You are on page 1of 26

Progress in Organic Coatings, 14 (1986) 193 - 218 193

MELAMINE/FORMALDEHYDE CROSSLINKERS: CHARACTERIZA-


TION, NETWORK FORMATION AND CROSSLINK DEGRADATION

DAVID R. BAUER
Research Staff, Ford Motor Company, P.O. Box 2053, Dearborn, MI 48121 (U.S.A.)

Contents

1 Introduction.. ................................................... 193


2 The characterization of melamine resins. ............................... 195
3 Crosslinking kinetics ............................................... 199
3 .l Types of reactions. ...................................... 199
3.2 Models for reactions at high conversion ............................. 202
4 Network formation ......................................... 204
4.1 Network models: functionality and crosslink efficiency ................ 204
4.2 Validation and application of network models ....................... 208
5 Crosslink degradation .............................................. 213
5.1 Hydrolysis ................................................... 213
5.2 Photodegradation ............................................. 214
6 Summary.. ..................................................... 216
References ........................................................ 216

1 Introduction

The need to formulate thermosetting coatings at higher and higher


solids levels has resulted in the development of new classes of aminoplast
resins. In particular, melamine/formaldehyde resins have been developed
with lower molecular weight and different functionality from the conven-
tional butylated melamine resins used in low-solids coatings. These changes
have required a better understanding of the crosslinking kinetics, network
formation and durability of these coatings. Several review papers addressing
various aspects of melamine crosslinking have recently appeared. The ap-
plication of melamine crosslinkers to high-solids coatings has been reviewed
by Santer [ 11. The characterization of melamine resins has been reviewed by
Christensen [ 2, 31. Hill and Wicks have discussed melamine crosslinking in a
review of high-solids coatings [4]. Blank has discussed in detail the cross-
linking mechanisms of melamine resins [ 51.
The basic melamine structure consists of an aromatic triazine ring with
six substituent sites, two on each terminal nitrogen. Typical substituents
include amine (i.e. unsubstituted NH), methyl01 (N-CH,OH), alkoxy

0033-0655/86/$9.60 0 Elsevier Sequoia/Printed in The Netherlands


194

MELAMINE/FORMALDEHYDE CROSSLINKERS

A. FULLY ALKY LATED ( HEXAMETHOXYMETHY LMELAM I NE 1

I
CH30CHfNhH20CH3

B. PARTIALLY ALKYLATED (MONOMERIC - METHYLATED)

$H20H
7

ti
H’ ‘CH20CH3

C. PARTIALLY ALKYLATED ( POLYMERIC - BUTYLATED 1

?H20H $H,OH

N
C, H9OCH ‘CH,OC,H,

C4Hg 0CH2 /NhHgO-CH2-N’


‘MELAMINE

Fig. 1. Structures of typical melamine/formaldehyde resins.

(N-CH,OR) and melamine-melamine ether linkages (N-CH,OCH,-N).


Examples of melamine/formaldehyde resins are shown in Fig. 1. Fully-
alkylated melamines are those where nearly all of the groups are alkoxy
groups (>80%). Partially-alkylated melamines are those where a substantial
fraction of the groups are amine or methyl01 groups. The reactivity of the
melamine crosslinker and the properties of coatings crosslinked with
melamine/formaldehyde resins depend in large part on the nature of the
substituents. The resins are synthesized by reacting melamine with formal-
dehyde under alkaline conditions [6]. The ratio of formaldehyde to
melamine determines the amount of residual amine and the concentration
of N-CH20H groups. N-Methyl01 groups can be converted to alkoxy groups
by reaction with alcohols. In high-solids coatings, low molecular weight
fully and partially alkylated melamines (A and B of Fig. 1) have replaced the
high molecular weight butylated melamines (C of Fig. 1) conventionally used
in low-solids solvent-based coatings. Crosslinking takes place by a number of
reactions which will be described in more detail below. The major reactions
195

are the condensation of melamine alkoxy groups with functional groups


on the polymer (hydroxy, carboxy or amide) to form polymer-melamine
crosslinks and the condensation of melamine methyl01 groups with methyl01
and/or amine groups to form melamine-melamine crosslinks. Due to the
complexity of the possible substituents, the characterization of melamine
resins is an important issue. In the section that follows, the characterization
requirements and modern spectroscopic techniques are discussed. The
details of the crosslinking chemistry are summarized in Section 3, where the
question of mechanism is discussed and a model for describing the kinetics
of cure at high conversion presented. In Section 4, the question of how the
crosslinking reactions form networks is addressed, high-solids and low-solids
coatings are compared, and quantitative network models are described and
applied. Finally, in Section 5, recent studies of degradation in melamine
crosslinked networks are discussed.

2 The characterization of melamine resins

The characterization of melamine/formaldehyde resins has been re-


viewed recently by Christensen [2, 31. As a result of the complexity and
diversity of melamine resins, several levels of characterization are possible.
In general, two quantities have to be characterized: the type of functionality
and the molecular weight. The molecular weight can be characterized by
light scattering (which gives a weight average molecular weight) [7] or by
gel permeation chromatography, which if properly calibrated can give the
molecular weight distribution [8]. The characterization of melamine func-
tionality can be more difficult. It is necessary, but not sufficient, to deter-
mine the average concentrations of amine, methyl01 and alkoxy groups. As
will be discussed in Section 3.1, the reactivity of a given group depends
strongly on its neighbors, i.e. NH-CH20CHs alkoxy groups react different-
ly from N(CH,OCHs), groups. Characterization of the distribution of pairs
of groups is important in determining the reactivity of the melamine cross-
linker. The different possible pairs of groups for a melamine with one alkylat-
ing alcohol are listed in Table 1. Mixed methylated and butylated melamines
would have even more types of pair groupings. In addition to the basic
groups described in Fig. 1 and Table 1, melamine resins are often found to
contain low levels of ‘impurity’ groups. One such group is the N-CH,-
OCH,OH group formed when formaldehyde reacts with a methyl01 group
instead of an amine group. The presence of such minor impurities further
complicates characterization. Ideally, the concentration of all pair groupings
should be determined. This in itself can be a formidable task. However, even
this level of characterization may be insufficient for some requirements. For
example, in order to characterize network structure, the distribution of
functionality (i.e. what fraction of melamine molecules have one alkoxy
group, what fraction have two, etc.) may be just as important as the average
functionality.
196

TABLE 1
Melamine/formaldehyde pair groups

H CHzOH
1. -N’ 6. -N’ N-
‘H ‘CH20CH,’

H CHzOH
2. -N’ 7. -N’
‘CHzOH ‘CHzOR

CH20CH2--N-
/H
3. -N\ /N- 8. -N’
CHzOCHz ‘CH,OCH,-N-

CH20CHz-N-
9. -N’
‘CH20R

CHzOH CH20R
5. -N’ 10. -N’
‘CH*OH ‘CH20R

The original methods for the determination of the ratio of formal-


dehyde and alkoxy groups to triazine ring groups were based on chemical
titrations. These methods, though accurate, can be tedious and do not give
additional information about structure. Christensen [ 2] has summarized
the pros and cons of the different chemical methods. Spectroscopic tech-
niques, particularly ‘H NMR spectroscopy, have, by and large, supplanted
the various chemical methods since they are generally more convenient and
more powerful [2,3,7,9 - 111. ‘H NMR chemical shifts are reported in
Table 2. As can be seen from the table, ‘H NMR spectroscopy is particularly
useful for determining the amount of NH and NH2 groups, the amount of
residual methylol, and distinguishing between the different alkoxy groups.
Integration of ‘H NMR peaks gives a quantitative measure of the amount
of each group. For example, the area under the peak at 4.6 - 5.1 ppm yields
the amount of formaldehyde (total CH,O) used. Methods for quantitation
have been discussed and ‘H NMR spectroscopy has been used together with
other techniques to characterize commercial melamine/formaldehyde resins
131.
Unfortunately, ‘H NMR spectroscopy does not have sufficient resolu-
tion (except in the case of NH and NH?) to distinguish pair groupings. It
cannot for example, distinguish between NHCH,OCHs and N(CH20CH&
alkoxy groups. 13C NMR spectroscopy, on the other hand, has much greater
potential for resolving such subtle differences. Where ‘H chemical shifts have
a range of some 10 ppm, i3C chemical shifts range over 200 ppm. The i3C
chemical shifts in model melamine compounds as well as commercial prod-
ucts have been studied by a number of researchers [12 - 141, and typical %
chemical shift data are reported in Table 3 [ 31. The ability to resolve subtle
changes in structure is remarkable. Chemical shift differences of less than
197

