þ S ¼ So1 ,

where S is styrene.
b. Propagation:
So1 þ S ¼ So2 ,
So1 þ P ¼ SPo etc:,

where P is polyester.
4. Air inhibition
So1!n þ O2 ¼ SOOo ,

where So1!n represents a styrene free radical of varying molecular size from degree
of polymerization of 1 to n:
SOOo þ S ¼ SOOSo
SOOSo þ nS þ mP ¼ Sn Pm Sn OOSn Pm Sn OOSon :

The rate of reaction of a styrene radical, S•, with O2 is ~107 greater than the reaction
with styrene for continued cure. The product is also thermally unstable, releasing
free radicals in an uncontrollable mechanism that can initiate further reaction with
styrene and O2.

Stage 2 Copolymerization of UPE Through Reaction of Maleate (M) and

Fumarate (F) Groups with Styrene (S)
Figure 3.26 illustrates the copolymerization of the maleate groups with styrene, but in
reality we need to consider the relative rates of reaction of the maleate and fumarate
groups with styrene, and styrene with styrene [30,31].
The following reactions occur with the given individual rate constants (k):
Ro þ S ¼ So ki ,
So þ S ¼ So2 þ nS ¼ Snþ2 kSS ,
S o þ M ¼ Mo kSM ,
Mo þ M ¼ Mo2 þ nM ¼ Monþ2 kMM ,
So þ F ¼ Fo kSF ,
So þ F ¼ Fo2 þ nF ¼ Fonþ2 kFF :

The relative rates of styrene-to-styrene radicals, styrene-to-maleate radicals,

maleate-to-styrene radicals, and maleate-to-maleate radicals are given by the ratios
of rate constants:
k SS
¼ r SM ¼ 6:5,
k SM
k MM
¼ r MS ¼ 0:005,
k MS

[Link] Published online by Cambridge University Press

 114 Matrices

where r SM and r MS are monomer reactivity ratios for a styrene free radical with
maleate groups and a maleate free radical with styrene.
From the values of r SM and r MS , we see that styrene (S) adds to a styrene radical
(S ) significantly more rapidly than to a maleate (M•) radical.
•

The analogous analysis for fumarate groups is:

k SS
¼ r SF ¼ 0:3,
k SF
k FF
¼ r FS ¼ 0:07:
k FS
Thus, fumarate radicals will tend to add styrene, and styrene radicals will tend to
add to fumarate unsaturated groups. Thus, an alternating copolymer will tend
to form.
Methyl methacrylate (m) is an alternative reactive diluent for UPE; the monomer
reactivity ratios given below provide understanding [32,33]:
kmm
¼ r mM ¼ 354,
k mM
k MM
¼ r Mm ¼ 0:
kMm
This data shows that methyl methacrylate will tend to homopolymerize, with only
the occasional maleate links.
Analogously, the reaction of methyl methacrylate with fumarate (F) groups has
the following monomer reactivity ratios:
k mm
¼ r mF ¼ 40,
kmF
kFF
¼ r Fm ¼ 0:03:
k Fm
Thus, there is still a high probability for methyl methacrylate to homopolymerize,
but the chain length between fumarate groups will tend to be shorter. Therefore,
methyl methacrylate is not an appropriate reactive diluent, but could be included as
an additional comonomer.

Stage 3 Diffusion of Monomer

The microstructure of the cured resin is determined in Stage 2. Although copolymerization
achieves an approximate alternating molecular structure from the two components,
the concentration of styrene usually exceeds the molar concentration of maleate and
fumarate groups. Therefore, the excess styrene present after initial gelation will
diffuse to the free radical sites for further reaction during Stage 3. Figure 3.27 shows
the consumption of the polyester (P) and styrene (S) components throughout the
curing of maleate and fumarate unsaturated polyesters. Post-curing is required for
full cure. The highest heat distortion temperature (HDT) is achieved with high post-
cure temperatures.

[Link] Published online by Cambridge University Press

 3.6 Curing of Polyester Resins 115

FUMARATE MALEATE

S
0.3
S
[C=C]

0.2 P P

39 39
25
0.1
18 15
36 30 12

10 102 103 80 110 140 170 10 102 103 50 80 110 140

t/min T/°C t/min T/°C

Figure 3.27 The schematic of the decrease in the concentration of unsaturated groups from
styrene (S) and maleate and fumarate (P) in the cure of a UPE, illustrating the differing
stages: (I) free radical formation; (II) copolymerization; (III) diffusion of monomers during
post-cure. The numbers represent the percentage of residual styrene (S) and UPE unsaturated
groups [24].

Post-Curing
During chain addition polymerization, polymers form instantaneously so that free
monomers are always present unless the reaction is pushed to completion. Therefore,
post-curing is usually necessary to achieve the optimum thermomechanical properties.
Figure 3.27 shows that, after curing, free styrene is still present in the moulding and
that post-curing reduces the free styrene concentration. It is also shown that in the
fumarate UPE the residual concentration of unsaturated groups is also lower.
For effective post-curing, the glass transition temperature of the cold-cured resin
should be exceeded so that the rate of diffusion of monomers is increased. Post-curing
can also involve the decomposition of any residual peroxide into free radicals for
further reaction.

