Viscoelasticity behaviors of lightly cured natural ... - Wiley Online Library

4 downloads 24776 Views 3MB Size Report
School of Mechanical and Automotive Engineering, South China University of Technology,. Guangzhou 510640, China. 3. College of Material Science and ...
Viscoelasticity Behaviors of Lightly Cured Natural Rubber/Zinc Dimethacrylate Composites

Yukun Chen,1,2 Chuanhui Xu,3 Yanpeng Wang1,2 1 The Key Laboratory of Polymer Processing Engineering, Ministry of Education, China (South China University of Technology) Guangzhou 510640, China 2

School of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou 510640, China

3

College of Material Science and Engineering, South China University of Technology, Guangzhou 510640, China

Zinc dimethacrylate (ZDMA) can be polymerized during peroxide curing to form the polymerized ZDMA (PZDMA) at nanoscales. At the same time, the covalent crosslink of the rubber matrix and ionic crosslink introduced by the graft-PZDMA also are formed. The structure evolution of this type of composites is complex. In this article, the dynamic viscoelasticity characteristics of lightly cured ZDMA/natural rubber (NR) composites were investigated using a Rubber Process Analyzer 2000 (RPA2000). Our goal was to study the internal structures of this type of composites in an early curing stage. The dynamical viscoelasticity of the composites cured for 1 min was focused. The results of RPA2000 indicated that the PZDMA could act as particles to form a strong filler–filler structure which resulted in apparent Payne effect. A ‘ primary network’’ structure might be formed which contained covalent crosslink points, ionic crosslinks, physical adsorption, and PZDMA. The stress-softening behavior was also investigated. At last, the scanning electron microscope analysis verified that most of the ZDMA had been polymerized to form PZDMA ‘ nanoparticle’’ when the composites were cured for 1 min. POLYM. COMPOS., 33:967–975, 2012. ª 2012 Society of Plastics Engineers

INTRODUCTION Dispersing solid fillers into rubbers is a very efficient way of improving the mechanical properties of rubbers [1, 2]. It is well known that, when the conventional reinforcing filler such as silica or carbon black particles is incorporated into rubber, the filler aggregates have a tendCorrespondence to: Yukun Chen; e-mail: [email protected] DOI 10.1002/pc.22189 Published online in Wiley Online Library (wileyonlinelibrary.com). C 2012 Society of Plastics Engineers V

POLYMER COMPOSITES—-2012

ency to congregate due to their high specific surface area and high surface energy [3]. Thus a secondary, physical network can be formed. These filled rubbers show a typical nonlinear viscoelastic behavior. At low strains of about 1%, a significant decrease of storage modulus occurs [4]. This phenomenon is defined as the well-known Payne effect [5–8], which is interpreted as the result of breakage and reforming of physical bonds between the filler aggregates. After curing, the chemical network is formed by covalent crosslinks in the vulcanizate. These two networks coexist in vulcanizates and interact, which contribute to the mechanical properties of vulcanizates. In recent years, metal salts of unsaturated carboxylic acids are used as new reinforcing agents for rubbers [9– 16]. When peroxides are used as curing agents, the metal salts of unsaturated carboxylic acids will be polymerized to form ionic crosslinks by the graft copolymerization [17, 18]. Obviously, the ionic crosslinks lead to formation of a complex internal crosslink structure [9]. A variety of studies [9–15] based on this kind of composites have revealed that different kinds of rubbers can be reinforced by zinc dimethacrylate (ZDMA), though reinforcing effect varies from the types of rubber matrix. Many researchers have observed the nanoscale polymerized ZDMA (PZDMA) in ZDMA/rubber composites by transmission electron microscopy (TEM) technology [9, 13, 16, 19]. Although many studies have been carried out on ZDMA/ rubber systems, there are questions still remaining unclear, such as, the detailed microstructure of ZDMA/ rubbers, polymerization process of ZDMA during curing, structure evolution of this type of composites, and the relationship between PZDMA and rubber molecules. Dynamic viscoelasticity is related to the internal molecular motion, and can help us to better understand

