The comparison of microstructure and oxidation behaviors of (SiC-C)/PyC/SiC and C/PyC^HT/SiC composites in air

2.1 Preparation of samples

The properties of the raw materials are summarized in Table 1. Two kinds of fibers, Hi-Nicalon SiC fiber from Nippon Carbon (Takanchi, Japan) and T300 carbon fiber from Toray (Tokyo, Japan) consisting of bundles of 500 and 1000 filaments with the diameter of 14 μm and 7 μm per-filament, respectively, were used to fabricate the hybrid composite. The volume fraction ratio of SiC fiber to carbon fiber was 2:1. Two kinds of fiber preforms were weaved by four-step three-dimensional techniques at Nanjing Fiberglass Research and Design Institute, China. One contained both SiC and carbon fibers, in which SiC and carbon fibers were back-to-back as one bundle for braiding the hybrid preform. The other one only contained carbon fibers. The volume fraction of the fibers was approximately 40% in both preforms. PyC interphase and SiC matrix were deposited by CVI process. PyC interphase was deposited on the surfaces of the fibers from C₃H₆ at 900°C for 144 h at reduced pressure of 5 kPa in a CVI reactor. The preform only contained carbon fiber with the PyC interphase was heat-treated at 1800°C in argon for 1 h to improve the crystalline degree of interphase and lower the fracture energy of the interphase [17]. The SiC matrix was prepared at 1000°C with a reduced pressure of 5 kPa using methyltrichlorosilane (MTS, CH₃SiCl₃) with a molar ratio of 10 between H₂ and MTS, which was carried by bubbling hydrogen gas phase and argon as the dilute gas to slow down the chemical reaction rate during deposition. Density and porosity were obtained by the Archimedes method using distilled water according to ASTM C-20 [18].

The SiC matrix composite reinforced by co-mingling SiC and carbon fibers was named as (SiC-C)/PyC/SiC and designated as sample A with a density of 2.56 g/cm³. The other one was named as C/PyC^HT/SiC composite and designated as sample B with a density of 2.25 g/cm³, as shown in Table 2. The superscript HT means the PyC interphase was heat treated at 1800°C after the deposition of PyC interphase. It should be pointed out that PyC interphase in co-mingling preform could not be heat-treated because of the strength of Hi-Nicalon SiC fiber deteriorates at high temperatures [19, 20].

2.2 Experimental procedure

The microstructure of samples was observed by Scanning Electron Microscopy (SEM, S-2700, Hitachi, Tokyo, Japan). The CTEs of the uncoated composites and CVD SiC material were tested in order to estimate the mismatch between the SiC coating and the uncoated composites by using a Thermo dilatometer (DIL 402C, Netzsch Company, Selb, Germany). The composites were oxidized under air condition in a tube furnace at 700, 1000, 1200, and 1300°C for 10 h. The coating’s components of the oxidized samples were analyzed by energy dispersive X-ray spectroscopy (EDS, Genesis XM2, NJ, USA). A cumulative weight change was obtained according to the weight before and after oxidation tests by analytical balance (e=0.1 mg, AG 204, Mettler, Schwerzenbach, Switzerland) and recorded as a function of oxidation time. The flexural strength of the samples before and after oxidation was measured using 3 mm×4 mm×40 mm rectangular coupons by a Universal Flexural Machine (SANS CMT 4304, Sans Materials Testing Co., Shenzhen, China), the cross-head speed was 0.5 mm/min. Three to six samples were measured in each case for statistical significance.

3.1 Microstructure

3.1.1 The matrix morphologies

Some residual pores unavoidably existed in the composites owing to the “bottleneck effect” during CVI process [21]. The porosity of samples A and B was 10% and 11.6% (Table 2), respectively. No matrix microcracks were found in sample A as shown in Figure 1A. However, some matrix microcracks were observed in sample B (Figure 1B). The SiC matrix microcracks resulted from TRS, which had been elaborated by the following equations [22, 23]:

Figure 1

SEM images of polished morphologies of the SiC matrix in both composites: (A) SiC matrix in sample A; (B) SiC matrix in sample B.

(1)σm=Emλ2λ1[EfE][Vf1-νm]Ω (1)

where

(a)λ1=1-(1-E/Ef)(1-νf)/2+Vm(νm-νf)/2-(E/Ef)[νf+(νm-νf)VfEf/E]2(1-νm)Ψ (a)

(b)λ2=[1-(1-E/Ef)/2](1+νf)+(1+Vf)(νm-νf)/2Ψ (b)

(c)Ψ=1+νf+(νm-νf)VfEf/Em (c)

(d)Ω=(αf-αm)ΔT (d)

(e)E=VfEf+VmEm. (e)

σ, E, ν, α and V represent the TRS, Young’s modulus, Poisson ratio, the CTE and volume fraction, with the subscript m and f for SiC matrix and fiber, respectively.

