Column-grid-array (CGA) technology could lead to a highly reliable package design

Abstract

An analytical stress model is developed for a short cylinder (beam) with clamped and offset ends. The offset is caused by an external lateral force that has to be determined from the known offset. It is envisioned that such a beam can adequately mimic the stresses and strains in a column of a column-grid-array (CGA) solder joint interconnection. Such an interconnection is characterized by an elevated stand-off height of the solder joints compared to the joints in a widely used today ball-grid-array (BGA) system. In a simplified and conservative analysis the ends offset in a CGA joint can be determined beforehand as a product of the known thermal mismatch strain between the IC package and the printed circuit board (PCB), and the position of the joint with respect to the package mid-cross-section. The objective of our analysis is to use the developed model to demonstrate that the application of CGA technology enables one not only to significantly relieve the thermally induced stresses in the solder material, but to do that to an extent that the induced stresses remain within the elastic range. The analysis is limited therefore to elastic deformations. It should be emphasized that while the classical Timoshenko short-beam theory seeks the beam's deflections caused by the combined bending and shear deformations for the given loading, an inverse problem is considered here: the lateral force and the induced stresses are sought for the given end offset. In short beams this force is larger than in long beams, since, in order to achieve the given displacement (offset), the applied force has to overcome both bending and shear resistance of the beam. The carried out analysis and the numerical data indicate that by employing beam-like CGA solder joints one could possibly manage to remain within the elastic range, i.e. to avoid inelastic strains and, hence, low cycle fatigue conditions. If this is achievable, the fatigue lifetime of the solder material will increase dramatically, because the low-cycle fatigue condition will be replaced by the linear accumulation of elastic damages. The numerical example indicates that the stand-off of the CGA joint should be rather large to make the shearing stress low compared to the normal bending stress. The height-to-diameter ratio of the CGA joint should be increased to the level of about 12–13. In such a situation the Bernoulli beam model, could be used instead of Timoshenko model for stress evaluations. The further increase in the stand-off height over this ratio, even if it is technologically achievable, is not advisable, since this will not lead to an appreciable further stress reduction. Future work will include, but might not be limited to, the finite-element-analysis (FEA) computations and experimental evaluations (such as, e.g., shear-off testing and/or temperature cycling) of the induced stresses in, and the fatigue lifetime of, typical BGA and novel CGA joints. We would like to point out that a solder joint in isolation is neither reliable nor unreliable, and that reliability has meaning only in the context of interconnections either within package or outside of package onto PCB. For this reason the future work should include also better understanding of how to translate the stress relief in a single joint into the improved reliability of an interconnection as a whole, thereby leading to a highly-reliable solder joint interconnections and, since these interconnections are the bottle-neck of the today's IC packaging technologies, to a highly reliable package design as a whole. Particularly, for BGA/CGA package assemblies, it is critical to determine the effective offset values for the peripheral joints.