A detailed description of the Cu-Al wire bond interface is presented, which can possibly explain the often observed corrosion failures in humidity reliability tests. Using micro-structural analysis techniques, it is shown that the unstressed interface contains up to three intermetallic phases, where the Cu-rich phases are located at the Cu-ball interface. Upon humidity stress test only the high-Cu containing intermetallic layers close to the Cu wire ball bond undergo a corrosion process, whilst the Cu-lean layers are stable in all environment stress tests. The failing layers consist out of an amorphous Al-based oxide matrix with embedded Cu precipitates. The failure process can be explained as galvanic corrosion of the intermetallic phases. The Cu-rich phases corrode faster compared to Al-rich phases, since they have an electrochemical potential lower than the Cu cathode and form, hypothetically, a less stable self-passivation oxide. The observed time-to-failure is then determined by the composition, thickness and volume of the intermetallic layers. This was verified by reliability test results performed on open wire bonds and on plastic encapsulated products. ©2010 IEEE

No abstract

We prepared 41 different superalloy compositions by an arc melting, casting, and heat treatment process. Alloy solid solution strengthening elements were added in graded amounts, and we measured the solidus, liquidus, and c¢-solvus temperatures of the samples by DSC. The c¢-phase fraction increased as the W, Mo, and Re contents were increased, and W showed the most pronounced effect. Ru decreased the c¢-phase fraction. Melting temperatures (i.e., solidus and liquidus) were increased by addition of Re, W, and Ru (the effect increased in that order). Addition of Mo decreased the melting temperature. W was effective as a strengthening element because it acted as a solid solution strengthener and increased the fraction of fine c¢-precipitates, thus improving precipitation strengthening. Experimentally determined values were compared with calculated values based on the CALPHAD software tools Thermo-Calc (databases: TTNI8 and TCNI6) and MatCalc (database ME-NI). The ME-NI database, which was specially adapted to the present investigation, showed good agreement. TTNI8 also showed good results. The TCNI6 database is suitable for computational design of complex nickel-based superalloys. However, a large deviation remained between the experiment results and calculations based on this database. It also erroneously predicted c¢-phase separations and failed to describe the Ru-effect on transition temperatures.

No abstract

No abstract