TABLE 2
‘H NMR chemical shift@

Proton Chemical shift Structure

-NH2* 5.8 - 6.2 broad singlet


-NH* 7.2 - 7.4 broad singlet
-N-CH20H* 5.4 - 5.6 broad triplet
-N-CH2*OH
-N-CH2*OR 4.6 - 5.1 broad peak
-N-CH2*-N-
-0-CH3* 3.2 singlet
-0-CH2*CH3 3.0 - 3.2 quadruplet
-0--CH&H? 1.2 triplet
-O-CHX*CH2CH3 3.5 triplet
-0-CH&H2*CH3 1.5 multiplet
-0-CH2CH2CH3* 1.1 triplet
-0-CH* (CH& 3.6 - 4.0 multiplet
-0-CH( CH;), 1.0 - 1.1 doublet
-0-CH2*CH&H2CH3 3.0 - 3.3 triplet
TO--CH~CH~*CH~*CH, 1.0 - 1.5 multiplet
-0-CH2CH2CH2CH3* 0.7 - 1.0 triplet
-O-CH2*CH(CH& 3.5 multiplet
-0-CH$H*(CH& 1.5 multiplet
-O-CH&H( CH:), 0.8 doublet

YXemical shifts in ppm relative to TMS.

0.25 ppm can usually be resolved. With continued improvements in NMR


spectrometers, early problems of low sensitivity have been eliminated.
13C NMR spectroscopy is likely to be widely used in the future in conjunc-
tion with ‘H NMR spectroscopy to characterize melamine resins. Unlike
‘H NMR spectra for which quantitation is fairly straightforward, care must
be take in the quantitation of 13C NMR spectra. For example, nuclear
Overhauser effects (which are a result of proton decoupling) as well as
ratios of spin-lattice relaxation times to the pulse delay time must be con-
sidered. Nevertheless, quantitative 13C NMR spectra of urea/formaldehyde
resins have been reported [ 151.
One group that is important in the cure and subsequent degradation
of melamine crosslinked coatings is the NCH*N group. This group is not
formed during the synthesis of melamine resins. When melamine resins are
prepared in alkaline solution, the methyl01 groups which polymerize form
only NCH,OCH*N linkages [16 - 191. For this reason no solution 13Cchem-
ical shift data have been reported for the NCHzN group. The NCHzN linkage
is formed only on cure or hydrolytic degradation under acidic conditions
[20,21]. At this point the melamine resin is crosslinked and conventional
solution NMR spectroscopy cannot be employed. Bauer et al. have used
198

TABLE 3
13C NMR chemical shiftsa

Carbon atom Chemical shift

N\
N//C*--NHz 167.4

N\
N/C*-NH( CH,O-) 166.0 - 166.6

N\
N//C*-N(CH20-)2 165.4 - 165.8

-NHCH20C*H20- 93.0

-N( C*H20CH3)2 76.8


-NHC*H20CH, 72.6
-N(C*H2OC&)2 74.4
-NHC*HzOC&19 71.0
-NHC*H20C*H2NH- 68.0 - 69.0
-NHC*H20H 64.5

-N(CH20C*H3)2 55.4
-NHCH20C*H3 54.5

-NCH20C*H2CH2CH2CH3 66.9
-NCH20CH2*CH2CH2CH3 31.4
-NCH20CH2CH2C*H2CH3 18.9
-NCH20CH2CH2CH2C*H3 13.7

-NCH20CH2C*H(CH3)2 28.1
-NCH20CH2CH( C*H3)2 18.9

Whemical shifts in ppm relative to TMS.

magic angle 13C NMR spectroscopy to obtain high-resolution NMR spectra


of hydrolyzed melamine [22]. They found that the chemical shift of the
NCHzN linkage ranged over 47 - 54 ppm depending on the substituent
pattern. Magic angle NMR spectroscopy is a promising new technique for
studying crosslinked polymers and has been reviewed by Havens and Koenig
1231.
Although 13C NMR spectroscopy is much more sensitive to subtle
changes in structure than ‘H NMR spectroscopy or the wet chemical tech-
niques, the complex structures of commercial melamine resins may still be
too difficult to fully resolve and characterize by conventional 13C NMR
spectroscopic methods. Novel NMR pulse sequences are becoming available
which may greatly increase the power of NMR spectroscopy to sort out
these subtle differences in structure. In particular, two-dimensional (2-d)
NMR spectroscopy which uses combinations of different pulse sequences
has the potential for determining which species are connected to one another
in highly complex spectra [ 241. Another technique that may be particularly
199

applicable to characterization of melamine resins and is only now becoming


commercially practical is “N NMR spectroscopy. To the author’s knowl-
edge, no 15N NMR spectra of melamine/formaldehyde resins have been
reported. Chuang et al. have recently reported solid-state measurements of
‘sN NMR spectra of urea/formaldehyde resins [ 251. Even though the infor-
mation available from NMR spectra is large, it can only determine the aver-
age concentrations of pairs of groups. The distributions in concentrations of
pairs of groups about those averages cannot be determined by NMR spectros-
copy alone.

3 Crosslinking kinetics

3.1 Types of reactions


The most extensive studies of the crosslinking reactions of melamine/
formaldehyde resins are those of Blank and coworkers [5,26 - 281. The
possible crosslinking reactions that can occur during the cure with hydroxy
functional polymers are summarized in Table 4. They can be split up into
two classes, i.e. those which form polymer-melamine crosslinks and those
which form melamine-melamine crosslinks. The actual amounts of the
different reactions that occur will depend on the functionality of the
melamine crosslinker, the degree of catalysis and the bake conditions em-
ployed. Fully-alkylated melamines (A of Fig. 1) contain almost exclusively
N(CHZOCH,), or other alkoxy groups. For this melamine, only reactions
1, 2, 4 and 6 need be considered. Of these, reaction 1 is by far the most
important under normal conditions of cure. Reactions 2 and 6, which
involve water, are generally not important except under very humid bake
conditions for coatings with an excess of alkoxy to polymer hydroxy groups.
Reaction 4, which involves the self-condensation of alkoxy groups to yield a
melamine-melamine crosslink and methylal, is only observed at high bake
temperatures [ 201. Thus, to a first approximation, the cure of fully-alkylated