It can be supposed that there was no TRS in the SiC_f/SiC composite because the CTE of SiC fiber is nearly the same as that of the SiC matrix [15]. Hence, only the carbon fibers in the two composites resulted in TRS. The volume fractions of carbon fiber are 13.3% and 40%, respectively, in samples A and B as mentioned in Section 2.1. The parameters concerning the calculation of TRS for the composites are listed in Table 1. The TRS was calculated by using equation (1) for samples A and B. The results showed that the TRS in sample A was decreased by 69.5% compared with those in sample B.

3.1.2 The morphology of SiC coating

Only few fine microcracks were found on the surface region of SiC coating for sample A (Figure 2A). However, many microcracks were found on the surface of SiC coating for sample B (Figure 2B). Determined through SEM observation, the SiC coating microcracks density (∼50 cracks/m) for sample A is much lower than that (∼1090 cracks/m) for sample B, as shown in Table 2. Furthermore, the coating microcracks width (∼1 μm) for sample B is five times larger than that (∼0.2 μm) for sample A determined from Figure 2A and B.

Figure 2

The morphologies of the SiC coating of the two composites: (A) the coating microcrack of sample A, (B) the coating microcrack of sample B.

The SiC coating microcracks are related to the mismatch of CTEs between SiC coating and uncoated composite. The CTEs of the uncoated composites for samples A and B and bulk of SiC were 4.05, 3.31 and 4.6×10^-6/°C (Tables 1 and 2). The TRS in the coating can be simplistically derived if the coating thickness (about 60 μm in this study) is much smaller than the substrate thickness (3 mm in this study) by the following equation [24]:

(2)σc=Ec(αs-αc)ΔT1-νc. (2)

Here, the subscripts c and s mean the coating (the values were substituted by the bulk of the SiC material) and the substrate composites (uncoated samples of A and B in this study). ΔT is the temperature difference between the calculated temperature and preparation temperature. The results according to equation (2) showed that the TRS in SiC coating for sample A was decreased by 62.2% compared with those for sample B.

3.2 Mass loss and the oxidation kinetics

The weight change of samples A and B after 10 h oxidation at different temperatures is shown in Figure 3. Weight loss of sample A is much smaller than that of sample B at each tested temperature. The weight loss is largest at 1000°C for sample A. However, the weight loss of sample B at 700°C is the largest one compared with that at the other three temperatures.

Figure 3

The weight change of the two composites after 10 h oxidation at different temperatures.

The weight change during oxidation should consider the oxidation of constituents. First, oxidation of carbon fiber and PyC interphase led to the weight loss. According to the previous studies [25] and considering the temperature range (700–1300°C) in the present study, the reactions between oxygen and carbon is described in the following equation:

(3)2C+O2→2CO. (3)

Second, the oxidation of SiC matrix, SiC fiber and SiC coating resulted in weight gain and the oxidation reaction according to the following equations [26]:

(4)SiC+2O2→SiO2+CO2, (4)

(5)SiC+32O2→SiO2+CO. (5)

For the oxidation of SiC, reaction (5) is more likely to occur than reaction (4) at high temperatures (above 1200°C), whereas reaction (4) may occur at intermediate temperatures (800–1200°C).

Besides, weight change of both composites during oxidation was also related to the defects existing in the composites. Pores and microcracks in SiC matrix and coating of composites seemed to be a connected network of passageways for the inward diffusion of oxygen and the outward diffusion of the gaseous oxides at the tested temperatures.

The evolution of the crack opening as a function of temperature may be represented by a linear relation of the form [26, 27]:

(6)e(T)=e0(1-TT0), (6)

where e₀ is the crack width at room temperature, T is the test temperature, and T₀ is the temperature where the crack opening is zero (1000°C in this study).

Few microcracks were found on the SiC coating of sample A. Below the deposition temperature (1000°C) of the SiC matrix, the defects in the coating and the open pores in the matrix may be the diffusion channels for the inward diffusion of oxygen. However, the width of the coating microcracks was 0.06 μm and 0 μm at 700°C and 1000°C according to equation (6), respectively. These defects were too narrow and less for oxygen diffusion to attack the carbon phases for sample A. No SiO₂ film was detected on the surfaces of sample A by the EDS analysis at 700°C (Figure 4, black line). The oxidation kinetics was controlled by gaseous diffusion kinetics. There was no distinct weight loss at 700°C and 1000°C for sample A. The oxygen could not attack the carbon fibers and PyC interphase owing to there being no matrix cracks in sample A. The carbon fiber and PyC interphase were protected at 700°C (Figure 5A).

