J. Mater. Sci. Technol., 2011, 27(1), 69-73.

Influence of Voids on the Tensile Performance

of Carbon/epoxy Fabric Laminates

Hongyan Zhu1)† , Baochang Wu2) , Dihong Li1) , Dongxing Zhang1) and Yuyong Chen1)
1) School of Materials Science and Engineering, Harbin Institute of Technology, Harbin 150001, China
2) Harbin Aircraft Industry Co., Ltd., Harbin 150060, China
[Manuscript received August 21, 2008, in revised form December 4, 2009]

This paper aims at investigating the effect of voids on the tensile properties of [(±45)4 /(0,90)/(±45)2 ]S
and [(±45)/04 /(±45)(0,90)]S composites. Specimens with void contents in the range of 0.4% to 9.0% were
fabricated from carbon/epoxy fabric. The void content was determined by ultrasonic attenuation and optical
image analysis, and microscopic inspection was also used to analyze the shape and size of the voids. The
influence of voids on the tensile strength and modulus of both stacking sequence is compared in terms of the
size and the shape of the voids. The effect of voids on the initiation and propagation of tensile failure of both
stacking sequence composite was investigated.

KEY WORDS: Composite; Void; Tensile strength

1. Introduction properties decrease as the void content increases, the

magnitude of the effect of void content on the me-
Carbon fiber reinforced plastic (CFRP) compos- chanical properties of CFRP composites is different.
ites have been widely used in aerospace field for the Most work studied the effect of voids in the literature
unique advantages with regard to their high strength by quantifying voids by content (%). Recently, it was
to weight ratio. Guarantee of the reliability, safety realized that this approach is too simplistic. The mi-
and durability of aeronautic structures is the most im- crostructure (shape and size) and location of voids are
portant problem. For this specific aspect, void eval- important as through their stress effect they can influ-
uation will play an essential and fundamental role. ence whether or not a crack emanates from voids and
Voids are inevitable in the fabrication of fiber rein- propagates and finally result in failure of the compos-
forced polymer composites, and the effect of voids on ites. Michelle et al.[19] found triangular voids and the
the material properties is detrimental[1,2] . cracks emanate from the triangular voids.
The effect of voids on the mechanical properties Wisnom et al.[20] studied the effect of discrete
of CFRP laminates has been widely studied. The and distributed voids by embedding PTFE at the
matrix-dominated properties such as: interlaminar mid-plane of unidirectional glass fiber and carbon-
shear strength[3–6] , bending properties[7–9] , compres- fiber/epoxy plates on the interlaminar shear strength.
sive strength and modulus[10,11] , fatigue[12–14] and The reduction of the interlaminar shear strength was
fracture toughness[15-16] are influenced by voids, found to be between 8% and 31% depending on the
while the fiber-dominated mechanical properties are void length.
not significantly influenced by voids[8,9,17,18] . The purpose of this research is to study the mi-
Although all the works indicate that mechanical crostructure of voids and investigate the effect of voids
on the tensile strength and modulus. Two differ-
ent stacking sequences ([(±45)4 /(0,90)/ (±45)2 ]S and
† Corresponding author. Ph.D.; Tel.: +86 451 86418802; E-mail
address: [email protected] (H.Y. Zhu). [(±45)/04 /(±45)(0,90)]S ) which are widely used in
70 H.Y. Zhu et al.: J. Mater. Sci. Technol., 2011, 27(1), 69–73

aeronautic structures were fabricated for this study. (GB/T 3354-1999) was used.
A significant range of porosity (0.4%–9.0%) was pro- Ten specimens of each type with di-
duced. The panels were ultrasonically inspected, and mensions of 230 mm×25 mm×3.3 mm and
sectioned into tensile specimens. 230 mm×25 mm×4.3 mm (length×width×thickness)
were tested to assess the effect of porosity on the
2. Experimental tensile strength. The tests were performed in an In-
stron tensile testing machine under a test speed of
2.1 Fabrication of the specimens and lay-ups 2 mm/min.

