Structural heterogeneity and its influence on the tensile fracture of PAN-based carbon fibers

Abstract

This communication reported an intuitive and convenient approach to detect the structural heterogeneity of carbon fibers by using Raman spectroscopy. As indicated by both the linear and mapping modes, the skin–core difference was enhanced with the rising temperature during the graphitization. This enhanced structural heterogeneity was observed to strongly influence the tensile fracture mode and the tensile properties of the graphitized fibers.

---

Carbon fiber is an important reinforcing material applied in advanced composites (e.g., metal, ceramic, polymer, carbon-based, etc.) due to its light weight, relative flexibility and excellent mechanical properties.^1–4 The properties of carbon fiber are largely determined by the fine structures featured by the turbostratic graphitic crystallites, amorphous regions, nanoscale microvoids as well as the skin–core structure.^1,5,6 Among them, the influence as well as the existence of skin–core difference within carbon fibers has been debated by Bennett and Johnson, who had reported no structural heterogeneity was found in the 1000 °C and 1500 °C-treated samples.⁷ However, the most majority information from many kinds of characterization techniques (e.g., electron microscopy, X-ray scattering and diffraction, Raman spectrum and dynamic mechanical analysis, etc.) supported the existence of structural heterogeneity.^6–11 Based on the ordinary point scanning or linear scanning mode, Raman spectrum was widely used in the characterization of the stress and structural conditions of carbon fibers.^5,12–15 In light of previous researchers' results, this work concentrates on the investigation of the skin–core structure and its transformation during the critical graphitization process using Raman techniques in multiple modes. The work will provide a more intuitive and convenient approach to detect the skin–core structure, and also bring us a better understanding on the topic how this structural feature influence the tensile fracture mode of carbon fibers.

Carbon fiber for this study was PAN-based T300B derived from company (Toray Co., Japan). The fiber had probably been heated up to ∼1300 °C in the carbonization process in factory, and subjected to extensional load during fiber drawing to align the graphitic layer structure parallel to the fiber axis.¹⁶ The as-received sample was de-resinated in 2-butanone through a Soxhlet extraction method to remove the coated resin before the experiment was performed.¹⁷ The graphitization of T300B fiber was carried out in a graphite element resistance furnace manufactured by Tongxin Electric Heating Apparatus Co., Ltd (Xi'an, China). The procedure for the process was to heat up a bundle of fibers to the desired temperature at 5 °C min⁻¹ in the helium atmosphere, and then held at that temperature for one hour to ensure the structural transformation is completely finished. In this study, four final heat treatment temperature (HTT) points (i.e., 1800, 2000, 2300 and 2500 °C) were chosen and used to note the corresponding graphitized samples.

The Raman experiment was performed on Raman spectrometer (Jobin-Yvon, LabRam HR Evolution). The wavelength of the exciting laser beam was 532 nm, the laser spot diameter was about 1 μm, and the spectra were recorded at 0.5 cm⁻¹ resolution. The samples were cut in liquid nitrogen and fastened onto the platform with the cross-section faceup. The exposed transverse sections were scanned in both linear scanning and mapping mode as illustrated in Scheme 1. For the continuous linear scanning mode, the spectrometer scanned along a straight line through the center of the cross-section and recorded the spectrum with an interval at about 1 μm. Spectra were taken for three times for the samples to confirm the reproducibility of the data. For the mapping mode, the whole cross-section observed in the microscope was chosen and scanned. The heterogeneous map was drawn using graphitization degree as the index parameter, which was obtained through a Lorentz fitting to the Raman data.¹⁸

Scheme 1 The schematic representation of the Raman measurement in the cross-section of carbon fiber. The gray areas indicate the scanning areas in the corresponding mode.

The tensile test was performed on the AG-1 universal material tester (Shimadzu Co., Ltd, Japan) based on GB/T3362-2005 “Test methods for tensile properties of carbon fiber multifilament”. The fracture samples were also collected for the cross-section morphology observation using field emission scanning electron microscope (SEM: JSM-6320F, JEOL). The bulk density was measured by flotation method based on ISO 10119:2002 “Carbon fiber-Determination of density”.

