Titanium oxide nanoparticle increases shallow traps to suppress space charge accumulation in polypropylene dielectrics

Abstract

Polymer nanocomposite dielectrics always attract widespread attention in electrical and electronic fields. Space charge suppression under direct current electric field is one of the key issues in developing high performance insulation materials. This paper reports a potential mechanism of space charge suppression in polypropylene/titanium oxide nanocomposites. Trap level distribution and space charge accumulation are obtained by thermally stimulated current and pulsed electro-acoustic method, respectively. The micro morphology and structure of the nanocomposites are examined by differential scanning calorimetry, X-ray diffraction and positron annihilation lifetime spectroscopy. The results indicate that doping of titanium oxide introduces numerous shallow traps and reduces the number of deep traps, which significantly suppresses space charge accumulation and increases conductive current. The results can be explained by the fact that shallow traps, resulting from the interfaces between nanoparticles and polymer matrix and the interaction between nanoparticles and the crystalline/amorphous interfaces, could increase the charge carrier mobility and reduce potential barrier for charge carrier transport. This potential mechanism is of great importance to understand the space charge suppression in polymer nanocomposites and in designing high performance nanodielectric materials.

---

Introduction

Polymer nanocomposites filled with various nanoparticles have drawn great attention in scientific research and industrial applications in many fields, such as biomedicine, optical, electronic and electrical materials and devices.^1,2 The application of nanocomposites in electrical insulation is especially called nanodielectrics, which was firstly raised by Lewis in 1994.³ After this, extensive research has been carried out on nanodielectrics. Recent advances have demonstrated that nanodielectrics exhibit dramatically improved breakdown strength,^4–6 space charge suppression,^7–9 partial discharge resistance,^10,11 mechanical performance,¹² and dielectric permittivity¹³ compared with the original polymer materials. Doping of nanoparticles has been a popular and effective way for designing and preparing insulation materials with specific properties. Interfacial zones between nanoparticles and the polymer matrix are thought to be the origin of the high performance of nanodielectrics.¹⁴ Various models considering the structure and characteristics of the interfaces have been proposed including DLVO (Derjaguin and Landau, Verwey and Overbeek) model,¹⁵ water shell model,¹⁶ and multi-core model.¹⁷ Nevertheless, these models are based on experiences, which are not clear enough to explain all the experimental results and should be modified taking into account the physical mechanism.

Space charge accumulation under direct current (DC) electric field in polymer dielectrics is one of the key issues for new insulation material development, especially for high voltage direct current (HVDC) applications.¹⁸ Space charge accumulation would seriously affect the operation performance and service life of the HVDC equipment as the result of affecting the dielectric strength, local electric field distortion, electrical conduction and aging performance of the insulation dielectrics.¹⁹ Thus, methods of space charge suppression and related mechanism are of great importance to be studied. Experimental results have shown that introducing nanoparticles can greatly suppress the space charge accumulation in various polymer materials,^7,20 but few studies deal with the physical mechanism of space charge accumulation and suppression.

Previous studies have shown that titanium oxide (TiO₂) nanoparticles have excellent chemical stability, thermal stability and corrosion resistance, which is a good choice in the development of nanocomposites. For the use of insulation materials, the electrical, thermal and mechanical properties of the material should be considered. Various studies have shown that introduction of TiO₂ nanoparticles can improve the stability of polymeric materials under UV exposure, improve the mechanical properties before and after thermal aging and suppress space charge accumulation.^21–25 Because the electrical properties of nanocomposites especially space charge accumulation depend seriously on the nanoparticle dispersion,²⁶ the TiO₂ nanoparticles are surface modified with the most commonly used silane coupling agent KH550.

In this research, surface modified titanium oxide nanoparticles are introduced into polypropylene and a suppression of space charge accumulation is observed in polypropylene/titanium oxide (PP/TiO₂) nanocomposites and the effect varies with the doping content of nano-TiO₂. In order to explain the physical mechanism of space charge suppression, trap level distribution in the nanocomposites is determined by thermally stimulated current (TSC) tests and the results demonstrate that nano-TiO₂ dopants introduce shallow carrier traps and reduce original deep carrier traps in PP. Reference to the “multi-core” model,¹⁷ shallow carrier traps in PP/TiO₂ nanocomposites should be located at the interface of nanoparticles and polymer matrix. These shallow carrier traps would result in higher charge carrier mobility to assist the transport of charges injected from the electrodes so as to suppress space charge accumulation. The higher charge carrier mobility of the nanocomposites is further confirmed by the increased conduction current. This improved model reveals a potential physical mechanism of space charge suppression from the view of charge transport and carrier trap level distribution.

