![Figure 8 c mode relaxation times as a function of temperature after hydrothermal aging. [Color figure can be viewed in the online issue, which is available at www. interscience.wiley.com.]](/figures/figure-8-c-mode-relaxation-times-as-a-function-of-39eh86o9.webp)
![Figure 10 b1 mode relaxation times after 80% RH aging as a function of temperature. [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.]](/figures/figure-10-b1-mode-relaxation-times-after-80-rh-aging-as-a-u8h1rzeb.webp)
![Figure 9 b1 mode relaxation times after 80 C aging as a function of temperature. [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.]](/figures/figure-9-b1-mode-relaxation-times-after-80-c-aging-as-a-15pkdb4r.webp)
![Figure 4 Storage modulus G0 (squares) and loss modulus (circles) as function of temperature for the bulk adhesive at initial state. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]](/figures/figure-4-storage-modulus-g0-squares-and-loss-modulus-circles-3owc8qqj.webp)
![Figure 6 Isothermal dielectric loss at initial state as a function of frequency showing b1 relaxation. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]](/figures/figure-6-isothermal-dielectric-loss-at-initial-state-as-a-a3tlavhl.webp)
![Figure 1 Fracture appearance for three single lap shear samples at initial state. [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.]](/figures/figure-1-fracture-appearance-for-three-single-lap-shear-2i2f4bmq.webp)
![Figure 3 Fracture appearance for three single lap shear samples after 7 days under 80 C, 95%RH. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]](/figures/figure-3-fracture-appearance-for-three-single-lap-shear-i8vpiu2g.webp)
![Figure 11 b1 enthalpy–entropy compensation after hydrothermal aging conditions. [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.]](/figures/figure-11-b1-enthalpy-entropy-compensation-after-2vtz44mg.webp)
Any correspondence concerning this service should be sent to the repository administrator:
staff-oatao@inp-toulouse.fr
Open Archive Toulouse Archive Ouverte (OATAO)
OATAO is an open access repository that collects the work of
Toulouse researchers
and makes it freely available over the web where possible.
This is an author -deposited version published in: http://oatao.univ-toulouse.fr/
Eprints ID: 3844
To link to this article: DOI: 10.1002/app.31253
URL:
http://dx.doi.org/10.1002/app.31253
To cite this document : Chevalier, M. and Dantras, Eric and Tonon, Claire and Guigue,
Pascale and Lacabanne, Colette and Puig, C. and Durin, Christian ( 2010) Correlation
between Sub-Tg relaxation processes and mechanical behavior for different
hydrothermal ageing conditions in epoxy assemblies. Journal of Applied Polymer
Science, vol. 115 (n° 2). pp. 1208-1214. ISSN 0021-8995
Correlation Between Sub-T
g
Relaxation Processes and
Mechanical Behavior for Different Hydrothermal Ageing
Conditions in Epoxy Assemblies
M. Chevalier,
1
E. Dantras,
1
C. Tonon,
2
P. Guigue,
3
C. Lacabanne,
1
C. Puig,
2
C. Durin
3
1
Laboratoire de Physique des Polyme
`
res, Institut Carnot CIRIMAT, Universite
´
Paul Sabatier,
Toulouse Cedex 31062, France
2
Astrium, 31 avenue des Cosmonautes, Toulouse 31402, France
3
Centre National d’Etudes Spatiales, 18 avenue Edouard Belin, Toulouse 31401, France
DOI 10.1002/app.31253
ABSTRACT: The
aim of this study is to understand aging
phenomena by monitoring physical parameters after real
and simulated aging experiments. This study focuses on alu-
minum-epoxy assemblies, which are commonly used on
spacecraft structures. Different samples are submitted to
simulated aging tests. Influence of temperature and mois-
ture is analyzed. Evolution with aging is characterized at
two different scales. The macroscopic behavior of the assem-
blies is studied by single lap shear test. A decrease in the
shear rupture stress is observed with increasing temperature
and relative humidity. It is demonstrated that temperature
has more important influence. The molecular behavior in
the adhesive joint is studied by dynamic dielectric spectros-
copy measurements. This experiment gives access to molec-
ular mobility in the adhesive. Dipolar entities are identified
as evolving with aging conditions. The temperature is more
effective than moisture at this scale. An interpretation of the
molecular mobility before and after aging shows that water
is an important parameter of this study. A link between
mechanical and molecular behavior with hydrothermal
aging is found. The decrease of mechanical properties occurs
while failures become interfacial. In the same time, the inter-
actions between hydroxyether and water increase. The
evolution of the macroscopic behavior of the bonded assem-
blies is due to this combination observed at different scales.