Exotherms
The curing of UPE is exothermic, which can be used to manage the degree of post-
cure in a ‘cold-cured’ component. However, uneven exothermic heat dissipation
throughout a structure may lead to differential curing and variable thermal residual
stresses. Typical exothermic temperatures can reach 200
C and above, as shown in
Table 3.6. The most reactive resins have high fumarate concentrations and exhibit the
highest exotherms. This is associated with the presence of the trans –C¼C– bonds
resulting from the high degree of cis–trans isomerization of the maleate groups in
propylene glycol resins (Figure 3.23). Management of the exotherm can be achieved
by choosing the correct catalyst/accelerator ratio. A low ratio provides curing without
a heat effect. Thus, structures should be designed with heat dissipation in mind so that
rapid production is possible.

[Link] Published online by Cambridge University Press

 116 Matrices

Table 3.6. The effect of maleate/fumarate isomerization on the gel time and exotherms observed for propylene glycol
(PG) and diethylene glycol (DEG) isophthalates (40% styrene) measured using SPI gel test at 82
C [25]

Flexural Flexural
Gel time Peak exotherm Peak exotherm modulus strength
Acid Glycol (min:s) (
C) time (min:s) HDT (
C) (GPa) (MPa)

Fumaric PG 5:20 210 7:47 100 3.7 126

Maleic PG 5:20 210 7:44 101 3.7 124
Fumaric DEG 4:30 202 6:51 52 2.7 106
Maleic DEG 4:40 183 8:17 43 2.1 84

Figure 3.28 Instantaneous copolymer composition of diethyl maleate (M) and diethyl fumarate
(F) with styrene (S). xps is the mole fraction of S in the copolymer. xos is the mole fraction of S in
the monomer mixture [30,31,34].

3.7 Resin Microstructure

The structure of a cured polyester resin is determined by the relative molar concen-
trations of fumarate groups and the reactive diluent (styrene) and the reactivity ratios
(i.e. the rate constants for the copolymerization). The instantaneous copolymer
composition can be calculated using the analysis given in Section 3.6.1.
Figure 3.28 gives the theoretical instant compositions of diethyl fumarate/styrene
and diethyl maleate/styrene copolymers as a function of the composition of the
monomer mixture [34].
Assuming that the instantaneous composition of a cured UPE would be similar, it
shows that the maleate groups are incorporated less efficiently than the fumarate
groups into the molecular structure. Furthermore, a typical UPE employs a mole
fraction of styrene of ~0.6 (Table 3.3), which would promote the formation of an
alternating copolymer.

[Link] Published online by Cambridge University Press

 3.8 Mechanical Properties of Cast Resins 117

Figure 3.29 A schematic of the conformation of a polyester random coil dissolved in styrene (S)
showing the propagation of the polystyryl free radical and the origin of the non-
homogeneous microstructure.

The molecular structure of a cured general UPE is considered to have polystyrene

crosslinks with an average length of 2–3 monomer units (n in Figure 3.20), which
reflects the degree of maleate/fumarate isomerization achieved in the polyester syn-
thesis. Further complications arise through the conformation of the polyester molecule
in the styrene solution. Styrene has a solubility parameter of 19 MPa½, whereas a
typical polyester has a value of ~21–22 MPa½, which means that in styrene solution
the polyester will exist as a highly conformed coil.
Figure 3.29 shows a schematic of a randomly coiled polyester molecule in styrene
solution at the onset of curing. The styryl free radical propagates through adjacent
unsaturated groups on the polyester random coil. This results in a ‘nodular’ micro-
structure to the cured resin. Funke [35] and Bergman and Demmler [36] have
demonstrated by nuclear magnetic resonance (NMR) and infrared (IR) spectroscopy
that a cured polyester resin has a defined nodular structure. The development of the
microstructure in a UPE is schematically described in Figure 3.30.

3.8 Mechanical Properties of Cast Resins

The mechanical properties of cast UPE resins are a function of the average crosslink
density and the ‘flexibility’ of the molecular chains between the crosslinks. These
variables are not fully independent, but provide the options for formulating resins for
different applications.
As shown in Table 3.7, increasing the length of the ‘diol’ chain between the ester
groups leads to a decrease in flexural modulus. The rotational flexibility of the ‘diol’
chain is reduced in the presence of a side group (see EPG vs DEG). Trimethylene
glycol (TMG) appears inconsistent with this trend, but the degree of isomerization is

[Link] Published online by Cambridge University Press

 118 Matrices

Table 3.7. Flexural properties of cured styrenated UPE with differing diol [25]

Property PG DEG EPG TMG EDEG

Chain length (bonds between esters) 3 6 6 4 9

Flexural modulus (GPa) 4.2 3.4 4.1 2.5 2.1
Flexural strength (MPa) 129 119 100 92 83
DEG, diethylene glycol; EDEG, ethylene diethylene glycol; EPG, ethylene propylene glycol; PG, propylene
glycol; TMG, trimethylene glycol.

