Correlation-driven machine learning for accelerated reliability assessment of solder joints in electronics

摘要: The quantity and variety of parameters involved in the failure evolutions solder joints under a thermo-mechanical process directs reliability assessment electronic devices to be frustratingly slow expensive. To tackle this challenge, we develop novel machine learning framework for systems; propose correlation-driven neural network model that predicts useful lifetime based on materials properties, device configuration, thermal cycling variations. results indicate high accuracy prediction shortest possible time. A case study will evaluate role material joint thickness devices; illustrate variations strongly determine type damage evolution, i.e., creep or fatigue, during operation. We also demonstrate how an optimal selection balances types considerably improves lifetime. established set stage further exploration processing offer potential roadmap new developments such materials.