A novel dielectric elastomer by constructing dual-network structure of carbon nanotubes and rubber nanoparticles in dynamically vulcanized thermoplastic elastomer

Abstract

Thermoplastic vulcanizates (TPVs), as a special class of high-performance thermoplastic elastomers (TPEs), consist of a high content (60–80 wt%) of crosslinked rubber particles as the dispersed phase and a low content of a thermoplastic as the matrix. In this study, inspired by the special microstructure of TPVs, we prepared carbon nanotubes (CNTs)/TPV dielectric elastomer composites with a high dielectric constant (k) and low dielectric loss by constructing a dual network formed by rubber and CNTs. The rubber network was formed by a high content of agglomerates of rubber nanoparticles in the TPVs, which simultaneously promoted the formation of a CNTs network at a low content of CNTs in the matrix, to increase the value of k, and hindered the direct connection of CNTs with one another, to decrease the dielectric loss. As a result, the CNTs/TPV composites simultaneously possessed a high value of k and low dielectric loss. Moreover, the elasticity of the composites was improved by the CNTs because of the nanosprings of CNTs. This study provides a new simple and effective strategy for preparing a high-performance dielectric elastomer with a high value of k, low dielectric loss, good mechanical properties, high elasticity, high processability and easy recyclability.

---

1. Introduction

Dielectric elastomers (DEs) are a group of the most promising electroactive polymers for sensors and actuators because of their high flexibility, high energy density, large actuated strain, durability, reliability, fast response, high processability, etc.^1–4 The dielectric constant (k) of conventional elastomers such as silicone elastomers, acrylic elastomers, etc., is usually quite low (below 10).^5,6 One traditional method for increasing the value of k of DEs is to add ceramic powders with a high value of k such as TiO₂ to the matrix.² To obtain a high value of k, a high content (up to 50 vol%) of ceramics is usually required, which results in the deterioration of the mechanical properties and processability of the elastomer matrix.⁷ In addition, metal particles (magnesium, etc.)¹ and ionic conductors (polyethylene glycol oligomers, etc.)⁸ have also been added to the elastomer matrix to increase the value of k. In this case, a particular disadvantage is that the dielectric loss of DEs dramatically increases and the electric breakdown strength dramatically decreases. Another widely used strategy is to add carbon nanotubes (CNTs) to the elastomer matrix because of the high value of k and large aspect ratio of CNTs.^9,10 A high value of k can be obtained by the addition of a small amount of CNTs, so the mechanical properties can be maintained. However, the large increase in the value of k of the DEs is still accompanied by an increase in dielectric loss, which limits the practical applications of DEs.^11,12

Several innovative techniques have been proposed for obtaining CNTs/polymer composites with a high value of k and low dielectric loss. For example, unique core–shell structured CNTs (MWCNTs), in which the outer wall of CNTs was covalently functionalized to become non-conductive, whereas the inner wall was not functionalized and remained electrically conductive, were prepared and used to increase the value of k and reduce the dielectric loss of CNTs/polymer composites.^3,7 The non-conductive outer wall of the MWCNTs facilitated the decrease in dielectric loss, and the electrically conductive inner wall facilitated the increase in the value of k.^10,13–15 In addition, CNTs/polymer composites with a high value of k and low dielectric loss were obtained by introducing insulating layers on the surface of CNTs, because coating by an insulating shell can effectively prevent the direct connection of CNTs with one another and stop leakage current.¹⁶ On the other hand, CNTs/polymer composites with a high value of k and low dielectric loss can also be obtained by controlling the alignment of CNTs in the polymer matrix.^7,17 In our previous study, we achieved the alignment of CNTs in the matrix by using an easy and controllable electrospinning in situ film-forming (EF) technique and obtained CNTs/polymer composites with a high value of k and low dielectric loss.⁷ In addition, we have controlled the alignment of CNTs by simply controlling the shearing force during processing and produced composites with a high value of k and low dielectric loss.¹⁷

