Graduate School of Engineering, Osaka Prefecture University, 1-1 Gakuen-cho, Naka-ku, Sakai, Osaka, Japan

Transcription

1 Advances in Science and Technology Submitted: ISSN: , Vol. 88, pp Accepted: doi: / Online: Trans Tech Publications, Switzerland Transmission Electron Microscopy of Interfaces in Diffusion-Bonded Silicon Carbide Ceramics Hiroshi Tsuda 1,a*, Shigeo Mori 1,b, Michael C. Halbig 2,c, Mrityunjay Singh 3,d and Rajiv Asthana 4,e 1 Graduate School of Engineering, Osaka Prefecture University, 1-1 Gakuen-cho, Naka-ku, Sakai, Osaka, Japan 2 NASA Glenn Research Center, Cleveland, Ohio, USA 3 Ohio Aerospace Institute, Cleveland, Ohio, USA 4 University of Wisconsin-Stout, Menomonie, WI, USA a tsuda@mtr.osakafu-u.ac.jp, b mori@mtr.osakafu-u.ac.jp, c michael.c.halbig@nasa.gov, d mrityunjaysingh@oai.org, e AsthanaR@uwstout.edu Keywords: diffusion bonding, TEM, CVD-, SA-Tyrannohex, interlayer, Ti, Mo B, interfacial phase formation Abstract Diffusion bonding was used to join silicon carbide () to substrates using three kinds of interlayers: physical-vapor-deposited (PVD) Ti coatings (10 and 20 µm) on the substrate, Ti foils (10 and 20 µm), and a Mo B foil (25 µm). Two types of substrates were used: chemical-vapor-deposited and fiber bonded ceramic (SA-Tyrannohex ), the latter having a microstructure consisting of fibers and a carbon layer. The microstructures of the phases formed during diffusion bonding were investigated using transmission electron microscopy (TEM) and selected-area diffraction analysis. TEM samples were prepared using a focused ion beam, which allowed samples to be taken from the reacted area. The effect of the interlayer material and the direction of the fibers in the substrate with respect to the interlayer was evaluated. Scanning electron microscopy and TEM revealed good diffusion bonds in all samples; however, some samples exhibited small amounts of microcracking. The diffusion bonded CVD sample using the 10-µm-thick PVD-Ti interlayer formed more of the stable phase and less of the intermediate phases than the sample using the Ti foil. This behavior was caused by the presence of columnar Ti grains in the interlayer, which may have enhanced the migration of Si and C atoms in the interlayer. In the SA-Tyrannohex samples using the Ti-foil interlayer, the chemical reaction proceeded more rapidly when the fibers were parallel to the interlayer than when they were perpendicular. This behavior was likely caused by the hexagonal carbon layer always facing the Ti interlayer in the sample with perpendicular fibers; this peculiar microstructure reduced the mobility of Si and C migrating into the interlayer. The SA-Tyrannohex sample using the Mo B foil as the interlayer had excellent diffusion bonds with no microcracks or voids. In this system, Mo5Si3C, Mo2C, and Mo5Si3 formed. While phases have anisotropic coefficient of thermal expansion (CTE), the CTE mismatch between those phases and the substrate was apparently smaller than the mismatch in the samples using Ti interlayers. Introduction Silicon carbide () is a very promising material for structural applications in extreme environments, including high temperatures, because of its excellent high-temperature mechanical properties, oxidation resistance, and thermal stability. These advantages have led to be developed and tested for many aerospace and energy applications [1-2], not only as a monolithic material, but also as a matrix and reinforcing fibers in composites. When fabricating complex, large components from brittle ceramics, machining and net-shape forming can be difficult; to overcome these difficulties, simpler units can be joined into a whole part. The most commonly used joining methods include reaction bonding [3 5] and brazing [6-7]. Diffusion bonding techniques are also very promising [8-9]. All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans Tech Publications, (# , Pennsylvania State University, University Park, USA-19/09/16,17:54:02)

