Low Cycle Fatigue Behaviour of Multi-joint Sample in Mechanical Testing

This paper explores the behaviour of a copper test vehicle with multiple reflowed solder joints, which has direct relevance to ball grid arrays (BGA) and high density interconnect structures. The paper explores the relative stress conditions on the distributed joints and the sensitivity to ball joint shape. The joints were exposed to isothermal fatigue, which was produced by a mechanical load that induced a cyclic shear stress across all the joints. The same structures were modelled using finite element analysis. The loading response distribution profile through the joints was analysed. The regions of likely failure were identified to be along the shear band and at the stress concentration areas in the corners of the joints. Failure of the individual joints was analysed by quantifying the accumulated creep strain per cycle. Solder joint models of three different shapes were investigated: rectangular, convex and concave shapes. This analysis has shown that less damage is found in concave shaped joints, indicating that BGAs would have more damage than the rectangular joints tested here. Results have also shown that more damage occurs in the outer joints as a vertical component appears due to a turning moment on the copper test vehicle. This behaviour could affect the external joints of large components, where the same vertical stress component may arise due to the differential CTE of the PCB and BGA component. Introduction Mechanical testing of solder joints in a cyclic mode provides materials’ data as the solder experiences fatigue. These data can be informative in discriminating between various solders, but can additionally be combined with finite element modelling in characterising behaviour. This paper investigates the properties of a multi-joint sample, studying the properties as the number of joints was increased. A motivation for this work was the relevance of this structure to area array packages. The popular Ball Grid Array (BGA) package offers high interconnection density, easy assembly, and low thermal resistance. Typically, a square array of solder balls connect a laminate substrate and a semiconductor die. Because of their geometry, the ball-shaped solder joints in BGAs are not able to easily deform to accommodate thermal-mechanical and other stresses [1]. Due to the Coefficient of Thermal Expansion (CTE) mismatch between Printed Circuit Board (PCB) substrate and the package, solder joints experience a cyclic strain in temperature varying conditions that induce fatigue in the solder joints and other parts of the assembly [1,2]. In this paper, we shall focus on isothermal fatigue testing of specimens containing multiple solder joints. The loading conditions on solder joints are similar to those found in BGAs that are subject to thermal-mechanical loading. The aim of the work is to establish an effective low cost test system that can be used for the evaluation of solder alloys that are used for BGAs, and to provide material data that can be used in physics-offailure based reliability prediction for BGA or other similar devices with solder interconnects. Experimental Procedure Solder interconnect samples with multiple joints were fabricated from copper coupons of size 5161 mm. A low speed diamond saw was first used to produce a 0.3 mm cut running centrally along almost the entire length of the coupon. This slit was placed over a jig comprising an oxidised stainless steel shim with a specified number of slots corresponding to the size and number of joints required in the sample. Before soldering, this space was filled in excess with the desired solder paste. In this study, 96.5Sn3.0Ag0.5Cu lead free solder was used. The jig was then placed onto a hot plate for solder reflow, after 5 seconds into the reflow process, the jig was removed from the hot plate and placed on a cold metallic surface to cool such that the time above the reflow temperature was between 10 to 30 seconds. Abrasive paper was used to remove excess solder and slots were cut in the sample (Figure 1) such that when pushed/pulled, a shear force is applied to the solder joints. Samples with 1 to 8 joints were prepared using this procedure and these are referred to as M1-M8. The samples were tested by applying isothermal cyclic displacement loading using the Interconnect Properties Testing Machine (IPTM), an instrument described in detail elsewhere [3, 4]. The ends of the samples were clamped and pulled/pushed along the length to produce cyclic deformation in solder joints. A load cell was used to monitor the supported force and laser displacement sensors were used to measure the distance between the fixed and moving stages of the instrument. The displacement was controlled through the thermal expansion of a stainless steel tube connected to the moving stage. As such, the machine can accurately control the strain applied to the sample, whilst measuring the supported As originally published in the IPC APEX EXPO Proceedings.