Carbon nanotube-induced morphological transformation for toughening of benzoxazole-containing semi-crystalline polyimide

Abstract

Carbon nanotube (CNT)/semi-crystalline polyimide (PI) nanocomposites have been prepared by incorporating different concentrations of carboxylic acid-functionalized multi-walled CNTs. The incorporation of CNTs not only results in remarkable strength improvement, but also increases the toughness of nanocomposites. The toughness shows a maximum at the CNT concentration of 0.75 wt%, about twofold increment in toughness compared to neat PI. To explore the origin of this toughening behaviour, dispersion of nanotubes in the PI matrix and their interfacial interactions were investigated by TEM, FT-IR spectroscopy and UV-visible absorption spectroscopy, respectively. Morphological transformation of the crystalline phase induced by CNTs was monitored by polarized optical microscopy and WAXD. The fracture surfaces of CNT reinforced PI nanocomposites were investigated by SEM. It was suggested that the abrupt change of toughness with respect to CNT weight fraction is attributed to the percolation effects and concomitant morphological change, which was accounted for by a proposed model for nanotube-induced morphological transformation. The model provides novel insight into the design of strengthened and toughened materials by tailoring the structure at the nanoscale.

---

1. Introduction

Extraordinary mechanical, electrical and thermal properties of CNTs have sparked strong interest in developing novel CNT/polymer nanocomposites by translating their properties to the polymer matrix.^1–7 Furthermore, the nucleating ability of CNTs for the polymer matrix results in an interfacial structure different from the bulk polymer, which offers a promising route toward stiff, strong and tough materials.⁸ Nanoscale dispersion of nanotubes, nanotube-induced polymer conformation and packing alternation, and polymer–nanotube interface structure effectively affect the macro properties of CNT/polymer nanocomposites.⁹ After a decade of research, significant advancement has been achieved on improving the dispersibility of nanotubes in polymer matrices and tailoring the CNT filler–polymer matrix interactions.^10–16 However, the potential of CNTs as novel nanofillers remains to be further explored to maximize their strength and toughness as the remarkable reinforcements in tensile strength and Young's modulus of CNT/polymer nanocomposites are mostly achieved at the cost of the decrease in flexibility,^14,15 which frequently limits their practical application. Thus, it is important to establish effective approaches that can simultaneously improve the strength and toughness as well as the flexibility of CNT reinforced polymer composites. Some fundamental aspects of the strengthening and toughening mechanisms for polymer nanocomposites have been investigated by many researchers. Lu et al. demonstrated the structure–property relationship in rod-like nanoparticle-filled polyimide nanocomposites, and proposed a possible mechanism.¹⁷ Yudin et al. reported that only a small concentration of CNTs is sufficient to induce the crystallization, resulting in improved properties of the semi-crystalline PI nanocomposites.¹⁸ Kumar et al. reported that the CNT-induced crystallization gives rise to significant increases in strength, Young's modulus and elongation at break of polypropylene.¹⁹ Argon et al. showed that the toughness of high-density polyethylene can be improved by incorporating calcium carbonate particles.²⁰ To the best of our knowledge, however, few papers have reported on the simultaneous strengthening and toughening mechanisms of semi-crystalline PIs by incorporating CNTs. The fundamental understandings related to these improvements remain unexplored. Our present work intends to discern the underlying mechanism in view of the morphological transformation of PI matrices, which is induced by nanotubes affecting their failure behaviour.

In this work, PI containing benzoxazole rings is chosen as a semi-crystalline PI matrix due to its high thermal stability and mechanical properties.^21–24 Modification of CNT with surface-bound carboxylic acid (–COOH) groups aims to achieve well-dispersed CNTs in PI matrix during the preparation of CNT/PI nanocomposites by in situ polymerization. Nanotubes dispersion in PI matrix and their interfacial interactions are examined, and the effect of morphological transformation of semi-crystalline PIs induced by nanotubes on their mechanical behaviour and failure mode are investigated. Furthermore, a percolation model of morphological transformation is suggested to explain the strengthening and toughening of semi-crystalline PIs during the incorporation of CNTs.

