Carbon fiber-reinforced epoxy filament-wound composite
laminates exposed to hygrothermal conditioning
Jose
´
Humberto S. Almeida Jr.
1
•
Samia D. B. Souza
2
•
Edson C. Botelho
2
•
Sandro C. Amico
1
Received: 14 November 2015 / Accepted: 25 January 2016 / Published online: 5 February 2016
Springer Science+Business Media New York 2016
Abstract This study focuses on the evaluation of the
effect of hygrothermal conditioning on tensile, compres-
sive, in-plane and interlaminar shear properties, and also on
the viscoelastic characteristics of carbon fiber/epoxy lam-
inates. Flat unidirectional laminates were manufactured by
dry filament winding and cured under hot compression.
The laminates were later exposed to hygrother mal condi-
tioning in a chamber, following the recommendations of
ASTM D5229M. All composite coupons were tested before
and after conditioning. An analytical Fickian model was
used to fit experimental data, showing very good estimates.
Shear strength and modulus reduced to about 30 and 38 %,
respectively. All specimens presented acceptable failure
modes; shear specimens failed at the gage section with
delaminations and fiber/matrix debonding, whereas short
beam specimens failed via delaminations at the specimen
mid-plane. Moisture penetration through the carbon/epoxy
surface lead to interfacial debonding and matrix plasti-
cization. Puck’s failure envelope accurately predicted
failure under compressive and shear loading.
Introduction
Carbon fiber-reinforced epoxy composites are commonly
used in applications with high structural demand, such as
marine, aeronautical, and aerospace sectors [1]. Among the
manufacturing processes available for advanced polymer
composites, filament winding (FW) stands out, being able
to produce from flat laminates to elbows and curved-sur-
face structures. The high fiber volumetric fraction and
precision in angle deposition, along with the use of a ten-
sioned and continuous reinforcement are the main features
of FW.
The oil and gas sector, in particular, is increasingly
replacing metallic-based structures with composites based
on the need to increase payload in marine structures by
reducing weight and increasing corrosion resistance.
Indeed, pipelines, subsea systems, and submersible struc-
tures operate under aggressive environments, which mod-
ify the mechanical response of the composite, decreasing
damage tolerance and sometimes leading to premature
failure [2].
Prediction of failure and damage in composites is typi-
cally complex mainly due to their orthotropic characteristic,
and this becomes even more complex when the material is
under extreme weathering conditions. When under these
environments, composites can absorb moisture, which
affects their long-term durability and properties [3]. Since
high-performance carbon fibers absorb virtually no mois-
ture, absorption is largely matrix-dominated [4], suggesting
that matrix-dominated failure mechanisms will be more
significantly affected. Also, if the matrix content is high or if
a fiber more prone to absorption is being used (e.g., pol-
yaramid), the greater the expected water absorption.
Moisture permeation is dominated by diffusion, capil-
larity, and/or transport by micro-cracks, and the rate of
& Jose
´
Humberto S. Almeida Jr.
jhsajunior@globomail.com
1
PPGE3M, Federal University of Rio Grande do Sul, Av.
Bento Gonc¸alves 9500, Porto Alegre, RS 91501-970, Brazil
2
Department of Materials and Technology, Universidade
Estadual Paulista (UNESP), Av. Ariberto Pereira da Cunha
333, Guaratingueta
´
, SP 12516-410, Brazil
123
J Mater Sci (2016) 51:4697–4708
DOI 10.1007/s10853-016-9787-9
moisture uptake varies with the type of matrix, fiber ori-
entation, water temperature, moisture [5], and fiber content
(V
f
). Moisture wicking can promote changes in mechanical,
thermo-mechanical, and thermo-physical characteristics of
the matrix by plasticization, swelling, cracking, and
hydrolysis, and it can also degrade fiber-matrix bonding [3,
6]. When com bined with high temperatures, it may induce
irreversible structural change s in the polymer network. The
extent of damage may change swelling, since cracking and
blistering cause higher uptake, whereas the leaching of
small molecu le components results in gradual decrease in
weight gain [7].
Ultraviolet radiation (UV) can also degrade the matrix
and cause irreversible structural changes. UV radiation can
trigger photolytic, photo-oxidative, and thermo-oxidative
reactions within the matrix. Degradation can cause from
simple discoloration to substantial loss of mechanical
properties as a result of C–C polymer chain scission [8] due
to photo-oxidation [9]. Another effect is the cross-linking,
which restricts molecular mobility and reduces the ability
of the material to accommodate externally applied strain
[10].