TABLE 4
Melamine/formaldehyde reactions

1 -NCHzOCHB + ROH + -NCH20R + CH30H


2 2 -NCH20CHB + Hz0 --f -NCH2N- + H&=0 + 2 CH30H
3 -NCHzOCH3 + -NH + -NCH2N- + CH30H
4 2 -NCH20CH, + -NCH2N- + CH30CH20CH3
5 -NCH20CH3 + -NCH20H -+ -NCH20CH2N- + CH30H
6 -NCH20CH3 + H20 + -NCH20H + CH30H
7 -NCH20H + -NH + H2C=0
8 -NCH20H + -NH -+ -NCH2N- + Hz0
9 2 -NCH20H + -NCH2N- + H2C=0 + Hz0
10 -NCH20H + ROH -+ -NCH20R + Hz0
11 2 -NCH20H -+ -NCH20CH2N- + H20
200

melamines can be described by a single reaction, the formation of polymer-


melamine crosslinks via the reaction of alkoxy groups on the melamine with
hydroxy groups on the polymer. Of course, this approximation breaks down
at high bake temperature or high levels of catalysts.
Alkoxy groups also react with other functional groups including car-
boxylic acids and amides [26]. Amide functionality is not widely used in
high-solids coatings. Carboxylic acid groups are often incorporated into the
polymer to promote cure. The relative reactivity of carboxylic acid groups
and hydroxy groups depends on the level of catalyst and the type of cross-
linker. Bauer and Dickie found that hydroxy groups tend to be more reactive
with fully-alkylated melamines with strong acid catalysis, while carboxy
groups are more reactive with partially-alkylated melamines [ 7, 291.
Partially-alkylated melamines (types B and C of Fig. 1) contain signif-
icant amounts of alkoxy, methyl01 and amine functionality. Thus, in prin-
ciple, all 11 reactions of Table 4 have to be considered. As was the case for
the fully-alkylated melamines, the reactions can be separated into major and
minor reactions. For reactions involving alkoxy groups, only reaction 1 is a
major reaction and under normal baking conditions the other reactions
involving alkoxy groups can be ignored. The reactions involving methyl01
and amine groups are more complex and less well understood. Reaction 11
is probably unimportant for acid-catalyzed cures. Reaction 10 was not
observed by Bauer and Dickie [ 71 during the cure of a partially-alkylated
melamine. Reaction 5 is also probably minor. The major reactions involving
methyl01 and amine groups involve self-condensation to form an NCH,N
melamine-melamine crosslink. It is not clear whether two methyl01 groups
can react directly to form the crosslink (reaction 9) or whether one of
the groups must deformylate (reaction 7) to amine before the crosslink can
form via reaction 8. In any case, the two major reactions of partially-alkylated
melamines are the formation of polymer-melamine crosslinks from alkoxy
groups and the formation of melamine-melamine crosslinks from methyl01
and/or amine groups.
A variety of techniques have been used to follow the crosslinking
reactions in melamine/formaldehyde based coatings. Early studies often used
measures of the physical properties (e.g. hardness) to characterize cure
[28, 30 - 341. These measurements can give a qualitative picture of the
crosslinking and are useful in relating formulation variables to final film
properties. However, the use of such measurements to infer crosslinking
kinetics or mechanism is highly dangerous, simply because the physical
properties depend primarily on network structure which is a function of
more variables than just the extent of crosslinking. It is far better to actually
measure the extent of crosslinking directly than to try to infer it from
measurements of physical properties. Two main techniques have been em-
ployed to measure this chemistry; IR spectroscopy [7, 29,31, 35 - 371 and
gas chromatography [27, 381. IR spectroscopy is a convenient technique
for following major reactions such as reactions 1, 8 and 9 by following
changes in functional group band intensities. The measurements are quanti-
201

SPECIFIC ACID CATALYSIS


Ii
_N/cH20cn3 +HS A _N/CH2zCH3
-
FAST
‘CH,OCH, k-l kHpOCHI

H
_ N/W$CH~
+ Ct$OHt SLOW
‘CH2QCH3 7y \CHZOCH 3

H
_ /H: _N/CHZgR
-t-ROH kj FAST
‘CH,OCH, +%y \CH20CH3

H
p-q R kq _y+@R +H*
FAST
-N\ -k_,
CH,OCH, ‘CH,OCH,

Fig. 2. Cure by specific acid catalysis.

tative (2 - 10% depending on the reaction). Bands of interest include the


hydroxy, amine and carboxy region between 3200 and 3500 cm-“, the
methoxy baud at 915 cm-’ and the melam~e triazine ring band at 810
cm-’ (an internal standard). Some care must be taken in interpreting the
hydroxy region due to band overlap [7]. IR spectroscopy is not terribly
sensitive to the presence of the minor reactions listed in Table 4, These
reactions have been followed using gas chromatographic detection of gases
evolved during cure (e.g. methanol, water, formaldehyde and methylal)
[ 27,381. This technique requires a somewhat more complex experimental
setup than for IR measurements, but can yield potentially more detailed
information concerning the cure chemistry.
The types of reactions are different in fully- and partially-alkyiated
melamines. In addition, the mechanism of reaction 1, which occurs with
both melamine types, is also different leading to different bake and catalyst
requirements. For fury-~kyla~d melamines, a specific acid-catalyzed mech-
anism has been proposed as shown in Fig. 2 IS]. This mechanism involves
reactions which are all reversible. The reaction is driven to completion by
the volatilization of CHsOH. Analysis of the kinetics of crosslinking via this
mechanism as well as alternative mechanisms will be presented below. The
mechanism for partially-alkylated melamines is shown in Fig. 3 IS]. The
crosslinking of this group requires general acid catalysis and goes through a
different intermediate from that of the fully-alkylated melamine. This
difference in cure mechanism is the main reason why it is important in the
characterization of melamine resins to be able to distinguish the different
pair groupings. Of course it is possible that a p~i~ly-~ky~~d melamine
will have both types of alkoxy groups and thus undergo crosslinking by
both mechanisms. Under normal conditions, weak (e.g. carboxylic) acids
can be used to catalyze cure of groups which react via the mechanism
depicted in Fig. 3, while strong acids (e.g. sulfonic) have to be used to
202

GENERAL ACID CATALYSIS

_ N/%0CH3 _ k3 _N/wCH3
+H$=Ot
‘CHzOH 7y \H

,CH,OCH, ,CH2-:CH3
kg, +
-N +HA -N CA- FAST
\H - \H

-NiCHZbH3+A- k7 -N=CH2+CH30H)+HA SLOW


‘H 7iy

,CH20R
__&+ FAST
- N=CH,+ ROH -N
z ‘H

Fig. 3. Cure by general acid catalysis.

catalyze cure of groups which react via the mechanism depicted in Fig. 2.
The difference in acid strength required is determined by the difference in
mechanism and is also related to the difference in ease of protonation of
partially-alkylated melamines uersus fully-alkylated melamines [ 391.

3.2 Models for cure at high conversion


Development of kinetic models for cure are generally based on an
understanding of the mechanism of the reaction. In the case of the reaction
of hydroxy groups with alkoxy groups of a fully-alkylated melamine (reac-
tion l), Berge [40] and Blank [5] have proposed the Snl mechanism of
Fig. 2. Holmberg [32] and others [37] have argued in favor of an SN2
mechanism. This is an important consideration since the mechanisms predict
different orders of reaction which would lead to different cure and network
formation behavior, particularly at high conversion where desirable physical
properties develop. Support for the SN2 mechanism comes from three sources.
First, the hardness of coatings increases with increasing hydroxy level as well
as melamine level [32]. Second, the extent of reaction appears to depend on
the nature of the polymer functionality [37]. Finally, Lazzara found a good
fit of his kinetic data to second-order kinetics [ 381. On the other hand, the
Snl mechanism is favored by solution work [40]. Also Bauer and Budde
found that their kinetic data fit models based on the S,l mechanism better
than the SN2 [ 411. It should be noted that the increasing hardness of a coat-
ing with increasing hydroxy functionality may have nothing to do with
kinetics, but rather with the fact that the crosslink density is proportional
to the hydroxy level at a given extent of reaction. Resolution of this con-
troversy may be possible by a closer examination of the implications of
the reaction mechanism depicted in Fig. 2. Steady-state analysis of the reac-
tions listed in Fig. 2 results in the following expression for the rate of
disappearance of the polymer hydroxy group:
203

