Figure 3 (a) presents the macro-morphologies of the wax-impregnated carbon fiber preforms with needle-punched unidirectional channels. It shows that the preform is properly cured by the wax, and the prefabricated round channels are fairly maintained neatly arranged. Figure 3 (b) shows the cross section of the preform with space-holders filled in the channels. At this stage, the space-holders are composed of SiO2 (major ingredients) and NaF (minor ingredients) according reaction (2) and the contents of the starting materials.
![](https://myfiles.space/user_files/58895_8739fc6c57c1c19a/58895_custom_files/img1602498884.png)
The space-holders exhibits a solid and dense morphology, and are well confined in the channels. Obviously, wax-impregnated carbon fibers prevent the space-holder particles trapped into the preform body beyond the channel. Based on this preform treating method, many other ingredients (E.g. phenolic resin-Si powder slurry for SiC formation) can be designed and combined into the final C/C composite with a columnar configuration. But this has been out of the scope of the present paper. Figure 3 (c) demonstrate the morphologies of preform after wax-removing. In this stage the NaF in the space-holders should have been largely dissolved into the boiling water, and thus the SiO2 content in the cylindrical space-holder increases. It clearly shows that the space-holders are still tightly embedded in the preform, which has become ventilated and clean and ready for densification.
Figure 4 (a) shows the photograph of the deposited state of the composites. After densification the outermost layer is covered with a bright white PyC crust. Figure 4 (b) reveals the morphology of the composite after removing the thin PyC crust. Arrayed columnar SiO2 firmly embedded in the C/C matrix; this composite itself can be regarded as a new type of C/C-SiO2 composite. After leaching out SiO2 (the space-holder), the unidirectional porous C/C composite S5, 2 is displayed in Figure 4 (c). The C/C matrix look dense and the cross-sectional inner wall of the pores appear smooth, indicating that the channels are properly preserved by the space-holders.
Figure 5 depicts the PLM microstructures of the porous C/C composites with different channel diameters and row-spacings. PyC is normally classified into three basic types based on the morphology of C/C composites under polarized light microscope (PLM): Rough laminar (RL), smooth laminar (SL) and optical isotropic (ISO) structure [11, 12]; or according to the extinction angle (Ae) under polarized light, the textures of PyC are quantitatively classified into isotropic PyC (ISO, Ae = 0°), low-textured PyC (LT, 0° < Ae ≤ 12°), medium-textured PyC (MT, 12° < Ae ≤ 18°) and high-textured PyC (HT, Ae > 18°) [13, 14].
Fig. 5 (a) shows the PLM photomicrograph of the composite with row-spacing of 5 mm and channel-diameters of 2 mm (S5, 2). Growth cones initialized from the surface of carbon fibers and disrupted Maltese extinction crosses with the Ae of 18.9° ± 0.6° are observed, characterizing a high texture or RL PyC matrix. The textures of S5, 2.5 [Fig. 5 (b)] and S5, 3 [Fig. 5 (c)] are similar as that of S5, 2, indicating that different channel-diameters with the row-spacing of 5 mm have little effects on the microstructures of the porous C/C composites. The grain size of the pyrolytic carbon of S4, 2 and S3, 2 is significantly refined compared with that of S5, 2, as evidenced in Figure 5 (d) and (e), respectively. This apparently due to preform densities of S4, 2 and S3, 2 are higher than that of S5, 2, S5, 2.5, and S5, 3, because the channels-row spacing in S4, 2 and S3, 2 is lower than that in S5, 2, S5, 2.5, and S5, 3. Apart from fine PyC grains with RL or high-textures, larger PyC grains (marked with arrow) with annular cracks and distinct extinction crosses that characterize SL or low- to middle-textured PyC also present in Figure 5 (d). PyC with two or more matrix structures is called a banded structure [12]. Many factors, such as precursor gas used, temperature, partial pressure, area/volume (A/V) value, etc., affect the PyC texture [12, 15], among which, the volume related concentration of active sites (A/V ratio) directly influence the kinetics of deposition as well as the texture of PyC matrix [16]. It is easy to see that the larger PyC grains mainly appear in areas where carbon fiber is relatively sparse.
