A comparative assessment of metal-Al₂O₃ joints formed using two distinct transient-liquid-phase-forming interlayers

This assumes that the diffusion of Ni into Nb during heating to 1300–1320 °C is negligible, and that all the Ni is available to form a liquid. Interdiffusion during heating can affect the temperature at which liquid formation initiates and the amount of liquid formed.

The contact-angle measurements were performed on droplets of fixed composition in a range of furnaces, none of which offered explicit control over, or assessment of the oxygen partial pressure. For cases where experiments were done in both a refractory metal element vacuum furnace, and in the graphitic environment of the vacuum bonding furnace, no differences in contact angle were detected, suggesting that the oxygen partial pressures sampled were in the plateau region where contact angles are insensitive to the oxygen partial pressure [41]. The conditions that arise in a sessile-drop experiment and in TLP bonding experiments of the type described in this paper differ in a number of important ways. In TLP bonding, the liquid is present as a very thin film that is in direct contact with the ambient atmosphere only along the bond perimeter. The ratio of the liquid/vapor to liquid/solid interface areas in a sessile-drop experiment and in TLP bonding typically differ by several orders of magnitude. At the perimeter, the solid refractory metal core layer is also in contact with the ambient atmosphere and can locally modify the oxygen partial pressure. Core layer dissolution occurs during bonding. Most refractory metals contain dissolved oxygen, and thus the oxygen activity along the interlayer/ceramic interface is quite likely more strongly impacted by the core layer’s dissolved oxygen content than the ambient oxygen pressure. Finally, in contrast to a typical sessile-drop experiment, the composition of the liquid changes with temperature and in ternary systems can also adjust with time. Thus, the impact of the ambient oxygen partial pressure in a sessile-drop experiment and in a TLP bonding experiment may be very different.

The dominance of failures within the ceramic indicates that the intergranular intermetallic phases seen within the interlayer do not provide a preferred fracture path, and do not appear to play a critical role in defining the joint strength.