- Email: [email protected]
Cu interlayers
Available online at www.sciencedirect.com
CERAMICS INTERNATIONAL
Ceramics International 41 (2015) 2140–2149 www.elsevier.com/locate/ceramint
Diffusion bonding of SiC fiber-bonded ceramics using Ti/Mo and Ti/Cu interlayers M.C. Halbiga, R. Asthanab,n, M. Singhc a NASA Glenn Research Center, Cleveland, OH, USA University of Wisconsin-Stout, 326 Fryklund Hall, Menomonie, WI 54751, USA c Ohio Aerospace Institute, Cleveland, OH, USA
b
Received 4 August 2014; received in revised form 1 October 2014; accepted 1 October 2014 Available online 13 October 2014
Abstract A SiC fiber-bonded ceramic (SA-Tyrannohex™) was diffusion bonded using Ti/Mo and Ti/Cu interlayers. The influence of metallic interlayers and SiC fiber orientation in the ceramic substrate with respect to the interlayers on joint microstructure, elemental composition, and microhardness in diffusion bonds was investigated using Optical Microscopy (OM), Field Emission Scanning Electron Microscopy (FE-SEM), Energy Dispersive Spectroscopy (EDS), and Knoop microhardness test. Compared to the Ti/Mo bilayers, the Ti/Cu bilayers yielded higher quality joints. The reaction products distributed more homogeneously across the joint thickness in Ti/Cu bonds than in Ti/Mo bonds. The reaction layers adjacent to the SiC substrate in both parallel and perpendicular SA-THX/Mo/Ti/SA-THX joints were twice as hard as the joint center where the Mo interlayer had remained untransformed during diffusion bonding. In SA-THX/Cu/Ti joints, hardness distribution was uniform across the joint thickness consistent with a more homogeneous reaction phase distribution across the joint. & 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
Keywords: B. Electron microscopy; Diffusion bonding; Silicon carbide; Knoop hardness; Reaction layers
1. Introduction Silicon carbide (SiC) based materials have excellent hightemperature mechanical properties, oxidation and heat resistance, and thermo-chemical stability. These materials are leading candidates for various thermostructural applications at high temperatures in harsh environments in aerospace and energy sectors [1–3]. A variety of advanced SiC-based ceramics such as CVD SiC, sintered SiC, hot-pressed SiC, and SiC fiber-bonded ceramic (SA-Tyrannohex™) as well as ceramic composites such as SiC/SiC and C/SiC have been developed for such applications. Implementing these advanced ceramics and composites in real components demands robust ceramic joining and integration technologies. The most commonly used ceramic joining methods include reaction bonding [4] and brazing [5,6]. Diffusion bonding techniques have been successfully utilized to bond n
Corresponding author. Tel.: þ1 715 232 2152; fax: þ 1 715 232 1330. E-mail address: [email protected] (R. Asthana).
http://dx.doi.org/10.1016/j.ceramint.2014.10.014 0272-8842/& 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
a wide variety of SiC ceramics. Sintered, hot-pressed and CVD silicon carbide ceramics have been diffusion bonded using refractory metal interlayers of titanium [7–10], molybdenum [11,12], tantalum [13], tungsten [14], niobium [13], zirconium [10], nickel [15] and Inconel 600 [16]. The diffusional transformation of a metal interlayer at high temperatures and under high mechanical pressures into carbides, silicides, or complex ternary and higher order compounds produces strong joints. Carbides and silicide compounds of refractory metals are also thermodynamically more stable than SiC. As a result, diffusive conversion of metal insert into carbides and silicides provides a pathway for strong bond formation. Previously, some of the present authors [17–19] investigated the diffusion bonding of SiC/Ti/SiC joints and identified the phases that form in the bonded area using scanning electron microscopy (SEM), X-ray diffraction (XRD) analysis, and energy dispersive spectroscopy (EDS). Recently, the authors conducted a detailed transmission electron microscopy (TEM) study for the structural evaluation of the phases formed during
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
diffusion bonding with Ti as the interlayer [20,21]. All prior studies investigated diffusion bonding of sintered, hot pressed, and CVD SiC ceramics, and apart from the present authors’ recent research [22] on diffusion bonded SA-Tyrannohex™ (also referred to as SA-THX) using Ti interlayers, no published reports currently exist on diffusion bonding of SiC fiberbonded ceramic, SA-THX. The liquid-phase joining of SATHX via brazing was also only recently demonstrated [23–25]. SA-THX is a dense and tough ceramic consisting of a highly ordered, closed-packed structure of fine hexagonal columnar fibers composed of crystalline β-SiC with a thin (10–20 nm) interfacial layer of turbostatic carbon between the fibers [26,27]. SA-THX is a product of Ube Industries, Japan, and is obtained by converting an amorphous Si–Al–C–O fiber into Tyranno-SA fiber with near-stoichiometric composition. The final SiC fiber-bonded ceramic, SA-THX, is made by hotpressing of fiber weave lay-ups into various configurations. The resulting material exhibits good thermomechanical performance, high thermal conductivity (33 W/m K at 1600 1C), high strength up to 1600 1C, high flexural strength ( 300 MPa), high fracture toughness (1200 J/m2 at room temperature), and good machinability and low cost. In this study, SA-THX was diffusion bonded to itself using either Ti/Mo or Ti/Cu bilayers to examine the effect of the interlayers and SiC fiber orientation (perpendicular and parallel relative to the joining plane) on the microstructure, composition and hardness of joints using Optical Microscopy (OM), Field Emission Scanning Electron Microscopy (FE-SEM), Energy Dispersive Spectroscopy (EDS), and the Knoop microhardness test. 2. Experimental SA-Tyrannohex™ (SA-THX) SiC fiber-bonded ceramic was obtained from Ube Industries (Ube, Japan). The SA-THX is made from SA-Tyranno fiber bundles woven into an eightharness satin weave that results in fibers being oriented in parallel (‖) and perpendicular ( ? ) directions. Joints were made using like SA-THX composite pairs (‖-to-‖, or ? -to- ? ). Ti foil (10 μm), Mo foil (12.7 μm), and Cu foil (5 μm) were obtained from Goodfellow Corporation (Glen Burnie, MD, USA). Before joining, all materials were ultrasonically cleaned in acetone for 10 min. Joints were diffusion bonded at 1200 1C with a pressure of 30 MPa. Joint processing was conducted in vacuum for 4 h at the peak temperature under load, followed by cooling at a rate of 2 1C per min. Joints with Ti/Cu bilayers had three Cu interlayers (total thickness: 15 μm) with 10 μm thick Ti foil on two sides. Similarly, for Ti/Mo bilayers, a 12.7 μm Mo foil was sandwiched by 10 μm thick Ti foil on two sides. The nomenclature used to identify the joints for subsequent discussion is as follows. A SA-THX joint with Ti and Mo interlayers having ? fibers coincident on the foil is denoted as follows: ? SA-THX/Ti/Mo/Ti/ ? SA-THX. Similarly, a SA-THX joint with ‖ fibers coincident on the foil is denoted as: ‖ SA-THX/Ti/Mo/Ti/‖ SA-THX. Optical Microscopy (OM) was done on an Olympus BX51, and SEM observations and elemental analyses were performed
2141
with a Hitachi S-4700-I Field Emission Scanning Electron Microscope (FE-SEM) coupled with an Energy Dispersive Spectrometer (EDS). Joint hardness was characterized using Knoop indenter on a Struers Duramin-A300 machine under a load of 200 g and loading time of 10 s. 3. Results and discussion 3.1. Joint microstructure and composition 3.1.1. SA-THX/Ti/Mo/Ti SA-THX joints In designing the SA-THX joints using Ti/Mo bilayers, dual objectives were sought to be accomplished. One objective was to take advantage of a low coefficient of thermal expansion (CTE) of Mo interlayer (CTE: 5.1 10 6 K 1) to reduce residual stresses in diffusion bonded joints. A small CTE mismatch mitigates the residual stress buildup from joining. Experience has shown that judiciously chosen interlayers positioned in the joint prior to bonding can reduce the strain energy and cracking propensity in ceramic joints. A second objective was to explore the feasibility of lowering the bonding temperature relative to direct bonding using only Mo foils. Past research [11,14] has shown that SiC/Mo joints diffusion bonded under proper conditions have excellent bond quality; in fact, the quality of the SiC/Mo/SiC bond was significantly better than the diffusion bonds produced with Ti. Cockeram [14] observed that high temperatures (1500 1C) and long contact times (10 h) produced sound SiC/Mo joints under 3.4 to 18 MPa pressures with 12.7 mm thick Mo foils. Similar joints processed at 1200 1C led to poor bond quality. In Ref. [14], it was also noted that use of thicker molybdenum foil (25.4 mm) resulted in a significantly higher density of cracks in the bond region in a manner similar to the diffusion bonding of SiC using Ti where use of thin rather than thick foils improved the quality of the interfacial bond [14]. Thus although Mo (CTE: 5.1 10 6 K 1) and SiC are better CTE matched than are Ti (CTE: 8.6 10 6 K 1) and SiC, thin Ti (10 mm) and Mo (12.7 mm) metallic foils were both used together in the present work with the goal to achieve sound diffusion bonds at relatively low (1200 1C) operating temperatures. Titanium and molybdenum form an isomorphous binary alloy system and Ti and Mo are completely miscible above 882 1C. Below 882 1C, the solubility of Mo in Ti is restricted (e.g., the maximum solubility of Mo in α-Ti is less than 0.5% at 600 1C [11]). It was envisioned that formation of a Ti(Mo) solid solution in conjunction with the carbides and silicides from the reactions between Ti and SiC and between Mo and SiC shall reduce the residual stresses in SiC joints while still forming strong bonds and lowering the energy required for diffusion bonding as a result of reduced process temperatures (from 1500 1C to 1200 1C). A 12.7 μm Mo foil was sandwiched by 10 μm Ti foils on two opposite sides in pairs of parallel and perpendicular substrates. Optical images of parallel (‖ SA-THX/Ti/Mo/Ti/‖ SA-THX) and perpendicular ( ? SA-THX/Ti/Mo/Ti/ ? SATHX) joints are shown in Fig. 1a and b, respectively. The ends
2142
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
Fig. 1. Optical photomicrographs of (a) a parallel and (b) a perpendicular SA-THX/Ti/Mo/Ti/SA-THX joint showing bonded regions near left edge, center, and right edge of flat-back bonded SA-THX samples.
