Page images
PDF
EPUB
[subsumed][subsumed][ocr errors][subsumed][merged small][merged small][merged small][merged small][merged small][merged small]

Figure 8. Stirrers used to vary microstructure of pitch-based carbon fibers (refs. 16 and 17).

Hamada also measured the transverse magnetoresistance Ap/p of these fibers. As the magnetic field increases, a highly graphitized fiber shows a positive and increasing Ap/p, while a poorly graphitized or turbostratic fiber has a negative and decreasing Ap/p. From these studies, he found that the radial microstructure showed a high degree of graphitization and that microstructures produced with stirring were turbostratic.

The layer planes in Hamada's fibers with radial microstructures (refs. 16 and 17), similar to those in high-modulus Thornel fibers, appear to be more perfectly oriented to the fiber axis and less folded than the layer planes in the Carbonic fibers Endo studied (ref. 15). However, the most significant result of Hamada's work was the demonstration that interrupting the preferred flow pattern during mesophase extrusion modifies the fiber microstructure, resulting in a smaller average graphite crystal size, a larger layer plane spacing, and a lower degree of 3-D order or graphitization.

Effect of Microstructure on Fiber Properties

Currently, neither PAN nor pitch precursor fibers develop balanced physical properties upon heat treatment. While PAN-based fibers dominate the market in high tensile strength applications, their tensile modulus is much lower than that of pitch-based carbon fibers. Conversely, current pitch-based fibers are the

predominate fiber used in high stiffness applications, but their strength normally is less than that of PAN-based carbon fibers. An analysis of the microstructure in terms of brittle fracture explains why the tensile properties of PAN-based and pitch-based carbon fibers differ and the reason for their imbalance in properties. Because carbon fibers fail by brittle fracture, this analysis also will provide insight as to how much, and how readily, the physical properties of these fibers can be modified.

Effect of Microstructure on Tensile Properties of Carbon Fibers

Assuming that the brittle fracture mechanism proposed by Reynolds and Sharp (see Johnson, ref. 13) applies, the results of microstructural studies might explain the differences in strength and modulus of pitch-based and PAN-based carbon fibers. Given the strength and modulus of any carbon fiber, the Griffith relationship (ref. 18) can determine the critical crack size.

[blocks in formation]

where σ is the ultimate tensile strength, E is the modulus, Ya is the apparent surface energy, and C is the critical crack length. The most generally accepted value of a is 4.2 J/m2; although higher values of this constant have been reported, they serve only to increase the critical crack size and, therefore, are not of interest in the limiting case.

PAN-Based Carbon Fibers

As was previously mentioned, Endo (ref. 15) found the ultimate tensile strength of Torayca M46 to be 2.4 GPa, the Young's modulus to be 450 GPa, and the average crystallite thickness to be 6.2 nm. According to equation (1), the critical crack length for this fiber would be 209 nm. Johnson (ref. 13) notes that the large difference between critical crack length and crystallite thickness is typical of PAN-based carbon fibers, except in regions of enhanced crystallization surrounding inclusions. Because commercial producers use a high-purity PAN solution for spinning the precursor fibers to minimize such inclusions, the chance of failure by exceeding this critical size in a single crystal is also minimized. Thus, if failure is to occur by the Reynolds and Sharp failure mechanism, the crack must propagate outside the initiating crystallite into neighboring crystallites.

The large interlayer spacing and lack of 3-D order in PAN-based carbon fibers lower the probability that such propagation will occur, and the folding nature of the crystallites will probably also hinder crack growth. Thus, the lack of inherent orientation of the precursor, which hinders the development of graphitic structures, actually serves as a crack-stopping mechanism, increasing the final fiber strength.

The Reynolds and Sharp failure mechanism may also explain the higher failure strains of PAN-based fibers. The combined effect of many small crystallites, which can fail and relieve the applied stress without causing a catastrophic failure, could yield this increased elongation.

Even the relatively low-tensile modulus of PAN-based carbon fibers is a direct result of its microstructure. Recall that the high modulus of carbon fibers is a direct result of strong bonds in the layer planes oriented parallel to the fiber axis. In PAN-based fibers, the layer planes are less oriented with the fiber axis than they are in pitch-based fibers. In addition, the low degree of graphitizability of PAN-based fiber implies that these planes are small. A lower modulus for PAN-based carbon fibers, therefore, would be predicted from crystallographic analysis and the microstructural model proposed by Johnson (ref. 13). Because this folded, turbostratic microstructure is created during initial fiber formation and is characteristic of many solution-spun polymers, major improvements in the modulus or graphitizability of PAN-based carbon fibers are unlikely unless a totally different spinning technique is used.

