Materials and Design 53 (2014) 209216

Contents lists available at SciVerse ScienceDirect

Materials and Design

journal homepage: www.elsevier.com/locate/matdes

Reduction of thermal residual stresses of laminated polymer composites by addition of carbon nanotubes
M.M. Shokrieh , A. Daneshvar, S. Akbari
Composites Research Laboratory, Center of Excellence in Experimental Solid Mechanics and Dynamics, School of Mechanical Engineering, Iran University of Science and Technology, Tehran 16846-13114, Iran

a r t i c l e

i n f o

a b s t r a c t
This paper studies the effects of multi-walled carbon nanotubes (MWCNTs) on the thermal residual stresses in polymeric brous composites. Reinforced ML-506 epoxy nanocomposites with different amounts of homogeneously dispersed MWCNTs (0.1 wt.%, 0.5 wt.% and 1 wt.%) were fabricated using the sonication technique. Thermo-mechanical analysis and tensile tests of the specimens were carried out to characterize the thermal and mechanical properties of MWCNTs/epoxy composites. Due to the negative thermal expansion and high modulus of MWCNTs, addition of MWCNTs resulted in a great reduction of the coefcient of thermal expansion (CTE) of epoxy. The MWCNTs also moderately increased the Youngs modulus of the epoxy. Then, the effects of adding MWCNTs on micro and macro-residual stresses in carbon ber (CF)/epoxy laminated composites were investigated using the energy method and the classical lamination theory (CLT), respectively. The results indicated that the addition of low amounts of MWCNTs leads to a considerable reduction in thermal residual stress components in both micro and macro levels. 2013 Elsevier Ltd. All rights reserved.

Article history: Received 30 December 2012 Accepted 2 July 2013 Available online 16 July 2013 Keywords: Polymer composites Thermal residual stresses Multi-walled Carbon nanotubes Coefcient of thermal expansion Youngs modulus

1. Introduction In laminated polymer composites, due to their inherent inhomogeneous nature, residual stresses are generated during the curing process. These stresses adversely inuence dimensional stability and strength of composite laminates and could result in premature failure, delamination, warpage and matrix cracking [13]. Thus, it is important to develop techniques to reduce these stresses. The most approaches employed so far for this purpose are conned to modifying curing cycles [48]. Residual stresses in laminated composites, based on their resources, are studied from two points of view; micro residual stresses in a unidirectional ply and macro residual stresses in laminated composites. Micro and macro residual stresses originate from independent sources. The residual stress in micro-scale is mainly the result of the mismatch in coefcient of thermal expansion (CTE) and Youngs modulus between the bers and the matrix, while at the macro-mechanical level, expansion and contraction of different layers with different orientations is the major source of residual stresses. Therefore, residual stresses in each layer of laminated polymer composites can be determined separately on the microand macro-scales.

Corresponding author. Tel./fax: +98 21 7749 1206.

E-mail address: Shokrieh@iust.ac.ir (M.M. Shokrieh). 0261-3069/$ - see front matter 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.matdes.2013.07.007

In recent years, an increasing number of materials with negative thermal expansion have been discovered. One of the most important applications of the negative thermal expansion materials is to compensate for undesirable effects of high thermal expansion of other materials. Regarding this fact, addition of nanoadditives with negative CTE to polymer matrix is a potential approach to effectively reduce thermal residual stresses in ber-reinforced polymer composites. The thermal residual stresses are mainly functions of CTE and Youngs modulus of the composite constituents. Compared with polymer matrix, nano-additives like CNTs and CNFs have much higher Youngs modulus and much lower CTE. Thus, these nano-additives could be dispersed into the polymer to modify its thermal and elastic behavior. Numerous experimental studies demonstrated the great capability of CNTs for the modication of thermal and mechanical properties of the polymer matrix [912]. Compared with two-phase composites (nano-additive and matrix), less studies have been carried out on the thermal and mechanical properties of three-phase composites (nano-additive, bers and matrix). Some researchers reported signicant reduction of CTE of brous composites due to addition of small amounts of nano-additives [13,14]. On the other hand, very little research has so far been conducted on the nano-additives effects on thermal residual stresses of composites. Nishino et al. [15] studied the effect of tungstate zirconium phosphate (ZWP) particles on CTE of polyether ether ketone (PEEK) polymers. They experimentally showed that by