2. Experiment

2.1 Materials

5-Amino-2-(p-aminophenyl)benzoxazole (AAPB, purity 99.71%, melting point 230.6 °C) was obtained from Harbin Institute of Technology. 4,4′-Oxydiphthalic anhydride (ODPA, purity 99.7%, melting point 227.3 °C) was received from Shanghai Research Institute of Synthetic Resins. N,N-Dimethylacetamide (DMAc) was redistilled prior to use. Carboxylic acid-functionalized multi-walled CNTs (COOH–CNTs, purity 99.9%) with diameters of 30–50 nm and lengths of 10–20 μm were obtained from Chengdu Organic Chemicals Co. Ltd, Chinese Academy of Science.

2.2 Synthesis of carboxylic acid-functionalized CNT/PI nanocomposites

The semi-crystalline PI was synthesized by a traditional two step method. First, the precursor, poly(amid acid) (PAA) was synthesized by a polycondensation of AAPB (3.1022 g) and ODPA (2.2632 g) in DMAc (23 mL). Then, the precursor PAA was thermally imidized at 350 °C for 1 h in a vacuum oven and subsequently cooled to room temperature to obtain PI. The CNT/PI nanocomposites with different concentrations of COOH–CNTs (i.e., 0, 0.25, 0.5, 0.75, 1 wt%) were prepared by in situ polymerization. The AAPB was dissolved in DMAc at room temperature with magnetic stirring, and then, the solution was sonicated in an ultrasonic bath at 47 kHz for 3 h after the addition of appropriate amounts of nanotubes. The resulting suspension of CNTs and AAPB mixture was immediately transferred to a three-neck round bottom flask equipped with a mechanical stirrer and drying tube outlet filled with calcium sulfate. After adding molar equivalent of ODPA to AAPB, the mixture was stirred at 300 rpm under nitrogen atmosphere at room temperature for 6 h to obtain a stable highly viscous solution. The resulting CNT–PAA solution was cast onto clean glass plates and stored in a dry air-flowing chamber at 90 °C for 10 h. Then the films, with the thickness of 40–60 μm, were heated step-wisely up to 350 °C in a vacuum furnace and cool to room temperature to obtain solvent-free PI films.

2.3 Characterization

The dispersion of nanotubes in PI matrix was examined by using transmission electron microscopic (TEM) (Hitachi H-7650 field emission TEM, Japan) at the acceleration voltage of 100 kV. The nanocomposite films were embedded into epoxy resin Epon-812 (USA) and cut into approximately 80 nm thick slices using a RMC PowerTomeXL Ultramicrotome with a diamond knife. The slices were collected on a 400 mesh copper grid before TEM measurement. Crystallization of the PI and CNT/PI nanocomposite films were investigated under a polarized optical microscopic (POM) equipped with Zeiss Axio Imager A2m by transmission mode. The Fourier transform infrared (FT-IR) spectra of the CNT/PI nanocomposite films were recorded at room temperature with Bruker Vector 22 system. Ultraviolet and visible absorption (UV-vis) spectra were recorded by using UV-2450 SHIMADZU spectrophotometer at a scan rate of 60 nm min⁻¹ over the range 190–500 nm for PI and CNT/PI nanocomposite films coated on a quartz plate. Scanning electron microscope (SEM) analysis was performed on a Quanta 200F SEM. The fracture surfaces of CNT/PI nanocomposites were coated with a thin gold layer by sputtering to suppress the surface charging effects. Tensile properties of the nanocomposites were measured with universal mechanical test system UTS-10, UTStestsysteme (Germany) at room temperature in a uniaxial extension mode. The films cut into strips of 30 mm by 2 mm were tested with an extension speed of 10 mm min⁻¹.

