The lead-free, Sn-3.5Ag-xCu and Sn-3.5Ag-0.7Cu alloys were prepared in two different forms; firstly, bulky as-cast alloys were cold rolled and thermally stabilized before the creep tests so that there would be very small amount of microstructural change during creep (TS), and secondary thin specimens were water quenched from the melt (WQ) to simulate microstructures of as-reflowed solders in flip chips.
The TS-alloys generally showed the secondary and the tertiary creep characteristics only without those of the primary stage, and the minimum strain rates $(\dot ε_min)$ were lowest for the TS-Sn3.5Ag0.75Cu specimens. The stress exponents (n) of $\dot ε_min$ were usually around 4, with the exception of the TS-Sn3.5Ag0.5Cu and TS-Sn3.5Ag0.75Cu alloys, which showed somewhat higher values of n. Therefore, the high temperature creep dominated by the lattice diffusion controlled dislocation climb was thought to be the main deformation mechanism. Additions of Cu had little effects on the creep ductility of TS-Sn3.5Ag based ternary alloys up to 1%, but further addition of Cu ( 1.5% ) deteriorated ductility substantially. Fractographic analyses revealed typical creep rupture by the nucleation and growth of cavities in the matrix except the TS-Sn3.5Ag1.5Cu specimens that showed cavity nucleation at brittle $Cu_6Sn_5$ particles.
Cooling rates of the WQ specimens ranged between 140K/sec and 150K/sec, and the resultant β-Sn globule size was about 5~10 times smaller than that of the TS specimens. Subsequent creep tests showed $~10^2$ times lower $\dot ε_min$ and $~10^2$ times higher $t_f$ for the WQ specimens. Rupture time analyses based on the Kachanov equation proved to be accurate within a factor of three for all cases, however the model based on necking was good only for the TS specimens. Fractographic analyses raised a need to conduct creep tests with thicker specimens for the thin WQ specimens.
