Page images

In Fitzer's work, subsequent pitch densification cycles increased the strength of the untreated SIGRAFIL HM material to almost 100 percent (calculated assuming no contribution of the matrix), whereas only a 40-percent strength realization was obtained from the nonsurface-treated HF material. A reasonable conclusion appears to be that even without surface treatment, the SIGRAFIL HF fibers reacted to form a strong bond with the phenolic novolac matrix, whereas the HM fiber did not. A similar effect had been reported by Thomas and Walker (ref. 103) working with phenolic resin subsequently densified with CVI.

It is reasonable to conclude from the discussed data that matrix precursor-fiber combinations that promote strong bonds (with or without surface treatment) will not produce a high-strength composite when measured parallel to the direction of the fibers.

The addition of graphite powder is often used to decrease the shrinkage tendency of phenolics during carbonization and to improve the carbon yield (ref. 106). Since the decrease in the thermal contraction should also produce a corresponding decrease in the internal stress generated in the matrix on cooling from carbonization temperatures, the strength of composites whose failure is matrix dominated should be improved. The results of Fitzer et al. (ref. 107) confirmed this when they demonstrated that the strength translation of HF fiberreinforced material increased from 40 to 60 percent when 50-percent graphite powders were added to the resin matrix before carbonization. We have shown in a previous section (Microstructure of CC Matrices) that (in the absence of fiber surface treatments) weak interfaces will occur naturally in some matrixfiber combinations, while strong interfaces will occur in others (refs. 67, 77, 78, and 95). Specifically, the nature and strength of the fiber-matrix bond appear dependent on the reactivity of the fiber (i.e., fiber type), the reactivity of the matrix, and the type and degree of fiber-surface treatment. Fitzer has shown that an improvement in properties is achieved with densification (ref. 107). He found that the strength of composites containing nonsurface-treated HM fibers could be increased from 10 percent after the initial carbonization to 90 percent of the theoretical strength after four densification steps, while the strength of the composite containing nonsurface-treated HF fibers increased to only about 40 percent. These data suggest that a reaction occurred naturally between the fiber and matrix in the case of HF fibers and a limited reaction occurred throughout the densification process for the HM materials. Perry and others (refs. 97, 108, and Herrick*) working with phenolics and different fibers found a maximum of only 40- to 50-percent translation of strength indicating some degree of reaction; however, since their preparation technique involved a number of graphitizations, other factors may have contributed to the low strength values.

*J. W. Herrick, Fiber Materials. Inc. Private communication.

Fitzer also found (ref. 98) that resins which form strong fiber-matrix bonds and also exhibit the highest shrinkage should be avoided if high translation of fiber strengths is required. He suggests the use of high carbon-yielding precursors that do not form strong bonds and that exhibit minimum shrinkage. Data presented indicate that pitches (rather than resins) will produce better translation of fiber strengths for fewer densification cycles.

Matrix Dominated Properties

The out-of-plane, off axis and transverse properties of unidirectionally reinforced composites depend mostly on the properties of the matrix or matrix-fiber bond. Specific properties of interest include transverse modulus, strength, and interlaminar shear strength. Using a simple model, an approximate value for the transverse modulus of a unidirectionally reinforced composite Eqj can be estimated

ECT = EmEj/{VmE} + VfEm) (5)

where Em and Ej are modulus values for the transverse direction of the composite; Vm and Vr are the respective volume fractions of the matrices. Although this expression neglects porosity and some other important effects (ref. 109), it can be used to reach the reasonable conclusion that in well-bonded composites, for the usual fiber volume fractions (50 to 60 percent), the transverse modulus will never be greater than about twice the modulus of the matrix. In the absence of constraint, transverse strength can never be greater than the strength of the matrix (and will degrade at a rate that depends on volume fraction of poorly bonded fibers). It is believed that in well-bonded composites the transverse strength will be independent of volume fraction of fibers and approximately equal to the strength of the matrix. Conversely, in a poorly bonded composite, a dependence on volume fraction will be observed and strengths appreciably lower than the strength of the matrix will be expected.