3. Results and discussion

3.1 CNT dispersion state and interfacial interactions of PI with CNT

The dispersion of nanotubes in the PI matrices was examined by TEM (Fig. 1). The results indicate that modified nanotubes were dispersed uniformly and homogeneously in the PI matrix at the concentration of 0.25 and 0.5 wt% (Fig. 1a and b). However, with increasing nanotubes concentration up to 0.75 wt%, the nanotubes begin to aggregate (in the circles) (Fig. 1c). Due to carboxylic acid groups on the surface of the nanotubes improving their compatibility with PI main chains, nanotubes were well-dispersed in PI matrix. At low concentration, the nanotubes do not interact with each other as they are sufficiently far apart, and consequently, a homogeneous dispersion can be achieved. However, at high concentration beyond a critical value, at which carbon nanotubes interact with each other, CNT aggregation takes place.

Fig. 1 TEM micrographs of CNT/PI nanocomposites with (a) 0.25 wt% CNTs; (b) 0.5 wt% CNTs; (c) 0.75 wt% CNTs with circles indicating CNT aggregation (scale bar: 500 nm).

The interfacial interactions between CNT and PI were further investigated by FT-IR spectroscopy. Substantial changes can be identified by comparing the characteristic peaks of PI with those of CNT/PI nanocomposites in the FT-IR spectra shown in Fig. 2. The characteristic peaks of imide ring are assigned at 1776 and 1712 cm⁻¹ (C=O asymmetrical and symmetrical stretching in imide ring, respectively), 1369 cm⁻¹ (C–N stretching in imide ring) and 744 cm⁻¹ (C=O bending in imide ring) for the neat PI. On the other hand, the CNT/PI nanocomposites exhibited the corresponding peaks at slightly different wavenumbers. The peaks corresponding to symmetric stretching vibrations of C=O and stretching vibrations of C–N were shifted to 1716 and 1365 cm⁻¹, respectively. These shifts are indicative of the non-covalent interactions between PI macromolecules and surface modified CNTs.

Fig. 2 FT-IR spectra of (a) PI and (b) CNT/PI nanocomposites with 0.5 wt% CNTs in the middle-wave area and in the long-wave area (inset).

To further investigate the interfacial interactions between PIs and CNTs, UV-vis absorption spectra of PI and CNT/PI nanocomposites were compared in Fig. 3. Intensity of the absorption band at 244 nm, which was assigned to be π–π* transition from phenyl ring to imide ring due to a charge-transfer complex between AAPB and ODPA units, decreased after introduction of carbon nanotubes, indicating the existence of non-covalent interactions between the PI backbone and the CNT surface. The UV-vis spectroscopic results are in close agreement with FT-IR data presented previously and provide a firm evidence that the polymer main chains in PI interact with nanotubes in the molecular level.

Fig. 3 UV-vis absorbance spectra of (a) PI and (b) CNT/PI nanocomposites with 0.5 wt% CNTs.

3.2 Morphological transformation induced by nanotubes in semi-crystalline PI

The crystalline structures of PI and CNT/PI nanocomposites were investigated by POM, as shown in Fig. 4. Apparent stars-like structures composed of radial lamellar aggregations can be seen in the neat PI in Fig. 4a. A similar phenomenon has also been reported by Cheng et al.²⁵ The size of lamellar aggregations ranges from 3 to 6 μm. POM images of CNT/PI nanocomposites (Fig. 4b) exhibited uniform distribution of fine lamellar crystals which should be induced by well-dispersed nanotubes. The nanotubes act as heterogeneous nucleates and dramatically increases the number of nucleation centres to form randomly oriented lamellar crystals. Such effect has been reported previously for other PI structures.^26,27 On increasing the nanotube content from 0.5 to 0.75 wt%, several large lamellar crystals appeared (Fig. 4c), which is due to the aggregation of nanotubes in the centre of large crystals. These results suggest that the aggregation of nanotubes could be more effective as nucleation centres than single nanotubes for growing lamellae.

Fig. 4 POM images of (a) neat PI, CNT/PI nanocomposites containing (b) 0.5 wt%, and (c) 0.75 wt% CNTs (scale bar: 10 μm).