CFRP composite laminates were produced with 3. Results and Discussion

T300/914 preimpregnated fabric (prepreg). The fiber
content of the prepreg was 58%–62%. The laminates 3.1 Effect of autoclave pressure on the void content
were produced using a vacuum bag and autoclave cure
technique. The standard cure cycle recommended by Table 1 presents the measurement of the void con-
manufactures was used. The temperature was in- tent at different autoclave pressures. As can be seen,
creased to 135◦ C at a rate of 1.5◦ C/min and dwelled the void content increases from 0.4% to 8.0% and
for 30 min. Then it was increased to 180◦ C at the 9.0% as the autoclave pressure goes from 0.4 MPa
same rate and dwelled for some hours. The purpose to 0.0 MPa.
of this paper was to study the effect of voids on the Table 1 Measurement of void content at different
tensile properties. In order to acquire a range of voids autoclave pressures
(0.4%–10.5%), test laminates were produced by us-
Void content/%
ing autoclave pressures of 0, 0.1, 0.3 and 0.4 MPa. Cure pressure/MPa laminate A laminate B
0.4 MPa was the manufacture s recommended pres-
0.4 0.4 0.4
sure. The vacuum is held during the curing process.
0.3 1.2 1.8
Two different stacking sequence were fabricated for 0.1 5.4 6.0
this study, namely [(±45)4 /(0,90)/(±45)2 ]S (A) and 0.0 8.0 9.0
[(±45)/04 /(±45)(0,90)]S (B) lay-ups.
All panels were inspected by using an ultrasonic
detector to determine the distribution of the void con- 3.2 Microstructural analyses
tent. Then the mechanical test and image analysis
specimens were cut from the plates. The locations of To assess the relationship between voids and ten-
the two kinds of specimens were close to each other sile strength, a detailed study was performed to
to assure that they have the same void content. characterize the voids in the laminates. Reference
specimens manufactured at the autoclave pressure of
2.2 Microstructural analysis 0.4 MPa and 0.0 MPa were used for this purpose.
Figure 1 shows a typical photomicrograph of lami-
Microscopic image analysis was employed to ob- nate A and B at the pressure of 0.4 MPa and 0.0 MPa.
tain statistical data on the shape, size and location of As can be seen from the figure, the voids are prefer-
voids in the carbon/epoxy samples. For each stack- entially located at the ply interface and run along in
ing sequence, four samples of each porosity level were the direction parallel to the ply. And most large voids
analyzed. locate in contact with fibers. The voids are spherical
Samples with section parallel and perpendicular and small at the pressure of 0.4 MPa. And the void
respectively to the length direction of the specimens becomes larger and elongated as the pressure goes
were embedded in an epoxy resin and polished care- from 0.4 MPa to 0.0 MPa.
fully with a polishing machine using 400, 600, 800, To systematically assess the effect of voids on the
1500 grit size silicon carbide abrasive paper. And the tensile properties, detailed analyses of the size and
specimens were finished with 1 μm diamond paste. shape of voids are necessary. An image analysis soft-
The quality of the polished surface is very important ware is used to quantify void size (length) and void
for image analysis. The polished surfaces were ob- shape (aspect ratio). The length is the dimension of
served using an optical microscope (VNT-100). The voids in the direction of the major axis. And aspect
images were analyzed by using an image analysis soft- ratio is defined as the ratio of length to width of voids.
ware. A quantification of void content, void shape Figures 3 and 4 show the length distribution and the
(aspect ratio), void size was obtained through this shape distribution based on aspect ratio under differ-
analysis. ent autoclave pressures. Both the shape distributions
are lognormal distributions, while the length distrib-
2.3 Tensile strength tests utions are not. They all have one peak. As the au-
toclave pressure goes from 0.4 MPa to 0.0 MPa, the
To assess the influence of porosity on the tensile fraction of void with larger length and aspect ratio for
strength of the carbon epoxy laminates, tensile testing both laminates increases.
 H.Y. Zhu et al.: J. Mater. Sci. Technol., 2011, 27(1), 69–73 71

Fig. 1 Micrographs of samples before testing: (a) A, 0.4 MPa, (b) A, 0.0 MPa, (c) B, 0.4 MPa, (b) B, 0.0 MPa

Fig. 2 Void length distribution at different autoclave pressures: (a) laminate A, (b) laminate B

Fig. 3 Void shape distribution based on aspect ratio at different autoclave pressures: (a) laminate A, (b) laminate B
72 H.Y. Zhu et al.: J. Mater. Sci. Technol., 2011, 27(1), 69–73

Fig. 4 Tensile properties as a function of void content: (a) strength; (b) modulus

Fig. 5 Comparison of the post-failure mode through the sample front view and side view: (a) A, 0.4%, (b) A,
8.0%, (c) B, 0.4%, (d) B, 9.0%

In order to study the effect of void sizes on the the increasing void content. For laminate A and B,
tensile properties, two different void sizes are defined. the longitudinal tensile strength decreases by 2% and
Small voids are defined as those voids with a length 2.5% when the void content goes from 0.4% to 8.0%
lower than 400 μm; while others are regarded as large and 9.0%, respectively (Fig. 4(a)). The longitudi-
voids. The fractions of large void in laminate A are nal tensile modulus of laminate B is not affected by
0, 2.5% and 4.7% at the autoclave pressures 0.4 MPa, void content, while the longitudinal tensile modulus
0.1 MPa and 0.0 MPa, while the fractions of large void of laminate A is sensitive to the void content. It de-
in laminate B are 0%, 1.1% and 1.2%, respectively. creases by 14% when the void content increases from
Commonly, the aspect ratio of the longer voids 0.4% to 8.0% (Fig. 4(b)). This is maybe because the
is larger. So, it is not necessary to classify the void fraction of large voids in laminate A are larger than
shape in addition to the void length because they are that in laminate B. Longer voids may cause prema-
closely related. The difference between the length dis- ture failure to initiate from the voids. Moreover, con-
tribution and the shape distribution of two laminates sidering the coefficient of variation, this percent of
is little. difference in tensile modulus is not considered to be
large.
3.3 Effect of void content on tensile strength and
Figure 5 shows the post-failure mode. Tensile
modulus
failure of the laminates for both laminates A and B
Figure 4 presents the tensile strength and mod- was instantaneous and catastrophic. The failure was
ulus with scatter bars (coefficient of variation). As accompanied by an audible acoustic occurrence and
expected, the tensile strength decreases slightly with cracking sound was heard prior to the catastrophic
 H.Y. Zhu et al.: J. Mater. Sci. Technol., 2011, 27(1), 69–73 73