Experimental

Surface modification of TiO₂ nanoparticles

The TiO₂ nanoparticles (40 nm, Aladdin Industrial Inc. China) are vacuum dried at 120 °C for 12 h before surface modification. About 8 g dried TiO₂ nanoparticles are dispersed in 200 mL toluene and ultrasonic treated for 30 min at room temperature. About 3 mL silane coupling agent KH550 (Sinopharm Chimical Reagent Co., Ltd, China) is added into 50 mL toluene. Then TiO₂/toluene mixture is poured into a three-necked flask and kept at 60 °C in oil bath and electrically stirred. The KH550/toluene mixture is slowly added into the flask by dropping funnel. After that, the mixture is heated to 120 °C in oil bath and electrically stirred for 6 h to obtain the modified nanoparticles suspension. The suspension is centrifuged at 6000 rpm for 6 min, and the supernatant liquid is removed. Finally the modified nanoparticles are vacuum dried at 60 °C for 12 h and grinded into nanoparticles powder.

Preparation of PP/TiO₂ nanocomposites

PP/TiO₂ nanocomposites with various nanofiller content are prepared by melt blending using the masterbatch method. Firstly, the PP pellets and surface modified nanoparticles are dried in vacuum oven at 60 °C for 12 h prior to use. Then the masterbatch with 10 phr surface modified TiO₂ nanoparticles is mixed in a torque rheometer (RM-200C, HAPRO, China) at 200 °C for 10 min with a rotor speed of 60 rpm. Then the masterbatch is diluted with pure PP into PP/TiO₂ nanocomposites with 0.5 phr, 1 phr, 3 phr and 5 phr TiO₂ nanoparticles. The dilution process is carried out at the same condition of masterbatch preparation. In order to make comparison between pure PP and the nanocomposites, pure PP is also melted in the torque rheometer. Film samples with different thickness for properties measurements are prepared using compression molding at 200 °C for 8 min under a pressure of about 15 MPa and then cooled to room temperature under the same pressure. In order to eliminate the influence of residual internal stress, all film samples are annealed in vacuum oven at 100 °C for 2 h.

Characterization

Micro morphology and dispersion of the TiO₂ nanoparticles in PP matrix are observed by field emission scanning electron microscope (FE-SEM, ZEISS Sigma, Germany). Samples for FE-SEM observation are broken in liquid nitrogen and sputtered with gold on the fractured surface to avoid charge accumulation while observing.

Crystallinity and crystalline structure of PP and PP/TiO₂ nanocomposites are verified by differential scanning calorimetry (DSC) and X-ray diffraction (XRD), respectively. The DSC tests are performed under nitrogen atmosphere at a heating rate of 10 °C min⁻¹ between 30 and 200 °C. The XRD experiments are carried out in the range of 2θ = 5–35° at a scanning rate of 3° min⁻¹ at room temperature.

Positron annihilation lifetime spectroscopy (PALS) is used to determine the free volume fraction in PP/TiO₂ nanocomposites. The measurements are performed on an EG&G ORTEC fast–fast coincidence system with a time resolution of 210 ps. The ²²Na source with the intensity of 13 μCi is sandwiched between two flat samples with a thickness of about 1 mm. Over 2 million counts are recorded to obtain the spectrums and then the spectrums are resolved by LT9.0 program.

The space charge distribution is measured by pulsed electro-acoustic (PEA) method. Before PEA tests, samples are electrically short-circuited at 100 °C for 2 h in vacuum oven to eliminate the internal charges generated during sample preparation process. After that, the space charge accumulation during polarization and charge decay during depolarization are measured with the PEA method. Samples are firstly polarized under DC electric field of 50 kV mm⁻¹ for 30 min and the space charge distribution are recorded. After polarization, the samples are immediately electrically short circuited to record the depolarization space charge distribution.