Location 1
Location 2
Fig. 2. SEM images of a ‖ SA-THX/Ti/Mo/Ti//‖ SA-THX joint showing overall views of the joint from two different locations and a BSE image of location 2. The EDS elemental compositions at point markers shown above are presented in Table 1.
of the joints show some un-bonded regions due to one of the three tape layers being shorter. No large-scale cracks appear in the ceramic; however, toward the center of the joint, evidence of micro-cracking is noted. SEM images (Figs. 2 and 3) of the joints reveal microcracks, 8–10 μm long, in the vicinity of bonded interfaces within the interlayers in both parallel and perpendicular orientations. These microcracks originate at the second layer from the interface and terminate at the Mo foil that resides in the center of the joint. Some of the micro-cracks observed metallographically could be sample preparation artifacts but similar preparation for Ti/Cu bilayer joints did not produce cracks as discussed later. Redistribution of Si, C and Ti is evident from the EDS analysis of the joints presented in Tables 1 and 2 for parallel and
perpendicular joints, respectively. There is a 6–8 μm thick continuous dark gray band decorating the sides in contact with SiC in both parallel and perpendicular joints (Figs. 2 and 3). For the parallel joint, this band has an average elemental composition that is consistent with the presence of Ti3SiC2 (point markers 5 and 6, Fig. 2) and also possibly Ti5Si3 (point 7). There also seems to be a possibility of TiC formation (54 at% Ti and 44 at% C at point markers 10, 11 and 20, Fig. 2). There appears to be a carbon concentration gradient over 5–7 μm near the interface with C concentration decreasing from 33.7 at% (point markers 5 and 6, Fig. 2) to 12.2 at% (point markers 8 and 9). At points 8 and 9, EDS suggests the composition to be Ti5Si3Cx. The ternary compound Ti5Si3Cx is an intermediate phase that is transformed into Ti3SiC2 following completion of the reaction [8] during
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
2143
Location 1
Region A
Location 2 Layer II Layer I
Fig. 3. BSE SEM images of a? SA-THX/Ti/Mo/Ti/? SA-THX joint showing overall views of the joint from two different locations and a BSE image of location 2. The EDS elemental compositions at point markers shown above are presented in Table 2.
Table 1 Elemental composition in ‖ SA-THX/Ti/Mo/Ti/‖ SA-THX Joint of Fig. 2.
Table 2 Elemental composition in ? SA-THX/Ti/Mo/Ti/? SA-THX joint of Fig. 3.
Marker in Fig. 2
C, atom % Si, atom % Ti, atom % Mo, atom %
Marker in Fig. 3
C, atom %
Si, atom % Ti, atom % Mo, atom %
1 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
52.777 56.752 53.05 30.641 36.807 26.667 12.043 12.297 2.331 42.585 14.056 20.238 20.411 6.252 5.571 16.09 14.021 37.641 43.565 10.384
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
44.865 46.334 46.627 45.175 22.458 31.118 33.991 47.613 38.051 19.101 10.365 20.118 10.334 40.43 31.904 41.148 34.359 13.178 15.371 15.555 14.245 13.321 14.631 16.573 17.609 23.299 17.26 14.23
54.275 52.493 52.301 53.913 18.904 14.216 0.940 0.870 0.188 26.977 29.075 26.596 29.876 2.711 0.153 0 0.696 33.638 35.363 35.119 35.933 0 1.152 0.215 0.165 0.34 0.532 0.131
47.223 43.248 46.95 18.381 17.914 19.528 36.823 35.729 42.036 2.105 0.376 29.58 28.909 0.353 0.455 28.511 30.807 0.345 3.961 35.268
0 0 0 50.979 45.279 53.805 51.134 51.973 55.633 55.31 0.481 21.307 17.625 2.45 0.575 21.646 21.492 61.755 51.938 52.918
0 0 0 0 0 0 0 0 0 0 85.087 28.876 33.055 90.945 93.399 33.752 33.68 0.26 0.536 1.43
diffusion bonding, thus leaving Ti3SiC2 as the primary phase. The Ti3SiC2 phase has a large but stable CTE of 10 10 6 K 1 up to about 800 1C [14,28]. The intermediate phase Ti5Si3Cx has an anisotropic coefficient of thermal expansion (CTE) [29,30] that exacerbates the propensity for micro-cracking. It is evident that both C and Si have dissolved and diffused in the Ti layer but not in the Mo layer in the center (point markers 12, 15 and 16, Fig. 2) where the Mo concentration exceeds 90 at%. The interface between Mo and Ti is diffuse and delineates a band containing lenticular precipitates. These precipitates (point marker 19, Fig. 2) have a composition that is consistent with sub-stoichiometric carbide, TiC1.6. The perpendicular ( ? SA-THX/Ti/Mo/Ti/ ? SA-THX) joints (Fig. 3) also show transverse microcracks in the bonded region that are arrested in the diffuse layer between Ti and Mo.
0.744 1.076 1.072 0.913 58.601 54.264 65.069 51.508 61.762 21.976 28.891 22.965 25.981 56.707 67.761 58.741 64.688 52.797 48.637 48.188 49.711 20.457 24.71 22.445 20.003 2.836 0.609 0.691
0.116 0.096 0 0 0.037 0.402 0 0.009 0 31.946 31.669 30.321 33.809 0.153 0.182 0.112 0.258 0.387 0.629 1.138 0.11 66.222 59.507 60.767 62.223 73.524 81.599 84.948
Cracks originate in the second layer (marked on the figure) from the interface, which, according to the EDS data (Table 2), appears to be the Ti5Si3Cx phase. In a manner similar to parallel joints, a 6–8 μm thick continuous dark gray band has developed on either side of the joint in Fig. 3. The composition of the interface region is consistent with the formation of the following possible phases: Ti3SiCx (point markers 5 and 6, Fig. 3), TiC (point marker 8), and sub-stoichiometric
2144
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
carbides TiCx (x: 0.5–0.6, point markers 7 and 9). Interestingly, one side (point markers 5 and 6 in Fig. 3) has Si (14– 19 at%) but the opposite side (point markers 7, 8 and 9) does not have Si; there are only Ti and C on the Si-depleted side even though the interaction zone thickness is the same on the two sides. This difference could arise from the geometrical configuration during diffusion bonding as explained in Refs. [19,20]. In perpendicular orientation, the Ti foil could contact either the basal plane of the carbon layer in the hexagonal columns or the prismatic planes. In the plane of Fig. 3, the top side contacts the basal plane that limits C diffusion into Ti foil. On the bottom side of the joint in Fig. 3, a large fraction of prismatic planes contacts Ti thus facilitating diffusion of C. As a result of increased carbon content in the region, the relative percentage of Si drops. This is consistent with the formation of titanium carbide rather than titanium–silicon–carbide phase. In Figs. 2 and 3, the interfacial reaction products appear to have coalesced, aided by the high pressures and high temperatures of diffusion bonding, to yield a compact and homogeneous layer of the reaction product. No grain structure in the reaction product layer could be identified at the magnifications used. In Ref. [22], we had reported that in SA-THX/Ti/SA-THX diffusion bonds, many small grains had formed with the major and minor dimensions of 2–4 μm and 1–2 μm, respectively [22]. Figs. 2 and 3 do not show evidence of grains although confirmatory tests using the TEM shall be needed to validate this. Further out from the interface (point markers 19 through 21, Fig. 3), the elemental compositions again suggest a compound of the type Ti5Si3Cx which could indicate formation of TiSi and TiC phases. Further away from the interface (point markers 10 through 13, Fig. 3), the atomic percentages indicate the following ratios: Ti3Si3CMo3 (point markers 11 and 13, Fig. 3), and TiSiCMo2 (point markers 10 and 12). The Ti, Si and Mo contents are nearly identical at point markers 10–13; however, point markers 10 and 12 have twice the carbon than point markers 11 and 13. The first atomic ratio (Ti3Si3CMo3) is consistent with the formation of either Ti3SiC þ Mo3Si (cubic) or TiSi þ TiC þ Mo3Si. Likewise, the second atomic ratio (TiSiCMo2) could indicate the formation of TiSi þ Mo2C phases (Mo2C is known to form in SiC/Mo/ SiC diffusion bonds [11,12,14]). The dark gray plate-like precipitates (point markers 14 through 17, Fig. 3) are all likely to be carbides of Ti; their average composition suggests sub-stoichiometric compounds, TiC0.5 (point markers 15 and 17) and TiC0.7 (point markers 14 and 16). Unlike parallel joints, where the joint center was nearly pure Mo, the region (marked as region ‘A’, Fig. 3) between the gray interfacial band and the center in perpendicular joints (point markers 22 through 25) shows likely formation of β-Mo(Ti) solid solution with 20–25 at% Ti, and this is consistent with the binary Ti–Mo diagram at 1200 1C [29]. Right in the center of the joint (point markers 26–28, Fig. 3) is nearly pure untransformed molybdenum foil (with evidence of some diffusiondriven carbon, about 14–23 at%). Past research [11,12,14] on SiC/Mo/SiC diffusion bonds (T: 1300–1700 1C) has shown that diffusion of Si and C into Mo
forms distinct layers of Mo5Si3 þ carbon and molybdenum carbide (Mo2C). The formation of these phases shall be detrimental to the diffusion bond because of highly anisotropic expansion properties of Mo2C and Mo5Si3. The thermal expansion of Mo5Si3 (tetragonal structure) is strongly anisotropic and the CTE along ‘a’ and ‘c’ axes are αa=5.2 10 6 K 1 and αc=11.5 10 6 K 1, respectively [31] (these CTE values are essentially temperature independent over a wide range). Similarly, the CTE of Mo2C is anisotropic with values of 4.9 10 6 K 1 along the a-axis and 8.2 10 6 K 1 along the c-axis [11]. Conceivably, the strongly anisotropic expansion of Mo5Si3 and Mo2C that possibly form in the diffusion bond is responsible for the microcracking observed in joints with Ti/Mo metallic interlayers. It must be mentioned that other phases such as Mo5Si3C and MoC could also form in diffusion bonded joints; however, Mo5Si3C forms in SiC/Mo joints only above 1400 1C and MoC forms above 1700 1C [11]. These deleterious phases are, therefore, unlikely to form in our joints. It is noteworthy that the reaction layer, as it formed in the Ti/Mo bilayer joints during diffusion bonding, did not disintegrate under mechanical pressure of 30 MPa during diffusion bonding even though the joints developed micro-cracks (Figs. 2 and 3) to relieve the stress. The application of large compressive stresses (30 MPa) during diffusion bonding on softer Ti foil (yield strength: 170– 480 MPa, depending upon the grade) that resided between harder SA-THX substrate and a hard Mo foil (yield strength: 415–550 MPa) could also have caused micro-cracking especially when the formation of brittle intermetallic reaction products in joints fabricated under pressure is considered. In summary, even though use of Ti in conjunction with Mo lowered the diffusion bonding temperature from the acceptable value of 1500 1C to 1200 1C and yielded sound metallurgical bonding with SA-THX, the bimetallic interlayers failed to yield crack-free joints. It is conceivable that judicious selection of interlayer thickness ratios could limit the volume of deleterious intermetallic compounds that form thus yielding sound joints even at the low fabrication temperature of 1200 1C. Optimization of the interlayer thickness ratio shall constitute a focus of our continuing research effort.