Key words: epoxy; adhesive; dielectric spectroscopy;
aging; single lap shear
INTRODUCTION
Aging processes in materials used for space applica-
tions depend on various parameters (on ground and
space environments) and are not fully understood.
The influence of each factor remains obscur. A par-
ticular attention shall be paid on the aging of
bonded assemblies due to storage on ground and
thermal cycling in orbit.
During spacecraft integration, in clean room,
materials are used under humid and thermal con-
trolled conditions: 21 2
C, 50 5%RH. The inte-
gration activities and the storage can last several
years. Consequently, it is necessary to take into
account the time spent in clean room as an aging
factor. An aging process, which begins before the
launch could be the cause of an irreversible damage
during the flight. All the structures of the satellite
are su bmitted to these controlled environmental con-
ditions. The characterization and the simulation of
aging phenomena in bonded assemblies are the chal-
lenge to improve their conception and their durabil-
ity. Checking these assemblies after the launch is
impossible. One alternative is to use accelerated test-
ing on the basis of space industry knowledge and
ESA normative data.
1
Empirical laws are the only
way to obtain accelerated aging conditions. But, the
link between accelerated testing and real-time aging
is unknown. The aim of this study is to shed some
light on this correlation.
The methodology used for the work presented
later is to carry out an investigation of the mechani-
cal behavior and molecular dynamics evolutions in
thermoset bonded structures exposed to moisture.
The combined use of mechanical and dielectric
measurements has been chosen to evaluate epoxy–
water interactions at macroscopic and molecular
scales.
Epoxy–amine networks are relatively hydrophilic
materials. They are able to absorb 1–6% per
weight of water in many industrial formulations.
2
Numerous studies have been devoted to the water
Correspondence to: E. Dantras (dantras@cict.fr).
absorption process in these systems but this problem
remains only partly understood.
2–6
Water ingress in
polyepoxy networks leads to a range of effects
3
: bulk
dissolution by water in the polymer network, mois-
ture absorption on the surface of free volume holes
in the glassy structure, interaction of water with
hydrophilic polar moieties, such as amines and
hydroxyls by strong hydrogen bonds. It is also
reported that water is present in two different forms
in polyepoxy networks
4,7–9
: free water that fills the
microvoids of the netwo rk and bound water that
bonds with polar groups.
A decrease in mechanical properties of bulk and
bonded polyepoxy networks, such as modulus and
rupture strength under humid conditions is
observed. It is rela ted to plasticization.
9–11
But how
absorbed water affects the mobility of polymer
chains is not well know n. According to Mijovic
et al.,
5
the chemical and physical changes on the
molecular level during the early stage of environ-
mental exposure is the most important point in the
prediction of performance in adhesive joints.
The use of dynamic mechanical and dielectric
measurements allows us to investigate the molecular
dynamics. In poly epoxy systems, two to four relaxa-
tion regions have been reported by different experi-
mental techniques.
12–16
Some authors studied the
influence of water on molecular dynamics of amine
cured epoxy adhesives.
5,8,10,12
They reported that
some relaxational processes were influenced by the
ingress of water into the network. But some results
are contradictory, probably due to the complexity of
the commercial adhesive formulations.
Therefore, the main objective of this study is to
get a simultaneous evaluation of macroscopic and
molecular changes in bonded structures exposed to
hydrothermal environment.
EXPERIMENTAL
Materials
The studied system is an aluminum-epoxy assembly.
The adhesive is a commercial amine-epoxy bicompo-
sant adhesive packaged to be usable immediately.