Before TSC tests, samples are sputtered with gold on both sides to serve as electrodes and then the samples are electrically short-circuited at 100 °C for 2 h in vacuum oven to eliminate the internal charges generated during sample preparation process. After that, the tested sample is first polarized under 1 kV at 20 °C for 30 min and then cooled down to −100 °C quickly. After a delay of 3 min at −100 °C, the sample is depolarized for 5 min to release the polarization charges. Finally the sample is linearly heated in 3 °C min⁻¹ and the thermally stimulated current is recorded.

Conduction currents of the samples are measured with a standard three-electrodes system. DC electric field of 50 kV mm⁻¹ is applied across the samples and the charging current is immediately recorded after the electric field is applied. The charging current is measured for 10 min to reach its steady state.

Results and discussion

Micro morphology and structure

Dispersion of TiO₂ nanoparticles in PP matrix is shown by SEM images in Fig. 1. The TiO₂ nanoparticles are homogeneously dispersed in PP matrix and no obvious agglomerate is found and the border between the nanoparticles and polymer matrix is obscure, indicating the effectiveness of nanoparticle surface modification.

Fig. 1 FE-SEM images of PP and PP/TiO₂ nanocomposites with different filler content.

Fig. 2 gives the melting curves of pure PP and TiO₂/PP nanocomposites and the results show that there is not much differences between the melting curves. The crystallinity of the samples calculated according to the melting curves is listed in Table 1. It indicates that the incorporation of TiO₂ nanoparticles does not change the semi-crystalline characteristic of PP and the crystallinity of PP.

Fig. 2 DSC melting curves of PP and PP/TiO₂ nanocomposites with different filler content.

Table 1 Crystallinity of PP and PP/TiO₂ nanocomposites with different filler content

	0 phr TiO2/PP	0.5 phr TiO2/PP	1 phr TiO2/PP	3 phr TiO2/PP	5 phr TiO2/PP
Crystallinity	16.40%	16.37%	16.78%	16.33%	16.26%

---

The XRD curves given in Fig. 3 show a typical XRD spectrum of PP with five typical diffraction peaks located at 2θ = 14–22°. The diffraction peak at about 2θ = 25° corresponds to TiO₂ nanoparticles, the intensity of which increases with TiO₂ nanoparticle content. Comparing the spectrums of PP and PP/TiO₂ nanocomposites, there is almost no obvious changes, indicating that the crystalline structure does not change with the inclusion of TiO₂ nanoparticles.

Fig. 3 XRD spectrums of PP and PP/TiO₂ nanocomposites with different filler content.

The free volume in PP/TiO₂ nanocomposites is quantitatively checked with PALS according to ref. 27 and 28. According to the PALS theory, the ortho-positronium (o-Ps) would preferentially localizes in the free volume of the polymer materials and the annihilation of which make it a microprobe of free volume because the lifetime and intensity of o-Ps are related to the size and fraction of free volume in the system. The radius of the free volume cavity R (in nm) can be obtained by the Tao–Eldrup equation:^29,30

where τ₃ is the lifetime of long-lived o-Ps, Δ is the width of the electron layer at the internal surface of the potential well and equal to 1.656 Å. The free volume fraction can be obtained by:

f_v = 0.0018 × V × I₃

where V = 4πR³/3 is expressed in ų unit and I₃ is the intensity.

The positron annihilation lifetime spectrum and the free volume fraction calculated according to the PALS results are shown in Fig. 4 and Table 2, respectively. The results indicate that the free volume fraction is increasing with TiO₂ nanofiller content. The increase of free volume proves that the inclusion of nanoparticles could change the properties of the original material.

Fig. 4 Positron annihilation lifetime spectrums of PP and PP/TiO₂ nanocomposites with different filler content.

Table 2 PALS lifetime components and free volume fraction of PP and PP/TiO₂ nanocomposites

	τ1 (ns)	I1 (%)	τ2 (ns)	I2 (%)	τ3 (ns)	I3 (%)	Free volume fraction (%)
0 phr TiO2/PP	0.1774	51.18	0.4927	30.87	2.5699	17.95	4.95
0.5 phr TiO2/PP	0.1812	53.32	0.5110	28.63	2.5500	18.05	4.93
1 phr TiO2/PP	0.1819	52.28	0.5000	29.69	2.5820	18.03	5.02
3 phr TiO2/PP	0.1826	53.67	0.5150	27.67	2.5462	18.66	5.07
5 phr TiO2/PP	0.1764	50.39	0.4860	30.41	2.5420	19.20	5.20

