Aust. J. Chem.

, 1995, 48, 323-331

The Chemistry of Novolac Resins. I

A Review on the Use of Models*

Mark G. LooneyA and David H. ~ o l o r n o n * ) ~

A School of Chemistry, The University of Melbourne, Parkville, Vic. 3052.

* Author to whom correspondence should be addressed.

Abstract
The model systems ranging from the simple phenolic derivatives xylenol and cresol to higher
oligomers such as (13), (15) and (17) are being used to study the chemistry of the curing
of phenol-formaldehyde resins with hexamethylenetetramine. The key intermediates in these
systems have been identified and this knowledge is being used to develop an understanding
of the curing of the complex commercial resins.

Introduction
Phenol-formaldehyde polymers, the oldest known synthetic polymer system,
are used in large quantities in a wide variety of applications1 and offer the
combination of highly desirable properties at a relatively cheap price. Due to
the fact that subtle variations in reaction conditions have been shown to have a
profound effect on the mechanical and physical properties of the resulting resin,
the chemistry of phenol-formaldehyde resins has been of great interest since they
were commercially developed by Baekeland in 1907.~Some of the factors identified
in determining the final properties of these resins are the phenol-formaldehyde
ratio, pH, temperature and the type of catalyst (acid or alkaline) used in the
preparation of the resin.lt3 Although the properties of the final resin are readily
determined, relating these properties to the chemistry of the phenol-formaldehyde
resin is more difficult and extremely challenging.
The phenol-formaldehyde ratio ( P I F ) is a most important factor as it leads
to two different classes of synthetic polymers.3~4The first class of resins are the
resoles which, in general terms, are made under alkaline conditions, from phenol-
formaldehyde mixtures with P / F < 1. The resoles are classed as thermosetting
polymers, as they typically have free methyl01 (hydroxymethyl) groups which
enable them to cure or crosslink without further addition of other components.
* Dedicated to Athel Beckwith in recognition of his contribution to chemistry.
Knop, A., and Pilato, L. A., 'Phenolic Resins' (Springer: Germany 1985).
Baekeland, L. H., U.S. Pat. 942,699 (1907); U.S. Pat. 949,671 (1907).
Keutgen, W. A., in 'Encyclopedia of Polymer Science and Technology' Vol. 10, pp. 1-73
(Interscience: New York 1969).
Megson, N. J. L., 'Phenolic Resin Chemistry1 pp. 133-159 (Academic: New York-London
1958).

Manuscript received 11 April 1994 0004-9425/95/020323$05.00

 M. G. Looney and D. H. Solomon

Alternatively, resins generated with P / F > 1, often by using acidic catalysis in

the phenol-formaldehyde condensation, are referred to as novolacs. A random low
molecular weight novolac resin is illustrated in Fig. 1. Novolacs are thermoplastic
polymers which do not have free methyl01 groups, and thus require the addition
of a formaldehyde source, to enable further curing to insoluble and infusible
products.

Fig. 1. A random novolac resin.

Hexamethylenetetramine (HMTA)is commonly used as a source of anhydrous

formaldehyde in the manufacture of synthetic resins. The main advantages of
HMTA over formaldehyde is that it is a solid hence easily handled, reacts only in
the presence of catalysts and formaldehyde acceptors or under the influence of
heat, and is cheap being made from formaldehyde and ammonia. During curing
reactions with novolac resin, the latter was postulated as a possible catalyst
which was readily lost by vo1atilization.l
Technologists were aware of differences in properties of novolac resins cured with
HMTA and other commercially used formaldehyde sources such as paraformaldehyde
and trioxan. However, the correlation between these observations with the chemistry
involved in reaction with the novolac was unclear. Although the chemistry of
novolac-HMTAmixtures has been extensively studied, there still remains a great
deal of uncertainty, due to difficulties associated with analysing complex polymer
systems, as to the overall processes.
In this paper we review our efforts in studying the chemistry of novolac resins
with the crosslinking agent HMTA. Our approach involves reducing the complex
polymer structures to simple models. The chemistry of these model systems is
studied and this knowledge is then applied to the complex polymers themselves.

Model Phenol Systems

The commerically available novolac resins are of a relatively low molecular
weight (up to approximately 2000) and are made up of phenolic monomer
units that are linked randomly by methylene bridges (Fig. 2). The internal
phenolic rings can be simplified to either an ortho-ortho disubstituted phenol,
Chemistry of Novolac Resins. I

i.e. 2,6-xylenol, or an ortho-para disubstituted phenol, 2,4-xylenol. Similarly, the

terminal phenolic rings of the novolac can be simplified to either ortho-cresol
or para-cresol. Therefore it is appropriate to consider the cresols and xylenols
as suitable models to examine the chemistry of the curing process with the
crosslinking agent HMTA.

