Ablation of C/SiC-HfC composite prepared by precursor infiltration and pyrolysis in plasma wind tunnel

Carbon fiber reinforced silicon carbide-hafnium carbide (C/SiC-HfC) composite was prepared by precursor infiltration and pyrolysis process. Then, ablation behavior of C/SiC-HfC was evaluated in plasma wind tunnel. It was found that oxide layer formed during ablation significantly influenced the surface temperature. Formation of dense HfO2-SiO2 layer under low heat flux led to stable surface temperature. Silica (SiO2) on the surface was gradually consumed when heat flux increased, resulting in conversion of HfO2-SiO2 on the surface to HfO2. Converted HfO2 with high catalytic coefficient absorbed more energy, causing gradual increase in the surface temperature. Formed oxide layer was destroyed at high heat flux and high stagnation point pressure. After carbon fiber lost the protection of HfO2-SiO2 layer, it burned immediately, leading to surface temperature jump.


Introduction 
During the hypersonic flight of vehicles in the atmosphere, a large amount of heat is generated due to the friction between vehicles and the atmosphere, causing the surface temperature of the vehicles to exceed 2000 ℃ [1,2]. Thus, thermal protection systems (TPS) have become essential for hypersonic flights. Carbon fiber reinforced silicon carbide (C/SiC) composite ZrC [13,14], TaC [15,16], HfB 2 [17,18], and HfC [19,20]. Among them, HfC not only has the highest melting point (about 3890 ℃), but also its oxide HfO 2 has the highest melting point (about 2810 ℃) among the oxides of UHTCs [21,22]. Consequently, HfC was often introduced into C/SiC composites as an ideal UHTC to form modified C/SiC-HfC composite. In our previous study [20], the influence of catalysis on the ablation behaviors of C/SiC-HfC composite prepared by reactive melt infiltration (RMI) was revealed. However, the diffusion mechanism of volatiles in ablation process is still not clear, which is related to the microstructures of C/SiC-HfC composites. For the comparative analysis of the diffusion mechanism of different microstructures during ablation, HfC phase used for modification was introduced via precursor infiltration and pyrolysis (PIP) process in this study, because PIP process has advantages such as easy designability of microstructures, low preparation temperature, and high mechanical properties of final composite, and uniform distribution of modified UHTCs [23][24][25]. However, reaction of HfO 2 with carbon was involved in the formation of HfC from the precursor [26], which could lead to the damage of carbon fibers. Thus, SiC was first deposited on the preform by chemical vapor infiltration (CVI) to reduce the damage to fibers caused by PIP process. Consequently, C/SiC-HfC composite was prepared by CVI and PIP in this study.
The methods used for the evaluation of ablation performance of above mentioned materials mainly include laser beam [27,28], oxy-acetylene flame [17,25], arc-jet [29,30], and plasma wind tunnel [20,31]. Plasma wind tunnel technique offers the advantage of measuring the change of the surface temperature compared to laser beam and oxy-acetylene flame methods. More importantly, plasma wind tunnel can provide a pure plasma flow [32]. On one hand, plasma must be considered during hypersonic flights, because the high temperature during flight can lead to the generation of plasma [33]. On the other hand, pure plasma flow can avoid the influence of impurities in wind tunnel, such as copper in the arc wind tunnel, and thus it is highly desirable to study the ablation mechanism of C/SiC-HfC composite. Therefore, plasma wind tunnel was selected as the method for evaluating the ablation performance of C/SiC-HfC composite and investigating the diffusion mechanism of volatiles in this study.
The phases and microstructures of C/SiC-HfC composite before and after ablation were systematically analyzed. Based on these analyses, the ablation mechanisms of C/SiC-HfC composite in plasma wind tunnel were proposed.