---

Space charge distribution

The space charge distribution is measured by pulsed electro-acoustic (PEA) method under 50 kV mm⁻¹ during polarization and depolarization time. Details of PEA measurements and space charge density calculation can be found in ref. 31 and 32. It can be seen from Fig. 5(a) that immediately after the voltage is applied, large amounts of heterocharges accumulate at both the cathode and the anode, and the maximum charge density is about 5 C m⁻³ in pure PP. Also, space charge in pure PP decays very slowly in Fig. 5(b), indicating lower charge carrier mobility. However, space charge characteristics of PP/TiO₂ nanocomposites are quite different, which is effectively suppressed and varies with the filler content as the result of nanoparticles doping. This is of great importance for practical application in HVDC insulation. During the depolarization time, space charge decay in PP/TiO₂ nanocomposites is apparently faster than pure PP, indicating that the charge carrier mobility of the nanocomposites is higher than pure PP. Previous studies have shown that the space charge behavior of polymer nanocomposites is closely related to the charge transport characteristics and the charge transport is the process of trapping and detrapping of the charge carriers.^33,34 So the trap characteristics, which determines the carrier mobility, free carrier concentration, hopping distance and activation energy, are closely related to the charge transport process and space charge accumulation characteristics.

Fig. 5 Space charge distribution during (a) polarization and (b) depolarization in PP and PP/TiO₂ nanocomposites with different filler content under 50 kV mm⁻¹.

Carrier trap level distribution

In order to investigate the potential mechanism of space charge suppression, thermally stimulated current (TSC) method is adopted to evaluate the changes of trap level distribution in PP/TiO₂ nanocomposites. Fig. 6(a) gives the thermally stimulated current spectrums of pure PP and PP/TiO₂ nanocomposites. According to TSC theory, the thermally induced currents should be attributed to the release of trapped charges and the currents under different temperature correspond to different trap depth. Therefore, the corresponding trap level distribution is calculated from the TSC spectrum according to the modified TSC theory and the details can be found in ref. 35. The trap level distribution of PP/TiO₂ nanocomposites is shown in Fig. 6(b). Multi-peaks fitting is used for the trap level distribution to separate the trap peaks with different trap depth and the results are shown in Fig. 7. In pure PP, there are three notable trap peaks located at 0.61, 1.0 and 1.1 eV and about 90% of the traps are in the range of 0.9 to 1.2 eV, which should be attributed to deep traps in polymer materials.³⁶ But with the increased TiO₂ filler content, the number of deep traps decreases and that of shallow traps increases. In PP/TiO₂ nanocomposite with 5 phr TiO₂ nanoparticles, there are only two trap peaks located at 0.61 and 0.74 eV, corresponding to shallow traps, and almost all trap levels are under 0.9 eV, which is much lower than that of pure PP. So large amounts of shallow traps are introduced by the doping of TiO₂ nanoparticles and the number of deep traps decreases.

Fig. 6 (a) Thermally stimulated current and (b) trap level distribution in PP and PP/TiO₂ nanocomposites with different filler content.

Fig. 7 Multi-peaks fitting of trap level distribution in PP/TiO₂ nanocomposites with (a) 0 phr, (b) 0.5 phr, (c) 1 phr, (d) 3 phr, and (e) 5 phr TiO₂ nanoparticles.

PP is a kind of semi-crystalline material, crystalline phase (mostly spherocrystal) and amorphous phase coexist in PP, so there are interfaces between the crystalline phase and amorphous phase.^37,38 Because of the structural disorder at the interfaces between the crystalline phase and amorphous phase, there are numerous defects and these defects would act as charge trapping sites and strongly affect the transport of charge carriers in pure PP. In addition, interfaces between impurities and PP would also generate defects and traps, but these traps are negligible because there are few impurities in the PP used in this investigation. It has been found previously that the trap depth at the interfaces between crystalline phase and amorphous phase in pure polymer material is about 1.0 to 1.4 eV by using TSC technology aided with X-ray.³⁶ Our results of trap levels in pure PP are consistent with the existing results. Therefore the deep traps in pure PP should be attributed to the interfaces between crystalline phase and amorphous phase.

