Effects of Chemical Structure of Hardener on Curing

Evolution and on the Dynamic Mechanical

Behavior of Epoxy Resins

F. F. DE NOCRARO, P. CUERRERO, M. A. CORCUERA, and 1. MONDRAGON*

Escuela Universitaria de lngenieria Tecnica Industrial, Departarnento de lngenieria Quirnica Y Medio Arnbiente,
Universidad del Pais Vasco/Euskal Herriko Unibertsitatea, Avda. Felipe IV, 16 2001 1 San Sebastian, Spain

SYNOPSIS

A diglycidyl ether of bisphenol-A-type difunctional epoxy resin was cured with different
amine-type curing agents at stoichiometric ratios. The crosslink process was followed by
viscosimetry and differential scanning calorimetry. The gelation time and the apparent
activation energy were found to be strongly dependent on the structure of the hardener.
The heat of reaction did not vary significantly when the hardener was changed. An inter-
pretation based on structural aspects such as amine reactivity, steric hindrance, and chain
rigidity is proposed for the variations corresponding to the curing process. Master viscosity
curves have been built up for all mixtures. The effect of the hardener on the glass transition
temperatures of the different mixtures has been analyzed taking into account the crosslink
density, measured by the rubber modulus obtained by dynamic mechanical studies, and
the chemical structure of the hardener chains. 0 1995 John Wiley & Sons, Inc.

INTRODUCTION groups of the constitutive components, as well as

In the last decade numerous investigations have the cure extension, can also influence the cure evo-
been carried out on the cure behavior and the re- lution and also the final properties of epoxies.
The influence of factors mentioned above has
lationship between structure and ultimate properties
been reported in various publication^,^-^ but the re-
of epoxy resins.'V2 This interest is a consequence of
lationship between the chemistry of the epoxide and
the excellent thermal and chemical resistance and
amine monomers during the different stages of cur-
the superior electrical and mechanical behavior of
ing, and the properties of crosslinked networks, still
these resins, and therefore epoxy resins are used in
need to be completely understood. An interesting
a wide range of applications in the adhesive, elec-
question is the way in which the chemical structure
tronic, and aerospace industry. Among the several
and existing interactions among groups can modify
curing agents employed to formulate these materials,
the amines are undoubtedly the most extensively the chemorheology and the structure of the formed
used. When the same epoxide monomer is selected, network correlating the variations experimentally
obtained from these physico-chemical parameters.
the chemical structure of the amine is responsible
for the kinetic differences upon curing as well as for The main objective of this study is to elucidate
the effects of the chemical structure of the curing
the ultimate variation on the properties of these
agent on the chemorheology of curing and the phys-
materials. Nevertheless, other factors such as steric
restrictions to the epoxy-amine addition reactions ical properties of epoxy resins cured with amines
or physical interactions among different functional with different chain backbones. The variations
throughout curing and the thermal behavior of cured
* T o whom correspondence should be addressed.
systems is related to the structure of the curing agent
Journal of Applied Polymer Science, Vol. 56,177-192 (1995)
employed. Aliphatic as well as cycloaliphatic and
0 1995 John Wiley & Sons, Inc. CCC 0021-8995/95/020177-16 aromatic amines have been used.
177
178 DE NOGRARO ET AL.

EXPERIMENTAL Various aliphatic, cycloaliphatic, and aromatic

amines were used as curing agents. The chemical
Materials and Mixing structures of the amines and epoxide monomer are
shown in Figure 1. The reactants were used as re-
Diglycidyl ether of bisphenol A (DGEBA), Rutapox ceived in stoichiometric amine-epoxy equivalent
VE-3966, an epoxide monomer kindly provided by ratio. DETA, IPDA, mPDA, and DDM were Fluka
Bakelite, with an equivalent weight of epoxy groups quality. 3DCM, Laromin C260, was purchased from
equal to 184.5 g/equiv determined by chemical ti- BASF. Jeffamine D230 and T403 were gently sup-
tration, was used in this study. plied by Texaco.

