Abstract
ZrB2 particles were preset to the C-AlSi interface to improve oxidation resistance of C/C preform and adjust the microstructure of the interpenetrated C/C-AlSi composite prepared through pressure infiltration of eutectic AlSi into a fiber fabric based porous C/C skeleton. Micro-morphology investigations suggested that the AlSi textures were changed from dendritic to petals-like state, and the nano to micro-scale ZrB2 particles were dispersed into AlSi and affected the distribution of Al and Si nearby carbon. Tests demonstrated that C/C-AlSi have slight lower density and thermal expansion coefficient, and higher original compressive strength, while C/C-ZrB2-AlSi composites presented an outstanding strength retention rate after thermal shock. Fracture and micro-morphology indicated that the influence of the preset ZrB2 to the interface of carbon and alloy greatly affected the generation and propagation of cracks, which determined the diverse compression behaviors of the composites before and after thermal shock.
Introduction
Carbon fiber reinforced Al matrix composites (C/Al) have been considered to be a promising smart lightweight material due to their high specific strength, low coefficient of thermal expansion (CTE), outstanding thermal conductivity and self-lubrication characteristics.1–3 A lot of works have been done to solve the problems of poor wettability, harmful chemical reaction and high residual thermal stress between carbon and Al. Up to now, some effective methods have been successfully developed, such as fiber surface modification,4–6 matrix alloying7,8 and process optimization.9,10 However, the precise control of distribution and volume content of carbon fiber in Al matrix is also difficult for their different densities and the clustering of fiber.11–13
Carbon fiber reinforced carbon matrix (C/C) and ceramic matrix composites are also good lightweight materials, and have been greatly studied in the last decades.14–16 Various fabrics could be filled by carbon or ceramic gradually without fiber damage through chemical vapor infiltration and process of precursor infiltration and pyrolysis (PIP).17,18 Before completely infiltrated, the composites are porous and can be used as reinforcement of metal matrix. If the fabric based porous skeleton was used to reinforce Al matrix, the interface of carbon fiber-Al would be changed to be carbon matrix-Al or ceramic-Al. And then all the disadvantages of wettability, reaction and thermal stress between C-Al would be relieved or eliminated, and the rigid skeleton would not deform during Al infiltration which was helpful to obtain an accurate control of fiber distribution in matrix.
However, to our best knowledge, few relevant papers can be found about the Al-carbon or Al-ceramic hybrid matrix composites reinforced by continuous carbon fiber, although the composites are potential candidates for thermal structural components as pantograph sliders of high-speed railway, light weight brake pad and disc, piston of high power density diesel engine, and so on. Liu et al. 19 and Yu et al. 20 prepared C/C-Al composites and studied the mechanical and thermophysical property. Liao et al. 21 researched the compressive behavior of C/SiC-Al. All the works indicated that the composites are promising materials for industrialization. In our recent study of C/C-AlSi, 22 the composites were prepared through pressure infiltration of eutectic AlSi into a fiber fabric based porous C/C skeleton in air, and the effect of grain refinement on the microstructure and property was discussed. To achieve a better and deeper infiltration, the C/C skeleton should be preheated to above 400°C which would result in oxidation of carbon and fast decreased strength of the C/C preform. Thus, it is necessary to find a method to protect the C/C skeleton from oxidation during preparation of the composite.
It has been found that boride ceramic could provide oxidation protection for carbon skeleton 23 and was reactionless with molten aluminum. 24 Moreover, the compositionally structures and multiscale hybrid composites have outstanding comprehensive properties and got increasing attention in the last decade.25,26 A lot of works27–31 have been performed on the simulation and optimization of various structures and compositions, which provide some good guides for the improvement of the hybrid composites.
To modify the preparation process and further investigate the matrix adjustment synchronously, nano to micro-scale ZrB2 particles were preset to the C-AlSi interface of C/C-AlSi composite in the present work. Influence of the interfacial ZrB2 particles on the microstructure and strengths evolution in thermal shock was mainly studied to support their application at high temperature.