Figure 6 shows the compressive strengths of the porous C/C composites for P// and P ⊥. For porous C/C composites with fixed channel-row spacing, both P// and P⊥ decrease as the channel-diameter increases [Fig. 6 (a)]; for the porous samples with a fixed channel-diameter of 2 mm, the P// and P⊥ decrease with the decreasing of channels-row spacing [Fig. 6 (b)]. This reflects that the strength of porous materials decreases with increasing of the overall porosity [17-19]. The P// for S5,2, S5, 2.5 and S5, 3 are 104±6.6MPa, 75.8±4.3 MPa and 43.5±3.8 MPa, respectively; P⊥for the corresponding samples are 72.1±3.2 MPa, 57±3.5 MPa and 35.6±2.8 MPa, respectively. Strength anisotropy arise mainly due to the bearing area percentage of the cross-section for P// is larger than that of P⊥[8]. The P// for S5,2, S4, 2 and S3, 2 are 104±6.6 MPa, 45.1±2.7 MPa and 39.5±2.3 MPa, respectively; P⊥for the corresponding samples are 72.1±3.2 MPa, 35.3±2 MPa and 33.7±2.1 MPa, respectively. The decrease in strength is consistent with the decrease in density or increase in porosity. The average densities of C/C matrix part of S5,2, S5, 2.5, S5, 3, S4, 2 and S3, 2 are 1.71±0.04, 1.69±0.03, 1.66±0.04, 1.58±0.05, and 1.54±0.05 g/cm3, respectively. Figure 6 (a) and (b) clearly shows that densities of matrix of the unidirectional porous C/C composites decrease with the increasing of channel-diameter or decreasing of channels-row spacing. This is because the decrease in the channels-row spacing or increasing of channel-diameter of the same volume composites will relatively increase the number or volume of channels, causing the increase in porosity and reducing the density of the composite matrix.
Increasing of channel density (i.e., decreasing channel-row spacing) or increasing of channel-diameter result in an increasing density of carbon fiber preform outside the channels part, and, in turn, lead to a more lengthy and complicated gas flow path through the preform [20]. Thus, under identical CVI densification manner, less PyC is deposited in the higher density preforms within the same time.
Figure 7 (a) and (b) demonstrates representative load-displacement curves of the porous C/C composites for P// and P⊥testing, respectively. For the P// curves, S5,2, S5, 2.5, and S4, 2 display a catastrophic rather than non-catastrophic failure mode with unimportant pseudo-plastic behavior, except that the failure load for S4, 2 is much lower than that for S5,2 and S5, 2.5. S5, 3, and S3, 2, however, show pronounced pseudo-plastic fracture behavior for both P// and P⊥ tests. It is meaning to compare the load-displacement curves of S5, 3, and S4, 2. The failure loads for both are comparable, but S5, 3 shows a linear elasticity at low stress followed by a long yield plateau, behaving more like a typical metal porous material [18, 21]; whereas S4, 2 exhibits a more brittle behavior characterized with a quick load-drop after peak value in the load-displacement curve. The slope of the linear modulus line of S3, 2 is apparent lower than other porous C/C composite samples, which should be because S3, 2 has the largest porosity.
Figure 8 shows SEM fracture morphologies of the porous C/C composites. The fracture features of S5,2 [Fig. 8 (a)] and S5, 2.5 [Fig. 8 (b)] are alike and characterized by relatively flat fracture surfaces with very short pullout fiber length, indicating that a strong interface bonding between the fibers and the matrix, and also reasoning that S5,2 and S5, 2.5 possess very similar compressing behaviors of relatively high strength and brittleness. Cracks occur at the matrix defect when brittle fracture begins and propagates, as shown by the red arrow in Figure 8 (a), and eventually lead to the composite fail with the increase of load. The fracture features of S5,3 [Fig. 8 (c)] also demonstrate a brittle fracture with short pulled-out fibers, suggesting that its pseudo-plastic behavior shown in Figure 7 (a) and (b) originates from the unidirectional pores. S4, 2 [Fig. 8 (d)] displays brittle fracture surface with delaminated fiber and PyC matrix. Whereas S3, 2 [Fig. 8 (e)] exhibits a compression fracture characterized with very long pulled-out fibers, indicating that a weak bonding strength between the fibers and the matrix, which results from the low density of S3, 2. Under the loading stress, the interface between fiber and matrix debonded easily. Combined with a high density of unidirectional pores, S3, 2 presents the lowest compression strength.