Product Formula

Dyglicidyl ether of bisphenol A

DGEBA
R U ~ ~ VE-3966
OX
(Bakelite)
CHT CH- CH
'O' { 0r
b CH3
OH
0- CH2- C
I H-CH2l:o

CH3
0- CH2- CH
0
/ -\CH2

n=0.10

Diethylene triamine
DETA

Poly(oxypropy1ene) diamine
Jeffamine
D230
(Texaw)

Poly(oxypropy1ene) triamine CH2

I
f 0 - CH2- CH (CH3) ] NH2
X
Jeffamine H 3 C - H E - C -I C H 2 f O - CH~-CH(CH~)~NHZ X+y+z a 5.3

T-403 I Y
(Texaw) CH2 { 0 - CH2- CH (CH3) 2
NH2

Isophomne diamine
IPDA
(Mew
CH3

3,3'-Dimethyl-4,4'diaminodicyclohexylmethane
3DCM
H2N
(B=4

~~~~
m-Phenylene diamine
mPDA
rnuka)

4,4'-Diamino diphenyl methane

HT 972
a NH2

DDM
(Ciba-Geigy)

Figure 1 Characteristics of the different materials used.

 STRUCTURE OF HARDENER ON CURING EVOLUTION 179

Before mixing, the DGEBA epoxy monomer was heating rate of 3"C/min from 20 to 250°C. The di-
preheated at 60°C to melt any crystals present. mensions of specimens were 60 X 12 X 6 mm. Glass
Stoichiometric mixtures of the epoxy prepolymer transition temperatures were identified by the tem-
and DETA, IPDA, D230, T403, or 3DCM were cast perature corresponding to the maximum value of
at room temperature, and at 80°C for m-PDA or the loss factor, tan 6.
DDM containing mixtures, for 5 min, and subse-
quently degassing was made in a vacuum oven for
5 min. These times were selected in order to have RESULTS
the minimum possible of cure reactions between the
materials mixed. There are several ways to monitor the curing process
of thermosetting resins. In this study one of the most
widespread methods of investigating the initial
Steady Shearing Flow Behavior
stages of curing has been chosen to compare the
Viscosity measurements were carried out in a Ro- curing process of epoxy resins cured with several
tovisco, RV 20, viscometer at several isothermal amines. Viscosity variations during the curing pro-
temperatures by using a coaxial-cylinder device. The cess have been measured by a steady shearing tech-
weight of the samples was 60 g approximately. Mix- nique. As it is well k n ~ w n ,curing
~ . ~ of a thermoset
tures were poured out into a cell previously pre- resin supposes an increase of molecular weight at
heated to the desired cure temperature, from 35 to the earlier stages of the curing process as the initial
90°C. Viscosity variations were recorded a t a shear thermoplastic epoxy material grow and branch.
rate of 0.77 s-'. The gelation time was defined as Thereafter, the attainment of a gel formation
the time to reach a viscosity of 1000 Pa s. A larger threshold corresponds to the formation in the curing
bob was employed to collect accurate values of the material of a macroscopic crosslinked structure what
unreacted mixtures. constrains the mobility of chains, so supposing an
increasing loss of fluidity until viscosity rises toward
an infinite value.
Differential Scanning Calorimetry Measurements
As it has been shown in the literature survey?
The enthalpy of reaction, AH, and the values of the the viscosity of these unreacted stoichiometric mix-
initial, maximum, and final temperatures of the tures increased as a function of the hardener struc-
exothermic peak, TI,Tp,and T2,were obtained from ture in the following way: IPDA < D230 = T403
dynamic measurements at a heating rate of 10°C/ < DETA < 3DCM < m-PDA < DDM. The sur-
min in a Perkin-Elmer, DSC-7. preasingly low values for mixtures containing IPDA
or poly(propy1ene oxide) (PPO) amines is a conse-
quence of the pendant methyl groups that hinder
Dynamic Mechanical Thermal Analysis
the intermolecular interactions among different
Dynamic mechanical analysis was used to determine chains, and so the strength to flow of these mixtures
the glass transition temperature of the stoichio- decreases.
metric formulations of the epoxy mixtures cured. The evolution of viscosity on curing for the sev-
Cure schedules for each mixture, shown in Table I, eral epoxy mixtures is shown in Figure 2(a)-(g). The
have been selected by taking a time higher than the bob employed to determine the gelation process did
corresponding gelation time at the earlier stage of not permit accurate values of the viscosity data prior
the cure schedule, and a postcure temperature which to gelation, and thus these viscosity data are shown
permits attaining the maximum glass transition in Figure 2(a)-(g) as a zero value despite their in-