3.3 Crystallinity of carbon nanotube reinforced PI composites

As shown by WAXD pattern in Fig. 5, the neat CNT displays peak at 2θ = 26.5° corresponding to the (002) reflections of the carbon atoms. PI displays a semi-crystalline structure, with three peaks at 2θ = 13.9, 17.8 and 26.2° corresponding to the interlayer spacing of d = 6.35, 4.98 and 3.40 Å, respectively. Addition of nanotubes to PI changes not only the peak intensities but also the ratio of them. The sharp peaks with increased intensities at 2θ = 13.9 and 17.8° in the nanocomposites indicate that the crystallinity increased with the nanotubes contents. The peak at about 26° can be attributed to both the face to face distance (3.40 Å) between two aromatic rings of PI and the distance between aromatic ring of PI and surface of underlying CNT. The change of the relative intensities at 2θ = 13.9, 17.8 and 26.2° in nanocomposites reveals the significant change in crystalline arrangement of PI chains after the addition of CNT, which should be investigated further in detail in the future. Nonetheless, it is obvious that the overall crystallinity of the PI nanocomposites increases with the addition of CNT.

Fig. 5 Wide angle X-ray diffraction patterns for COOH–CNT, neat PI and PI-based nanocomposites with 0.25, 0.5 and 1 wt% COOH–CNTs.

3.4 Mechanical properties of CNT/PI nanocomposites

To investigate the strengthening and toughening effect of CNTs on nanocomposite films, strain–stress tests in a uniaxial extension mode were carried out for the PI and CNT/PI nanocomposite films having different nanotube concentrations as shown in Fig. 6. Young's modulus (E), yield strength (σ_y), tensile strength at break (σ_b) and elongation at break (ε_b) were measured from the averaged curve. Fracture toughness was estimated as the integration of energy under the stress–strain curve up to fracture. Substantial increases in E, σ_y, and σ_b were observed with incorporation of CNTs into PI matrix. An appreciable increase of E, σ_y, and σ_b takes place already at 0.25 wt% CNTs, and the maximum was found at 0.5 wt% CNTs. Further increase of CNT concentration caused a decrease of mechanical parameters, E, σ_y, and σ_b, probably due to the aggregation of nanotubes as previously mentioned in the TEM results. The most striking feature of the nanocomposite fracture behaviour is that the introduction of CNTs at concentrations below 0.5 wt% does not affect the elongation at break (ε_b). Moreover, if CNTs were introduced at the concentration more than 0.75 wt%, a substantial increment in ε_b up to 52% and approximately twofold increment in toughness were observed as displayed in Fig. 7. The significant increments in elongation at break and toughness in the range between 0.5 and 0.75 wt% suggest that there exists a percolation threshold between these concentrations, at which carbon nanotubes simultaneously strengthen and toughen PI.

Fig. 6 Representative stress–strain curves for PI nanocomposites.

Fig. 7 Fracture toughness of CNT/PI nanocomposite.

In order to further study the increase of toughness taking into account the failure behavior of nanocomposites, the fracture surfaces for CNT/PI nanocomposites containing 0.5 and 0.75 wt% of CNTs were examined by SEM and compared with that of neat PI (Fig. 8). Neat PI displays a surface with agglomeration of grains and plastic deformed streaks corresponding to a shrinking deformation. With introducing nanotubes to PI matrix and increasing their concentration, the fracture surface gradually exhibits a large amount of voids with rough surfaces. A rougher fracture surface observed in CNT/PI nanocomposites containing 0.5 wt% of CNT (Fig. 8b) than in the neat PI (Fig. 8a) reflects the increased extent of plastic deformation with the nanocomposites failure being dominated gradually by adhesive mode. In addition, the preferential alignment of the CNT fragment in the direction perpendicular to the fracture surface (in the stretching direction) is observed, which was probably caused by unidirectional orientation of nanotubes during the deformation of nanocomposites. As the nanotube content increase to 0.75 wt%, nanotubes are fully stretched during the deformation, and the extent of plastic deformation abruptly increase, which is evidenced from the depth of voids in Fig. 8c. This fact implies the failure behaviour is completely in an adhesive mode, which dissipates the fracture energy dramatically and dominates the large deformation.²⁸

Fig. 8 Fracture surfaces of (a) neat PI, CNT/PI nanocomposites containing (b) 0.5 wt% and (c) 0.75 wt% CNTs under uniaxial tensile test up to failure (scale bar: 2 μm).