= -_[H+l~‘{W,[ROHl [OCH,] - k-,k-,[CH,OH]([ROH”] - [ROH])}


k3[ ROH] + k_,[ CH,OH]

(1)

where [ ROH] , [ OCH,] and [ CHsOH] denote the concentrations of polymer


hydroxy, melamine methoxy and methanol in the film. It is also assumed
that kg = k-,, kz = k_, and k,/k_, = k-,/k, = K’. Finally, it is assumed that all
of the methoxy groups are equally reactive and that the reaction is not
diffusion-controlled. As can be seen from eqn. (l), the form of the rate
equation depends strongly on how the concentration of methanol varies with
time. The reaction is first order only if the methanol concentration is zero.
Under certain conditions, the rate of disappearance of hydroxy groups could
appear to be pseudo second order and still be consistent with eqn. (1).
Formally, the rate of reaction is sensitive to polymer functionality (i.e. the
value of k, and kM3). In order to use measurements of chemical reactions to
determine mechanism in these coatings, accurate measurements at high con-
version levels are required. At high conversion the assumptions used to derive
eqn. (1) may not be completely valid. Thus, the use of kinetic data to infer
mechanism is dangerous for this system, and the use of a particular kinetic
expression must be empirical in nature to some extent.
Bauer and Budde found that an expression based on eqn. (l), with
suitable assumptions concerning the methanol concentration, did give good
agreement with experimentally determined extents of reaction up to the
high conversions where final physical properties are achieved [41]. They
could accurately predict the extent of reaction in acrylic copolymer/fully-
alkylated melamine crosslinked coatings as a function of time and temper-
ature. They found that the rate of crosslinking was proportional to the level
of acid catalyst added and to the square root of the acid strength. This
dependence on acid strength is why strong acids are more effective than
weak acids in promoting cure of fully-alkylated melamines. They also found
that the rate constant for the crosslinking reaction had an activation energy
of 12.5 kcal mol-‘. This value can be compared to a value of cu. 15 kcal
mol-’ found by Santer and Anderson [42]. The kinetic expression was also
used to determine the effectiveness of various blocked catalysts. By compar-
ing the rate of crosslinking using blocked catalysts with that of p-toluene-
sulfonic acid, the effectiveness of deblocking could be determined as a func-
tion of temperature [43]. Further support for the SN1 mechanism comes
from the recent work of Meijer [44], who used polarimetry measurements
of the reaction of hexamethoxymethylmelamine with optically active alco-
hols to determine the forward and backward rate constants (k3 and k-& for
primary and secondary alcohols. He found that his data fitted an SN1 rather
than an SN2 mechanism, and that the rate of methanol evaporation played
a key role in the kinetics.
204

There is no equation equivalent to eqn. (1) capable of predicting the


extent of reaction for the crosslinking of partially-alkylated melamines. The
kinetics of the various reactions appear to be complex functions of bake
time, bake temperature and the catalyst strength. In general, high bake tem-
peratures and strong acids favor melamine-melamine crosslink formation
while low bake temperatures and weak acids favor polymer-melamine cross-
link formation [29]. Although Blank [5] has described a kinetic scheme to
interpret partially-alkylated melamine crosslinking, insufficient data on the
reactions of well-characterized partially-alkylated melamines are available to
develop a realistic model.

4 Network formation

4.1 Network models: functionality and crosslink efficiency


The extent of crosslinking is one of the key variables in determining the
final structure of the network; however, it is not the only variable. Other
variables such as the equivalent weight of functionality and the distribution
of functionality are also very important. The importance of equivalent
weight is obvious. At a given extent of reaction, decreasing the equivalent
weight increases the number of crosslinks. The role of the distribution of the
functionality (i.e. the number fraction of polymer and crosslinker with a
given number of functional groups) is particularly important in the case of
high-solids coatings which tend to have polymers and crosslinkers with low
functionality. The reason for this is that the number of polymer and cross-
linker molecules with low functionality (0 - 4) plays a key role in determin-
ing the efficiency of crosslinking. This effect can be understood by consider-
ing the reaction of a polymer of functional group n with a difunctional
crosslinker. If n < 2, no network can be formed. If n 2 3, a network can be
formed; however, for small values of n the gel point (i.e. the point at which
the infinite network is first formed) occurs at high extents of reaction. For
n = 3, for example, the gel point occurs at over 70% reaction. Thus, at most
30% of the bonds formed contribute to the network. Flory has described
methods of calculating, in detail, the structures that are formed on cross-
linking in simple systems [45]. Unfortunately, the broad distribution of
functionality and the multitude of potential crosslinking reactions make
such a detailed calculation of network structure impossible in the case of
practical coatings. One parameter that can be calculated simply from the
extents of reaction and equivalent weights of the different functional groups
is the total number of chemical crosslinks formed. As was noted above,
however, and as will be discussed in more detail below, not all crosslinks
formed contribute to the network structure. The key problem becomes one
of determining how efficiently a given amount of crosslinking creates net-
work structure. Kooistra has used the following empirical model to estimate
‘structural’ crosslink density [ 461:
205

rf-2
c, = -
MO
where MO is the functionality equivalent weight, r the extent of reaction and
f the functionality. In this model, the efficiency of crosslinking is (rf - 2)/rf.
The model ignores the fact that in practical coatings there is no single func-
tionality. Rather, there is a .distribution of functionality. It is not obvious
which weighted average to use to obtain a value of f. This expression also
assumes a single crosslinking reaction.
The only detailed effort to model network structure in practical
melamine crosslinked coatings has been by Bauer et al. [7,29, 41,47, 481.
These workers have developed a model based on the probabalistic arguments
of Miller and Macosko [49]. The advantage of the Miller and Macosko
approach is that the data required as input can be obtained for practical
coatings. Although the mathematics is somewhat complex, expressions
which calculate the efficiency of crosslinking can be derived. To illustrate
the Miller-Macosko approach, a derivation will be given for a simple sys-
tem - an ‘n’ functional polyol crosslinked with hexamethoxymethyl-
melamine. Extension to more complex systems will be discussed.
There are two key concepts in the Miller-Macosko formalism. The first
involves directionality. Sitting on a functional group on the polyol one can
imagine looking two ways: back in towards the polymer chain or out from
the chain. If the functional group has reacted, one looks out from the group
on the polyol across the crosslink to the crosslinker molecule. The definition
of out and in for a group on the crosslinker is analogous. The other key
concept involves the difference between finite and infinite chains. Up to the
gel point there are no infinite chains. Above the gel point, there is an infinite
network. However, not all functional groups look either in or out to the
infinite network. The Miller-Macosko approach calculates the probability
that a given group is connected (either in or out) to a finite chain or infinite
network. These concepts are illustrated diagramatically in Fig. 4. For the
purpose of a simple example, it is assumed that the polyol only reacts with
methoxy groups on the melamine and that no other reaction occurs. The
melamine is assumed to be pure six-functional hexamethoxymethylmelamine.
All six groups are assumed to be equally reactive as are the groups on the
polyol. If the extent of reaction of hydroxy is r and the ratio of methoxy
groups to hydroxy groups is m, then the extent of reaction of methoxy is
r/m. The parameter that is calculated by the model is the probability that
looking out from a given group (hydroxy or methoxy) is a finite chain.
This probability is given by P(FA Out) and P(Faout) where A denotes the
polyol and B denotes the melamine. The probability that looking out from
a group on A is finite is given by the sum of two terms, the probability that
the group has not reacted (1- r) and the probability that it has reacted and
that the group that is attached to A is finite. This is the same as r times the
probability that looking into group B is finite [P(FB”)]. Thus the probabil-
ity is:
206

MILLER - MACOSKO
FORMALISM

POLYMER CROSSLINKER

LOOKING OUT FROM ‘A’ TO Q

Fig. 4. Miller-Macosko network structure formalism.