temperature. Studies were carried out in a Metravib crease on cure time as branched molecules formed,
viscoanalyzer using a three-point bending device. but nevertheless viscosity remained almost constant
For all mixtures scans were made at 10 Hz and a for all mixtures at the earlier stages of curing because

Table I Cure Schedules Used for the Different Stoichiometric Epoxy Mixtures

Curing Agent DETA IPDA 3DCM D230 T403 m-PDA DDM

Cure schedule 50 (60) 50 (60) 80 (120) 60 (180) 60 (180) 80 (120) 80 (150)

T("C) [t(min)] 120 (120) 90 (120) 125 (120) 180 (120) 180 (120) 125 (120) 140 (90)
190 (840) 190 (840) 190 (360) 190 (360) 200 (120)
180 DE NOGRARO ET AL.

(a)
1200 - 50°C
0 45°C
H A 40°C
O
lo00 -
m
A
0
A
800 -
h A
2 0
A
3 H A
=6
lo o - A
o m
A
400- 0 .
A
A
0
200 - m
A A
O H

- 0
0
A
I
.
-
I
A .
-
A
-
,A

1200 - 0

0 50°C
A A A 45'C 0

- 0 A 40°C
lo00 0 35°C
A
0
O m A

-2 800 - 0

o m
A
A 0

a, - A
A 0
1600 0
A . 0

400 -
0
A
A 0
A
0 A
200 - A '
0

0
A A 0
0
A A
-- -
0 .
-,=.I * - - & M A I o 1
0
0 50 100 150 200
t (min)
Figure 2 Viscosity profiles at the isothermal temperatures for DGEBA cured with several
curing agents. (a) DETA, (b) IPDA, ( c ) SDCM, (d) D230, (e) T403, (f) mPDA, (g) DDM.

the average longest linear path dominated the rheo- isothermal cure temperatures. In every case an in-
logical behavior. At the onset of gelation there was crease in the cure temperature originated an increase
an increase in viscosity which rose rapidly toward in the rate of cure reaction. The viscosity increase
an infinite value as gelation occurred. In this work, around the gelation threshold was sharper at higher
gelation, measured by steady viscosimetry, has been cure temperatures, therefore the gelation time re-
taken as the time at which viscosity reached a 1000- corded was more accurate for every mixtures when
Pa s value. Each mixture was analyzed a t different they were cured to relatively high cure temperatures.
 STRUCTURE OF HARDENER ON CURING EVOLUTION 181

A
0 A
A

800 - m A
n A A
2 o
a, 8 A
A
t 6oo- 0
A
A
A A
A
400-
A t
O m A

t (min)

1200 - A
0 70°C A
8 60°C
A 5OoC
A 40°C A
'Oo0- A
0
A

800 -
0 8 A
n A
9
(d

a, A
=t 6oo- 0 8 A
A
0 8
400- A
A
0 m
200 - 8
A
A A
0 A
m A
O
0
W ' - - I AAA .
.AA
I .
0 100 200 300 400 500
t (min)
Figure 2 (Continued from the previous page)

For each temperature, gelation took place at differ- increase in the kinetic constant of the nucleophilic
ent times as a function of the chemical structure of reaction between the epoxy and the primary amine
the curing agent employed. Gelation times for all groups upon the temperature.
blends at some of the several cure temperatures are As it is well known,1°-15amine-cured epoxy net-
given in Table 11. For all mixtures, an increase in works generally form exclusively from epoxide-
the cure temperature produced a reduction in the amine nucleophilic addition reactions, above all at
gelation time. This variation is a consequence of the low cure temperatures. Although either the primary
 182 DE NOGRARO ET AL.

(e)
1200 8
A
A 50°C A
8 A 40°C
lo00 0 A A
8

A A
800 O D
A A

z
h

A
B 0 .
A A
s.600
A A
0 A
400 A
A
A
A
200 0 A
A A
A

0
a 0
8 A

0 100 200 300 400 500

t (min)

(f )
1200 - 0 A A
8 70°C
D A
A 50°C
lo00 - 0

A
0 8 A
800 A

z
h

A A
8 - 0
1600
rn A
0 A

400 - 0 A
0
A A

0
200 8 A A
0
0 8 A A
A

0 I"
. A
I

0 100 200 300 400

t (min)

Figure 2 (Continued from the preuwus page)

and secondary amine-epoxide reactions occur si- The dependence of the viscosity variation during
multaneously, the faster primary amine-epoxide re- curing a t 50°C with the chemical structure of the
action predominates in the earlier stages of curing curing agent is shown in Figure 3. Gelation times so
which conducts to the gelation threshold. Therefore, obtained for these epoxy mixtures are also shown
the variations on gelation times, along with the rel- in Table 11.
ative viscosity values of the unreacted mixtures, can Let us analyze the different times to gelation for
be employed to compare the reactivities of the pri- these amine-cured epoxy resins. At a constant cure
mary amines for the different curing agents. temperature, it becomes evident that gelation time
 STRUCTURE OF HARDENER ON CURING EVOLUTION 183

1200

lo00

0
A 0
800 A
0
0

0
600 0
0

400

200

0
0 100 200 300 400 500
t (min)

Figure 2 (Continued from the previous page)

is a function of the curing agent employed or, in to the methylene unit is more hindered than that