3.1.2. SA-THX/Ti/Cu/Ti/SA-THX joints In designing Ti/Cu bilayer joints, the goal was to accommodate residual stresses via plastic deformation of copper. Although Cu has a very large CTE (16.6 10 6 K 1) and a large CTE mismatch with Ti and SiC, its low yield strength (70 MPa) and high ductility (45% elongation for pure annealed Cu) are advantageous to managing residual stresses in bonded joints. It is known that a hard/soft interlayer combination enables plastic deformation of the second ductile interlayer between the first hard interlayer [32]. It is also known that multiple interlayers reduce the strain energy in the ceramic more than a single interlayer of the same total thickness. Bimetallic interlayers have been used to diffusion bond SiC/metal couples such as SiC/steel using W/Ni interlayers [33]. In the present work, three Cu foils, each with a thickness of 5 μm were
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
2145
Fig. 4. Optical photomicrographs of (a) a parallel and (b) a perpendicular SA-THX/Ti/Cu/Ti/SA-THX joint showing bonded regions near left edge, center, and right edge of flat-back bonded SA-THX samples.
sandwiched by 10 μm thick Ti foils on the two opposite sides for diffusion bonding. Optical photomicrographs of parallel (‖ SA-THX/Ti/Cu/Ti/‖ SA-THX) and perpendicular (? SA-THX/Ti/Cu/Ti/ ? SATHX) joints are shown in Fig. 4a and b, respectively. In parallel joints, the outer edges of the bonded SA-THX substrates showed some un-bonded regions and microcracking (Fig. 4a). No such cracks were evident at the outer edges of perpendicular SA-THX joints (Fig. 4b). The center of both parallel and perpendicular joints was crack-free and exhibited excellent bonding. SEM images (Figs. 5 and 6) of the middle portions of the bonded SA-THX substrates for both orientations revealed crackfree diffusion bonds together with a high volume fraction of a secondary phase throughout the joint volume. This is in contrast to the Ti/Mo bilayer joints where distinct continuous bands of reaction layers (6–8 μm thick) had formed at the interface and the reaction layers had remained confined to near-interface regions rather than extending to the center of the joint. The Ti/Cu bilayer joints reveal more extensive interfacial roughening in parallel joints (e.g., SEM images for locations 1 and 2, Fig. 5). The reaction phase has grown along inter-fiber boundaries because of increased availability of carbon for the reaction to occur in these regions. The mechanical anchor points appear to be more near the top of the joint (locations 1 and 2, Fig. 5) than the bottom. However, a polished section oriented perpendicular to the plane of the figure revealed similar incidence of anchor points near the bottom half. Such anchor points potentially could contribute to enhanced bonding via mechanical keying. Compared to parallel joints, the interfacial roughening is modest in perpendicular joints (Fig. 6) and seems to be related to the reduced reaction intensity in the perpendicular orientation as also discussed previously. Silicon and carbon atoms can
migrate readily when the SA-THX fibers are set parallel to the Ti layer, because the prismatic plane of the carbon layer in the hexagonal columns commonly does not face the Ti foil. This will cause a more extensive growth of the interfacial reaction layer in the parallel orientation as was actually observed. Tables 3 and 4 show EDS elemental compositions for SATHX/Ti/Cu/Ti/SA-THX parallel and perpendicular joints, respectively. In the parallel joint (Fig. 5), a dark gray phase is distributed throughout the joint region (point markers 11 through 15). This phase has an average composition that is consistent with the carbides such as TiC (point marker 13, Fig. 5) and TiC0.7 (point markers 11, 12, 14, and 15). Unlike the Mo interlayer in Ti/Mo joints that had remained untransformed and intact at the center, the Cu interlayer in the Ti/Cu joints has undergone near-complete transformation. Additionally, the titanium carbide phase(s) are distributed throughout the volume of the diffusion bond due presumably to plastic deformation of the interlayer together with the melting of any untransformed residual copper (M.P.: 1086 1C) during diffusion bonding. The liquid phase formed upon melting would permit rapid diffusion and reaction, and also facilitate fluidflow assisted rearrangement of precipitating phases throughout the joint volume under the applied pressure. The perpendicular joints (? SA-THX/Ti/Cu/Ti/ ? SATHX) shown in Fig. 6 exhibit features similar to parallel joints such as absence of microcracks and a homogeneous distribution throughout the joint volume of a dark-gray platelike reaction phase. The mechanical keying effect and interfacial roughening from the reaction layer formation in perpendicular joints are less extensive than in parallel joints (insets for locations 1, 2 and 3, Figs. 5 and 6). The EDS analysis of a perpendicular joint at locations marked in Fig. 6 is presented in Table 4. The point markers 3, 5 and 6 on the SA-THX side show Si and C as expected. Point 4 at the interface region is a
2146
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149 Location 1
Location 2
Location 3
Fig. 5. SEM images of a ‖ SA-THX/Ti/Cu/Ti/‖ SA-THX joint showing overall views of the joint from three different locations and a BSE image of location 2. The EDS elemental compositions at point markers shown above are presented in Table 3.
Location 1
Location 2 Shrinkage cavity
Location 3
Fig. 6. SEM images of a ?SA-THX/Ti/Cu/Ti/? SA-THX joint showing overall views of the joint from three different locations and a higher-magnification BSE image of location 2. The EDS elemental compositions at point markers shown above are presented in Table 4.
Si (39 at%), C (53 at%) and Ti (8 at%) phase. The point markers 7, 8 and 12 in the dark gray phase have an average composition range of 56–59 at% Ti and about 39–42 at% C, suggesting a possible carbide compound, probably the substoichiometric TiC0.7. The point markers 9, 10 and 11, that are close to the interface with SA-THX, show a small amount of Si (4–6 at%) besides larger quantities of Ti (51–54 at%) and C (41–43 at%). The light phase (point markers 13, 14 and 15) has Cu (31–39 at%), C (32–43 at%) and Ti (9–14 at%) besides Si (12–16 at%). A similar elemental distribution (Table 5)
is observed at another location of the same perpendicular joint shown in Fig. 7. The dark gray phase is composed of titanium and carbon, possibly a titanium carbide phase. Besides the likely formation of a titanium carbide phase, the Cu and Ti interlayers juxtaposed together in the joint can form Cu–Ti intermetallic compounds such as Ti2Cu, TiCu, Ti3Cu4, TiCu4, TiCu2, and Ti2Cu3 [34]. Some of the compounds have low melting points (e.g., Ti2Cu and TiCu melt at 1015 and 984 1C, respectively) whereas some others become unsta ble over different temperature ranges. Thus during diffusion
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
2147
Table 3 Elemental composition in ‖ SA-THX/Ti/Cu/Ti/‖ SA-THX joint of Fig. 5.
Table 5 Elemental composition in ? SA-THX/Ti/Cu/Ti/? SA-THX joint of Fig. 7.
Marker in Fig. 5
C, atom %
Si, atom %
Ti, atom %
Cu, atom %
Marker in Fig. 7
C, atom %
Si, atom %
Ti, atom %
Cu, atom %
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
53.37 50.953 67.75 63.823 39.192 28.171 16.058 25.885 32.112 28.515 40.039 30.22 42.344 38.826 33.006 51.122 56.824
46.419 48.907 32.063 35.899 18.638 20.835 22.273 21.321 18.514 20.967 1.462 19.128 3.182 3.15 19.446 48.555 43.012
0.107 0.067 0.037 0.115 2.294 16.144 15.043 5.056 48.689 49.904 57.482 48.095 49.68 55.203 46.143 0.061 0.06
0.104 0.073 0.15 0.163 39.876 34.85 46.626 47.738 0.684 0.614 1.017 2.557 4.794 2.82 1.405 0.262 0.103
1 2 3 4 5 6 7 8 9 10
56.323 56.421 54.044 41.765 42.882 41.116 42.446 34.688 30.457 33.839
43.417 40.583 45.857 6.202 2.177 1.276 2.536 15.671 16.846 13.573
0.171 2.897 0.03 51.187 54.563 56.795 52.644 5.091 5.148 11.261
0.09 0.099 0.068 0.846 0.379 0.812 2.374 44.549 47.549 41.327
Table 4 Elemental composition in ? SA-THX/Ti/Cu/Ti/? SA-THX joint of Fig. 6. Marker in Fig. 6
C, atom %
Si, atom %
Ti, atom %
Cu, atom %
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
59.437 56.686 55.181 52.579 63 58.897 41.837 39.177 41.113 42.045 43.111 39.223 38.957 32.245 42.938
40.369 42.944 43.736 39.362 35.183 39.314 1.754 1.144 5.761 3.533 5.93 1.314 13.068 16.497 12.431
0.084 0.37 0.976 8.009 1.775 1.714 55.466 58.895 52.566 53.783 50.514 58.387 8.446 12.339 13.839
0.11 0 0.107 0.049 0.041 0.075 0.942 0.784 0.56 0.639 0.446 1.076 39.529 38.918 30.793
bonding at 1200 1C, and during heat up and cool down, multiple reactions to form Cu–Ti intermetallics can occur in the bonded region besides the reaction with SiC. These reactions may also be accompanied by phase transformations in the Cu–Ti system, such as peritectic reactions that form Ti3Cu4, TiCu2, and TiCu4 at 918, 878, and 870 1C, respectively, and eutectic transformations at 1003 1C, 960 1C and 860 1C [34]. What emerges from this discussion is the fairly complex nature of the reactions and transformations that accompany diffusion bonding of SiC using Cu/Ti bimetallic interlayers. Such reactions and transformations not only modulate elemental distributions but also morphology, volume of precipitating phases, residual stress, and joint integrity. It should be noted that kinetics of solid-state diffusion in SiC joints under consideration are generally fast enough at or above 1200 1C to result in full conversion of thin metal inserts into more stable carbide and silicide phases. In the presence of a liquid
Fig. 7. Backscattered electron SEM image from location 1 of Fig. 6 of a? SATHX/Ti/Cu/Ti/ ?SA-THX joint. Elemental compositions at point markers shown above are presented in Table 5.
Table 6 Knoop hardness of diffusion bonded joints. Average HK (‖ joint) Average HK (? joint) SA-THX/Ti/Mo/Ti/SA717.77 273.6 THX SA-THX/Ti/Cu/Ti/SA-THX 816.57 43.9 SA-THX (un-bonded) 12447 176
758.97299.3 — 624 7205
phase (e.g., melted Cu interlayer), the diffusion and reaction kinetics could be substantially accelerated and yield a greater quantity of the reaction products. However, shrinkage cavities develop upon solidification of the liquid film (e.g., Fig. 6) and their presence in joints is evidence of melting of copper during bonding. Volumetric changes that result from solidification and from diffusive conversion of the metal foil into carbide or silicide phases also affect the stress state and joint integrity. 3.2. Microhardness The Knoop hardness of both parallel and perpendicular joints as well as un-bonded SA-THX substrates was characterized. The results are summarized in Table 6. All hardness values are
2148
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
averages of 15 to 20 measurements. The average Knoop values on un-bonded SA-THX with parallel and perpendicular fiber orientations were 12447176 HK and 6247205 HK, respectively. For parallel fibers, the tip of the indenter pushed into the ends of fiber bundles, which merely pushed the fibers into the sample. The elasticity of the material, however, led to springback with the fibers mostly bouncing back to their original shape upon removal of load. This gave a smaller indent and higher hardness values. The perpendicular fibers were indented in a different manner, by the tip of the indenter pushing fibers apart, where they mostly stayed, giving larger indents and smaller hardness values. The average hardness values of joint region in SA-THX/Ti/ Mo/Ti/SA-THX joints in parallel and perpendicular fiber orientations were 717.7 7 273.6 and 758.9 7 299.3, respectively. There were, however, substantial differences between the hardness of the reaction layer adjacent to the SiC substrate and the hardness in joint center. The reaction layer was over twice as hard as the joint center where the Mo interlayer had remained largely untransformed (e.g., Figs. 2 and 3). Fig. 8(a) shows typical indentation marks at these two locations in a perpendicular SA-THX/Ti/Mo/Ti/SA-THX joint. The average hardness values within the reaction layer adjacent to the SiC were 977.57 79.4 and 1040.57 80.9 for parallel and perpendicular joints, respectively whereas the average hardness values in the center of the joints for parallel and perpendicular joints were only 457.8 7 40.3 and 477.3 7 16.4, respectively (interference fringes developed around the indents, Fig. 8b, due to high refractive index of the SA-Tyranno fibers).
