Technical collection

Hygrothermal aging of a filled

epoxy resin

2007 - Conferences publications

E. Brun
P. Rain
G. Teissédre
C. Guillermin
S. Rowe
Schneider Electric 2007 - Conferences publications
2007 InternationalConference on Solid Dielectrics, Winchester, UK, July 8-13, 2007

Hygrothermal aging of a filled epoxy resin.

E. Brun2, P. Rain1*, G. Teissèdre1, C. Guillermin2, S. Rowe2

1
Grenoble Electrical Engineering Lab (G2Elab), CNRS- Université de Grenoble, Grenoble - France
2
Schneider Electric, Grenoble, France
* E-mail : [email protected]

Abstract: The hygrothermal conditioning of an epoxy propagating in the bulk material [11].
resin at 80°C under 80% RH has been followed by
weight measurements, thermogravimetric analysis For a better understanding of the overall mechanisms,
(TGA) and dynamical mechanical analysis (DMA). The the impact of the hygrothermal conditioning on the
samples are either filled with 60% by weight of silica physical properties have thus been carried out first.
flour or are not filled. Above an apparent saturation Electrical characterisations will ensue. In the following,
value of about 1.5% reached within a few days, a slight mass uptakes, thermogravimetric analyses and dynamic
but significant mass uptake was observed in the filled mechanical analyses are reported.
resin, especially after 50 days. The TGA showed an
evolution of the filled samples with conditioning after EXPERIMENTAL PART
50 days as well, which was not observed on unfilled
samples. For the filled samples, the elastic modulus in Materials, Sampling and Conditioning
the rubbery state decreased with conditioning. These
evolutions have been attributed to the formation of a The material used is a DGEBA based epoxy resin filled
degraded inter-phase region due to hydrolysis occurring or not with silica flour and cured with an anhydrid acid.
after the debonding of the filler-matrix interface caused The filler content is 60% by weight. The fillers’ sizes
by the absorbed water. range between a few 0.1 μm and 200 μm.

INTRODUCTION The components were mixed, cast in a mould then

maintained at 100°C during one hour. After cross-
Filled epoxy resin has been used for many years in linking, samples were demoulded and then cured at
electrical engineering. The material is submitted to 130°C during 16 hours. The sheets are 0.5 mm thick.
thermal and electrical stress and is in contact with the Glass transition temperatures of the filled and unfilled
environment. The mechanisms leading to the occurrence samples are 72°C and 70°C respectively.
of electrical breakdown are still not properly
understood. A great variety of electrical, thermal, Samples are cleaned with alcohol and dried in an oven
mechanical, chemical phenomena may be involved in at 50°C for 24 hours. After this preparation, their initial
the ageing of these polymeric insulations. Since the mass was measured. The samples were then conditioned
material may be exposed to a humid environment, the in a climatic chamber at 80°C under 80%RH.
present work focuses on the specific influence of water
on the material and its effects on the electrical rigidity. Characterisations

From an electrical point of view, the impact of water on Moisture uptake measurements: Samples were
the dielectric behaviour of filled epoxy resins has been periodically withdrawn from the climatic chamber.
extensively described. As concerns the electrical Before mass measurements were made, the temperature
rigidity, the breakdown voltages of wet materials may and hygrometry of the samples were stabilized. For this
fall by a factor of 5 to 10 in comparison with a dry purpose, they were laid in a small quantity of water
material [1-3]. In composites, the interfaces between the initially at 80°C. After 15 minutes, both water and
matrix and the mineral fillers are known to be zones of sample were at ambient temperature. Samples were then
weakness [4-6]. The shape of the fillers may also dried with a suitable paper. The mass uptake was
influence the breakdown voltage [7]. Furthermore, the measured with an electronic Ohauss balance Explorer.
deleterious influence of water on epoxy resin is well
known [8]. The main mechanisms leading to physical Thermogravimetric analysis (TGA): The thermal
and chemical degradation of epoxy resin have been stability of the materials has been evaluated throughout
illustrated or at least foreseen [9]. With the use of a conditioning by TGA with a TA instruments® 2050.
FTIR spectrometer, water layers of a few hundred nm The mass loss of the samples was measured during a
have been measured at an epoxy/glass interface [10]. temperature rise of 3°C/min between -40°C and 850°C.
This order of magnitude is in accordance with the MEB Experiments were carried under nitrogen to avoid added
observations reported in [3]. Water may accumulate at oxidation. A gas flow of 45 mL/min inside the oven
the interfaces and lead to a filler/matrix debonding, allowed the extraction of the thermolysis by-products.
which may be followed by mechanical cracks The relative mass loss of samples with initial mass of 10