3.5 Mechanism of strengthening and toughening of semi-crystalline PI composites

Generally, a semi-crystalline PI exhibits a complicated multiphase morphology, mainly including amorphous phase and crystalline phase. After carbon nanotubes being incorporated into PI, crystalline phase of PI in the vicinity of carbon nanotubes is favourably oriented along the length of nanotubes forming an ordered structure due to non-covalent interactions between the PI chains and graphene surface of carbon nanotubes. The CNTs and PI crystallites can form synergistic “rigid constraints” to connect the disordered amorphous PI nano regions. Therefore, amorphous nanophase distribution in the matrix might be influenced by the local crystalline phase. The morphological transformation of PI caused by nanotubes depends on the degree of dispersion and concentration of nanotubes.

Based on our experimental results, we propose a model for morphological transformation in the PI as shown schematically in Fig. 9. For a well-dispersed CNT–PI system, individual and homogenous dispersion of nanotubes results in better distribution of crystalline regions in the PI matrix. The total crystalline phase in the matrix volume can be divided into the bulk polymer crystallites and the lamellar crystallites induced by nanotubes, which are separated from each other by amorphous part of the matrix. By taking into account the interfacial interactions between PI polymer chains and the surface of nanotubes as shown previously by FT-IR and UV-vis spectroscopic results, it is believed that PI polymer chains adopt favourable conformation adjustment and orientation transformation to interact with CNTs. PI polymer chains bound to the nanotubes initiate the formation of lamellar crystalline structures which are different from those in the bulk polymer. The neat semi-crystalline PI contains homogeneous stars-like crystallites only (Fig. 9a), while CNT/PI nanocomposites exhibit a heterogeneous crystalline phase distribution due to the different crystalline morphology in the matrix and near the nanotubes. At very low nanotube concentration, the nanotube dispersion is sparse, and the average inter-tube distance is large. The corresponding crystalline phase induced by nanotubes occupies only a small part of total crystalline phase, and therefore, the crystalline entities around the nanotubes does not overlap with each other. In this case, the orientation of lamellae surrounding nanotubes is substantially random, and the nanocomposites exhibit a failure in cohesive/adhesive mode. When the content of carbon nanotubes is increased to 0.5 wt%, the crystalline morphology of the matrix is dominated by the nanotube-induced crystalline morphology, and consequently, the crystalline entities tend to overlap with each other (Fig. 9b). When the nanotube content continues to increase, the crystalline entities in the matrix volume are almost completely composed of nanotube-induced crystallites. The average inter-tube distance decreases to a critical value, below which the overlap of crystalline entities induces locally oriented crystalline lamellae (Fig. 9c). Further increase of nanotube contents, in addition to the CNT aggregation (as shown in Fig. 1), the overlap of these crystalline entities induces crystalline lamellae with preferred orientation. The preferentially oriented crystalline lamellae percolate throughout the whole nanocomposites, which may provide a path where plastic deformation propagates and absorbs fracture energy (Fig. 9d). Overlap of crystalline entities around neighbouring CNTs induces preferred orientation of crystalline layers with non-covalently bonded planes, or in other words, carbonyl–carbonyl dipolar attraction planes as shown schematically in Fig. 10. The carbonyl–carbonyl dipolar attraction planes are reported to be parallel to the polymer–nanotube interface.^29,30 Morphological changes of the fracture surface with respect to the CNT amounts in Fig. 7 also suggest that the failure mode of CNT/PI nanocomposites is transferred from cohesive/adhesive to adhesive mode. It is reported that non-covalently bonded planes such as the dipole–dipole interaction plane in polyimide or the hydrogen bonded planes in rubber-modified polyamides have the low slip resistance.³¹ Therefore, the CNT/PI nanocomposites display sharp increase in plastic deformation when the preferred orientation of crystalline lamellae percolates throughout the whole nanocomposites (Fig. 11).

Fig. 9 Proposed model for evolution of crystalline phase in: (a) neat PI, CNT/PI nanocomposites at (b) 0.5 wt% CNTs, (c) percolation concentration, and (d) 0.75 wt% CNTs.