210

M.M. Shokrieh et al. / Materials and Design 53 (2014) 209216

incorporating 40 vol.% ZWP particles with negative CTE, the CTE of the ZWP/PEEK composite reduces about 62% and becomes identical to the CTE of aluminum. This reduction in CTE was found to be effective for the decrease of the residual stress at the interface between aluminum plate and the ZWP/PEEK composites. Also, Hsiao and Gangireddy [16] showed that the use of CNFs results in the reduction of the spring-in phenomenon in L-shaped glass/polyester composite parts. Badrinarayanan et al. [17] could remarkably reduce the warpage of un-symmetric carbon ber-reinforced composite laminates by using zirconium tungstate (ZrW2O8) nanoparticles. Although these studies show that the enhancement of dimensional stability in composite components could be accomplished by using nano-additives, but no previous work, to the best of the authors knowledge, has quantitatively studied the effect of nano-additives on the micro and macro residual stresses distribution in the laminated polymer composites. This paper presents a study on MWCNTs effects on thermal residual stresses in polymeric brous composites. For this purpose, the CLT and the energy method are coupled with the experimental characterization results to study MWCNTs effects on the thermomechanical properties of MWCNTs/epoxy and MWCNTs/CF/epoxy nanocomposites as well as, ultimately, on the micro- and macroresidual stress components. A signicant trend of reduction in both micro and macro-residual stresses was found when the MWCNTs weight fraction increased. These ndings conrm that MWCNTs possess excellent potential to be used as a thermal expansion compensator for the modication of the thermal behavior of the epoxy and to reduce the thermal residual stresses of the ber/epoxy laminated composites. 2. Experimental procedure In this study, two-phase MWCNTs/epoxy composites and threephase MWCNTs/CF/epoxy composites were fabricated and characterized. The MWCNTs were dispersed into the epoxy matrix with three weight ratios of 0.1%, 0.5% and 1% which were then reinforced with carbon bers. A detailed description of these experiments is presented in the following sections. 2.1. Raw materials and nanocomposites fabrication The MWCNT (purity >95%) in this study was provided by Neutrino Company and had an average diameter of 815 nm, a length of 0.52 lm and a density of 1.65 g/cc. The diglycidyl ether of Bisphenol-B epoxy resin, ML-506 (Mokarrar Engineering Materials, Iran) and the curing agent, Aradur-830 (Huntsman, Germany) were used at 100:60 ratio for fabrication of nanocomposites. The ML-506 epoxy was mixed with Aradur-830 hardener thoroughly. The mixture was stirred for 30 min at 250 rpm. Then, the mixture was placed under a vacuum chamber for 30 min to remove air bubbles. The MWCNTs/epoxy nanocomposite specimens reinforced with three different contents of MWCNTs were fabricated with the aid of the sonication technique. First, ML-506 epoxy were premixed with MWCNTs by shaking and stirred for 30 min at 2000 rpm. Then, in order to break the residual aggregates and obtain a homogeneous dispersed mixture of epoxy resin and MWCNTs, the mixtures containing 0.1, 0.5 and 1 wt.% MWCNTs were sonicated (Hielscher UP400S, Teltow, Germany) at 200 W with a probe of 14 mm diameter for 40, 60 and 80 min, respectively [18]. During the sonication, the mixture container cooled in an ice-bath. Once the sonication was completed, the curing agent was added to the mixture and stirred for 20 min at 250 rpm. Then, air bubbles were removed by degassing the solution in a vacuum chamber for 15 min. Finally, the bubble free mixtures of MWCNTs/epoxy and neat resin were cast into steel molds

and cured for 6 h at 100 C followed by 6 h at 120 C. A moldreleasing agent was added to the mold surface to allow an easy release of the cured specimens. The cured specimens were then allowed to cool slowly. Three-phase MWCNTs/CF/epoxy composites were manufactured using the hand lay-up method. The MWCNTs/epoxy composite was used in the manufacturing of MWCNTs/CF/epoxy composites. Subsequently, the curing agent was added to MWCNTs/epoxy mixture and the multi-phase composite was fabricated by the hand lay-up method. In addition, the same cure process as the two-phase composites was employed for the threephase composites. A roller was used to remove the air entrapped during the hand lay-up process and to uniformly distribute the resin between all layers. The volume fraction of the carbon bers in nal laminates was approximately 45% for all composite specimens, measured by burn-off tests. 2.2. Characterization of nanocomposites 2.2.1. Youngs modulus characterization of nanocomposites The Youngs modulus of the fabricated MWCNTs/epoxy specimens was obtained using tensile tests, performed according to ASTM: D638-10. In order to perform the tensile test, a universal testing apparatus (STM-150, Santam, Iran) with a 50 kN load cell was used. The machine was run under the displacement control mode at a crosshead speed of 1.0 mm/min. An extensometer of 50 mm gauge length was used for strain measurement. Both ends of each specimen were clamped by the pneumatics grips of the testing machine with an inter-grip distance of 40.0 mm. Prior to the tensile test, all samples were mechanically polished to minimize the inuence of surface aws, especially the porosity. At least ve samples were tested for each MWCNTs contents; the nal property was the average result of the ve tests. Prepared samples for tensile tests are shown in Fig. 1. In the macro residual stress analysis by the CLT, the elastic constants of the unidirectional MWCNTs/CF/epoxy composites are required. Static strength tests in the longitudinal and transverse directions were performed for the unidirectional MWCNTs/CF/ epoxy laminates as three-phase composites, according to ASTM: D3039 M-08. The dimensions of 0 unidirectional specimens were about 250 mm in length, 15 mm in width and 1.2 mm in thickness. The dimensions of the 90 unidirectional specimens were about 175 mm in length, 25 mm in width and 2.1 mm in thickness. For each weight fraction of MWCNTs, ve specimens were tested. 2.2.2. CTE characterization of nanocomposites In order to investigate the MWCNTs effects on the CTE of composites, the CTE of the neat resin and, MWCNTs/epoxy specimens was measured by a thermo-mechanical analyzer (TMA). The CTE values were determined by measuring the inclination of the thermal strains vs the temperature according to ASTM: E831-12:

Fig. 1. (a) MWCNTs/epoxy, (b) Neat epoxy, (c) longitudinal MWCNTs/CF/epoxy and (d) transverse MWCNTs/CF/epoxy specimens prepared for the tensile test.

M.M. Shokrieh et al. / Materials and Design 53 (2014) 209216

211

a DL=L0 DT

szh srh 0

where, DL is the change in the length of the composite specimen due to heating, L0 is the initial length of test specimen and DT is the temperature difference, over which the change in the specimen length is determined. The TMA was carried out using a thermomechanical analyzer (TMA-120, Seiko Instruments, Tokyo, Japan). The tested specimens had dimensions of 7 7 6 mm. The test was run up to 140 C at a constant heating rate of approximately 0.1C/s. The effect of the weight fraction of MWCNTs was assessed by measuring the CTE of MWCNTs/epoxy composites at three different weight fractions of 0.1%, 0.5%, and 1%. 3. Residual stresses 3.1. Energy method Residual stresses may lead to the development of naturally local damages like matrix cracking. Thus, a micromechanical analysis is needed to study the effects of residual stresses in micro-level. For this purpose, a single bermatrix composite including the interphase between them is considered as the representative volume element (RVE). These three phases are modeled as concentric circular cylinders, as shown in Fig. 2. Using the energy method, a three-dimensional analytical solution for micro-residual stresses is obtained. The problem is axisymmetric and cylindrical coordinates (r, h, z) are selected with the z-axis corresponding to the axis of the ber. The radii of ber, matrix, and interphase are Rf, Rm, and Ri, respectively. At the origin, there is a radial plane of symmetry perpendicular to the ber axis. The boundaries are traction-free and the composite is subjected to a temperature decrease during the curing process. Due to the bigger coefcient of are thermal expansion of matrix in comparison with the ber, the matrix axial residual stress is tensile, whereas the ber is under compression. The ber, matrix and interphase are modeled as linearly elastic, isotropic materials. The stresses in the present problem can be described by using appropriate Airy stress functions, which satisfy not only the equilibrium conditions, but also the compatibility relationships. For an axisymmetric problem, the stress components in terms of the Airy stress functions u are dened as:

Because of the axi-symmetry, Airy stress functions can be assumed as [20].

uj r; z fn r:g n z

n 1; 2; 3 ;

j f ; m; i

The subscripts j = f, m, i refer to the ber, matrix and interphase, respectively. Also, the functions fn(r) and gn(z) are functions of radial and axial coordinates, respectively. In order to determine the stress components in the three phases, the unknown functions fn(r) and gn(z) must be initially obtained. Because of the stress continuity at the berinterphase and interphasematrix interfaces along radial direction, the number of unknown functions decreases. Due to symmetry, the axial stresses in composite constituents are maximum and shear stresses are zero at the plane of symmetry. Also, the stress-free conditions results in zero axial and shear stresses at the end of the ber. Using the total complementary energy principle, the remaining unknown functions are determined and nally, the stress components can be obtained in composite constituents. The details of these equations are presented in [19]. Because of the complexity of the equations, the mathematical software, Maple 12 [21] was used. 3.2. Classical lamination theory The CTE of a unidirectional ply along the longitudinal direction is much lower than its CTE in the transverse direction. This is because bers have much lower CTE in comparison with the matrix and matrix has more contraction during the curing. Therefore, a compressive stress along the longitudinal direction and a tensile stress along the transverse direction are generated. Fig. 3 shows the macro-residual stress formation during curing process of a [0/90]s laminate. Residual stresses at the macroscopic level could be estimated by the classical lamination theory (CLT) [22]. In this theory, each ply is simplied to a homogeneous orthotropic thin plate. Therefore, using CLT it is not possible to obtain the local micro-residual stress eld, for example the stress eld at the interface of the ber and matrix. A brief review of the CLT formulation for the calculation of macro-residual stresses in laminated composites is presented below. In order to describe macro-residual stress formation in a general form, the constitutive relation for each ply should be considered.
 k rR Q k e zk k ak DT
k