2 140 13th International Ceramics Congress - Part B However, forming diffusion bonds with good thermomechanical properties requires detailed knowledge of the phases formed during the reaction. For example, Gottselig et al. [8] and Naka et al. [9] have reported the bonding of with Ti using scanning electron microscopy (SEM), X-ray diffraction (XRD), and elemental analysis. They studied the phases formed and the diffusion path in the Si Ti C system. However, few have used transmission electron microscopy (TEM) to study the phases formed during diffusion bonding in this system, likely because it is difficult to prepare a TEM sample from the bonded area. However, we have previously used a focused ion beam (FIB) to obtain a cleaner, less damaged, and more precisely selected thin TEM specimen from diffusion bonds than possible using traditional techniques such as ion milling [10-11]. In addition to monolithic, there is also strong need for joining fiber reinforced composites. Ishikawa et al. have developed a tough ceramic, SA-Tyrannohex (SA-THX), that consists of a highly ordered, close-packed structure of fine hexagonal, columnar fibers composed of crystalline β- with a thin interfacial carbon layer between the fibers [12-13]. This material has excellent high temperature properties and thermal shock resistance. In the present study, we used diffusion bonding to join to substrates, using three types of interlayers: physical-vapor-deposited (PVD) Ti (10 or 20 µm) coated on the substrate, Ti foils (10 or 20 µm), and a Mo B foil (25 µm). Two types of substrates were used: chemical-vapor-deposited (CVD)- and SA-Tyrannohex (SA-THX). After diffusion bonding, we studied the microstructural effects of the interlayer type and thickness, and of fiber direction (parallel or perpendicular to the interlayer) in SA-THX by analyzing the microstructure and phase formation. The present paper presents a detailed microstructural analysis of the phases formed during diffusion bonding from TEM images and selected-area diffraction (SAD) analysis of the samples prepared by FIB, which allowed us to select microscopy specimens from the reacted area. Experimental Commercially available CVD β- substrates (Rohm & Hass, Woburn, MA, USA) and SA-THX (Ube Industries, Ube, Japan) were used. SA-THX is a fiber-reinforced composite made from Tyranno-SA fiber bundles woven in an eight-harness satin weave, orienting the fibers in transverse and longitudinal directions. To bond to the substrate, we used two types of interlayers: a Ti metallic foil (Goodfellow Corporation, Glen Burnie, MD) with a thickness of 10 or 20 µm, and a Ti coating with a thickness of 10 µm, applied in-house with physical vapor deposition (PVD). To obtain a 10 µm PVD Ti interlayer, we jointed a coated substrate to an uncoated one. To obtain a 20 µm PVD Ti interlayer, we joined two coated substrates together. To bond to the SA-THX substrate, we used a 25 µm Mo B foil, formed by rapid cooling and cold rolling. Before joining, all materials were ultrasonically cleaned in acetone for 10 min. Joints formed using PVD-Ti-coated were diffusion-bonded at 1250 C with a clamping pressure of 24 MPa. Joints formed using the 10 or 20 µm Ti foil as the interlayer were diffusion-bonded at 1200 C with a clamping pressure of 30 MPa. Joining was performed in vacuum for 2 h at the peak temperature under load, followed by slow cooling at 5 C/min. Elemental analysis and phase identification were performed on carbon-coated samples in two ways: electron probe micro-analyzer (EPMA; JEOL 8200 Super Probe) for the bonds formed with PVD Ti, and SEM (JEOL, JXA-8900) for bonds formed with Ti and Mo-B foil. Energy-dispersive spectroscopy (EDS) was also performed on these samples in conjunction with EPMA and SEM. TEM samples were prepared by FIB (FEI, Quant 3D), with details described in the literature. TEM (JEOL, JEM-2000FX) was performed at 200 kv.

