Reaction sintered SiC (RS-SiC) was used as the joining base material. Its thermal properties depend on the melting point of residual Si about 1410˚C and the coefficient of thermal expansion (CTE) of 4.4 × 10⁻⁶/K. Brazing material has some necessary conditions such as electrical insulation, phase compatibility with SiC, good wettability on SiC, melting temperature lower than the melting point of Si (1414˚C) and CTE value close to SiC. In this work, low thermal expansion borosilicate glass, Pyrex glass, was employed. The CTE of the heating material also should be similar to SiC, and must be a refractory metal with high oxidation resistance and electrical conductivity. Thus, Tungsten (W; melting point = 3387˚C, CTE = 4.3 × 10⁻⁶/K) was selected in this work.

The reaction sintering (RS) technique was used for fabricating bulk SiC ceramics, since it had several advantages such as low sintering temperature, near net-shape forming and nearly full density by the infiltration of the liquid Si, although the residual Si might affect the flexural strength of SiC [17]. Commercial SiC powder (GC100 grade, Maruto Ltd., Japan) was mixed with the resol-type phenolic resin (HP3000A, Asahi-Yukizai Kogyo Ltd., Japan) and curing agent (HA305k Asahi-Yukizai Kogyo Ltd., Japan), was cured at 55˚C for 12 hrs to give a resin composite with the SiC volume fraction about 45%, which was subsequently dewaxed in a flowing N₂ atmosphere to prepare a green body composed of SiC and carbon using a heating program shown in Figure 2. As shown in Figure 3, a dewaxed green body was put on to the bottom of a graphite susceptor, then surrounded by the Si powder (code No. 198-054555455, Wako Pure Chemical Crop., Japan) of which amount was equal to the 1.1 times its pore volume, and was heat treated in vacuum at 1500˚C for 1 hr for Si infiltration. The RS-SiC sintered bodies were cut into pieces (10 × 10 × 2 mm) using a high-speed diamond blade (CD-SD14OB, Maruto Ltd., Japan).

In this work, the brazing material, Pyrex glass, may react with SiC during bonding to generate the gaseous phases such as CO and/or SiO. Therefore, the formation of SiO₂ protective film on the surface of RS-SiC ceramics was attempted, since it could protect the SiC surface from the exposure of residual oxygen gas, and delay the oxygen diffusion from the Pyrex glass at high temperatures [18]. As-is SiC specimens were heat-treated at 1200˚C for 3 hrs in the air using a ceramic tube furnace (DSPSH-39, Siliconit Co. Ltd., Japan) to make surface-oxidized SiC specimens. Figure 4(a) and Figure 4(b) show SEM micrographs of as-is and surface oxidized SiC surfaces, respectively, and the latter clearly shows the existence of SiO₂ protective film on the SiC surface.

The composition of Pyrex glass was slightly modified to have a softening point around 900˚C and CTE value around 4.3 × 10⁻⁶/K; 64.0SiO₂-30.4B₂O₃-2.8Na₂O-1.8Al₂O₃-0.9K₂O (mol%) [19]. Reagent grade SiO₂, H₃BO₃, Na₂CO₃, Al₂O₃ and K₂CO₃ powders were weighed to make the total amount of 20 g mixture, which was ground using a mortar and pestle for 1 hr. The mixed powder was transferred to a platinum crucible, and was melted at 1500˚C for 1 hr using a ceramic tube furnace (DSPSH-39, Siliconit Ltd., Japan) in air.

A bulk of Pyrex glass was crushed into pieces with about a few mm size, and was placed on an as-is or surface-oxidized SiC plate. This assembly was placed in a ceramic tube furnace, subsequently was evacuated with an oil-sealed rotary vacuum pump, and was heat-treated to a target temperature and held for 30 min. The target temperatures are 900˚C to 1200˚C at 50˚C intervals, consequently, 12 samples were prepared. Each sample was cross-sectioned using a high-speed diamond blade (CD-SD14OB, Maruto Ltd.) to observe the specimen using an optical microscope (VH-5000; KEYENCE Inc., Japan).

Pyrex glass was pulverized with a WC vibrational mill (P-0, FRITSCH Ltd.,

Japan). The Pyrex glass powder was sieved using a 140 mesh’ (106 μm) sieve and was mixed with methanol to form paste. About 0.5 mm thickness of the paste was applied to each one side of two SiC plates, and W-mesh (30 mesh’, wire diameter = 0.1 mm; Nilaco Ltd., Japan) was sandwiched between them. This assembly was set in a vacuum deposition apparatus (HUS-5GB; Hitachi, Ltd., Japan) as shown in Figure 5. The vacuum chamber was evacuated using an oil-sealed rotary vacuum pump, and then, 40 A of electrical current was supplied for 90 sec from the two electrodes that are connected to the ends of a W-mesh. After cooling, the inside of the vacuum chamber was returned to the atmospheric pressure. A joined specimen was cross-sectioned using a high-speed diamond blade (Maruto Ltd., CD-SD14OB) to observe the SiC-Pyrex glass interface using an optical microscope (VH-5000; KEYENCE Inc., Japan). The SiC-Pyrex glass interface was polished using the cross-section polisher (IB-19500CP, JEOL Ltd., Japan) to observe the microstructure and chemical composition of the interface using the field-emission scanning electron microscope (FE-SEM) with an energy-dispersive X-ray (EDX) spectrometry (JSM-7401F and JED-2300, JEOL Ltd., Japan).