Thermoplastic vulcanizates (TPVs), as a special class of high-performance thermoplastic elastomers (TPEs) prepared by dynamic vulcanization, consist of a high content (60–80 wt%) of crosslinked rubber particles as the dispersed phase and a low content of a thermoplastic as the matrix.^18,19 Thus, TPVs combine the high elasticity and good mechanical properties of traditional thermosetting elastomers, as well as the good melt processability and easy recyclability of thermoplastics.^20,21 TPVs have attracted much attention as “green” polymers and have been widely used in industries such as the automotive industry, electronics, etc. Nowadays, TPVs have become one of the fastest growing elastomers because of the requirements for environmental protection and resource-saving.^19,22,23

In our previous study, we revealed that the rubber microparticles that were dispersed in the thermoplastic matrix in EPDM/PP TPV were actually agglomerates of rubber nanoparticles, and that they formed a rubber network in the TPV by connecting with each other.^24,25 In this study, inspired by the special microstructure of EPDM/PP TPV, we aimed to prepare a CNTs/TPV composite with a high value of k and low dielectric loss by constructing a dual network formed by rubber and CNTs. The rubber network, which occupied a large volume of the TPV, was expected to simultaneously promote the formation of a CNTs network at a low content of CNTs, to increase the dielectric constant, and hinder the formation of a conductive path, to decrease the dielectric loss. A special kind of aligned carbon nanotube bundles (CNTBs) prepared by electrostatic self-assembly was used for melt blending with TPVs, because these CNTBs could be well dissociated into single MWCNTs and dispersed uniformly in the matrix during mechanical shearing.³ More importantly, these MWCNTs could act as nanosprings to adjust the viscoelasticity and improve the elasticity of the elastomer matrix.²⁶ We studied the dual-network structure, dielectric properties, mechanical properties, and microstructure–properties relationship of the CNTs/TPV dielectric elastomer. This study provides a new, simple and effective technique for preparing high-performance DEs with a high dielectric constant, low dielectric loss, good mechanical properties, high processability and easy recyclability.

2. Experimental section

2.1 Materials

PP (MFI = 0.5 g (10 min)⁻¹, T_m = 159 °C) was supplied by Basell Co., Ltd. (Thailand), and EPDM rubber (ethylene content = 66%, ML (1 + 4) 125 °C = 61 Pa s) was supplied by Mitsui Chemicals Co., Ltd. (Japan). tert-Butyl peroxyacetate (TBPA) supplied by Lanzhou Auxiliary Agent Plant (China) was used as the crosslinking agent for the rubber phase in EPDM/PP TPV. Commercially available pentaerythritol tetrakis[3-(3,5-di-tert-butyl-4-hydroxyphenyl)propionate] (1010) was used as an antioxidant. CNTs bundles with a purity of greater than 93% and an average diameter of 8 nm were supplied by Beijing Cnano Technology Co., Ltd. (China).

2.2 Preparation of CNTs/TPV composites

2.2.1 Preparation of EPDM/PP TPV. EPDM/PP TPV was prepared in a Haake internal mixer (Haake Rheomix 600 OS, Thermo Fisher Scientific, USA) equipped with two counter-rotating rotors. The premix of the EPDM/PP blend in the Haake Rheomix was prepared by using the two-roll mill as follows. Firstly, EPDM rubber and PP (60 : 40) were fed into the two-roll mill at 180 °C, and 1010 (0.25 wt% based on EPDM rubber) was added to prevent the aging of PP. Subsequently, the homogeneous blend was transferred to another two-roll mill at ambient temperature, when TBPA (5.0 wt% based on EPDM rubber) was added. The premix obtained was fed into the Haake Rheomix at 170 °C and a rotor speed of 80 rpm, and EPDM/PP TPV was obtained after dynamic vulcanization for 10 min.