P(FAout) = 1 - r + rP(FBi”) (2)


Looking into the melamine molecule (group B) and seeing a finite chain
requires that looking out from each of the other five arms on the melamine
molecule is a finite chain. Thus, the probability P(Fs’“) is given by the
following:
P(F$“) = P(F$ut)s (3)
Similarly, the probability that a group on B looks out on a finite chain is
given by:
P(Faout) = 1 - (r/m) + (r/m)P(F,‘*) (4)
and the probability that looking in on A is finite is given by:
P(F,in) = ,(F*out)(n - 1)
(5)
Substituting eqn. (3) into (2), and eqn. (5) into (4) yields the following:
P(FAout) = 1 - r + r[P(FBout)15 (6)
P(FBout) = 1 -(r/m) + (r/m)[P(F,oUt)](“-l) (7)
Equations (6) and (7) cannot be solved analytically; however, they can be
solved numerically using such techniques as Newton’s method [7, 471. This
leads to a relation between the probabilities, the extents of reaction and the
functionality. Note the probability is 1 for extents of reaction below the gel
point. The probabilities are used to calculate a quantity termed the elastical-
ly effective crosslink density (C,,) [ 471. The crosslink efficiency is given by
dividing C,, by the total chemical crosslink density. As will be shown below,
C,, gives a good correlation with physical measures of cure. A calculation
207

of C,, is based on an analogy with rubber elasticity theory which states that
a crosslink must be at least three-functional to contribute to the elastic
modulus. Crosslinks are counted by looking out from both polymer func-
tional groups and crosslinker functional groups. For the polyol the func-
tional group must look out to the infinite network [probability = 1 -
P(FAoUt)]. In addition, there must be at least two other paths from polymer
functional groups which lead out to the infinite network. The same criteria
are used for the crosslinker groups. Since two groups yield one crosslink,
C,i is calculated by summing the polymer and crosslinker groups which are
elastically effective and dividing by two. The value of C,i for the network
described above is given by the following:

Cel = (1/2EQ)[l -P(FAoUt)] (1 - [P(FA”“t)](“-l) - (n - 1)

x [ 1 - P(F*““t)] [P(F*““t )]‘“-2’} + (m/2EQ)[l -P(Faout)]

x (1 - [P(FBout)+ 5[1 --P(FBout)] [P(FBout)]4} (8)


where EQ is the equivalent weight of the hydroxy groups on the total coat-
ing resin. This expression makes use of the fact that the probability that
there are two or more additional paths out to the infinite network can be
given by 1 less the probability that there are either no such paths or only
one such path. This definition is somewhat arbitrary and the correlation
between C,, and physical properties must be considered empirical.
Several extensions of the model to more realistic coatings have been
described and others can be envisioned [7, 471. The above model assumes
that the functionality of the polymer and crosslinker are monodisperse. The
Miller-Macosko method can be easily adapted to handle distributions of
functionality. The case of a free-radical initiated randomly-polymerized
copolymer has been treated by Bauer and Budde [47]. The method could
be extended to cover polydispersity in the functionality of the crosslinker
as well. More than one reaction can also be included. For example, in the
case of partially-alkylated melamines, a model which counts both polymer-
melamine and melamine-melamine crosslinks has been developed [7].
Incorporation of the minor reactions is also possible. All that is required is
that the extents of reaction be known and the amount and functionality
of each group be measured. The above model also assumes that all of the
groups are equally reactive. As was discussed in Section 3, this appears to
be a reasonable assumption for hexamethoxymethylmelamine. Miller and
Macosko have extended the model to cover the case on nonequal reactivity
[ 501. In order to use this extension, data concerning the relative reactivities
of the different groups are required. The model also ignores ring formation.
Stanford et al. have shown that the amount of ring formation can, in some
instances, have a large effect on the properties of a crosslinked polymer
[ 511. As will be discussed in Section 4.2, the model gives good agreement for
the physical measures of cure even though ring formation is ignored.
208

4.2 Validation and application of network models


The utility of any model depends on how well it correlates with ex-
perimental data. In the case of models of cure, the parameter that is derived
(in our case the elastically effective crosslink density) must correlate with
physical measures of cure independent of composition variables. Ideally
there will be a single value of C,, for which cure is optimized. Unfortunate-
ly, there is no single physical test which provides data on optimum cure.
Typically, acceptable cure is determined by measurements which are sus-
ceptable to underbake (solvent resistance, water sensitivity, cold cracking,
etc.) and to overbake (intercoat adhesion). Coatings are considered well
cured when they exhibit neither under or overbake problems. Hardness can
also be used to follow cure. Unfortunately hardness (as well as some of the
other physical tests) depends on variables that are unrelated to cure. For
example, hardness depends on the glass transition temperature of the
polymer, as well as the extent of the crosslink density. In principle, swelling
measurements or measurements of the modulus above the glass transition
temperature can be used to measure coating crosslink density. Swelling
measurements on high-solids urethanes have been reported [52]. They
appear to suffer from the fact that at low crosslink density unreacted species
can be extracted from the coating giving a false measure of the crosslink
density. Hill has used the modulus measurements with success [53]. Bauer
and Budde have used methyl ethyl ketone (MEK) solvent resistance as a
physical measure of cure [47]. As shown in Fig. 5, the minimum value of

Fig. 5. Chemical crosslink density and Cd required to achieve acceptable solvent resis-
tance in acrylic melamine coatings as a function of the number average molecular weight
of the acrylic polymer. The symbols containing an X denote non-acrylic polymers. See
ref. 46 for details.
209

16,

IZ-

I.0 -

3
i 08 -

n
b
; 0.6 -

04-

Fig. 6. Plots of C,l uersus the extent of reaction for a polymer of functionality n cross-
linked with hexamethoxymethylmelamine. The equivalent weight (EQ) was 625 and the
ratio of methoxy to hydroxy (m) was 2.0.

Cel necessary to achieve MEK solvent resistance is independent of coating


composition and has a value of 1.0 X 10m3mol g-’ * 10%. By contrast, the
total number of crosslinks required to meet this level of cure increases
drastically with decreasing molecular weight of the polymer. The efficiency
of crosslinking drops dramatically as the molecular weight of the polymer
decreases. Curiously it was found that as far as effective crosslink density
was concerned, polymer-melamine crosslinks were identical to melamine-
melamine crosslinks.
The decrease in crosslinking efficiency with decreasing molecular
weight of the polymer is due to the fact that the functionality also decreases
with decreasing molecular weight. The role of polymer functionality on Cel
is shown in Fig. 6, where C,, was calculated from eqns. (6) and (7) as a
function of the extent of reaction for polymers of different functionality
crosslinked with hexamethoxymethylmelamine. Polymers with only one
or zero hydroxy groups cannot form a network. The value of Cel for the
polymer with two groups is small since only the melamine groups can be
effective; in effect, the polymer acts as a chain extender. At high extents
of reaction (where physical properties are achieved), CL, becomes indepen-
dent of the polymer functionality above n = 6. Clearly, the presence of low
functionality material can have a large influence on the cure performance of
a high-solids coating. At low extents of reaction (i.e. the pre-gel region), C,,
I hETHOXY / HYDROXY
Fig. 7. Bake time necessary to achieve C,l = 1 .O x 1 Oe3 mol g-l plotted uersus the methoxy
to hydroxy ratio. The catalyst level and bake temperature were held constant. This figure
is reprinted with permission from ref. 41.