Figure 3.30 Development of the microstructure during the cure of a styrenated PG maleate
UPE, established by NMR spectroscopy according to Bergmann and Demmler
[36]. 1000 Å ¼ 100 nm ¼ 0.1 μm.

low (and hence so is crosslink density). Furthermore, etherification of TMG can occur
during the synthesis so that the length of the ‘diol’ chain between crosslinks is
probably larger.
Aliphatic saturated acids are also used to provide flexibility of the chain between
crosslinks. Terephthalic acid can enhance the chain rigidity between crosslinks, but
higher synthesis costs means that isophthalic acid is preferred for high-performance
resins. Phthalic anhydride is commonly used for general-purpose resins because of
lower synthesis costs. Table 3.8 gives the typical properties of cured UPE castings.
The theoretical tensile modulus of an organic glass is calculated to be 4 GPa, and
most thermosetting resins have similar moduli, depending on the degree of cross-
linking and the ‘flexibility’ of the molecular chain between them. Strength is not a
material property because it is dependent on the concentration and dimensions of
flaws in the casting. The tensile strength of castings is difficult to measure since flaw-
free coupons are difficult to prepare, especially as a result of the microstructure and a
high cure shrinkage of ~8% by volume.

[Link] Published online by Cambridge University Press

 3.11 Related Resins 119

Table 3.8. Typical properties of cured styrenated UPE and vinyl ester

Property UPE Vinyl ester

Density (kg m–3) 1,280 1,120

Hardness (Rockwell M) 110 –
Hardness (Barcol) 40 35
Tensile modulus (GPa) 3.5 3.3
Tensile strength (MPa) 70 70–80
Elongation at break (%) 2.5–4.5 5–6
Fracture toughness, K1c (MN m–3/2) 0.49–0.61 0.75
Compressive strength (MPa) 140 –
Refractive index (nD) 1.57 –
Volume contraction during cure (%) 8 8.3
HDT (
C) 75 100

3.9 Low-Profile Resins

While the addition of reinforcements reduces the volume of resin and hence the curing
shrinkage, it is still important to compensate further for the shrinkage to achieve a
good surface finish. Low-profile resins contain shrinkage control additives or low-
profile additives, which are soluble thermoplastics such as polyvinyl acetate and
polystyrene.

3.10 Gel Coat Resins

Gel coat resins are made thixotropic by the addition of 1% by weight of hydrophilic
fumed silica. In this way a uniform coating can be applied to the tooling at a thickness
of 0.5 mm and gelled prior to the subsequent structural laminating resin and glass
fibre. This ensures a good finish to the moulding.

3.11 Related Resins

3.11.1 Vinyl Ester Resins

These resins are designed to achieve the higher performance of epoxy resins while
maintaining the flexibility in processing associated with unsaturated polyester resin.
A typical vinyl ester is a vinyl-terminated epoxy, which is cured by copolymerization
with the styrene diluent using peroxides and accelerators. They are synthesized by
reacting an epoxy resin with an unsaturated acid such as methacrylic acid, as shown in
Figure 3.31.
Vinyl ester resins have a higher extensibility than standard UPEs, but with a
slightly lower tensile modulus. Other vinyl resins based on acrylic acid end-capped

[Link] Published online by Cambridge University Press

 120 Matrices

CH3 O OH CH3 OH CH3 OH O CH3

CH2 C C O CH2 CH CH2 O C O CH2 CH CH2 O C O CH2 CH CH2 O C C CH2
★ ★ ★ ★

CH3 n CH3

Figure 3.31 Methacrylate end-capped bisphenol A epoxy, an example of a vinyl ester resin.
** indicates the vinyl group.

epoxies are available. Specialist resins based on novolak and tetrabrominated bisphe-
nol A epoxies end-capped with methacrylic acid are also available.

3.11.2 Microstructure of Vinyl Ester Resins

These resins are cured similarly to unsaturated polyester resins; therefore, the analysis
given in Section 3.6.1 is relevant to understand the structure of the cast resin [24].
For the methacrylate vinyl ester, the reactivity ratios for the vinyl (V) groups with
styrene (S) are:
kSS
¼ r SV ¼ 0:4,
kSV
kVV
¼ r VS ¼ 0:37:
kSS
For the acrylic vinyl ester, the reactivity ratios for vinyl (V) groups with styrene (S) are:
k SS
¼ r SV ¼ 0:73,
kSV
kVV
¼ r VS ¼ 0:72:
k SS
Since the reactivity ratios for the vinyl group/styrene and styrene/styrene reactions are
essentially identical for both systems, the vinyl groups can react either with them-
selves or with styrene with equal probability. Thus, the chemical structure of the cured
resin, which is a copolymer, will be a random arrangement of the monomeric
components and a function of the monomer composition. To be random it will consist
of blocks of styrene and vinyl ester groups ranging from small (n 1) to large. The
microstructure should be homogeneous, which is in contrast to the cured polyesters,
which have a heterogeneous microstructure, as shown in Figure 3.30. This observation
helps explain the higher fracture toughness (K1c) of the vinyl ester resins.