2.2.2 Preparation of CNTs/TPV composite samples. CNTs/TPV composites with different contents of CNTs were prepared by mixing different amounts of CNTs bundles (1.2 wt%, 2.0 wt%, 2.5 wt%, 3.0 wt%, 3.5 wt%, and 4.0 wt%) with the prepared EPDM/PP TPV at 180 °C for 7 min in a two-roll mill with a roll diameter of 6 inches.

2.3 Electrical properties tests

Films with a thickness of 0.5 mm were compression-molded at 180 °C for testing their electrical properties. A high-resistance meter (EST 121, Beijing Huajinghui Co., Ltd., China) was used to measure the electrical conductivity of CNTs/TPV composites, which is the reciprocal of volume resistivity. The volume resistivity (R_V) was calculated by using the equation as follows:^3,8where R is the electrical resistance of the sample, d is the diameter of the sample, and L is the thickness of the sample.

The dielectric properties of the composites were measured by using broadband dielectric spectroscopy (Alpha-A, Novocontrol, Germany) in the frequency range of 20 Hz to 10⁶ Hz at room temperature. The samples had a diameter of 25 mm and a thickness of 0.5 mm.

2.4 RPA analysis

The extent of dispersion of the CNTs in the TPV can be indicated by the Payne effect of the CNTs/TPV composites, which were analyzed by a rubber process analyzer (RPA 2000, Alpha, America). In this study, the strain scanning conditions were 190 °C, 0.2 Hz, and a strain range of 0.7–1300%. Before the determination, the samples were preheated for 5 min.

2.5 Morphology studies

An H-800 transmission electron microscope (TEM) purchased from Hitachi Co. (Japan) was used to observe the dispersion and connection of the CNTs in the CNTs/TPV composites. Before observation, the samples were cryomicrotomed into sections with a thickness of 100 nm at −130 °C and then vapor-stained with ruthenium tetroxide for 25 min. In addition, an S-4800 scanning electron microscope (SEM) supplied by Hitachi Co. (Japan) was used to further observe the dispersion of the CNTs. Before SEM observations, the samples were first brittle-fractured by using liquid nitrogen, and then the fractured surface was coated with gold and observed.

2.6 Tensile testing

Films with a thickness of 1 mm were compression-molded at 180 °C, from which dumbbell-shaped tensile bars (25 × 6 × 1 mm) were punched.

Tensile tests were carried out according to ASTM D412-98 at a strain rate of 500 mm min⁻¹ by using a CMT 4104 tensile tester (Shenzhen SANS Testing Machine Co., Ltd., China).^27,28 Each sample was tested at least five times and the average values are presented. The elastic modulus of the samples was determined by calculating the slopes of stress–strain curves at a strain of 5%. Cyclic stretching stress–strain curves of the samples were also obtained by using a CMT 4104 tensile testing machine at a strain rate of 500 mm min⁻¹, and the largest strain was set at 50%. The residual strain when the tensile stress recovered to zero was considered as the permanent set of the samples. The hysteresis losses of the composites were calculated from the ratio of the area of the stress–strain loops during tension recovery to the area below the loading curves.²⁶

3. Results and discussion

3.1 CNTs and rubber nanoparticles dual network in CNTs/TPV composites

In order to demonstrate the formation of a dual network in the CNTs/TPV composites, the fractured surfaces of CNTs/TPV composites with different contents of CNTs were observed by using SEM (see Fig. 1). To show more clearly the dispersion state of CNTs in the composites, SEM images of the composites with higher magnification are shown in Fig. 1(a′) and (b′). We can see many agglomerates of rubber nanoparticles in the composites, which is in good agreement with our previous findings that the rubber microparticles in TPVs are actually agglomerates of rubber nanoparticles.²⁴ These agglomerates of rubber nanoparticles are interconnected with one another and form a rubber network in the PP matrix (indicated by the yellow dashed line in Fig. 1(a)). Furthermore, it can be observed that most of the CNTs are uniformly dispersed in the PP matrix, whereas some CNTs are dispersed among the rubber nanoparticles (see Fig. 1(a′)), which is highly consistent with the results revealed by Min et al.²⁹ This phenomenon indicates that CNTs are more easily dispersed in the PP matrix, which is caused by the good melt fluidity of PP. The same phenomenon can be observed in the composites with lower contents of CNTs (0.5 wt%), as shown in Fig. S1.† In the composites with higher contents, both the amount of CNTs dispersed among the PP matrix and the rubber phase increased with an increase in the content of CNTs (see Fig. 1(b)).

