Microstructures and mechanical properties of Ti₃Al/Ni-based superalloy joints brazed with AuNi filler metal

1. Introduction

Intermetallic compounds have attracted a great deal of attention because they have low densities and maintain high strength at elevated temperatures [1,2]. In particular, Ti₃Al-based alloy is a potential structural material with the long-term service temperature up to 923-973 K for aerospace applications due to its attractive properties such as relatively low density, high specific strength, excellent creep resistance and good oxidation stability at high temperature [3,4]. Obviously, in order to realize its practical engineering applications, joining technology of Ti₃Al alloy is extremely important. Joining of Ti₃Al to other materials such as Ni-based superalloys is of great interest to achieve the optimal material combination design.

In the past decades, several welding and joining techniques were reported on the joining of Ti₃Al alloys to themselves, including brazing [5], diffusion bonding [6], laser beam welding [7] and linear friction welding [8]. With regard to dissimilar joining, studies mainly focused on the joining of Ti₃Al to Ti-based alloys or TiAl intermetallics. For example, Shiue et al. [9] carried out the dissimilar brazing of α₂-Ti₃Al and Ti-6Al-4 V alloys using Ti-15Cu-25Ni and Ti-15Cu-15Ni (wt%) filler metals. It is found that the existence of the Ti₂Ni intermetallic compound is detrimental to the bonding strength of the joint. The shear strength of the brazed joint free of the blocky Ti₂Ni phase is comparable with that of the α₂-Ti₃Al substrate. Ren et al. [10] designed TiZrCuNi(Co) filler metal for the dissimilar joining between Ti₃Al alloy and TiAl intermetallics. Compared with traditional filler metals, this new filler improved the Ti₃Al/TiAl joint properties and the joint shear strength reached 278 MPa. Li et al. [11] investigated the microstructure and mechanical properties of Ti-22Al-25Nb/TA15 dissimilar joint fabricated by dual-beam laser welding.

Actually, the dissimilar joining of Ti₃Al alloy to Ni-based superalloy is more attractive for engineering applications because of its high-temperature service potential as well as the remarkable weight reduction effect. As a typical application example in aerospace field, there is a need to join nickel-based superalloy honeycomb to light-weight Ti₃Al alloy substrate, and in this case the brazing method should be almost unique solution. Although Chen et al. [12] and Cai et al. [13] studied the fusion welding between the two materials with arc welding technology, the research work on their brazing is still highly lacking.

Furthermore, the dissimilar joining of Ti₃Al-based alloy to Ni-based superalloy is extremely difficult due to the following reasons. First, the dissolution enthalpy of Ti in liquid Ni solvent is up to -170 kJ/mol [14], indicating the strong reaction tendency between Ti and Ni. On the other hand, Ti and Ni could react with each other and solidify in the form of Ni-Ti intermetallic compounds such as Ti₂Ni, TiNi and TiNi₃ [15]. These brittle phases would greatly decrease the joint mechanical properties and even make it sensitive to cracking during the welding process. Besides, the thermal expansion coefficient of Ni-based superalloy ($\widetilde{1}$3.0 × 10^-6 K^-1) is much higher than that of the Ti₃Al alloy ($\widetilde{9}$.1 × 10^-6 K^-1) [16]. This difference could influence the heat transfer during the welding and induce large residual thermal stress, which would easily lead to micro-cracks in the joint. Third, chemical composition would vary over a wide range across the dissimilar joint. Hence, it is difficult to obtain a single liquidus isotherm defining the solid-liquid interface as it does in similar joining of metals. Therefore, the joining of Ti₃Al alloy and Ni-based superalloy is not only an extremely difficult technology problem but also a key scientific issue. It is of great importance to depress the formation of brittle phases and to ensure good metallurgical quality at the joining interface.

An attempt was also made by Chen et al. [17] to directly join Ti₃Al alloy to Ni-based superalloy using a Ti-Zr-Cu-Ni filler alloy. However, micro-cracks were visible within the dissimilar joint and the corresponding joint strength was only 86 MPa. And within this joint, continuous Fe-Ti and Ni-Ti compounds were visible and caused the micro-cracks during the cooling process.