The micro-hardness of the parallel SA-THX/Ti/Cu/Ti/SATHX joints was independent of the location of the indent within the joint (Fig. 8c and d). This is consistent with the observation that the SA-THX/Ti/Cu/Ti/SA-THX joint microstructures were uniform throughout joint thickness unlike the SA-THX/Ti/Mo/ Ti/SA-THX joints where distinct regions had formed within the seam. The average hardness of the parallel SA-THX/Ti/Cu/Ti/ SA-THX joints (816.5743.91) was greater than the hardness of the parallel SA-THX/Ti/Mo/Ti/SA-THX joints (717.77273.6 and 758.97299.3). Due to small joint thickness, reliable measurements of hardness could not be made within perpendicular SA-THX/Ti/Cu/Ti/SA-THX joints in spite of multiple attempts. However, microstructural similarities between parallel and perpendicular SA-THX/Ti/Cu/Ti/SA-THX joints suggest that the hardness values would also be similar. 4. Conclusions A SiC fiber-bonded ceramic (SA-Tyrannohex, SA-THX) was diffusion bonded using either Ti/Mo layers or Ti/Cu layers. The influence of metallic interlayers and SiC fiber orientation in SATHX with respect to the interlayers on joint microstructure, elemental composition, and microhardness in diffusion bonded joints was investigated using Optical Microscopy (OM), Field Emission Scanning Electron Microscopy (FE-SEM), Energy Dispersive Spectroscopy (EDS), and Knoop test. Well-bonded joints formed in all samples; however, joints with Ti/Mo bilayers revealed microcracks within the diffusion bond in both parallel and perpendicular orientations. The diffusion bonds
Fig. 8. Photomicrographs showing indentation marks and stress fringes in joints of (a) a ?SA-THX/Ti/Mo/Ti/? SA-THX joint, (b) a ‖ SA-THX/Ti/Mo/Ti//‖ SATHX joint, and (c) and (d) a ‖ SA-THX/Ti/Cu/Ti/‖ SA-THX joint.
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
with Ti/Cu bilayers yielded higher quality joints devoid of cracks. Redistribution of Si, Ti and C within joints was revealed by EDS suggesting possible formation of carbides and silicides during diffusion bonding. The reaction products distributed more homogeneously in Ti/Cu bilayer bonds than in Ti/Mo bilayer bonds and this behavior was consistent with the hardness distribution within the joints. The reaction layers in parallel and perpendicular SA-THX/Ti/Mo/Ti/SA-THX joints were twice as hard as the core region comprised of untransformed Mo interlayer whereas in SA-THX/Ti/Cu/Ti/SA-THX joints, hardness distribution was uniform across the joint thickness. References [1] J. Lamon, Chemical vapor infiltrated SiC/SiC composites (CVI SiC/SiC), in: N.P. Bansal (Ed.), Handbook of Ceramic Composites, Kluwer, 2005, pp. 55–76. [2] M.C. Halbig, M. Jaskowiak, J.D. Kiser, D. Zhu, Evaluation of ceramic matrix composite technology for aircraft turbine engine applications, in: 51st AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition, pp 1–11, 2013. [3] G.S. Corman, K.L. Luthra, Silicon melt infiltrated ceramic composites (HiPerComp), in: N.P. Bansal (Ed.), Handbook of Ceramic Composites, Kluwer, 2005, pp. 99–115. [4] M. Singh, A reaction forming method for joining of silicon carbide-based ceramics, Scr. Mater. 37 (8) (1997) 1151–1154. [5] M.C. Halbig, B.P. Coddington, R. Asthana, M. Singh, Characterization of silicon carbide joints fabricated using SiC particulate-reinforced Ag–Cu–Ti alloys, Ceram. Int. 39 (2013) 4151–4162. [6] M.G. Nicholas, Joining Processes: Introduction to Brazing and Diffusion Bonding, Kluwer Academic Publishers, Dodrecht, 1998. [7] M. Naka, J.C. Feng, Trans. Mater. Res. Soc. Jpn 16B (1994) 1143–1146. [8] M. Naka, J.C. Feng, J.C. Schuster, Phase reaction and diffusion path of the SiC/Ti system, Metall. Mater. Trans. A 28 (1997) 1385–1390. [9] B. Gottselig, E. Gyarmati, A. Naoumidis, H. Nickel, Joining of ceramics demonstrated by the example of SiC/Ti, J. Eur. Ceram. Soc. 6 (1990) 153–160. [10] S. Morozumi, M. Endo, M. Kikuchi, K. Hamajim, Bonding mechanism between silicon carbide and thin foils of reactive metals, J. Mater. Sci. 20 (1985) 3976–3982. [11] A.E. Martinelli, R.A.L. Drew, Microstructural development during diffusion bonding of α SiC to molybdenum, Mater. Sci. Eng., A 19 (1995) 239–247. [12] T. Cheng, J.O. Kiggans Jr, Y. Katoh, L.L. Snead, Process development and optimization for SiC joining and irradiation studies (Dec 31), Fusion Reactor Mater. Program 53 (2012) 42–54 (DOE/ER-0313/53). [13] A. Joshi, H.S. Hu, L. Jesion, J.J. Stephens, J. Wadsworth, Hightemperature interactions of refractory metal matrices with selected ceramic reinforcements, Metall. Trans. A 21 (1990) 2829–2837. [14] B.V. Cockeram, The Diffusion Bonding of Silicon Carbide and Boron Carbide Using Refractory Metals (Report B-T-3255, USDOE Contract No. DE-ACI 1-98 PN38206), Bettis Atomic Power Laboratory, Mifflin, PA, USA, 1999. [15] K. Bhanumurthy, R. Schmidt-Felzer, Solid-state bonding of SiC [HIPSiC] below 1000 C, Mater. Sci. Eng., A 220 (1996) 35–40.
2149
[16] J. Li, P. Xiao, Fabrication and characterization of SiC/superalloy interfaces, J. Eur. Ceram. Soc. 24 (2004) 2149–2156. [17] M. Singh, M.C. Halbig, Bonding and integration of silicon carbide based materials for multifunctional applications, Key Eng. Mater 352 (2007) 201–206. [18] M.C. Halbig, M. Singh, Development and characterization of the bonding and integration technologies needed for fabricating silicon carbide-based injector components, Ceram. Eng. Sci. Proc. 29 (9) (2009) 1–14. [19] M.C. Halbig, M. Singh, Diffusion bonding of silicon carbide for the fabrication of complex shaped ceramic components, in: M. Singh, T. Ohji, R. Asthana, S. Mathur (Eds.), Ceramic Integration and Joining Technologies: From Macro- to Nanoscale, John Wiley & Sons, 2011, pp. 143–162. [20] M.C. Halbig, M. Singh, H. Tsuda, Integration technologies for silicon carbide-based ceramics for micro-electro-mechanical systems-lean direct injector fuel injector applications, Int. J. Appl. Ceram. Technol. 9 (2012) 677–687. [21] H. Tsuda, S. Mori, M.C. Halbig, M. Singh, TEM observation of the Ti interlayer between SiC substrates during diffusion bonding, Ceram. Eng. Sci. Proc. 33 (8) (2012) 81–89. [22] H. Tsuda, S. Mori, M.C. Halbig, M. Singh, Interfacial characterization of diffusion-bonded monolithic and fiber-bonded silicon carbide ceramics, Ceram. Eng. Sci. Proc. 34 (8) (2013) 133–141. [23] R. Asthana, M. Singh, H-T. Lin, T. Matsunaga, T. Ishikawa, Joining of SiC fiber-bonded ceramics using silver, copper, nickel, palladium, and silicon-based alloy interlayers, Int. J. Appl. Ceram. Technol. 10 (5) (2013) 801–813. [24] T. Matsunaga, H.T. Lin, T. Ishikawa, S. Kajii, R. Asthana, M. Singh, Microstructure and mechanical properties of joints in sintered SiC fiberbonded ceramics, Key Eng. Mater 484 (2011) 9–14. [25] M. Singh, T. Matsunaga, H.-T Lin, R. Asthana, T. Ishikawa, Microstructure and mechanical properties of joints in sintered SiC fiber-bonded ceramics brazed with Ag–Cu–Ti alloy, Mater. Sci. Eng., A 557 (2012) 69–76. [26] T. Ishikawa, S. Kajii, K. Matsunaga, T. Hogami, Y. Kohtoku, T. Nagasawa, A tough, thermally conductive silicon carbide composite with high strength up to 1600 1C in air, Science 282 (1998) 1265–1297. [27] T. Ishikawa, Y. Kohtoku, K. Kumagawa, T. Yamamura, T. Nagasawa, High-strength alkali-resistant sintered SiC fibre stable to 2200 1C, Nature 391 (1998) 773–775. [28] Y.S. Toulouki, R.K. Kirby, R.E. Taylor, T.Y.R. Lee, Thermrophysical Properties of Matter, vol. 13, IFI-Plenum, New York, 1977, p. 778–898. [29] L. Zhang, J. Wu, Thermal expansion and elastic moduli of the silicide based intermetallic alloys Ti5Si3(X) and Nb5Si3, Scr. Mater. 38 (2) (1998) 307–313. [30] J.H. Schneibel, C.J. Rawn, Thermal expansion anisotropy of ternary silicides based on Ti5Si3, Acta Mater. 52 (2004) 3843–3848. [31] F. Chu, D.J. Thoma, K. McClellan, P. Peralta, Y. He, Intermetallics 7 (1996) 611–620. [32] J.-W. Park, P.F. Mendez, T.W. Eager, Strain energy distribution in ceramic-to-metal joints, Acta Mater. 50 (5) (2002) 883–899. [33] Z. Zhong, T. Hinoki, H. Yang, Y. Park, A. Kohyama, Bimetallic interlayers of W/Ni to diffusion bond SiC to steel, Mater. Des 31 (2010) 1070–1076. [34] V.N. Eremenko, Yu. I. Buyanov, S.B. Prima, Phase diagram of the system titanium-copper, Sov. Powder Metall. Met. Ceram. 5 (6) (1996) 494–502.