3 Advances in Science and Technology Vol Results and Discussion Characterization of Interlayers and Substrate Intensity (arb. unit) (a) 0002 α-ti (b) β- α-ti 0004 α-ti 2θ (degree) Fig. 1 XRD profiles of (a) PVD-coated Ti on the substrate and (b) the Ti foil. We first characterized the 10-µm PVD-Ti-coated interlayer on the CVD- substrate and the 10-µm Ti-foil interlayer; Fig. 1 shows XRD profiles of two interlayers. As shown in Fig. 1(a), the sample using the PVD-Ti exhibited peaks of β- and strong reflections from the (0002) and (0004) planes of α-ti, suggesting that the PVD-coated Ti has a preferred orientation on the CVD- substrate. In contrast, as shown in Fig. 1(b), the sample using the Ti foil exhibited reflections from various crystal planes of α-ti, revealing randomly orientated grains. Fig. 2 shows TEM images of the PVD-Ti-coated interlayer and the Ti-foil interlayer, along with corresponding SAD patterns. The thin TEM specimen from the PVD-Ti interlayer was taken perpendicular to the CVD- substrate. In the PVD-Ti interlayer (Fig. 2(a)), we found columnar α-ti perpendicular to the substrate, growing along [0001] direction of α-ti, determined by SAD analysis. In contrast, in the Ti foil (Fig. 2(b)) we found randomly orientated grains, shown by the ringed SAD pattern. These TEM results agree with our XRD results. (a) PVD 10 µm Ti 0001 (b) 10 µm Ti foil 1010 [1120]α-Ti substrate Fig. 2 TEM images and SAD patterns of (a) PVD-coated Ti on the substrate and (b) the Ti foil. Figs. 3 and 4 show XRD and TEM with SAD patterns for the sample using a Mo B foil interlayer. The XRD pattern shows clear Mo peaks but no characteristic B peaks. Also, the intensities of some of these peaks differed from those of the Joint Committee for Powder Diffraction Standards (JCPDS) card: the intensity of the (011) plane was extremely weak, and those of the (002) and (211) planes were relatively strong, indicating the foil had a preferred orientation. This preferred orientation may have been caused by the cold rolling used to fabricate the 25-µm-thick foil after rapid cooling. Fig. 4 shows a representative TEM image and SAD patterns taken from grains in the Mo B foil interlayer, which exhibited a net-shape pattern and halo pattern, respectively. Based on SAD analysis, the net-shape pattern (e.g., Fig. 4(b)) revealed the presence of crystalline Mo, while the halo pattern (e.g., Fig. 4(c)) revealed the presence of amorphous grains. EDS confirmed the presence of Mo in the crystalline grains and the presence of B in the amorphous grains.

4 142 13th International Ceramics Congress - Part B 1 Intensity (arb. unit) Mo 211 Mo 011 Mo θ (degrees) Fig. 3 XRD profile of the Mo B foil. (a) (b) (1) (2) (c) [011] Mo 0.5µm Fig. 4 TEM image of (a) the Mo B foil and SAD patterns from (b) grain (1) and (c) grain (2). Fig. 5 shows TEM images of the two kinds of substrates, CVD- and SA-THX, revealing details on the substrate microstructures. The average grain sizes of samples with substrates prepared by CVD (Fig. 5(a)) were 3 5 µm. The SAD patterns (insert of Fig. 5(a)) had twin spots with streaks, indicating stacking faults or micro-twins located in the crystalline. In contrast, the average grain sizes of the substrates with SA-THX (Fig. 5(b)) were µm, and their SAD pattern (insert of Fig. 5(b)) revealed them to be crystalline β-. The SA-THX also exhibited twin spots with streaks. (a) (b) [011] [011] 1 μm 1 μm Fig. 5 TEM images and SAD patterns of (a) CVD- and (b) SA-THX. SEM of Diffusion-Bonded Samples Fig. 6 shows SEM images of the diffusion-bonded samples. All samples seemed to be successfully diffusion-bonded, but some had microcracks. Fig. 6(a) and (b) show SEM images of the diffusion bonds for the 10-µm- and 20-µm-thick PVD-Ti-coated samples of CVD. The bond formed with the thinner 10-µm layer (Fig. 6(a)) had no microcracks. EDS of this sample revealed two phases: Ti32 (phase A1: 56% Ti, 19% Si, 25% C) and TiSi2 (phase B1: 36% Ti, 61% Si, 3% C). In contrast, the bond formed with the thicker 20-µm layer (Fig. 6(b)) had microcracks. EDS of this sample revealed three phases: Ti32 (phase A2), TiSi2 (phase B2), and Ti5Si3Cx (phase C2). Fig. 6(c) and (d) show the diffusion bonds in CVD samples using Ti-foil interlayers with thicknesses of 10 and 20 µm, respectively. In Fig. 1(d), the composition of phase A was 51 55% C, 13 14% Si, and 32 35% Ti; for phase B, it was 38 47% C, 22 27% Si, and 31 35% Ti. In the sample using the 10-µm Ti foil (Fig. 6(c)), we observed little microcracking. However, in the sample using the 20-µm Ti foil, we observed significant microcracking, possibly caused by the presence of a phase with relatively low Si content (58% C, 7% Si, 35 %Ti; phase C).