The two parts are prepared and a nozzle allows us
to make and extrude the mix with an accurate
repeatability. The hardener (part A) is a mix of sev-
eral components, where aliphatic amine is prepon-
derant. The part B is based on diglycidyl ether of
bisphenol-A epoxy resin mixed with other compo-
nents (fillers, catalyst...). Parts A and B are mixed at
room temperature (ratio 2 : 1). The curing process is
7 days at 21 2
C, 50 5%RH. It leads to two
glassy transitions (T
g
) determined by Diathermal
Scanning Calorimetry measurements. One, equal to
60
C, is associated with the high-crosslink network.
The second one, equal to 10
C, corresponds to the
low-crosslink regions. Both substrates are wrought
aluminum alloys (2024 or 2017) widely used in aero-
space industry. By this way, differential thermal
expansions (DCTE) between the two substrates are
avoided. This assemb ly is considered as a model.
The joint thickness is ranging from 70 to 200 lm.
Hydrothermal aging conditions
During accelerated hydrothermal aging tests, the
joints are exposed to a warm humid environment
during 7 days. The temperature range is from 20 to
80
C (step ¼ 20
C) and the relative humidity is
ranging from 20%RH to vapor saturation by steps of
20%RH. Each temperature–relative humidity pair is
tested.
Methods
Single lap shear test
Single lap shear tests are performed at 25
C using
the tensile tension machine Instron 4505 with a 100
kN load cell. The crosshead speed is 1 mm/min.
Three specimens were studied for each test. Data
were recorded with the Serie IX software. The
ruptures were analyzed visually to determine the
percentage of cohesive failure.
Thermo mechanical analysis
Thermo mechanical measurements are performed
using an Advanced Rheometric Expansion System
strain controlled rheometer (TA Instuments) in the
torsion rectangular mode within the linear elasticity
range. Dynamic mechanical storage and loss modu-
lus G
0
and G
00
were recorded as function of tempera-
ture from 50 to 100
Cat3
C/min, for an angular
frequency equal to 1 rad/s.
Dynamic dielectric spectroscopy
Broadband dielectric measurements are performed
using a Novocontrol BDS 4000 covering a frequency
range from 10
2
Hz to 10
6
Hz with 10 points per
decade. Experiments were carried out isothermally
from 150 to 150
C by steps of 5
C. The tempera-
ture is controlled with an accuracy of 0.5
Cbya
nitrogen gas stream heated by a Quatro temperature
controller. The samples are round plate assemblies
(U ¼ 40 mm). We use the aluminum alloy substrates
as electrodes.
The real e
0
and imaginary e
00
parts of the rela tive
complex permittivity e* are measured as a function
of frequency f at a given temperature T. Experimen-
tal data are fitted by the Havriliak-Neg ami (HN)
[eq. (1)] function with an additional conductivity
term.
17,18
e
ðxÞ¼e
1
þ
e
s
e
1
ðÞ
1 þ ixs
HN
ðÞ
a
HN
ðÞ
b
HN
þ
r
0
ie
0
x
(1)
where e
1
is the real permittivity for high frequen-
cies, De is the relaxation strength, a
HN
and b
HN
are
the HN parameters, s
HN
is the relaxation time, x is
the angular frequency, r
0
is the d.c. conductivity,
and e
0
the dielectric permittivity of vacuum.
RESULTS
Mechanical behavior
Static mechanical behavior
The shear behavior is characterized at initial state.
The shear rupture stress r
R
is 27.5 0.3 MPa. The
fracture appearance is cohesive as shown in Figure
1. The cohesive surface is 89 1% of the bonded
surface. These values are considered as reference
values before aging.
Single lap shear tests are performed after different
aging conditions. The shear rupture stres ses are
shown in Figure 2 as function of aging temperature
for different relative humidity percentage. First, we
note that r
R
is higher than r
R
at initial state when
the relative moisture percentage of the tests is
40%RH, i.e. lower than clean room conditions. For
this moisture environment, r
R
value is independent
of the aging temperature. For a given relative
humidity percentage, r
R
exhibits a linear decrease as
function of temperature. The influence of relative
humidity is also analyzed. For isothermal aging con-
ditions, a nonlinear decrease of r
R
is observed as
function of moisture amount. The higher the temper-
ature is, the higher the moisture influence.