When the nanoparticles are introduced into polymers, there are probably two cases of changes in trap level distribution. One is the addition of new trap level distribution to the original trap level distribution in pure PP, the other one is the replacement of the original trap level distribution by new trap level distribution. Addition of new trap level distribution seems more intuitive and easy to think, but it is not consistent with the experimental results. If the increased shallower traps in PP/TiO₂ nanocomposites shown by the TSC results is the additional trap level distribution and the deep traps still exist, space charge in PP/TiO₂ nanocomposites should increase because charge carriers would be trapped by deep traps in the end, which forms space charges. On the contrary, space charge suppression is observed in PP/TiO₂ nanocomposites shown by Fig. 5. So modification of the original deep traps by nanoparticles doping and transform to shallower traps is more reasonable, because from the TSC measurements shown in Fig. 6, the trap depth decreases gradually with the increase of TiO₂ nanofiller content and transforms to shallower trap level distribution. Also the results of DSC and XRD experiments reveal that the semi-crystalline characteristic of PP and the degree of crystallinity do not change obviously. Such kind of trap level distribution modification should be attributed to the numerous interface regions generated in the nanocomposites as the result of the huge specific surface area of the nanoparticles. Properties of the interface regions are quite different from either the particles or the polymer matrix, which is regarded as the dominate feature of nanodielectrics and the excellent properties of nanodielectrics should be attributed to the interface regions.¹⁴ Local physical properties of the interfaces are different in crystallinity, free volume, chain arrangement and charge carrier mobility, etc. These differences may change the trap level distribution in the nanocomposites and modify the original deep traps in pure PP.

Mechanisms of the space charge suppression

A discussion on the changes of trap level distribution in PP/TiO₂ nanocomposites can be given based on the multi-core model proposed by T. Tanaka.¹⁷ According to the multi-core model, interfaces between spherical nanoparticles and polymer matrix are divided into three layers, a bonded layer (first layer), a bound layer (second layer), and a loose layer (third layer) with the thickness of 1 nm, several nm and several tens nm, respectively. In addition, an electric double layer overlaps the above three layers. Free volume becomes higher and density becomes lower from the first layer to the third layer and as a result, carrier traps are distributed from deep to shallow. Deep traps are present in the first and second layers and shallow traps are present in the third layer. Because of the third layer is much larger in volume friction than the second layer, the shallow traps become dominated traps in the nanocomposites. In addition, the nanoparticles and interfaces between the nanoparticles and polymer matrix would interact with the original traps in pure PP electrostatically due to the far-distance effect, resulting in pushing the deep traps to shallower levels.³⁹ Therefore the shallow traps increase in the nanocomposites and replace the original deep traps in pure PP. As the result of numerous shallow traps introduced, the charge carriers will transport in the shallower trap band to result in higher charge carrier mobility in PP/TiO₂ nanocomposites. So the space charge accumulation is suppressed. The effect of enhanced charge carrier mobility can also be verified from the space charge measurement during short-circuited condition shown in Fig. 5(b). The decay of space charge in PP/TiO₂ nanocomposites is conspicuously faster than pure PP, indicating that the space charge in PP/TiO₂ is trapped by shallow traps which is easy to detrap.

Fig. 8 gives a schematic diagram of charge carrier transport in PP/TiO₂ nanocomposites. In pure PP, almost all carrier traps are deep traps so that the charge carriers are difficult to detrap once they are trapped by these deep traps, and thus the permanent space charge accumulation forms. However, things are quite different in PP/TiO₂ nanocomposites because of the large amount of shallow traps introduced by nano-dopants, which would help the carriers to transport and suppress space charge accumulation. Three typical charge transport forms are given in Fig. 8 (the dash lines (a)–(c)). Path (a) describes the charge carrier transports between the shallow traps and eventually is captured by the deep traps to form space charge. Path (b) describes the charge carrier detraps from the shallow traps and then transports between the shallow traps. Path (c) describes the charge carrier transports between the shallow traps. Results of PALS have shown that free volume in PP/TiO₂ nanocomposites increases with the content of TiO₂ nanoparticles. Considering the high free volume and low density characteristics of the third layer, charge carriers are more likely to choose the low resistance path and transport between the shallow traps with higher carrier mobility, namely path (c) is the most common charge transport form. Results of TSC testes have shown that with the increasing nanoparticle content, shallow traps increase and deep traps decrease. So that the charges trapped by deep traps reduces significantly and more charges would transport in the bulk of the material and as a result the space charge accumulation is suppressed. Furthermore, inter particle distance would reduce with the increase of filler content and the shallow traps in the third layer are more likely to build up shallow trap band, which is easier for charge carriers to transport.

