Carbon 1971. Vol. 9, pp. 321-325. Pergamon Press. Printed in Great Britain

ELECTRON MICROSCOPE STUDY OF THE MICROSTRUCTURE OF CARBON AND

GRAPHITE FIBERS FROM A RAYON PRECURSOR*

ROGER BACON and A. F. SILVACGI Union Carbide Corporation, Parma Technical Center, Parma, Ohio 44130

(Received 1 September 1970)

Abstract-The microstructure of carbon and graphite fibers (including high modulus graphite fibers) made from a rayon precursor has been studied by both transmission and surface replica electron microscopy. On the ultrafine scale, hi h modulus graphite fibers contain a honeycomb of axially oriented micropores, - 60 f apart, separated by long, ribbon-like layer packets of turbostratic graphite. On a larger scale, the graphite fibers consist of bundles of fibrils each - 5OOA in diameter. Failure of the bonds between these fibrils is believed to be responsible for the grainy appearance of fracture surfaces and for the tendency of the fibers to split longitudinally. The fibrils apparently have a latent existence in the “carbon” fibers prepared at a lower heat treatment temperature, and they are probably derived from the rayon precursor.

1. INTRODUCTION graphite fibers from both rayon [2] and poly-

The microstructure of low modulus graph- acrylonitrile [3] precursors have also been ite fibers, prepared by pyrolysis of rayon examined by electron microscopy [J-7]. precursor fibers, was briefly investigated in In all cases, a much more highly oriented an earlier article [l]. A moderate degree of fibrous texture is observed than in the case preferred orientation of graphite layer planes of the low modulus fibers. The “grain sizes” parallel to the fiber axis was observed by X- of these fibers are remarkably similar, all ray scattering and found to be derived from falling in the range 50-lOOA. the preferred orientation present in the Johnson and Watt [6] and Badami et al. starting material. Transmission electron [7] have found evidence for a fibrillar struc- microscopy of thin sections of individual tural unit, 250-1OOOw in diameter, in high filaments revealed a slightly oriented fibrous modulus carbon fibers made from poly- texture which influenced their behavior acrylonitrile precursors. We have found a during microtome sectioning. This fibrous similar structure in graphite fibers made texture was attributed to the formation of from rayon precursors [5]. long, narrow graphite layers which form the This paper presents a detailed description bounding planes of longitudinally oriented of the microstructure of both low and high prismatic pores. The “grain size” (or pore separation distance) was under lOOA.

modulus carbon and graphite fibers made from a rayon precursor. Both the internal

The more recently available high modulus structure and the external surface mor- phology have been studied, the latter by

*This work was supported in part by the Air means of surface replicas.

Force Materials Laboratory, Nonmetallic Materials The starting material for all fibers used for Division, Fibrous Materials Branch. this study was Villwyte rayon (IRC Fibers

321

322 R. BACON and A. F. SILVAGGI

Company). In accordance with common High temperature “graphite” fibers, on the industrial usage, the term “carbon” will refer other hand, could be cut with relatively little to fibers heat treated to a temperature damage, providing one did not cut “across- typically in the range lOOO-1500X, whereas the-grain” in a well-oriented sample. Poorly the term “graphite” will refer to fibers heat oriented material (which had been graphi- treated to a temperature typically in the tized without stress) could be cut nearly as range 2000-3000°C. High modulus graphite well in transverse sections as in longitudinal fibers were prepared by a process of “stress- sections. Highly oriented high modulus fibers graphitization, ” i.e., the fibers were stretched yielded undamaged sections only in longitu- during graphitization. Low modulus graphite dinal cutting; transverse sections were very fibers were graphitized without stress. difficult to cut.

2. EXPERIMENTAL PROCEDURES

A. Sample preparation for transmission electron microscopy

Thin sections of carbon or graphite fibers were prepared for transmission electron microscopy by cutting in a microtome. Fibers in yarn form were embedded in plastic by placing them in a gelatin capsule and filling the capsule with a mixture of 20 parts methyl methacrylate and 80 parts butyl methacrylate. The mixture was polymerized (sometimes with the aid of a catalyst) by heating overnight at 40-60°C. The embedded sample was then trimmed with a sharp blade preparatory to sectioning. The fibers were oriented in such a way as to permit cutting either transverse sections or longitudinal sections. A thermal advance ultramicro- tome, LKB Model 4800 Ultratome, was employed, with a 51” diamond knife set at a clearance angle of 4-6”. Sections approxi- mately 40081 thick were obtained by using a cutting speed of 2 or 5 mm/set.