Experimental
Three-dimensional needled carbon fiber preform with 35% fiber volume fraction was prepared by needlepunching method with T-300 TM alternatively stacked with weftless piles and short-cut-fiber webs. Pyrolytic carbon was deposited on the surface of fibers as interface layer by CVI process. Then, SiC was further deposited on the preform by CVI using methyltrichlorosilane as precursor to reduce the damage to fibers caused by subsequent processes. The density and open porosity of C/SiC after deposition of the above mentioned layers were 1.56±0.03 g/cm 3 and 38.0%±1.1%, respectively. The C/SiC preform was then machined into samples with 20 mm in diameter and 50 mm in length, as shown in Fig. 1(a). www.springer.com/journal/40145 C/SiC-HfC composite was prepared by PIP process based on the as-prepared C/SiC samples. In this process, liquid HfC precursor (Institute of Process Engineering, Chinese Academy of Sciences, Beijing, China) and solid polycarbosilane (PCS, Xiamen University, Xiamen, China) as SiC precursor were used. The ceramic yield of liquid HfC precursor was about 20 wt%. More narrowly, the mass residue was 66.7% from liquid to solid at room temperature and 30% from solid to HfC after treatment at 1700 ℃ for 2 h. The ceramic yield of solid PCS was about 50 wt% after treatment at 1700 ℃ for 2 h. The xylene solution of PCS (50% mass concentration) was first mixed with HfC precursor at a mass ratio of 33:100 to obtain the final molar ratio of SiC:HfC as 2:1, which is the same as for C/SiC-HfC composite prepared by RMI process [20], and thus it was convenient to compare the ablation behaviors of the two composites. The mixed precursors were introduced into the as-prepared C/SiC samples by vacuum pressure impregnation (VPI). Subsequently, the impregnated samples were dried at room temperature for 12 h and then pyrolyzed at 1500 ℃ for 2 h. After eight circles of VPI and pyrolysis, the resultant samples were heat treated at 1700 ℃ for 2 h in vacuum, ensuring complete reaction between HfO 2 and carbon to form HfC. The open porosities of samples increased after heat treatment. Thus, the samples were further densified by VPI and pyrolysis. The final C/SiC-HfC composite samples were obtained after four heat treatment cycles, as shown in Fig. 1(b).
The ablation characteristics of C/SiC-HfC composite were tested in a high-frequency plasma wind tunnel (1 MW, Ultrahigh Speed Aerodynamics Research Institute, China Aerodynamics Research and Development Center, China), whose schematic illustration is presented in Fig. 1(c). The velocity of plasma gas was set to be 0.7 Mach. The volumetric proportions of oxygen and nitrogen in plasma gas were 21% and 79%, respectively, in accordance with the air environment. The distance between samples and torch exit was set as 600 mm. The temperature changes during the ablation process were measured using an infrared pyrometer with ±20 ℃ accuracy and 500-2500 ℃ range. More detailed descriptions of plasma wind tunnel can be found in literature report [20].
The density and porosity of samples were measured by Archimedes drainage method. The crystal phases of samples were characterized by X-ray diffraction (XRD, Cu Kα, RigakuD/Max-2400, Tokyo, Japan). The scanning rate and scanning range were 0.12 (°)/s and 10°-80° respectively. The microstructures of samples were analyzed by scanning electron microscopy (SEM, JEOL 6700F, Tokyo, Japan) system equipped with energy dispersive spectrometer (EDS), which was used to investigate the sample composition.

Results and discussion
The densities and open porosities of final C/SiC-HfC samples obtained by PIP process were 3.18±0.04 g/cm 3 and 13.2%±0.4%, respectively. Figure 2 exhibits XRD patterns of sample, showing the existence of SiC and HfC phases and indicating successfully introduction of HfC into the porous C/SiC composite and no impurities were introduced after the PIP process. The volume fraction of HfC phase can be calculated as follows: where V H represents the volume fraction of HfC phase in C/SiC-HfC composite; V mf and V ms represent the molar volumes of HfC and SiC, which are 15.3 and 12.5 cm 3 /mol respectively; P r and P f represent open porosities of initial C/SiC preforms and final C/SiC-HfC composite, which are 38% and 13.2% respectively. Thus, the volume fraction of HfC in C/SiC-HfC composite is about 9.4%. The backscatter images of cross-sectional microstructures for the C/SiC-HfC composite are shown in Fig. 3. Clearly, the carbon fibers were protected by the surrounding SiC layer deposited by CVI process, and no fiber corrosion was observed. The SiC and HfC phases introduced by PIP were filled between the fiber bundles as the matrix of C/SiC-HfC  composite. The HfC phase was uniformly distributed in SiC phase formed from PCS, because the liquid HfC precursor and xylene solution of PCS were mixed at the molecular scale. Figure 3(b) also shows the presence of some micropores, which could be due to the processes of drying and pyrolysis of precursors involving volume contraction.