the internal structure of the composite. In this article, our goal is to study the internal structures of the ZDMA/natural rubber (NR) composites in an early curing stage using RPA2000. Lu et al. [19] found that the in situ polymerization of ZDMA was almost completed at the beginning stage of curing. They observed that nanoparticles (10 nm) of PZDMA were formed even in the composite which was cured for 1 min. Nie et al. [9] reported that the apparent activation energy (Ea) of the polymerization of ZDMA was lower than that of curing the pure NR, this meant that the polymerization of ZDMA had priority over the curing of NR matrix in the same condition. In our previous study, we also found that the polymerization of ZDMA [20] or magnesium dimethacrylate (MDMA) [21] took place in advance of the crosslinking reaction of NR. A primary network dominated by the ionic bonds was formed in the early curing stage, while the backbone of the covalent crosslink network was not formed. However, our previous studies mainly focused on the crosslink network. This article will make a good supplement to our previous reports to show a structure evolution of the ZDMA/NR composites. Theoretically, if the nanoPZDMA [9, 13, 16, 19] has been formed when the NR network is not crosslinked, the aggregation of PZDMA should act to form filler–filler network, and the ZDMA/ NR will show a typical nonlinear viscoelastic behavior. As expected, the apparent ‘‘Payne effect’’ was observed when the ZDMA/NR composite was cured for 1 min. A ‘‘primary network’’ structure might be formed in this time, which contained covalent crosslink points, ionic crosslinks, physical adsorption, and PZDMA filler networks. Though limitations inherent in our experiment, we expect our study to contribute to further understanding of the microstructure of ZDMA/rubber, the polymerization process of ZDMA, and the relationship between PZDMA and rubber molecules. EXPERIMENTAL AND METHOD Raw Materials NR (Malaysia 1#) was provided by Guangzhou Rubber Industry Research Institute (China). Methacrylic acid (MAA) purchased from Guangzhou Xin’gang Chemical Factory (China) was purified by distillation under nitrogen at reduced pressure. Zinc oxide (ZnO) was purchased from Tianji Yaohua Chem (China). Dicumyl peroxide (DCP) purchased from Sinopharm Chemical Reagent (China) was purified by anhydrous alcohol recrystallization before use.

(parts per 100 parts of rubber) DCP, and 0–40 phr ZDMA. Rubber compounds were prepared in a two-roller mill. ZnO was added first in NR and mixed for 2 min, then MAA was added slowly while mixing, and the period was 4–5 min. When MAA was added, another 5 min mixing was followed, and DCP was added at last. The compound was stored at room temperature for 8 h. RPA2000 test The experiments, including frequency sweeps and strain sweeps, were conducted by RPA2000 (Alpha technologies, UK). The cavity house was a biconical test chamber closed by the action of a pneumatic ram operated at a pressure. A slight excess of test material was needed to ensure that the cavity house was full. Tests were thus made under pressurized conditions to make sure that porosity did not develop in the samples during the tests. Strain Sweeps 1. For the uncured compounds. The temperature and the frequency were kept constant at 608C and 60 cpm (cycles per minute), respectively. 2. For the 1-min cured composites. The temperature was kept constant at 1558C firstly. After the samples were subjected into the cavity house, the curing was started immediately. The curing time was 1 min. After that, the temperature was immediately reduced to 608C (or 1008C). Then conducted stress-softening tests. The frequency was kept constant at 60 cpm. When the test temperature was changed, the 1-min cured test sample also had to be placed by a new one. This measurement was done to avoid the decomposition of DCP at the conduct temperature for a long time. For example, the sample was cured for 1 min at 1558C, then the temperature was reduced to 608C immediately to conduct a sweep. When the temperature was rised to 1008C to conduct another sweep, an uncured sample was need to cure for 1 min at 1558C firstly, then the temperature was reduced to 1008C immediately to conduct this sweep. 3. For the vulcanizate. The temperature was kept constant at 1558C firstly. After the samples were subjected into the cavity house, the curing was started immediately. The curing times were their optimal cure time (T90). After that, the temperature was immediately reduced to 608C. Then conduct strain sweep. The frequency was kept constant at 60 cpm.

Stress-Softening Tests Samples Preparation Equivalent ZnO and MAA (ZnO/MAA molar ratio 0.5) was used to react in the NR matrix during mixing. The compounds contained 100 weight parts of NR, 1.5 phr 968 POLYMER COMPOSITES—-2012

After three consecutive scans of the samples, the temperature was raised to 1008C with a delay of 30 min, keeping the samples intact within the confines of the cavity of the RPA2000 and again reduced to 608C to carry out the fourth scan. This was done as a measure to DOI 10.1002/pc

FIG. 1. Frequency sweeps for ZDMA/NR vulcanizates. Strain amplitude: 0.18. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

determine the stress-softening of the samples. Thus, the fourth scan was defined as recovery sweep.