The results of Perry and Adams (ref. 97) who studied three types of resin matrices and different fiber types support the conclusion that larger values of transverse strength, modulus, and interlaminar shear are obtained from composites reinforced with well-bonded fibers. Although several fiber types were investigated, it appears that bonding is controlled by the reactivity of the matrix and the fiber. All properties improved with densification up to the fourth cycle and, in general, heating to graphitization, although degrading the transverse tensile strength caused no degradation in interlaminar shear values. Carbon-fiber-reinforced carbonized Borden SC-1008 phenolic exhibited a maximum transverse strength of about 870 psi (6 MPa) which degraded to about 700 psi (4.8 MPa) or less when heated to graphitization temperatures. A similar degradation from a maximum of 680 psi (4.7 MPa) (nongraphitized) to less than 500 psi (3.4 MPa) (graphitized) occurred for carbon-fiber-reinforced PPQ resin char and from 760 psi (5.2 MPa) to 450 psi (3.1 MPa) for the reinforced FF-26 resin char. In most cases, values of interlaminar shear varied between 3000 psi to 3600 psi (20.7 MPa to 24.8 MPa) for the SC1008, about 2100 psi (14.5 MPa) for the PPO, and about 2600 psi (19.9 MPa) for the FF-26. Interlaminar shear strengths for carbonized pitch densified materials appear to be superior to resin-char matrices. Values of 3915 psi (27 MPa) have been quoted for resin pitch-densified material (ref. 104).

In the late 1970's, Thomas and Walker (ref. 103) studied CVD-densified carbonized phenolic resin chars reinforced with a series of commercially available fibers, some of which were surface treated. Apart from naturally reactive fibers (high strength, low modulus), surface treatment improved the transverse flexural strength slightly while the interlaminar shear was increased significantly. Subsequent carbonization and redensification always produced composites with matrixdominated properties lower than the reinforced precursor; however, the superiority of the surface-treated material always remained after heat treatment. Unfortunately, as discussed before, a good fiber-matrix bond reduced the value of the fiberdominated longitudinal strength, and the fracture mode became very brittle. Depending on fiber type and the presence or absence of surface treatment, values of transverse flexural strength exhibited by the CC material varied from about 2030 psi (14 MPa) to 4350 psi (30 MPa); values of interlaminar shear varied from 2320 psi (16 MPA) to 3915 psi (27 MPa).

The apparent superiority in matrix-dominated properties of the CVD-densified material when compared to resin-densified material is supported by the work of Walker and Lee (ref. 110) who indicate that both transverse tension and interlaminar shear of 2-D CC are superior if the matrix is prepared by chemical vapor deposition. These authors indicated interlaminar tensile strengths of about 790 psi (5.4 MPa) and interlaminar shear strengths of 2340 psi (16.1 MPa) for carbonized resin precursor material compared with 1460 psi (10.1 MPa) and 2740 psi (18.9 MPa) for CVD material. In this work, it was pointed out that the superiority of CVD was only true for thin materials (0.1 in.) because reductions of both interlaminar tensile and interlaminar shear strengths occurred when the thickness of the CVD-densified material approached 0.5 in. This property degradation was shown to be related to the degree of difficulty in fully densifying thick sections using CVD.

The results reported in the literature are difficult to interpret because varying one parameter while keeping other factors constant is difficult. The matrix-dominated properties depend on fiber type, surface reactivity, matrix type (resin precursor, CVI microstructure, and pitch precursor), matrix reactivity, fiber volume fraction, processing condition, degree of densification, fiber orientation, and laminate thickness. By optimizing all these parameters, the matrix-dominated properties can be maximized; however, the greatest improvement in these properties is obtained by placing fibers in the appropriate directions.

Two-Dimensional Reinforcements

Two-dimensional composites can be fabricated from unidirectionally reinforced prepreg material in which individual plies or fabrics are stacked in precomputed orientations.