5 Advances in Science and Technology Vol Fig. 6(e) and (f) show SEM images of the diffusion bonds for the SA-THX fibers parallel and perpendicular, respectively, to the 10-µm-thick Ti foil, revealing adequate diffusion bonds in both of these samples. As shown in Fig. 6(e) and (f), hexagonal and straight carbon layers were present on the surface of the SA-THX. The morphology of the interface between the bond and the SA-THX with parallel fibers was smoother than that with perpendicular fibers, and this bond exhibited no microcracking. The width of the diffusion bond with SA-THX fibers parallel to the Ti interlayer, 15 µm, was wider than that of the bond with perpendicular fibers, 9 µm. Thus, the speed of the chemical reaction depends much on the fiber direction. In the sample with fibers perpendicular to the Ti layer, as shown in Fig. 6(f), a small amount of microcracking appeared. (a) (b) A2 (c) (d) A1 B2 A B B1 Carbon layer C2 (e) (f) Si Carbon (g) (h) layer 15µm 9µm 35µm Remaining Mo-B foil 35µm Remaining Mo-B foil 20µm Fig. 6 SEM images of diffusion-bonded samples, (a) PVD-Ti 10µm, (b) PVD-Ti 20 µm, (c) Ti foil 10µm, (d) Ti foil 20µm, (e) 10µm Ti foil SA-THX Parallel, (f) 10µm Ti foil SA-THX Perpendicular, (g) 25µm Mo-B foil SA-THX Parallel and (h) 25µm Mo-B foil SA- THX Perpendicular. Fig. 6(g) and (h) show SEM images of diffusion-bonded SA-THX using 25-µm Mo B foil with fibers parallel and perpendicular to the interlayer, respectively. These images show that the Mo B interlayer formed a better diffusion bond, free from microcracking and voids, than bonds formed using the Ti or Ti Mo interlayers.[14] This diffusion bond was sound for fibers oriented both parallel and perpendicular to the interface, showing that fiber direction in SA-THX did not affect bond quality. Additionally, with this interlayer the bonded area widened to 35 µm, even though the interlayer was only 25 µm wide before diffusion bonding. The SEM images of this bond revealed the presence of at least two phases in the diffusion bonds, differentiated by contrast. In the central areas of the bonds shown in Fig. 6(g) and (h), the Mo B foil interlayer remained unchanged after diffusion bonding. TEM of Diffusion-Bonded Samples TEM was performed to examine the detailed microstructures of phases formed in the diffusion-bonded samples. Figs show TEM micrographs of the samples; Figs. 7 and 8 show TEM images of CVD- joined using the 10-µm PVD Ti and 10-µm Ti foil interlayer, respectively. Figs. 9 and 10 show TEM of the SA-THX, with fibers parallel to the interlayer, joined using the 10-µm Ti foil and 25-µm Mo B foil interlayers, respectively. When a Ti interlayer was used, the diffusion bonds consisted of many small reaction-formed grains with lengths of 2 4 µm and widths of 1 2 µm. When the Mo B foil was used, the formed grains appeared larger than those formed from the Ti interlayers. These figures denote the locations of the SAD pattern as numbers. We performed detailed phase identification by analyzing the SAD patterns from grains in each sample. Table 1 shows the calculated proportions of phases formed during diffusion bonding. Fig. 7 shows the TEM image and identified phases from the sample using the 10-µm PVD-Ti-coated interlayer, revealing the presence of Ti32, TiSi2, and Ti5Si3Cx. The calculated fractions of the phases formed during diffusion bonding were 91.4% Ti32, 2.9% Ti5Si3Cx, and 5.7% TiSi2. These phases are consistent with the results of Gottselig et al. [8]. Naka et al. [9] also

6 144 13th International Ceramics Congress - Part B confirmed the presence of Ti32 and TiSi2 by XRD during final bonding of to using a Ti interlayer. Ti32 was the primary phase and the fractions of the other phases were extremely small, suggesting that the reaction between the substrate and the 10-µm PVD Ti coating interlayer may have been completed. Naka et al. [9] also reported that Ti32 is a stable phase and Ti5Si3Cx is an intermediate phase that contains a low concentration of carbon in Ti5Si3. Although our TEM results for the sample using the 20-µm PVD Ti interlayer are not shown here, we identified 75.9% Ti32, 13.8% Ti5Si3Cx, and 10.3% TiSi2 in the interlayer: less Ti32 and considerably more Ti5Si3Cx than in 20-µm-Ti foil sample. For the sample with the thinner 