Similar as articles reported, the Raman spectra of our samples presented the main E_2g2 line near 1584 cm⁻¹ (G band) and additional lines at about 1360 cm⁻¹ (D band) and 1620 cm⁻¹ (D′ band) as shown in Fig. 1.^12,19 The intensity of G band near the skin was obviously higher than that near the core region. The intensity ratio, I_D/I_G, was likewise lower in the skin than that in the core region. These together implied the existence of skin–core structure in carbon fibers.^8,11 Three peaks (G, D and D′ band) were then fitted using Lorentz function for the Raman spectrum of carbon fiber.¹⁸ The structural heterogeneity can be further studied through the parameters obtained from the Lorentz fitting to the data as shown in Fig. 2. Referred as the degree of graphitization, A_D/A_G (the integral area ratio of D to G band) of skin was observed to be lower than that in the core for the heat-treated fibers. This difference was enhanced with the increase of HTT as shown in Fig. 2(a). Furthermore, the G band shifts were all smaller in the skin than those in the core regions as shown in Fig. 2(b). It had been reported that the shifts of G band had relevance with the microstrain within the graphite layers and the compressive strain was the most common strain mode for carbon fibers.^12,20 In this case, the skin–core structure can be interpreted as structural heterogeneity caused by the radial asynchronous relaxation of compressive stress within the crystallites during the heat treatment.

Fig. 1 The Raman spectra of the cross-section of 2500 °C-treated carbon fiber. The insect is the magnification of G band.

Fig. 2 The structural heterogeneity of carbon fibers treated at various temperatures as indicated by the values of (a) A_D/A_G and (b) G band wavenumber.

The mapped images using A_D/A_G as the index parameter might be more intuitive to show the structural heterogeneity. The as-received carbon fiber (Fig. 3(a)) with a graphitization degree at about 2.5 was not so heterogeneous as was expected. The micro-domains of structural difference rather than the skin–core heterogeneity were easier to be recognized. The graphitized fiber, however, developed into a heterogeneous structure with a dramatic skin–core difference. Fig. 3(b) indicates that the thickness of the skin is about 1–2 μm with an average graphitization degree at about 0.6.

Fig. 3 The Raman mapped images of (a) the as-received and (b) 2500 °C-treated carbon fibers. The color bars based on the values of A_D/A_G are given to the right of each image. The white blanks are singular points due to the zero signal area outside the cross-section of the fiber.

The coefficient of variance of A_D/A_G was introduced in the evaluation of the structural difference between the skin and core region. The values in Table 1 indicate that the heterogeneity as well as the mean degree of graphitization was enhanced with the increasing HTT. It is believed the non-synchronization of the recrystallization between the skin and core region is the main reason for the enhanced heterogeneity.⁶ With the application of high temperature, non-structural elements like N, O, H and other impurity elements begin to escape outward from the fiber. The recrystallization is thus taken place preferentially near the skin where the removement of non-carbon elements is easier and thorough. Meanwhile, new C–C bonds which give rise to the higher G peak generate in these regions as the sp³ carbons rapidly transform into sp² ones.²¹ The thermal expansion of the interlayer spacing further provides possible space for the strain relaxation and the crystalline rearrangement.²² However, the crystallized skin after that might become an obstacle for the outward diffusion and removement of the non-carbon elements in the core region, especially when the skin is densified with higher degree of graphitization (Table 1). In this case, the regions near the skin become well-crystallized while these in the core are less perfect in crystallinity.⁶ This radial difference as reflected by Raman spectra is the skin–core structure. It is believed that the structural difference might be reduced a lot through advanced technics control but is not likely to be eliminated during the graphitization.^6,8,13

Table 1 The parameters for carbon fibers treated at various temperatures

Sample series	As-received	1800 °C	2000 °C	2300 °C	2500 °C
a Graphitization degree was the mean value of AD/AG (the integral area ratio of D to G band) for the cross-sections based on the Lorentz fitting to the Raman data.b Heterogeneity index was the coefficient of variance of AD/AG for the whole cross-sections of carbon fibers.
Graphitization degree^a (a.u.)	2.57	1.70	1.27	1.17	1.05
Heterogeneity index^b (a.u.)	0.069	0.088	0.110	0.125	0.128
Carbon content (wt%)	93.6	99.2	99.5	99.7	99.8
Density (g cm^−3)	1.77	1.83	1.84	1.85	1.88
Tensile strength (GPa)	3.68 ± 0.02	1.99 ± 0.08	1.94 ± 0.06	1.83 ± 0.09	1.79 ± 0.06
Tensile modulus (GPa)	221 ± 3	226 ± 10	249 ± 13	275 ± 20	290 ± 9
Fracture elongation (%)	1.67 ± 0.02	0.77 ± 0.05	0.76 ± 0.04	0.68 ± 0.08	0.67 ± 0.06

---

The heterogeneous structure greatly affects the mechanical properties of carbon fibers as indicated by the SEM images of the fracture surfaces in Fig. 4. For the as-received fiber, the fracture surface was of great roughness and the fracture source, crack propagation extent as well as the ultimate fracture region could be clearly observed from the tensile fracture morphologies. The graphitized fibers, however, had changed in fracture mode which showed no specific fracture source. As observed from their fracture morphologies, the fracture was likely to start in the whole skin regions and then transferred inward to the core. The heterogeneous stress distribution (Fig. 2(b)) might have resulted in the easier breakage of the surface part. In fact the carbon fiber with lower band shifts or higher graphitization degree was observed to be more brittle.⁵ In this way the mechanical property of carbon fibers might be interpreted on the basis of heterogeneous stress distribution among the various parts of the fiber as suggested by Kobayashi et al.⁵ Owing to the changes in fracture mode as well as other structural factors (e.g., the crystalline, void and aggregation structures.^4,6,23), both the tensile strength and fracture elongation of the heat-treated fibers dropt a lot with the increase of HTT as shown in Table 1.