CERAMICS INTERNATIONAL
Ceramics International 41 (2015) 2140–2149 www.elsevier.com/locate/ceramint
Diffusion bonding of SiC fiber-bonded ceramics using Ti/Mo and Ti/Cu interlayers M.C. Halbiga, R. Asthanab,n, M. Singhc a NASA Glenn Research Center, Cleveland, OH, USA University of Wisconsin-Stout, 326 Fryklund Hall, Menomonie, WI 54751, USA c Ohio Aerospace Institute, Cleveland, OH, USA
b
Received 4 August 2014; received in revised form 1 October 2014; accepted 1 October 2014 Available online 13 October 2014
Abstract A SiC fiber-bonded ceramic (SA-Tyrannohex™) was diffusion bonded using Ti/Mo and Ti/Cu interlayers. The influence of metallic interlayers and SiC fiber orientation in the ceramic substrate with respect to the interlayers on joint microstructure, elemental composition, and microhardness in diffusion bonds was investigated using Optical Microscopy (OM), Field Emission Scanning Electron Microscopy (FE-SEM), Energy Dispersive Spectroscopy (EDS), and Knoop microhardness test. Compared to the Ti/Mo bilayers, the Ti/Cu bilayers yielded higher quality joints. The reaction products distributed more homogeneously across the joint thickness in Ti/Cu bonds than in Ti/Mo bonds. The reaction layers adjacent to the SiC substrate in both parallel and perpendicular SA-THX/Mo/Ti/SA-THX joints were twice as hard as the joint center where the Mo interlayer had remained untransformed during diffusion bonding. In SA-THX/Cu/Ti joints, hardness distribution was uniform across the joint thickness consistent with a more homogeneous reaction phase distribution across the joint. & 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
Keywords: B. Electron microscopy; Diffusion bonding; Silicon carbide; Knoop hardness; Reaction layers
1. Introduction Silicon carbide (SiC) based materials have excellent hightemperature mechanical properties, oxidation and heat resistance, and thermo-chemical stability. These materials are leading candidates for various thermostructural applications at high temperatures in harsh environments in aerospace and energy sectors [1–3]. A variety of advanced SiC-based ceramics such as CVD SiC, sintered SiC, hot-pressed SiC, and SiC fiber-bonded ceramic (SA-Tyrannohex™) as well as ceramic composites such as SiC/SiC and C/SiC have been developed for such applications. Implementing these advanced ceramics and composites in real components demands robust ceramic joining and integration technologies. The most commonly used ceramic joining methods include reaction bonding [4] and brazing [5,6]. Diffusion bonding techniques have been successfully utilized to bond n
Corresponding author. Tel.: þ1 715 232 2152; fax: þ 1 715 232 1330. E-mail address: [email protected] (R. Asthana).
http://dx.doi.org/10.1016/j.ceramint.2014.10.014 0272-8842/& 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
a wide variety of SiC ceramics. Sintered, hot-pressed and CVD silicon carbide ceramics have been diffusion bonded using refractory metal interlayers of titanium [7–10], molybdenum [11,12], tantalum [13], tungsten [14], niobium [13], zirconium [10], nickel [15] and Inconel 600 [16]. The diffusional transformation of a metal interlayer at high temperatures and under high mechanical pressures into carbides, silicides, or complex ternary and higher order compounds produces strong joints. Carbides and silicide compounds of refractory metals are also thermodynamically more stable than SiC. As a result, diffusive conversion of metal insert into carbides and silicides provides a pathway for strong bond formation. Previously, some of the present authors [17–19] investigated the diffusion bonding of SiC/Ti/SiC joints and identified the phases that form in the bonded area using scanning electron microscopy (SEM), X-ray diffraction (XRD) analysis, and energy dispersive spectroscopy (EDS). Recently, the authors conducted a detailed transmission electron microscopy (TEM) study for the structural evaluation of the phases formed during
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
diffusion bonding with Ti as the interlayer [20,21]. All prior studies investigated diffusion bonding of sintered, hot pressed, and CVD SiC ceramics, and apart from the present authors’ recent research [22] on diffusion bonded SA-Tyrannohex™ (also referred to as SA-THX) using Ti interlayers, no published reports currently exist on diffusion bonding of SiC fiberbonded ceramic, SA-THX. The liquid-phase joining of SATHX via brazing was also only recently demonstrated [23–25]. SA-THX is a dense and tough ceramic consisting of a highly ordered, closed-packed structure of fine hexagonal columnar fibers composed of crystalline β-SiC with a thin (10–20 nm) interfacial layer of turbostatic carbon between the fibers [26,27]. SA-THX is a product of Ube Industries, Japan, and is obtained by converting an amorphous Si–Al–C–O fiber into Tyranno-SA fiber with near-stoichiometric composition. The final SiC fiber-bonded ceramic, SA-THX, is made by hotpressing of fiber weave lay-ups into various configurations. The resulting material exhibits good thermomechanical performance, high thermal conductivity (33 W/m K at 1600 1C), high strength up to 1600 1C, high flexural strength ( 300 MPa), high fracture toughness (1200 J/m2 at room temperature), and good machinability and low cost. In this study, SA-THX was diffusion bonded to itself using either Ti/Mo or Ti/Cu bilayers to examine the effect of the interlayers and SiC fiber orientation (perpendicular and parallel relative to the joining plane) on the microstructure, composition and hardness of joints using Optical Microscopy (OM), Field Emission Scanning Electron Microscopy (FE-SEM), Energy Dispersive Spectroscopy (EDS), and the Knoop microhardness test. 2. Experimental SA-Tyrannohex™ (SA-THX) SiC fiber-bonded ceramic was obtained from Ube Industries (Ube, Japan). The SA-THX is made from SA-Tyranno fiber bundles woven into an eightharness satin weave that results in fibers being oriented in parallel (‖) and perpendicular ( ? ) directions. Joints were made using like SA-THX composite pairs (‖-to-‖, or ? -to- ? ). Ti foil (10 μm), Mo foil (12.7 μm), and Cu foil (5 μm) were obtained from Goodfellow Corporation (Glen Burnie, MD, USA). Before joining, all materials were ultrasonically cleaned in acetone for 10 min. Joints were diffusion bonded at 1200 1C with a pressure of 30 MPa. Joint processing was conducted in vacuum for 4 h at the peak temperature under load, followed by cooling at a rate of 2 1C per min. Joints with Ti/Cu bilayers had three Cu interlayers (total thickness: 15 μm) with 10 μm thick Ti foil on two sides. Similarly, for Ti/Mo bilayers, a 12.7 μm Mo foil was sandwiched by 10 μm thick Ti foil on two sides. The nomenclature used to identify the joints for subsequent discussion is as follows. A SA-THX joint with Ti and Mo interlayers having ? fibers coincident on the foil is denoted as follows: ? SA-THX/Ti/Mo/Ti/ ? SA-THX. Similarly, a SA-THX joint with ‖ fibers coincident on the foil is denoted as: ‖ SA-THX/Ti/Mo/Ti/‖ SA-THX. Optical Microscopy (OM) was done on an Olympus BX51, and SEM observations and elemental analyses were performed
2141
with a Hitachi S-4700-I Field Emission Scanning Electron Microscope (FE-SEM) coupled with an Energy Dispersive Spectrometer (EDS). Joint hardness was characterized using Knoop indenter on a Struers Duramin-A300 machine under a load of 200 g and loading time of 10 s. 3. Results and discussion 3.1. Joint microstructure and composition 3.1.1. SA-THX/Ti/Mo/Ti SA-THX joints In designing the SA-THX joints using Ti/Mo bilayers, dual objectives were sought to be accomplished. One objective was to take advantage of a low coefficient of thermal expansion (CTE) of Mo interlayer (CTE: 5.1 10 6 K 1) to reduce residual stresses in diffusion bonded joints. A small CTE mismatch mitigates the residual stress buildup from joining. Experience has shown that judiciously chosen interlayers positioned in the joint prior to bonding can reduce the strain energy and cracking propensity in ceramic joints. A second objective was to explore the feasibility of lowering the bonding temperature relative to direct bonding using only Mo foils. Past research [11,14] has shown that SiC/Mo joints diffusion bonded under proper conditions have excellent bond quality; in fact, the quality of the SiC/Mo/SiC bond was significantly better than the diffusion bonds produced with Ti. Cockeram [14] observed that high temperatures (1500 1C) and long contact times (10 h) produced sound SiC/Mo joints under 3.4 to 18 MPa pressures with 12.7 mm thick Mo foils. Similar joints processed at 1200 1C led to poor bond quality. In Ref. [14], it was also noted that use of thicker molybdenum foil (25.4 mm) resulted in a significantly higher density of cracks in the bond region in a manner similar to the diffusion bonding of SiC using Ti where use of thin rather than thick foils improved the quality of the interfacial bond [14]. Thus although Mo (CTE: 5.1 10 6 K 1) and SiC are better CTE matched than are Ti (CTE: 8.6 10 6 K 1) and SiC, thin Ti (10 mm) and Mo (12.7 mm) metallic foils were both used together in the present work with the goal to achieve sound diffusion bonds at relatively low (1200 1C) operating temperatures. Titanium and molybdenum form an isomorphous binary alloy system and Ti and Mo are completely miscible above 882 1C. Below 882 1C, the solubility of Mo in Ti is restricted (e.g., the maximum solubility of Mo in α-Ti is less than 0.5% at 600 1C [11]). It was envisioned that formation of a Ti(Mo) solid solution in conjunction with the carbides and silicides from the reactions between Ti and SiC and between Mo and SiC shall reduce the residual stresses in SiC joints while still forming strong bonds and lowering the energy required for diffusion bonding as a result of reduced process temperatures (from 1500 1C to 1200 1C). A 12.7 μm Mo foil was sandwiched by 10 μm Ti foils on two opposite sides in pairs of parallel and perpendicular substrates. Optical images of parallel (‖ SA-THX/Ti/Mo/Ti/‖ SA-THX) and perpendicular ( ? SA-THX/Ti/Mo/Ti/ ? SATHX) joints are shown in Fig. 1a and b, respectively. The ends
2142
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
Fig. 1. Optical photomicrographs of (a) a parallel and (b) a perpendicular SA-THX/Ti/Mo/Ti/SA-THX joint showing bonded regions near left edge, center, and right edge of flat-back bonded SA-THX samples.