other words, of the chemical structure of the com- attached to the aliphatic cycle, can lead to mistakes
ponents, which are reacting, as well as of the vis- in the interpretation of results since viscosity in-
cosity of the mixture. Indeed, when comparing ge- creasing a t low cure temperatures could occur as a
lation times of systems based on DETA and IPDA, consequence of vitrification more than gelation (the
a variation from aliphatic to cycloaliphatic frame- conversion of the epoxy group was not high at this
work supposed an important delay in the gelation stage of curing, and therefore the real gelation should
threshold as a consequence of the loss on chain flex- happen at higher cure times). Thus, as stiffness of
ibility introduced by the cycloaliphatic cycle. This the amine chain increased (when mixtures cured
behavior becomes more evident when the rigidity of with cycloaliphatic, IPDA and SDCM, and aromatic
the cycloaliphatic amine was increased, as it was for m-PDA and DDM, amines were compared), gelation
the SDCM. Nevertheless, it is worth noting that for appeared at longer times for the aromatic ones. In
the IPDA containing mixture, the different reactiv- this latter comparison, besides the stiffness of the
ities of the amine groups, because of that attached hardener, in which the side chains or groups have

Table I1 Gelation Times (min) at Different Cure Temperatures

\
T ("C)
Curing Agent
DETA IPDA 3DCM D230 T403 m-PDA DDM

35 53 175 -
40 36 103 435 445
45 26 85 -

50 22 50 108 216 221 386 441

60 - 26 66 85 96 192 250
70 - 36 38 46 108 143
75 28 -
80 - 81
90 35 47
184 DE NOGRARO ET AL.

1200 A
A 0

lo00 [Im A
A
*o

A 0

800 0
h
v) A
(d 0
a, 600
I A
0 8
a A
A A 0
0 * n

400

0
200

0
0 100 200 300 400 500
t (min)
Figure 3 Viscosity profiles for the epoxy mixtures cured at 50°C. (0)DETA, (B) IPDA,
(A)3DCM, (A)D230, (0) T403, (4) mPDA, (El) DDM.

to be considered, it must be taken into account that the gelation behavior has to be analyzed. Neverthe-
in aromatic amines the nucleophilicity lowers with less, undoubtedly other factors have to be considered
respect to that for cycloaliphatic amines. Conse- as well. Indeed, as shown in Figure 3, gelation times
quently, the chain stiffness and the amine reactivity corresponding to the aliphatic PPO-based curing
of the curing agent are the main features whenever agents were conducted in a very different way since

2,7 2,8 2,9 3,O 3,1 3,2 3,3

lOOO/T (OK-'
)

Figure 4 Arrhenius plots for gelation times. Symbols as in Figure 3.

 STRUCTURE OF HARDENER ON CURING EVOLUTION 185

for these amine-cured mixtures gelation occurred at ing the accessibility of other reactive groups like
very long times. Therefore, the variation of cure be- epoxy with respect to that existing in mixtures with
havior on the polyether di- or triamine based for- aliphatic amines unsubstituted. Indeed, the high ge-
mulations must be connected with other structural lation times for mixtures cured with PPO-based
features than chain rigidity. polyamines seem to be basically due to the decreas-
According to Dusek and Matejka," the different ing amine reactivity due to the presence of the ether
reactivities of the terminal primary amino groups group.
of these PPO-based amines differ from that of NH2 On the other hand, the slight longer gelation times
groups of aliphatic amines because of competition of T403 cured mixtures with respect to that for D230
between the epoxy and ether groups of the polymer containing ones, especially at low cure temperatures
backbone in an interaction with proton donors; be- [figs. 2(d) and 2(e)] possibly is due to the higher
sides, the methyl substitution effect in the amine oxygen-active hydrogen atoms ratio in the T403
group reactivity is more negative than in aliphatic triamine, which would increase the hydrogen bond-
amines. ing between oxygen and amine groups in the T403
Therefore, one significant reason for the delay in rather than for the D230 amine.
gelation for mixtures based on PPO amines must The influence of cure temperature on the gelation
arise from the inter- and intramolecular interactions time has been also studied for each epoxy mixture
by means of hydrogen bonding, in addition to those [figs. 2(a)-(g)]. As it is well known, the dependence
existant for other epoxy-amine sy~tems.'~ In fact, of gelation time on the cure temperature can be well