Fig. 8 Charge carrier transport in PP/TiO₂ nanocomposites based on multi-core model.

Conduction current and carrier mobility

In order to prove the assumption of shallow traps increase the charge carrier mobility, conduction current under 50 kV mm⁻¹ is measured and shown in Fig. 9. The conduction current density is increasing with the increased TiO₂ nanoparticle content. The current density of the PP/TiO₂ nanocomposites with 5 phr TiO₂ nanoparticles is about 4 times higher than that of pure PP. The increase of conduction current proves that the charge carrier mobility of the nanocomposites is much higher than the polymer matrix. From the results of TSC tests, shallow traps are introduced in PP/TiO₂ nanocomposites to replace the deep traps in pure PP. Compared with the charge carriers trapped by deep traps in pure PP, charge carriers trapped by shallow traps are more easily to detrap. As a result, charge carrier mobility in the nanocomposites is much higher and performs as the increased conduction current in macroscopy. On the basis of the mechanism proposed above, charge carriers are more likely to transport between the shallow traps and result in higher carrier mobility.

Fig. 9 Conduction current of PP and PP/TiO₂ nanocomposites with different filler content under 50 kV mm⁻¹.

Conclusions

In summary, this investigation proposes a potential mechanism of space charge suppression in PP/TiO₂ nanocomposites. In order to reveal the mechanism of space charge suppression in PP/TiO₂ nanocomposites under high electric field, the space charge distribution under polarization and depolarization condition, and TSC spectrum which corresponds to trap level distribution are investigated. The results indicate that numerous shallow traps are introduced into PP/TiO₂ nanocomposites to replace the original deep traps in PP. It is demonstrated that the interfaces between nanoparticles and polymer matrix possess different properties with the original polymer and have significant impact on the properties of the nanocomposites. The increased shallow traps would enhance the charge carrier mobility, which contributes to the charge transport and suppresses the space charge accumulation. The increase of conduction current in PP/TiO₂ nanocomposites further confirms that the charge carrier mobility is increased in the nanocomposites. This investigation is of great importance on understanding the space charge suppression mechanism in polymer nanocomposites and the development of high performance HVDC nanodielectric materials.

Acknowledgements

This work is supported in part by the National Basic Research Program of China (973 Project) under grant 2014CB239504.

