OMPAC Package C5 Reliability—Parametric Study

Abstract

Over molded pad array carrier (OMPAC) also called as ball grid array (BGA) has been used in products, such as hand held two way radio within Motorola. Now it has gained interest outside Motorola in companies like Compaq for portable computers and AT&T for the telephone systems. The reduced size, thickness, and increased I/O density at the board level are the attractive features of the OMPAC package over other competing package types. In Ref. [1, 2], life time of C5 solder interconnects under temperature cycle loading was reported. That study was based on nonlinear FEM simulation incorporating creep behavior of the solder. In the current study, influence of four key package design variables on C5 solder reliability is addressed using analyses methodology described in Ref. [1, 2]. The selected design variables are thickness of the substrate, the C5 pitch, solder ball diameter, and height to diameter (h/d) ratio of the C5 interconnect. For each design variable, three levels (low, middle, and high) are considered. The full study required a total of 81 simulation runs. However, statistical design of experiments were used and an L9 design was selected in the current study. The maximum permanent strain range for stabilized temperature cycle was calculated for all the nine case studies. In all the cases it occurs at the substrate interface in the solder ball just inside the edge of the die. The response (maximum total permanent strain range and thus maximum number of temperature cycles before failure) as a function of the four design variables are studied and the trends are: as the substrate thickness, C5 ball diameter and C5 (h/d) ratio increases and C5 pitch decreases, the maximum total permanent strain range decreases and thus cycles to failure increases. The order of importance of the design variables for the C5 reliability are (i) the C5 (h/d) ratio, (ii) the C5 ball diameter, (iii) the C5 pitch, and (iv) the substrate thickness. The C5 (h/d) is the key design variable. The maximum total permanent strain range decreases from 0.0055 for (h/d) ratio of 0.5 to 0.0045 for (h/d) ratio of 0.7. This in other words increases cycles to failure from 3000 cycles to 4400 cycles, an improvement of 47 percent.