the presence of the ether group, the oxygen atom is fitted with an Arrhenius-type e q ~ a t i o n . ~ .Thus,
'~~'~
a strong H-acceptor, in the PPO-based polyamines the time required to reach a viscosity of 1000 Pa s,
appears to be responsible for the decrease on the t,, can be written:
cure rate by the occurring intramolecular bonding,
but also by intermolecular interactions which slow
down the reactivity of the active hydrogens of the
amine groups, and so their nucleophilicity. There- where to is a constant for each mixture, E , is the
fore, despite the chain flexibility of these PPO-based activation energy, R is the gas constant, and T is
polyamines, the gelation threshold appeared at very the temperature expressed in K.
long times. Indeed, when the curing agent reactivity Indeed, Figure 4 shows a linear dependence of the
must be analyzed, besides the nucleophility of the logarithm of gelation time on the inverse cure tem-
reacting group, factors such as chain rigidity, inter- perature for all epoxy mixtures analyzed. The ap-
and intramolecular interactions, steric hindrance, parent activation energies for gelation determined
and others have to be considered as well. from the slopes of these straight lines are shown in
Morgan et a1.l' have suggested that the methyl Table 111. The values obtained are similar to those
group adjacent to the amine group in PPO-based reported in the literature survey [for instance, Grillet
polyamines would sterically interfere, especially in et al.5 using a Contraves viscosimeter obtained an
the rate of secondary amine reactions, and thus the activation energy of 53 kJ mol-' for a DGEBA (DER
methyl groups should not influence enough the ge- 332)/DDM mixture].
lation time with respect to that for aliphatic amines The introduction of methyl substituents in the
since before gelation reactions with primary amines carbon atom located in the P position with respect
occur preferentially to those with secondary to the reactive groups, 3DCM or IPDA cases, gives
amines.15*16 Nevertheless, these methyl groups con- rise to shielding and steric hindrance of the reactive
tained in the PPO-based polyamines structurally groups. Indeed, as it has been recently shown,6 the
obstruct the nearest-neighbor amines and so the in- introduction of alkyl substituents in carbon atoms
termolecular interactions, therefore decreasing the near to the reactive groups leads to an increase of
viscosity of the mixtures and consequently improv- the activation energy of the chain segments implied,

Table I11 Apparent Activation Energies and geITs'sfor the Epoxy Mixtures

Curing Agent DETA IPDA 3DCM D230 T403 m-PDA DDM

Activation Energy (kJ/mol) 47.7 62.3 50.9 73.4 68.1 54.5 54.1
CeiTZ("C) - 32 47 42 43 40 47
186 DE NOGRARO ET AL.

(a)
A
1200
- 45°C a0
A 40°C I
0
A
H
A
0

800 -
A
m
A
0
t 4
= 600-
m
A
400- m
A
A
200 -
0

:
0.

\
0
z
z
-c
-
A
A
0
50°C
45°C
40°C
35°C
1 0

A
so
mo
A
n
A0
0

0
0

Figure 5 Master curing curves of viscosity for epoxy mixtures cured under isothermal
conditions for the different curing agents used. (a) DETA, (b) IPDA, (c) 3DCM, (d) D230,
(e) T403, (f) mPDA, (g) DDM.

and so the similarity of activation energies for mix- on mobility of the hydrocarbon cycle introduced by
tures cured with alkyl-substituted cycloaliphatic the methyl groups located far away from the reactive
amines or with aromatic amines arises from the sites. The surprising activation energy values for
comparable increase on chain rigidity introduced ei- mixtures with PPO-based amines must be a con-
ther by the alkyl substituents or by the aromatic sequence of both the steric hindrance of the methyl
rings. Moreover, the high activation energy for IPDA group and the intramolecular interactions between
containing mixture is possibly due to the lowering the hydrogens of the amine groups and the ether
 STRUCTURE OF HARDENER ON CURING EVOLUTION 187

(d)
1200

lo00

800
0
z
\

1-600

400

200

0
0,o 0,4 0,8 12 1,6 2 ,o
t/tx
Figure 5 (Continued from the previous page)

groups of the amine backbone. Intramolecular in- ing curves by using a time shift factor, a(t).In order
teractions between hydrogen and nitrogen atoms of to do that, cure times have been divided by an ar-
the amine groups are also possibly responsible of bitrary time, t,., taken as time to which the viscosity