Fabric is the preferred material used to process CC material. Fabrics contain undulating fibers, however, and some loss in strength is realized when compared to simple cross-plied unidirectionally reinforced types. Typically, strengths and modulus of 2-D CC composites measured parallel to the fibers average 30 000 psi (206 MPa) and 16 Msi (110 GPa), respectively, while interlaminar tension and shear strengths are 1000 psi (6.9 MPa) and 1200 psi (8.3 MPa). These results represent an increase of more than 30 times in transverse strength and modulus depending on fiber volume fraction. The data showed a drop to half, or even a quarter, of the longitudinal tension strength values formerly exhibited by unidirectionally reinforced composites. Although the properties in the transverse direction have improved significantly, the properties through the thickness are still relatively poor (ref. 110).

Three-Dimensional Reinforcements

Three-dimensional composites have been designed in order to improve the poor, through-the-thickness, interlaminar tension and shear strength values exhibited by the 1-D and 2-D CC composites. Various methods of doing this include pattern nesting, angle interlock, and ply interlock as described by Walker and Lee (ref. Ill) and 3-D weaving by Ransone et al. (ref. 110). Methods for producing thick sections have also been discussed by McAllister and Lachman (ref. 4). For an orthogonal construction, Schmidt (ref. 112) quoted by McAllister and Lachman (ref. 4) obtained tensile strengths of 15 ksi (103 MPa) in the Z (through-thethickness) direction and 5 ksi (34.5 MPa) in both the X and Y directions on pierced fabric-reinforced CC. Girard (ref. 113) obtained 10 ksi (70 MPa) in all directions for CVD-densified material with slightly better properties obtained from resindensified CVD precursor material. Levine* reported values of 45 ksi (310 MPa) in the Z direction and 15 ksi (103 MPa) in both the X and Y directions of the 3-D materials that he examined. Correspondingly high values for modulus were also reported.

The absolute values of strengths for the 3-D materials cannot be directly compared with unidirectionally 2-D or other 3-D materials because the fiber volume fraction in each direction varies; however, the conclusion can definitely be made that placing fibers in the three orthogonal directions will improve otherwise poor, matrix-dominated properties.

Levine, A.: High Pressure Dcnsified Carbon-Carbon Composites—Part II: Testing. Extended Abstracts12th Biennial Conference on Carbon, American Carbon Soc., 1973. Unpublished paper.

The following points can be made in summary:

1. The elastic modulus of CC composites is dependent on the modulus of the carbon matrix and the modulus of the fibers.

2. Good bonding between the fiber and the carbon-matrix precursor will result in a better bonded CC composite.

3. Bonding depends on surface treatments, type of fiber, and type of matrix precursor.

4. Fibers subjected to higher processing temperatures (high modulus) generally do not bond as well as those subjected to lower processing temperatures.

5. Good bonding between the fibers and matrix results in a composite that fails in longitudinal tension at the failure strain of the matrix.

6. Good bonding tends to prestress the matrix and can cause failure to occur at strains lower than the failure strain of the unrestrained matrix.

7. High longitudinal strength is obtained by providing weak interfaces (or cracks) perpendicular to the path of the catastrophically propagating crack.

8. Better matrix-dominated properties are obtained from well-bonded materials.

9. The matrix-dominated mechanical properties of CVD-densified material are superior to that densified with phenolic resin.

10. Compared to resins and thick CVD-densified composites, pitch matrix materials produce higher translation of fiber strengths and higher interlaminar shear strengths.


We wish to acknowledge J. He, C. Kocher, and E. Forrester for their contributions in preparing the specimens for examination in the optical scanning and electron microscopes, J. Bazola for his help as Director of the Electron Microscope Center at Southern Illinois University at Carbondale (SIUC), and the staff of the Electron Microscope Laboratory of the Materials Research Laboratory, University of Illinois, Champaign-Urbana. We also wish to thank K. Rankin, who diligently typed this manuscript.

« PreviousContinue »