A Bayesian Neural Network-based approach for multistate reliability assessment of solder joints exposed to various failure mechanisms

Abstract

The evaluation of the operational lifespan of solder joints in electronic devices, especially when subjected to intricate mission profiles, stands as a pivotal imperative in the domain of electronic material design. In response to this critical concern, this study introduces an innovative methodology designed to discern between gradual and sudden failures, offering a nuanced exploration of the distinct failure mechanisms at play. Distinguishing itself through a sophisticated treatment of failure mechanisms—specifically, the incorporation of thermomechanical fatigue and vibration-induced mechanisms—within the context of solder joint degradation, the proposed approach significantly elevates the precision and accuracy of reliability evaluations. This refined methodology emerges as a valuable contribution to the field, promising more insightful and reliable results in the assessment of electronic device durability under diverse and complex operational conditions.