rrr rhh rzz

1 @u @2u 2 r @r @z @2u @ z2 @2u 1 @u r @r @ r2 @2u @ r@ z

srz

where rR is the vector of residual stresses and k is the layer number. Q k is the stiffness matrix in off-axis coordinate system. e and  k are the strain and curvature of the mid-plane, respectively. zk is the distance to the mid-plane. DT is the temperature difference between the ambient and the stress-free temperatures. Moreover, ak is the vector of coefcients of thermal expansion in off-axis coordinate system.

Fig. 2. Three-phase composite model [19].

212

M.M. Shokrieh et al. / Materials and Design 53 (2014) 209216

(a)

T x
(b)

T y
R xx
R yy

(c)

Fig. 3. Macro-residual stress formation in a [0/90]s laminate. (a) Initial stress free state, (b) hypothetical unconstrained nal state and (c) actual nal state.

Fig. 4. Youngs modulus comparisons for MWCNTs/epoxy composites with three different weight fractions of MWCNTs.

The strain and the curvature of the mid-plane are calculated as follows:

e
k


A B

B D

1 "

N T M T

# 9

A, B, and D are the extensional, coupling, and bending stiffness matrices, respectively. The thermal forces and the moment resulT T T T T tants are dened as N T N T M T x ; N y ; N xy and M x ; M y ; M xy ; respectively. For a laminate, the thermal loads are dened as follows:
T N T 1 ; M1 N X k k k k k Q k 11 :a1 Q 12 :a2 Q 16 :a6 D T :t k ; t k zk k 1 N X k k k k k Q k 21 :a1 Q 22 :a2 Q 26 :a6 D T :t k ; t k zk k 1 N X k k k k k Q k 61 :a1 Q 26 :a2 Q 66 :a6 D T :t k ; t k zk k 1

10

T N T 2 ; M2

11

The specimens were coated by a thin layer of gold prior to the observation. Figs. 5(ad) depict the dispersion state of MWCNTs in the neat epoxy and MWCNTs/epoxy composites contained 0.1, 0.5 and 1 wt.% MWCNTs, respectively. The fracture surface of the neat epoxy is smooth. However, by addition of MWCNTs, the fracture surface roughness increases and the MWCNTs broken stem can be seen. Figs. 5(bd) indicate that MWCNTs was uniformly dispersed in the matrix resin blend. Thus it can be assumed that MWCNTs played its role as a nano-additive. Fig. 6 represents the CTE values at the room temperature for the MWCNTs/epoxy composite specimens at three different weight fractions as well as the neat epoxy. It is observed that the CTE values of all composites signicantly decreases by increasing the weight fraction of the MWCNTs, which is readily expected because the MWCNTs have a much lower CTE than that of the neat epoxy. For composites with MWCNTs of weight fractions of 0.1%, 0.5%, and 1%, the CTE decreased by 6%, 21%, and 23.7% respectively. Results of characterization tests showed that adding MWCNTs to the epoxy has a negligible impact on the matrix Youngs modulus. On the contrary, MWCNTs have a signicant effect on the matrix CTE.

T N T 6 ; M6

12
4.2. Youngs modulus and CTE of three-phase MWCNT/CF/epoxy nanocomposites The stressstrain behaviors for the three-phase composites were almost linear. The results showed that the longitudinal Youngs modulus is not affected seriously by the MWCNTs dispersion. On the contrary, the transverse Youngs modulus is somewhat affected by MWCNTs dispersion. Three-phase composites with 0.1, 0.5 and 1 wt.% MWCNTs showed an increase in the transverse modulus by about 2.6%, 5.9%, and 9%, respectively. Table 1 shows the averaged results of the Youngs modulus and the CTE for all types of MWCNTs/CF/epoxy composites. The inconsistent behavior of the Youngs modulus in the longitudinal and transverse directions is due to the role of the CFs and matrix. In the longitudinal direction, the Youngs modulus of composite depends essentially on the stiffness of bers which carry the load, while in the transverse direction the bers have actually no reinforcing effect and the modulus value depends on the matrix stiffness. In such a case, by addition of MWCNTs, it was shown that the Youngs modulus of MWCNTs/epoxy composites increased. The changes in the transverse Youngs modulus of MWCNTs/CF/epoxy composites are very close to those of the MWCNTs/epoxy composites.