Location 1
Location 2
Fig. 2. SEM images of a ‖ SA-THX/Ti/Mo/Ti//‖ SA-THX joint showing overall views of the joint from two different locations and a BSE image of location 2. The EDS elemental compositions at point markers shown above are presented in Table 1.
of the joints show some un-bonded regions due to one of the three tape layers being shorter. No large-scale cracks appear in the ceramic; however, toward the center of the joint, evidence of micro-cracking is noted. SEM images (Figs. 2 and 3) of the joints reveal microcracks, 8–10 μm long, in the vicinity of bonded interfaces within the interlayers in both parallel and perpendicular orientations. These microcracks originate at the second layer from the interface and terminate at the Mo foil that resides in the center of the joint. Some of the micro-cracks observed metallographically could be sample preparation artifacts but similar preparation for Ti/Cu bilayer joints did not produce cracks as discussed later. Redistribution of Si, C and Ti is evident from the EDS analysis of the joints presented in Tables 1 and 2 for parallel and
perpendicular joints, respectively. There is a 6–8 μm thick continuous dark gray band decorating the sides in contact with SiC in both parallel and perpendicular joints (Figs. 2 and 3). For the parallel joint, this band has an average elemental composition that is consistent with the presence of Ti3SiC2 (point markers 5 and 6, Fig. 2) and also possibly Ti5Si3 (point 7). There also seems to be a possibility of TiC formation (54 at% Ti and 44 at% C at point markers 10, 11 and 20, Fig. 2). There appears to be a carbon concentration gradient over 5–7 μm near the interface with C concentration decreasing from 33.7 at% (point markers 5 and 6, Fig. 2) to 12.2 at% (point markers 8 and 9). At points 8 and 9, EDS suggests the composition to be Ti5Si3Cx. The ternary compound Ti5Si3Cx is an intermediate phase that is transformed into Ti3SiC2 following completion of the reaction [8] during
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
2143
Location 1
Region A
Location 2 Layer II Layer I
Fig. 3. BSE SEM images of a? SA-THX/Ti/Mo/Ti/? SA-THX joint showing overall views of the joint from two different locations and a BSE image of location 2. The EDS elemental compositions at point markers shown above are presented in Table 2.
Table 1 Elemental composition in ‖ SA-THX/Ti/Mo/Ti/‖ SA-THX Joint of Fig. 2.
Table 2 Elemental composition in ? SA-THX/Ti/Mo/Ti/? SA-THX joint of Fig. 3.
Marker in Fig. 2
C, atom % Si, atom % Ti, atom % Mo, atom %
Marker in Fig. 3
C, atom %
Si, atom % Ti, atom % Mo, atom %
1 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
52.777 56.752 53.05 30.641 36.807 26.667 12.043 12.297 2.331 42.585 14.056 20.238 20.411 6.252 5.571 16.09 14.021 37.641 43.565 10.384
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
44.865 46.334 46.627 45.175 22.458 31.118 33.991 47.613 38.051 19.101 10.365 20.118 10.334 40.43 31.904 41.148 34.359 13.178 15.371 15.555 14.245 13.321 14.631 16.573 17.609 23.299 17.26 14.23
54.275 52.493 52.301 53.913 18.904 14.216 0.940 0.870 0.188 26.977 29.075 26.596 29.876 2.711 0.153 0 0.696 33.638 35.363 35.119 35.933 0 1.152 0.215 0.165 0.34 0.532 0.131
47.223 43.248 46.95 18.381 17.914 19.528 36.823 35.729 42.036 2.105 0.376 29.58 28.909 0.353 0.455 28.511 30.807 0.345 3.961 35.268
0 0 0 50.979 45.279 53.805 51.134 51.973 55.633 55.31 0.481 21.307 17.625 2.45 0.575 21.646 21.492 61.755 51.938 52.918
0 0 0 0 0 0 0 0 0 0 85.087 28.876 33.055 90.945 93.399 33.752 33.68 0.26 0.536 1.43
diffusion bonding, thus leaving Ti3SiC2 as the primary phase. The Ti3SiC2 phase has a large but stable CTE of 10 10 6 K 1 up to about 800 1C [14,28]. The intermediate phase Ti5Si3Cx has an anisotropic coefficient of thermal expansion (CTE) [29,30] that exacerbates the propensity for micro-cracking. It is evident that both C and Si have dissolved and diffused in the Ti layer but not in the Mo layer in the center (point markers 12, 15 and 16, Fig. 2) where the Mo concentration exceeds 90 at%. The interface between Mo and Ti is diffuse and delineates a band containing lenticular precipitates. These precipitates (point marker 19, Fig. 2) have a composition that is consistent with sub-stoichiometric carbide, TiC1.6. The perpendicular ( ? SA-THX/Ti/Mo/Ti/ ? SA-THX) joints (Fig. 3) also show transverse microcracks in the bonded region that are arrested in the diffuse layer between Ti and Mo.
0.744 1.076 1.072 0.913 58.601 54.264 65.069 51.508 61.762 21.976 28.891 22.965 25.981 56.707 67.761 58.741 64.688 52.797 48.637 48.188 49.711 20.457 24.71 22.445 20.003 2.836 0.609 0.691
0.116 0.096 0 0 0.037 0.402 0 0.009 0 31.946 31.669 30.321 33.809 0.153 0.182 0.112 0.258 0.387 0.629 1.138 0.11 66.222 59.507 60.767 62.223 73.524 81.599 84.948
Cracks originate in the second layer (marked on the figure) from the interface, which, according to the EDS data (Table 2), appears to be the Ti5Si3Cx phase. In a manner similar to parallel joints, a 6–8 μm thick continuous dark gray band has developed on either side of the joint in Fig. 3. The composition of the interface region is consistent with the formation of the following possible phases: Ti3SiCx (point markers 5 and 6, Fig. 3), TiC (point marker 8), and sub-stoichiometric
2144
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
carbides TiCx (x: 0.5–0.6, point markers 7 and 9). Interestingly, one side (point markers 5 and 6 in Fig. 3) has Si (14– 19 at%) but the opposite side (point markers 7, 8 and 9) does not have Si; there are only Ti and C on the Si-depleted side even though the interaction zone thickness is the same on the two sides. This difference could arise from the geometrical configuration during diffusion bonding as explained in Refs. [19,20]. In perpendicular orientation, the Ti foil could contact either the basal plane of the carbon layer in the hexagonal columns or the prismatic planes. In the plane of Fig. 3, the top side contacts the basal plane that limits C diffusion into Ti foil. On the bottom side of the joint in Fig. 3, a large fraction of prismatic planes contacts Ti thus facilitating diffusion of C. As a result of increased carbon content in the region, the relative percentage of Si drops. This is consistent with the formation of titanium carbide rather than titanium–silicon–carbide phase. In Figs. 2 and 3, the interfacial reaction products appear to have coalesced, aided by the high pressures and high temperatures of diffusion bonding, to yield a compact and homogeneous layer of the reaction product. No grain structure in the reaction product layer could be identified at the magnifications used. In Ref. [22], we had reported that in SA-THX/Ti/SA-THX diffusion bonds, many small grains had formed with the major and minor dimensions of 2–4 μm and 1–2 μm, respectively [22]. Figs. 2 and 3 do not show evidence of grains although confirmatory tests using the TEM shall be needed to validate this. Further out from the interface (point markers 19 through 21, Fig. 3), the elemental compositions again suggest a compound of the type Ti5Si3Cx which could indicate formation of TiSi and TiC phases. Further away from the interface (point markers 10 through 13, Fig. 3), the atomic percentages indicate the following ratios: Ti3Si3CMo3 (point markers 11 and 13, Fig. 3), and TiSiCMo2 (point markers 10 and 12). The Ti, Si and Mo contents are nearly identical at point markers 10–13; however, point markers 10 and 12 have twice the carbon than point markers 11 and 13. The first atomic ratio (Ti3Si3CMo3) is consistent with the formation of either Ti3SiC þ Mo3Si (cubic) or TiSi þ TiC þ Mo3Si. Likewise, the second atomic ratio (TiSiCMo2) could indicate the formation of TiSi þ Mo2C phases (Mo2C is known to form in SiC/Mo/ SiC diffusion bonds [11,12,14]). The dark gray plate-like precipitates (point markers 14 through 17, Fig. 3) are all likely to be carbides of Ti; their average composition suggests sub-stoichiometric compounds, TiC0.5 (point markers 15 and 17) and TiC0.7 (point markers 14 and 16). Unlike parallel joints, where the joint center was nearly pure Mo, the region (marked as region ‘A’, Fig. 3) between the gray interfacial band and the center in perpendicular joints (point markers 22 through 25) shows likely formation of β-Mo(Ti) solid solution with 20–25 at% Ti, and this is consistent with the binary Ti–Mo diagram at 1200 1C [29]. Right in the center of the joint (point markers 26–28, Fig. 3) is nearly pure untransformed molybdenum foil (with evidence of some diffusiondriven carbon, about 14–23 at%). Past research [11,12,14] on SiC/Mo/SiC diffusion bonds (T: 1300–1700 1C) has shown that diffusion of Si and C into Mo
forms distinct layers of Mo5Si3 þ carbon and molybdenum carbide (Mo2C). The formation of these phases shall be detrimental to the diffusion bond because of highly anisotropic expansion properties of Mo2C and Mo5Si3. The thermal expansion of Mo5Si3 (tetragonal structure) is strongly anisotropic and the CTE along ‘a’ and ‘c’ axes are αa=5.2 10 6 K 1 and αc=11.5 10 6 K 1, respectively [31] (these CTE values are essentially temperature independent over a wide range). Similarly, the CTE of Mo2C is anisotropic with values of 4.9 10 6 K 1 along the a-axis and 8.2 10 6 K 1 along the c-axis [11]. Conceivably, the strongly anisotropic expansion of Mo5Si3 and Mo2C that possibly form in the diffusion bond is responsible for the microcracking observed in joints with Ti/Mo metallic interlayers. It must be mentioned that other phases such as Mo5Si3C and MoC could also form in diffusion bonded joints; however, Mo5Si3C forms in SiC/Mo joints only above 1400 1C and MoC forms above 1700 1C [11]. These deleterious phases are, therefore, unlikely to form in our joints. It is noteworthy that the reaction layer, as it formed in the Ti/Mo bilayer joints during diffusion bonding, did not disintegrate under mechanical pressure of 30 MPa during diffusion bonding even though the joints developed micro-cracks (Figs. 2 and 3) to relieve the stress. The application of large compressive stresses (30 MPa) during diffusion bonding on softer Ti foil (yield strength: 170– 480 MPa, depending upon the grade) that resided between harder SA-THX substrate and a hard Mo foil (yield strength: 415–550 MPa) could also have caused micro-cracking especially when the formation of brittle intermetallic reaction products in joints fabricated under pressure is considered. In summary, even though use of Ti in conjunction with Mo lowered the diffusion bonding temperature from the acceptable value of 1500 1C to 1200 1C and yielded sound metallurgical bonding with SA-THX, the bimetallic interlayers failed to yield crack-free joints. It is conceivable that judicious selection of interlayer thickness ratios could limit the volume of deleterious intermetallic compounds that form thus yielding sound joints even at the low fabrication temperature of 1200 1C. Optimization of the interlayer thickness ratio shall constitute a focus of our continuing research effort.