Whole, relatively perfect sections could be obtained only if the material had been heat treated at a temperature higher than -lSOO”C, i.e., “graphitized.” Lower tempera- ture “carbon” fibers, such as Union Carb- ide Grade VYB, could be sectioned, but each section broke into many small “chips” which, with luck, could often be retained in proper relative position, like the pieces of a jig-saw puzzle. The internal structure of “carbon” fibers was quite uniform, and the material seemed to be glass-like, i.e., hard and brittle.

Undamaged sections may be obtained from a brittle material by using a microtome only if an extensive crack front forms ahead of the knife edge, preventing large bending strains which would result in the cracking of the section into small pieces. Undamaged sections may be obtained easily when “graph- ite” fibers are cut in longitudinal sections (i.e., “with-the-grain”), but they are not obtained from well-oriented fibers cut in transverse sections. The fact that good transverse sec- tions have been obtained from poorly oriented “graphite” fibers suggests the existence of microscale-level plasticity. Evidence for such plasticity will be given in Section III-D.

Samples for transmission microscopy were also prepared by grinding in a mortar and pestle. Resulting fragments were nonuniform in thickness, and only one example (Fig. 2) is shown in this article.

B. Preparation of surface replicas

Surface replicas were obtained from exter- nal surfaces of the fibers as well as from trans- verse fracture surfaces. The primary replica was obtained by using cellulose acetate tape (Faxfilm) which had been presoftened with acetone. A carbon replica was then obtained by evaporating a carbon film onto the tape and dissolving away the tape in acetone.

Replication of lateral surfaces was achieved by placing the fibers on the presoftened tape and removing them after drying. Prior to evaporating the carbon film onto the tape, the latter was shadowed with platinum in a direction 30” from the fiber axis.

Fig. 1. Microtome section of low modulus “carbon” modulus, E. -

fiber (heat treated below 15iWC, Young’s 6 x 10’ psi). Cutting direction horizontal. cracks parallel to knife edge.

Fig. 2. Fragment of ground low modulus “graphite” fiber (heat treated above 25OO”C, Young’s modulus. E. - 6 X 10” psi).

[Facing page 322 ]

Fig. 3. Microtome sections of high modulus “stress-graphitized” fibers. Cutting direction horizontal. (a) E = 60 X lo6 psi, longitudinal section. (b) E = 90 X lo6 psi, longitudinal section. (c) E = 50 X lo6 psi, transverse section. In (c) the section is cracked, with the cracks roughly parallel to knife edge; within the fragments one sees an irregular honeycomb arrangement of

pores in cross-section, spaced - 60-90A apart.

Fig. 4. Replicas of external fiber surfaces. (a) Rayon precursor, “Villwyte” (IRC Fibers Company). (b) Low modulus “carbon” fiber. (c) Low modulus “graphite” fiber; note surface ridges. (d) High modulus “stress-graphitized” fiber, E = 90 X 10” psi; note surface

ridges are aligned.

Fig. 5. Replicas of transverse fracture surfaces showing also a portion of external fiber surface. (a) High modulus “stress-graphitized” fiber, E = 60 X 10” psi. (b) Low modulus

“graphite” fiber (c) Low modulus “carbon” fiber.

Fig. 6. “Broken” microtome sections of high modulus “stress-graphitized, ” fiber showing

-‘s()() ,A diameter fibAs partially or completely separated. E > 75 X 10” PSI. (a) Attempted transverse section. (b) End of longitudinal section.

Fig. 7. Optical photomicrographs of high modulus “stress-graphitized” fiber (E = 65 X 10” psi) in two successive stages of fracture, showing tendency to split longitudinally. Filament diameter

-6pm.

ELECI‘RON MICROSCOPE STUDY OF CARBON AND GRAPHITE FIBERS 323

Transverse fracture surfaces were repli- cated by embedding the end of a bundle of fibers in the tape and pulling them out after drying. A certain fraction of the embedded filaments was thus broken in tension, and the process was repeated until virtually all filaments had been fractured. A final replica was then made. No platinum shadowing was required for these replicas.

All electron micrographs were taken with an R.C.A. EMU 3B electron microscope operating at either 50 or 100 kV.

3. RESULTS A. Internal structure of carbon and graphite fibers

Attempts to obtain microtome sections of low modulus “carbon” fibers (heat treatment temperature under 1500°C) resulted in a series of thin chips (Fig. 1). Evidently the material is hard and brittle, and no extensive crack front formed ahead of the knife edge. The internal structure is relatively feature- less; no pores are resolvable. Measurement of the pore-size distribution by nitrogen desorption shows that very fine pores pre- dominate. X-ray studies[8] show that this material is highly disordered and that the average graphite-like layer diameter is only - 30 A.