Figure 6(a) and Figure 6(b) show the cross-sections of specimens after wetting experiments at 900 (softening point of Pyrex glass), 950˚C, 1000˚C and 1050˚C on (a) as-is and (b) surface oxidized SiC, respectively. All wetting angles of Pyrex glass sessile drops on the as-is SiC plate were greater than 90 degrees (non-wetting) and showed little change with raising temperature, while the wetting angles on the surface-oxidized SiC plates gradually decreased with raising temperature. Figure 7(a) and Figure 7(b) show the cross-sections of specimens after wetting experiments at 1100˚C, 1150˚C and 1200˚C on (a) as-is and (b) surface oxidized SiC, respectively. The wetting angles of Pyrex glass sessile drops on the surface-oxidized SiC plates became less than 90 degrees (wetting) at 1100˚C, although a Pyrex glass sessile drop on the as-is SiC was observed to be deformed at the same temperature. At more than 1100˚C, all Pyrex glass sessile drops were

observed to be severely deformed, and further, large pores were observed only on the surface-oxidized SiC. The severe deformation and pore generation in the sessile drops would be due to the gas evolution that caused by the reaction between RS-SiC and Pyrex glass and/or residual oxygen gas, which will be discussed later in detail. The current results suggest that the desirable condition for joining would be at 1100˚C using the surface-oxidized SiC.

After cutting using a high-speed diamond blade, all joined specimens were survived without debonding, which clearly indicates that all the joined bodies made by the Flash-bonding technique possessed certain levels of strength high enough for machining. Figure 8(a) and Figure 8(b) show the optical micrographs of the cross sections of the joining bodies made of as-is and surface-oxidized SiC plates, respectively. It was observed that the SiC/joining layer made of the as-is SiC plates had almost no pores, and its color was blackened, and in contrast, the SiC/joining layer made of the surface-oxidized SiC had a large amount of pores, and its color remained colorless and transparent.

Figures 9(a)-(d) show the microstructures of SiC/joining layers made of the surface-oxidized SiC plate observed by using the FE-SEM. As observed in the optical microscope, many pores are visible in the whole interlayer (a). In the vicinity of the SiC interface shown in Figure 9(b), many petal-like precipitates are observed in the interlayer, and some bright spots exist at the SiC interface. As shown in Figure 9(c), an apparent layer structure is observed between the interlayer and SiC, which would be the SiO₂ protective film of the surface-oxidized SiC. At the W/interlayer interface shown in Figure 9(d), the interfacial debonding and some cracks are clearly observed. Figure 10(a) and Figure 10(b) demonstrate the EDX elemental analysis results at points 1 and 2, respectively, in Figure 14(b). C and O signals at point 1 are relatively stronger than those at point 2, and Si signals at point 2 are much stronger than at point 1. Figure 11 shows the EDX line analysis result along the line shown in Figure 9(c). It is observed that the Si Kα1 signal decreases in two steps indicating the existence of SiC, SiO₂ and Pyrex glass along this line. It is interesting that C is segregated between SiO₂ and Pyrex glass.

Figures 12(a)-(c) show the microstructures of SiC/joining layers made of the as-is SiC plate observed by using the FE-SEM. As observed in the optical microscope, large pores are not visible in the whole interlayer (a), however, some micron-size pores exist in the interlayer as shown in Figure 12(b), which also indicate that the interlayer is composed of black and white two phases. At the W/interlayer interface shown in Figure 12(c), the interfacial debonding is not observed as opposed to that in Figure 9(d). Figure 13(a) and Figure 13(b) demonstrate the EDX elemental analysis results at points 3 and 4, respectively, in Figure 12(b). The intensities of C, O and Si signals at point 3 (white phase) are almost the same as those at point 1, however, that of C signal at point 2 (black phase) is much stronger than those of O and Si. The elemental mapping shown in Figure 14 demonstrates that the white phase mainly composed of Si and O, and C are segregated both in the black phase and the SiC/interlayer interface.

Song and Smith [20] proposed a phase diagram for the interaction of oxygen with SiC. In this work, the joining experiments seem to be conducted in the temperature range around 1100˚C - 1200˚C by comparing with the results obtained by the wetting experiments. The vacuum level in the vacuum chamber would be around a few Pa, since only an oil-sealed rotary vacuum pump was used for evacuation. According to the phase diagram [20], there are regions 3 and 4 in which the equilibrium reactions represented by Equations (1) and (2), respectively, would proceed considering the current experimental conditions above mentioned.

SiC ( s ) + 3 / 2 O 2 ( g ) = SiO 2 ( s ) + CO ( g ) (1)

SiC ( s ) + O 2 ( g ) = SiO 2 ( s ) + C ( s ) (2)

When the joining experiment was performed using the surface-oxidized SiC plates, oxygen diffusion from the chamber atmosphere and/or interlayer to the bulk SiC would be inhibited by the SiO₂ film on the SiC surface, and oxygen partial pressure would be less than 1.3 Pa at around 1100˚C, which is namely the region 3. As a result, the reaction (1) proceed to generate CO gas that resulted in the formation of many large pores, and to precipitate SiO₂ and probably the reaction product of Si, C and O, namely Silicon Oxicarbide, as demonstrated in Figures 9-11. When the joining experiment was performed using the as-is SiC plates, the SiC/interlayer assembly would be in equilibrium with the chamber atmosphere in which the oxygen partial pressure would be more than 1.3 Pa at around 1100˚C, which is namely the region 4. As a result, the reaction (2) proceed to generate no gas phase, and to precipitate SiO₂ and carbon as demonstrated in Figures 12-14.

Cracks were observed between and the interlayer and W as shown in Figure 9(d), which might be caused due to the excessive volume expansion of WO₃ [21]. When the as-is SiC was used, the carbon precipitate in the interlayer would lower the oxygen activity to prevent the oxidation of W resulting in no crack formation. In this work, the identification of W was not investigated in detail, and thus, requires further investigation.