3.1.2. SA-THX/Ti/Cu/Ti/SA-THX joints In designing Ti/Cu bilayer joints, the goal was to accommodate residual stresses via plastic deformation of copper. Although Cu has a very large CTE (16.6 10 6 K 1) and a large CTE mismatch with Ti and SiC, its low yield strength (70 MPa) and high ductility (45% elongation for pure annealed Cu) are advantageous to managing residual stresses in bonded joints. It is known that a hard/soft interlayer combination enables plastic deformation of the second ductile interlayer between the first hard interlayer [32]. It is also known that multiple interlayers reduce the strain energy in the ceramic more than a single interlayer of the same total thickness. Bimetallic interlayers have been used to diffusion bond SiC/metal couples such as SiC/steel using W/Ni interlayers [33]. In the present work, three Cu foils, each with a thickness of 5 μm were
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
2145
Fig. 4. Optical photomicrographs of (a) a parallel and (b) a perpendicular SA-THX/Ti/Cu/Ti/SA-THX joint showing bonded regions near left edge, center, and right edge of flat-back bonded SA-THX samples.
sandwiched by 10 μm thick Ti foils on the two opposite sides for diffusion bonding. Optical photomicrographs of parallel (‖ SA-THX/Ti/Cu/Ti/‖ SA-THX) and perpendicular (? SA-THX/Ti/Cu/Ti/ ? SATHX) joints are shown in Fig. 4a and b, respectively. In parallel joints, the outer edges of the bonded SA-THX substrates showed some un-bonded regions and microcracking (Fig. 4a). No such cracks were evident at the outer edges of perpendicular SA-THX joints (Fig. 4b). The center of both parallel and perpendicular joints was crack-free and exhibited excellent bonding. SEM images (Figs. 5 and 6) of the middle portions of the bonded SA-THX substrates for both orientations revealed crackfree diffusion bonds together with a high volume fraction of a secondary phase throughout the joint volume. This is in contrast to the Ti/Mo bilayer joints where distinct continuous bands of reaction layers (6–8 μm thick) had formed at the interface and the reaction layers had remained confined to near-interface regions rather than extending to the center of the joint. The Ti/Cu bilayer joints reveal more extensive interfacial roughening in parallel joints (e.g., SEM images for locations 1 and 2, Fig. 5). The reaction phase has grown along inter-fiber boundaries because of increased availability of carbon for the reaction to occur in these regions. The mechanical anchor points appear to be more near the top of the joint (locations 1 and 2, Fig. 5) than the bottom. However, a polished section oriented perpendicular to the plane of the figure revealed similar incidence of anchor points near the bottom half. Such anchor points potentially could contribute to enhanced bonding via mechanical keying. Compared to parallel joints, the interfacial roughening is modest in perpendicular joints (Fig. 6) and seems to be related to the reduced reaction intensity in the perpendicular orientation as also discussed previously. Silicon and carbon atoms can
migrate readily when the SA-THX fibers are set parallel to the Ti layer, because the prismatic plane of the carbon layer in the hexagonal columns commonly does not face the Ti foil. This will cause a more extensive growth of the interfacial reaction layer in the parallel orientation as was actually observed. Tables 3 and 4 show EDS elemental compositions for SATHX/Ti/Cu/Ti/SA-THX parallel and perpendicular joints, respectively. In the parallel joint (Fig. 5), a dark gray phase is distributed throughout the joint region (point markers 11 through 15). This phase has an average composition that is consistent with the carbides such as TiC (point marker 13, Fig. 5) and TiC0.7 (point markers 11, 12, 14, and 15). Unlike the Mo interlayer in Ti/Mo joints that had remained untransformed and intact at the center, the Cu interlayer in the Ti/Cu joints has undergone near-complete transformation. Additionally, the titanium carbide phase(s) are distributed throughout the volume of the diffusion bond due presumably to plastic deformation of the interlayer together with the melting of any untransformed residual copper (M.P.: 1086 1C) during diffusion bonding. The liquid phase formed upon melting would permit rapid diffusion and reaction, and also facilitate fluidflow assisted rearrangement of precipitating phases throughout the joint volume under the applied pressure. The perpendicular joints (? SA-THX/Ti/Cu/Ti/ ? SATHX) shown in Fig. 6 exhibit features similar to parallel joints such as absence of microcracks and a homogeneous distribution throughout the joint volume of a dark-gray platelike reaction phase. The mechanical keying effect and interfacial roughening from the reaction layer formation in perpendicular joints are less extensive than in parallel joints (insets for locations 1, 2 and 3, Figs. 5 and 6). The EDS analysis of a perpendicular joint at locations marked in Fig. 6 is presented in Table 4. The point markers 3, 5 and 6 on the SA-THX side show Si and C as expected. Point 4 at the interface region is a
2146
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149 Location 1
Location 2
Location 3
Fig. 5. SEM images of a ‖ SA-THX/Ti/Cu/Ti/‖ SA-THX joint showing overall views of the joint from three different locations and a BSE image of location 2. The EDS elemental compositions at point markers shown above are presented in Table 3.
Location 1
Location 2 Shrinkage cavity
Location 3
Fig. 6. SEM images of a ?SA-THX/Ti/Cu/Ti/? SA-THX joint showing overall views of the joint from three different locations and a higher-magnification BSE image of location 2. The EDS elemental compositions at point markers shown above are presented in Table 4.
Si (39 at%), C (53 at%) and Ti (8 at%) phase. The point markers 7, 8 and 12 in the dark gray phase have an average composition range of 56–59 at% Ti and about 39–42 at% C, suggesting a possible carbide compound, probably the substoichiometric TiC0.7. The point markers 9, 10 and 11, that are close to the interface with SA-THX, show a small amount of Si (4–6 at%) besides larger quantities of Ti (51–54 at%) and C (41–43 at%). The light phase (point markers 13, 14 and 15) has Cu (31–39 at%), C (32–43 at%) and Ti (9–14 at%) besides Si (12–16 at%). A similar elemental distribution (Table 5)
is observed at another location of the same perpendicular joint shown in Fig. 7. The dark gray phase is composed of titanium and carbon, possibly a titanium carbide phase. Besides the likely formation of a titanium carbide phase, the Cu and Ti interlayers juxtaposed together in the joint can form Cu–Ti intermetallic compounds such as Ti2Cu, TiCu, Ti3Cu4, TiCu4, TiCu2, and Ti2Cu3 [34]. Some of the compounds have low melting points (e.g., Ti2Cu and TiCu melt at 1015 and 984 1C, respectively) whereas some others become unsta ble over different temperature ranges. Thus during diffusion
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
2147
Table 3 Elemental composition in ‖ SA-THX/Ti/Cu/Ti/‖ SA-THX joint of Fig. 5.
Table 5 Elemental composition in ? SA-THX/Ti/Cu/Ti/? SA-THX joint of Fig. 7.
Marker in Fig. 5
C, atom %
Si, atom %
Ti, atom %
Cu, atom %
Marker in Fig. 7
C, atom %
Si, atom %
Ti, atom %
Cu, atom %
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
53.37 50.953 67.75 63.823 39.192 28.171 16.058 25.885 32.112 28.515 40.039 30.22 42.344 38.826 33.006 51.122 56.824
46.419 48.907 32.063 35.899 18.638 20.835 22.273 21.321 18.514 20.967 1.462 19.128 3.182 3.15 19.446 48.555 43.012
0.107 0.067 0.037 0.115 2.294 16.144 15.043 5.056 48.689 49.904 57.482 48.095 49.68 55.203 46.143 0.061 0.06
0.104 0.073 0.15 0.163 39.876 34.85 46.626 47.738 0.684 0.614 1.017 2.557 4.794 2.82 1.405 0.262 0.103
1 2 3 4 5 6 7 8 9 10
56.323 56.421 54.044 41.765 42.882 41.116 42.446 34.688 30.457 33.839
43.417 40.583 45.857 6.202 2.177 1.276 2.536 15.671 16.846 13.573
0.171 2.897 0.03 51.187 54.563 56.795 52.644 5.091 5.148 11.261
0.09 0.099 0.068 0.846 0.379 0.812 2.374 44.549 47.549 41.327
Table 4 Elemental composition in ? SA-THX/Ti/Cu/Ti/? SA-THX joint of Fig. 6. Marker in Fig. 6
C, atom %
Si, atom %
Ti, atom %
Cu, atom %
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
59.437 56.686 55.181 52.579 63 58.897 41.837 39.177 41.113 42.045 43.111 39.223 38.957 32.245 42.938
40.369 42.944 43.736 39.362 35.183 39.314 1.754 1.144 5.761 3.533 5.93 1.314 13.068 16.497 12.431
0.084 0.37 0.976 8.009 1.775 1.714 55.466 58.895 52.566 53.783 50.514 58.387 8.446 12.339 13.839
0.11 0 0.107 0.049 0.041 0.075 0.942 0.784 0.56 0.639 0.446 1.076 39.529 38.918 30.793
bonding at 1200 1C, and during heat up and cool down, multiple reactions to form Cu–Ti intermetallics can occur in the bonded region besides the reaction with SiC. These reactions may also be accompanied by phase transformations in the Cu–Ti system, such as peritectic reactions that form Ti3Cu4, TiCu2, and TiCu4 at 918, 878, and 870 1C, respectively, and eutectic transformations at 1003 1C, 960 1C and 860 1C [34]. What emerges from this discussion is the fairly complex nature of the reactions and transformations that accompany diffusion bonding of SiC using Cu/Ti bimetallic interlayers. Such reactions and transformations not only modulate elemental distributions but also morphology, volume of precipitating phases, residual stress, and joint integrity. It should be noted that kinetics of solid-state diffusion in SiC joints under consideration are generally fast enough at or above 1200 1C to result in full conversion of thin metal inserts into more stable carbide and silicide phases. In the presence of a liquid
Fig. 7. Backscattered electron SEM image from location 1 of Fig. 6 of a? SATHX/Ti/Cu/Ti/ ?SA-THX joint. Elemental compositions at point markers shown above are presented in Table 5.
Table 6 Knoop hardness of diffusion bonded joints. Average HK (‖ joint) Average HK (? joint) SA-THX/Ti/Mo/Ti/SA717.77 273.6 THX SA-THX/Ti/Cu/Ti/SA-THX 816.57 43.9 SA-THX (un-bonded) 12447 176
758.97299.3 — 624 7205
phase (e.g., melted Cu interlayer), the diffusion and reaction kinetics could be substantially accelerated and yield a greater quantity of the reaction products. However, shrinkage cavities develop upon solidification of the liquid film (e.g., Fig. 6) and their presence in joints is evidence of melting of copper during bonding. Volumetric changes that result from solidification and from diffusive conversion of the metal foil into carbide or silicide phases also affect the stress state and joint integrity. 3.2. Microhardness The Knoop hardness of both parallel and perpendicular joints as well as un-bonded SA-THX substrates was characterized. The results are summarized in Table 6. All hardness values are
2148
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
averages of 15 to 20 measurements. The average Knoop values on un-bonded SA-THX with parallel and perpendicular fiber orientations were 12447176 HK and 6247205 HK, respectively. For parallel fibers, the tip of the indenter pushed into the ends of fiber bundles, which merely pushed the fibers into the sample. The elasticity of the material, however, led to springback with the fibers mostly bouncing back to their original shape upon removal of load. This gave a smaller indent and higher hardness values. The perpendicular fibers were indented in a different manner, by the tip of the indenter pushing fibers apart, where they mostly stayed, giving larger indents and smaller hardness values. The average hardness values of joint region in SA-THX/Ti/ Mo/Ti/SA-THX joints in parallel and perpendicular fiber orientations were 717.7 7 273.6 and 758.9 7 299.3, respectively. There were, however, substantial differences between the hardness of the reaction layer adjacent to the SiC substrate and the hardness in joint center. The reaction layer was over twice as hard as the joint center where the Mo interlayer had remained largely untransformed (e.g., Figs. 2 and 3). Fig. 8(a) shows typical indentation marks at these two locations in a perpendicular SA-THX/Ti/Mo/Ti/SA-THX joint. The average hardness values within the reaction layer adjacent to the SiC were 977.57 79.4 and 1040.57 80.9 for parallel and perpendicular joints, respectively whereas the average hardness values in the center of the joints for parallel and perpendicular joints were only 457.8 7 40.3 and 477.3 7 16.4, respectively (interference fringes developed around the indents, Fig. 8b, due to high refractive index of the SA-Tyranno fibers).