Frequency Sweeps Frequency sweeps were performed from 1 to 1000 cpm. All tests were repeated three times (using untested samples). All results showed very little scatter. Thus, the plotted values in the figures were not the mean values of all the experimental results, but one of the chosen test results.

SEM Morphology of the compounds was observed using a Philips XL-30FEG scanning electron microscope (SEM). SEM analysis was performed on samples fractured in liquid nitrogen and the surface of samples was covered with a thin layer of gold to prevent electrostatic charging build-up during observation. In order to represent detailed morphological information of the specimens, 35000 and 340,000 magnifications were utilized.

to characterize the interaction of filler–filler and filler– rubber. In general, the effect at high frequency is equivalent to the effect at low temperature because of reduce of the chains responsibility. Thus, the increase of modulus is attributed to a decrease in time available for molecular relaxation. As shown in Fig. 1, the storage modulus (G0 ) of vulcanizates increases as a function of the concentration of ZDMA. This is due to the increase in filler volume fraction (F) and the enhanced crosslink density by ZDMA [9]. The slopes of G0 curves are very small. This suggests that a presence of strong network in the vulcanizates leads to a more time-independent elastic response [22]. The case of the 1-min cured sample is quite different from that of vulcanizate. As shown in Fig. 2, the relative large slope of storage modulus represents a strong timedependent elastic response. The curve of the 1-min cured NR seems to be coincident with the uncured one, which indicates that the covalent crosslinked network of NR is not formed at this time. For the filled samples, the G0 are close to each other, which imply that the fundamental crosslinked network is not mature (see Fig. 1, the G0 curves of the vulcanizates separate from each other). In a low frequency region, the G0 of the sample with high concentration of ZDMA (40 phr) is lower than that of the sample with low concentration of ZDMA (10 phr). When the frequency is close to 1000 cpm, the G0 of the sample with high ZDMA concentration shows an apparent increase. This result can be explained by the dilution effect, that is, the amount of the deformable rubber is diluted by the un-deformable rigid filler [23]. As shown in Fig. 3, the tendency of G0 at the strain amplitude of 18 is similar to that at the strain amplitude of 0.18 (Fig. 2). However, in the low frequency region, values of G0 from the highest to the lowest are in follow sequence: 20 phr, 10 phr, 30 phr, and 40 phr. This specific behavior is due to the internal structure of the 1-min cured sample.

RESULTS AND DISCUSSION Frequency Sweep The dynamic properties are usually expressed as complex modulus, which consists of a storage modulus and a loss modulus. A storage modulus or in-phase component represents an immediate response to the application of the force. A loss modulus or out-of-phase component represents the energy dissipated as heat. In this article, we mainly focus on the storage modulus which is used DOI 10.1002/pc

FIG. 2. Frequency sweeps for ZDMA/NR compounds cured for 1 min. Strain amplitude: 0.18. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

POLYMER COMPOSITES—-2012 969

FIG. 3. Frequency sweeps for ZDMA/NR compounds cured for 1 min. Strain amplitude: 18. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Nie et al. [9] clearly described the reaction process of ZDMA in NR. Peroxide radical abstracted hydrogen from methylene of molecular chains of NR, produced rubber radicals. When two rubber radicals met, a crosslinking bond was formed. Simultaneously, ZDMA underwent polymerization initiated by peroxide radicals. Because of the double bonds in the rubber, PZDMA radicals also reacted with NR radicals, forming NR-graft-PZDMA [9]. Moreover, when two PZDMA radicals met or one PZDMA radical abstracted hydrogen from a rubber chain, an ungrafted PZDMA molecule appeared [9]. As stated in the introduction, the apparent activation energy (Ea) of polymerization of ZDMA was lower than that of curing the pure NR, which meant that the ability to capture peroxide radicals of ZDMA was stronger than that of the pure NR in the same condition. Thus, it can be explained that, at a high concentration of ZDMA, majority of the limited DCP radicals can be captured by ZDMA to form PZDMA while a small quantity of DCP radicals is consumed by crosslinking of NR; at a low concentration of ZDMA, the majority of the limited DCP radicals can be consumed by crosslinking of NR. Our previous study [20, 21] have verified that the covalent crosslinked network of NR is not formed at this time, but some primary covalent crosslink points can be formed. When the strain amplitude is raised to 18, the deformation of rubber chain occurs. As a result, the fact that the 1-min cured sample with 20 phr ZDMA show the highest G0 as a result of the density of NR crosslinks and the concentration of PZDMA aggregate and their interaction. In the high frequency region, the rapid increase of G0 of the sample with 40 phr ZDMA is due to the high concentration of rigid filler-PZDMA. Strain Sweep Figure 4 exhibits the elastic torque (S0 ) of the ZDMA/ NR vulcanizates versus strain amplitudes. The S0 overlap 970 POLYMER COMPOSITES—-2012

