Carbon fibers are manufactured from synthetic fibers through heating and stretching treatments, and a simplified schematic of this manufacturing process can be seen in Figure 1 (on p. 2). Polyacrylonitrile (PAN) and pitch are the two most common raw products used to produce carbon fibers. PAN is a synthetic fiber that is pre-manufactured and wound onto spools, and pitch is a coal-tar petroleum product that is melted, spun, and stretched into fibers. First, in the thermoset treatment, the fibers are stretched and heated to no more than 400° C. This step cross-links carbon chains so that the fibers will not melt in subsequent treatments. Second, in the carbonize treatment, the fibers are heated to about 800° C in an oxygen free environment. This step removes non-carbon impurities, and for PAN based fibers Table 1 (on p. 2) shows the fibers’ composition after different treatment steps. Third, the fibers are graphitized; this step stretches the fibers between 50 to 100% elongation, and heats them to temperatures ranging from 1100° C to 3000° C. The stretching ensures a preferred crystalline orientation (See Figure 2, on p. 3), which results in the desired Young’s modulus around 300-600 GPa. Finally, the last two treatment steps, surface treatment and epoxy sizing, are preformed to enhance the carbon fiber / epoxy bonding strength. Several different methods can be used in these last steps, but will not be discussed in this paper.
Figure 1. Schematic of PAN and pitch based carbon fiber manufacturing procedure.
(Source: A. R. Bunsell, Fibre Reinforcements for Composite Materials, Amsterdam, The Netherlands: Elsevier Science Publishers B.V., 1988, p. 90.)
Table 1. The composition of PAN fibers at different treatment steps.
(Source: A. R. Bunsell, Fibre Reinforcements for Composite Materials, Amsterdam, The Netherlands: Elsevier Science Publishers B.V., 1988, p 93)
Figure 2. Structural model for carbon fibers during graphitization.
(Source: A. R. Bunsell, Fibre Reinforcements for Composite Materials, Amsterdam, The Netherlands: Elsevier Science Publishers B.V., 1988, p. 120.)
During this quarter, I ran an unknown sample, a graphite powder sample, and several different carbon fiber samples in the x-ray diffraction (XRD) machine. The results of these samples are discussed below.
- Unknown Sample
- Graphite Powder Sample
- Carbon Fiber Samples
An unknown sample, consisting of a black powder, was place on a no back-round noise quartz plate and x-rayed; the x-ray diffraction diagram seen in Figure 3 (on p. 4) was produced. The indexing of this diagram can be seen in Table 2 (on p. 4), and the sample has a face-centered cubic structure with a basic unit cell length of 5.93 Å. This corresponds perfectly with lead sulfide (PbS), and its card is in Figure 4 (on p. 5).
Figure 3. X-ray diffraction diagram of unknown sample.
Table 2. Indexing of the unknown sample’s x-ray diffraction diagram.
Figure 4. Lead Sulfide (PbS) card.
A sample of graphite powder was analyzed in the XRD machine; this sample was mounted on an alumina oxide (Al2O3) plate. The graphite diagram can be seen in Figure 5 (on p.6), and the peaks from the Al2O3 plates (See Figure 6, on p. 6) are not included in the graphite indexing in Table 3 (on p. 7). Graphite powder has a hexagonal unit cell with dimensions a = 2.46 Å and c = 6.71 Å.
Figure 5. X-ray diffraction diagram of graphite powder sample.
Figure 6. X-ray diffraction diagram of alumina oxide (Al2O3) sample.
Table 3. Indexing of the graphite powder’s x-ray diffraction diagram.
I analyzed several different carbon fiber samples in the XRD machine. Two samples were prepared for me from carbon fibers imbedded in epoxy: 1) with the fibers lying along the sample surface (Sample 1), and 2) with the fibers’ ends "coming out of" the samples surface (Sample 2), and their x-ray diffraction diagrams can be seen in Figure 7 (on p 8) and Figure 8 (on p. 9), respectively. These were interesting results, especially since there are strong peaks in the middle of the large amorphous humps on both Sample 1 and Sample 2. I spend a lot of time trying to figure out the crystalline structure of the carbon fibers based on these graphs. The strong peaks in the middle of the humps suggested that an organized crystalline structure exists in an otherwise amorphous substance. Since some of the peaks might originate from the epoxy, I ran a sample with epoxy free carbon fibers lying down (similar to Sample 1) on an alumina oxide plate (See Figure 9, on p. 9) in order to eliminate potential epoxy peaks. However, it turns out that every peak on this x-ray diffraction diagram corresponds perfectly with the peaks of the alumina oxide (See Figure 6, on p. 6). The conclusion is that all the peaks on Sample 1 and Sample 2 originated from contaminants that were introduced during the sample preparation procedures. This leave only three carbon fiber humps: Sample 1 has a hump centered over 2q = 26.6° that corresponds to the hkl = 002 index; and Sample 2 has one hump centered over 2 q = 19° that is midway between hkl = ¼ ¼ 0 and hkl = ½ 00 indexes, and another hump centered over 2 q = 43° that corresponds to the hkl = 100 index.