3.11.3 Vinyl Urethanes

Since ester groups can hydrolyse in aqueous environments, especially in the presence
of acidic or alkaline conditions, the number of ester groups should be minimized for
high durability in aqueous environments. Therefore, hydrolysis of isolated ester
groups will not disrupt the network structure significantly. The urethane group is

[Link] Published online by Cambridge University Press

 3.12 High-Temperature Resin Systems 121

CH3 O CH3 CH3 CH3 O

CH2 C NH C O CH CH2 O C O CH2 CH O C CH3 CH3 CH3 O CH3
CH CH
CH3 C O CH CH2 O C O CH2 CH O C NH C CH2
O CH3 q

Figure 3.32 Methacrylate end-capped bisphenol A epoxy fumarate, an example of a vinyl

urethane [37–39].

more hydrolytically resistant, so vinyl urethanes offer long-term durability. The

methacrylate vinyl urethane given in Figure 3.32 is a vinyl end-capped bisphenol
epoxy fumarate (originally Atlac 580) that achieves high crosslink density after curing
and provides excellent long-term chemical durability in comparison to the isophtha-
late, bisphenol A polyesters, and vinyl esters. The high crosslink density is reflected in
the low fracture toughness (K1c ¼ 0.45 MN m–3/2). This resin is recommended for
long-term durability in unstressed applications, but not when environmental stress
cracking is the most likely failure mechanism, where toughened isophthalate fumarate
UPEs are recommended.

3.12 High-Temperature Resin Systems

Many polymeric systems are employed in specialist applications. Here we briefly

consider resins for composites used in high-temperature applications.

3.12.1 Thermosetting Polyimides

Polyimides represent the largest class of high-temperature polymers, which are often
used for composites. Polyimides are available as thermoplastics or thermosets [40].
We will consider only the main thermosets, PMR and bisimides.

[Link] PMR Systems

PMR refers to polymerization of monomeric reactants, whereby the materials are
polymerized in situ during the processing of a component. The prepregs are made by
impregnating the fibre with a solution of the monomers in methanol. Figure 3.33
illustrates the most common type, PMR-15, which was developed at NASA [41]. The
number ‘15’ refers to the molecular weight of the imidized prepolymer, 1,500 g mol–1,
which forms from the monomer composition.
The three components shown are:
1. MDA: 4,40 -diamino-diphenylmethane
2. BTDE: benzophenone tetracarboxylic acid dimethyl ester
3. NE: 5-norbornene-2,3-dicarboxylic acid monomethyl ester.
Residual methanol is used to provide the prepreg with tack and drape, and careful
cure schedules in an autoclave are employed. The final stage of curing involves

[Link] Published online by Cambridge University Press

 122 Matrices

CO2Me MeO2C O CO2Me

H2N NH2
2 +3 CH2 +2 C
CO2H HO2C CO2H
–H2O, –CH3OH

O O O
O
C O C C
C
N N C N N
C H2C C C CH2 C
O O O 2 O

Heat, pressure

O O O O
C O C C
C
N N C N N
C H2C C C CH2 C
O O 2 O
O

Figure 3.33 The PMR thermosetting polyimide system.

Figure 3.34 CDA: 4,40 -methylenebis-(5-isopropyl-2-methylaniline), a potential replacement for

MDA in a low- or non-toxic PMR system.

polymerization through the norbornene end-caps at 316
C at 200 psi. While this resin
is thermally stable and can withstand temperatures up to 350
C, composites are prone
to microcracking resulting from the generation of thermal stresses on cooling. The
process appears to be time-dependent because of the loss of volatiles [42].
The major inhibitor to application of PMR resins is the toxicity of the MDA
component. Research on replacing MDA is ongoing. For example, Harvey et al.
[43] have recently reported the use of a diamine, CDA, with low toxicity for a PMR
resin and a Tg approaching that for PMR-15 (Figure 3.34).