where N is the total number of layers.tk and zk are the ply thickness and distance to the reference plane, respectively. A subroutine was written in the commercial matrix software MATLAB [23] for the implementation of the residual stress calculation procedure described above. 4. Results and discussion 4.1. Youngs modulus and CTE of two-phase MWCNTs/epoxy nanocomposites Fig. 4 shows the averaged tensile test results for all three types of composites as well as the neat resin. The Youngs modulus for each specimen was extracted from the average of the slope of the stress vs strain curves in the linear region. It is observed in Fig. 4 that the dispersion of MWCNTs into the epoxy has a moderate effect on the Youngs modulus of the two-phase composites. At 1% loading, the Youngs modulus of MWCNTs/epoxy composites was 9% larger than that of the neat resin. To investigate the dispersion state of MWCNTs in the matrix, scanning electron microscopy (SEM) was performed using Merlin eld emission electron microscope (Carl Zeiss, Jena, Germany).

M.M. Shokrieh et al. / Materials and Design 53 (2014) 209216

213

Fig. 5. Fracture surface of (a) neat epoxy (b) 0.1 wt.% (c) 0.5 wt.% and (d) 1 wt.% MWCNTs.

4.3. Micro-residual stresses analysis In this section, the energy method is used to calculate microresidual stresses of MWCNTs/CF/epoxy composite. Investigation of micro-residual stress distribution in a single ber/MWCNT/ epoxy nanocomposite requires the variation of nanocomposite constituents properties (e.g., the Youngs modulus and CTE of ber, matrix and interphase). Since nanollers are randomly dispersed in the epoxy, it would be reasonableto assume the MWCNTs/epoxy as an isotropic matrix [24,25]. Properties of the matrix with different MWCNTs weight loading were experimentally characterized in the previous section. Thermal and mechanical properties of the interphase are different from the ber and matrix properties, but depend on them. Consequently, before performing a micromechanical analysis it is necessary to predict the interphase characteristics in terms of ber and matrix properties. Papanicolaou et al. [26] presented a detailed two-dimensional model for the interphase in ber reinforced composite materials to study their micromechanical behavior under thermal loading. Based on this hybrid interphase model, a generalized exponential variation for thermal and mechanical properties in terms of material inhomogeneity, ber volume fraction, and imperfect adhesion is ob-

Fig. 6. CTE comparisons for MWCNTs/epoxy composites with three different weight fractions of MWCNTs.

Table 1 Youngs modulus and CTE of three-phase MWCNT/CF/epoxy nanocomposites with different nanoller contents. 0 wt.% Longitudinal Youngs modulus of MWCNT/CF/epoxy (GPa) Transverse Youngs modulus of MWCNT/CF/epoxy (GPa) Longitudinal CTE of MWCNT/CF/epoxy (1/C) Transverse CTE of MWCNT/CF/epoxy (1/C) 104.1 7.43 0.912 106 40.98 106 0.1 wt.% 104.8 7.63 0.933 106 38.96 106 0.5 wt.% 105.2 7.84 1.028 106 34.29 106 1 wt.% 105.3 8.1 1.053 106 31.44 106

214

M.M. Shokrieh et al. / Materials and Design 53 (2014) 209216

tained. Assuming the interphase bounding efciency (ki = 0.2), the Youngs modulus and CTE of the interphase were predicted using Papanicolaous model. The results are shown in Table. 2. As shown in Table. 2, increasing MWCNTs weight loading, properties of carbon ber remain constant but those of the matrix and the interphase change accordingly. Also, the temperature reduction was assumed to be 100 C. All material properties were considered to be constant throughout the applied temperature range. In micromechanical analysis, the radius of ber, interphase and matrix (Rf, Ri and Rm) was assumed to be 6, 6.5 and 8.5 microns respectively. The axial residual stresses are symmetric about the composite mid-plane. Therefore, these stresses are presented for half length of the composite. Figs. 7 and 8 compare the axial stress distribution in the ber and matrix for different weight fractions of the MWCNTs. As observed in these gures, the matrix axial stress is tensile, while the ber has compressive axial stress. The peak of axial stresses occurs at the composite mid-plane. Also, they reach the zero magnitude at both ber ends. Increasing the MWCNTs content leads to decrease in the ber and matrix axial stresses. Comparing with the CF/epoxy composite, addition of 1 wt.% MWCNT leads to about 18% decrease in maximum value of the ber and matrix axial stresses. The berinterphase interfacial shear stress, shown in Fig. 9, reaches the maximum magnitude in the vicinity of the ber ends. An increase in the MWCNTs content causes the berinterphase interfacial shear stress to decrease. For the 1 wt.% MWCNTs, the maximum berinterphase shear stress decreases about 21%. The interphasematrix interfacial shear stress shows a similar trend to berinterphase shear stress. The peak shear stress is of critical importance, since ber debonding may occur, if the maximum shear stress exceeds the interfacial shear strength.