In a previous study [18], the dissimilar joining of a Ti₃Al-based alloy to a Ni-based superalloy was studied using Ag-21Cu-25Pd (wt%) filler metal. Due to the strong affinity, element Pd mainly reacted with Ti₃Al substrate and formed intermetallic compounds of TiPd, Ti₃Pd₅ and AlPd. Element Ag was enriched in the joint central area as (Ag,Cu) solid solution. Even the joint tensile strength reached 404 MPa at room temperature, the strength at high temperature was deserved to be improved further. Besides, no other reports about the dissimilar brazing between Ti₃Al alloy and Ni-based superalloy were found.

Based on the previous research results that Au-based filler metals could wet TiAl system materials [19]. In this study Au-17.5Ni (wt%) filler metal was attempted for the joining of Ti₃Al-based alloy and Ni-based superalloy. The mechanism of the reactions between this Au-Ni filler and the parent materials during the brazing process was studied. Microstructure evolution across the dissimilar joint and the interface metallurgical behaviors as well as the mechanical properties of the joints were investigated. The diffusion behavior of elements during brazing process was discussed as well. The research results should be able to provide theoretical basis for the joining of γ-TiAl to Ni-based superalloy in future.

2. Experimental procedures

The Ti₃Al-based alloy with a nominal composition of Ti-13Al-28Nb-2Mo (wt%) was used in this experiment. Its backscattered electron (BSE) image is illustrated in Fig. 1(a). This alloy was composed of lath α₂-Ti₃Al, fine lamellar O-Ti₂AlNb and the β/B2 matrix. The lamellar O-Ti₂AlNb phase precipitated from the β/B2 matrix during cooling process [20]. This alloy was prepared by the following steps: vacuum-consumable electrode arc melting, breaking down in the β/B2 phase fields, forging and rolling in the α₂+β/B2 phase field, and heat treatment at 1253 K for 1 h followed by cooling in air. Another base material to be joined was GH536 Ni-based superalloy with a long-term service temperature of 1173 K. Its backscattered electron (BSE) image and chemical composition are shown in Fig. 1(b) and Table 1, respectively. The microstructure of GH536 consisted of face centered cubic (fcc) γ-matrix and precipitated M₆C carbide [21]. The Au-17.5Ni (wt%) alloy foil with a thickness of about 100 μm was used as the filler metal in the brazing experiment.

Cylindrical samples were machined from Ti₃Al and GH536 alloys as shown in Fig. 2(a). Then the AuNi filler foil was sandwiched between the two base metals to form a butt joint (Fig. 2(a)). Prior to brazing, the joined samples and the filler metal were ultrasonically cleaned in acetone. The brazing experiment was conducted at 1253 K for 5-20 min. During the brazing experiment the vacuum was kept between 7.0 × 10^-3 Pa and 2.0 × 10^-3 Pa and the heating rate was 10 K/min. After the brazing experiment the brazed joints were cooled down to 773 K with a low rate of 4.5 K/min followed by furnace cooling.

The microstructures of the brazed specimens were examined using a scanning electron microscope (SEM, JXA-8100) equipped with an electron probe micro-analyzer (EPMA). An X-ray diffraction (XRD) spectrometer with CuK radiation (D/max-RB) was carried out to identify the reaction phases within the joint. Furthermore, the morphological structure of the reaction phases was also examined using a transmission electron microscopy (TEM, JEM-2010). The preparation steps of TEM samples were as follows. A cylinder sample was cut from the as-brazed Ti₃Al/GH536 joint, and the brazed seam was located in the centre of the cylinder sample. Then slice samples were cut from the cylinder sample and ground to the thickness of 30 μm. Finally, the ground slice samples were further prepared by an ion thinning instrument (PIPS-691).