The micro-hardness of the parallel SA-THX/Ti/Cu/Ti/SATHX joints was independent of the location of the indent within the joint (Fig. 8c and d). This is consistent with the observation that the SA-THX/Ti/Cu/Ti/SA-THX joint microstructures were uniform throughout joint thickness unlike the SA-THX/Ti/Mo/ Ti/SA-THX joints where distinct regions had formed within the seam. The average hardness of the parallel SA-THX/Ti/Cu/Ti/ SA-THX joints (816.5743.91) was greater than the hardness of the parallel SA-THX/Ti/Mo/Ti/SA-THX joints (717.77273.6 and 758.97299.3). Due to small joint thickness, reliable measurements of hardness could not be made within perpendicular SA-THX/Ti/Cu/Ti/SA-THX joints in spite of multiple attempts. However, microstructural similarities between parallel and perpendicular SA-THX/Ti/Cu/Ti/SA-THX joints suggest that the hardness values would also be similar. 4. Conclusions A SiC fiber-bonded ceramic (SA-Tyrannohex, SA-THX) was diffusion bonded using either Ti/Mo layers or Ti/Cu layers. The influence of metallic interlayers and SiC fiber orientation in SATHX with respect to the interlayers on joint microstructure, elemental composition, and microhardness in diffusion bonded joints was investigated using Optical Microscopy (OM), Field Emission Scanning Electron Microscopy (FE-SEM), Energy Dispersive Spectroscopy (EDS), and Knoop test. Well-bonded joints formed in all samples; however, joints with Ti/Mo bilayers revealed microcracks within the diffusion bond in both parallel and perpendicular orientations. The diffusion bonds
Fig. 8. Photomicrographs showing indentation marks and stress fringes in joints of (a) a ?SA-THX/Ti/Mo/Ti/? SA-THX joint, (b) a ‖ SA-THX/Ti/Mo/Ti//‖ SATHX joint, and (c) and (d) a ‖ SA-THX/Ti/Cu/Ti/‖ SA-THX joint.
M.C. Halbig et al. / Ceramics International 41 (2015) 2140–2149
with Ti/Cu bilayers yielded higher quality joints devoid of cracks. Redistribution of Si, Ti and C within joints was revealed by EDS suggesting possible formation of carbides and silicides during diffusion bonding. The reaction products distributed more homogeneously in Ti/Cu bilayer bonds than in Ti/Mo bilayer bonds and this behavior was consistent with the hardness distribution within the joints. The reaction layers in parallel and perpendicular SA-THX/Ti/Mo/Ti/SA-THX joints were twice as hard as the core region comprised of untransformed Mo interlayer whereas in SA-THX/Ti/Cu/Ti/SA-THX joints, hardness distribution was uniform across the joint thickness. References [1] J. Lamon, Chemical vapor infiltrated SiC/SiC composites (CVI SiC/SiC), in: N.P. Bansal (Ed.), Handbook of Ceramic Composites, Kluwer, 2005, pp. 55–76. [2] M.C. Halbig, M. Jaskowiak, J.D. Kiser, D. Zhu, Evaluation of ceramic matrix composite technology for aircraft turbine engine applications, in: 51st AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition, pp 1–11, 2013. [3] G.S. Corman, K.L. Luthra, Silicon melt infiltrated ceramic composites (HiPerComp), in: N.P. Bansal (Ed.), Handbook of Ceramic Composites, Kluwer, 2005, pp. 99–115. [4] M. Singh, A reaction forming method for joining of silicon carbide-based ceramics, Scr. Mater. 37 (8) (1997) 1151–1154. [5] M.C. Halbig, B.P. Coddington, R. Asthana, M. Singh, Characterization of silicon carbide joints fabricated using SiC particulate-reinforced Ag–Cu–Ti alloys, Ceram. Int. 39 (2013) 4151–4162. [6] M.G. Nicholas, Joining Processes: Introduction to Brazing and Diffusion Bonding, Kluwer Academic Publishers, Dodrecht, 1998. [7] M. Naka, J.C. Feng, Trans. Mater. Res. Soc. Jpn 16B (1994) 1143–1146. [8] M. Naka, J.C. Feng, J.C. Schuster, Phase reaction and diffusion path of the SiC/Ti system, Metall. Mater. Trans. A 28 (1997) 1385–1390. [9] B. Gottselig, E. Gyarmati, A. Naoumidis, H. Nickel, Joining of ceramics demonstrated by the example of SiC/Ti, J. Eur. Ceram. Soc. 6 (1990) 153–160. [10] S. Morozumi, M. Endo, M. Kikuchi, K. Hamajim, Bonding mechanism between silicon carbide and thin foils of reactive metals, J. Mater. Sci. 20 (1985) 3976–3982. [11] A.E. Martinelli, R.A.L. Drew, Microstructural development during diffusion bonding of α SiC to molybdenum, Mater. Sci. Eng., A 19 (1995) 239–247. [12] T. Cheng, J.O. Kiggans Jr, Y. Katoh, L.L. Snead, Process development and optimization for SiC joining and irradiation studies (Dec 31), Fusion Reactor Mater. Program 53 (2012) 42–54 (DOE/ER-0313/53). [13] A. Joshi, H.S. Hu, L. Jesion, J.J. Stephens, J. Wadsworth, Hightemperature interactions of refractory metal matrices with selected ceramic reinforcements, Metall. Trans. A 21 (1990) 2829–2837. [14] B.V. Cockeram, The Diffusion Bonding of Silicon Carbide and Boron Carbide Using Refractory Metals (Report B-T-3255, USDOE Contract No. DE-ACI 1-98 PN38206), Bettis Atomic Power Laboratory, Mifflin, PA, USA, 1999. [15] K. Bhanumurthy, R. Schmidt-Felzer, Solid-state bonding of SiC [HIPSiC] below 1000 C, Mater. Sci. Eng., A 220 (1996) 35–40.
2149
[16] J. Li, P. Xiao, Fabrication and characterization of SiC/superalloy interfaces, J. Eur. Ceram. Soc. 24 (2004) 2149–2156. [17] M. Singh, M.C. Halbig, Bonding and integration of silicon carbide based materials for multifunctional applications, Key Eng. Mater 352 (2007) 201–206. [18] M.C. Halbig, M. Singh, Development and characterization of the bonding and integration technologies needed for fabricating silicon carbide-based injector components, Ceram. Eng. Sci. Proc. 29 (9) (2009) 1–14. [19] M.C. Halbig, M. Singh, Diffusion bonding of silicon carbide for the fabrication of complex shaped ceramic components, in: M. Singh, T. Ohji, R. Asthana, S. Mathur (Eds.), Ceramic Integration and Joining Technologies: From Macro- to Nanoscale, John Wiley & Sons, 2011, pp. 143–162. [20] M.C. Halbig, M. Singh, H. Tsuda, Integration technologies for silicon carbide-based ceramics for micro-electro-mechanical systems-lean direct injector fuel injector applications, Int. J. Appl. Ceram. Technol. 9 (2012) 677–687. [21] H. Tsuda, S. Mori, M.C. Halbig, M. Singh, TEM observation of the Ti interlayer between SiC substrates during diffusion bonding, Ceram. Eng. Sci. Proc. 33 (8) (2012) 81–89. [22] H. Tsuda, S. Mori, M.C. Halbig, M. Singh, Interfacial characterization of diffusion-bonded monolithic and fiber-bonded silicon carbide ceramics, Ceram. Eng. Sci. Proc. 34 (8) (2013) 133–141. [23] R. Asthana, M. Singh, H-T. Lin, T. Matsunaga, T. Ishikawa, Joining of SiC fiber-bonded ceramics using silver, copper, nickel, palladium, and silicon-based alloy interlayers, Int. J. Appl. Ceram. Technol. 10 (5) (2013) 801–813. [24] T. Matsunaga, H.T. Lin, T. Ishikawa, S. Kajii, R. Asthana, M. Singh, Microstructure and mechanical properties of joints in sintered SiC fiberbonded ceramics, Key Eng. Mater 484 (2011) 9–14. [25] M. Singh, T. Matsunaga, H.-T Lin, R. Asthana, T. Ishikawa, Microstructure and mechanical properties of joints in sintered SiC fiber-bonded ceramics brazed with Ag–Cu–Ti alloy, Mater. Sci. Eng., A 557 (2012) 69–76. [26] T. Ishikawa, S. Kajii, K. Matsunaga, T. Hogami, Y. Kohtoku, T. Nagasawa, A tough, thermally conductive silicon carbide composite with high strength up to 1600 1C in air, Science 282 (1998) 1265–1297. [27] T. Ishikawa, Y. Kohtoku, K. Kumagawa, T. Yamamura, T. Nagasawa, High-strength alkali-resistant sintered SiC fibre stable to 2200 1C, Nature 391 (1998) 773–775. [28] Y.S. Toulouki, R.K. Kirby, R.E. Taylor, T.Y.R. Lee, Thermrophysical Properties of Matter, vol. 13, IFI-Plenum, New York, 1977, p. 778–898. [29] L. Zhang, J. Wu, Thermal expansion and elastic moduli of the silicide based intermetallic alloys Ti5Si3(X) and Nb5Si3, Scr. Mater. 38 (2) (1998) 307–313. [30] J.H. Schneibel, C.J. Rawn, Thermal expansion anisotropy of ternary silicides based on Ti5Si3, Acta Mater. 52 (2004) 3843–3848. [31] F. Chu, D.J. Thoma, K. McClellan, P. Peralta, Y. He, Intermetallics 7 (1996) 611–620. [32] J.-W. Park, P.F. Mendez, T.W. Eager, Strain energy distribution in ceramic-to-metal joints, Acta Mater. 50 (5) (2002) 883–899. [33] Z. Zhong, T. Hinoki, H. Yang, Y. Park, A. Kohyama, Bimetallic interlayers of W/Ni to diffusion bond SiC to steel, Mater. Des 31 (2010) 1070–1076. [34] V.N. Eremenko, Yu. I. Buyanov, S.B. Prima, Phase diagram of the system titanium-copper, Sov. Powder Metall. Met. Ceram. 5 (6) (1996) 494–502.