Four C/SiC-HfC composite samples were tested under four different conditions in the plasma wind tunnel, which was the same as our previous study [20], in order to compare the results. The main parameters included cold-wall heat flux q cw and stagnation point pressure P s , which ranged in 3.5-5.1 MW/m 2 and 4.5-6.2 kPa, respectively. The sample temperatures varied with the ablation conditions. The steady wall temperature and the maximum temperature were denoted as T w and T m , respectively. Detailed parameters of the four ablation conditions labeled as HF1 to HF4, and the ablation rates including mass ablation rate and linear ablation rate are listed in Table 1. The ablation rate increased with the increase in heat flux and stagnation point pressure. The sample expanded slightly at low heat flux due to oxidation of SiC and HfC. In contrast, ablation resulted in the shortening of the sample at high heat flux. Figure 4 demonstrates the variations of temperature with ablation time, exhibiting the rapid increase in the surface temperature after samples were exposed to heat flow. Then, the surface temperature under HF1 and HF2 conditions reached the maximum and remained stable during the ablation process. Higher value of P s resulted in higher surface temperature at the same heat flux. This is attributed to the fact that a higher pressure would cause more energy to reach the surface of the sample. Different from HF1 and HF2 samples, the surface temperature of HF3 sample first reached a steady-state, then gradually increased, and finally reached another steady-state. The final temperature was about 2454 ℃. When the heat flux was further increased to 5.1 MW/m 2 , namely the sample HF4, a sudden temperature jump occurred at about 40 s during the ablation process. The temperature after jump was above 2500 ℃, which could not be accurately measured in this experiment due to the limitation of infrared temperature measurement. Besides, the surface temperature of HF4 dropped slightly before the jump. In order to protect the surface microstructures of sample, the ablation  process was stopped immediately after the temperature jump.
The macroscopic images of C/SiC-HfC composite after ablation are shown in Fig. 5. When the heat fluxes were 3.5 and 4.5 MW/m 2 , a white oxide covered the sample surface and no fiber was exposed, although the surface temperature of the sample reached 2454 ℃ at heat flux of 4.5 MW/m 2 . This indicated that C/SiC-HfC composite prepared by PIP exhibited good ablation resistance under these ablation conditions. When the heat flux was further increased to 5.1 MW/m 2 , the fibers at the stagnation point were obviously ablated, which showed that the sample failed during the ablation process. Figure 6 shows the surface microstructures of the stagnation point region after ablation, exhibiting the  coating of the oxide layers on the surface of samples HF1-HF3 and no fibers are exposed after ablation. A few holes were observed on the surface of HF1. Considering that the surface after ablation was not machined and the stagnation point pressure in this state was low, the holes resulted from oxidation of fibers in short-cut-fiber webs. The oxide layers of HF1 and HF2 were clearly composed of SiO 2 and HfO 2 formed due to oxidation of SiC and HfC, respectively. Further, the content of SiO 2 on the surface decreased with the increase in ablation temperature. No SiO 2 was found on the surface of sample HF3, due to high temperature during the ablation process. Different from HF1-HF3, fibers of sample HF4 were uncovered and exhibited severe ablation. HfO 2 was detected among the fiber bundles and no SiO 2 was found in the stagnation point region after ablation.