Fig. 1 SEM micrographs of CNTs/TPV composites with different contents of CNTs (a′ and b′ are the magnified images of a and b, respectively): (a) 1.2 wt%; (b) 3.0 wt%.

Because only a small part of the CNTs can be observed at the fractured surface by using SEM, we further observed the dispersion of CNTs in the composites by using TEM and the results are shown in Fig. 2. In this figure, the hollow tubes are CNTs and the lighter and darker regions represent the PP and stained EPDM phases, respectively. In order to clearly observe the CNTs in TEM images, we used TEM images with high magnification (see scale bars in Fig. 2). Therefore, the interconnection of agglomerates of EPDM nanoparticles cannot be observed in Fig. 2, because only part of the EPDM rubber network is shown in the micrograph. The interconnection between agglomerates of rubber nanoparticles and the rubber network can be observed in TEM micrographs with lower magnification, as shown in Fig. S2 in the ESI.† From Fig. 2, we can observe more clearly that most CNTs are dispersed in the PP matrix rather than the EPDM phase. There is almost no interconnection between the CNTs in the composite with 1.2 wt% of CNTs (see Fig. 2(a)), which indicates that no CNTs network was formed in the composite. CNTs started to interconnect with one another in the composite with 2.0 wt% of CNTs (see Fig. 2(b)). The interconnection between CNTs increased significantly with a further increase in the content of CNTs and a CNTs network formed in the composite (see Fig. 2(c)). On the other hand, we can also see that the rubber domains in the composites acted as barriers, which hindered the formation of a conductive path of CNTs.

Fig. 2 TEM micrographs of CNTs/TPV composites with different contents of CNTs: (a) 1.2 wt%; (b) 2.0 wt%; (c) 3.0 wt%. The hollow tubes are CNTs and the lighter and darker regions represent the PP and stained EPDM phases, respectively.

Thus, Fig. 1 and 2 demonstrate that the rubber network is formed by agglomerates of rubber nanoparticles and a CNTs network is formed by the interconnection of CNTs with an increase in the content of CNTs in the CNTs/TPV composites. Thus, the rubber nanoparticles and CNTs together form a dual network in the composites. In order to further demonstrate the formation of a dual network in the composites with an increase in the content of CNTs, RPA was performed on pure PP, EPDM/PP TPVs, and the CNTs/TPV composites with different contents of CNTs (see Fig. 3). We can see from Fig. 3(a) that the elastic modulus (G′) of the pure TPV exhibited a plateau at low shear strain before it declined, which is attributed to the flexible rubber network in the TPV. The plateau in the value of G′ of the composites was shortened with an increase in the content of CNTs, which indicates a change in the network in the composites from flexible to stiff. The plateau in the value of G′ still existed when the content of CNTs increased to 4.0 wt%, which indicates that the dual network of CNTs/rubber agglomerates in the composites was still flexible at a high content of CNTs.

Fig. 3 Variations in the EPDM rubber network and conductive network of CNTs in the CNTs/TPV composites with different contents of CNTs: (a) storage modulus vs. shear strain for PP and CNTs/TPV composites; (b) variations in ΔG′ and conductivity of the CNTs/TPV composites.