Marine structures sometimes operate under combined
UV and seawater influence both being detrimental to
polymeric composites. Botelho et al. [11] evaluated the
effect of different hygrothermal conditionings on the shear
behavior of carbon/epoxy composites and Mouzakis et al.
[9] studied accelerated environmental aging of glass/
polyester composites, reporting a significant decrease in
compression and shear strength. Thus, this paper addresses
the effect of environment al conditioning (accelerated water
immersion) on the tensile, compressive, in-plane shear, and
viscoelastic properties of flat filament-wound carbon fiber/
epoxy laminates.
Experimental
Materials and manufacturing
Carbon fiber/epoxy prepreg tow (towpreg) from TCR
Composites, with Toray T700-12 K-50C carbon fiber and
UF3369 epoxy resin system, was used in this work. The flat
laminates were manufactured using a rectangular stainless
steel mandrel (327 9 228 9 12 mm
3
) and a KUKA robot
KR 140 L100 with control and peripheral devices from
MFTech, with an achievable tolerance in the winding angle
lower than 0.5 %. Hoop flat laminates were produced on
the designed mandrel, as presented in Fig. 1. Two simul-
taneous rovings (towpregs) were used to produce the
laminates. The towpregs were then wound onto the man-
drel. After winding, a polyester-based shrink tape was used
to wrap and help consolidating the laminate during the
curing process that followed. The tape begins to shrink
around 70 C and reaches a maximum shrink force at
150 C. This aids ply compaction and the elimination of
voids, yielding a resin-rich part surface.
The laminates were cured by hot compression under six
ton for 4 h at 130 C. The final fiber volume fraction was
&72 % (measured by acid digestion following ASTM
D3171-11) and mean ply thickness was 0.35 mm in the
4-layer laminate (overall thickness of 1.40 mm).
Weathering
The carbon fiber/epoxy specimens were exposed to a
combination of temperature and humidity for 60 days at
80 C under a relative humidity of 90 % inside a
hygrothermal climatic chamber Marconi, model MA835/
UR. The hygrothermal exposure was carried out followi ng
the recommendations of ASTM D5229M-14. Th e samples
went through a drying procedure to guarantee that they
were dry and in mass equilibrium before aging, preventing
that initial and atmosphere humidity affected water uptake
values [1, 12]. The drying cycle can be summarized as
follows: the sample was weighed, placed in an oven at
110 C for 24 h, removed from the oven and placed in a
desiccator, weighed again, and placed in an oven at 110 C
for 3 h. These stages were repeated until the specimen
reached mass equilibrium, and only then, they were placed
in the hygrothermal chamber for 60 days at 80 C and
90 % humidity. The sample mass was periodically moni-
tored, and moisture absorption (M) was calculated using
M ¼ðM
w
M
d
Þ=M
d
, where M
w
and M
d
are the wet and
dry masses, respectively.
In order to better understand moisture absorption and
diffusion response of the composites herein analyzed,
Fick’s analytical model [13], shown in Eq. (1), was applied
to the data.
Fig. 1 Manufacturing of a flat unidirectional laminate by filament
winding
4698 J Mater Sci (2016) 51:4697–4708
123
M
t
¼ M
1
1 exp 7:3
Dt
h
2
0:75
"#()
ð1Þ
where M
t
is water uptake at a particular time t, M
1
is the
mass at a quasi-equilibrium state, D is the diffusion coef-
ficient, and H is the specimen thickness. The diffusion
coefficient is calculated from the absorption curve, as
shown in Eq. (2):
D ¼ p
h
4M
1
2
M
2
M
1
ffiffiffiffi
t
2
p
ffiffiffiffi
t
1
p
2
ð2Þ
where the subscripts in M and t refer to a particular mass
and time, respectively.
Characterization
The unidirectional composites were cut longitudinally and
transversely to the fiber direction. Mechanical testing was
carried out using an Instron 3382 universal testing
machine, with 100 and 5 kN load cells, in samples before
and after the environmental conditioning. The tests were:
• Tensile This test was performed at a cross-he ad speed
of 2 mm/m in in five tabbed coupons of controlled
geometry dimensions following ASTM D3039-14.
From this test, the elastic moduli, tensile strengths,
and Poisson’s ratio were obtained. Two extensometers,
one longitudinal and another transversal, were coupled
to the specimens for testing and later removed just prior
to rupture .