The evolution of the failures is studied. Figu re 3
shows the failure after aging at 80
C, 95%RH. The
cohesive failure percentage is reported in the
Table I. As function of increasing aging temperature
and humidity, we note that the rupture becomes
more interfacial. For 80
C and 95%RH aging condi-
tions, we observe that the ratio between interfacial
and cohesive failures is about one.
Dynamic mechanical behavior
The storage modulus G
0
and loss modulus G
00
are
shown in Figure 4 as function of temperature for the
bulk adhesive at initial state. The conservative mod-
ulus decreases from 1 GPa at 45–0.8 GPa at 40
C.
From 50 to 80
C, the decrease is stronger. G
0
reaches
10 MPa. This transition is associated with the
mechanical manifestation of glass transition. The
loss modulus exhibits two relaxation phenomena
Figure 1 Fracture appearance for three single lap shear
samples at initial state. [Color figure can be viewed in the
online issue, which is available at www.interscience.
wiley.com.]
Figure 2 Evolution of the shear rupture stress as a func-
tion of temperature and humidity percentage.
Figure 3 Fracture appearance for three single lap shear
samples after 7 days under 80
C, 95%RH. [Color figure
can be viewed in the online issue, which is available at
www.interscience.wiley.com.]
TABLE I
Cohesive Failure Percentage as a Function of
Hydrothermal Ageing Conditions
Aging moisture (%RH)
Aging
temperature (
C)
20 40 60 80
40 N/A 87 478 3 N/A
60 N/A 77 11 65 12 79 9
80 80 275 660 461 7
95 N/A 70 11 61 145 7
called x and a. x peak reaches a maximum at 10
C.
This relaxation is associated with heterogeneities in
the adhesive.
19
a reaches a maximum at T
a
tempera-
ture equal to 55
C.
Dielectric behavior
Molecular mobility at initial state
The dielectric loss from 150 to 150
C of an assem-
bly at initial state shown in Figure 5. Six dielectric
phenomena are pointed out. In the low temperature
range a weak and broad relaxation mode is
observed, labeled c. At higher temperature, a more
intense and narrow atypical mode b
2
is found. Then,
a weak and broad mode is detected: this mode is
called b
1
. The dielectric loss corresponding to this
relaxation is shown in Figure 6 for temperatures
ranging from 40 to 15
C. In the high temperature
range, three additional relaxations occur: x, a, and
MWS modes.
Secondary relaxation modes have various molecu-
lar origins. The c mode is related to the CH
2
mobil-
ity in aliphatic chain sequences.
12,13
The molecular
origin of b
2
and b
1
is more complex. b modes are a
combination of several molecular entities. According
to Ochi et al.,
15
b is mainly associated with the
mobility of hydroxyether and diphenyl propane
groups. At higher temperatures, x is associated with
heterogeneities in the adhesive as observed previ-
ously during thermo mechanical measurements.
19
a
is attributed to the dielectric manifestation of glass
transition. The MWS relaxation is attributed to
macrodipoles.
Relaxation times s
HN
are determined using the
Havriliak-Negami function [eq. (1)]. The temperature
dependence of c, b
2
, and b
1
relaxation times at initial
state is reported in the Arrhenius diagram presented
in Figure 7. The increase of dynamic dielectric spec-
troscopy (DDS) measurement temperature shows
that c, b
2
, and b
1
modes follow an Arrhenius law as
expected [eq. (2)].
sðTÞ¼s
0
exp
DH
RT
(2)
where DH is the activation enthalpy, s
0
is the pre-
exponential factor, and R is the universal gas constant.
The x, a, and MWS evolution is more difficult to
analyze. At high temperatures, conductivity appe ars.
Consequently, the use of the HN equation is more
complex: we observed a merging between high
temperature relaxation modes and conductivity
phenomenon. The behavior of these relaxation
Figure 4 Storage modulus G
0
(squares) and loss modulus
(circles) as function of temperature for the bulk adhesive
at initial state. [Color figure can be viewed in the online
issue, which is available at www.interscience.wiley.com.]
Figure 5 Dielectric loss at initial state reported on a loga-
rithmic scale as a function of temperature and frequency.
Figure 6 Isothermal dielectric loss at initial state as a
function of frequency showing b1 relaxation. [Color figure
can be viewed in the online issue, which is available at
www.interscience.wiley.com.]