[Link] Bisimides
The most common members of this group are the bismaleimides (BMI). The funda-
mental structure of a BMI is shown in Figure 3.35, together with a typical example.
The cured BMI in Figure 3.35b has a glass transition temperature of 342
C.
A variety of BMI resins are available, where R in Figure 3.35a can differ by using a
range of diamines [43]. Commercial BMIs are available for continuous use at

[Link] Published online by Cambridge University Press

 3.12 High-Temperature Resin Systems 123

Table 3.9. Glass transition temperatures of high-temperature matrix polymers (polyimide, cyanate ester, and
related resins)

Tg Flexural Flexural
Resin Type (
C) modulus (GPa) strength (MPa)

PMR-15 PMR 340 – 176

Thermid MC-600 BMI 320 4.5 145
Avimid NR 150B2 BMI 340 – –
Compimide 796 BMI 300 4.6 76
Compimide 796 – toughened BMI/propenyl copolymer 261 3.7 132
Matrimid 5292 Thermoplastic polyimide 273 4.0 –
PT resin Phenolic triazine resin 360 4.7 97
AroCy l-10 Cyanate ester 259 2.8 162
AroCy l-10/epoxy blend Cyanate ester/novolak epoxy 214 – –
RTX 366 Cyanate ester 175 2.8 121
PBOX Bisoxazoline-phenolic >200 – –
TGDDM-DDS Epoxy 250 3.6 90

(a) (b)
O O
O O

N R N N N

O O
O O

Figure 3.35 (a) Structure of a bismaleimide, where R is either an alkyl or aryl chain; (b) a typical
component 4,40 - bismaleimidodiphenylmethane [40,44], Tg ¼ 342
C.

temperatures in the range 200–230
C. To achieve processability for composites a

eutectic mixture is often employed. While BMI can be cured by thermal radical
polymerization, copolymerization with a range of other monomers provides many
formulation options, including toughening with thermoplastic imides. These details
are discussed elsewhere [40].

[Link] Cyanate Resin

Cyanate ester resins have good high-temperature performance, which is only sur-
passed by that of BMI. Further details can be found elsewhere [45]. The approximate
service temperature is related to the Tg of the resin. Table 3.9 provides the Tg of some
exemplars of the different resin groups, together with some mechanical properties.
The advantages are their rheological properties and the low moisture absorption, but
the main disadvantage of cyanate ester resins is the sensitivity of the cyanate groups to
hydrolysis. This aspect can be compensated for by copolymerization with epoxy
resins [46].

[Link] Published online by Cambridge University Press

 124 Matrices

[Link] Other Potential Matrix Resins

Other high-temperature performance resins such as the bisoaxazoline phenolics,
benzocyclobutene (BCB), and phthalonitrile resins are also available. Details are
given by Hay [47,48].

3.13 Ceramic Matrices

Ceramics are inorganic materials, and are employed in high-temperature applications –

that is, at temperatures >1000–2500
C, which is in excess of that available to metal
and polymer matrices. We observed earlier that these materials have moduli greater
than typical reinforcements (e.g. alumina at 340 GPa) and exhibit brittle behaviour
with failure strains <1%. As a result, the role of the reinforcement is to toughen the
ceramic. Toughness can be introduced by adding fibrous ‘reinforcements’, which
induce matrix cracking. When the matrix is well bonded to the fibres (i.e. a strong
interface) the matrix cracks propagate instantaneously through the fibres without
introducing additional energy absorption. However, when the cracks are pinned by
the fibres (i.e. with a controlled interface) further cracks form as the load increases.
Each time a matrix crack occurs or grows, free surfaces form with energy absorption.
Thus, the composite exhibits tough behaviour and the stress–strain curves show an
extended displacement as a result of multiple matrix cracking. Figure 3.36 illustrates
the extended displacement curve for a fibre composite in comparison to particulate
reinforcement [49].
Processing of ceramic matrix composites (CMCs) involves the following stages:
(1) impregnating the fibres; (2) consolidation; and (3) formation of the ceramic at high

Figure 3.36 Schematic force displacement curves for monolithic ceramic, particulate-reinforced
ceramic, and fibre-reinforced ceramic composite. The areas under the curves equate to the
energy absorbed through fracture. The fibre composite absorbs energy via multiple matrix
cracking [49].

[Link] Published online by Cambridge University Press

 3.13 Ceramic Matrices 125

temperature. Generally, melt infusion is impractical because of the high temperatures

involved. Therefore, indirect routes are employed in which the fibres are infiltrated
with matrix precursors. We will only briefly discuss the available routes.

3.13.1 Slurry Processing

In slurry processing the inorganic powder precursor(s) is dispersed in a carrier liquid,
which is often water. For particulate or whisker or short fibre reinforcements a fine
dispersion can be prepared ultrasonically then dried. Consolidation is through hot
pressing or cold pressing followed sintering. For example, plaster of Paris from
CaSO4·2H2O or ordinary Portland cement involve reaction with the water and subse-
quent crystal formation for the ceramic matrix to form. A porosity-free matrix is
difficult to achieve in this way.
For continuous fibres, the tows are pulled through a slurry of the precursor in an
aqueous solution of a polymeric binder to create a ‘prepreg’, which can be dried and
subsequently consolidated by thermal degradation of the binder and hot pressing.
Typical matrices formed in this way are glass and glass ceramics. With these matrices,
the glass can flow and porosity can be minimized, but that limits the available
ceramics.