remains a strong function of the polymer functionality above n = 20. This


implies that the chemorheological behavior of high-solids coatings will be
vastly different from low-solids coatings. This, in fact, has been observed
[541.
There are several possible applications of network models to coatings.
The model can be used to screen formulations to determine whether or not
acceptable crosslink density can be achieved. This will not eliminate the
necessity of physical measurements; however, it may allow the evaluation of
coatings over a narrower range of variables by eliminating compositions
which cannot achieve an acceptable cure. The model can also be used to
evaluate the kinetics of network formation. The network structure model
has been used in conjunction with the crosslinking kinetic expression to
determine the cure response of a hexamethoxymethylmelamine crosslinked
coating as a function of the ratio of methoxy to hydroxy, holding other
variables such as catalyst level constant [41]. As can be seen from Fig. 7,
minimum cure times (defined by the time required to reach the optimum
value of C,r at a bake temperature of 130 “C) are achieved when the methoxy
to hydroxy level lies within the range 1.5 - 2.5. Other composition variables,
such as the effect of the addition of low functionality reactive diluents, can
also be studied.
Another application of the network model is to develop correlations
between network structure and bake conditions. Bauer and Dickie have
found that there is a range of values of C,, for which an acceptable cure
(for automotive application) is achieved [48]. The low end (0.8 X lop3
mol g-r) is determined by solvent resistance, humidity resistance, etc. Over-
211

100 110 120 130 140 150 160


Bake TemperoturePC)
Fig. 8. Plots of Cel uersus bake temperature as a function of polymer molecular weight.
The number average molecular weights were 1000 (A), 2000 (B), 400 (C) and 10 000
(D). The equivalent weights were adjusted so that a value of Cel of 1.0 X 10e3 mol g-l
was achieved at an extent of reaction of 85% (corresponding to 17 min at 130 “C).
Reprinted with permission from ref. 48.

bake is determined by intercoat adhesion failure. This occurs when a second


coat is sprayed over an already baked coat as in a tutone or repair. If the
first coat is over-baked, the second coat will not adhere well. It was found
that a value of Cel greater than 1.25 X 10P3 mol g-l is sufficient to cause
intercoat adhesion failure in the coatings studied. The value was indepen-
dent of the coating composition variables studied including melamine
composition. The onset of intercoat adhesion failure was found to be
abrupt - occurring over a range of about 0.03 X 10e3 mol g-l crosslink
density.
A range of crosslink density can be used to define a cure window as
in Fig. 8. The coatings in Fig. 8 differed in the molecular weight of the
copolymer. The equivalent weight and catalyst level were adjusted so that all
coatings achieved a value of C1 of 1.0 X lop3 mol g-’ after a 17 min bake
at 130 “C. Lowering the molecular weight of the copolymer is seen to reduce
the cure window of the coating. This is a result of the decreased functional-
ity of the low molecular weight polymers. As shown in Fig. 6, the lower the
functionality the steeper the dependence of C,i on the extent of reaction.
Thus, everything else being equal, low molecular weight polymers yield
coatings whose state of cure is more sensitive to small changes in the extent
of reaction. Using polymeric melamines as opposed to monomeric melamines
increases the cure window since the average melamine functionality is in-
creased. Ultimately, increasing the molecular weight of the melamine will
reduce the solids level of the coating. However, Santer has found that some
degree of polymerization can be tolerated without loss of solids [42].
Another way to widen the cure window is to adjust the composition so that
optimum cure is achieved at a higher extent of conversion. In the case of
3
2
9 0 I I I 1 I

5 100 110 120 130 140 150 160

d BAKE TEMPERATURE(‘C)

Fig. 9. Plots of C’,l versus bake temperature for different methoxy to hydroxy ratios.
The catalyst level was held constant and the polymer equivalent weight adjusted to
achieve Cd = 1.0 X lop3 mol g-l after 17 min at 130 “C. The calculation ignores methoxy
self-condensation (reaction 4 of Table 4). Reprinted with permission from ref. 41.

Fig. 8, the degree of hydroxy conversion was 85% at 130 “C and the ratio
of methoxy to hydroxy was 1.5. Increasing the hydroxy equivalent weight
and increasing the catalyst level (by 40%) so that cure is achieved at 90%
conversion would yield a coating with a 25% larger cure window. Of course,
increasing the acid level in the coating can lead to shelf stability problems.
Cure windows can also, in principle, be increased by increasing the ratio of
methoxy to hydroxy (see Fig. 9). In practice this advantage may be over-
stated since the model ignores methoxy self-condensation (reaction 4 of
Table 4). This reaction is more likely at high methoxy levels and high bake
temperatures leading to higher values of crosslink density at high ratio.
Finally, it was found that the cure windows of high-solids coatings are more
sensitive to composition variation. For example, a +20% variation in molec-
ular weight of the copolymer has almost no effect on the cure window of a
low-solids coating, but the same variation is found to reduce the cure win-
dow of a high-solids coating by as much as 25% [48]. In general, it is found
that cure response in high-solids coatings tends to be more sensitive to batch-
to-batch variations than it is in low-solids coatings. This implies that quality
control will be particularly important in high-solids coatings.
In the above discussion, there was no mention of how wide a cure
window is required for a given application. It was simply assumed that a
wider cure window was better. The crosslinking kinetic model and the net-
work structure model have been used by Bauer and Dickie to determine
the relationship between process variables (heat transfer rates and bake
213

schedules), coatings variables (cure window) and cure response [55]. Again
the application was specific to automotive painting, though the method-
‘ology is quite general. First, air and car body temperatures were used to
derive heat transfer rates for different parts of the car body. The heat trans-
fer rates were used to determine the response of the car body to arbitrary
oven bake schedules. The cure response of a variety of high- and low-solids
paints was determined both for experimentally measured car body temper-
ature profiles and for profiles generated from ‘optimized’ ovens where the
air temperature was adjusted to give optimum cure. The coatings were
characterized in terms of their cure window (in “C) for a 17 mm bake. The
high-solids paints were chosen to have cure windows less than that of the
conventional low-solids paint studied (50 “C). From the experimental car
body profiles, it was found that the low-solids paint generally achieved an
acceptable cure, i.e. there were very few over and underbakes and these were
not large in magnitude. However, as the cure window was narrowed the
number and magnitude of the under and overbakes increased. Thus, coatings
with narrow cure windows will require some change in process. From the
studies of the optimized ovens, it could be shown that the two process
changes which would accommodate paints with narrow cure windows are (1)
increasing the bake time and (2) increasing the minimum heat transfer rate.
Heat transfer rates are complex functions of air flow in the bake oven and
car body design. It may be difficult in practice to maintain a sufficiently
large enough heat transfer rate to compensate for the likely decrease in cure
window with the use of high-solids coatings. If this is in fact the case, in-
creasing the bake time may be the only viable option for achieving accept-
able cure response.