When carbon yarn is graphitized by heat- ing to - 2000°C or higher, the internal structure becomes highly textured (Fig. 2). A well-defined pore structure develops. Porosimetry measurements of a heat treat- ment series (1500”-2800°C) showed that the graphitization process results in a reduction in the number of small pores present in “car- bon” fibers and in the development of larger pore sizes[9]. As these larger pores are formed, they also become closed, i.e., inacces- sible even to helium. This pore structure has been studied quantitatively and in detail by small angle X-ray scattering techniques by Perret and Ruland [lo].

The growth at high temperatures of the graphitic layers in carbon fibers[8] and in

other nongraphitizing carbons [ 1 l] has been studied by Ruland and Perret by wide-angle X-ray scattering techniques. The layers grow to - 60 8, width and probably hundreds or thousands of A length. At the same time, uniform, parallel stacking of layers occurs so that layer packets, although still “turbo- stratic [ 11, 121, are otherwise relatively “crystalline”. In transmission electron microscopy, these layer packets strongly diffract the electron beam in the (002) diffrac- tion mode; consequently, very good diffrac- tion contrast is provided. The dark streaks are due to layer packets (which constitute the walls separating the pores) oriented with their planes nearly parallel with the incident electron beam. The lighter areas are due to the presence of pores and of pore walls oriented with their layers more nearly per- pendicular to the electron beam.

The internal microstructure of graphitized (but unstretched) carbon fibers made from rayon tire yarns[l] appears to be essentially identical to that derived from Villwyte rayon (Fig. 2). The only obvious difference is the presence of elongated voids, - 500 A in diameter and axially oriented, which are known to be present in such precursor rayons as tire yarns and Fortisan (Celanese Cor- poration).

When carbon yarn is graphitized under stress to produce a high modulus fiber [2], the fiber structure becomes highly oriented, as in Fig. 3a. In this case, it is possible to identify individual pores extending several thousands of w in length. A more highly stretched, and therefore even higher modulus fiber, is shown in Fig. 3b. The pore structure is even more highly oriented.

An end-on look at this pore-and-pore-wall structure is shown in Fig. 3c. Since Fig. Yc is a transverse microtome section, it is not a perfect, continuous slice but, rather, one which contains many cracks. These cracks lie predominantly parallel with the edge of the advancing knife blade. Within each piece of undamaged material, however, the struc-

CARBON Vol. 9, No. 3-C

324 R. BACON and A. F. SILVAGGI

ture is clearly evident. The light areas (pores) texture of the fracture surface of a stress- are no longer long streaks but, rather, appear graphitized filament is shown in Fig. 5a. In as small dots in a generally dark matrix. The this micrograph (which was unavoidably structure apparently consists of an irregular warped and torn in several places during honeycomb of axially oriented pores, separ- ated laterally by 60-90 A.

preparation of the replica), the external lateral surface appears in the bottom of , ,

B. External surface structure of carbon and the picture, and the fracture surface (only

~a~hite~#s about haIf of which was replicated) consti-

The shape of the cross-sections of all of tutes the upper portion of the picture.

these carbon and graphite filaments is highly Frequent “bumps” and “dimples”, - 300 A

irregular. This shape is derived from that in diameter, are visible; they appear as dots

of the textile rayon precursor. The external which, at the magnification of the fig.

surface is &ted; Fig. 4a shows a surface are 4-l mm in diameter. We assume these

replica of a rayon precursor filament. The to be the broken ends of fibrils. A dominant

sharp ‘creases” or valleys between flutes feature of the fracture surface is the heavy,

appear as dark lines running parallel with dark, worm-like “lines.” These “lines” are

the filament axis. Ignoring these gross actually cleavage steps and other steep con-

features, one notes that the fiber surface is tours in a very irregular fracture surface.

relatively smooth, although slight undula- That this description is correct was made

tions can sometimes be seen running parallel very clear in an unpublished study by R.

with the filament axis. Sprague (formerly of these laboratories),

The generally smooth surface of the pre- in which steroscopic pairs of pictures of

cursor is preserved when the rayon is fiber fracture surfaces were taken. Striking three-dimensional views resulted and con-

carbonized (Fig. 4b). Surface flaws present h rmed in the precursor are apparently preserved,

the description of the surface features

also. When the fiber is ~aphitized, however, presented here.

a marked surface roughness develops (Fig. When the carbon fiber is graphitized with-

4c), which is ordered into well-aligned ridges out stretching, the internal and external

if the fiber is stretched during graphitization structural features are not highly aligned,

(Fig. 4d). The ridges appear to be separated as we have already noted in Fig. 2 and Fig.

by 300-500 A. They probably “develop” as 4c+ Th’ IS reduction in oriented texture is also

a result of the graphitization process, and we manifested by the fracture surface (Fig. 5b),

believe that they are the surface manifestation which is smoother and less dominated by 1

of an internal fibrillar structure which, in the arge cleavage steps than that of the stress-

case of high modulus (stress-graphitized) graphitized fiber.