the relatively high activation energy for the of the mixture steadied. As shown in Figure 5, for
DGEBA-DETA mixture. each mixture a single generalized curve described
On the other hand, it was noted that the curing the viscosity evolution during isothermal curing,
curves of viscosity under various isothermal condi- thus indicating that the reaction mechanisms did
tions had similar shapes for the different mixtures not change for the intervals of cure temperatures
analyzed. Attempts have been made to shift the cur- studied. However, at the early stage of the sharp
188 DE NOGRARO ET AL.

Figure 5 (Continued from the previous page)

variation on viscosity some deviation in the curves tures indicated Arrhenius-like relationships from
corresponding to the higher cure temperatures was which the apparent activation energies were calcu-
observed, especially for PPO-based amine mixtures, lated. The values so obtained, though not shown,
possibly as a consequence of competitiveness be- are equal to those shown in Table 111.
tween gelation and vitrification at low cure temper- Furthermore, the temperature at which gelation
atures. and vitrification take place simultaneously, gelTg,has
The linear dependence of the time shift factor been calculated for all mixtures studied. The values
upon the inverse temperature for the different mix- of the slopes on the viscosity versus time curves cor-
 STRUCTURE OF HARDENER ON CURING EVOLUTION 189

1200

800
0
\
2
=600

400

200

Figure 5 (Continued from the previous page)

responding to both regions before and after the sharp overall heats of reaction to epoxy equivalent were
rising on viscosity during curing have been measured quite the same for all epoxy-amine mixtures ana-
and plotted against reciprocal absolute cure tem- lyzed, thus indicating the similarity in the mecha-
peratures. For every mixtures all experimental val- nisms of the reactions for these systems.
ues were well fitted by straight lines, which means On the other hand, the maximum glass transition
that, within the experimental error, the method was temperature of these epoxy mixtures, obtained for
adequate to obtain the gelTgvalues of the studied each mixture with the cure schedules shown in Table
mixtures. Data so obtained are shown in Table 111, I, has been measured by dynamic mechanical studies
and they are similar to those shown in the literature as the temperature corresponding to the maximum
survey.15 Despite the evident dependence on the value of the a relaxation, To.Results are shown in
structure of hardener, at this time we do not have Figure 6 and Table V. Instead the rheological be-
an interpretation of the values obtained. More in- havior, the mixtures containing PPO di- or triamines
vestigations with other hardeners and epoxy mono- presented the lower glass transition temperatures
mers with different molecular weights are necessary
in order to understand this behavior.
Dynamic differential scanning calorimetry (DSC)
scans have also been carried out for all epoxy mix- Table IV Initial, Peak, Final Temperatures,
tures studied. Results are reported in Table IV. TI, and Heats of Reaction Obtained from DSC
Analysis for Stoichiometric Mixtures
T p ,and T2 are the temperatures corresponding to
the initial, the peak, and the final temperatures of Curing
the cure exotherm through the dynamic scan, re- Agent TI("C) T, ("C) T2("C) AH (kJ/equiv)
spectively. The value of Tpis strongly dependent on
the structure of the curing agent. Thus, aromatic DETA 34 98 200 100
amines like rn-PDA or DDM are less reactive than IPDA 40 114 210 104
the aliphatic and cycloaliphatiques. Besides, the 3DCM 50 131 215 106
steric hindrance existing in D230 and T403, as well D230 45 108 176 94
as in the cycloaliphatic amines, IPDA and 3DCM, T403 44 113 222 97
also increases Tp.Consequently it indicates a lower m-DPA 85 159 240 104
DDM 80 168 240 104
reactivity in the corresponding epoxy mixtures. The
190 DE NOGRARO ET AL.

0 50 100 150 200 250

TW)
Figure 6 Variation of loss factor as a function of temperature. Symbols as in Figure 3.
The scale is correct for T403. For the rest of the hardeners, the scale is offset vertically
from that of its immediate neighbors by 1.0.

of all systems analyzed. As is well known, for an to know the crosslink density of the formed network.
epoxy network the maximum Tg is reached when Therefore, in order to known the crosslinking den-