5 Crosslink degradation

5.1 Hydrolysis
The crosslinking reactions listed in Table 4 are, in general, reversible.
For example, soaking a crosslinked film in acidified methanol would even-
tually break all of the polymer-melamine crosslinks regenerating a methyl-
ated melamine and the hydroxy functional polymer. Melamine linkages are
also susceptable to attack by water. Berge et al. have studied the hydrolysis
of melamine linkages in solution and found that the reaction of alkoxy
groups on melamine with water proceeds by the same mechanism as reaction
with hydroxy groups [40]. Partially-alkylated melamines react via general
acid catalysis while fully-alkylated melamines react via specific acid catalysis.
Although these hydrolysis reactions have been known for many years, the
rate of hydrolysis in crosslinked films exposed to water has only recently
been measured. Bauer has measured the hydrolysis rates of coatings with
both fully- and partially-alkylated melamines exposed to condensing humid-
ity as a function of temperature, acid level and other coating variables [21].
The results were completely consistent with the solution measurements of
214

Berge et al. [40]. Under mild acid conditions, partially-alkylated melamines


hydrolyzed much faster than fully-alkylated melamines, while in the pres-
ence of strong acid fully-alkylated melamines hydrolyzed faster. The activa-
tion energy for hydrolysis was found to be 21 kcal mol-‘, which is approx-
imately the sum of the activation energy for the crosslinking reaction plus
the activation energy for the vaporization of water. The rate was also in-
versely proportional to crosslink density and was also a function of such
composition variables as polymer glass transition temperature.
Hydrolysis of polymer-melamine bonds or residual alkoxy groups
results in the formation of melamine methylol. The reverse reaction of
the methyl01 group with hydroxy to re-form the polymer-melamine cross-
link is apparently slow. Methyl01 groups either deformylate to amine groups
or condense with themselves or melamine amine groups to form NCH,N
linkages. Formation of these linkages on exposure to condensing humidity
has been inferred from IR spectral changes for partially-alkylated melamines
[21]. Such changes were not observed for fully-alkylated melamines pos-
sibly because the rate of hydrolysis was slow. The net result of exposure
of a partially-alkylated melamine crosslinked film is the scission of polymer-
melamine crosslinks and the formation of melamine-melamine crosslinks.
Network structure calculations indicate that this leads to only a small
change in the net crosslink density. However, magic angle NMR spectral
studies revealed that there is a change in the mobility of the melamine
triazine ring in the coating [ 22, 561. Melamine-melamine methylene linkages
are much more rigid than polymer-melamine ether linkages. Since the
number of crosslinks is roughly the same, the melamine triazine ring be-
comes much more rigid as hydrolysis proceeds.

5.2 Photodegradation
Despite the significant hydrolysis chemistry that occurs in coatings
exposed to condensing humidity, there is usually little perceived change in
the physical appearance of the coating. Coatings that are well cured rarely
exhibit humidity problems such as blushing. This may be due to the fact
that scission of polymer-melamine crosslinks is compensated by the forma-
tion of melamine-melamine crosslinks. Normally the durability of melamine
crosslinked coatings is determined by their resistance to ultraviolet (UV)
light [ 57 - 601. The mechanism of UV light degradation is generally thought
to be by photoinduced free-radical attack on the polymer [61]. Specific
attack on the polymer-melamine crosslinks has only recently been con-
sidered. It should be noted, however, that humidity does play a role in the
degradation of melamine crosslinked coatings. Coatings are found to weather
more rapidly in the humid atmosphere of Florida than in the dry climate of
Arizona even though the intensity of UV light is higher in Arizona [ 591.
Gerlock et al. have reported that the hydrolysis reactions observed
during condensing humidity were found to proceed at an accelerated rate in
the presence of UV light [62]. This was subsequently studied in detail by
Bauer and Briggs [63] and by English and Spinelli [ 64,651, and later by
215

Bauer et al. [ 541 and Gerlock et a2. [ 661. English and Spinelli measured the
IR spectrum of a fully-alkylated melamine crosslinked coating exposed for
two years in Florida [64,65]. They found evidence for methoxy hydrolysis
but no evidence for polymer-melamine crosslink hydrolysis nor melamine-
melamine condensation. They proposed that the observed acceleration in
hydrolysis of methoxy groups was due to the formation of carboxylic acid
functionality as a result of the photo-oxidation of the coating. Bauer and
Briggs measured the IR spectra of both partially- and fully-alkylated
melamine crosslinked acrylics as a function of UV light intensity, time and
humidity of exposure [63]. They found that scission of both the methoxy
and polymer-melamine crosslinks occurred at an accelerated rate in the
presence of UV light. Some scission occurred in the absence of water and
was suggested to be due to free-radical attack on the crosslink. The rate of
scission increased with increasing humidity and the rate was larger than
could be explained by simple dark hydrolysis. Melamine-melamine crosslink
formation was only observed for the partially-alkylated melamine cross-
linked coating under humid conditions. This was later observed by Bauer
et al. for fully-alkylated melamine crosslinked coatings but only after exten-
sive exposure [ 541. The rate of methoxy loss of an aqueous solution of hexa-
methoxymethylmelamine was also found to be accelerated in the presence
of UV light, though not by as much as in the coatings. On the basis of these
results, Bauer and Briggs proposed that the acceleration was due to melamine
excited-state chemistry. Melamine has a weak absorbance at 300 nm and the
excited state is more easily protonated than the ground state [63], and thus
the excited state could be expected to hydrolyze faster than the ground
state. This explanation did not, however, explain the observation that the
rate of methoxy loss decreases by a factor of two when a hindered amine
light stabilizer is added, or the observation that the rate is largest in coatings
with high rates of photo-oxidation. Recently, Gerlock et aE. have measured
the emission rates of formaldehyde and methanol from coatings with and
without nitroxide free-radical scavengers [ 663. These data conclusively
demonstrate that the mechanism of ‘photoenhanced hydrolysis’ is free
radical in nature. They also showed that not all of the formaldehyde pro-
duced on self-condensation escapes from the coating. Rather, a substantial
amount is retained and is able to participate in the free-radical oxidation.
Formaldehyde is a peracid precursor. Formaldehyde and performic acid have
been suggested to play a key role in the photodegradation of melamine cross-
linked coatings, and may also affect the chemistry of stabilization by hin-
dered amines.
Although changes in crosslink structure may not have much effect on
the gloss of pigmented coatings, they may play a key role in the durability of
clear coats. Increased or excessive crosslinking has been implicated in crack-
ing of clear coats [67 1. The formation of rigid melamine-melamine cross-
links may result in premature failure of melamine crosslinked clear coats. It
is generally observed that clear coats based on fully-alkylated melamines are
more durable than ones based on partially-alkylated melamines. This may be
216

due to the fact that self-condensation in fully-alkylated melamines does


not occur until extensive exposure. It should be noted, however, that it
does eventually occur and when it does the increase in rigidity is abrupt with
very little warning as to when the change will occur. Thus, the prediction
of service life in such systems may be difficult.

6 Summary

The characterization, crosslinking chemistry, network formation and


degradation of melamine crosslinked high-solids coatings have been dis-
cussed. At this point it might be beneficial to speculate on future studies in
these areas. It has already been mentioned that modern NMR spectroscopic
techniques should play an increasingly important role in the characterization
of melamine resins and of melamine crosslinked coatings. In crosslinking
chemistry, most of the detailed chemical studies have used fully-alkylated
melamines (e.g. hexamethoxymethylmelamine). Accurate data for the
kinetics of well-characterized partially-alkylated melamine crosslinking are
required. The use of kinetic models to study crosslinking in the presence of
blocked catalysts could lead to more shelf-stable, lower curing coatings. The
use of network structure models as both formulation and process aids should
become more widespread. In the area of degradation chemistry, the role of
formaldehyde in the degradation and stabilization of these coatings is an area
ripe for study. The development of improved photostabilizers based on these
studies should also be possible.