1-4244-0750-8/07/$20.00©2007 IEEE.
239
Schneider Electric 2007 - Conferences publications

to 20 mg is reported hereafter. accumulates between the two phases, part of which

possibly reacts with the resin.
Dynamic Mechanical Analysis (DMA): The evolution
of mechanical properties and thermal transitions during TGA results
conditioning were measured by DMA with a TA
instruments® 2980 using the 3-point bending mode. Thermogravimetric analysis was carried out for filled
Taking into account the elastic domain of the material, and not filled samples after until 70 days of
the measurement conditions were the following: a conditioning. The thermal behaviour of unaged sample
dynamic magnitude of 50µm and a static force of 140% are mentioned as “reference” in the figures. The dots on
of the dynamic force. The samples were rectangular in the graphics are not measurements points but are added
shape 40×10×0.5mm3. The frequency was 1Hz and the to improve visualisation of the data.
temperature rise was 3°C/min from -40°C to 140°C.
The thermograms are displayed in Figure 2 for unfilled
RESULTS AND DISCUSSION samples. The derivatives of the mass loss curves are
also displayed to highlight the differences between the
Moisture uptake results samples. A detail of the curves between 50°C and
200°C is enlarged. This shows a significant decreasing
Three different samples, either filled or not filled, were of the mass between 50°C and 120°C in aged samples,
measured for each conditioning duration. Mean values which is not observed in reference samples. This mass
and standard deviations are reported in Figure 1. For a loss is slightly lower than the water uptake measured by
better comparison of the two materials, mass variations gravimetry. It corresponds to the evaporation of
of the filled samples were calculated taking into account absorbed water. Xu et al [13] observed an initial, weak,
the initial mass of unfilled resin, evaluated by the mean peak of the derivative weight in the same temperature
filling content of 60%. range that they attributed to the evaporation of water
contained only in the free volumes. They considered
Water saturation of not filled samples is observed after that the bound water was released at 200-300°C where
about 5 days. The water uptake was then of 1.5%. The they observe a slight difference between as-cured and
mass uptake of filled samples is lower at the beginning aged samples. Our results do not confirm this point. The
than the one of unfilled ones: the kinetics of water main mass loss occurred between 300 and 450°C. For
absorption is slower. The water diffusion is impeded by all aged unfilled samples investigated so far, the
the silica grains, which slow down its propagation derivative weights display a maximum at about 398°C
throughout the whole material. After 5 days and a shoulder (maximum of the second derivative) at
conditioning, the mass variation was close to the one of 364°C. This corresponds to resin decomposition. At
unfilled samples but measurements tend to indicate that 800°C, a residual mass of about 6% was measured
the mass uptake continues to progress after the quasi– which was not observed in similar experiment under
saturation has been reached. This specific behaviour of oxygen flow. MEB analysis showed the presence of
filled samples has already been reported in [12]. carbon with low quantities of oxygen. MEB photos
Furthermore, a slope increasing after about 50 days has showed blackish sheets. This residue was then mainly
been observed in a similar way to that shown in [13]. composed of carbon graphite.
These measurements will be completed to confirm this
tendency. This difference observed between the two For filled samples, residual masses were scattered and
materials can be attributed to phenomena taking place at not correlated with the duration of conditionning. The
the epoxy/silica interfaces. The following mechanism mean value was 64%. Taking into account the residual
can be proposed. Firstly, water molecules break the
physical bounds between the silica grains and the resin
and form H-bounds with the polymer. Then, some water

2
1.8
1.6
mass uptake (%)

1.4
1.2
1
0.8
0.6
0.4 Unfilled samples
0.2 Filled samples
0
0 20 40 60 80
Conditioning time (days)
Figure 2: Weight loss during a TGA dynamic test for
Figure 1: Weight gain in unfilled (Ŷ) and filled (Ÿ) unfilled samples before (•) and after conditioning at
samples as a function of conditioning time. 80°C and 80%HR during 5 (+), 14 (Ŷ) and 50 (Ƈ) days.

