球栅阵列
焊接
材料科学
蠕动
温度循环
共晶体系
合金
冶金
复合材料
结构工程
热的
工程类
物理
气象学
作者
Joshua Adeniyi Depiver,Sabuj Mallik,Emeka H. Amalu
标识
DOI:10.1016/j.engfailanal.2021.105447
摘要
Abstract As temperature cycling drives fatigue failure of solder joints in electronic modules, characterisation of the thermal fatigue response of different solder alloy formulations in BGA solder joints functioning in mission-critical systems has become crucial. Four different lead-free and one eutectic lead-based solder alloys in BGA solder joints are characterised against their thermal fatigue lives (TFLs) to predict their mean-time-to-failure for preventive maintenance advice. Five finite elements (FE) models of the assemblies of the BGAs with the different solder alloy compositions are created with SolidWorks. The models are subjected to standard IEC 60749–25 temperature cycling in ANSYS mechanical package environment. Plastic strain, shear strain, plastic shear strain, and accumulated creep energy density responses of the solder joints are obtained and inputted into established life prediction models – Coffin Manson, Engelmaier, Solomon and Syed – to determine the lives of the models. SAC405 joints have the highest predicted TFL of circa 13.2 years, while SAC387 joints have the least life of circa 1.4 years. The predicted lives are inversely proportional to the magnitude of the areas of stress–strain hysteresis loops of the BGA solder joints. The prediction models are significantly not consistent in predicted magnitudes of TFLs across the solder joints. With circa 838% variation in the magnitudes of TFL predicted for Sn63Pb37, the damage parameters used in the models played a critical role and justifies that a combination of several failure modes drives solder joints damage. This research provides a technique for determining the preventive maintenance time of BGA components in mission-critical systems. It proposes developing a new life prediction model based on a combination of the damage parameters for improved prediction.
科研通智能强力驱动
Strongly Powered by AbleSci AI