Micro-hardness test of the typical microzones across the Ti₃Al/GH536 joints was performed using a Vickers hardness tester (450-SVD) with a load of 50 g and load time of 15 s. At least three indentations were performed for each microzone. The tensile strengths for the brazed joints were measured at room-temperature and 923 K, respectively. Detailed geometric dimensioning of the tensile specimen is illustrated in Fig. 2(b) machined from the as-brazed joint in Fig. 2(a). The reported average strength was obtained from at least three joint specimens.

3. Results and discussion

3.1. Microstructures of Ti₃Al/GH536 joints

The liquidus temperature of the AuNi filler alloy was about 1223 K. Fig. 3 shows the backscattered electron images of Ti₃Al/GH536 joints brazed at 1253 K with this filler. Sound brazing seams were achieved and no micro-cracks or other defects were visible. As can be seen in Fig. 3, due to the great difference in chemical composition between the two joined materials, and the diffusion behavior during the brazing process, all of the brazed joints were characterized by multiple layer structures. Fixed at the brazing temperature of 1253 K, the thickness of the brazed seams increased with the increase of the dwell time. For example, under the dwell time of 5 min, the joint thickness was only 50 μm (Fig. 3(a)), but its thickness was increased to 80 μm for 15 min (Fig. 3(c)).

As presented in Fig. 3(e), a reaction layer consisting of black matrix (“1”) and gray laths (“2”) were formed adjacent to the Ti₃Al substrate. According to the EPMA analysis results (Table 2), though microzone “1” presented lower concentration of dissolved Ni, the phase constituents of both microzones “1” and “2” should be the mixture of O-Ti₂AlNb and NiTi phase. Similarly, for microzone “3”, it exhibited the same phase constituents but was dissolved with about 10.0 at.% Au. The XRD peaks associated with the phases of O-Ti₂AlNb and NiTi were confirmed by Fig. 4(a). Particularly, the selected area electron diffraction pattern corresponded to NiTi phase was also detected through TEM (Fig. 5(a) and (b)).

The EPMA area scanning results in Fig. 6(a) indicated that the quasi-continuous black phase in microzone “4” was enriched with element Ni, and according to the EPMA analysis data presented in Table 2 this phase was identified as ternary AlNi₂Ti compound dissolved with about 6.2 at.% Nb. It was concurrently detected through XRD analysis on the fractured surface at the joined GH536 side (Fig. 4(b)) and also confirmed by TEM photograph (Fig. 5(a) and (c)).

In the present study it was not strange to observe the AlNi₂Ti ternary compound at the joining interface. For example, in the Ti₃Al/Ti₃Al joint brazed with Ni-8Cr-5Si-2B-2Fe (wt%) filler foil [22], AlNi₂Ti compound was visible due to the reaction between the Ni-based filler metal and parent metal. Besides, when a Ni/Al nanolayer was used to join γ-TiAl alloy AlNi₂Ti compound was also detected [23]. For the Ni-8Cr-5Si-2B-2Fe (wt%) filler foil [22] or Ni/Al nanolayer [23], Ni was the main element, and the foil or layer was in contact with the Ti₃Al or γ-TiAl base metal. Hence, element Ni was easy to react with Ti and Al, leading to the formation of AlNi₂Ti compound. Furthermore, in the TiAl/GH99 joint brazed with Ti foil [24], the Ti foil could react with TiAl and GH99 alloys, element Al and Ni diffused into the joining interface. In the area adjacent to GH99 alloy with a high gradient of element Ni, AlNi₂Ti compound was detected.

However, when Ti-15Cu-15Ni and Ti-15Cu-25Ni (wt%) filler metals, with similar Ni content to the AuNi filler alloy, were used to join TiAl alloys, not ternary AlNi₂Ti compound but Ti₂Ni phase was detected within the TiAl/Ti₃Al [25] and TiAl/Ti-6Al-4 V [26] joints. During the brazing process, parent metals also reacted with the Ti-Cu-Ni filler metals. Nevertheless, Ti was the main element in both parent metals and filler metals, leading to the reaction zone was always dominated by Ti. As a result, Ti₂Ni phase rather than AlNi₂Ti compound was formed.