each other at low strain amplitudes (not more than 0.18); when the strain amplitude exceed 0.18, the elastic torques begin to be separated. This implies that the crosslinked network get involved in the deformation of structure when the strain amplitudes exceed 0.18. It is also clearly seen that the elastic torque increases as a function of increasing ZDMA concentration at high strain amplitudes. This is due to that the ZDMA increases the crosslink density. Conversely, the S0 of uncured compounds shows a decrease with increasing ZDMA concentration, as shown in Fig. 5. The case of the 1-min cured samples is quite different from that of uncured and cured samples. As shown in Fig. 5, the elastic torque of the NR is the lowest which overlaps the uncured one. This again indicates that the crosslinked network of NR is not formed in such a short curing time. However, compared with the uncured ones, all the filled samples show an increase of the elastic torque, which imply the existence of crosslinks. This is good evidence that the ZDMA can improve the crosslinking efficiency [9] of rubbers. Similar to the G0 of frequency sweep (Fig. 3), sequence of the S0 from the highest to the lowest is below: 20 phr, 10 phr, 30 phr, and 40 phr. This demonstrated that the density of crosslinks in the sample with 20 phr ZDMA is higher than that of the others, and the density of crosslinks with 40 phr ZDMA was the lowest. As discussed before, the more ZDMA added in the rubber, the less DCP radicals consumed in NR crosslinking. Theoretically, the sample with 10 phr ZDMA should obtain the highest density of NR crosslinks and result in the highest elastic torque. However, the elastic torque of sample with 10 phr ZDMA is lower than that of the sample with 20 phr ZDMA. Here, it is necessary to take the PZDMA into account. The sample cured for 1-min contains uncrosslinked NR macromolecules, primary covalent crosslink points, residual ZDMA, PZDMA, and NR-graft-PZDMA. NR-graft-PZDMA was inclined to exist in the interface of NR and PZDMA.

FIG. 4. Strain sweeps for S0 of the ZDMA/NR vulcanizates. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

DOI 10.1002/pc

FIG. 5. Strain sweeps for S0 of the ZDMA/NR samples. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

PZDMA could also independently act with rubber in aggregation formed by physical adsorption [9]. Because of the large numbers of ion pairs in PZDMA molecules and the strong electrostatic interaction between ion pairs, aggregates consisting of several pairs called ‘‘multiplets’’ could be formed, restricting the mobility of adjacent polymer chains [9, 14]. Moreover, the PZDMA also can form a filler network, which is similar to a black carbon network [23]. Thus, a considerable PZDMA and crosslinks of NR form a relatively developed ‘‘primary network’’ [20] (containing covalent crosslink points, ionic crosslinks, physical adsorption, and filler–filler joints) in the sample with 20 phr ZDMA. Thus, the sample with 20 phr ZDMA obtains the highest S0 in this particular experiment. Figure 6 presents the change in G0 of uncured compounds against the strain. At high concentration of

FIG. 7. Strain sweeps for G0 of the ZDMA/NR samples cured for 1 min. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