3.13.2 Liquid Processing

A melt is used to infiltrate the fibres. This could be a polymer precursor, which can be
converted into a ceramic during thermal degradation prior to consolidation and firing,
such as phenolic resin for carbon, polycarbosilane for silicon carbide, and polysila-
zane for silicon nitride. An important example is the use of petroleum pitch as a
carbon precursor, where the carbon yield can be as high as 80%. In order to achieve
high degrees of consolidation and minimum porosity, the pitch normally needs to be
converted into a mesophase (or liquid crystal) pitch to maximize the carbon yield
during thermal processing. Normally, multiple infiltrations are required to achieve a
high-density composite, but the liquid infiltration will not penetrate all the pores left
by the loss of volatiles during heat treatment. Full densification is generally not
achievable using this approach, and further gaseous infiltration such as chemical
vapour infiltration (CVI) or chemical vapour deposition (CVD) can be used to
penetrate the remaining pores.

3.13.3 Sol-Gel Processing

In this technique, a monomer reacts with the dispersant, which is mainly water, to
provide effectively a solution or a colloid, referred to as a gel, in which the
‘particles’ are of molecular or nanometre dimensions [50]. As with the slurry
technique, the fibres can be infiltrated to prepare a ply, which can be stacked in
the manner of prepreg. The advantage is that multiple infiltrations can be used to
ensure that the porosity is minimized. The classic example is the use of tetraethoxy

[Link] Published online by Cambridge University Press

 126 Matrices

Figure 3.37 Formation of a silica network using sol-gel chemistry from TEOS.

silane (TEOS), which in the presence of an aqueous acid or alkali hydrolyses into a
silica sol. The polymerization by condensation of the silanol and ethoxy silane
groups causes particles of differing size and structure to form and provides the sol
with an increase in viscosity until a gel forms. In this way the infiltrated fibres can
have a handleable form for composite manufacture by hot pressing. A number of
configurations of molecular structure of the silica are possible, and these can be
controlled by TEOS/H2O ratio as well as the pH. As a result, the silica product will
form a matrix that remains either amorphous or crystallizes during hot pressing and
firing (Figure 3.37).
A number of oxide-based matrices can be prepared by this method. For example,
by mixing silica and alumina sols, mullite can be prepared [51]. Other examples are
titanium oxide (titania) or zirconium oxide (zirconia) from titanium or zirconium
ethoxides.

3.13.4 Chemical Vapour Infiltration

Chemical vapour deposition is the technique in which chemical reactions in the
gaseous state are used to deposit a solid onto a heated surface. Boron and silicon
carbide fibres are prepared by deposition onto carrier fibres such as a tungsten wire.
For application to composites, the CVD technique is used for CVI of an array of fibres
(often in the form of a textile) by gaseous compounds, which are deposited and form
the matrix. For example, carbon/carbon fibre composites are prepared in this way from
the vapour-phase deposition of carbon using the thermal decomposition of a hydro-
carbon. Methane is preferred because it has the highest carbon content for a volatile
precursor:
CH4ðgÞ ! CðsÞ þ 2H2ðgÞ :

The substrate temperature for CVD and CVI of carbon is generally maintained at
1100
C, which is the temperature at which the carbon atoms just become mobile.
Under isothermal conditions, the infiltration is carried out at low pressures (5–50 torr),
whereas with a thermal gradient atmospheric pressures are employed. A pressure
gradient technique is used when the porosity deep in the structure needs to be filled.

[Link] Published online by Cambridge University Press

 3.14 Metal Matrices 127

3.13.5 In-Situ Processing

In this method the fibres are infiltrated with a molten metal, which is then reacted with
a gas to create the required matrix. In the Lanxide process the fibres are infiltrated by
molten aluminium, which reacts with oxygen at the outside surface of the preform to
form an alumina–ceramic matrix.

3.14 Metal Matrices

A metallic matrix can be formed using the following techniques.

3.14.1 Solid State

This technique refers to diffusion bonding and powder metallurgy routes. In the
former a polymer-bound fibre mat is sandwiched by metal foils to form a stack, which
is hot pressed into a consolidated composite. The latter involves mixing powdered
metal with short fibres, whiskers, and/or particulates, which are pressed and sintered at
elevated temperatures.

3.14.2 Liquid State

Discontinuous fibres or whiskers can be simply mixed into a molten metal for casting
into an artefact. Rheocasting employs a more viscous melt, which is stirred at
temperatures closer to the melting point so that the reinforcement remains
uniformly dispersed.
Continuous fibres can be infiltrated using the technique of squeeze casting of the
molten metal into a preform. An example is the pressure infiltration of aluminium into
random long discontinuous Saffil alumina fibres for locally reinforcing piston crowns
for automotive engines.

3.14.3 Deposition
In this technique, spray co-deposition of molten metal and the reinforcement are
collected for subsequent processing into an artefact.

