A review of data-driven methods on predicting solder joint fatigue lifetime in microelectronics packaging

As a crucial interconnection structure in electronics packaging, evaluation of the lifetime and damage conditions of solder joints has become an important theme in related fields. To meet the requirements of the electronics reliability in the design stage, rapid estimation of the lifetime of solder joints accurately has drawn considerable attentions in electronic packaging. With the development of data-driven methods in reliability assessment, some advances have been achieved on the lifetime prediction of solder joints in electronics packaging with Artificial Neural Network (ANN), Radial Basis Function Neural Network (RBFNN), Recurrent Neural Network (RNN), or Convolutional Neural Networks (CNN) based models. This new trend has brought new hope and possibility on the establishment of the rapid and accurate lifetime prediction models for solder joints. The main data-driven methods developed in electronics packaging have been reviewed and discussed in this paper. The future trends and challenges in data-driven methods on predicting fatigue lifetime of solder joints are also presented.