Experimental procedure
Composites preparation
As shown in Figures 1 and 2, 5D needle punched carbon fiber felt disks (0.45 g/cm3, Φ80 × 10 mm3) were filled by pyrocarbon to 0.6 g/cm3 through thermal gradient chemical vapor infiltration (TCVI). And then the disks were divided into two groups. The first group was further densified to 0.8 g/cm3 by pyrocabron while ZrB2 particles were introduced into the second one to the same densities through PIP. Details of the fiber felts, TCVI and PIP have been described in our previous work.32,33 Afterwards, cubic skeletons with size of 10 × 10 × 10 mm3 were cut from the two disks. Then eutectic AlSi melt was filled into the C/C and C/C-ZrB2 skeletons through pressure infiltration (PI) in air which was economical and effective. 1 During PI, the cubic skeletons were preheated to 500–700°C in static air and then fixed into the AlSi melt at 750–850°C. Holding 50–80 MPa pressure for 2–3 min ended the infiltration, and the subsequent solidification was also finished in the period. Finally, the C/C-AlSi and C/C-ZrB2-AlSi composites were taken out from the AlSi cylinder by machining.

Schematic preparation processes of the two composites.

Backscattered electron morphologies and XRD patterns of the prepared composites: (a) C/C-AlSi; (b) C/C-ZrB2-AlSi; (c) XRD patterns.
Tests and characterization
Densities of the two composites were measured by drainage according to the Archimedean principle. Compression tests were performed on the electron universal testing machine (CMT 5304, Suns Co. China). The sample size was 10 × 10 × 10 mm3 and the loading speed was 0.5 mm/min. CTE between room temperature and 500°C was determined with sample size of Φ5 × 5 mm3 by a thermal mechanical analyzer (TMA.SDTA840) of METTLER Company. The directions of the above two tests were parallel to the plane of non-woven layer. To investigate the thermal stability of the prepared composites, thermal shock test was performed by alternant 5 min of room temperature and 500°C in air. Weights of the samples were measured by a precision balance with sensitivity of 0.1 mg. Cumulative weight changes (weight loss percentage, %) were calculated and reported as a function of thermal cycle times. Differential scanning calorimetry (DSC) analysis was conducted in simulated air using a METTLER TOLEDO TGA/DSC 1 Thermal Analysis System. The samples were heated from room temperature to 600°C at a heating rate of 5°C/min.
The micro-morphology, chemical composition and phase of the prepared composites were analyzed by a light optical microscope (OM), a scanning electron microscopy (SEM, JSM6460) with energy dispersive spectroscopy (EDS) and a X-ray diffraction (XRD, X’Pert Pro MPD).
Results and discussion
Microstructure of the C/C-ZrB2-AlSi composites
Cross section morphologies and XRD patterns of the prepared two composites are shown in Figure 2. Corresponding to the fiber fabric, lamellar structures with some penetrating textures are clear in both composites. Attested by EDS analysis, the black phase is carbon and the white phase is a mixture of Al and Si. Obviously, most of the infiltrated AlSi is located at web layer and passes through the non-woven layer with needle punched fibers. Moreover, both composites are well infiltrated and compacted at the magnification. XRD patterns indicate that the two composites are mainly composed of carbon, Al and Si while more carbon in C/C-AlSi and some unique ZrB2 in C/C-ZrB2-AlSi. This is in accordance with the fabrication process.