ZDMA, G0 is more sensitive to the shear strain, which is a typical characteristic of weak filler structure breakdown [5–8]. All the filled 1-min cured samples show a pronounced Payne effect [5–8], as shown in Fig. 7. Once again, the sample with 20 phr ZDMA shows the highest G0 . The G0 curves of 1008C were lower than that of 608C, which is due to the softening of NR matrix. Please note that, at very high strain amplitudes (108), the G0 curves of 10 phr and 20 phr are very close, whereas the 40 phr and pure NR are very close. In this high strain amplitudes region, the filler networks have been destroyed completely and rubber is stretched to deform. The value of G0 decreases to a low level that is mainly contributed by the rubber matrix. Thus in the case of the samples with 10 phr and 20 phr, the higher G0 is attributed to the higher density of crosslinks; whereas in the case of the samples with 0 phr and 40 phr, the G0 reveals that there is very little crosslinks of NR matrix. These results are consistent with the conclusion which is deduced from the analysis of the elastic torque. What must be pointed out here is that the half-life of DCP at 1008C is up to about 50 h, and the scan test time is not more than 5 min. Therefore, traces of DCP may be decomposed during the test at 1008C, but the whole tendency of the result is not affected.

Stress-Softening Behavior

FIG. 6. Strain sweeps for G0 of the uncured ZDMA/NR samples. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

DOI 10.1002/pc

The stress-softening behavior of the G0 is shown in Fig. 8. It is observed that the higher the ZDMA loading, the greater the softening of the G0 . After the first scan, the linear viscoelastic region (LVE) of G0 is reduced with increasing ZDMA concentration. It is accepted that reinforcement is related to the combinational contribution of the filler–filler and the filler–rubber interactions [24, 25] and breakdown of these interactions is considered to be responsible for strain softening. After rest at 1008C for POLYMER COMPOSITES—-2012 971

FIG. 8. Stress-softening of G0 of ZDMA/NR samples cured for 1 min. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

30 min, all the filled samples show a remarkable recovery of G0 which even exceed that in the first scan. The recovery of G0 is due to the thermal movement of the uncured NR chains. Because the cure extent of the filled samples is very low (especially the sample with high loading of 972 POLYMER COMPOSITES—-2012

ZDMA), the NR matrix retains some fluidity. When the temperature is raised, the PZDMA can aggregate together to rebuild the filler–filler structures which contributes to the recovery of G0 . On the other hand, the residual DCP may decompose to generate radicals to continue the crossDOI 10.1002/pc

FIG. 9. Stress-softening of S0 of ZDMA/NR samples cured for 1 min. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

link reaction during rest at 1008C for 30 min. As a result, the G0 of the fourth scan exceeds that of the first scan. Figure 9 exhibits the stress-softening behavior of S0 . Generally speaking, rupture of the crosslinked network in a cured rubber occurs at about 108 (equal to about 140%). DOI 10.1002/pc

In this particular experiment, the overlapping of the first three scans of S0 indicates no rupture of the rubber network. Please note that, the significant stress-softening behavior and pronounced Payne effect can be observed in G0 (Figs. 7 and 8), but no stress-softening behavior is POLYMER COMPOSITES—-2012 973

observed in S0 (Fig. 9), this supports that a developed ‘‘filler structure’’ exists in the uncrosslinked rubber matrix. In another word, the fundamental covalent crosslinked network of NR matrix is not formed in the 1-min cured NR/ ZDMA samples, and the formed PZDMA can act as particles to form filler–filler structures. After rest at 1008C for 30 min, the fourth scan of the neat NR does not show any change, while the filled samples exhibit an apparent increase of S0 at high strain amplitudes. This suggests that some crosslinks are really formed. Thus, the remarkable recovery of G0 in Fig. 8 is mainly attributed to the crosslinks due to the decomposition of DCP during rest at 1008C for 30 min. We consider that the increased crosslinkings are mainly attributed to the ionic crosslinks, cooperating with a little of primary covalent crosslink points and some physical adsorption [20, 21].

SEM Analysis In order to give a direct observation of the morphology of the ZDMA/NR compound, the SEM images were taken from the fracture surface of uncured sample (Fig. 10) and 1-min cured sample (Fig. 11). As shown in Fig. 11, the large-scale ZDMA which can be observed in Fig. 10 disappears, indicating that most of ZDMA have polymerized to form PZDMA. In a large magnifying SEM picture (Fig. 11b), nano-PZDMA ‘‘particle’’ can be observed clearly. CONCLUSIONS The nonlinear viscoelasticity of 1-min cured ZDMA/NR composites is associated with the ionic crosslinks, some primary covalent crosslink points, PZDMA–PZDMA interactions, physical adsorptions depending on ZDMA concentration, and amplitude deformation. The polymerization of ZDMA resulted in nanoscaled PZDMA which could act as nanoparticles to form a filler–filler network. More ZDMA could result in the less rubber crosslinks because ZDMA captured more DCP radicals to produce PZDMA. A

FIG. 10. SEM micrographs of the uncured NR/30phr ZDMA compound.