10-µm Ti-foil interlayer, the reactions between and Ti appeared to have been completed after holding for 2 h due to the presence of the stable phase. In contrast, when the thicker 20-µm Ti interlayer was used, the reaction appeared incomplete when processed at the same temperature and hold time. Fig. 8 shows a TEM image and the phases of the sample using the 10-µm Ti-foil interlayer, revealing 63.5% Ti32, 18.2% Ti5Si3Cx, 6.1% TiSi2, 6.1% TiC, and 6.1% unknown phases. For the sample using the 20-µm Ti-foil interlayer, the fractions were 37.5% Ti32, 43.7% Ti5Si3Cx, 3.1% TiSi2, 9.4% TiC, and 6.3% unknown phases. Comparing the samples using PVD-Ti and Ti-foil interlayers with the same thickness, we found the chemical process proceeded rapidly in the PVD-Ti, forming Ti32 and TiSi2 quickly. This behavior may have been caused by the structure of the Ti: 1 µm wide, columnar, and perpendicular to the substrate. This structure allowed Si and C atoms to migrate freely in the columnar interlayer substrate µm Fig. 7 TEM image and identified phases for sample with 10-µm PVD-Ti interlayer. (CVD- substrate) 4 5 Ti 3 2 Ti 5 Si 3 Cx TiSi Ti 3 2 Ti 5 Si 3 C x TiSi 2 TiC unknown μm Fig. 8 TEM image and identified phases For sample with 10-µm Ti-foil interlayer. (CVD- substrate) Ti 3 2 Ti 5 Si 3 Cx TiSi 2 TiC 8 unknown Fig. 9 shows TEM image of the SA-THX fibers parallel to the 10-µm-Ti foil, with SAD patterns taken at the labeled locations to determine the probable phases; Table 1 shows the calculated percentages of these phases. TEM indicated that the diffusion bonds in samples with parallel SA-THX fibers consisted of 84.2% Ti32, 5.3% Ti5Si3Cx, and 10.5% TiSi2. In this diffusion bond, we found more Ti32 and less Ti5Si3Cx; as mentioned before, Ti32 was the primary phase and the fractions of other phases were extremely small. This behavior suggests that the reaction between the SA-THX with parallel fibers and the 10-µm Ti-foil interlayer had finished. In contrast, from TEM of the sample with SA-THX fibers perpendicular to the Ti foil (data not shown), we identified 63.7% Ti32, 9.1% Ti5Si3Cx, and 13.6% TiSi2, as well as an unidentified phase. Compared with the sample with parallel fibers, the sample with perpendicular fibers had a smaller Ti32 fraction and a much greater Ti5Si3Cx fraction. For the sample with fibers parallel to the Ti foil, the reactions between the and Ti completed after a 4 h hold. In contrast, for the sample with perpendicular fibers the reaction remained incomplete when processed at the same temperature and hold time. This difference can be explained by assessing the sample geometry; when the fibers are perpendicular to the Ti layer, the basal plane of the carbon layer in the hexagonal columns will always face the Ti interlayer, just

7 Advances in Science and Technology Vol like a grain boundary. This structure likely reduced the mobility of Si and C atoms migrating into the Ti interlayer during diffusion bonding. In contrast, when fibers are parallel to the Ti layer, the Si and C atoms can migrate readily, because the prismatic plane of the carbon layer in the hexagonal columns commonly does not face the Ti foil. This facing enhanced the mobility of the Si and C atoms into the Ti interlayer during diffusion bonding. Fig. 10 shows TEM image of the sample with SA-THX fibers parallel to the Mo B foil. Mostly Mo5Si3C and Mo2C formed in these bonds. Sometimes, two polymorphs of Mo2C coexisted in the same grain: β-mo2c, a hexagonal structure stable at higher temperature, and α-mo2c, an orthorhombic structure stable at lower temperatures [15]. According to previous report [16], α-mo2c may precipitate from β-mo2c during cooling after diffusion bonding at 1500 C. In the sample with fibers parallel to the Mo B foil, the volume fraction of Mo5Si3C was 52.6%, and that of Mo2C was 47.4%. In contrast, in the sample with perpendicular fibers, Mo5Si3C formed less and Mo2C formed more, comparing the sample with parallel fiber, and Mo5Si3 and a small amount of an unknown phase formed. Martinelli et al. [17] reported that, when α- was joined to Mo by solid-state bonding at C, Si and C atoms from the decomposing first diffused into Mo, producing a reaction layer with two main phases: Mo3Si and Mo2C. As the reaction progressed, further