240
Schneider Electric 2007 - Conferences publications

mass of 6% measured for the unfilled samples, this DMA results

value leads to a filler content of 58% which corresponds
well with the “rated” filler content of 60%. To simplify The dynamic mechanical analysis was carried out on
the comparisons, the variations of the residual masses four filled and four unfilled samples after 0, 5, 14 and
were neutralized in the results displayed in Figure 3. 50 days of conditioning. The evolution of the elastic
This shows the relative mass losses between the initial moduli E’ is displayed in Figure 4a. The magnitudes for
mass values of the samples and the residual masses. the filled samples were about three times larger in the
glassy state and five times larger in the rubbery state.
A mass loss between 50°C and 200°C can be observed, The elastic modulus decreased during conditioning,
as in unfilled samples. In these cases, the variations especially in the glassy state. In practice, water
correspond well with the water uptake measured before. molecules break the hydrogen bonds established inside
The resin decomposition can be observed in the same the network. The bound water increases also the
temperature range as above. As shown in the zoom in mobility of the polymer chains and causes a decrease of
Figure 3, decomposition started at lower temperatures the elastic modulus. This is the well-known water-
for longer conditioning periods. This is clear if we induced plasticization effect.
compare the curves for 50 and 70 days, which fit the
curve of 14 days until about 200°C and decrease faster In the rubbery state, we would expect the effect to be
between 200°C and 300°C. This indicates an evolution less marked since the chains are already mobile. Thus in
of the material produced by the hygrothermal this case, the modulus E’ depends theoretically on the
conditioning after about 50 days. Since this effect was density of cross-links in the epoxy network [14].
not encountered in samples without fillers, the Nevertheless, the conditioning induced a decrease of the
epoxy/silica interface regions are necessarily involved. elastic moduli E’ for the filled samples. For unfilled
We have already mentioned that the mass measurements samples, the measurements were conducted at the lower
suggest an accumulation of matter close to these limit of sensitivity of the DMA instrument. Such a
interfaces. We may consider that this evolution is due to decrease of the rubbery modulus has already been
an hydrolysis of the filler/matrix interface regions which observed in [15]. It is not due to the plasticization but
creates a degraded inter-phase region. Since the filler may be attributed to a degradation of the polymer,
content is large, this degraded region may also fill the which probably results from hydrolysis of the resin.
whole gap between neighbouring fillers and constitute
“weak” path between them. This hydrolysis occuring
after about 50 days is probably correlated with the
increase of the mass uptake observed at the same time.
The possible explanation of this is as follows: free space
is created by the hydrolytic reactions consisting mainly
in the attack of the ester linkages and creation of acid
groups [4]. This allows extra water molecules to
accumulate between the filler and the degraded region.
However, the extra quantity of water involved in this
mechanism would be low in comparison with the total
quantity of absorbed water, which was almost entirely
lost between 50°C and 150°C. Further experiments are (a)
under way to confirm this hypothesis.

100
90 REF (b)
80 5 days
14 days
M(Minitial-Mfinal)%

70
50 days
60
70 days
50
REF
40
5 days
30 14 days
20 50 days
70 days
10
0
0 100 200 300 400 500 600
T(°C)

Figure 3: Weight loss during a TGA dynamic test for

Figure 4: Elastic modulus E’ (a) and loss modulus E’’
filled samples before (•) and after conditioning at 80°C
(b) vs temperature of unfilled (---) samples and filled
and 80%HR during 5 (+), 14 (), 50 (Ƈ) and 70 () days
(___) samples before (•) and after conditioning at 80°C,
with a zoom in the temperature range 0-350°C between
80%HR during 5(+), 14(Ŷ) and 50 (Ƈ) days.
92 and 100%.