As previously reported, the value of G′ decreases with an increase in the strain amplitude (Payne effect) for filled rubbers, and eventually reaches quite a low value and then remains stable when the shear strain approaches a certain level, which is regarded as the minimum value of G′. The difference between the maximum and minimum values of G′ (ΔG′) is a measurement of the filler network.^30–32 In this study, both the crosslinked rubber nanoparticles and the CNTs can be regarded as fillers dispersed in the CNTs/TPV composites. A larger value of ΔG′ indicates a stronger dual network in the composite and a stronger Payne effect.^30,31 In Fig. 3(a), the values of G′ of all these composites decreased to almost the same as that of PP as the shear strain approached 1300%, which indicates the complete breakdown of the dual networks at high strain rates. Therefore, the value of G′ of the composites under a strain of 1300% can be regarded as the minimum value of G′, and the difference between the initial value of G′ and that under a shear strain of 1300% is the value of ΔG′ of the composites. The value of ΔG′ of EPDM/PP TPV is higher than that of PP, which is attributed to the network of rubber nanoparticles in the TPV. The value of ΔG′ of the composites further increases with an increase in the content of CNTs, and the change in the value of ΔG′ of the composites is demonstrated in Fig. 3(b). It can be seen that ΔG′ increased significantly with an increase in the content of CNTs, which indicates that the CNTs network became denser.

The conductivity of the composites also increased with an increase in the content of CNTs, which indicates the gradual formation of a conductive network of CNTs in the composites, which is highly consistent with the TEM results (see Fig. 2). A sharp increase in conductivity could be observed when the content of CNTs increased from 2.0 wt% to 3.0 wt%. Within this range, the interconnection between the CNTs increased rapidly and the formation of the conductive network of CNTs was completed. Therefore, we define a content of CNTs of 3.0 wt% as the percolation threshold of the CNTs/TPV composites. On the other hand, the conductivity of the composites was still low (2.5 × 10⁻⁵ S m⁻¹) when the content of CNTs reached 4.0 wt%. This is because the network of rubber agglomerates isolated a large number of the CNTs dispersed in the PP matrix (see Fig. 1), and consequently hindered the direct interconnection of many CNTs and the formation of a conductive pathway.

Based on the above results, we propose a schematic representation of the microstructure of the CNTs/TPV composites, as shown in Fig. 4. The agglomerates of rubber nanoparticles in EPDM/PP TPV were interconnected with one another and formed a rubber network in the PP matrix. During the mixing of the CNTBs with the TPV at high temperatures, the CNTBs dissociated easily into single CNTs, which were uniformly dispersed in the PP matrix because of the good melt fluidity of PP.²⁹ Moreover, a small part of the CNTs were dispersed between the crosslinked rubber nanoparticles when break-up and reagglomeration of the rubber agglomerates occurred simultaneously in the rotational shear field during mixing, which led to the dispersion of a small fraction of CNTs in the rubber phase (see Fig. 4(a)). With an increase in the content of CNTs, the CNTs that were dispersed in the PP matrix gradually interconnected with one another and formed a network of CNTs in the composites. Thus, both the storage modulus (G′) and the electrical conductivity increased sharply, and we define this point as the percolation threshold of the composite. Therefore, a dual network composed of a rubber network and a network of CNTs formed in the CNTs/TPV composites after the percolation threshold, as shown in Fig. 4(b). On the other hand, the rubber network hindered the direct connection of some CNTs.

Fig. 4 Schematic of the formation of a dual network with an increase in the content of CNTs.

