Reliability Evaluation of BGA Solder Joints by Using Thermal Fatigue Models

Abstract

The use of Ball Grid Array (BGA) interconnects utilizing the BGA solder joint has grown rapidly because of its small volume and diversity of its application. Therefore, the continuous quantification and refinement of BGA solder joint in terms of its reliability are required. The creep and cyclically applied mechanical loads generally cause metal fatigue on the BGA solder joint which inevitably leads to an electrical discontinuity. In the field application, the BGA solder joints are known to experience mechanical loads during temperature changes caused by power up/down events as the result of the Coefficient of Thermal Expansion (CTE) mismatch between the substrate and the Si die. In this paper, extremely small resistance changes in the lead free joints corresponding to the through-cracks generated by the thermal fatigue were measured and the failure was defined in terms of anomalous changes in the joint resistance. Furthermore, the reliability of BGA solder joints under thermal cycling was evaluated by using a criterion that may define and distinguish a failure in the solder joint. Any changes in circuit resistance according to the accumulated damage induced by the thermal cycling in the joint were recorded and evaluated by the First Order Reliability Method (FORM) procedure in order to quantify the reliability of solder joint. The first order Taylor series expansion of the limit state function incorporating with thermal fatigue models is used in order to estimate the failure probability of solder joints under heated condition. Various thermal fatigue models are utilized in this study. Models based on various plastic-strain rates such as Coffin-Manson fatigue model, total strain fatigue model and Solomon fatigue model are utilized in this study. The effects of random variables such as the CTE, the pitch of solder joint, the diameter of solder joint, and the CTE difference solder joints on the failure probability of the solder joint are systematically investigated by using a failure probability model with the FORM.