Handbook of Polymer Synthesis Second Edition Episode 15 pdf

15
Crosslinking and Polymer Networks
Manfred L. Hallensleben
Institut fu
¨
r Makromolekulare Chemie, Universita
¨
t Hannover, Hannover, Germany
I. INTRODUCTION
In this article chemical crosslinking reactions are dealt with respect to network formation
starting from individual monomeric, oligomeric or polymeric molecules. Any type of
chemical reaction of functional groups may be used for the purpose of polymer network
formation that allows for almost quantitative conversion of these functionalities; if not,
considerable difficulties with respect to network stability, long term stability of physical
properties of the network, and chemical transformation of unreacted functionalities may
arise. Since the chemical and physical properties of a respective polymer network strongly
depend on the chemical nature of the monomeric units in the chains and also depend on the
crosslink density, a wide variety of physical properties of crosslinked polymeric materials
is available and considerable technological input is made to design the network in order to
match the demands.
This contribution to chemical crosslinking does not include the use of electron
beam or g-irradiation. These methods have some advantages over the use of chemical
crosslinking agents as they do not leave behind toxic, elutable agents. Also it does not
include peroxide initiated radical crosslinking of saturated polymers which proceeds
randomly by hydrogen abstraction from chain segments and coupling reactions of these
radical sites.
This contribution does also not include ‘physical’ almost reversible crosslinking
due to microphase separation of block copolymers, to strong hydrogen bonding or to ionic
interactions or to crystallite formation.
II. DEFINITION OF POLYMER NETWORKS
Any formation of a polymer network starts from monomeric, oligomeric or polymeric

individual molecules which react in solution, in melt or in the solid state. It is necessary
that at least a small fraction of these molecules has a functionality f  3 to undergo bond
formation with another individual. From each individual molecule may emanat e zero to f
bonds to neighboring molecules and thus this molecule may participate in the formation
of a large cluster of molecules which is called a macromolecule. In the so-called sol–gel
transition, an infinitely large macromolecule is formed. This infinitely large macromolecule
Copyright 2005 by Marcel Dekker. All Rights Reserved.
is called a gel whereas a collection of finite clusters is called a sol independently from
the fact that the gel may be formed by crosslinking the molecules in the solid state. A gel
usually coexists with a sol: the finite clusters are then trapped in the interior of the gel.
Gelation is the phase transition from a state without a gel to a state with a gel, i.e., gelation
involves the formation of an infinite network [1,2,5,6–9].
The conversion factor p (see W. H. Carothers, p ¼ extent of reaction) is the fraction
of bonds which have been formed between the monomer s of the system, i.e., the ratio of
the actual number of bonds at the given moment to the maximally possible number of such
bonds. Thus, for p ¼ 0, no bonds have been formed and all monomers remain isolated
1-clusters. In the other extre me, p ¼ 1, all possible bonds between monomers have been
formed and thus all monomers in the system have clustered into one infinite network, with
no sol phase left. Thus for small p no gel is present wher eas for p close to unity one such
network exists. The gel is, in fact, considered as one molecule. Therefore, there is in general
a sharp phase transition at some intermediate critical point p ¼ p
c
, where an infinite cluster
starts to appear: a gel for p above p
c
, a sol for p below p
c
. This point p ¼ p
c
is the gel point

and may be the analog of a liquid –gas critical point: For p below p
c
, only a sol is present
just as for T above T
c
only a supercritical gas exists. But for p above p
c
, sol and gel coexist
with each other; similarly for T below T
c
vapor and liquid coexist at equilibrium on the
vapor pressure curve. However, we do not assert that these thermal phase transitions
and gelation have the same critical behavior. Also, in gelation there is no phase separation:
Whereas the vapor is above the liquid, the sol is within the gel. The liquid–gas transition is
a thermodynamic phase transition whereas gelation deals with geometrical connections
(i.e., with bonds). At least in simple gelation models the temperature plays only a minor
role co mpared with its dominating influence on the thermodynamic phase transitions.
Such simple gelation theories often make the assumption that the conversion p alone
determines the behavior of the gelation process, though p may depend on temperature T,
concentration c of monomers, and time t.
Early theoretical approaches to the gel-formation [1–4] as the Flory–Stockmayer
theory do not take into account several aspects which naturally occur as the individual
molecules grow to form the gel, such as cyclic bond formation, excluded volume effects
and steric hinderance. The Flory–Stockmayer theory assumes that in the gelation process
each bond between two individual monomeric, oligomeric or polymeric molecules is
formed randomly. Thus this theory assumes point-like monomers. This apparently is
not the case when already existing macromolecules are crosslinked, i.e., in vulcanization
reactions as well as in copolymerization reactions of macromolecules with the
functionality f  3 with bifunctional monomers.
Besides the polymer networks which are generated from homogeneous solution or in

bulk either by crosslinking processes of already existing pre-polymers or by crosslinking
copolymerization reactions and which are completely insoluble in any solvent, there are
also existing network particles in much smaller dimensions which are called microgels and
which form in very dilute solution in copolymerizing a monofunctional and a difunctional
vinyl monomer, e.g., such as styrene and divinylbenzene, or which are formed in emulsion
copolymerization of such comonomers. A microgel is an intramolecularly crosslinked
macromolecule which is dispersed in normal or colloidal solutions, in which, depending on
the degree of crosslinking and on the nature of the solvent, it is more or less swollen [10].
The IUPAC Commission on Macromolecular Nomenclature recommended mic ronetwork
as a term for microgel [11] and defi ned it as a highly ramified macromolecule of colloidal
dimensions. However, ‘micro’ refers to dimensions of more than one micrometer whereas
the dimensions of the so-called microgels are in the range of nanometers.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Probably most network structures obtained by copolymerization reactions of
bifunctional monomers and larger fractions of monomers with a higher functionality
are inhomogeneous, consisting of more densely crosslinked domains embedded in a less
densely crosslinked matrix, often with fluent transitions.
Besides the inhomogeneity due to a non-uniform distribution of crosslinks,
other inhomogeneities due to pre-existing orders, network defects (unreacted groups,
intramolecular loops and chain entanglements) or inhomogeneities due to phase
separation during the crosslinking process may contribute to network structures [7].
It may be concluded therefore that network inhomogeneity is a widespread structural
phenomenon of crosslinked polymers.
For any existing polymer network the most important parameters are the crosslink
density, the functionality of the crosslinks, that is the number of elastic network chains
tied to one given crosslink, the number of dangling chains (with only one end attached to
the network), molecular weight and molecular weight distribution of the elastic chains in
the network, the number of loops and the number of trapped entanglements.
III. THEORETICAL CONSIDERATIONS
Polymerization reactions comprising monomers of the A–B plus A

