Chapter 4 Results
4.4 The effect of brazing environment
After comparing the AlN-Ticusil-AlN system, graphite-Ticusil-AlN system prepared in flowing nitrogen at 1050 oC and prepared under vacuum at 900 oC, an overview of the joining system was observed. However, there were still hard to directly compare the graphite-Ticusil-AlN system in the two different environment. The joining temperature and atmosphere were different in the two joining system. As the result, a series of step changes of environment from vacuum at 900 oC, then within nitrogen at 900
oC, to within nitrogen at 1050 oC was conducted to observe the influence of atmosphere and temperature.
The observation of the influence of atmosphere was started from the specimen prepared in flowing nitrogen at 900 oC. Since the fact that reaction phases mainly concentrated at AlN-Ticusil interface so that the change at this interface was more obvious, the observation was focused on the AlN-Ticusil interface. As the image shown in Fig. 4-25, the morphology of column layer was changed after the flowing nitrogen was induced.
The column structure significantly decreased and lots of particles dispersed near the TiN layer. The containing range of these dispersed particles was around 6-7 um from the TiN layer, while the thickness of the TiN layer was only 1.5 um. The composition of those dispersed particles contained not only titanium and copper but some aluminum. Some of the darker particles within the back-scattering image contained of some oxygen or high nitrogen content. Besides, there was still some column structure grains formed near those particles.
When the joining temperature raised up to 1050 oC, the amount of column structure grains further decreased, and could only be found near the TiN layer. The titanium-copper reaction product layer became mainly composed of the particles. These particles
contained some of nitrogen. However, there was no obvious dependence between the composition of these particles and their position. After the joining temperature was raised, the containing range of the particles increased to around 8-9 um, while the thickness of the TiN layer was around 2.5 um.
The factor of the change of the joining atmosphere was also investigated. The atmosphere of 5% H2 – N2 and argon were chosen to observe the influence of different atmosphere. The specimen joined in 5% H2 – N2 at 1050 oC was shown in Fig. 4-27. The thickness of TiN layer was still around 2.4 um, while the containing range of those particles was increased to 11-12 um. The titanium-copper particles and nitrides were mixed together and could only be distinguished by the color of the particles under back-scattering images. The oxides and nitrides would be in darker color since their average atomic mass were low. Besides, there was almost no column grains could be detected.
Only little column grains could be found after etching the specimen with ion beam to eliminate most of the particles, as shown in Fig. 4-28. The size of those column grains was also smaller comparing to those prepared in vacuum.
The specimen prepared in argon was shown in Fig. 4-29. The thickness of the TiN layer was around 2.3 um. A thin layer composed of small column grains was attached to the TiN layer. The thickness was around 2.5 um, with the composition of those small grains was also titanium nitrides. Some of the particles formed with a small gap after the column grain layer. These particles were mainly composed of the titanium-copper oxides,
Figure 4-25 AlN-Ticusil interface for graphite-Ticusil-AlN system prepared in nitrogen at 900 oC for 15 min, subscript P and L denote for particles and layer, respectively.
Figure 4-26 AlN-Ticusil interface for graphite-Ticusil-AlN system prepared in nitrogen at 1050 oC for 15 min, subscript P and L denote for particles and layer, respectively.
Figure 4-27 AlN-Ticusil interface for graphite-Ticusil-AlN system prepared in 5% H2 – N2 at 1050 oC for 15 min, subscript P and L denote for particles and layer, respectively.
Figure 4-29 The AlN-Ticusil interface for graphite-Ticusil-AlN system prepared in argon at 1050 oC for 15 min, in which subscript P and L denote for particles and layer,
respectively.
4.5 Characterization of reaction products
In order to have a fully characterization of the reaction products formed on the interface, the specimens were grinded to the interfaces, following polished and cut by FIB to prepare TEM sample. The interfaces were then observed by TEM. The composition and the structure of the reaction products were identified by TEM-EDS and selected area diffraction pattern (SADP).
The characterization of reaction products started from the vacuum specimen, since the morphology of the reaction product was relatively regular and the joining environment was stable. Fig. 4-30 showed the observation on the AlN-Ticusil interface of the vacuum specimen. The layer structure beside the AlN substrate was first identified, as shown in Fig. 4-30 (b) and (c). The crystal structure was found to be cubic structure and the d-spacing was near 0.220 nm for its [010] zone axis, which corresponded to the structure of TiN. The EDS composition analysis also showed a matched result. This reaction product layer was identified as TiN.
The TEM observation result of the column structure at AlN-Ticusil interface of the vacuum specimen was shown in Fig. 4-30 (d) and (e). The crystal structure was found to be a near-cubic structure, with the low index plane showed in Fig. 4-30 (d). However, the characterization of the column structure was relatively difficult. The column structure was
matched for the composition-location dependence that Ti decrease and Cu increase with the distance to the TiN layer.