In the present experiment, on the one hand, element Ni from the AuNi filler metal could directly react with Ti and Al at the Ti₃Al/AuNi interface. On the other hand, the diffusion activity element of Au to Ti₃Al substrate was lower as illustrated in Fig. 4(b). Element Ni diffused to the Ti₃Al/AuNi interface and became enriched in this area (Fig. 6(a)). As shown in Fig. 3(e), NiTi phase and AlNi₂Ti compound were distributed alternately and both of them were characterized by a hairbrush-like morphology. It was deduced that they were formed through the following reaction:

α₂-Ti₃Al + 4Ni = AlNi₂Ti+ 2NiTi (1)

Concerning microzone “5” (Fig. 3(f)) higher Au concentration was detected as shown in Fig. 4(b) whereas for element Al its concentration was at rather low level (Fig. 6(c) and Table 2), indicating that its composition was a little close to the original AuNi filler alloy (Au-41.6Ni at.%), but the dissolved Ti content in it was rather high, about 26.394 at.%. That is, due to the strong inter-diffusion between the filler alloy and the Ti₃Al parent material during the brazing process, pure residual filler alloy was hardly retained within the brazed joint. In addition, selected area electron diffraction patterns associated with TiAu phase was detected through TEM (Fig. 5(d) and (e)). Hence, the phase composition in microzone “5” might be the mixture of TiAu and (Ni,Au) solid solution ((Ni,Au)ss).

The compositions of the isolated particle “6” were similar to those of the continuous reaction layer “7” (Fig. 3(f)), in which the area distribution of three elements Ni, Au and Nb were in good agreement with each other (Fig. 6(a), (b) and (d)). In general its composition was greatly deviated from the original AuNi filler alloy. In particular, the concentration of element Au was decreased to only 15.085 at.% (Table 2) whereas that for Ni increased to 58.504 at.%. Besides the large variation in Au and Ni concentration, microzone “6” was also dissolved with 16.082 at.% Nb which evidently came from the Ti₃Al alloy substrate, signifying that within this area the diffusion behavior had become more complicated. However, as can be seen in Fig. 6(e) little element Ti was detected in microzones “6”and “7”.

Considering that Ni₃Nb phase was detected at the both fractured surfaces (Fig. 4) and also confirmed by TEM (Fig. 5(d) and (f)), it is reasonable to deduce that the phase in microzone “6” is Ni₃Nb(Au) and that in microzone “7” should be (Ni,Au)ss dissolved with Nb. It is known that the dissolution enthalpy of Nb in liquid Ni solvent is up to -143 kJ/mol [14], indicating a strong reactive tendency between them. In a previous study [27], when Nb and Ni foils was used during the diffusion bonding between Ti₂AlNb and GH4169 superalloy, the inter-diffusion and reaction between Nb and Ni foils occurred and caused Ni₃Nb reaction product within the Ti₂AlNb/Nb/Ni/GH4169 joint. According to Ni-Nb binary alloy diagram [28], three phases between them could be formed: Ni₈Nb, Ni₃Nb and Ni₆Nb₇. Among of them Ni₈Nb was an unstable phase and it would be decomposed into Ni and Ni₃Nb at 808 K. Concerning the phase Ni₆Nb₇, it was easy to form when the atom proportion between Nb and Ni was close or above 1:1. However, in microzone “6”, the atom proportion of Nb and Ni was only about 2:7. For this atom proportion, the driving force towards Ni₃Nb formation was larger than that of Ni₈Nb and Ni₆Nb₇ [29]. Thus, only Ni₃Nb was detectable in this experiment.

In the case of microzones “8” and “9”, they exhibited a characteristic of eutectic microstructure. Their main elements were Ni and Au and they should be identified as (Ni,Au) solid solution dissolved with Ti, Fe and Cr (Table 2). Similarly, the phase in microzone “10” was also (Ni,Au)ss dissolved with Ti, Fe and Cr as presented in Table 2. Finally, the black phase in microzone “11” should be Ni-rich phase dissolved with Ti and Al.