3.2 Effect of dual network on dielectric properties of CNTs/TPV composites

To study the effect of the dual network on the dielectric properties of the composites, we measured the dielectric properties of pure EPDM/PP TPV and all the CNTs/TPV composites in the frequency range of 20–10⁶ Hz and the results are shown in Fig. 5. As an example, the dielectric properties at 1000 Hz of the CNTs/TPV composites are shown in Fig. 6. As is seen, the dielectric constant (k) increased slightly from ∼2 to ∼10 at a content of CNTs lower than 3.0 wt%, whereas k increased sharply to ∼140 when the content of CNTs increased to 4.0 wt%. This again indicates that a content of CNTs of 3.0 wt% is the percolation threshold of the composites. The dielectric loss at 1000 Hz increased slightly with an increase in the content of CNTs when the content of CNTs was lower than 2.0 wt%. The dielectric loss increased almost linearly from 0.1 to 0.6 when the content of CNTs was within the range of 2.0–4.0 wt%. The dielectric loss of CNTs/TPV composites is not high compared with other dielectric composites with similar contents of conductive filler.^7,17 The underlying mechanism of the dielectric behavior of the CNTs/TPV composites is attributed to the dual network, which simultaneously increased the value of k via the network of CNTs and decreased the dielectric loss by hindering the formation of a conductive pathway of CNTs by the rubber network. The results indicate that CNTs/TPV composites with a high value of k and low dielectric loss can be successfully prepared by simply mixing aligned CNTBs with a TPV, owing to the formation of the specific dual-network structure.

Fig. 5 Dielectric constant and dielectric loss of CNTs/TPV composites with different contents of CNTs.

Fig. 6 Dielectric constant and dielectric loss of CNTs/TPV composites at 1000 Hz.

The dependence on frequency of the dielectric properties of a dielectric material is also very important for its wide application; therefore, the dielectric properties of the composites at different frequencies were also analyzed (see Fig. 5). The dielectric constant (k) of the pure TPV and the composites with contents of CNTs of 2.0 wt% and 2.5 wt% remained constant with an increase in frequency, which indicates that the value of k is frequency-independent for the composites with contents of CNTs below the percolation threshold, as shown in the inset of Fig. 5(a). The values of k of the composites with contents of CNTs of 3.0 wt%, 3.5 wt% and 4.0 wt% sharply decreased with an increase in frequency (see Fig. 5(a)), which is attributed to the space charge polarization of the CNTs.⁸ In addition, the values of k of the composites with higher contents of CNTs are much more sensitive to frequency than those with low contents of CNTs. The dependence on frequency of the dielectric loss of the composites is almost the same as that of the value of k.

3.3 Effect of dual network on mechanical properties of CNTs/TPV composites

The tensile properties of pure EPDM/PP TPV and the CNTs/TPV composites with different contents of CNTs were measured, and the stress–strain curves are shown in Fig. S3(a).† We can see from the figure that the tensile strength slightly increased from 8.1 MPa for the pure TPV to 9.4 MPa for the composite with a content of CNTs of 4.0 wt%, which indicates that the addition of the CNTs only provided a slight increase in the tensile strength. Furthermore, the addition of CNTs had almost no influence on the elongation at break of the composites. Fig. S3(b)† shows the elastic modulus of the composites determined by calculating the slopes of the stress–strain curves at a strain of 5%. The elastic modulus of the composites significantly increased from 42 MPa to 63 MPa with an increase in the content of CNTs from 0 wt% to 4.0 wt%, which is attributed to the strengthening of the network of CNTs.

The changes in the tensile properties of the CNTs/TPV composites with different contents of CNTs are mainly attributed to the deformation mechanism of the TPV and the dispersion of the CNTs. The deformation of the TPV at low tensile strain (≤20%) was dominated by the stretching of the PP ligament between the rubber agglomerates.^21,33,34 The elastic modulus at low tensile strain was significantly affected by the addition of CNTs because the CNTs were mainly dispersed in the PP matrix and gradually formed a network with the increase in their content. When the tensile strain of the TPV samples was above 20%, the deformation behavior was dominated by the distortion of the agglomerates of rubber nanoparticles. The strain increased while the stress remained constant due to the slippage and orientation of the agglomerates of rubber nanoparticles, as reported in our previous studies.³⁵ Because most of the CNTs were dispersed in the PP matrix, the deformation process of the CNTs/TPV composites was similar to that of the TPV at high tensile strain. Therefore, both the tensile strength and the elongation at break of the composites were hardly affected by the contents of CNTs.