Fig. 2. Random linking of phenolic monomer units by methylene bridges.

Another factor is that commerical novolacs always contain some unchanged

phenol in amounts which range from 2 up to 10% of the total weight. The phenol
content is of particular importance as it has to been shown to affect properties
such as the novolac gel time with the curing agent HMTA, as well as, to a small
extent, its v i s c ~ s i t Also,
~.~~ Sprung7
~ ~ ~ has shown that phenol is more reactive
than the cresols or xylenols towards formaldehyde (Table 1); however, no similar
studies with HMTA are reported. Therefore it is considered that the study of the
curing process with phenolic model systems should also include phenol.
Katovic, Z., in 'Weyerhaeuser Science Symposium' Vol. 2, pp. 85-104 (Weyerhaeuser Company:
Washington 1981).
Takamitsu, N., Okamura, T., and Nishimura, M., Nzppon Kagaku Kaishi, 1973, 11, 2196.
Sprung, M. M., J. Am. Chem. Soc., 1941, 63, 334.
 M. G. Looney and D. H. Solomon

The model studies were carried out with 2,4- and 2,6-xylenol as representative
of the internal phenolic units, o- and p-cresol as the terminal phenolic units, and
phenol. The results of the HMTA reactions with these models will be reported in
detail8 but a summary is provided in this review.

Table 1. Relative reaction rates of various phenols with formaldehyde1"

Compound Relative reactivity Compound Relative reactivity
2,6-Xylenol 0.16 phenol 1.00
o-Cresol 0.26 2,3,5-trimethylphenol 1.49
p-Cresol 0.35 m-cresol 2.88
2,5-Xylenol 0.71 3,5-xylenol 7.75
3,4-Xylenol 0.83

The preliminary results from reaction of 2,4- and 2,6-xylenol [(I) and (3),
respectively] with HMTA, at 130°C, indicate that the major product formed
through attack at the ortho position is the cyclic benzoxazine (2) (Scheme 1)
which has been found to be relatively stable, still being present after prolonged
periods of time (24 h) at 130°C or after heating at elevated temperatures over
several hours. The reaction at the para position of 2,6-xylenol (3) gives rise to
di- or tri-benzylamines (4) and ( 5 ) , respectively. These first-formed products are
relatively unstable, and decompose readily to form the methylene-bridged dimer
(6) (Scheme 2). The study of the reaction of the xylenols (1) and (3) with
HMTA was then extended into examining the system where we have competitive

Scheme 2

* de Bruyn, P. J., Looney, M. G., Potter, A. C., and Solomon, D. H., unpublished data.
Chemistry of Novolac Resins. I

processes. The results indicate that, in the reaction of a 1: 1 mixture of 2,4-xylenol

(1) and 2,6-xylenol (3) with HMTA, the reaction at the ortho position in (1) is
preferred to the reaction at the para position in (3).
The study was then extended to the more functional systems o- or p-cresol,
and it was found that in both cases the majority of the linkages were based on
the benzoxazine structure (Fig. 3). In the case of p-cresol, with two free ortho
positions on the molecule, the major product from reaction with HMTA was the
polybenzoxazines (7). The major product isolated from the reaction of o-cresol,
under similar conditions, was the simple benzoxazine (8). This reaction also
highlights the difference in reactivity with EMTA of the ortho and pclrcl positions
of a phenolic system; the majority of the HMTA has been consumed in benzoxazine
formation, through reaction in the ortho position, in preference t o benzylamine
formation.

I (7)
Fig. 3. Benzoxazine structures.

The reaction of phenol with HMTA is a far more complex system to study
due to the availability of three reactive sites on the molecule. The preliminary
results indicate that the benzoxazine linkage is the most common first-formed
intermediate although the assignment is complicated by formation of high molecular
weight products. However, from solid-state 13C n.m.r. experiments, these higher
molecular weight products appear to resemble the polybenzoxazines (7) formed
in the p-cresol reaction.
Water (up to 2%) is generally present in commercial novolac resins, and it has
been reported that varying the amount of water can have a substantial effect on
the viscosity of the mixture, thus highlighting that its effect on the chemistry
needs to be s t ~ d i e d . ~It* has
~ $ ~been
~ reported that as the concentration of
water in these reactions decreases the gel time becomes progressively longer. For
example, at 150°C in the presence of 10% HMTA, there is a 67% increase in the
value for the rate of gelation going from a 1 . 4 to 0.1% water content. Dienes,ll
in 1949, claimed that an increase of 1% in the water content of a novolac resin
decreases the viscosity about 5 times. The influence of water on the mechanism
of the curing of novolac resins is currently being studied by using the models
discussed above in an effort to clarify the observed effects.
Tonogai, S., Sakaguchi, Y., and Seto, S., J. Appl. Polym. Sci., 1978, 22, 3225.
Tonogai, S., Hasegawa, K., and Kondo, H., Polym. Eng. Sci., 1980, 20, 1132.
Dienes, G. J., J. Colloid Sci., 1949, 4, 257.
 M. G. Looney and D. H. Solomon