f
type (with f > 2) in the
presence of B–B monomers will lead not only to branching but also to a crosslinked polymer
structure. Branches from one polymer molecule will be capable of reacting with those of
another polymer molecule because of the presence of the B–B reactant. Crosslinking can be
pictured as leading to the structure I in which two polymer chains have been joined together
(crosslinked) by a branch. The branch joining the two chains is referred to as a crosslink.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
A crosslink can be formed whenever there are two branch es that have different
functional groups at their ends, that is, one has an A group and the other a B group.
Crosslinking will also occur in other polymerization reactions involving reactants with
functionalities f greater than two. These include the polymerizations
A A þ B
f
!
A A þ B B þ B
f
!
A
f
þ B
f
!
In order to control the crosslinking reaction so that it can be used properly it
is important to understand the relationship be tween gelation and conversion, that is
consumption of monomers and/or functional groups, that is also called extent of reaction.
Two general approaches have been used to relate the extent of reaction at the gel point to
the composition of the polymerization system based on calculating when X
n
and X

w
,
respectively, reach the limit of infinite size.
X
n
!1
The first one considering the gel point when the number average degree of
polymerization X
n
becomes infinite X
n
!1in a polycondensation reaction was given by
the pioneer W. H. Carothers himself [12]. This approach is based on the simple assumption
that the reactive groups in the system only are consumed by chemical reaction; no
branching or cyclization events are taken into account. If the average functionality of all
functional groups present in the system of two monomers A and B in equimolar amounts
is named f
avg
, the average functionality of a mixture of monomers is the average number of
functional groups per monomer molecule and is given by
f
avg
¼
X
N
i
f
i
.
X

N
i
which of course is the general formula to calculate the average specifics of a great number
of individuals. Thus for a system consisting of 2 moles of lycerol (a triol, f ¼ 3) and 3 moles
of adipic acid (a diacid, f ¼ 2), the total number of functional groups is 12 per 5 monomer
molecules, and f
avg
therefore simply is 12/5 or 2.4. For a system consisting of equimolar
amounts of glycerol, adipic acid, and acetic acid (a monoacid), the total number of
functional groups is 6 per 3 monomer molecules and f
avg
simply is 6/3 or 2.
In a system containing stiochiometric numbers of A and B groups, the number
of monomer molecules present initially is N
0
and the corresponding total number of
functional groups is N
0
f
avg
.IfN is the number of molecules after reaction has occurred,
then 2(N
0
N) is the number of functional groups that have reacted. The extent of reaction
p is the fraction of functional groups lost
p¼ 2ðN
0
 NÞ=N
0
f

avg
while the degree of polymerization is
X
n
¼N
0
=N
Copyright 2005 by Marcel Dekker. All Rights Reserved.
This is the so-called Carothers equation which relates the degree of polymerization to the
number of mo lecules present in the polymerizing system. From combination of both these
equations it follows that
X
n
¼ 2=2  pf
avg
or by rearrangement
p¼ 2=f
avg
 2=X
n
f
avg
This equation is equivalent to the Carothers equation, and in this expression it relates to
the extent of reaction and degree of polymerization to the average functionality f
avg
of the
system.
At the gel point the number average degree of polymerization X
n
becomes infinite

and therefore the secon d term in the previous equation is zero. Thus, the critical extent of
reaction p
c
at the gel point is given by
p
c
¼ 2=f
avg
This equation allows us to calculate the extent of reaction to which the reaction has to be
pushed to reach the onset of gelation in the reaction mixture of reacting monomers from
its average functionality.
In the example given above of reacting a dibasic acid, adipic acid, with a
trifunctional alcohol, glycerol, which is of the type A
2
B
3
, we have to take 2 moles of
glycerol and 3 of adipic acid, or 5 altogether, containing 12 equivalents and f
avg
¼ 12/5 ¼
2.4. Then at X
n
¼1, p ¼ 2/2.4 and the limit of reaction will be 5/6 ¼ 0.833. This, in fact,
represents the maximum amount of reaction that can occur before gelation under any
distribution of combinations, provided only, that the reaction is all intermolecular.
X
w
!1
Flory [1,2] and also Stockmayer [3,4] used a statistical approach to derive an expres-
sion for predicting the extent of reaction at the time where gelation will occur by calculating

when X
w
approaches infinite size. This statistical approach in its simplest form assumes
that the reactivity of all functional groups of the same type is the same and independent of
molecular size and shape. It is further assumed that there are no intramolecular reactions
between functional groups on the same molecule such a s cyclizatio n reactions.
For the ease of demonstration how the branching reaction in a step-growth
polymerization reaction of A–A þ B–B þ A
f
molecules proceeds, Flory has used a simple
picture to sketch the branching procedure which at some critical point finally leads to
gelation [13]
A A þ B B þ A
f
! A
ð f1 Þ
AðB BA AÞ
n
B BA A
ð f1 Þ
The center unit in Figure 1 is given by the segme nt to the right of the arrow with
the two A
f
at the end as branching sites. Infinite networks are formed when n number of
chains or chain segments give rise to more n chains through branching of some of them.
The criterion for gelation in a system containing a reactant of functionality f is that at least
Copyright 2005 by Marcel Dekker. All Rights Reserved.
one of the ( f  1) chain segments radiating from a branch unit will in turn be connected to
another branch unit (note: f is not identical to f
avg