241
Schneider Electric 2007 - Conferences publications

The glass transition temperatures have been taken at the [4] M.K. Antoon and J.L. Koenig, “Irreversible effects
maxima of the curves of loss moduli E’’ (Figure 4b). of moisture on the epoxy matrix in glass-reinforced
For the reference samples, the Tg were 70°C and 72°C composites”, J.Polym. Sci. Phy. Ed., vol. 19, p.
respectively for the unfilled and filled samples. These 197-212, 1981.
maxima shifted towards the low temperatures during [5] Janssen H., Seifert J.M. and Kärner H.C., Interfacial
conditioning, thus highlighting water-induced phenomena in comopsite high voltage insulation,
plasticization. This decrease occurred mainly within the IEEE Transactions on dielectrics an Electrical
first days of conditioning and is more important in the Insulation, vol. 6, 1999, p. 651-659.
unfilled samples. After a 50 days conditioning, the shift
of the Tg values were of 14K and 11K for the unfilled [6] D.H. Kaelble and P.J. Dynes, “Hygrothermal aging
and filled samples respectively. Similar evolutions have of composite materials”, J. Adhesion, vol. 8, p. 195-
been already mentioned [16]. A secondary peak, which 212, 1977.
would have indicated the heterogeneity of the material [7] M. Ezoe, M. Nakanishi and J. ShouGou, “Effects of
[6, 12, 17], was not visible here. water absorption on interfacial phenomena in HV
insulating materials”, IEEE High Voltage
CONCLUSION Engineering Symposium, 1999, 4.244-4.247, 1999.
[8] A.C. May, Epoxy resins, New York, Dekker
Evidence of aging of filled epoxy resin conditionned at Edition, 1988.
80°C and 80%RH were provided by weight
measurements, thermogravimetric analysis (TGA) and [9] J. Verdu, Action de l’eau sur les plastiques, Paris,
dynamical mechanical analysis (DMA). Techniques de l’ingénieur, vol AM 3 165.
[10] T. Nguyen, E. Byrd and D. Bentz, “In situ
Above an apparent saturation value of about 1.5% measurement of water at the organic
reached within a few days, a slight but significant mass coating/substrate interface”, Progress in organic
uptake was observed in the filled resin. An coatings, vol. 27, p. 181-193, 1996.
accumulation of water at the epoxy/silica interface can [11] K.A. Kasturiarachchi and G. Pritchard, Scanning
be inferred, part of which may have chemically reacted electron microscopy of epoxy-glass exposed to
with the polymer. humid condition, J. of Mat. SC., vol. 20, p. 2038-
2044, 1985.
Thermogravimetric analysis showed a degradation of
filled samples after a conditioning period longer than 50 [12] M.J. Adamson, “Thermal expansion and swelling
days, which does not occur in unfilled samples. of cured epoxy resin used in graphite/epoxy
composite materials”, J. Mater. Sci., vol. 15, p.
The decrease of the elastic modulus E’ and of the glass 1736-1745, 1980.
transition temperature Tg underlined the plasticization [13] S. Xu and D.A. Dillard, “Environmental aging
of the samples during conditioning. For the filled effects on thermal and mechanical properties of
samples, the elastic modulus in the rubbery state electrically conductive adhesives”, J. Adhesion, vol.
decreased with conditioning. 79, p. 699-723, 2003.
[14] T. Murayama and J.P. Bell, “Relation between the
These results suggest the creation of a degraded inter- network structure and dynamic mechanical
phase region between the silica and the epoxy matrix properties of a typical amine-cured epoxy
due to the hydrolysis of the resin. polymer”, J. Polym. Sci., vol. 8, p. 437-445, 1970.

REFERENCES [15] K.I. Ivanova, R.A. Pethrick and S. Affrossman,

“Investigation of hydrothermal aging of a filled
rubber toughened epoxy resin using dynamic
[1] H.C. Kärner and M. Ieda, “Technical aspects of
mechanical thermal analysis and dielectric
interfacial phenomena in solid insulating”, Proc. of
spectroscopy”, Polymer, vol. 41, p. 6787-6796,
the 3rd IEEE Int. Conf. On Prop and Appl of Diel
2000.
Mat, p. 592-597, 1991.
[16] A. Apicella, L. Egiziano, L. Nicolais and V. Tucci,
[2] T. Kumazawa, M. Oishi and M. Todoki, “High-
“Environmental degradation of the electrical and
humidity deterioration and internal structure change
thermal properties of organic insulating materials”,
of epoxy resin for electrical insulation”, IEEE
J. Mater. Sci., vol. 23, p. 729-735, 1988.
Transactions on dielectrics an Electrical
Insulation, vol. 1, n°1, p. 133-138, 1994. [17] L. Wang, K. Wang, L. Chen, C. He, L. Wang et al,
“Hygrothermal effects on the thermomechanical
[3] T. Kumazawa, M. Oishi, M. Todoki and T.
properties of high performance epoxy/clay
Watanabe, “Physical and chemical structure change
nanocomposites”, Polymer Engineering and
of filled epoxy due to water absorption”, 5th IEEE
Science, p. 215-221, 2006.
int.Conf. on Prop and Appl of Diel Mat, p. 491-494,
1997.

242