The cross-section microstructures of HF1-HF3 in the stagnation point region are shown in Fig. 7. A thick SiO 2 -HfO 2 layer was observed in all three samples. The oxide layers of HF1 and HF2 were consistent with the above mentioned results of above surface analysis. In contrast, large amounts of SiO 2 were observed in the cross-section microstructure of HF3, while SiO 2 was not found on the surface of the same sample. Moreover, there were some voids in the oxide layer, which got enlarged in sample HF3 and formed larger holes. Based on the above mentioned observations and results of our previous study [20], the ablation behaviors of C/SiC-HfC prepared by PIP process can be explained as follows (Fig. 8). At heat flux of 3.5 MW/m 2 , the ablation behaviors of C/SiC-HfC composites were independent of the preparation methods and similar to the oxidation of UHTCs [34,35]. HfC and SiC on the surface of the sample were first oxidized during the ablation process to form HfO 2 and SiO 2 , respectively.  The formed HfO 2 has a high melting point (about 2810 ℃). Thus, it acted as a porous solid skeleton during ablation. SiO 2 with a low melting point was filled in the HfO 2 skeleton in liquid state. The liquid SiO 2 and HfO 2 skeleton formed the oxide layer, protecting the substrate from further ablation. Then, oxygen diffused from the surface of sample to the substrate through the oxide layer, while silicon oxide diffused from the oxide layer to the surface [20,36]. SiO 2 on the surface would be lost under the combined effects of atomic oxygen, high-temperature, and highspeed flow [37]. Besides, the active oxidation of SiC likely occurred under the HfO 2 -SiO 2 layer due to lower oxygen partial pressure here, which may be one of the reasons for the formation of voids under the oxide layer. The loss of SiO 2 and oxidation of SiC led to consumption of SiC, and finally caused mass loss after ablation. However, owing to the low heat flux, the consumption rate of SiC was low and the HfO 2 -SiO 2 layer could protect the substrate composite for a long time.
When the heat flux of ablation increased to 4.5 MW/m 2 , a unique phenomenon was observed. The surface temperature of C/SiC-HfC composites reached a stable stage, then gradually increased, and finally reached another stable stage. This phenomenon was not found during ablation under the same conditions for C/SiC composite prepared by CVI process and C/SiC-HfC composite prepared by RMI process [20,31,37]. Considering that there was no significant change in energy output (mainly including radiation and conduction) from the sample after the temperature reached the first stable stage, the energy arriving at the sample likely increased, and eventually caused the surface temperature of the sample to rise. It is known that the flow in plasma wind tunnel contains molecular oxygen and atomic oxygen (dissociated oxygen) [38]. When the atomic oxygen reached the surface of material, it was partially converted to molecular oxygen. The ratio of converted atomic oxygen to total atomic oxygen is defined as catalytic coefficient γ [39]. The conversion is exothermic, and thus higher catalytic coefficient of the material causes higher amount of energy to reach the surface of the sample. Our previous study [20] found that when the sample surface composition changed from SiO 2 with low catalytic coefficient to HfO 2 with high catalytic coefficient, it caused the sample to absorb more heat from plasma gas, causing a jump of temperature on the sample surface. However, the temperature jump was not observed for the C/SiC-HfC composite prepared by PIP and a gradual temperature change was found instead. Consequently, it can be inferred that a gradual change of the sample surface SiO 2 to HfO 2 led to increase in the surface catalytic coefficient, and eventually caused gradual increase in surface temperature of samples.