The study of the chemistry of model systems was first reported by Zinke12
and Hultzsch,13 and the formation of benzylamines in these reactions was well
documented. However, it was not until the work of Kopf and Wagner14 that
the benzoxazine structure was first proposed as an intermediate in the curing
process. Our studies so far have illustrated that the formation of benzoxazine
is a major process in these syst.ems and work is continuing to define the overall
process that occurs during the curing of these simple model systems.

Model Novolac Resins

-7

l h e model systems demonstrated that the curing process with HMTA can
proceed through various intermediates and that the reaction pathway leading to
these different products is determined by the first point of attack, whether it is
at the ortho or para position. This raises the possibility that the structure of
the novolac resin could be of vital importance in relation to the properties of
the final cured polymer.
Therefore the study was extended to include model novolac resin systems with up
to eight phenol residues (octnmers) in which we selectively controlled the number
of free ortho and free para positions. The synthetic scheme for the preparation of
the model resins is based on the ion-assisted ortho-specific phenol-formaldehyde
oligomerization reported by Casiraghi and coworker^.^^^^^ They prepared the
series (10) ( n = 2, 4 and 6) selectively linked by ortho-ortho coupling, and which
contains the maximum number of free para positions (Scheme 3).

(10) n = 2 , 4 and 6 Scheme 3

We have recently prepared model novolac resins which contain the maximum
number of free ortho positions.17>18This required the development of an efficient
protection/deprotection sequence to control the regioselective ortho-ortho coupling
reaction late in the synthetic pathway. Hence, monoprotection of the commercially
available para-linked dimer (11) followed by ortho-ortho coupling afforded, after
deprotection, the tetramer (13). The protected tetramer (12) was also the
key intermediate in the synthesis of oligomers containing more than four units.
Protection of (12) and selective deprotection followed by a second coupling reaction
gave the desired linear octamer (15) which contains only free ortho positions
(Scheme 4).
l2 Zinke, A., J. Appl. Chem., 1951, 257.
l3 Hultzsch, K., Chem. Ber., 1949, 82, 16.
l4 Kopf, P. W., and Wagner, E. R., J. Polym. Sci., Polym. Chem. Ed., 1973, 11, 939.
l5 Casiraghi, G., Cornia, M., Sartoria, G., Casnati, G., and Bocchi, V., Makromol. Chem.,
1982, 183, 2611.
l6 Dradi, E., Casiraghi, G., Sartori, G., and Casnati, G., Macromolecules, 1978, 11, 1295.
l7 de Bruyn, P. J., Lim, A. S. C., Looney, M. G., and Solomon, D. H., Tetrahedron Lett.,
1994, 35, 4627.
l8 de Bruyn, P. J., Foo, L. M., Huang, Y., Lim, A. S. C., Looney, M. G., Naug, D. J., and
Solomon, D. H., unpublished data.
Chemistry of Novolac Resins. I

The methodology developed was used to expand the series of model resins to
include structures which vary in the degree of chain branching. For example,
octamer (17), which, like octamer (15), has only free ortho positions, has
been synthesized in two steps, coupling and deprotection, from tetramer (12)
(Scheme 5).

-
OR' OR'
(14) R' = Bu1Ph2Si
(15) R' = H
Scheme 4

(16) R = Bu'MezSi
(17) R = H
Scheme 5

15N,13C-Labelled Hexamine/Model Novolac Reactions

Many of the analytical tools available have been used to study phenol-
formaldehyde resins and their reaction with curing agents such as HMTA. The use
of l H and 13C solution-state n.m.r., however, has been primarily responsible for
developing our understanding of the curing process. For example, in 1973, Kopf
and Wagner14 published their use of lH n.m.r. to identify reactive intermediates
and the cure mechanism through the use of model phenols. Although the model
phenols had been studied previously, Kopf and Wagner were the first group to
identify the benzoxazine as one of the first-formed intermediates. The use of 13C
 M. G. Looney and D. H. Solomon