3.4 Effect of dual network on elasticity of CNTs/TPV composites

To examine the effect of the dual-network structure on the elasticity of the composites, we studied the hysteresis loss and permanent set of the composites via a tension recovery test. A large permanent set means significant slippage and internal friction between elastomeric macromolecule chains, which consequently results in high hysteresis loss.²⁶ Fig. 7(a) shows the tension recovery stress–strain curves of CNTs/TPV composites with different contents of CNTs, and we can see that the permanent sets of all the samples range from 11% to 15%. According to ASTM D1566-07a, our CNTs/TPV composites are elastomers with high elasticity, which is attributed to the rubber network structure of the CNTs/TPV composites. In addition, it can be seen that both the permanent set and the hysteresis ratio decreased with an increase in the content of CNTs (see Fig. 7(b)), which indicates that the elasticity increased with an increase in the content of CNTs. The reason is that the CNTs dispersed in the composites behaved like nanosprings in the low strain region, which further increased the elasticity of the composites.²⁶ Thus, the network structure of the CNTs also contributed to the high elasticity of the composites.

Fig. 7 Elastic performance of CNTs/TPV composites with different contents of CNTs: (a) the tension recovery stress–strain curves for the composites; (b) the hysteresis loss and permanent set as functions of the content of CNTs.

4. Conclusions

Based on the special microstructure of a TPV and the nanospring effect of carbon nanotubes, we successfully prepared CNTs/TPV composites with high values of k, low dielectric loss, improved elasticity, high processability and easy recyclability for the first time by dissociating aligned CNTBs into a TPV and constructing a dual-network structure of rubber nanoparticles/CNTs. The rubber network was formed by agglomerates of rubber nanoparticles, which occupied most of the volume of the composites. The CNTs were more easily dispersed among the PP matrix and gradually formed a network of CNTs. The rubber network promoted the formation of the network of CNTs at a low content of CNTs. On the other hand, the agglomerates of rubber nanoparticles acted as barriers to hinder the direct interconnection of some CNTs with one another. As a result, the CNTs/TPV composites simultaneously possessed high values of k and low dielectric loss. Moreover, the CNTs acted as nanosprings to improve the elasticity of the composites. This study provides a new technique for preparing a high-performance DE by using a simple and effective method.

Acknowledgements

We gratefully acknowledge the National Science Fund for Distinguished Young Scholars of China (Grant No. 51525301), the Basic Research Program of China (Grant No. 2011CB606003), the National Natural Science Foundation of China (Grant No. 51221002), and the Doctoral Science Research Foundation of the Education Ministry of China (Grant No. 20130010110005) for financial support.