References

1 J. 0. Santer, Prog. Org. Coat., 12 (1984) 309.


2 G. Christensen,Prog. Org. Coat., 5 (1977) 255.
3 G. Christensen, Prog. Org. Coat., 8 (1980) 211.
4 L. W. Hill and Z. W. Wicks, Jr., Prog. Org. Coat., 10 (1982) 55.
5 W. J. Blank, J. Coat. Technol., 54 (1982) 26.
6 M. Gordon, A. Halliwell and T. Wilson, J. Appl. Polym. Sic., 10 (1966) 1153.
7 D. R. Bauer and R. A. Dickie, J. Polym. Sci., Polym. Phys. Ed., 18 (1980) 1997.
8 D. G. Anderson, D. A. Netzel and D. J. Tessari, J. Appl. Polym. Sci., 14 (1970) 3021.
9 M. Chiavarini, N. de1 Fanti and R. Bigatto, Angew. Makromol. Chem., 56 (1976) 15.
10 I. H. Anderson, M. Cawley and W. Stedman, Br. Polym. J., 1 (1969) 24.
11 B. Feuer and A. Gourdenne, XZZZth FATZPEC Congress, Congress book, (1976) pp.
210 - 214.
12 A. J. J. de Breet, W. Dankelman, W. G. B. Huymans and J. de Witt, Angew. Makro-
mol. Chem., 62 (1977) 7.
13 J. R. Ebdon and P. E. Heaton,Polymer, 18 (1977) 971.
14 M. Dawbarn, J. R. Ebdon, S. J. Hewitt, J. E. B. Hunt, I. E. Williams and A. R.
Westwood, Polymer, 19 (1978) 1309.
15 B. Tomita and S. Hatono, J. Polym. Sci., Polym. Chem. Ed., 16 (1978) 2509.
16 G. Zigeuner, Monatsh. Chem., 82 (1951) 175.
17 G. Zigeuner, Monatsh. Chem., 83 (1952) 1099.
217

18 A. Takahashi and I. Yamazaki, Kobunshi Kagaku, 15 (1958) 327.


19 D. Braun and F. Bayersdorf, Angew. Makromol. Chem., 89 (1980) 183.
20 J. N. Koral and J. C. Petropoulos, J. Paint Technol., 38 (1966) 600.
21 D. R. Bauer,J. Appl. Polym. Sci., 27 (1982) 3651.
22 D. R. Bauer, R. A. Dickie and J. L. Koenig, J. Polym. Sci., Polym. Phys. Ed., 22
(1984) 2009.
23 J. R. Havens and J. L. Koenig, Appl. Spectrosc., 37 (1983) 226.
24 W. P. Aue, E. Bartholdi and R. R. Ernst, J. Chem. Phys., 64 (1976) 2229.
25 L.-S. Chuang, B. L. Hawkins, G. E. Maciel and G. E. Myers, Macromolecules, 18
(1985) 1482.
26 W. J. Blank and W. L. Hensley, J. Paint Technol., 46 (1974) 46.
27 W. J. Blank,& Coat. Technol., 51 (1979) 61.
28 W. J. Blank, J. N. Koral and J. C. Petropoulos, J. Paint Technol., 42 (1970) 609.
29 D. R. Bauer and R. A. Dickie, J. Polym. Sci., Polym. Phys. Ed., 18 (1980) 2015.
30 R. Saxon and J. H. Daniel, Jr., J. Appl. Polym. Sci., 8 (1964) 325.
31 R. Saxon and F. C. Lestienne, J. Appl. Polym. Sci., 8 (1964) 475.
32 K. Holmberg, J. Oil Colour Chem. Assoc., 61 (1978) 359.
33 J. W. Lorimer, J. Paint Technol., 40 (1968) 586.
34 R. Seidler and H. G. Stolzenbach, Furbe + Lacke, 81 (1975) 281.
35 J. Dorffel and U. Biethan, Farbe + Lacke, 82 (1976) 1017.
36 K. H. Hornung and U. Biethan, Farbe + Lacke, 76 (1970) 461.
37 U. Biethan, K. H. Hornung and G. Peitscher, Chem.-Ztg., 96 (1972) 208.
38 M. G. Lazzara, J. Coat. Technol., 56 (1984) 19.
39 J. K. Dixon, N. T. Woodberry and G. W. Costa, J. Am. Chem. Sot., 69 (1947) 599.
40 A. Berge, B. Kvaeven and J. Ugelstad, Eur. Polym. J., 6 (1970) 981.
41 D. R. Bauer and G. F. Budde, J. Appl. Polym. Sci., 28 (1983) 253.
42 J. 0. Santer and G. J. Anderson, J. Coot. Technol., 52 (1980) 33.
43 M. S. Chattha and D. R. Bauer, Ind. Eng. Chem., Prod. Res. Dev., 22 (1983) 440.
44 E. W. Meijer, J. Polym. Sci., Polym. Chem. Ed., submitted for publication.
45 P. J. Fiery, Principles of Polymer Chemistry, Cornell University Press, Ithaca, NY,
1953, pp. 347 - 392.
46 M. F. Kooistra, J. Oil Colour Chem. Assoc., 62 (1979) 432.
47 D. R. Bauer and G. F. Budde, Znd. Eng. Chem., Prod. Res. Dev., 20 (1981) 674.
48 D. R. Bauer and R. A. Dickie, J. Coat, Technol., 54 (1982) 57.
49 D. R. Miller and C. W. Macosko, Macromolecules, 9 (1976) 206.
50 D. R. Miller and C. W. Macosko, Macromolecules, 13 (1980) 1063.
51 J. L. Stanford, R. F. T. Stepto and R. H. Still, in S. S. Labana and R. A. Dickie
(eds.), Characterization of Highly Crosslinked Polymers, ACS Symp. Ser. NO. 243,
Am. Chem. Sot., Washington, DC, 1984, p. 1.
52 K. Takeuchi, in G. D. Parfitt and A. V. Patsis (eds.), Organic Coatings Science and
Technology, Vol. 6, Dekker, New York, 1984, p. 151.
53 L. W. Hill and K. Kozlewski, 12th Znt. Conf. Org. Coat. Sci. Technol., Proc., Athens,
Greece, July 7 - 11, 1986.
54 D. R. Bauer and L. M. Briggs, J. Coat. Technol., 56 (1984) 87.
55 D. R. Bauer and R. A. Dickie, in T. Provder (ed.), Computers in Applied Polymer
Science, ACS Symp. Ser., in the press.
56 D. R. Bauer, R. A. Dickie and J. L. Koenig, Ind. Eng. Chem., Prod. Res. Dev., 24
(1985) 121.
57 M. L. Ellinger, J. Coat. Technol., 49 (1977) 44.
58 J. L. Scott, J. Coat. Technol., 49 (1977) 27.
59 R. A. Kinmonth, Jr. and J. E. Norton, J. Coat. Technol., 49 (1977) 37.
60 G. W. Grossman, J. Coat. Technol., 49 (1977) 45.
61 B. Ranby and J. F. Rabek, Photodegradation, Photooxidation, and Photostabilization
in Polymers, Wiley, New York, 1975.
62 J. L. Gerlock, H. van Oene and D. R. Bauer, Eur. Polym. J., 19 (1983) 11.
218

63 D. R. Bauer and L. M. Briggs, in S. S. Labana and R. A. Dickie (eds.), Characteriza-


tion of Highly Crosslinked Polymers, ACS Symp. Ser. No 243, Am. Chem. Sot.,
Washington, DC, 1984.
64 A. D. English, D. B. Chase and H. J. Spinelli, Macromolecules, 16 (1983) 1422.
65 A. D. English and H. J. Spinelli, J. Coat. Technot., 56 (1984) 43.
66 J. L. Gerlock, M. J. Dean, T. Korniski and D. R. Bauer, in preparation.
67 M. Both and W. Uerdingen, Org. Coat. Plast. Chem., 43 (1980) 59.

You might also like