Fig. 10 Schematic diagram of PI chain–chain interaction in the crystalline entities.

Fig. 11 Schematics of proposed model for deformation mechanism of an idealized overlap of CNT-induced oriented lamellae under tensile test: top view (a) section view (b) before and section view (c) after deformation of nanocomposites.

The morphological transformation induced by nanotubes gradually changes the failure behavior of CNT/PI nanocomposites from cohesive/adhesive mode to adhesive mode. The change of failure mode attributes the key factors which influence strengthening and toughening semi-crystalline PI. The simultaneous strengthening and toughening of semi-crystalline PI can be attributed to high strength and toughness of CNTs, homogeneous dispersion of isolated nanotubes and accompanying nanotubes-induced homogeneous distribution of crystallites. Additionally, it is also attributed to the increased crystallinity in PI matrix, strong interfacial interaction between nanotubes and PI chains, and epitaxial formation of crystalline polymer coating on the nanotube surface. Most of all, the preferred orientation of crystalline lamellae induced by the overlap of these crystalline entities, which percolates throughout the whole nanocomposites, might play the most significant role.

4. Conclusions

In this research, an approach to explore novel CNT/PI nanocomposites with enhanced strength and toughness was developed. The isolated nanotubes and their homogeneous dispersion in the PI matrix volume can act as a nucleation site giving rise to the formation of homogeneous crystalline PI regions with improved crystallinity in the nanocomposite volume. The strong interfacial interactions between PI backbone and CNTs contribute to the effective load transfer from the PI matrix to nanotubes. The maximum strength and toughness of nanocomposites were obtained at the concentrations of 0.5 and 0.75 wt% CNT, respectively. The well-dispersed nanotubes and preferentially oriented local lamellae in the inter-nanotubes region are mainly ascribed to the dramatic increase in toughness. It is believed that there exists a percolation threshold of nanotube concentration between 0.5 and 0.75 wt%, at which the oriented crystalline lamellae percolate throughout the whole nanocomposites. The morphological transformation induced by carbon nanotubes has a significant impact on the failure mode transition of semi-crystalline PI, from cohesive/adhesive mode to adhesive mode, which consequently results in dramatic improvements in toughness. Further research should be carried on the effect of nanotubes size and PI conformation on the thickness of crystalline entities around the CNTs to optimize nanostructure for strengthening and toughening semi-crystalline PI. These findings are, however, of significance to design the macroscopic nanocomposites of novel and significant performance with controlled morphological transformation at a nano scale.

Acknowledgements

This work has been financially supported by National Natural Science Foundation of China (no. 51010005, 91216123, 51174063), the program for New Century Excellent Talents in University (NCET-08-0168), and Natural Science Funds for Distinguished Young Scholar of Heilongjiang Province. The project of International Cooperation supported by Ministry of Science and Technology of China (2013DFR10630) is also acknowledged. The authors thank Dr Iosif Gofman from Institute of Macromolecular Compounds of Russian Academy of Sciences for experimental assistance and valuable discussion on strain–stress measurement. XDL is also grateful to the financial support from the National Science Foundation of China (no. 51243005). MHL thanks to the support from Bilateral International Collaborative R&D Program (Project number: GT-2009-CL-OT-0058, Ministry of Knowledge & Economy, Republic of Korea).