used by Carothers [12]). The probability
for this occurring is simply 1/( f  1) and the critical branching coefficient a
c
for gel
formation is
a
c
¼ 1ð f  1Þ
When a( f  1) equals 1, a ch ain segment will, on average, be succeeded by a( f  1)
chains. Of these a( f  1) chains a portion a will each end in a branch point so that
a
2
( f  1)
2
more chains are creat ed. The branching process continues with the number of
succeeding chains becoming progessively greater through each succeeding branching
reaction.
If all groups (of the same kind) are equally reactive, regardless of the status of other
groups belonging to the same unit, the probability P
A
that any particular A group has
reacted equals the fraction of the As which have reacted; similarly, P
B
is defined. If r is the
ratio of all A to all B groups, then
P
B
¼rP
A
since the number of reacted A groups equals the number of reacted B groups.

The probability that a given functional group (A) of a branch unit is connected to a
sequence of 2n þ 1 bifunctional units followed by a branch unit is
½P
A
P
B
ð1  rÞ
n
P
A
P
B
r
where r is the ratio of As belonging to branch units to the total number of As. Then
a ¼
X
1
n¼0
P
A
P
B
ð1  rÞ½
n
P
A
P
B
r
¼ P

A
P
B
r=½1  P
A
P
B
ð1  rÞ
¼ rP
2
A
r=½1  rP
2
A
ð1  rÞ ¼P
2
B
r=½r  P
2
B
ð1  rÞ
Figure 1 Schematic representation of a trifunctionally branched three-dimensional polymer
molecule [13].
Copyright 2005 by Marcel Dekker. All Rights Reserved.
It will depend on the analytical circumstances which of the unreacted groups, A or B, is the
one to determine which of the equations will be used.
Combination of a
c
¼ 1( f  1) and a ¼ rP
2

A
r /[1rP
2
A
(1  r)] ¼ P
2
B
r /[r  P
2
B
(1  r)]
yields a useful expression for the extent of reaction (of the A functional groups) at the gel
point
p
c
¼1=fr½1 þ rð f  2Þg
1=2
When the two functional groups are present in equivalent numbers, r ¼ 1 and P
A
¼ P
B
¼ P,
then
a ¼ P
2
r=½1  P
2
ð1  rÞ
and
p

c
¼ 1=½1 þ rð f  2Þ
1=2
In the reaction of glycerol, f ¼ 3, with equivalent amounts of several diacids, the gel
point was observed [14,15] at an extent of reaction of 0.765. The predicted values of p
c
are
0.709 and 0.833 calculated from [13] (Flory, statistical) and [12] (Carothers), respectively.
Flory [13] studi ed several syst ems composed of diethylene glycol ( f ¼ 2), 1,2, 3-propane-
tricarboxylic acid ( f ¼ 3), and either succinic or adipic acid ( f ¼ 2) with both stoichiometric
and nonstoichiometric amounts of hydroxyl and carboxyl groups, see Table 1.
The observed p
c
values as in many other similar systems fall approximately midway
between the two calculated values. The Carothers equation [12] gives a high v alue for p
c
.
The experimental p
c
values are close to but always higher than those calculated from
the Flory equation [13]. Two reasons can be given for this difference: first the occurence of
intramolecular cyclization and second unequal functi onal group reactivity. Both factors
were ignored in the theoretical derivations for p.
Although both the Carothers and statistical approaches are used for the practical
prediction of gel points, the statistical approach is the more frequently employed. The
statistical method is preferred, since it theoretically gives the gel point for the largest sized
molecules in a size distribution.
Some theoretical evaluations of the effect of intramolecular cyclization on gelation
have been carried out [6,16,17]. The main conclusion is that, although high reactant
concentrations decrease the tendency toward cyclization, there is at least some cyclizati on

occurring even in bulk polymerizations. Thus, even after correcting for unequal reactivity
of functional groups, one can expect the actual p
c
in a crosslinking system to be larger than
a calculated p
c
value.
Table 1 Gel point for polymers containing tricarboxylic acid [13].
Extent of reaction at gel point ( p
c
)
r ¼ [CO
2
H]/[OH] r Calculated from [12] Calculated from [13] Observed
1.000 0.293 0.951 0.879 0.911
1.000 0.194 0.968 0.916 0.939
1.002 0.404 0.933 0.843 0.894
0.800 0.375 1.063 0.955 0.991
Copyright 2005 by Marcel Dekker. All Rights Reserved.
IV. CROSSLINKING — CONCEPT
Among all crosslinking strategies which are used to synthesize polymer networks, three
different classes are in common application:
1. One-shot crosslinking of multifunctional monomers or copolymerization with
difunctional monomers,
2. two-stage crosslinking via prepolymers,
3. crosslinking of high molecular weight polymers.
Into the first category of crosslinking strategies fall the formation of poly(styrene-co-
divinylbenzene) resins, the methacrylic resins and some others, and among those also
a small fraction of the so-called microgels. In general, these resins are formed of monomers
which in linear polymerization lead to thermoplastic polymers such as poly(styrene),