References

1. D. Yang, L. Zhang, H. Liu, Y. Dong, Y. Yu and M. Tian, J. Appl. Polym. Sci., 2012, 125, 2196–2201.
2. H. Liu, L. Zhang, D. Yang, Y. Yu, L. Yao and M. Tian, Soft Mater., 2013, 11, 363–370.
3. N. Ning, L. Ji, L. Zhang, J. Liu, Y. Lu, S. Wu, H. Zou, M. Tian and T. W. Chan, Compos. Sci. Technol., 2015, 118, 78–84.
4. R. Trujillo, J. Mou, P. Phelan and D. Chau, Int. J. Adv. Des. Manuf. Technol., 2004, 23, 176–182.
5. F. Carpi and D. D. Rossi, IEEE Trans. Dielectr. Electr. Insul., 2005, 12, 835–843.
6. S. P. Lacour, H. Prahlad, R. Pelrine and S. Wagner, Sens. Actuators, A, 2004, 111, 288–292.
7. N. Ning, X. Bai, D. Yang, L. Zhang, Y. Lu, T. Nishi and M. Tian, RSC Adv., 2014, 4, 4543–4551.
8. M. Tian, B. Yan, Y. Yao, L. Zhang, T. Nishi and N. Ning, J. Mater. Chem. C, 2014, 2, 8388–8397.
9. K. Yuse, D. Guyomar, M. Kanda, L. Seveyrat and B. Guiffard, Sens. Actuators, A, 2011, 165, 147–154.
10. C. Yang, Y. Lin and C. Nan, Carbon, 2009, 47, 1096–1101.
11. X. Zhang, G. Liang, J. Chang, A. Gu, L. Yuan and W. Zhang, Carbon, 2012, 50, 4995–5007.
12. Q. Li, Q. Xue, L. Hao, X. Gao and Q. Zheng, Compos. Sci. Technol., 2008, 68, 2290–2296.
13. Q. Li, Q. Xue, Q. Zheng, L. Hao and X. Gao, Mater. Lett., 2008, 62, 4229–4231.
14. P. Pötschke, S. M. Dudkin and I. Alig, Polymer, 2003, 44, 5023–5030.
15. Z. M. Dang, S. H. Yao and H. P. Xu, Appl. Phys. Lett., 2007, 90, 012907.
16. M. Tian, Z. Wei, X. Zan, L. Zhang, J. Zhang, Q. Ma, N. Ning and T. Nishi, Compos. Sci. Technol., 2014, 99, 37–44.
17. S. Liu, M. Tian, L. Zhang, Y. Lu, T. W. Chan and N. Ning, J. Mater. Sci., 2016, 51, 2616–2626.
18. A. Coran and R. Patel, Rubber Chem. Technol., 1983, 56, 210–225.
19. M. v. Duin, Macromolecules, 2006, 233, 11–16.
20. H. Sun, P. He, S. Zhao and Z. Chen, J. Macromol. Sci., Part B: Phys., 2011, 50, 1423–1429.
21. G. de Schaetzen, J. Oderkerk, B. Goderis, L. Hellemans and G. Groeninckx, Macromolecules, 2002, 35, 6623–6629.
22. K. Severyns, Medical Device Technology, 2000, 11, 28–32.
23. S. Abdou-Sabet, R. Puydak and C. Rader, Rubber Chem. Technol., 1996, 69, 476–494.
24. N. N. Hanguang Wu, L. Zhang and Y. W. a. M. T. Hongchi Tian, Soft Matter, 2014, 10, 1816–1822.
25. H. Wu, M. Tian, L. Zhang, H. Tian, Y. Wu, N. Ning and T. W. Chan, ACS Sustainable Chem. Eng., 2014, 3, 26–32.
26. J. Liu, Y. L. Lu, M. Tian, F. Li, J. Shen, Y. Gao and L. Zhang, Adv. Funct. Mater., 2013, 23, 1156–1163.
27. R. l'Abee, H. Goossens and M. v. Duin, Polymer, 2008, 49, 2288–2297.
28. U. Basuli, T. Chaki and K. Naskar, J. Appl. Polym. Sci., 2008, 108, 1079–1085.
29. K. T. Min and G. H. Kim, Polym. Adv. Technol., 2011, 22, 2273–2278.
30. J. Fröhlich, W. Niedermeier and H.-D. Luginsland, Composites, Part A, 2005, 36, 449–460.
31. A. R. Payne, J. Appl. Polym. Sci., 1962, 6, 57–63.
32. L. Wang, N. Ning, L. Zhang, Y. Lu, M. Tian and T. Chan, Composites, Part A, 2013, 47, 135–142.
33. M. C. Boyce, K. Kear, S. Socrate and K. Shaw, J. Mech. Phys. Solids, 2001, 49, 1073–1098.
34. K. K. Mary, C. Boycea, S. Socrate and O. Yeh, J. Mech. Phys. Solids, 2001, 49, 1323–1342.
35. T. A. Huy, T. Luepke and H. J. Radusch, J. Appl. Polym. Sci., 2001, 80, 148–158.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra01463a

---