Notes and references

1. S. Kango, S. Kalia, A. Celli, J. Njuguna, Y. Habibi and R. Kumar, Prog. Polym. Sci., 2013, 38, 1232–1261.
2. A. C. Balazs, T. Emrick and T. P. Russell, Science, 2006, 314, 1107–1110.
3. T. J. Lewis, IEEE Trans. Dielectr. Electr. Insul., 1994, 1, 812–825.
4. J. K. Nelson and J. C. Fothergill, Nanotechnology, 2004, 15, 586–595.
5. Y. Zhou, J. L. He, J. Hu and B. Dang, J. Appl. Polym. Sci., 2016, 133, 42863.
6. K. Y. Lau, A. S. Vaughan and G. Chen, IEEE Electr. Insul. Mag., 2015, 31, 45–54.
7. G. C. Montanari, D. Fabiani, F. Palmieri, D. Kaempfer, R. Thomann and R. Mülhaupt, IEEE Trans. Dielectr. Electr. Insul., 2004, 11, 754–762.
8. Y. Murakami, M. Nemoto, S. Okuzumi, S. Masuda, M. Nagao, N. Hozumi and Y. Sekiguchi, IEEE Trans. Dielectr. Electr. Insul., 2008, 15, 33–39.
9. K. Y. Lau, A. S. Vaughan, G. Chen, I. L. Hosier, A. F. Holt and K. Y. Ching, IEEE Trans. Dielectr. Electr. Insul., 2014, 21, 340–351.
10. T. Tanaka, Y. Ohki, M. Ochi, M. Harada and T. Imai, IEEE Trans. Dielectr. Electr. Insul., 2008, 15, 81–89.
11. C. Nyamupangedengu and D. R. Cornish, IEEE Trans. Dielectr. Electr. Insul., 2016, 23, 85–94.
12. M. Takala, M. Karttunen, J. Pelto, P. Salovaara, T. Munter, M. Honkanen, T. Auletta and K. Kannus, IEEE Trans. Dielectr. Electr. Insul., 2008, 15, 1224–1235.
13. S. Singha, M. J. Thomas and A. Kulkarni, IEEE Trans. Dielectr. Electr. Insul., 2010, 17, 1249–1258.
14. T. J. Lewis, IEEE Trans. Dielectr. Electr. Insul., 2004, 11, 739–753.
15. T. J. Lewis, J. Phys. D: Appl. Phys., 2005, 38, 202–212.
16. C. Zou, J. C. Fothergill and S. W. Rowe, IEEE Trans. Dielectr. Electr. Insul., 2008, 15, 106–117.
17. T. Tanaka, M. Kozako, N. Fuse and Y. Ohki, IEEE Trans. Dielectr. Electr. Insul., 2005, 12, 669–681.
18. R. N. Hampton, IEEE Electr. Insul. Mag., 2008, 24, 5–13.
19. G. C. Montanari, IEEE Trans. Dielectr. Electr. Insul., 2011, 18, 339–364.
20. X. Y. Huang, P. K. Jiang and Y. Yin, Appl. Phys. Lett., 2009, 95, 242905.
21. Y. Feng, J. Yin, M. Chen, X. Liu, B. Su, W. Fei and Q. Lei, IEEE Trans. Dielectr. Electr. Insul., 2014, 21, 1501–1508.
22. J. W. Zha, Z. M. Dang, H. T. Song, Y. Yin and G. Chen, J. Appl. Phys., 2010, 108, 094113.
23. T. J. Turton and J. R. White, Polym. Degrad. Stab., 2001, 74, 559–568.
24. S. K. Esthappan, S. K. Kuttappan and R. Joseph, Polym. Degrad. Stab., 2012, 97, 615–620.
25. V. G. Nguyen, H. Thai, D. H. Mai, H. T. Tran and M. T. Vu, Composites, Part B, 2013, 45, 1192–1198.
26. L. Zhang, Y. Zhou, X. Cui, Y. Sha, T. H. Le, Q. Ye and J. Tian, IEEE Trans. Dielectr. Electr. Insul., 2014, 21, 1554–1564.
27. J. K. Nelson, Y. Huang, T. M. Krentz, L. S. Schadler, J. Dryzek, B. C. Benicewicz and M. Bell, Proceedings of IEEE ICPADM, 2015, pp. 40–43.
28. Y. Zheng, Y. Zheng and R. Ning, Mater. Lett., 2003, 57, 2940–2944.
29. S. J. Tao, J. Chem. Phys., 1972, 56, 5499–5510.
30. M. Eldrup, D. Lightbody and J. N. Sherwood, Chem. Phys., 1981, 63, 51–58.
31. P. Morshuis and M. Jeroense, IEEE Electr. Insul. Mag., 1997, 13, 26–35.
32. C. L. Cheng, J. L. He and J. Hu, Appl. Phys. Lett., 2014, 105, 133508.
33. D. Ma, T. A. Hugener, R. W. Siegel, A. Christerson, E. Mårtensson, C. Önneby and L. S. Schadler, Nanotechnology, 2005, 16, 724–731.
34. G. Chen and Z. Xu, J. Appl. Phys., 2009, 106, 123707.
35. F. Tian, W. Bu, L. Shi, C. Yang, Y. Wang and Q. Q. Lei, J. Electrost., 2011, 69, 7–10.
36. M. Ieda, IEEE Trans. Electr. Insul., 1984, EI-19, 162–178.
37. S. Brückner, S. V. Meille, V. Petraccone and B. Pirozzi, Prog. Polym. Sci., 1991, 16, 361–404.
38. M. L. Cerrada, M. J. Polo-Corpa, R. Benavente, E. Pérez, T. Velilla and R. Quijada, Macromolecules, 2009, 42, 702–708.
39. T. Tanaka, IEEE Trans. Dielectr. Electr. Insul., 2005, 12, 914–928.

---