polyacrylics or methacrylics a.s.o. High glass trans ition temperature of the linear
polymers and high melt viscosity makes it unattractive to process premade linear
thermoplastics prior to a second step of crosslinking reaction. Incorporation of pendant
C–C– double bonds into the linear chains by copolymerization with small quantities
of a difunctional monomer and thereby avoiding early stage crosslinking is difficult to
handle and such polymers would be very sensitive to undergo uncontrolled network
formation.
One-shot crosslinking of multifunctional monomers and copolyme rization therefore
is limited to the radical induced copolymerization of styrene and some derivatives with
divinylbenzene or of methacrylates with ethyleneglycol dimethacrylate as crosslinker
in suspension polymerization to form densely crosslinked polymer beads for applications
such as ion exchange resins, Merrifield resins, polymer supports for chemical reagents
especially with the aspect of combinatorial syntheses.
Into the second category of crosslinking strategies fall the processes of preparing
polymer networks which make use of prepolymers. These are two-stage processes in which
in the first stage, overhelmingly in step-growth polymerization reactions, prepolymers
are prepared with molecular weight mostly ranging from 1 to 6  10
3
which are soluble in
organic solvents, fusible and have low melt viscosity. The second stage curing is achieved
either by heat — thermosetting — or, when necessary, by the addition of appropriate
curing agents. Most prominent examples are epoxy resins, phenol-formaldehyde resins,
unsaturated polyesters, and the polyurethane networks.
Into the third catagory fall the vulcani zation reactions of elastomers. These polymers
expose C–C double bonds incorporated in the main chain segments which are necessary
for the crosslinking process referred to as vulcani zation. Natural rubber and the synthetic
elastomers have glass transition temperatures far below the temperature range in which
the crosslinked rubbers are used. The molecular weight of the applied polymers is in the
range of 2–5  10
5

, and natural rubber with an upper molecular weight fraction of
2–4  10
6
has to be degraded to this molecular weight level by mechanical treatment
referred to as mastication. The basis of all processes that come after mastication and
before vulcanization are the operations of blending rubber mixtures, mixing with all
the vulcanization ingredients, calendering, frictioning, extrusion, moulding and combining
with textile fabrics or cords is the flow or viscous deformation of the rubber, more
precisely the rheological behavior. Extrusion, calendering and frictioning all involve
vigorous mechanical working in large machines and hence enormous energy consumption
and heat generation.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
A. General Class ification of Prepolymers [18]
Curing reactions applied to epoxy prepolymers, unsaturated polyesters, resoles, and
novolacs make use of three general classes of prepolymers which are distinguished by the
number and location of sites of functional groups available for subsequent crosslinking reac-
tions. These three general classes have been defined as discussed in the following sections.
1. Random prepolymers. Random prepolymers are those built up from polyfunc-
tional step-growth monomers which have been reacted randomly and which are
capable of forming crosslinked polymers directly. Monomer conversion in the
first-stage polymerization reaction for the formation of these prepolymers is
stopped short and kept below the critical conversion at which network
formation woul d occur. Crosslinking in the second-stage, step-growth poly-
merization reaction is achieved simply by heating to carry the original reaction
past the critical conversion. For this reason, the term thermoset is applied to
these prepolymers, and these are exemplified by the phenol-formaledehyde
resole resins and the glycerol polyesters. The term structoset has been applied to
the other two classes of prepolymers to distinguish them from the thermoset type
because in the other two classes the second-stage crosslinking reaction requires
the addition of a catalyst or monomer, and generally proceeds by a reaction

different from the first-stage reaction.
2. Structoterminal prepolymers. Structoterminal prepolymers are those in which the
reactive sites are located at the ends of the polymer chains. These first-stage
polymers give maximum control of the length and type of chain in the final
network polymer. The epoxy prepolymers may be considered examples of
this class if the second-stage reaction occurs overwhelmingly through reaction of
the terminal epoxide functional groups. If the aliphatic hydroxyl groups along
the chain in epoxy prepolymers become significantly involved in the crosslinking
reaction, then these polymers are more properly included in the third class of
prepolymers.
3. Structopendant prepolymers . Structopendant prepolymers are those in which the
crosslink sites are distributed in either a regular or random order along the chain.
Examples of this class are the unsaturated polyesters and the novolac resins.
V. PHENOL-FORMALDEHYDE RESINS
Phenol-formaldehyde condensates were among the first synthetic polymeric materials on
the market. It was Baekeland at the beginning of the 20th century who in 1907 defined the
differences between basic or acidic react ion conditions and the different molar ratios on
the reaction procedure and the resulting molecular structure. He was able to manufacture
a thermosetting resin and made applications for a patent [19] (Bakelite).
Most phenolic resins are heat hardenable or thermosetting. The resin may be
delivered to the user ready to be cured or it may be in the temporarily thermoplastic
novolac form to which a hardener, commonly hexamethylenetetramine–urotropin, will be
added. The major categories of uses for phenolics are
 Molding compounds
 Coatings
 Industrial bonding resins.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
The latter includes resins for grinding wheels and coated abrasives, laminating, plywood
adhesives, glass wool thermal insulation and bonded organic fiber patting, foundry sand
bonding, wood waste bonding, and other miscellaneous applications.

A. Reaction of Phenol and Formaldehyde Under Basic Conditions
The base-catalyzed first-step reaction of phenol ( f ¼ 3, because reaction can take place in
two ortho and one para position) and formaldehyde ( f ¼ 2) with an excess of formaldehyde
of about 15 mol% closely resembles an aldol addition and yields mixtures of
monomolecular methylolphenols and also dimers, trimers and the corresponding
polynuclear compounds according to a generalized reaction scheme given in (1b). In
commercial processes formaldehyde is added in aqueous solution. Sodium hydroxide,
ammonia and hexamethylenetetramine–urotropin, sodium carbonate, calcium-,
magnesium-, and barium-hydroxide and tertiary amines are used as catalysts. After the
hydroxybenzyl alcohol has been formed in the first step, the condensation steps to
form oligomers are likely to be a Michael type of addition to a base-induced dehydration
product of the hydroxybenzyl alcohol. Detailed studies have been presented by Martin [20]
and Megson [21].
ð1Þ
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Such mixtures, whose exact composition depend on the phenol–formaldehyde ratio
and the reaction conditions employed, are termed resoles or resole prepolymers. The
resoles are generally neutralized or made slightly acidic before the second-stage reaction is
accomplished by heating. The second-stage polycondensation and crosslinking takes place
by the formation of methylene and dibenzyl ether linkages between the benzene rings to
yield a network structure of type I. The relative importance of the methylene and ether
bridges is not well established, although both are definitely formed. Higher reaction
temperatures favor the formation of the methylene bridges.
B. Curing of Resol Prepolymers
Heat curing of resols usually is carried out at temperatures in the range 130–200