Fig. 7. Fiber axial residual stress distribution for different contents of MWCNTs.

4.4. Macro-residual stresses analysis Similar to the comparisons made for micro-residual stresses, here macro-residual stresses are analyzed based on the CLT. The results are shown for symmetric cross-ply and un-symmetric angle-ply laminates in Figs. 10 and 11. With increase in the MWCNTs content, a systematic decrease in the normal and shear macroresidual stresses is seen for all laminate congurations. This can be explained by the reduction of the difference between the longitudinal and transverse properties of a unidirectional MWCNTs/CF/ epoxy ply. In [0/90]s laminated composites, shear stress component in each ply is zero and 1 wt.% MWCNTs leads to 16% decrease in normal residual stress in each ply, as shown in Fig. 10. In unsymmetric [452/452] laminated composites, 1 wt.% MWCNTs leads to 18% decrease in shear residual stress in 45 plies (Fig. 11). The stress results demonstrate that, in general, the addition of MWCNTs has a broadly similar effect on the stress in different plies. The results indicate that the reduction of the CTE mismatch between the polymer matrix and ber reinforcements could effectively reduce the thermal residual stresses in

Fig. 8. Matrix axial residual stress distribution for different contents of MWCNTs.

laminated polymer composites, and increase the dimensional stability of the composite specimen. The CLT predicts that un-symmetric laminates have a saddle shape at the room-temperature [27]. The experimental observation conrms this prediction, Fig. 12. This gure indicates that a neat [02/902] CF/epoxy laminate is warped or curved after cooling down from the curing temperature to the ambient temperature. The buildup of thermal residual stresses is the main reason of this phenomenon. The effect of MWCNTs loading on the shape of the rectangular composite laminates with un-symmetric layup after curing is shown in Fig. 13. As shown in this gure, in presence of 1 wt.% MWCNTs in the matrix, the warpage of the un-symmetric laminate

Table 2 Properties of single ber/MWCNTs/epoxy nanocomposites constituents with different nanoller weight loadings. 0 wt.% Youngs modulus of CF (GPa) Longitudinal CTE of CF (1/C) Transverse CTE of CF (1/C) Youngs modulus of matrix (GPa) CTE of matrix (1/C) Youngs modulus of interphase (GPa) CTE of interphase (1/C) 230 0.41 106 15 106 3.13 62.46 106 9.39 46.85 106 0.1 wt.% 230 0.41 106 15 106 3.23 58.72 106 9.69 44.04 106 0.5 wt.% 230 0.41 106 15 106 3.3 49.33 106 9.9 37 106 1 wt.% 230 0.41 106 15 106 3.42 47.65 106 10.26 35.74 106

M.M. Shokrieh et al. / Materials and Design 53 (2014) 209216

215

Fig. 12. Room-temperature shape of CF/epoxy laminate with un-symmetric [02/ 902] layup.

Fig. 9. Fiberinterphase interfacial shear stress distribution for different contents of MWCNTs.

Fig. 13. Effect of MWCNTs loading on the warpage and curvature of MWCNTs/CF/ epoxy laminates with un-symmetric [02/902] layup.

Fig. 10. Normal residual stress vs MWCNT in each ply of [0/90]s MWCNTs/CF/epoxy nanocomposites.

Fig. 11. Shear residual stress vs MWCNT in each ply of ply of [452/452] MWCNTs/ CF/epoxy nanocomposites.

decreases noticeably. This nding is consistent with the earlier reports of the enhanced dimensional stability of the composite com-