The observation of Graphite-Ticusil interface of the vacuum specimen was also conducted, as shown in Fig. 4-31. A clear image of a thin layer of reaction product formed between graphite paper and Ticusil filler was collected. Thickness of this layer was about 200 nm and still bonded to the substrate after joining and etching by FIB. Moreover, some distortion within graphite paper beside the product layer was also found. The distortion was concentrated near the interface and sorted parallel to the planar direction. The distortion was speculated to be formed by the thermal induced stress. The characterization of the reaction product was also conducted, as shown in Fig. 4-31 (b) and (c). The crystal structure was a cubic structure. By comparing the SADP and composition result with database, this reaction product was identified as TiC.
The further observation of the specimen joined in atmosphere was also conducted.
Fig. 4-32 showed the observation on the AlN-Ticusil interface of specimen joined in 5%
H2 – N2 atmosphere. The layer beside the AlN substrate was also identified as TiN layer.
However, the other part of reaction product no longer column structure but became dispersed particles. The particles dispersed near the interface and showed a non-regular position. The crystal structure and composition of these particles were analyzed, as shown in Fig. 4-32 (b) and (c). These particles were identified as TiN particles. However, some of the TiN particles dispersed far from the TiN layer contained relatively higher copper content than other TiN particles, which was mainly due to the diffusion of copper.
Figure 4-30 The (a) TEM image of AlN-Ticusil interface for graphite-Ticusil-AlN system prepared under vacuum at 900 oC for 15 min. (b) The diffraction pattern and
corresponding simulated pattern and (c) EDS composition analysis of TiN layer structure. (d) The diffraction pattern and corresponding simulated pattern and (e) EDS
composition analysis of (Ti, Cu, Al)6N column structure
Figure 4-31 The (a) TEM image of Graphite-Ticusil interface for graphite-Ticusil-AlN system prepared under vacuum at 900 oC for 15 min. (b) The diffraction pattern and
corresponding simulated pattern and (c) EDS composition analysis of TiC layer structure.
Figure 4-32 The (a) TEM image of AlN-Ticusil interface for graphite-Ticusil-AlN system prepared in 5% H2 - N2 gas at 1050 oC for 15 min. (b) The diffraction pattern and
4.6 Four-point bending test
The four-point bending test was applied to the specimen to discover the fracture behavior of the joined system. The test started from the ceramic substrate AlN first. The load was applied directly on the AlN substrate. After the applied load reached the flexural strength of AlN plate, the fracture behavior was catastrophic and the sample broke down suddenly into several pieces, as the result shown in Fig. 4-33. The recorded maximum flexural strength of the AlN substrate was 455 MPa. The flexural strength of the joined AlN-Ticusil-AlN specimen was also measured, as shown in Fig. 4-34. The fracture behavior was similar to the AlN substrate. The specimen suddenly broke down after reached the maximum strength. However, the flexural strength of the AlN-Ticusil-AlN was slightly decreased to 425 MPa.
The fracture behavior of the graphite-Ticusil-AlN specimen was also observed.
However, since the two side of the joined specimen was different, the direction of the applied load would affect the fracture behavior. When the AlN side was on the top side, the load was directly applied on the AlN substrate, the result was shown in Fig. 4-35. The first component that fractured during the bending test was the graphite paper. However, since the flexural strength between the graphite paper layers too low to be detected, the loading curve was smooth when the graphite paper ruptured. The maximum recorded flexural strength was 237 MPa, which was significant lower than the AlN substrate.
Nonetheless, the fracture behavior was changed in this system. After the loading reached the maximum strength, the specimen did not break into pieces. The ruptured component was still attached to the Ticusil layer. The further load onto the specimen made the specimen, or precisely to say, the Ticusil layer bend. The attached AlN and graphite paper ruptured continuously when the load still applied. As the result, the loading curve would
decrease smoothly after a drop when the load reached the flexural strength limit. Further extension of the specimen was attributed to the bending of the Ticusil layer. The fracture behavior was then become smooth. However, when the specimen was put in the opposite direction, in which the graphite paper was on the top side. The fracture behavior became catastrophic again, as shown in Fig. 4-36. The specimen broke into pieces when the flexural strength reached the peak. The maximum flexural strength was recorded as 199 MPa.
Figure 4-33 Loading curve for AlN substrate. The AlN substrate was broken catastrophically after reaching the maximum load.
Figure 4-34 Loading curve for the AlN-Ticusil-AlN specimen. The joined system was broken catastrophically after reaching the maximum load.
Figure 4-35 Loading curve for the graphite-Ticusil-AlN specimen. The graphite was under the tensile stress. The drops of load indicated one rupture during loading
4.7 Thermal conductivity
Besides the mechanical properties, the thermal conductivity was also an important factor of the performance of specimen. The thermal diffusivity was measured by the laser flash method conducted in the vertical direction to the plane of the specimen. The measured range was from 25 oC to 300 oC. The heat capacity was also obtained by DSC.
Thus, the thermal conductivity could be calculated by equation (3.1). The calculated thermal conductivity values were shown as below.