The microstructure of C/SiC-HfC by PIP (Fig. 3(b)) indicated that HfC had a small size (< 2 μm) and was uniformly dispersed in HfC-SiC matrix. During ablation, HfC was oxidized to form HfO 2 skeleton, through which SiC was oxidized and consumed. Owing to the small size and uniform distribution of SiC, the formed passages for the diffusion of oxidative species and silicon oxide in HfO 2 skeleton were less than 2 μm in size. The results of Holcomb's study [40] showed that when the pore size of diffusion medium was less than 10 μm, the effect of Knudsen diffusion could not be ignored, and increased with the decrease of pore size. Yang et al. [41] reported that Knudsen diffusion tended to contribute dominantly when the average pore size of diffusion medium was 2 μm. Thus, the similar pore size in HfO 2 skeleton was small enough to enable Knudsen diffusion for the outward species of silicon oxide. Knudsen diffusion with much lower diffusion rate than molecular diffusion retarded the erosion rate of SiC at the heat flux of 4.5 MW/m 2 . Moreover, the silica in the top oxide layer during ablation was gradually consumed and finally transferred into pure HfO 2 as shown in Fig. 6(c). In this case, the surface catalytic coefficient should slowly increase to the value of pure HfO 2 , which should respond to the high temperature at final stage. This was confirmed by the fact that the final stable surface temperature of HF3 sample was 2454 ℃, similar to the reported temperature of ablation for HfC-based ceramics in arc-jet or plasma wind tunnel (2400 ℃) [42]. At such a high temperature, HfO 2 on the surface was sintered, and some holes with size less than 2 μm were left, as shown in Fig. 6(c). This was because that the Tammann temperature of HfO 2 was about 2000 ℃ (about 0.75 of its melting point [43]), lower than the final stable surface temperature. However, the whole HfO 2 porous network could still be stable for some time due to the temperature gradient in the oxide layer. Furthermore, SiO 2 was filled in the porous HfO 2 structure, as shown in Fig. 7(c). The stable HfO 2 porous network provided channels for Knudsen diffusion, which retarded the loss of SiO 2 in the HfO 2 -SiO 2 layer.
When the heat flux was further increased to 5.1 MW/m 2 , a jump in surface temperature occurred during the ablation process. The temperature jump was also found during the ablation of C/SiC-HfC prepared by RMI and C/SiC under high heat flux condition [20,31,37]. When the sample HF4 was moved into the plasma flow, the surface temperature rapidly climbed to 1735 ℃. The atomic oxidation of SiC and HfC occurred under this condition, similar to that of HF1 and HF2 at low heat flux. The Knudsen diffusion of outward silicon oxide was still believed to be the controlling process. However, higher heat flux for HF4 samples resulted in a faster loss of SiO 2 than that for HF3. The outward diffusion of silicon oxide was not fast enough to catch up its consumption in the top layer, leaving the pure HfO 2 to face the plasma flow. High catalytic surface of HfO 2 could lead to the rapid increase in temperature, in turn accelerating the consumption of SiO 2 in the top layer to form loose HfO 2 , which was likely scoured away by the high-speed flow. After the coating on C/SiC-HfC was ablated away, the carbon fiber directly faced the plasma flow and began to burn. The burning of fibers included not only the reaction between C and O 2 , but also reactions of C with atomic O and N, which exhibited faster reaction rates [44,45]. A large amount of heat was released by the burning of fibers, which finally caused a jump in the surface temperature. After the temperature jumped, SiC was decomposed rapidly at temperature above 2500 ℃ and the remaining HfC that was not scoured was oxidized to HfO 2 , as shown in Fig. 6(d). However, the final temperature (after jumping) during ablation of C/SiC prepared by CVI [37] and final temperature of HF3 were lower than 2500 ℃. Therefore, the temperature jump was likely due to the synergy between carbon fiber combustion and the catalysis of HfO 2 . The residual HfO 2 absorbed more energy during ablation due to its high catalytic coefficient and in turn accelerated the burning of carbon fibers.

Conclusions
C/SiC-HfC composite was successfully prepared by precursor infiltration and pyrolysis process. The ablation behavior of the C/SiC-HfC composite was studied in plasma wind tunnel. The results showed that the synergistic effect between the stable HfO 2 porous network and the sustainable diffusion of silicon oxide played a crucial role in the ablation process. The low diffusion rates of oxygen and silicon oxide through the HfO 2 -SiO 2 layer at heat flux of 3.5 MW/m 2 led to a good ablation performance. SiO 2 was gradually removed through Knudsen diffusion when the heat flux was increased to 4.5 MW/m 2 . The stable HfO 2 porous network stabilized by SiO 2 provided the channel for Knudsen diffusion. At heat flux of 5.1 MW/m 2 , the stable network was destroyed due to the rapid oxidation of SiC and the scouring of HfO 2 . The subsequent burning of carbon fibers and the catalysis of HfO 2 caused temperature jump conjointly. Therefore, this study provides a reference for the microstructure control of composites used in TPS.