Figure 3 shows the OM and maginificated backscattered electron morphologies of the prepared composites and relative EDS analysis. In OM, black carbon fiber and circumambient pyrocarbon matrix, grey eutectic Si and white Al are all clear. There is no crack between carbon and AlSi alloy. The metal matrix of C/C-AlSi is composed of dendritic Al and co-exist Si, whereas the Al in C/C-ZrB2-AlSi is radial around carbon and like petals. Moreover, the Al texture in C/C-ZrB2-AlSi is bigger and more agminated, which indicates that the ZrB2 particles acted as the nucleating agents during solidification of AlSi melt. In the view of mechanical property, the coarse microstructure of C/C-ZrB2-AlSi is disadvantageous which would lead to crack generation for the reason of stress concentration. In the magnified backscattered electron morphologies, black carbon fiber and pyrocarbon matrix are still discriminable. However, both Al and Si are grey and hard to be identified for their similar atomic quality. Having a much higher atomic number of Zr, ZrB2 particles in C/C-ZrB2-AlSi are apparent. It can be seen that the particles are in nano to micro-scale and dispersed into AlSi nearby carbon. A good strength and ductility of AlSi around carbon can be inferred, since the alloy layer can be regarded as ceramic nano-particles reinforced metal matrix composite. 34 Through linear EDS analysis, the difference of element distribution at the interface of carbon and AlSi was ascertained. The width of transition zone was about 3–4 μm. With the reducing of carbon, the increase of Al was always prior to Si in the linear EDS analysis of C/C-AlSi. Nevertheless, the rise of Al and Si was almost simultaneously when ZrB2 particles adhere to carbon in C/C-ZrB2-AlSi. Besides, fewer Si was tested in the place without ZrB2 particles. Obviously, the ZrB2 particles not only changed the texture of AlSi in the composite but also determined the distribution of Al and Si nearby carbon. In another word, the ZrB2 particles mainly acted as the nucleating agents of Si. Combined Figure 3(b) and (d), it can be concluded that the chemical reaction between carbon and Al should also be impeded by the ZrB2 particles as they could be a barrier between them, which would result in a weakened interface bonding.

OM (a, b) and backscattered electron morphologies of the prepared composites at high magnification and relative EDS analysis: (a, c) C/C-AlSi; (b, d) C/C-ZrB2-AlSi; (e–h) relative EDS analysis corresponding to lines in (c, d).
Properties of the C/C-ZrB2-AlSi composites
The density, CTE and compression strength before (BTS) and after (ATS) 50 times of thermal shock of the prepared composites are shown in Figure 4. Firstly, make a comparison with relative researches, it can be found that the two prepared composites have lower density and CTE than graphite reinforced Al-12Si composites, 35 and possess lower density and higher compressive strength than Al-sintered-carbon composites. 36 Obviously, the two prepared C/C-AlSi composites are attractive in application of light weight thermal structural components. Secondly, compared the C/C-AlSi with the C/C-ZrB2-AlSi, it can be concluded that the C/C-AlSi has a higher stress BTS, lower density and CTE, which seems to be a better material for application. However, the high temperature could not be avoided for many components as narrated in the introduction, and the material must endure cyclic high temperature. Thus the stability of the composite after thermal shock is particularly important for these applications. Doubtlessly, the C/C-ZrB2-AlSi is more suitable for relevant structural parts since the residual stress is above 95% and both the density and CTE are slightly higher. Moreover, the higher CTE may indicate a weakened interface bonding of C/C-ZrB2-AlSi in comparison with C/C-AlSi.37,38

Properties of the prepared two composites.
Failure analysis before and after thermal shock
Figure 5 shows the representative stress–strain curves of the two composites before and after thermal shock. It can be concluded that ductility of AlSi matrix was greatly restrained by C/C or C/C-ZrB2 skeletons for the low strain at fracture. Before breakage, all stress–strain curves rise up continuously to a maximum, and have durations of increasing linearly which were marked as I, II, IV and VII. The straight lines in stress–strain curve mean elastic deformation of the composites depending on synergy of different components. Once interface debonding, fiber and matrix smashed, or sliding along cracks occurred, the linear curve will be changed. To original C/C-AlSi, abruptly falling followed linear rising suggest a strong interfacial bonding between carbon and AlSi. This is similar to the compression of C/SiC-Al. 21 The brittle fracture behavior should be avoided for engineering components since it always cause sudden failure. Differently, after the first decline, the curve of C/C-ZrB2-AlSi goes up again (Marked as III) and then decreases slowly. The serrated curve and lower slop suggest a pseudoplastic fracture and medium interface bonding of C/C-ZrB2-AlSi. Undoubtedly, the pseudoplasitc fracture behavior is more acceptable as it would not result in unpredictable failure.

Stress–strain curves of the two composites before and after thermal shock.