ponents reinforced with nano-additives with negative thermal expansivity [1517]. It is expected that the introduction of the higher weight fractions of MWCNTs into matrix will lead to more reduction in the residual stresses. However, attaining acceptable quality of the dispersion of the MWCNTs in higher weight fractions could create serious practical challenges. This is because when the content of MWCNTs reaches a certain limit, the possibility of MWCNTs aggregation signicantly increases, and the MWCNTs become in homogeneous in the matrix. The present study clearly explain how the decrease of thermal residual stresses in polymer composites is achieved through the decrease of the matrix CTE and increase of matrix Youngs modulus, which accordingly results in the decrease of the transverse CTE of the unidirectional ply and increase of its Youngs modulus. Prior to the use of nano-additives for the residual stress decrease in polymer composites, it is very important to ensure that the addition of the selected nano-additives results in an enhancement of the thermal and mechanical properties of the polymer. This is because the incorporation of nano-additives does not always improve the properties of polymer composites. For example, Yokozeki et al. [28] reported that the addition of the cup-stacked CNTs has no considerable effect on the CTE and Youngs modulus of CF/epoxy prepregs. Therefore, the optimization of thermal residual stresses signicantly depends on a clear understanding of the respective role played by the type of the matrix and nano-additive as well as the interaction between the matrix and nano-additive. The transfer of the excellent thermal and mechanical properties of nano-additives into weaker polymer matrix and the subsequent transforming into ber-reinforced composite is still a practical challenge. The effective use of nano-additives in composite applications extremely depends on the homogeneous dispersion of

216

M.M. Shokrieh et al. / Materials and Design 53 (2014) 209216 [4] Hodges J, Yates B, Darby MI, Wostenholm GH, Clement JF, Keates TF. Residual stresses and the optimum cure cycle for an epoxy resin. J Mater Sci 1989;24:198490. [5] White SR, Hahn HT. Cure cycle optimization for the reduction of processinginduced residual stresses in composite materials. J Compos Mater 1993;27:135278. [6] Kim JW, Lee JH, Kim HG, Kim HS, Lee DG. Reduction of residual stresses in thick-walled composite cylinders by smart cure cycle with cooling and reheating. Compos Struct 2006;75:2616. [7] Kim SS, Murayama H, Kageyama K, Uzawa K, Kanai M. Studyon the curing process for carbon/epoxy composites to reduce thermal residual stress. Compos Part A: Appl Sci Manuf 2012;43(8):1197202. [8] Kim HS, Yoo SH, Chang SH. In situ monitoring of the strain evolution and curing reaction of composite laminates to reduce the thermal residual stress using FBG sensor and dielectrometry. Compos Part B: Eng 2013;44(1):44652. [9] Qian D, Dickey EC, Andrews R, Rantell T. Load transfer and deformation mechanisms in carbon nanotube-polystyrene composites. Appl Phys Lett 2000;76(20):286870. [10] Gojny FH, Wichmann MHG, Kpke U, Fiedler B, Schulte K. Carbon nanotubereinforced epoxy-composites: enhanced stiffness and fracture toughness at low nanotube content. Compos Sci Technol 2004;64:236371. [11] Yeh MK, Tai NH, Lin YJ. Mechanical properties of phenolic-based nanocomposites reinforced by multi-walled carbon nanotubes and carbon bers. Compos Part A: Appl Sci Manuf 2008;39(4):67784. [12] Wang S, Liang Z, Gonnet P, Liao YH, Wang B, Zhang C. Effect of nanotube functionalization on the coefcient of thermal expansion of nanocomposites. Adv Func Mater 2007;17(1):8792. [13] Godara A, Mezzo L, Luizi F, Warrier A, Lomov SV, Van Vuure AW, et al. Inuence of carbon nanotube reinforcement on the processing and the mechanical behaviour of carbon ber/epoxy composites. Carbon 2009;47(12):291423. [14] Green KJ, Dean DR, Vaidya UK, Nyairo E. Multiscale ber reinforced composites based on a carbon nanober/epoxynanophased polymer matrix: synthesis, mechanical, and thermomechanical behavior. Compos Part A: Appl Sci Manuf 2009;40(9):14705. [15] Nishino T, Kotera M, Sugiura Y. Residual stress of particulate polymer composites with reduced thermal expansion. J Phys: Conf Ser 2009;184:12026. [16] Hsiao KT, Gangireddy S. Investigation on the spring-in phenomenon of carbon nanoberglass ber/polyester composites manufactured with vacuum assisted resin transfer molding. Compos Part A: Appl. Sci. Manuf. 2008;39:83442. [17] Badrinarayanan P, Rogalski MK, Kessler MR. Carbon ber-reinforced cyanate ester/nano-ZrW2O8 composites with tailored thermal expansion. ACS Appl Mater Interfaces 2012;4(2):5107. [18] Chitsazzadeh M, Shahverdi H, Shokrieh MM. Fabrication of multi-walled carbon nanotube/vinyl ester nanocomposites: dispersion and stabilization. Def Diff Forum. 2011;312315:4605. [19] Shokrieh MM, Safarabadi M. Three-dimensional analysis of micro-residual stresses in brous composites based on the energy method: a study including interphase effects. J Compos Mater 2012;46(6):72735. [20] Quek MY. Analysis of residual stresses in a single bermatrix composite. Int J Adhes Adhes 2004;24:37988. [21] Maple 12, copyright 19812008 by Waterloo Maple Inc. [22] Tsai SW, Hahn HT. Introduction to Composite Materials. 1980. [23] MATLAB Release 14, The Math Works, Natick, MA. 2004. [24] Na X, Qingjie J, Chongguang Z, Chenglong W, Yuanyuan L. Study on dispersion and electrical property of multi-walled carbonnanotubes/low-density polyethylene nanocomposites. J Mater Des 2010;31:167683. [25] Montazeri A. The effect of functionalization on the viscoelastic behavior of multi-wall carbon nanotube/epoxy composites. J Mater Des 2013;45:5107. [26] Papanicolaou GC, Michalopoulou MV, Anifantis NK. Thermal stresses in brous composites incorporating hybrid interphase regions. Compos Sci Technol 2002;62:188194. [27] Hyer MW. Calculation of the room-temperature shapes of unsymmetric laminates. J Compos Mater 1981;15:296310. [28] Yokozeki T, Iwahori Y, Ishiwata S. Matrix cracking behaviors in carbon ber/ epoxy laminates lled with cup-stacked carbon nanotubes (CSCNTs). Compos Part A: Appl Sci Manuf 2007;38(3):91724. [29] Ren Y, Li F, Cheng HM, Lio K. Tension fatigue behavior of unidirectional single walled carbon nanotube reinforced epoxy composite. Carbon 2003;41(11):21779. [30] Breton Y, Desarmot G, Saivetat JP. Mechanical properties of multi-walled carbon nanotube/epoxy composite: inuence of network morphology. Carbon 2004;42:102730. [31] Shokrieh MM, Akbari S, Daneshvar A. A comparison between the slitting method and the classical lamination theory in determination of macroresidual stresses in laminated composites. Compos Struct 2013;96:70815.