In general, under the brazing temperature of 1253 K, with the variation of dwell time the change of the fundamental microstructure characteristic of the joints did not happen. The interfacial microstructure within the Ti₃Al/GH536 joint brazed with AuNi filler metal from Ti₃Al side to GH536 side can be described as the follows: Ti₃Al/O-Ti₂AlNb + NiTi/AlNi₂Ti/TiAu + (Ni, Au)ss/Ni₃Nb(Au)/(Ni,Au)ss dissolved with Nb/(Ni, Au)ss dissolved with Cr(Al), Fe and Ti/Ni-rich phase dissolved with Ti and Al/GH536.

3.2. Microstructure evolution mechanism of joints

Undoubtedly, the joint microstructure will dominate the joint mechanical performance. It is a significant scientific issue to reveal the formation mechanism of interfacial microstructure and reaction behaviors between different elements during brazing process. Based on the EPMA analyzed results in Table 2 and the element area distribution maps of Ni and Au in Fig. 6, the diffusion and distribution characteristics about the two elements could be described as follows. The diffusion effects of elements Au and Ni to the jointed Ti₃Al substrate were different. 12-20 at.% Ni was detected in microzones “1”-“3”, whereas the amount of detected Au was only 2.2-10.9 at.%. And Ni was even the predominant element in microzone “4”. This signified that the diffusion driving force of Ni to Ti₃Al substrate was much larger than Au.

On the basis of Ti-Ni and Ti-Au binary diagrams [28], the solubility of Ni in β-Ti is about 8 at.% at 1273 K, but that of Au is about 5 at.%, evidently lower than the former. On the other hand, diffusion coefficient (D₀) and activation energy (Q) of Ni in Ti at 1209-1511 K are 9.2 × 10^-7 m²/s and 123.9 kJ/mol, whereas those of Au in Ti at 1173-1823 K are 9.32 × 10^-4 m²/s and 348.0 kJ/mol [30], respectively. Hence, the diffusion velocities (D) of Ni and Au in Ti at 1253 K are 6.29 × 10^-12 m²/s and 2.89 × 10^-18 m²/s, respectively, calculated through Arrhenius equation:

where R is the universal gas constant and T is the temperature. Distinctly, atoms Ni can diffuse much faster than Au in Ti. As a result, Ni atoms rather than Au diffused intensively into Ti₃Al alloy, and NiTi and AlNi₂Ti compounds were formed adjacent to Ti₃Al parent metal.

According to Au-Ni binary diagram [28], elements Ni and Au were completely miscible. Therefore, during the brazing process, element Au diffused from the filler metal towards the GH536 substrate and (Ni,Au) solid solution was formed in microzones “8”-“10”.

In order to describe the brazing process intuitively, a simplified schematic diagram of the microstructure evolution was proposed based on the previous analysis results, as displayed in Fig. 7. When the temperature reached about 1223 K, the AuNi brazing filler began melting and then the base metals partially dissolved into the liquid brazing filler. Due to the concentration gradient, element Au diffused towards the two base alloys and element Ni mainly diffused to the Ti₃Al substrate (Fig. 7(a)). In the meantime the diffusion phenomenon of elements Ti, Nb, Ni, Fe and Cr from the base metals into the brazed seam also occurred.

Along with the element diffusion between AuNi brazing filler and base metals, metallurgical reaction happened, and some new phases and even reaction layers were formed as shown in Fig. 7(b). Owning to the diffusion of Ni towards Ti₃Al substrate, phases TiNi and AlNi₂Ti were detectable. However, at this stage the AlNi₂Ti phase was distributed only dispersedly (Fig. 7(b)). Besides, element Ti diffused into the brazed seam and reacted with Au, resulting in the formation of TiAu and (Ni,Au)ss (microzone “5” in Fig. 3(e)). Similarly, the reaction between element Nb and AuNi brazing filler led to the formation of (Ni,Au)ss and Ni₃Nb (microzone “7” in Fig. 3(f)). And next, an area characterized by (Ni, Au) eutectic microstructure appeared (microzones “8” and “9” in Fig. 3(f)). Finally, (Ni, Au)ss (microzone “10” in Fig. 3(f)) and Ni-rich phase (microzone “11” in Fig. 3(f)) were detected adjacent to GH536 substrate.