C.
Below 150

C the formation of dibenzyl ether bridges is predominant whereas at higher

temperatures methylene bridge formation is favored. This was nicely shown by the
investigations of Ka
¨
mmerer et al. who carried out polycondensation reaction of 2,6-
bis(hydroxymethyl-4-methylphenol to the corresponding poly(benzyl ether) [24] with
molecular weights ranging from 2500 to 20,000.
Although at lower temperatures only water is liberated but also water and
formaldehyde at temperatures above 150

C [25], the water to formaldehyde ratio is not
an exact measure of the ratio of benzyl ether to methylene bridge formation, because it is
known that the yield of isolable formaldehyde is considerably less than the theoretical
yield [21].
If curing is carried out above 180

C in the presence of air, some oxidation reaction
takes place which gives a reddish color to the final product. Quinone structures are
responsible for the color and researchers were able even to isolate quinone methi des
formed in pyrolysis reactions [26].
C. Reaction of Phenol and Formaldehyde Under Acidic Conditions
The reaction between phenol and formaldehyde under strongly acidic conditions can be
regarded as an electrophilic substitution reaction, route (b) in Scheme 1 [28]. The catalysts
most frequently used are sulfuric acid, oxalic acids or p-toluene sulfonic acid. By the
addition of a proton to formaldehyde a hydroxymethylene carbenium ion is formed which
Copyright 2005 by Marcel Dekker. All Rights Reserved.
undergoes an electrophilic hydroxyalkylation reaction mostly in the o-position of phenol.
From this o-methylol phenol compound water is eliminated by reaction of the methylol
group with a proton thus yielding a benzylium type carbenium ion which then undergoes
very fast alkylation reaction of a second phenol molecule in the o-position with the
generation of a new proton [20–22,27]. Continued methylolation and methylene bridge

formation by these reactions leads to the formation of polynuclear compounds of
considerable complexity. Under strongly acidic conditions, methylol substitution and
methylene-bridge formation both occur predominantly at p-positions [29]. The pH most
favorable for the formation of the o-products is between 4 and 5.
D. Curing of Novolac Prepolymers
Novolacs require an auxiliary chemical crosslinking agent. The most widely used
crosslinker is hexamethylenetetramine, and the products in this curing reaction are
influenced by the molar ratio of phenol nuclei to hexamethylenetetramine. At a phenol
nucleus to hexamethylenetetramine ratio of 6 : 1, the products turn out to contain little or
even no nitrogen, and the reaction appears to an almost entirely one of methylene-bridge
formation. At a mole ratio 0.5 : 1 or higher, nitrogen enters into the product, and the
nitrogen content of the products can come close to 10% with the amount of ammonia
evolved proportionately decreased.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
The reaction of curing is not clear. It is known that under controlled conditions
phenol and hexamethylenetetramine form a crystalline salt of the stiochiometric
composition C
6
H
12
N
4
 3C
6
H
5
OH [30] which, when heated, evolves ammonia with the
formation of an insoluble, infusible polymer [31]. In the presence of water, hexamethyl-
enetetramine hydrolyzes with the formation of two moles of dimethylolamine DMA, one
mole of formaldehyde and two moles of ammonia. Water is ubiquitious in novolacs and

therefore under basic reaction conditions in the presence of tert and sec amines and also
ammonia as shown in the chart, methylene bridges are formed by entering formaldehyde
into the reaction. With increasing amounts of hexamethylenetetramine, the benzylamine
type bridges become predominant.
Cured novolacs show a more or less slightly yellow color. There is some indication in
the literature that the benzylamine type bridges are converted to azomethines by hydrogen
elimination under heating conditions applied in the curing reaction [20].
Bender et al. found that the o,o
0
-compounds have a much more rapid cure rate than
isomeric ‘novolacs’ [23]. The gel times for the 2,2
0
, 4,4
0
, and 2,4
0
isomers at 160

C have
been reported to 60, 175, and 240 sec, respectively.
VI. UREA- AND MELAMIN-FORMALDEHYDE RESINS
Urea 1 ( f ¼ 4) and melamin 2, 2,4,6-triamino-1,3,5-triazin ( f ¼ 6) under basic or acidic
conditions react with formaldehyde ( f ¼ 2) rather similar to the phenol–formaldehyde
reaction. The reaction products are called aminoplastics.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Polymerization of urea and formaldehyde in a 1.5 : 1 ratio in the first-stage reaction
yields various methylolureas as prepolymers [32–36], which in a second-stage reaction are
cured by heat (thermosetting) under neutral or slightly acidic conditions. Control of the
extent of reaction is achieved by pH (by the use of buffers) and temperature control.
The reaction rate increases with increasing acidity [37, 38]. The prepolymer can be made at