nano-additives throughout the matrix without destroying their integrity. On the other hand, in order to achieve load transfer across the nano-additive-matrix interface, good interfacial bonding is a necessary condition [29,30]. The accuracy of the stress results presented in this study could be improved by considering more realistic material properties, such as viscoelastic behavior and chemical shrinkage of resin. Moreover, possible MWCNTs existence on the interphase region and also reaction of MWCNTs particles with the polymer chain were ignored in the energy model. Such phenomena may lead to presence of the residual stress. Therefore, considering them needs further studies. In general, experimental methods such as slitting and incremental hole-drilling can give a more realistic estimation of the macro-residual stresses [31]. In the next phase of this research, the effects of different amounts of MWCNTs on the residual stresses of the laminated composites will be experimentally studied by the slitting method. 5. Conclusion The use of a stiff and low CTE nano-additive as a ller for the reinforcement of the compliant and high CTE matrix is a new approach to decrease thermal residual stresses in ber-reinforced polymer composites. For this purpose, the CNT is a good candidate, because it has much higher Youngs modulus and much lower CTE compared with conventional polymers, and it is expected that MWCNTs could reduce the residual stresses in laminated polymer composites. First, the overall thermo-elastic properties of two-phase MWCNTs-reinforced epoxy and three-phase MWCNTs-reinforced carbon ber/epoxy composites were measured by performing the tensile tests and TMA analysis. As expected, the results indicate that addition of MWCNTs to epoxy makes its properties closer to the ber ones. The results also indicated that the properties of a unidirectional ply in longitudinal direction are ber-dominant. As a result, addition of MWCNTs has negligible effect on the longitudinal properties. On the other hand, the transverse properties are matrix-dominant and addition of MWCNTs can considerably reduce the transverse CTE and increase transverse Youngs modulus of the unidirectional ply. Consequently, the difference between the longitudinal and transverse properties decreases. In the nal step, in order to quantify the nano-additives effects on thermal residual stresses, the classical lamination theory and the energy method were used to calculate macro and micro thermal residual stresses in carbon ber-reinforced composite containing low weight contents of MWCNTs. The results indicate that the incorporation of small amounts of MWCNTs leads to a signicant reduction in residual stress in the carbon/epoxy composites in both micro and macro-levels. The results of this study show that the MWCNT has excellent potential as a ller for development of polymer composites with low thermal residual stresses. References
[1] Gascoigne HE. Residual surface stresses in laminated cross-ply ber-epoxy composite materials. Exp Mech 1994;34:2736. [2] Hosseini-Toudeshky H, Mohammadi B. Thermal residual stresses effects on fatigue crack growth of repaired panels bounded with various composite materials. Compos Struct 2009;89:21623. [3] Ersoy N, Vardar O. Measurement of residual stresses in layered composites by compliance method. J Compos Mater 2000;34:57598.