diffusion of Si transformed Mo3Si into Mo5Si3. At >1400 C, Mo5Si3C also formed by diffusion of C through the previously formed Mo5Si3. Although MoSi2 can form early in the reaction, the presence of C prevents this phase from forming. Additionally, MoC can form at >1700 C. In the present study, diffusion bonding was performed at 1500 C, and our TEM results agree with those of Martinelli et.al. [17]. Ti 3 SA-THX 2 Ti 5 Si 3 C x TiSi 2 Diffusion bond Carbon layer µm 11 8 interface Mo 5 Si 3 C Mo 2 C 13 8 Mo 2 C+α-Mo 2 C μm Ti32 Ti5Si3Cx TiSi2 Mo5Si3C Mo2C Mo2C+α- Mo2C Fig. 9 TEM image of diffusion-bonded sample with 10-µm Ti foil. (SA-THX substrate, fibers parallel to the interlayer.) Fig. 10 TEM images of diffusion-bonded samples with Mo B foil interlayer (SA-THX substrate, fibers parallel to the interlayer). Table 1 Calculated volume fraction of phases formed during diffusion bonding. (%) Substrate CVD- SA-THX Interlayer PVD-Ti Ti foil Ti foil Mo-B foil thickness (µm) fiber direction Parallel Perpen- Perpen- Parallel dicular dicular Ti Ti 5Si 3C x TiSi TiC unknown Mo 5Si 3C Mo 2C Mo 5Si unknown Total

8 146 13th International Ceramics Congress - Part B When we consider the formation of microcracks in the diffusion bonds, we must consider the mismatch of the coefficients of thermal expansion (CTEs) of the phases formed during diffusion bonding and that of the substrate, because CTE mismatches and anisotropy can cause microcracking from thermal stresses forming during cooling after diffusion bonding. As mentioned above, many microcracks were observed when the Ti5Si3Cx intermediate phase formed more and the stable Ti32 phase formed less. The CTE and anisotropy of Ti5Si3Cx (almost the same as Ti5Si3) are both much larger than those of other phases, as shown in Table 2 [15, 18-20]. This correlation suggests that the large CTE mismatch between Ti5Si3 and as well as the strong anisotropy of the CTE may have led to significant microcracking. In contrast, in the Mo system, Mo5Si3C and Mo2C formed; these phases have anisotropic CTE, but both the mismatch and anisotropy of their CTE are less than those of Ti5Si3 in the Ti system, leading to less significant microcracking. Table 2 Coefficient of thermal expansion (CTE) and anisotropy of various phases. Phase axis CTE, α ( 10-6 K -1 ) Anisotropy of CTE (α c / α a) Reference 4.4 [15] Ti 5Si 3 Ti 3 2 Mo 5Si 3 Mo 5Si 3C Mo 2C a 6.11 c a 8.9 c 10.0 a 5.82 c a 6.4 c 12.6 a 4.9 c [18] 1.12 [19] 2.01 [18] 1.97 [20] 1.67 [15] Conclusions 1. SEM and TEM revealed that all samples had suitable diffusion bonds. 2. The sample using the 10-µm-thick PVD-Ti interlayer formed more of the stable phase (Ti32) and less of the intermediate phases (Ti5Si3Cx) than the sample using the Ti-foil interlayer. This behavior was caused by the presence of the columnar Ti in the interlayer formed on the CVD- substrate, which may have enhanced the migration of Si and C atoms in the interlayer. 3. In the sample using the Ti-foil interlayer to join SA-THX, the chemical reaction proceeded more rapidly when fibers were parallel to the interlayer than when they were perpendicular. This behavior was likely caused by the presence of the hexagonal carbon layer that always faced the Ti interlayer in the sample with perpendicular fibers; this peculiar microstructure reduced the mobility of Si and C atoms into the interlayer. 4. Samples using the Mo B foil as the interlayer to join SA-THX had excellent diffusion bonds with no microcracks or voids. In this system, Mo5Si3C, Mo2C, and Mo5Si3 were formed. These samples had anisotropic coefficients of thermal expansion (CTE), but the mismatch of CTE of substrate and those phases was apparently smaller than those of Ti interlayer samples, leading to few microcracks in the sample using the Mo B interlayer. References [1] P. J. Lamicq, G. A. Bernhart, M. M. Dauchier, and J. G. Mace, / Composite Ceramics, Am. Ceram. Soc. Bull., 65(2), (1986). [2] M. Halbig, M. Jaskowiak, J. Kiser, and D. Zhu, Evaluation of Ceramic Matrix Composite Technology for Aircraft Turbine Engine Applications, proceedings of the 51st AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition (2013). [3] M. Singh, A Reaction Forming Method for Joining of Silicon Carbide-based Ceramics, Scr. Mater., 37(8), (1997).