The fibrillar structure is much less evident fibers, is highly aligned. The fibrillar texture seems to be latent in the carbon fiber (see

in the fracture surface of an ungraphitized “carbon”

Fig. 5c) and is probably derived from the fiber (Fig. 5~). The surface is

rayon precursor. Additional evidence for the generally smooth, with only occasional low

existence of 300-500 A fibrils is given in the cleavage steps. However, a bumpy or grainy

following sections. surface texture, with a grain size in the range 250-500 A, is still evident. The reality of this

C. ~ra~verse~~t~re surf~es ofcu~bo~ and grain structure (which is well resolved in the

g-rafihite$bers original photographs) has been confirmed by detailed comparisons between two successive

An apparent relationship between the replicas made from the carbon fiber fracture aligned surface ridges of Fig. 4d and the surface of Fig. 5c.

ELECTRON MICROSCOPE STUDY OF CARBON AND GRAPHITE FIBERS 325

Further evidence for the existence of fibrils in very high modulus graphite fibers is shown in Fig. 6. These pictures show some of the debris resulting from fiber microfracture which occurred during attempted micro- tome cutting of a transverse sect.ion (Fig. 6a) and of a lon~tudinal section (Fig. 6b) of the same fiber.

1. The ultrafine microstructure of graphi- tized rayon fibers consists of a honeycomb of micropores, - 60-90 A apart, separated by long, ribbon-like layer packets of turbo- stratic graphite. This structure is highly aligned in the case of stress-graphitized, high modulus fibers.

The upper portion of Fig. 6a shows fibrils - 500 A across separated from each other. Sometimes severe bends, approaching 90”, can be seen in the fibrils.

Figure 6b shows the end of a longitudinal section. The slice terminates in rounded nodules, which appear to be the ends of - 500 A fibrils. The upper right hand corner of Fig. 6b shows fibrils which have been torn off from the main slice. These fibrils, severeiy bent and twisted, evidently had undergone considerable plastic deformation. Evidence of inelastic behavior during bending of carbon and graphite fibers has been reported by Williams et al.[ 131.

2. On a larger scale, the structure of these fibers consists of bundles of fibrils - 500 8, in diameter. Failure of the bonds between these fibrils is responsible for the grainy appearance of transverse fracture surfaces and for the frequently noted occurrence of longitudinal cleavage or splitting during fracture of high modulus graphite fibers,

Acknowledgements-The authors are grateful to Messrs. W. H. Smith and B. D. Middleton for providing optical photomicrographs of split high modulus fibers (Fig. 7).

In both Fig. 6a and Fig. 6b, the ultrafine structure of oriented micropores and pore walls is evident within the librils.

REFERENCES

E. Macrofracture of high modulusjbers

The influence of the fibrillar structure on the topography of transverse fracture sur- faces in carbon and graphite fibers has been discussed in Section IIIC. The influence of this structure on microfracture of microtome sections of high modulus fibers was des- cribed in Section IIID. Figure 7 shows optical photomicrographs of a high modulus fila- ment in two successive stages of fracture induced by pressing on the filament with a blunt tool. A marked tendency exists for longitudinal splitting over distances of several filament diameters. Thus, the ori- ented fibrillar structure apparently in- fluences fracture of these fibers on a scale of several microns as well as on a submicron scale.

5.

6.

7.

8. 9.

10.

Il.

12. 13.

R. Bacon and M. M. Tang, Cnrbon 2,221 (1964). R. Bacon and W. A. Schalamon, A@ Ydym.

Symp. No. 9,285 (1969). W. Watt and W. Johnson, Ap$. Polym. Symp. No. 9,215 (1969). R. Bacon, A. A. Pallozzi and S. E. Slosarik, Proc. 21st Annual Meeting, Reinforced Plas- tics Division, Sot. Plastics Ind., Chicago, Ill., Feb. 8-10, 1966. R. Bacon and A. F. Silvaggi, Paper No. 147, Abstracts, Eighth Conf. on Carbon (1967), Gar&tr 6.199 (1968). W. Johnson and W. Watt, Nature, Lond. 215, 384 (1967). D. V. Badami, J. C. Joiner and G. A. Jones, izi;zture, Lond. 215,386 (1967). W. Ruland, private communication. R. J. Bobka and R. Bacon, unpublished data. R. Perret and W. Ruland, J. appl. Cryst. 2, 209 (1969). R. Perret and W. Ruland, j. a#. Ctyxf. 1, 257 (1968). B. E. Warren, Pkys. Rev. 59,693 (1941). W. S. Williams, D. A. Steffens and R. Bacon, J. ‘4ppl. Pky. 41.4893 (1970).