Measurement and prediction of reliability for double-sided area array assemblies

Abstract

The increasing demand for smaller and faster products in the electronics industry has encouraged the use of CSPs and BGAs assembled on printed circuit boards (PCBs) in a backto-back double-sided fashion. A critical issue in doublesided assemblies is the thermal cycling reliability performance of the device-to-board level attachment. In this paper, the effects of components placed on both sides of the test board were studied. The components included CSPs, BGAs, QFPs and chip capacitors. To assess the assembly flexural stiffness, a three-point bendmg measurement was performed. The effect of assembly stifkess on thermal cycling reliability was investigated. Cycles-tdailure were documented for all assemblies and the data were used to calculate the characteristic life. In addition, moire interferometry was used to study the displacement dishibution in the solder joints at specitic locations in the packages. Data from the reliability and moire measurements were correlated with predictions generated fiom three-dimensional f~te element models of the assemblies. The models incorporated nonlinear and timetemperature dependent solder material properties and they were used to estimate the fatigue life of the solder joints and to obtain an estimate of the overall package reliability using Darveaux’s crack propagation method. Various configurations of double-sided assemblies were studied. These included symmetrically double sided assemblies (i.e., “mirror image” assemblies), 50% overlap assemblies, and assemblies with an area array component on one side, and chip capacitors (of various sizes) on the other side. In general, a 2X to 3X decrease in reliability was observed for mirror image assemblies when compared to single-sided assemblies for both BGAs and CSPs on 62 mil test boards. The reliability of mirror image assemblies when one component was an area array device and the other was a 208 VO quad flat pack (QFP) was comparable to the reliability of the single-sided area array assemblies alone, that is, the QFP had almost no influence on the double-sided reliability when used with an area array component. Generally, the fmite element models predicted the assembly reliability within the range expected for Darveaux’s method. As building test assemblies and running reliability assessments is both expensive and time consuming, the finte element models can provide a cost-effective altemative to investigating the impact of double sided components on the assembly reliability.