• Compression The compressive strength was obtained
with the combined loading compression test according
to ASTM D6641-09 in five tabbed specimens. The size
of the specimens was 140 9 12 9 1.4 mm
3
, the gage
length was 12 mm, and the speed rate was 1.3 mm/min.
• In-plane shear The V-notched rail shear method
(ASTM D7078-12) test was chosen. The specimen
(dimensions: 76 9 56 9 2.8 mm
3
) was centrally
V-notched on both sides. The notch angle is 90 and
the radius is 1.3 mm. Cross-head speed was 1.5 mm/
min and the shear modulus was determined on six
samples using a strain gage rosette aligned at ± 45 at
the mid-section of the sample (as seen in Fig. 2).
• Short beam Interlaminar shear strength was evaluated
through short beam tests, following ASTM D2344-13,
using a span-to-depth thickness (s:t) ratio of 4:1.
Length and width of the samples followed the recom-
mendations of the standard (6 9 t and 2 9 t,
respectively).
• Dynamic mechanical analysis (DMA) DMA was used
to evaluate the glass transition temperature (T
g
) of the
polymeric matrix. Analysis was carried out in a TA
Instruments 2980 DMA Dynamic Mech anical Analyzer
under single-cantilever bending mode at 1 Hz fre-
quency, for a maximum displacement of 10 mm, under
N
2
atmosphere, with a heating rate of 3 C/min, on
20 9 10 9 2.8 mm
3
specimens.
• Failure analysis Fractographic studies were carried out
using optical micrographs in a Carl Zeiss Axio scope
and scanning electron mi croscopy (SEM), in a Phenom
ProX equipment.
Failure envelope
Classical failure criteria (Ma ximum stress , Tsai-Wu [14],
Tsai-Hill [15], Hashin [16], Christensen [17], and Puck
[18]) were used to predict failure envelope in the r
22
–s
12
plane, assuming the stress in fiber direction being r
1
, stress
in the transverse r
2
, and s
12
the shear stress.
Results and discussion
Moisture uptake
Figure 3 presents the mass uptake for the 0 unidirectional
carbon fiber/epoxy coupons (average of five coupons) after
hygrothermal conditioning. Water uptake is influenced by
(i) the hydrophilic character of the matrix and fibers, (ii)
fiber/matrix adhesion, (iii) micro-cracks, and (iv) voids. The
resin network controls water uptake and, in turn, the absorp-
tion of water influences the network. Formation of voids and
micro-cracks is induced by water absorption, becoming more
important as the saturation level increases [13].
Fig. 2 V-notched shear test with a bonded rosette
J Mater Sci (2016) 51:4697–4708 4699
123
Hot/wet environment accelerates the deterioration pro-
cess of an epoxy-based composite. Reproducibility of the
experimental data is well predicted by the Fickian model,
both experimental and analytical data being representative
in terms of water content of the material. The initial stage
till pseudo-equilibrium is typical of a thermally activated
Fickian response, whereas, at longer times, the uptake
occurs at a lower rate mainly due to a combination of
mechanisms such as relaxation of the glassy epoxy net-
work, filling of micro-voids, and debonded zones with
water by wicking [19]. Thus, moisture absorption is
expected to take place via diffusion.
Kinetics of the diffusion process depends on tempera-
ture and relative moisture absorption. The higher the rel-
ative moisture absorption, the greater the absorption rate.
Water saturation (&0.37 %) reached at around 30 days,
and maximum mass uptake, about 0.40 %, was noticed
after 42 days of conditioning, both are typical of epoxy
resins. Moreover, mass uptake values were relatively low
compared to literature data for the same matrix and fiber [8,
12, 13, 20], which can be justified by the high fiber volu-
metric fraction and good compaction of the layers.
Viscoelastic properties
Figure 4 shows the storage modulus (E
0
) for the uncondi-
tioned and conditioned composites studied. As expected,
storage modulus for 0 specimens was higher than for 90
in all states (glassy, elastomeric, and rubbery). This par-
ticular DMA analysis was performed under bending mode,
which is markedly dependent on fiber orientation, which is
stiffer for fibers aligned along the length of the specimen.
Unconditioned specimens showed higher storage modulus
compared to the weathered ones, which is related to the
plasticizing effect promoted by moisture uptake.
The aged composites showed similar behavior at the
glassy state, perhaps due to a post-curing that may take
place during aging. The relatively high temperatures
involved, although not severe enough to break chemical
bonds of the polymer [8], may contribute to the appearance
of free radicals in the epoxy molecules, leading to further
cross-linking [9].