With the prolongation of dwell time, the reaction layer of O-Ti₂AlNb and TiNi was thickened, AlNi₂Ti phase grew up and they connected with each other gradually (Fig. 7(c)). Because of the further diffusion of Ni and Au, more Au atoms diffused to the interface between AuNi brazing filler and GH536. And in the meantime more Ni atoms from GH536 diffused to this area and element Ni reacted with Au leading to the formation and growth of (Ni,Au) eutectic microstructure. However, the area of (Ni,Au)ss and Ni₃Nb (microzone “7” in Fig. 3(f)) decreased, on the contrary, (Ni,Au)ss (microzone “10” in Fig. 3(f)) and Ni-rich phase (microzone “11” in Fig. 3(f)) were thickened (Fig. 7(c)).

3.3. Mechanical properties of Ti₃Al/GH536 joints

Fig. 8 presented the tensile strength of the Ti₃Al/GH536 joints brazed at 1253 K as a function of brazing time. The joints brazed for 5 min offered an average tensile strength of 372 MPa at room temperature. With the prolonging of brazing time, the joint strength exhibited an increase tendency. For the brazing time of 15 min the joint strength reached the maximum value of 434 MPa (Fig. 8). However, too long dwell time (20 min) conversely resulted in a slight decrease of the joint strength (362 MPa). In addition, for the joints brazed at 1253 K for 15 min, the tensile strength of 314 MPa was maintained at 923 K, about 72% of the strength value at room temperature. As mentioned previously, with the variation of dwell time the joint microstructure did not change fundamentally. Hence, the joint tensile strengths were maintained from 356 MPa to 434 MPa, implying a wide process window.

Cai et al. [13] reported the laser joining of Ti₃Al-based alloy to Ni-based superalloy using a titanium interlayer. A large amount of complex intermetallic compounds with a width of 70 μm was formed adjacent to the Ti₃Al base metal. The presence of AlNi₂Ti and Ti₂Ni brittle intermetallic compounds in the Ti₃Al/weld interface limited the joint properties. The average tensile strength of the laser welded joint was only 177 MPa, obviously lower than the present results. Table 3 summarized the types of formed brittle phases and thickness of the brittle regions in different Ti₃Al/Ni-based superalloy joints, together with the joint strengths. It can be seen that when arc welding using only one kind of alloy as filler material such as Ti-_51.5-54.5Ni-_7.3-9.3Nb [12] and Ni-_32-38Cu [31] alloys, the thickness values of the region containing brittle phases were about 150-200 μm. Both of the joints exhibited tensile strengths lower than 250 MPa at room temperature. When gradient fillers were used, the thickness of brittle phases decreased to 50-70 μm (Table 3) and thus the joint strength was improved to 353 MPa [32].

Concerning the brazing method, in the previous study Ag-21Cu-25Pd (wt%) filler metal was used to join Ti₃Al to Ni-based superalloy (GH536) [18], and the corresponding thickness of brittle phases in Ti₃Al/AgCuPd/GH536 joint was about 57 μm. In this work, that thickness in Ti₃Al/AuNi/GH536 joint was further decreased to 40 μm (Table 3). Therefore, it was not strange to notice that the tensile strength of Ti₃Al/AuNi/GH536 joint at room temperature was slightly higher than that of Ti₃Al/AgCuPd/GH536 joint. More importantly, the Ti₃Al/AuNi/GH536 joint strength at 923 K was increased evidently.

As the results in Table 3 illustrated, with the decrease of brittle region thickness, the dissimilar joint strengths correspondingly increased. Due to the thinner brittle region, Ti₃Al/AuNi/GH536 joint maintained its high-temperature advantages. Hence, the selection of filler metal and the reaction control between the filler metal and the base metals are extremely important to inhibit the formation of brittle phases.