the cure is complete. However, as previously shown sity of the epoxy mixtures, most a ~ t h o r s ~use
~-~an'
for different epoxy mixture^,^'-^^ despite using the approximation of the simple rubber elasticity theory,
optimum cure schedule some epoxy groups still along with the experimental equilibrium rubber
remain unreacted. The storage moduli, E', for +
modulus, E,, taken as elastic modulus at T, 30°C.
these stoichiometric DGEBA-hardener mixtures are Thus, the average molecular weight between cross-
plotted against temperature in Figure 7. E' conducted links can be approximated through the following
in a relatively similar way for all mixtures since at relati~nship~~:
temperatures lower than the TgE' values were high,
showing a sharp decrease in the glass transition re-
gion corresponding to each mixture. Afterward, the
storage modulus reached a near constant value be- where 6 is the front factor, p the density at absolute
cause all materials became like a rubber. temperature T, R the gas constant, and E, the equi-
The average molecular weight between strands librium elastic modulus. Although E, is usually taken
of the different crosslinked networks can be theo- as the tensile modulus (or shear modulus, G,), values
retically calculated by assuming a full conversion of of flexural rubber modulus, shown in Table V, can
epoxy resin in the curing process and afterward its be used in order to compare the physical rigidity of
value can be compared with the experimental one the several mixtures studied.

Table V Glass Transition Temperatures and Rubber Moduli for the Different Epoxy Mixtures

Curing Agent DETA IPDA 3DCM D230 T403 m-PDA DDM

Tg("0 138 161 160 90 91 169 170

E, (N/m2) 6.0 5.0 2.3 2.8 5.0 6.0 6.0
 STRUCTURE OF HARDENER ON CURING EVOLUTION 191

9-
........... *\
8 -

7 -
m
2 6-
X
h
N 5 -
&
4 -
2,
iil 3 -
0 0 0 0 0 0 0 ~A
0 .
2 - i
OoO A A
1 - 0"
,
80
0- aabcao 000

0 50 100 150 200 250

T(OC)
Figure 7 Storage modulus variation upon temperature. Symbols as in Figure 3. The scale
is correct for T403. For the rest of the hardeners, the scale is offset vertically from that of
its immediate neighbors by 1.0 (for mPDA the offset is 1.5).

Although the crosslink density of the cured net- of the steric restrictions to motion between chains
work is the main factor to determine the glass tran- led by the bulky methyl groups in the PPO poly-
sition behavior for an epoxy resin crosslinked with amines based mixtures have been seen to increase
a given hardener and cured at different conditions,6." the Tgin other epoxy blends.34The T430-cured sys-
other factors such as chemical structure of the har- tem showed a slightly higher Tgthan that for the
dener backbone must be considered to compare dif- D230 containing mixture; this change is certainly
ferent amine-cured epoxy mixtures. With the ex- related to the more complicated network formed for
ception of PPO-amine based mixtures, while rubber the T430-cured mixture." On the other hand, the
modulus values did not show significant changes for glass transition temperatures for mixtures cured
the mixtures cured with the different hardeners, the with cyclic amines increased as rigidity of amine
glass transition temperatures of the fully cured mix- cyclic chains did, i.e., Tg's were higher for the mix-
tures showed evident variations from one mixture tures cured with aromatic amines than those for cy-
to another. Therefore, the variation on Tg'sbetween cloaliphatic amines containing mixtures, that de-
the different mixtures did not have a clear relation spite the high Tgof the latter mixtures because of
with their crosslink density or rubber modulus, as the steric restrictions to chain motions of the pen-
shown in Table V. Instead, the chemical structure dant methyl groups.
of the amine explained the differences in the glass
transitions of these mixtures in spite of other factors
such as packing density3' or that variations on the CONCLUSIONS
free v o l ~ m e ~could
~ , ~vary
~ - ~slightly
~ the reported The cure behavior of epoxy mixtures cured with
values. Thus, the networks constructed with ali- hardeners with different backbone stiffness and
phatic amines showed the lower glass transition pendant methyl groups has been investigated by
temperatures because of the flexibility of their rheological techniques. Gelation times and apparent