Notes and references

1. S. Iijima, Nature, 1991, 354, 56.
2. P. Calvert, Nature, 1999, 399, 210.
3. C. Mercader, C. Jaillet, N. Lachman, C. Zakri, M. Maugey, H. D. Wagner and P. Poulin, Macromol. Rapid Commun., 2011, 32(24), 1993.
4. P. Pötschke, S. Pegel, M. Claes and D. Bonduel, Macromol. Rapid Commun., 2007, 29(3), 244.
5. J. N. Coleman, U. Khan, W. J. Blau and Y. K. Gun'ko, Carbon, 2006, 44(9), 1624.
6. R. Andrews and M. C. Weisenberger, Curr. Opin. Solid State Mater. Sci., 2004, 8(1), 31.
7. J. H. Lee, Y. Lee and J. D. Nam, Macromol. Rapid Commun., 2008, 30(1), 52.
8. V. E. Yudin, V. M. Svetlichnyi, A. N. Shumakov, D. G. Letenko, A. Y. Feldman and G. Marom, Macromol. Rapid Commun., 2005, 26(11), 885.
9. J. Chen, R. Ramasubramaniam, C. Xue and H. Liu, Adv. Funct. Mater., 2006, 16(1), 114.
10. P. C. Ma, N. A. Siddiqui, G. Marom and J. K. Kim, Composites, Part A, 2010, 41(10), 1345.
11. S. S. Wong, E. Joselevich, A. T. Woolley, C. L. Cheung and C. M. Lieber, Nature, 1998, 394, 52.
12. R. J. Chen, Y. Zhang, D. Wang and H. Dai, J. Am. Chem. Soc., 2001, 123(16), 3838.
13. B. P. Grady, Macromol. Rapid Commun., 2009, 31(3), 247–257.
14. T. Ogasawara, Y. Ishida, T. Ishikawa and R. Yokota, Composites, Part A, 2004, 35(1), 67.
15. R. Srivastava, S. Banerjee, D. Jehnichen, B. Voit and F. Böhme, Macromol. Mater. Eng., 2009, 294(2), 96.
16. L. Qu, Y. Lin, D. E. Hill, B. Zhou, W. Wang, X. Sun, A. Kitaygorodskiy, M. Suarez, J. W. Connell, L. F. Allard and Y.-P. Sun, Macromolecules, 2004, 37(16), 6055.
17. L. An, Y. Pan, X. Shen, H. Lu and Y. Yang, J. Mater. Chem., 2008, 18(41), 4928.
18. V. E. Yudin, A. Y. Feldman, V. M. Svetlichnyi, A. N. Shumakov and G. Marom, Compos. Sci. Technol., 2007, 67(5), 789.
19. S. Kumar, T. D. Dang, F. E. Arnold, A. R. Bhattacharyya, B. G. Min, X. Zhang, R. A. Vaia, C. Park, W. W. Adams, R. H. Hauge, R. E. Smalley, S. Ramesh and P. A. Willis, Macromolecules, 2002, 35(24), 9039.
20. Z. Bartczak, A. Argon, R. Cohen and M. Weinberg, Polymer, 1999, 40(9), 2347.
21. M. I. Bessonov, M. M. Koton, V. V. Kudryavtsev and L. A. Laius, Polyimides – Thermally Stable Polymers, Plenum Publishing Corp., New York, 1987.
22. M. K. Ghosh and K. L. Mittal, Polyimides: Fundamentals and Applications, CRC, New York, 1996, vol. 36.
23. B. Wunderlich, Prog. Polym. Sci., 2003, 28(3), 383.
24. B. Zhang, V. Bershtein and T. Sukhanova, J. Macromol. Sci., Part B: Phys., 2012, 51, 1794.
25. D. P. Heberer, S. Z. D. Cheng, J. S. Barley, S. H. S. Lien, R. G. Bryant and F. W. Harris, Macromolecules, 1991, 24(8), 1890.
26. K. Lu, N. Grossiord, C. E. Koning, H. E. Miltner, B. van Mele and J. Loos, Macromolecules, 2008, 41(21), 8081.
27. H. Harel and G. Marom, Acta Polym., 1998, 49(10–11), 583.
28. J. A. W. van Dommelen, D. M. Parks, M. C. Boyce, W. A. M. Brekelmans and F. P. T. Baaijens, J. Mech. Phys. Solids, 2003, 51(3), 519.
29. O. K. Muratoglu, A. S. Argon, R. E. Cohen and M. Weinberg, Polymer, 1995, 36(5), 921.
30. A. Reiser, L. Leyshon, D. Saunders, M. V. Mijovic, A. Bright and J. Bogie, J. Am. Chem. Soc., 1972, 94(7), 2414.
31. L. Lin and A. Argon, Macromolecules, 1992, 25(15), 4011.

---