Influence of BGA Design Parameters on Solder Joint Reliability

Autonomous vehicles are predicted to constitute 90% of all U.S. vehicle sales by the year 2050. This will greatly expand the use of semiconductor devices in an already growing market. Electronic components in automotive environments are subjected to harsh conditions: including large temperature variations, mechanical shock, vibration and other wear-out failures. Of the different types of failures, second level interconnects are most vulnerable to fatigue failures. These solder interconnects are particularly sensitive to thermal and mechanical loading due to Coefficient of Thermal Expansion (CTE) mismatch with printed circuit boards (PCBs). To counter this effect, a robust package design approach is necessary across all material layers that can survive the required field life. Accelerated tests are run to assess the package performances and if the package survives a certain target cycles in -40C to 125C environment, it is deemed qualified to meet the customer requirements. This is at least a 5x harsher than the performance requirements for mobile or compute type applications. The cycle of design iterations and qualification tests take several months from start to finish which is very inefficient and doesn’t nearly keep up with the customer needs. A robust and accurate life prediction model [1] is an effective engineering tool to improve designs and accurately predict fatigue life of the package within a short time without need to perform expensive and time-consuming reliability testing. However, the life prediction accuracy of finite element simulation models is dependent on modeling assumptions, geometrical complexity, creep constitutive model of solder and fatigue life model calculation [2–9]. In the current study we first build a simulation framework that can sufficiently predict the semiconductor package life with varying design and material parameters (see Figure 1 and 2). This model is fine-tuned for material behavior, predicted life and test vehicle data. Following that, we used the model to understand the influence of various design parameters to the predicted life. This provided the first level screening of identifying the right knobs that will be critical in enhancing package life. The information was very critical to guide the design of next generation packages as well as modify some of the existing packages that are not meeting the customer requirements. This paper details the parametric study and key parameters for improved solder joint reliability in harsh automotive environments.