chains, higher for the PPO di- and triamines than activation energies of the several mixtures have been
for DETA-cured mixtures since the higher aliphatic related to the different characteristics of the curing
segment-amine nitrogen ratio in the PPO di- and agent such as amine nucleophylity, backbone stiff-
triamines with respect to that for DETA, rigidized ness, steric hindrance, and inter and intramolecular
by the central amine group after its reaction, in spite interactions between different groups.
192 DE NOGRARO ET AL.

DSC dynamic analysis has shown the relation- 11. R. J. Morgan, F. M. Kong, and C. M. Walkup, Polymer,
ships between the reactivity in the mixtures as a 2 5 , 3 7 5 (1984).
function of the chain stiffness of the curing agent 12. D. W. Schiering, J. E. Katon, L. T. Drzal, andV. B.
and the pendant groups of the amines. Gupta, J. Appl. Polym. Sci., 3 4 , 2367 (1987).
13. A. Sabra, T. M. Lam, J. P. Pascault, M. F. Grenier-
The glass transition temperatures of cured epoxy
Loustalot, and P. Grenier, Polymer, 2 8 , 1030 (1987).
mixtures, dependent on the crosslink density for a 14. C. C. Riccardi and R. J. J. Williams, in Crosslinked
given formulation, are a function of the chemical Epoxies, W. de Gruyter, Hawthorne, NY, 1987.
structure of the curing agent employed. The analysis 15. D. VerchBre, H. Sautereau, J. P. Pascault, C. C. Ric-
of the rubber modulus in stoichiometric epoxy mix- cardi, S. M. Moschiar, and R. J. J. Williams, Macro-
tures, obtained with various curing agents, has molecules, 23, 725 (1990).
shown that for a similar crosslink density the Tg 16. K. Dusek, and L. Matejka, Adu. Chem. Sci., 2 2 2 , 3 0 3
varies upon the chemical backbone structure be- (1989).
tween crosslink sites. 17. V. Bellenger, J. Verdu, J. Francillette, P. Hoarau, and
This study has been concerned with the analysis E. Morel, Polymer, 2 8 , 1079 (1987).
of the relationships between cure evolution and ul- 18. J. B. Enns and J. K. Gillham, J. Appl. Polym. Sci.,
2 8 , 2567 (1983).
timate properties of epoxy systems and the chemical
19. S. Lunak, J. Vladyka, and K. Dusek, Polymer, 19,
structure of the amine-type hardener employed. 931 (1978).
Such studies can be used to control the network for- 20. E. F. Oleinik, Adu. Polym. Sci., 80, 49 (1986).
mation and to develop structure-property relations 21. I. Mondragon and C. Bucknall, Plast. Rubb. Compos.
for thermosetting mixtures. Proc. & Applic., 21, 275 (1994).
22. F. Meyer, G. Sanz, I. Mondragon, and J. Mijovic,
Polymer, to appear.
23. G. Sanz, J. Garmendia, M. A. AndrBs, and I. Mon-
REFERENCES dragon, J . Appl. Polym. Sci., to appear.
24. A. V. Tobolsky, D. W. Carlson, and N. J. Indicator,
1. C. A. May, Epoxy Resins. Chemistry and Technology, J . Polym. Sci., 5 4 , 175 (1961).
Marcel Dekker, New York, 1988. 25. J. M. Charlesworth, Polym. Eng. Sci., 2 8 , 2 3 0 (1988).
2. B. Ellis, Chemistry and Technology of Epoxy Resins, 26. M. S. Vratsanos and R. J. Farris, Polym. Eng. Sci.,
Chapman and Hall, New York, 1993. 2 9 , 8 0 6 (1989).
3. B. A. Rosenberg, Adu. Polym. Sci., 7 5 , 113 (1986). 27. C. D. Wingard and C. L. Beatty, J . Appl. Polym. Sci.,
4. K. Dusek, M. Ilavsky, S. StokrovL, L. Matejka, and 4 1 , 2 5 3 9 (1990).
S. Lunhk, in Crosslinked Epoxies, B. Sedlacek and J. 28. T. Sasuga and A. Udagawa, Polymer, 3 2 , 4 0 2 (1991).
Kahovec, Eds., W. de Gruyter, 1987. 29. J. A. Schroeder, P. A. Madsen, and R. T. Foister,
5. A. C. Grillet, J. Galy, J. Pascault, and I. Bardin, Poly- Polymer, 2 8 , 9 2 7 (1987).
mer, 3 0 , 2 0 9 4 (1989). 30. V. Bellenger, W. Dhaoui, E. Morel, and J. Verdu, J.
6. H. Stutz, and J. Mertes, J . Polym. Sci. Part A: Polym. Appl. Polym. Sci., 35, 563 (1988).
Chem., 3 1 , 2031 (1993). 31. E. F. Oleinik, Adu. Polym. Sci., 80, 49 (1986).
7. A. Apicella, P. Masi, and L. Nicolais, Rheol. Acta, 2 3 , 32. E. S. W. Kong, Adu. Polym. Sci., 80, 125 (1986).
291 (1984). 33. V. G. Gupta and C. Brahatheeswaran, Polymer, 3 2 ,
8. D. Harran, M. F. Grenier-Loustalot, and Ph. Monge, 1875 (1991).
Eur. Polym. J., 2 4 , 225 (1988). 34. E. Balizer, J. V. Duffy, and G. F. Lee, ACS (PMSE),
9. S. M. Lee, International Encyclopedia of Composites, 6 0 , 8 3 7 (1989).
Vol. 2, VCH Publishers, New York, 1990.
10. K. Horie, H. Hiura, M. Sawada, I. Mita, and H. Received May 19, 1994
Kambe, J. Polym. Sci.: Part A - I , 8, 1357 (1970). Accepted October 27, 1994