The micro-hardness profile along the joint brazed at 1253 K for 15 min is displayed in Fig. 9. In general, the whole joint exhibited a higher micro-hardness than the two parent materials. From the right side of Fig. 9, the micro-hardness was gradually increased from 272 HV to 626 HV due to the formation of Ni₃Nb, Ni-rich phases or (Ni,Au)ss (corresponded to microzones “7”-“11” in Fig. 3(f)). Then it was slightly decreased to 560 HV, and this value should correspond to the TiAu + (Ni,Au)ss phases in microzone “5”. Finally, the micro-hardness reached the maximum value of 847 HV (microzones “1”-“4” in Fig. 3(e)) at the interface adjacent to the Ti₃Al substrate alloy. The phase constituents in microzones “1”-“4” included AlNi₂Ti and NiTi, in which the hardness of AlNi₂Ti was even high up to 14.1 GPa (1437.7 HV) [23].

Fig. 10(a) presents the fracture surface (Ti₃Al side) of the Ti₃Al/GH536 joint brazed with AuNi filler metal at 1253 K for 15 min. Two kinds of typical fracture characteristics were visible at the surface. Based on the EPMA analysis results (Table 4) the square “1” in Fig. 10(a) could be identified as the mixture of O-Ti₂AlNb and TiNi phases (microzones “1”-“3” in Fig. 3(e)) whereas the square “2” should be (Ni,Au)ss and Ni₃Nb phase (microzone “7” in Fig. 3(f)). Consequently, the phases of O-Ti₂AlNb, Ni₃Nb and NiTi were detected by XRD (Fig. 4(a)) on this fracture surface. Furthermore, Fig. 10(b) shows the cross section of the fracture surface. The microstructure (indicated by the arrow) similar to microzone “4” in Fig. 3(e) was observed on the cross section of GH536 side. Naturally, AlNi₂Ti phase was confirmed by XRD (Fig. 4(b)) on the fracture surface of the joined GH536 side. Once again, concerning the phase AlNi₂Ti, it was a hard intermetallic compound with a B2 structure. Evidently, the fracture of the joint subjected to the tensile test mainly occurred at the two high-hardness areas. Furthermore, for the total specimens fracture, the area percentage of the fracture surface relevant to AlNi₂Ti compound was about 66%.

4. Conclusions

(1)Sound Ti₃Al/GH536 joints were achieved at the brazing temperature of 1253 K for 5-20 min using Au-17.5Ni (wt%) filler metal. All of the brazed joints were characterized by multiple layer structures. The typical brazed seam mainly consisted of O-Ti₂AlNb, NiTi, AlNi₂Ti, TiAu, Ni₃Nb, (Ni,Au)ss dissolved with Cr(Al), Fe and Ni-rich phase dissolved with Ti and Al.

(2)The brazed seam presented a higher micro-hardness than the two base materials due to the formation of phases AlNi₂Ti, NiTi and Ni₃Nb. The average tensile strength of all the joints brazed at 1253 K for 5-20 min was above 356 MPa at room-temperature, in which the joints brazed at 1253 K/15 min exhibited the maximum tensile strength of 434 MPa at room temperature and 314 MPa at 923 K.

(3)Sound metallurgical bonding was obtained at AuNi/GH536 interface. Due to the diffusion of Ni atoms into Ti₃Al base alloy, complicated reaction products were formed at Ti₃Al/AuNi interface, including AlNi₂Ti, Ni₃Nb and NiTi compounds. These compounds especially the AlNi₂Ti ternary compound corresponded to the maximum hardness along the joint, and the narrow AlNi₂Ti reaction band appeared as the weak link of the dissimilar joint.

Acknowledgements

This work was sponsored by the National Natural Science Foundation of China (No. 51705489) and the State Key Laboratory of Advanced Brazing Filler Metals and Technology, Zhengzhou Research Institute of Mechanical Engineering (No. SKLABFMT201603).

The authors have declared that no competing interests exist.

---