Figure 4

The EDS analysis of oxidized SiC coating at different temperatures.

Figure 5

The morphology of the carbon fiber and PyC interphase after 10 h oxidized of samples A (A) and B (B) at 700°C.

Above 1000°C, the width of microcracks in sample A became very narrow due to the thermal expansion of the SiC matrix, and the SiC matrix was oxidized to form SiO₂ film (Figure 4, green line), thus the recession kinetics of sample A was controlled by oxygen diffusion through the SiO₂ film, and the oxidation was then superficial. There was some weight increase after 10 h oxidation for sample A at 1200°C and 1300°C because of the oxidation of SiC to form the SiO₂ film. The width of coating microcracks in sample B was about 0.3 μm at 700°C according to the equation (6). The same as sample A, no SiO₂ film was detected on the surfaces of sample B by the EDS analysis at 700°C (Figure 4, black line). The reaction of carbon phase and oxygen controlled the oxidation of sample B. The fiber/matrix interphase debonding took place because of mismatch of the radial direction CTE between the fiber and matrix and the matrix cracks acted as the channel for oxygen diffusion. Consequently, the oxidation preferentially propagates on the PyC interphase, that is therefore rapidly consumed, when oxidation is further diffused, carbon fibers become partly oxidized in turn (Figure 5B). Meanwhile, because of the rapid diffusion of oxygen and reactivity of the carbon phases (carbon fiber and PyC interphase) and silicon carbide, the oxygen rapidly reaches inside samples B that is thus homogeneously oxidized.

The weight loss was smaller than those of sample B at 700°C when the oxidation temperature was as high as 1000°C. The recession kinetics controlled by oxygen diffusion implies a crack geometry effect. As the oxidizing temperature increased, the thermal expansion of silicon carbide became significant resulting in the decrease of the microcracks width [zero theoretically by equation (6) in this study]. The amount of oxygen diffusing through these cracks was decreased. However, the carbon fibers near to the SiC coating consumed more oxygen than at lower temperatures because of the oxidation rate because the carbon fibers increased noticeably. An oxygen gradient and consequently a carbon consumption gradient take place between inside the composites and surface of the materials. Oxidation kinetics of the composites was governed by diffusion regime. A small quantity of SiO₂ was detected by EDS analysis at 1000°C (Figure 4, red line).

Above 1000°C, the microcracks in SiC coating and matrix was appreciably narrowed owing to the thermal expansion of the SiC matrix. Besides, the oxidation rate of SiC started to increase significantly [28]. The silica layers grew from the crack walls. The diffusion of oxygen through the coating microcracks was thus modified by the oxidation of silicon carbide (Figure 4, green line), which induces a decrease of carbon oxidation with time and temperature. The amount of oxygen reaching carbon fibers was limited and immediately consumed. The oxidation took place equally on the fibers and the interphase because the rate of reaction of carbon with oxygen was high, and only the carbon phases in the vicinity of the microcracks tips were consumed, the oxidation was therefore superficial. The oxidation of composite was controlled by the oxygen diffusion through SiO₂ film. As a result, the weight loss of sample B at 1300°C was the least than those at other three temperatures.

3.3 The effect of oxidation recession on the mechanical property

The initial flexural strength and residual strength for samples A and B after oxidizing at different oxidation temperatures is shown in Table 2 and Figure 6. The change in trend of the residual strength began to be inversed for the two kinds of composites after 10 h oxidation in the air. The residual strength of sample B was increased with raising the temperatures because the weight loss of sample B decreased with the increasing oxidation temperature. However, the residual strength did not decrease remarkably below 1200°C for sample A, and the strength decreased notably at 1300°C, even lower than those of sample B at the same temperature. The reason was that the SiC microcrystal in the Hi-Nicalon SiC fiber increased at high temperatures, which made the strength of Hi-Nicalon fiber decrease [19, 20]. Researchers found that the flexural strength reduced by 15% because of the recession of Hi-Nicalon SiC fiber when the composite was heat treated at 1300°C for 1 h in N₂ [29]. Hence, the strength decrease of SiC fiber at 1300°C was the main reason for the lower residual strength of sample A than that of sample B. As a result, (SiC-C)/PyC/SiC composite could have been used below 1200°C for long time.

Figure 6

The residual flexural strength vs. the oxidation temperatures.