After thermal shock, strength of both composites decreased and residual strength of C/C-ZrB2-AlSi is inverted higher than that of C/C-AlSi. Moreover, unique nonlinear increasing of stress–strain curve appeared for both composites, labeled as V, VI and VIII. In combination with the smooth decrease over inflection point and referencing the compressive behavior of porous C/SiC, 21 it can be inferred that interface of carbon and AlSi was weakened to both composites during thermal shock.
Figure 6 shows the fracture of composites before and after thermal shock. Before thermal shock, the fracture surfaces of both composites are relative flat and the broken of C/C in C/C-AlSi was in bundle mode while that in C/C-ZrB2-AlSi was random and isolated. After thermal shock, the fracture became uneven. Some pull-out C/C were smashed and distributed randomly on the surface. Obviously, the interface of carbon and AlSi was weakened. Moreover, the fracture surface of C/C-ZrB2-AlSi was coarser than that of C/C-AlSi.

Fracture of the two composites before and after thermal shock: (a) C/C-AlSi before thermal shock; (b) C/C-ZrB2-AlSi before thermal shock; (c) C/C-AlSi after thermal shock; (d) C/C-ZrB2-AlSi after thermal shock.
The magnified fracture in backscattered electron mode is shown in Figure 7. They are evidently different. The surface also became coarser after thermal shock at this magnification. Besides, more AlSi (white phase) is located at the surface of C/C-ZrB2-AlSi, which is in accordance with the pyrocarbon content and microstructure in Figure 2. To C/C-AlSi, although plastic deformed with broken carbon and cracked (indicated by arrow), the residual AlSi still kept good adhesion to carbon and smooth. However, after thermal shock, fewer AlSi was left and became folded (indicated by ellipse) which should seriously deform during breakage of carbon. To C/C-ZrB2-AlSi, the AlSi also deformed with broken carbon (indicated by dashed line). Differently, some matrix particles containing grooves of C/C are left on the surface (indicated by arrow), which indicate that the interface bonding strength between carbon and AlSi in C/C-ZrB2-AlSi is inhomogeneous. After thermal shock, many fibers were pulled out and the surrounding pyrocarbon smashed (indicated by ellipse). The pyrocarbon should be crushed by ZrB2-AlSi nearby during compression. Thus it can be deduced that the pyrocarbon in C/C-ZrB2-AlSi might be damaged in the thermal shock.

Fracture of the two composites before and after thermal shock at high magnification: (a) C/C-AlSi before thermal shock; (b) C/C-ZrB2-AlSi before thermal shock; (c) C/C-AlSi after thermal shock; (d) C/C-ZrB2-AlSi after thermal shock; inserted figure is the secondary electron image of relative position.
Figure 8 shows the weight change curves during thermal shock, DSC analysis and microstructures after thermal shock of the two composites. The weight change curves can be regarded as straight lines, whose different slopes indicate different thermal shock resistance of C/C-AlSi and C/C-ZrB2-AlSi. Obviously, C/C-ZrB2 skeleton reinforced AlSi composites possess better resistance to thermal shock than the C/C with same fabric and density reinforced ones. The DSC analysis curves suggest that the phase transitions and chemical reactions of the two composites were same when temperature rose from room temperature to 600°C. In another word, the ZrB2 particles in the composites do not chemically react with other element below 600°C, and they play stable reinforcement at the tested temperature in this work. It is clear that the microstructures after thermal shock are distinct for the two composites, although cracks generated as that in C/Al composites. 39 Crack in C/C-AlSi is much wider than that in C/C-ZrB2-AlSi. Besides, the crack in C/C-AlSi propagated through pyrocarbon around carbon fiber rather than along the interface between carbon and AlSi, which implies that the interface bonding is stronger than pyrocarbon and AlSi selves. In C/C-ZrB2-AlSi, part of cracks propagated through pyrocarbon while others along the interface of carbon and ZrB2-AlSi, which indicates that the interfacial bonding strength is not uniform.

Weight change during thermal shock (a), DSC analysis (b) and microstructures after thermal shock of C/C-AlSi (c) and C/C-ZrB2-AlSi (d).