3.14.4 In situ
In this technique, the reinforcement is a fibre-like crystal introduced into the metal
matrix using directional solidification. A movable induction-heating coil is used to
directionally crystallize a eutectic alloy to introduce parallel needle crystals, which
provide the reinforcement.

[Link] Published online by Cambridge University Press

 128 Matrices

3.15 Conclusions

The choice of matrix material is discussed. While the emphasis is on polymeric

matrices, we describe metal and ceramic systems for completeness. The fabrication
routes described in Chapter 4 demonstrate the range of polymers employed, and we
give the reasons for the choice. The metal and ceramic composite matrices cannot be
readily divorced from the processing routes, so we use these techniques to explain the
choice of material.

3.16 Discussion Points

1. What is the approximate maximum value of tensile modulus of a glassy linear

polymer? What molecular force resists deformation?
2. What aspect of the structure of a linear polymer is responsible for a higher modulus
than that of a glassy polymer?
3. Thermosetting polymers can have slightly higher modulus than a glassy linear
polymer. Why?
4. What are advantages and disadvantages of using a linear thermoplastic matrix or
thermosetting matrix?
5. Why have epoxy resins and unsaturated polyester resins commonly been used as
matrix resins?
6. What mechanisms are available for curing epoxy and unsaturated polyester resins?
7. Consider the solubility parameter as a method of predicting the compatibility of
polymers, and also with diluents.

References

1. J. M. G. Cowie and V. Arrighi Polymers, Chemistry and physics of modern materials, 3rd
ed. (Bocca Raton, FL: CRC Press, 2007).
2. L. R. G. Treloar, The physics of rubber elasticity (London: Oxford University Press, 1958).
3. P. Flory, Principles of polymer chemistry (Ithaca, NY: Cornell University Press, 1953).
4. H. Ishida and T. Agag, eds., Handbook of benzoxazine resins (New York: Elsevier, 2011).
5. B. Ellis, ed., The chemistry and technology of epoxy resins (London: Chapman and Hall,
1993).
6. C. A. May, Epoxy resin: chemistry and technology, 2nd ed. (New York: Marcel Dekker,
Wiley, 1988).
7. K. Hoy, New values of solubility parameter from vapour pressure data. J. Paint Technol. 42
(1970), 76.
8. D. W. van Krevelen and K. Te Nijenhuis, Properties of polymers, 4th ed. (Amsterdam:
Elsevier, 2009).
9. F. R. Jones, Thermally induced self healing of thermosetting resins and matrices in smart
composites. In Self healing materials: an alternative approach to 20 centuries of materials
science, ed. S. van der Zwaag (Dordrecht: Springer, 2007), pp. 69–93.

[Link] Published online by Cambridge University Press

 3.16 Discussion Points 129

10. R. Scott, The thermodynamics of high polymer solutions. V. Phase equilibria in the ternary
system: polymer 1—polymer 2—solvent. J . Chem. Phys. 17 (1949), 279.
11. R. J. Young and P. A. Lovell, Introduction to polymers, 2nd ed. (London: Chapman and
Hall, 1991), pp. 241–309.
12. P. J. Flory, Statistical mechanics of chain molecules (New York: Interscience, 1989).
13. X. Peng and J. K. Gillham, Time–temperature–transformation (TTT) cure diagrams:
Relationship between Tg and the temperature and time of cure for epoxy systems.
J. Appl. Polym. Sci. 30 (1985), 4685–4696.
14. M. T. Aronhime and J.K. Gillham, Time-temperature-transformation (TTT) cure diagram
of thermosetting polymeric systems. Adv. Polym. Sci. 78 (1986), 83–113.
15. J. K. Gillham, Formation and properties of thermosetting and high Tg polymeric materials.
Polym. Eng. Sci. 26 (1986), 1429–1433.
16. B. Ellis, Time-temperature-transformation diagrams for thermosets. In Handbook of
polymer-fibre composites, ed. F. R. Jones (Harlow: Longman, 1994), pp. 115–121.
17. J-P. Pascault, H. Sautereau, J. Verdu, and R. J. J. Williams, Thermosetting polymers (New
York: Marcel Dekker, 2002).
18. R. B. Prime, Thermosets. In Thermal characterisation of polymeric materials, ed. E. A.
Turi (New York: Academic Press, 1981).
19. X. Wang and J. K. Gillham, J. Coatings Technol. 64 (1992), 37–45.
20. F. R. Jones, Molecular modelling of composite matrix properties. In Multi-scale modelling
of composite material systems, ed. C. Soutis and P. W. R. Beaumont (Cambridge:
Woodhead, 2005), pp. 1–32.
21. D. Porter, Group interaction modelling of polymer properties (New York: Marcel Dekker,
1995).
22. J. P. Foreman, S. Behzadi, S. A. Tsampas, et al. Rate dependent multiscale modelling of
fibre reinforced composites. Plast. Rubber Compos. 38 (2009), 67–71.
23. B. Parkyn and E. Bader, British Patent 713332 (1954)
24. F. R. Jones, Unsaturated polyester resins, In Brydson’s Plastic Materials, ed. M. Gilbert,
8th ed. (Oxford: Elsevier, 2017), pp 743–772.
25. H. V. Boenig, Unsaturated polyesters: structure and properties (Amsterdam: Elsevier, 1964).
26. A. A. K. Whitehouse, Resin systems. In Glass reinforced plastics, ed, B. Parkyn (London:
Iliffe-Butterworth, 1970), pp. 109–120.
27. R. A. Panther, Unsaturated polyester resins. In Handbook of polymer-fibre composites, ed.
F. R. Jones (Harlow: Longman, 1994), pp. 121–127.
28. J. A. Brydson, Plastics materials, 5th ed. (London: Butterworth, 1989), pp. 653–696.
29. Scott Bader, Polyester handbook. Available at: [Link]
3381_crystic-[Link].
30. F. W. Billmeyer Jr., Textbook of polymer science (New York: Interscience, 1962).
31. F. R. Mayo and C. Walling, Copolymerisation. Chem. Rev. 46 (1950), 191–287.
32. W. I. Bengough, D. Goldrich, and R. A. Young, The copolymerisation of methyl metha-
crylate with diethyl maleate and diethyl fumarate. Eur. Polym. J. 3 (1967), 117–120.
33. F. R. Mayo, F. M. Lewis, and C. Walling, Copolymerization. VIII. The relation between
structure and reactivity of monomers in copolymerization. J. Am. Chem. Soc. 70 (1948),
1529–1133.
34. F. Rodriguez, Principles of polymer systems (New York: McGraw-Hill, 1970).
35. W. Funke, Über Vernetzte Makromolekulare Stoffe mit Inhomogener Netzwerkdichte.
J. Polym. Sci. C. 16 (1967), 1497.