Fig. 6 Micrographs of tested specimen showing cracks emanating from voids: (a) A, 8.0%; (b) B, 9.0%

failure. When failure occurred, the specimen of B cracks were found emanating from the voids in the
parted into two pieces. The angles between the frac- specimens with the void content of 0.4%.
ture surfaces and horizontal direction were 10–37 deg..
As for the specimen of A, some samples parted into REFERENCES
two pieces, others were not. And the failure was com-
bined with interlaminar cracking. The angles between [1 ] D.E. Stone and B. Clark: NDT, 1975, July, 137.
the fracture surfaces and horizontal direction were 35– [2 ] N.C. W. Judd and W.W. Wright: SAMPE J., 1978,
14, 10.
43 deg.. The void content seems not to affect the
[3 ] H. Jeong: J Compos. Mater., 1997, 31, 276.
failure mode for both laminates significantly. Cracks [4 ] S.F.M. Almeida and A.C.M. Santacreu: Polym.
emanating from the voids in both laminate samples Polym. Compos., 1995, 3, 193.
with the void content of 8.0% (laminate A) and 9.0% [5 ] S.F.M. Almeida and Z.S.N. Neto: Compos. Struct.,
(laminate B) after tensile strength test are depicted in 1994, 28, 139.
Fig. 6. One crack emanates from one void and prop- [6 ] Z. Güerdal, A.P. Tomasino and S.B. Biqqers: SAMPE
agates in the direction of another void. However, no J., 1991, 27, 39.
cracks were found emanating from voids in the spec- [7 ] P.O. Hagstrand, F. Bonjour and J.A.E. Måson: Com-
imens with the void content of 0.4%. That is to say pos. Part A-Appl. S., 2005, 36, 705.
the voids with the length larger than 400 μm promote [8 ] L. Liu, B.M. Zhang, D.F. Wang and Z.J. Wu: Com-
initiation and propagation of the cracks because the pos. Struct., 2006, 73, 303.
[9 ] P. Olivier, J.P. Cottu and B. Ferret: Composites, 1995,
length of voids at the void content of 0.4% is below
26, 509.
400 μm. [10] J.C. Suarez, F. Molleda and A. Güemes: in Proc. 9th
Int. Conf. on Composite Materials, Madrid, Spain,
4. Conclusions
1993, 589.
(1) An experimental program aiming at es- [11] J. Cinquin, V. Triquenaux and Y. Rousne: in Proc.
tablishing the relationship between the voids and 16th Int. Conf. on Composite Materials, Kyoto,
Japan, 2007.
the tensile strength is described. Laminates with
[12] A.R. Chambers, J.S. Earl, C.A. Squires and M.A.
stacking sequence of [(±45)4 /(0,90)/(±45)2 ]S and Suhot: Int. J. Fatigue, 2006, 28, 1389.
[(±45)/04 /(±45)(0,90)]S were produced with inten- [13] M.N. Bureau and J. Denault: Compos. Sci. Technol.,
tionally high void content. The voids were charac- 2004, 64, 1785.
terized for reference samples. The results show that [14] C.W. Dill, S.M. Tipton, E.H. Glaessgen and K.D.
the size and aspect ratio of the voids increase with in- Brasnscum: ASTM STP 1128, 1992, 152.
creasing void content. The voids are typically located [15] P.A. Olivier, B. Mascaro, P. Margueres and F. Collom-
at the ply interface and run along the fibers. bet: in Proc. 16th Int. Conf. on Composite Materials,
(2) The results suggest that both the tensile Kyoto, Japan, 2007.
strengths decrease with increasing void content. The [16] V.I. Rizov: in Proc. 12th Eur. Conf. on Composite
modulus of the [(±45)4 /(0,90)/(±45)2 ]S laminates de- Materials, Biarritz, France, 2006.
[17] J. Varna, R. Joffe, L.A. Berglund and T.S. Lundström:
crease with increasing void content while the modu-
Compos. Sci. Technol., 1995, 53, 241.
lus of the [(±45)/04 /(±45)(0,90)]S laminates is not [18] M. Ricotta, M. Quaresimin and R. Talreja: Compos.
affected by the void content. Cracks emanate from Sci. Technol., 2008, 68, 2616.
the voids in both stacking sequence samples with the [19] M.L. Costa, S.F.M. Almeida and M.C. Rezende: Com-
void content of 8.0% and 9.0% after tensile strength pos. Sci. Technol., 2001, 61, 2101.
test. One crack emanates from one void and propa- [20] M.R. Wisnom, T. Reynolds and N. Gwilliam: Com-
gates in the direction of another void. However, no pos. Sci. Technol., 1996, 56, 93.