varying pH levels depending on the reaction temperature. Polymerization is stopped by
bringing the pH close to neutral and cooling.
The second-stage, crosslinking reaction of the prepolymers unde r acidic conditions
causes the formation of a network containing principally a random mixture of linear
and branched substituted trimethylenetriamine repeating units and, to some extent, also
methylene ether bridges and methylene bridges [35,39]. The latter are exclusively formed
under strongly acidic conditions [40].
Copyright 2005 by Marcel Dekker. All Rights Reserved.
The formation and crosslinking of random prepolymers from melamine, 2,4,6-
triamino-1,3,5-triazin, and formaldehyde follows in a similar manner [33,34,41–43], but,
unlike urea, melamin readily forms polymethylol compounds with two methylol groups
on a single nitrogen atom. Paper chromatographic separation of the products of
this reaction, in which an excess of formaldehyde greater than 2.1 was used, revealed
the presence of all possible methylol compounds from the monosubstituted to the
hexasubstituted derivatives [44].
VII. EPOXY RESINS
Epoxy resins as a class of crosslinked polymers are prepared by a two-step polymerization
sequence. The first step which provides prepolymers, or more exactly: preoligomers,
is based on the step-growth polymerization reaction of an alkylene epoxide which contains
a functional group to react with a bi- or multifunctional nucleophile by which prepolymers
are formed containing two epoxy endgroups. In the second step of the preparation of the
resins, these tetrafunctional (at least) prepolymers are cured with appropriate curing
agents. Table 2 co mpiles a representative selection of di- and multi-epoxides both as alkyl
and cycloalkyl epoxides and the most widely used curing reagents.
The most widely used pair of monomers to prepare an epoxy prepolymer are 2,2
0
-
bis(4-hydroxyphenyl)propane (referred to as bisphenol-A) and epichlorohydrin, the
epoxide of allylchloride. The formation of the prepolymer can be seen to involve two
different kinds of reactions. The first one is a ba se-catalyzed nucleophilic ring-opening

reaction of bisphenol-A with excess of epichlorohydrin to yield an intermediate b-chloro
alcoholate which readily loses the chlorin anion reforming an oxirane ring. Further
nucleophilic ring-opening reaction of bisphenol-A with the terminal epoxy gro ups leads to
oligomers with a degree of polymerization up to 15 or 20, but it is also possible to prepare
high molecular weight linear polymers from this reaction by careful control of monomer
ratio and reaction conditions [45]. The two ring-opening reactions occur almost exclusively
by attack of the nucleophile on the primary carbon atom of the oxirane group [46].
Depending on the conditions of the polymerization reaction, these low molecular
weight polymers can contain one or more branches as a result from the reaction of
the pendant aliphatic hydroxyl groups with epichlorohydrin monomer. In most cases,
however, the chains are generally linear because of the much higher acidity of the phenolic
hydroxyl group. At high conversions, when the concentration of phenolic hydroxyl groups
drops to a very low level, under the base-catalyzed reaction conditions formation and
reaction of alkoxide ions become competitive and polymer chain branching may occur.
Polymers of this type with molecular weight exceeding 8000 are undesirable because
of their high viscosity and limited solubility, which make processing in the second-
stage, crosslinking-reaction difficult to perform. The oligomers of the diglycidylether of
bisphenol-A (DGEBA) are the most commonly epoxy resins, therefore a great deal of
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Table 2.
Aliphatic epoxy monomers and
pre-polymers (selection)
Curing agents
prim./sec. Amines
tert. Amines
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Aliphatic-cycloaliphatic epoxy compounds Acid anhydrides
Polymerization catalysts
such as amine complexes of Lewis acids [65]
or diaryliodonium salts [66], photocrosslinking [67]

Copyright 2005 by Marcel Dekker. All Rights Reserved.
investigations with respect to the processibility behavior before crosslinking is focused on
this oligomer [47].
A. Aliphatic-Cycloaliphatic Epoxy Compounds and Prepolymers [48]
Aliphatic-cycloaliphatic epoxy compounds (ACECs) contain different epoxy groups in the
molecule: glycidyl, i.e., 2,3-epoxypropyl groups, and cycloaliphatic, i.e., 1,2-epoxycyclo-
pentane or 1,2-epoxycyclohexane rings, for which molecules 3 and 4 are cha racteristic.
The most important feature of ACECs is the different reactivity of the cycloaliphatic
epoxy group and the glycidyl epoxy group with various curing agents. This property
affects some important properties of ACECs. Table 2 contains a good selection of ACECs
which have been described in the literature. It is possible to consume different epoxy
groups consecutively in the course of curing [49,5 0]. In the early stages of curing, reaction
of carboxyl groups with cycloaliphatic epoxy groups prevails, resulting in the formation of
a polymer chain with a loose crosslinking. In later stages, the chain extension and dense
crosslinking proceeds as a result of the conversion of glycidyl groups and of the remaining
cycloaliphatic epoxy groups. Eventually, the network density is achieved. The network
density is determined by the ACEC–hardener ratio and by the conditions of the curing
process.
The reaction sequence is different if amines are applied as curing agents. In the first
stage the glycidyl groups react followed by the cycloaliphatic epoxy groups which then
enter into the reaction with the curing agent.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Nevertheless, the sequential entering of different epoxy groups into the reaction,
irrespective of the acidic or basic character of the curing agent, is a very important feature
of the crosslinking process of ACECs because it conditions the formation of a regular
polymer network [51].
B. Curing
The epoxy prepolymers are considered as structopendant prepolymers because of the
pendant aliphatic hydroxyl groups or as structoterminal prepolymers with respect to the
terminal epoxy groups [52].

An acid anhydride as curing agent is bifunctional ( f ¼ 2) and crosslinking occurs
primarily through the hydroxyl groups. In this reaction, the prepolymer acts as a
structopendant prepolymer. Maleic anhydride introduces C–C– double bonds into the
resin. Mostly phthalic anhydride and pyromellitic anhydride are used.
Anhydrides react initially with the hydroxyl groups in the prepolymers to form half-
esters, and the generated carboxyl groups in this half-ester can condense with another
hydroxyl group. Also the reaction of the carboxyl group with an ep oxy group is possible
[53,54], but these reactions are much slower than the initial alcohol–anhydride reaction
and are not shown in the above picture. For these reasons dianhydrides are very effective
crosslinking agents, and because of the great number of hydroxyl groups in the
prepolymer, curing with dianhydrides can form very densely crosslinked, second-stage
polymers if used in relatively high concentrations.
The prepolymer is a structoterminal prepolymer when amines are used as
crosslinkers. Crosslinking in this case involves the base-catalyzed ring-opening of the
oxirane groups. Both primary and secondary amines are used as crosslinking agents [55].
Since each N–H bond is reactive in this process, primary and secondary amine functional
groups have a crosslinking functionality f equal to two and one, respectively. A variety of
amines such as diethylene triamine ( f ¼ 5), triethylene tetramine ( f ¼ 6), m-phenylene-
diamine ( f ¼ 4) and others are used as crosslinking agents. The presence of other reactants
is required to foster this ring-opening reaction because the nucleophilic ring-opening
reaction of an amine with an oxacyclopropane is not only accelerated by, but, in fact,
requires the presence of an active proton-donor [56]. Anhydrous diethylamine and
Copyright 2005 by Marcel Dekker. All Rights Reserved.
oxacyclopropane do not react, but the reaction proceeds readily in the presence of
catalytic amounts of proton-donating agents like water, methanol or ethanol [57].
Similarly, the reaction of epoxybenzylacetophenone with morpholine or with piperidine
in benzene or ether is extremely slow, but proceeds smoothly in methanol at room
temperature [58]. The reaction of phenyl glycidyl ether with diethylamine in the absence of
solvents shows a sigmoidal rate curve, which can be attributed to the autocatalytic effect
of the hydroxyl groups in the product [59], while in proton-donating solvents the reaction