Pitch-Based Carbon Fibers

Using the data from Endo's study of Thornel and Carbonic pitch-based carbon fibers (ref. 15), the prediction from equation (1) shows that their critical crack lengths will be approximately 130 nm. Although this is smaller than the critical crack length estimated for the Torayca PAN-based fiber studied by Endo, it is still a factor of 10 larger than the average crystallite thickness. However, Endo found that high-modulus Thornel fibers had a high degree of 3-D order, and thus, a continuous medium for crack propagation. He also postulated that the flat-layered structure of Thornel fibers will propagate cracks more easily than the folded crystallites found in PAN-based fibers. If the crack must propagate outside the initiating crystallite for a fiber to fail, these observations would explain why the high modulus Thornel fibers showed inferior tensile strengths. On the other hand, Endo found the modulus of Thornel P120 to be nearly 83 percent of that theoretically possible. This extremely high modulus is a direct result of the nearly perfect orientation of the closely spaced layer planes found in the Thornel pitch-based fibers.

Endo's study (ref. 15) revealed that, although both the Carbonic and the Thornel pitch-based carbon fibers are melt-spun from a mesophase pitch precursor, they are quite different. The low degree of graphitization and the presence of crystallite folding appear to be responsible for the increased strength of Carbonic fibers. Endo also found that, as the X-ray diffraction results of these fibers become more similar to those of PAN-based fibers, the strength increases. Carbonic fibers also have significantly higher failure strains than Thornel fibers, a phenomenon that also might be explained as the cumulative effect of numerous noncatastrophic failures in misoriented crystallites.

Based on the similarities between Carbonic and PAN-based fibers, one would expect the Carbonic pitch-based fiber to have a lower modulus than the Thornel pitch-based fiber. However, while Carbonic HM80 is 46 percent stronger than Thornel P120, its modulus is only 5 percent lower. Evidently, a high degree of graphitization is not necessary to develop a high modulus. Instead, high preferred orientation, which is characteristic of a melt-spun fiber using a liquid-crystalline precursor, is largely responsible for the resulting fiber modulus.

Hamada et al. (refs. 16 and 17) also found that a lower degree of 3-D order leads to increased strength in pitch-based carbon fibers. He found that fibers with a nonradial, turbostratic microstructure, produced by disturbing the flow profile of the mesophase prior to extrusion, were significantly stronger than those with a radially oriented microstructure and the same modulus. Interestingly, the pitch-based fibers recently introduced by Nippon Steel Corporation and E. I. du Pont de Nemours and Company* have a random microstructure. This microstructure probably results in a lower degree of 3-D order within the fiber. Although their moduli are similar to those of the Amoco Performance Products, Incorporated, Thornel pitch-based carbon fibers, these new fibers show considerably higher tensile strengths (see fig. 9, refs. 19 and 20).

Although these fibers have a random microstructure when viewed perpendicular to the fiber axis, the graphite layer planes still have excellent alignment along the fiber axis. This orientation perpendicular to the fiber axis usually is determined during the flow through the extrusion capillary. Nazem (ref. 21) demonstrated one method of creating a random orientation during fiber formation with mesophase. By extruding mesophase through a round extrusion capillary that had a small lengthto-diameter ratio and contained a porous media, he prevented a stable, parabolic flow profile from developing during extrusion. The resulting fiber had a totally random microstructure perpendicular to the fiber cross section. The melt-spun fiber still developed the excellent modulus that is characteristic of pitch fibers because the drawdown of the molten mesophase after having been extruded through the capillary still oriented the mesophase molecules parallel to the fiber axis.

Edie et al. (ref. 22) also showed that changing the extrusion capillary can readily change the microstructure of the melt-spun mesophase fibers. In this study noncircular extrusion capillaries were used to melt-spin a mesophase precursor into noncircular, pitch-based carbon fibers. These fibers had a line-origin microstructure that emanated from the center lines of the lobes of the noncircular fibers. Even when the noncircular fibers were cooled slowly (permitting them to collapse into a circular shape before solidification), they retained a lobal microstructure, confirming that the microstructure was developed during flow through the extrusion capillary.

*Sato, K., Nippon Steel Corporation, Tokyo, Japan, personal communication, 1989.

[blocks in formation]

Figure 9. Mechanical properties of various pitch-based carbon fibers (Nippon Steel data (Sato, K., Nippon Steel Corporation, Tokyo, Japan, personal communication, 1989), Amoco data (ref. 19), du Pont data (ref. 20)).

Thus, pitch fibers appear to be a very tailorable material. Just as the nature of the PAN precursor enhances the final fiber strength, the liquid-crystalline nature of the pitch precursor gives the resulting fiber high modulus. It is even possible that precursor characteristics do not limit the strength of pitch-based fibers as severely as they do the modulus of PAN-based fibers. Studies have shown that disturbing the flow profile of the molten mesophase can hinder the development of long-range order across the fiber cross section and increase the final fiber strength.

Effect of Microstructure on Compressive Properties of Carbon Fibers

Often it is the compressive strength of the composite material, rather than the tensile strength, which limits its use in structural applications. However, the compressive strength of the composite is directly proportional to the compressive strength of the fibers themselves. A common and serious mistake is to assume that the compressive strength of a carbon fiber is equal to its tensile strength. As figure 10 illustrates, this is far from correct, especially for both pitch-based and PAN-based high-modulus carbon fibers (ref. 23). The ultimate factors determining the compression strength of a carbon fiber differ from those determining its tensile

« PreviousContinue »