initially complemented lH n.m.r., but through the efforts of Sojka,lg Kim20 and
Casiraghi16 it has become the main tool in extensively characterizing novolac
resins. However, the products of the n o v o l a c / ~reaction
~ ~ ~ are insoluble and
many of these studies were carried out on the model systems. More detailed
information about the mechanism of the curing process has been determined
through the use of solid-state 13C n.m.r, with cross polarization (up) and magic
angle spinning MAS).^'-^^ Even with this powerful technique, the reactions are
still quite complex and difficult to analyse.
There are reports on the use of 15N-labelled H M T A ~ to *'~identify
~ reaction
mechanisms in other systems, but the iabei has only been used in the
study of ~ O V O ~ ~ C / H M Tsystems
A once. The use of 13C-labelled or 15N-labelled
HMTA to examine the curing of phenolic resins was reported by Maciel who
illustrated the effectiveness of labels in the study of these systems.26 Therefore,
we prepared 13C,15N-labelled HMTA to use in the curing studies as it offered
significant advantages, notably that the same reaction product could be used
to evaluate both the nitrogen and carbon residues of the HMTA that had been
incorporated into the cured resin. The work with 13C,15N-labelled HMTA was
carried out on the simplest model, phenol, and has proceeded onto the resins
(9) and (11) which contain only ortho-ortho or only para-para linkages. To
complement the series the ortho-para linked dimer (18) was also studied.27 The
results from these experiments indicate that the preferred site of attack during
the curing process is still the ortho position of the phenolic ring. Also, from this
study, there is strong evidence to suggest that in cases where two types of ortho
positions (a or b) are available (Fig. 4), as is the case in dimer (18), the preferred
site of attack is the ortho position in the ring resembling para-cresol (b).

Fig. 4. Type a and b ortho positions.

The labelled HMTA is also being used to study commercial novolac systems,28
particularly to determine how the rate of cure affects the final novolac resin
in terms of its structure. Although these systems are extremely complex, the
Sojka, S. A., Wolfe, R. A., Dietz, E. A., Jr, and Dannels, B. F., Macromolecules, 1979, 12,
767.
20 Kim, M. G., Tiedeman, G. T., and Amos, L. W., in LWeyerhauserScience Symposium' Vol.
2. p. 263 (Weyerhaeuser Company: Washington 1981).
Sojka, S. A., Wolfe, R. A., and Guether, G. D., Macromolecules, 1981, 14, 1539.
2 2 Bryson, R. L., Hatfield, G. R., Early, T. A., Palmer, A. R., and Maciel, G. E., Macromolecules,
1983, 16, 1669.
23 Maciel, G. E., Chuang, I.-S., and Gollob, L., Macromolecules, 1984. 17, 1081.
24 Crampton, M. R., Jones, M., Scranage, J . K., and Golding, P., Tetrahedron, 1988, 44, 1679.
25 Cooney, A. P., Crampton, M. R., Jones, M., and Golding, P., J. Heterocycl. Chem., 1987,
24, 1163.
26 Hatfield, G. R., and Maciel, G. E., Macromolecules, 1987, 20, 608.
27 Caulfied, M.J., Lim, A. S. C., Looney, M. G., Solomon, D. H., and Zhang, X., unpublished
data.
28 Zhang, X., Looney, M. G., Potter, A. C., Solomon, D. H., and Whittaker, A. K., unpublished
data.
Chemistry of Novolac Resins. I

observed first-formed intermediates are easily identified by comparison with the

simpler models. Studies are currently under way to identify the pathways that the
first-formed intermediates follow during their breakdown after prolonged heating.
In summary, the double-labelled HMTA enabled us to confirm that the chemistry
observed in the simple compounds was occurring in the higher oligomers. Namely,
the free ortho positions lead to benzoxazine structures, and the free para
positions formed the initial crosslink via benzylamines which then decompose to
methylene-bridged aromatic residues.

Synthetic polymers such as phenol-formaldehyde resins, due to their good

physical and mechanical properties, are used in a large range of commercial
applications. However, an understanding of reaction mechanisms and reactive
intermediates in polymer reactions should provide an insight into how polymers
could be modified to tailor those properties required for particular applications.
To date, the difficulties in handling polymer systems and the limited range of
spectroscopic and analytical methods available, which can be used to study these
systems, have prevented detailed explanations of how properties are directly
related to polymer structure.
The approach to polymer chemistry described here typifies our general philosophy
to the study of the chemistry of complex polymer systems. First, a detailed
study leading to a thorough understanding of the simplest models is required.
Next, and vital to our approach, is the study of more complex systems using as
a basis the knowledge gained from the model studies.
In the specific study on n o v o l a c / ~chemistry,
~~~ we have been able to show
the vast difference in the type of product formed at ortho and at para positions of
the phenolic ring systems, and we have related this difference to other properties.
These are discussed in detail in refs 8, 17, 18, 27 and 28.

Acknowledgments
We thank the Australian Research Council and the Industry, Research and
Development Board (grant number 15068) for supporting this research and
Comalco Aluminium Limited for financial support.