The loss modulus (E
00
) is related to energy dissipation in
the material, and composites with poor interfacial bonding
are prone to dissipate more energy. Some specimens pre-
sented a more flattened loss peak, where the hygrother-
mally conditioned specimens had lower loss modulus, as
seen in Fig. 5. Analysis of the curves shows that the
behavior was more affected by fiber orientation than by
weathering, and the specimens with fibers at 0
o
presented
much higher peaks. For the 90
o
specimens, the aged lam-
inate dissipated slightly less energy. This can be attributed
Fig. 3 Mass gain obtained from experimental measurements and
Fickian diffusion prediction
Fig. 4 Storage modulus of the unconditioned and weathered
specimens
Fig. 5 Loss modulus of the unconditioned and weathered composites
4700 J Mater Sci (2016) 51:4697–4708
123
to the inhibition of relaxation processes in the composites,
decreasing mobility at the fiber/matrix interface.
Tan d curves are shown in Fig. 6. The unconditioned
specimens showed similar temperature at the peak (i.e.,
T
g
), and the conditioned samples showed similar and lower
T
g
values. The weathering effect on the composites is more
clearly observed in this figure, i.e., the conditioned speci-
mens show larger areas under the peak and lower T
g
,
suggesting matrix degradation [21].
Table 1 presents the T
g
extracted from (i) intersection
between the extrapolation of the elastomeric plateau and
the glass state from the storage modulus curve, (ii) the loss
modulus peak, and (iii) the tan d peak. The unconditioned
0
o
specimen showed the highest T
g
, but the aged 0
o
spec-
imen showed the lowest value in all methods; thus this
specimen is more damaged by aging, confirming that fiber
orientation has little effect on the glass transition of the
polymer.
Mechanical properties
Figure 7 presents the typical load versus displacement
curves for all samples. Based on the mean results of five
specimens with acceptable failure, conditioning reduced
tensile strength of the laminates from 1409 ± 131 to
1091 ± 114 MPa, around 29 %. Although the 0-oriented
specimens presented a slight flattening near the failure,
these specimens presented a brittle behavior, typical for
this type of loading/sample.
Figure 7 also shows tensile behavior of the 90
o
-oriented
composites. Brittle behavior was dominant for both non-
aged_[90]
4
and aged_[90]
4
specimens and the ultimate load
was significantly low compared to longitudinally oriented
laminates. In addition, aging reduced in about 25 % their
tensile strength.
Even though the load versus displacement curves
somewhat differ for the various longitudinally oriented
specimens, all coupons showed similar failure mode (see
Fig. 8a), with a sudden failure of the fibers. All 10 fractures
(five for aged and five non-aged) were characterized as
explosive gage middle. The failure mode for all 90
o
sam-
ples, conditioned and non-conditioned, was also similar—
lateral gage middle failure (see Fig. 8a), with only one
lateral at the top tab failure. Since the optical micrographs
indicated a good quality laminate, essentially free of voids
(Fig. 8c, d), a weakened fiber/matrix interface may have
appeared after aging (Fig. 8d), corroborated by some resin-
rich areas due to water uptake and more brittle fractu re at
the carbon/epoxy interface, which reduced tensile strength.
Furthermore, Fig. 8d suggests that broken fibers tend to
initiate failure in adjacent fibers.
Stress versus strain data for the tensile testing of non-
aged and aged composites are shown in Fig. 9. The final
properties may be summarized as follows: (i) Non-aged:
E
1,t
= 129.8 ± 5.6 GPa and E
2,t
= 9.1 ± 0.5 GPa, and
Fig. 6 Tan delta of the unconditioned and weathered specimens
Table 1 T
g
determined by
different methods
Specimen T
g
(C) from onset E
0
drop T
g
(C) from E
00
peak T
g
(C) from tan d peak
Non-aged_[0]
4
78.3 91.0 98.1
Aged_[0]
4
35.2 47.7 57.9
Non-aged_[90]
4
68.9 84.8 94.7
Aged_[90]
4
44.7 57.8 67.4
Fig. 7 Typical load 9 displacement tensile curves for the studied
families of unidirectional laminates
J Mater Sci (2016) 51:4697–4708 4701
123
![Fig. 8 Optical micrographs are shown for the fractured non-aged_[0]4 a, aged_[0]4 b, non-aged_[90]4 c, aged_[90]4 d tensile coupons](/figures/fig-8-optical-micrographs-are-shown-for-the-fractured-non-2z1g57sa.webp)
