Tin-Lead solder has been widely used in microelectronic packaging for its beneficial electrical and mechanical properties. But, the motions of banning Pb which is toxic for human health and environment are expanded in a worldwide and specially for electronic products. Recently, Sn-3.5Ag eutectic based ternary alloys have been gaining wide support from industry, but data on their creep properties are rather rare and experiments with solder joint specimens are hardly found in literature.
In the present work, Sn-3.5Ag-xCu or -xBi alloys with five different levels of Cu (0, 0.5, 0.75, 1.0, 1.5wt%) and Bi (0, 2.5, 4.8, 7.5, 10wt%) were prepared. Firstly, conventional creep tests using rolled and stabilized bulk specimens were conducted in order to understand the creep mechanism of the alloys under the stable microstructure. Creep tests were performed under constant loads for the stress levels lower than 10MPa at 373K. For the Bi containing alloys, creep strength showed the maximum around 2.5wt% and tended to decrease with more Bi content. The stress exponent of the alloy was around 4, suggesting typical dislocation creep, but the exponent was 2 for the 10wt% Bi alloy, suggesting creep assisted by grain boundary sliding. For the Bi containing alloys, the brittle fracture mode appeared showing small amount of reduction of area, while the ductile fracture mode was true for the Bi free alloy. Microstructural examination of ruptured specimens showed cavitations on grain boundaries normal to the load axis, and a significant of grain boundary sliding particularly for the 10wt% Bi alloy. For the Cu bearing alloys, the creep strength saturated around the Cu content of 0.75-1.0wt% but decreased with further additions. The alloys also showed the stress exponent of 4, and creep voids formed around intermetallic particles. On the whole, Cu containing alloy showed better creep resistance than that of Bi containing alloy.
Secondly, ball shear tests which are widely used on industry were performed in order to find their relevance to the creep characteristics of lead-free alloys. The results showed that the variation of ball shear strengths of Cu containing alloy was not observed but ball shear strengths of Bi alloy decreased as Bi content increased after aging. It was attributed to the fact that the failure paths were between the intermetallics and solder matrix for the Bi alloys, while those paths went through the solder matrix for the Cu alloys. And, Bi containing alloy showed the higher ball shear strengths than those of the Cu alloys. Therefore, the ball shear strength is nothing but the degree of strength of solder alloy and it is not related to the solder joint reliability.
Thirdly, in order to simulate the creep deformation of solders under the real package situation, single lap shear creep tests were conducted. Solder balls within the two PCB with Cu or Ni(Au) metallization were reflowed for the test. Results showed that Cu containing alloy had the lower creep rate than that of Bi containing alloy the same as the result of bulk creep test. But, the amount of Cu content which affected the creep rate differed from the each solder pad. 0.75Cu alloy showed best creep resistance on Ni pad as expected while 0Cu alloy showed best on Cu pad implying that Cu element already dissolved in Sn-3.5Ag alloy during reflow. Main strengthening of Cu containing alloy was the dispersed precipitations of $Cu_6Sn_5$. Therefore, Sn-3.5Ag alloy on Ni or Au/Ni pad which have not contain the fine $Cu_6Sn_5$ precipitates shows higher creep rates. As Cu content increased up to 1.5wt%, the creep rate decreased due to the enlarged $Cu_6Sn_5$ or increased plate-like $Ag_3Sn$ intermetallics which caused the interphase sliding.
Conventional bulk specimen didn’t consider the effects of substrate elements, while lapshear specimen have a difficulty in the mechanical analysis for its complex geometry but it is very adaptive for analyzing the real package product.