Figure 7. X-ray diffraction diagram of carbon fibers lying along the sample surface (Sample 1).
Figure 8. X-ray diffraction diagram of carbon fibers with the fibers’ ends "coming out of" the sample surface (Sample 2).
Figure 9. X-ray diffraction diagram of epoxy free carbon fibers with the fibers lying along the sample surface.
From the peak broadening, the crystallite size on Sample 1 can be determined. The width, B’, of the alumina oxide peak at 2q = 25.6° (See Figure 10, on p. 11) is first calculated.
B’ is subtracted from B of the carbon fiber hump centered over 2q = 26.6° .
The crystallite size, t, is then calculated from equation
where
Figure 10. X-ray diffraction diagram over the 2
q = 25.6° peak of alumina oxide.
I looked at Sample 1 and Sample 2 in a scanning electron microscope (SEM). Sample 1, with the fibers lying along the sample surface, did not give any visual information, and Sample 2, with the fibers’ ends "coming out of" the samples surface, can be seen in Figure 11 (on p. 12). The sample was so well prepared that no inside features of the carbon fibers were visible. Figure 11 c shows about 8% of the fiber’s cross-sectional area; however, there are no signs of the fiber’s interior structure.
(a) (b)
(c) (d)
Figure 11. Scanning electron microscope pictures of Sample 2 (a, b, and c), and Sample 1 (d).
Fibers that had failed in tension where more interesting to look at (See Figure 12, on p. 14), and they show more details of carbon fibers’ interior structure. It can be seen in Figure 12 (a) and (b) that the fibers’ surfaces begin to deform by rippling before they fail; unstretched fibers have a smooth surface (See Figure 11 d). However, these pictures did not capture the interior structure of the carbon fibers in the same way as previous SEM pictures have done (See Figure 13, on p. 15), which suggest that carbon planes are systematically organized within the fiber. The SEM pictures of the carbon fibers that I took have a more random structure. Each "ball" like end sticking up out of the fiber has diameters ranging between 500-4000 Å. This indicates that the model in Figure 14 and 15 (on p. 16 and 17 respectively) is more appropriate, and that the "balls" are the broken off ends of the microtexture seen in Figure 15 (b). Each microtexture string is constructed from sheets of aromatic carbon rings (like the ones seen in Figure 2) that are folded, twisted, and inter-linked with each other (See Figure 14 and 15 (a)). Since it is highly unlikely that these microtexture strings are as long as the fibers, the microtexture strings are probably inter-linked in a similar way as the carbon sheets, as seen in Figure 14 (b) and (c), in order to transfer tension loads throughout the fibers.
The flaw with the above model is that the carbon sheets are constructed from aromatic carbon rings. The XRD diagrams of these carbon fibers do not show any large peaks that would support the presence of highly organized crystalline structure; instead the large broad humps indicates that its structure is amorphous. After all, the raw product for the carbon fibers is PAN or pitch, which consist of large hydrocarbon chains, and it is too optimistic to think that the carbon atoms will rearrange themselves into perfect sheets of aromatic carbon rings during manufacturing. Instead it is more likely that these sheets have an amorphous structure where the carbon atoms are trying to form aromatic rings. This would explain that the crystallite size is only 9 Å, that carbon fibers are brittle—their maximum strain is 2 %, and their high tensile strength (which ranges between 3.5-6.5 GPa).
Figure 12. Scanning electron microscope pictures of carbon fibers that failed in tension.
Figure 14. Lamellar model; (a) model, (b) interlinking of aromatic layers, and (c) tensile failure.
(a) (b)
Figure 15. Microtexture model of carbon fiber: (a) top view, and (b) side view.