9 Advances in Science and Technology Vol [4] M. Singh, Joining of Sintered Silicon Carbide Ceramics for High Temperature Applications, J. Mater. Sci. Lett., 17(6), (1998). [5] M. Singh, Microstructure and Mechanical Properties of Reaction Formed Joints in Reaction Bonded Silicon Carbide Ceramics, J. Mater. Sci., 33, 1 7 (1998). [6] V. Trehan, J. E. Indacochea, and M. Singh, Silicon carbide brazing and joint characterization, J. Mech. Behav. Mater., 10(5 6), (1999). [7] M. G. Nicholas, Joining Processes: Introduction to Brazing and Diffusion Bonding, Kluwer Academic Publishers, Dodrecht, [8] B. Gottselig, E. Gyarmati, A. Naoumidis, and H. Nickel, Joining of Ceramics Demonstrated by the Example of /Ti, J. Eur. Ceram. Soc., 6, (1990). [9] M. Naka, J. C. Feng, and J. C. Schuster, Phase Reaction and Diffusion Path of the /Ti System, Metall. Mater. Trans. A, 28A, (1997). [10] H. Tsuda, S. Mori, M. C. Halbig, and M. Singh, TEM observation of the Ti Interlayer between Substrates during Diffusion Bonding, Proceedings of ICACC 2012, (2012). [11] M. C. Halbig, M. Singh, and H. Tsuda, Integration Technology for Silicon Carbide-BasedCeramics for Micro-Electro-Mechanical Systems-Lean Direct Injector Fuel Injector Applications,Int. J. Appl. Ceram. Tec., 9, (2012). [12]T. Ishikawa, S. Kajii, K. Matsunaga, T. Hogami, Y. Kohtoku and T. Nagasawa, A Tough, Thermally Conductive Silicon Carbide Composite with High Strength up to 1600 C in Air, Science, 282, (1998). [13]T. Ishikawa, Y. Kohtoku, K. Kumagawa, T, Yamamura and T. Nagasawa, High-Strength Alkali-Resistant Sintered Fiber Stable to 2200 C, Nature, 391, (1998). [14] M. C. Halbig, M. Singh, and R. Asthana, Diffusion Bonding of Fiber-Bonded Ceramics using Ti/Mo and Ti/Cu Interlayers. To be published. [15]A. E. Martinelli, PhD. Dissertation, Diffusion Bonding of Silicon Carbide and Silicon Nitride to Molybdenum, McGill University (Nov. 1995). [16]J. Kouvetakis and L. Brewer, Temperature Stability Range of the Binary MoC phase, J. Phase Equilib., (1992). [17]A. E. Martinelli and R. A. L. Drew, Microstructural Development during Diffusion Bonding of α- to Molybdenum, Materials Science and Engineering, A, 19, (1995). [18]Schneibel, C. J. Rawn, E. A. Payzant and C. L. Fu, Controlling the Thermal Expansion Anisotropy of Mo5Si3 and Ti5Si3 Silicides, Intermetallics, 12, , (2004). [19]T. H. Scabarozi, S. Amini, O. Leaffer, A. Ganguly, S. Gupta, W. Tambussi, S. Clipper, J. E. Spencer, M. W. Barsoum, J. D. Hettinger and S. E. Loflamd, Thermal Expansion of Select M>n+1AX>n (M=early Transition Metal, A=A Group Element, X=C or N) Phases Measured by High Temperature X-ray Diffraction and Dilatometry, Journal of Applied Physics, 105, (2009). [20]T. Hayashi, K. Ito and K. Tanaka, Physical and Mechanical Properties of Single Crystals of the Mo5Si3C phase, Intermtallics, 11, , (2003).