Table 4.1 showed the thermal conductivity of AlN substrate, AlN-Ticusil-AlN system and graphite-Ticusil-AlN system. The joined systems were conducted in flowing nitrogen joined at 1050 oC for 15 minutes. The thermal conductivity at 25 oC of AlN substrate was around 233 W/mK. Though the thermal conductivity of AlN-Ticusil-AlN system at 25 oC to 136 W/mK, joining with graphite paper lead to a higher thermal conductivity at 25 oC to 177 W/mK. On the other hand, the thermal conductivity at higher temperature showed another trend. Though the thermal conductivity at 25 oC of AlN substrates was the highest among the three systems, the thermal conductivity at 300 oC was the lowest one. Joining with another AlN substrate would increase the thermal conductivity to 84 W/mK and joining with graphite paper would be even higher to 102 W/mK.
The thermal conductivity of the specimens prepared in different joining atmosphere was shown in Table. 4-2. The specimen joined in 5% H2 – N2 showed a slightly decreased thermal conductivity at 25 oC while the value at 300 oC was similar to that of in pure nitrogen. However, the specimen prepared in flowing argon showed a significantly lower value at both 25 oC and 300 oC. The thermal conductivity of specimen joined under argon was 81 W/mK at 25 oC and 53 W/mK at 300 oC. The value of the specimen joined under
argon was almost just half of the value prepared in flowing nitrogen. The thermal conductivity of vacuum specimen was lower than that of joined in nitrogen-rich atmosphere. However, the value was slightly higher than in argon atmosphere. The composition and morphology of the reaction products would influence the path of the thermal flow and thus influence the thermal conductivity.
The specimen joined in prolonged holding time at 1050 oC in flowing nitrogen was also prepared. The holding time was increased to 60 minutes to observe the influence of the holding time to the thermal conductivity. The thermal conductivity result was shown in Fig. 4-3. The thermal conductivity was significantly decreased after joining in the prolonged holding time. The values were decreased to 84 W/mK and 60 W/mK for 25 oC and 300 oC, respectively.
A simple test on the stability of the specimen in the thermal cycling was also conducted. The specimens were taken to repeat the process of joining, which was to heat to 1050 oC to hold 15 minutes and then cooled down to room temperature with the heating rate and cooling rate both 3 oC/min, for 5 times, 10 times and 15 times. The thermal conductivity of the specimens showed a gradually decreased value at both 25 oC and 300
oC as the cycles increased. After 15 cycles of the process, the thermal conductivity of the specimen was only 71 W/mK and 48 W/mK for 25 oC and 300 oC, respectively. The value of the thermal conductivity was less than the half of the value of the specimen with no repeated process.
characteristic. The behavior of graphite paper along planar direction was a great thermal conductor while along the vertical direction was a thermal insulator.
Table 4-1 The thermal conductivity at 25 oC and 300 oC for AlN substrate, AlN-Ticusil-AlN system and graphite-Ticusil-AlN-Ticusil-AlN system
Table 4-2 The thermal conductivity at 25 oC and 300 oC for graphite-Ticusil-AlN system joined in flowing N2, 5% H2 – N2 and argon atmosphere
Table 4-3 The thermal conductivity at 25 oC and 300 oC of graphite-Ticusil-AlN system joined for 15 minutes and 60 minutes holding time
Table 4-4 The thermal conductivity at 25 oC and 300 oC for graphite-Ticusil-AlN system after different thermal cycles
Figure 4-37 Thermal conductivity of graphite paper along xy-plane
Chapter 5 Discussion
5.1 Brazing conditions
In this study, the aim was to achieve good joining of graphite-Ticusil-AlN joining system within flowing atmosphere. Most of the joining process in the previous studies was conducted in vacuum instead of in flowing atmosphere [25, 63, 67]. Joining in atmosphere or under vacuum would affect not only the phases formed on the interfaces, but also the joining parameters that needed to achieve a good bonding between the components. The parameters such as joining temperature, holding time, the condition of braze foil and the atmosphere used to join would all have influence on the performance of the specimen. Especially for the joining temperature, the wettability of the braze foil would be significantly affected by the joining temperature. As the braze foil Ticusil used in this study, the composition of Ticusil is Ag 68.8 wt.% - Cu 26.7 wt.% - Ti 4.5 wt.%, the corresponding phase diagram was shown in Fig. 5-1 [27]. The solidus point was around 780 oC for this composition. However, joining at 780 oC could not achieve a good bonding at both interfaces since the wettability to AlN or graphite at this temperature was not enough, especially for the graphite side. The joining temperature had to be increased as the result. As the temperature increased to 900 oC in flowing nitrogen, it was found that the AlN side was wetted by Ticusil, while the graphite side remained partially wetted.
As the result, to achieve fully wetted interfaces at both sides, the increased temperature to 1050 oC was then used for the flowing atmosphere cases. However, the temperature needed to achieve fully wetted interfaces under vacuum (5 x 10-5 torr) was much lower.
Around 850 oC – 900 oC could achieve the fully wetted interfaces when joined under vacuum. Meanwhile, higher joining temperature would lead to the boiling of the low boiling point component under high vacuum, silver in Ticusil braze foil, then be extracted
by the vacuum pump. The joining temperature above the boiling point under vacuum of silver should be avoided. In order to compare with the specimen joined in atmosphere, the joining temperature under vacuum was then set to be 900 oC.