974 POLYMER COMPOSITES—-2012

FIG. 11. SEM micrographs of NR/30 phr ZDMA compound cured for 1 min: (a) 50003; (b) 40,0003.

‘‘primary network’’ structure which contained covalent crosslink points, ionic crosslinks, physical adsorption, and filler–filler networks was formed in the 1-min cured ZDMA/NR composites. However, the backbone of the covalent crosslinked NR network was not formed. REFERENCES 1. M. Wang, Rubber Chem. Technol., 71, 520 (1998). 2. L. Chazeau, J. Brown, L. Yanyo, and S. Sternstein, Polym. Compos., 21, 202 (2000). 3. S. Wolf, M.J. Wang, and E.H. Tan, Kautsch Gummi Kunstst, 47, 870 (1994). 4. F. Zhao, X.Y. Shi, X. Chen, and S. Zhao, J. Appl. Polym. Sci., 117, 1168 (2010). 5. A.R. Payne, J. Appl. Polym. Sci., 6, 368 (1962). 6. A.R. Payne and W.F. Watson, Rubber Chem. Technol., 36, 147 (1963). 7. J.A.C. Harwood, L. Mullins, and A.R. Payne, J. Appl. Polym. Sci., 9, 3011 (1965). 8. G. Kraus, Appl. Polym. Symp., 39, 75 (1984). 9. Y.J. Nie, G.S. Huang, L.L. Qu, P. Zhang, G.S. Weng, and J.R. Wu, J. Appl. Polym. Sci., 115, 99 (2010). 10. D. Yin, Y. Zhang, Y. Zhang, Z. Peng, Y. Fan, and K. Sun, J. Appl. Polym. Sci., 85, 2667 (2002). DOI 10.1002/pc

11. D. Yin, Y. Zhang, Z. Peng, and Y. Zhang, Eur. Polym. J., 39, 99 (2003). 12. Z. Peng, X. Liang, Y. Zhang, and Y. Zhang, J. Appl. Polym. Sci., 84, 1339 (2002). 13. Y.L. Lu, L. Liu, C. Yang, M. Tian, and L.Q. Zhang, Eur. Polym. J., 41, 577 (2005). 14. Y.L. Lu, L. Liu, M. Tian, H.P. Geng, and L.Q. Zhang, Eur. Polym. J., 41, 589 (2005). 15. X.H. Yuan, Z.L. Peng, Y. Zhang, and Y.X. Zhang, J. Appl. Polym. Sci., 77, 2740 (2000). 16. A.H. Du, Z.L. Peng, Y. Zhang, and Y.X. Zhang, J. Polym. Sci. Part B: Polym. Phys., 42, 1715 (2004). 17. A. Nomura, J. Takano, A. Toyoda, and T. Saito, J. Jpn. Rubber Soc., 66, 830 (1993).

DOI 10.1002/pc

18. Y. Saito, K. Nishimura, M. Asada, and A. Toyoda, J. Jpn. Rubber Soc., 67, 867 (1994). 19. Y. Lu, L. Liu. D. Shen, C. Yang, and L. Zhang, Polym. Int., 53, 802 (2004). 20. Y.K. Chen and C.H. Xu, Polym. Compos., 32, 1505 (2011). 21. C.H. Xu, Y.K. Chen, and X.Y. Zeng, J. Appl. Polym. Sci., 125, 2449 (2012). 22. N. Phewthongin, P. Saeoui, and C. Sirisinha, J. Appl. Polym. Sci., 100, 2565 (2006). 23. T. Sajjayanukul, P. Saeoui, and C. Sirisinha, J. Appl. Polym. Sci., 97, 2197 (2005). 24. G. Heinrich and M. Kluppel, Adv. Polym. Sci., 160, 1 (2002). 25. Y.K. Chen and C.H. Xu, J. Appl. Polym. Sci. 123, 833 (2012).

POLYMER COMPOSITES—-2012 975