is greatly accelerated and the sigmoidal form of the rate curve disappears. By protonation
of the oxacyclopropane oxygen, an intermediate oxonium ion is formed which facilitates
the nucleophilic attack on the carbon atom. In the case of the epoxy end groups of the
prepolymers, this nucleophilic attack is exclusively directed to the sec carbon atom. Phenol
has been found to be a particularly useful proton-donating accelerator. And it has been
shown also that the reaction of oxacyclopropane with aniline in the presence of small
amounts of water [60] or acids [61] is proportional to the concentration of water or to the
strength of the acid. Different mechanisms have been proposed by Smith [56], Tana ka [62],
and King et al. [63], but they have not yet been confirmed [64].
VIII. CROSSLINKING–POLYURETHANE NETWORKS
Structoterminal prepolymers with two isocyanate endgroups prepared by reaction of
polyethers containing two hydroxyl endgroups with diisocyanates are the basis for the
formation of polyurethane networks. They can be made either in melt or in solution,
but polyurethanes with melting points much above 200

C are difficult to prepare in melt
because of the thermal instability of the urethane linkage above 220

C [68]. The molecular
weight of the prepolymers generally is in the range of 1–10  10
3
.
The fundamental reactions of an isocyanato group which proceed easily at room
temperature or slightly above are reaction with (i) an alipha tic or aromatic hydroxyl group
in a reactivity order primary > secondary > tertiary OH-group, and with (ii) primary or
(iii) secondary amines. With carboxylic acids (iv) an amide is form ed and CO
2
is liberated,
Copyright 2005 by Marcel Dekker. All Rights Reserved.
and with water (v) isocyanates give substituted carbamic acids which decarboxylate with

extreme ease to give an amine which is recycled into reaction (ii). Thus, in (i)–(iii) linkages
are formed which directly help to build up a polymer chain, and in (iv) and (v) functional
groups are created which can further react.
At elevated temperatures (120–140

C), the structures formed in (i)–(iii) are able
to undergo further reaction with isocyanate groups according to (vi)–(viii), which, in this
way, can be used for crosslinking.
A. Crosslinking
One-shot crosslinking is a step-growth polymerization of a difunctional alcohol with a
diisocyanate in the presence of a small amount of a polyfunctional alcohol. In the presence
of small quantities of water, carbon dioxide is liberated from hydrolysis of some
isocyanate groups and acts as a foaming agent in polyurethane foam production.
Two-stage crosslinking, in which in the first stage is the synthesis of a prepolymer
containing two isocyanato endgroups in the classical way of reaction (a diol either of low
or of high molecular weight with an excess of diisocyanate) and the second step to form
the network, can be accomplished by
1. addition of multifunctional alcohols, and the resulting bridges are urethane
linkages,
2. addition of diamines which extend the linear prepolymer chain via urea linkages,
which, in turn, add to other isocyanate endgroups to form biuret branching sites
and eventually crosslinks,
3. excess diisocyanate, and the network is formed by the reaction of isocyanate
with the preformed urethane linkages according to reaction (vi), and the bridges
are of an allophanate structure,
Copyright 2005 by Marcel Dekker. All Rights Reserved.
4. at elevated temperatures at which the isocyanate end groups of the prepolymer
react intermolecularly with urethane linkages in the main chain thus also
forming allophanate bridges.
Again, if water is present, the network is expanded by the carbon dioxide liberated from

hydrolysis reaction of isocyanate groups, and the resulting primary amino groups are
recycled into the reaction.
The relative rates of the different types of chain extension and crosslinking reactions
will depend in part on the structure of the diisocyanate monomer envolved as indicated by
the rate constants for reactions of several diisocyanate monomers with water and with
various functional groups which can be found in polyurethanes [69]. It is noteworthy that
hexamethylene diisocyanate reacts very slowly with urethane groups and therefore would be
a very poor crosslinking agent. In addition, it should be mentioned that the relative rates of
the various reaction can be changed significantly by the presence of a catalyst and by the
type of the catalyst which, in general, is a base, i.e., an amine, or a metal salt (Table 3).
IX. UNSATURATED POLYESTERS UPs
Unsaturated polyesters have a widespread field of applications. In almost all cases,
unsaturation in these materials is introduced by the acid component when the prepoly-
mers are manufactured. These prepolymers can be of either the structoterminal or
structopendant type depending on the location of the unsaturated linkages. The average
molecular weight is in the range of 1– 5  10
3
.
Table 3 Rate constants for reactions of diisocyanate monomers with different substrates [69].
Rate constant,
a
k  10
4
, liters mole
1
sec
1
Diisocyanate monomer Hydroxyl Water Urea Amine Urethane
p-Phenylene 36.0 7.8 13.0 17.0 1.8
2-Chloro-1,4-phenylene 38.0 3.6 13.0 23.0 –