Fig. 4 The SEM images of the fracture surface of carbon fibers treated at various temperatures. The arrow indicates the fracture source during the tensile experiment.

Conclusions

Raman spectra were found capable and convenient in the characterization of skin–core structure of PAN-based carbon fiber. The spectra showed an intuitive result that the skin–core difference was enhanced during the graphitization. This enhanced structural heterogeneity was observed to influence the tensile fracture mode and the tensile property for the graphitized fibers. The results of this work suggest skin–core structure had very bad influence on the mechanical performance, and the whole manufacturing process should maintain rigorous inspections on this structural feature in view of structural control.

Acknowledgements

The authors wish to thank Mr Zhen Li and Shifeng Liu of Horiba company for the assistance in the Raman measurement.

Notes and references

1. W. Ruland, Adv. Mater., 1990, 2, 528.
2. Z. Liu, C. Tang, P. Chen, Q. Yu and W. Li, RSC Adv., 2014, 4, 26881.
3. Y. Chen, C. Hong and P. Chen, RSC Adv., 2013, 3, 13734.
4. W. Li, D. H. Long, J. Miyawaki, W. M. Qiao, L. C. Ling, I. Mochida and S. Yoon, J. Mater. Sci., 2012, 47, 919.
5. T. Kobayashi, K. Sumiya, Y. Fukuba, M. Fujie, T. Takahagi and K. Tashiro, Carbon, 2011, 49, 1646.
6. F. Liu, H. Wang, L. Xue, L. Fan and Z. Zhu, J. Mater. Sci., 2008, 43, 4316.
7. S. C. Bennett and D. J. Johnson, Carbon, 1979, 17, 25.
8. G. H. Zhou, Y. Q. Liu, L. L. He, Q. G. Guo and H. Q. Ye, Carbon, 2011, 49, 2883.
9. O. Paris, D. Loidl and H. Peterlik, Carbon, 2002, 40, 551.
10. D. J. Johnson and C. Frank, Philos. Trans. R. Soc., A, 1980, 294, 443.
11. X. Liu, C. Zhu, J. Guo, Q. Liu, H. Dong, Y. Gu, R. Liu, N. Zhao, Z. Zhang and J. Xu, Mater. Lett., 2014, 128, 417.
12. D. H. Li, C. X. Lu, G. P. Wu, Y. Yang, Z. H. Feng, X. T. Li, F. An and B. P. Zhang, J. Mater. Sci. Technol., 2014, 30, 1051.
13. D. F. Li, H. J. Wang and X. K. Wang, J. Mater. Sci., 2007, 42, 4642.
14. C. J. Su, A. J. Gao, S. Luo and L. H. Xu, Carbon, 2013, 51, 436.
15. H. Wang, Y. Wang, T. Li, S. Wu and L. Xu, Prog. Nat. Sci.: Mater. Int., 2014, 24, 31.
16. G. Wu, D. Li, Y. Yang, C. Lu, S. Zhang, X. Li, Z. Feng and Z. Li, J. Mater. Sci., 2012, 47, 2882.
17. Z. Cheng, L. Meng, D. Fan and Y. Huang, Synth. Fiber China, 2008, 10, 21.
18. A. Sadezky, H. Muckenhuber, H. Grothe, R. Niessner and U. Poschl, Carbon, 2005, 43, 1731.
19. N. Melanitis, P. L. Tetlow and C. Galiotis, J. Mater. Sci., 1996, 31, 851.
20. C. Galiotis and D. N. Batchelder, J. Mater. Sci. Lett., 1988, 7, 545.
21. F. Tanaka, D. Wolverson, T. Okabe, I. A. Kinloch and R. J. Young, Carbon, 2014, 66, 742.
22. C. Sauder, J. Lamon and R. Pailler, Carbon, 2004, 42, 715.
23. D. Li, C. Lu, G. Wu, J. Hao, Y. Yang, Z. Feng, X. Li, F. An and B. Zhang, J. Appl. Crystallogr., 2014, 47 DOI:10.1107/S1600576714020081.

---