[Link] Published online by Cambridge University Press

 130 Matrices

36. K. Bergman and K. Demmler, Koll. Ztscht. Ztscht. f. Polym. 252 (1974), 193–206.
37. A. Kandelbauer, G. Tondi, and S. H. Goodman, Unsaturated polyesters and vinyl esters. In
Handbook of thermoset plastics, ed. H. Dodiuk and S. H. Goodman (San Diego, CA:
Elsevier, 2014).
38. R. G. Weatherhead, FRP technology (London: Applied Science, 1980).
39. F. R. Jones, Vinyl ester resins. In Handbook of polymer-fibre composites, ed. F. R. Jones
(Harlow: Longman, 1994), pp. 129–132.
40. D. Wilson, H. D. Stenzenberger, and P. M. Hergenrother, eds., Polyimides (Glasgow:
Blackie, 1990).
41. T. T. Serafini, P. Delvigs, and G. R. Lightsey, Thermally stable polyimides from solutions
of monomeric reactants. J. Applied Polym. Sci. 16 (1972), 905–915.
42. M. Simpson, P. M. Jacobs, and F. R. Jones, Generation of thermal strains in carbon fibre
reinforced bismaleimide (PMR-15) composites, Pt 3, A simultaneous thermogravimetric
mass spectral study of residual volatiles and thermal microcracking. Composites 22 (1991),
105.
43. B. G. Harvey, G. R. Yandek, J. T. Lamb, et al., Synthesis and characterization of a high
temperature thermosetting polyimide oligomer derived from a non-toxic, sustainable bisani-
line. RSC Adv. 7 (2017), 23149–23156.
44. R. J. Iredale, C. Ward, and I. Hamerton, Modern advances in bismaleimide resin technol-
ogy: a 21st century perspective on the chemistry of addition polyimides. Progr. Polym. Sci.
69 (2017), 1–21.
45. I. Hamerton, ed., Chemistry and technology of cyanate ester resins (Glasgow: Blackie,
1994).
46. S. K. Karad, D. Attwood, and F. R. Jones, Mechanisms of moisture absorption by cyanate
ester modified epoxy resin matrices, part III: effect of blend composition. Composites A 33
(2002), 1665–1675.
47. J. N. Hay, High temperature resins-thermosetting polyimides. In Handbook of polymer-
fibre composites, ed. F. R. Jones (Harlow: Longman, 1994), pp. 96–101.
48. J. N. Hay, High temperature resins other thermosets. In Handbook of polymer-fibre
composites, ed. F. R. Jones (Harlow: Longman, 1994), pp. 101–106.
49. F.L. Matthews and R.D. Rawlings, Composite materials: engineering and science
(London: Chapman and Hall, 1994).
50. L. L. Hench and J. K. West, The sol-gel process. Chem. Rev. 90 (1990), 33–72.
51. J. Wu, M. Chen, F. R. Jones, and P. F. James, Mullite and alumina-silica matrices for
composites by modified sol-gel processing. J. Non Cryst. Solid. 162 (1993), 197–200.

[Link] Published online by Cambridge University Press