To sum up, rigid skeletons in this study greatly restricted the ductility of penetrated AlSi alloy. Thus the fractures of the prepared composites were mainly breakage of the skeletons, and the interface state of carbon and AlSi played a key role in crack generation and propagation. Considering the great CTE difference between carbon and AlSi alloy, it can be inferred that the residual stress at interface of carbon and AlSi was higher than other parts of the composites. If the matrix was stronger than the interface bonding, micro-cracks would apt to generate at the interface, or else they would appear at matrix where stress concentrated.
The interface bonding between carbon and AlSi of C/C-AlSi was strong in the present work. When the stress–strain curve went up to maximum, crack initiated and propagated through the whole composite with few deflections, which caused the sudden fracture and high strength. To C/C-ZrB2-AlSi, nano to micro-scale ZrB2 particles located at AlSi around carbon formed a composite layer of ZrB2 reinforced AlSi between carbon and AlSi. Besides, the particles changed the element distribution nearby carbon and texture of AlSi alloy. The interface bonding was weakened by the particles through hindering the contact of carbon and AlSi. In another word, the weakened interface bonding was only present to location of ZrB2 particles. Thus when stress rose to a certain value, micro-crack would initiate from the weakened interface and then spread through pyrocarbon or AlSi to bypass the strong bonding part. Meanwhile, some broken parts slid and the up stress–strain curve started to decrease step by step. Once the crack tip was stopped by fiber, ductile matrix or other interfaces, the whole stress would be carried by residual parts of the composites and the curve would increase again (as part III in Figure 5). With stress rising, the micro-crack grew up continually and encountered with others to be bigger crack. The bigger crack further extended and induced final failure of the composite. Obviously, ZrB2 particles prolonged the route of crack and changed the fracture process.
During thermal shock, no chemical reaction occurred for ZrB2. Thus, the interfacial bonding and element distribution, and structure of ZrB2-AlSi layer were important factors to the microstructure evolution of the composites under thermal stress. There was no doubt that the partial weakened interface changed the propagation of crack and more deflections came into being, which was similar to that in compression. Besides, the high thermal stress between carbon and AlSi was reconciled by ZrB2-AlSi layer effectively, which greatly suppressed the crack widening and reduced the weight loss of carbon oxidation. After thermal shock, the wide crack of C/C-AlSi was first closed by compression stress, which was corresponding to the part VI in Figure 5. And then all components of the composites carried stress together. When sliding of different components happened and then paused, the stress–strain curve started part VIII. Finally, the curve decreased as the growth and intersection of cracks resulted in fracture. The folded AlSi at fracture of C/C-AlSi after thermal shock implied a large slippage in the breakage. Differently, the frequent deflected crack resulted in a crack-rich pyrocarbon in C/C-ZrB2-AlSi. When stress exceeded the elastic limit, the fibers were pulled out and defective pyrocarbon was crushed by nearby ZrB2-AlSi, which formed the part V in Figure 5. With growing and encountering of microcracks, the stress–strain curve increased to maximum and decreased gradually.
Conclusion
Effect of preset interfacial ZrB2 particles on the microstructure and properties of fiber fabric based C/C-AlSi composites prepared through pressure infiltration was studied. The nano to micro-scale ZrB2 particles were dispersed into AlSi nearby carbon, affected the distribution of Al and Si, and changed the textures of AlSi matrix from dendritic to petals-like state. The C/C-AlSi composites have slight lower density and thermal expansion coefficient, and higher original compressive strength, while C/C-ZrB2-AlSi possess better strength retention rate after thermal shock. Fracture and micro-morphology indicated that the influence of ZrB2 to the interface of carbon and alloy greatly affected the generation and propagation of cracks, which determined the diverse compression behaviors of the composites before and after thermal shock.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Science Foundation of China (51902239), the Science and Technology Fund of Shaanxi Province (Grant No. 2020JQ-808), the Education fund of Shaanxi Province (Grant No. 19JK0400 and 19JK0402), the innovation and entrepreneurship training program for college students (Grant No. S202010702020).