2,4-Tolylene 21.0 5.8 2.2 36.0 0.7
2,6-Tolylene 7.4 4.2 6.3 6.9 –
1,5-Naphthalene 4.0 0.7 8.7 7.1 0.6
Hexamethylene 8.3 0.5 1.1 2.4 2  10
5
a
For reactions at 100

C, except for 1,5-naphthalene diisocyanate, 130

C.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Structopendant unsaturated polyesters, containing double bonds within the
polymer chain, are produced by step-growth polycondensation reaction of an unsaturated
diacid or anhydride, such as fumaric acid or maleic anhydride, with a diol. Structural
unsymmetry in the diol component lowers the viscosity of the prepolymer. Mostly,
crosslinking of the structopendant unsaturated polyester is accomplished by copolymer-
ization with alkene monomers such as styrene, methyl methacrylate, or others using
radical initiators.
Structoterminal polyesters have terminal C–C– double bonds which are introduced
by terminating the step-growth polycondensation reaction by the addition of an
unsaturated monocarboxylic acid. The monocarboxylic acid is usually a fatty acid derived
from linseed oil, and the polyester is referred to as an alkyd resin. Crosslinking is
accomplished most simply by oxidation with atmospheric oxygen.
X. SILICON RUBBER
Network formation to build up crosslinked silicones is based on linear polysiloxane
precursors. In most cases, poly(dimethylsiloxane) is used, a smaller fraction of products
also contains phenyl substituents to silicon. The precursors are all prepared by the usual
way of ring-opening polymerization of cyclic tri- or tetrasiloxanes which are previously
prepared by cyclocondensation of the corresponding dichlorosilanes [70]. Silicon rubbers

are very flexible because of the very low glass transition temperature T
g
of – 100

C.
Higher stiffness is achieved by the addition of fillers such as silicates which by means of
their HO–Si-groups at the surface interact with the silicon Si–O–Si-linkage via hydrogen
bonding.
A. Curing
Curing is achieved either by random radical crosslinking of polysiloxanes by heating with
peroxides or by room temperature vulcanization techniques making use of reactive end
groups of the precursors.
B. Radical Crosslinking
The radical crosslinking method involves heating the polysiloxane with dicumyl peroxide,
ditertiary butyl peroxide, ben zoyl peroxide, or bis-2,4-dichlorobenzoyl peroxide. The
peroxide radical abstracts hydrogen from the polymer chain and creates a radical site on the
interior of the chain. Two such sites interact to randomly form the crosslink. The major
disadvantage of this technique is its commercial inefficiency. Obviously, vulcanization can
only be carried out in a mould to produce the final silicon rubber product.
C. Crosslinking Via Reactive Structoterminal Precursors
Platinum catalyzed anti-Markoffnikov addition of hydrosilanes to C–C– double bonds is
a widely applied reaction to form Si–C linkages. For this hydrosilylation reaction the
platinum based catalyst has to be added only in the ppm scale. Two different polysiloxane
components are necessary to achieve network formation by the so-called addition
vulcanization of polysiloxanes, one stru ctoterminal polysiloxane precursor providing vinyl
endgroups and a polysiloxane crosslinker providing hydrosilane groups as chain segments
Copyright 2005 by Marcel Dekker. All Rights Reserved.
in the main chain.
The advantage of this addition vulcanization is that no revision occurs because
no byproducts are formed which might interfere with the network in terms of a reversible

network degradation. Furthermore, although this vulcanization reaction is considerably
accelerated at elevated temperatures. For a given receipe, the curing characteristics at
different temperatures are shown in Table 4.
A second group of room temperature vulcanization techniques ha ve been developed
based upon linear polysiloxane chains terminated by hydroxyl groups. Curing can be
achieved by two ways which both make use of hydrolyzation reactions of labile Si–O–R
bonds. Two-component vulcanization RTV-2.
The so-called RTV-2 method — room temperature vulcanization of a two
component system — adds a crosslinking agent such as tri- or tetraalkoxysilane and a
metallic salt catalyst to hydroxyl terminated polysiloxane precursors. The hydroxyl end
groups react with the silicic ester, e.g., tetraethyl silicate, in a condensation reaction and
ethanol is liberated. This condensation reaction is catalyzed by stannous-based catalysts
such as dibutyltindilaurate.
Table 4 Curing times for a typical addition vulcanization reaction of
silicones [71], probe thickness 1 cm.
Processing time at room temperature 60 min
Demoulding at room temperature After 10 hr
Final hardness at room temperature After 24 hr
Final hardness at 50

C After 1 hr
Final hardness at 100

C After 10 min
Final hardness at 150

C After 5 min
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Water is provided by atmospheric moisture. The alcohol liberated from the
condensation reaction has to be removed and this is sufficiently achieved by diffusion

into the environment. If the vulcanization is carried out in a closed system at elevated
temperature, there is the danger of revision.
The so-called RTV-1 method — room temperature vulcanization of a one component
system — is based on the finding that hydroxyl terminated siloxanes do not react with
certain crosslinking agents under strictly dry conditions. Technically, this is achieved by the
addition of an excess of crosslinker which reacts much faster with water than with the silanol
groups and thereby acts as a drying agent. As atmospheric moisture diffuses into the system,
crosslinking starts to occur. Therefore, these one-component silicon rubber precursors are
stored in one-compartment cartridges and can be applied very easily.
The different react ion steps envolved are demonstrated below for an acetoxy system:
in the first step, under dry conditions the hydroxyl terminated polymer reacts with
triacetoxymethylsilane to form a diacetoxy-terminated siloxane:
By the addition of water, the silylacetoxy end groups are hydrolyzed and a silanol
end group is set free which in the next step of reaction can undergo condensation reaction
with an acetoxy group of a second polymer molecule. By consecutive condensation
reactions the polysiloxane network is formed.
Copyright 2005 by Marcel Dekker. All Rights Reserved.