Application of deep neural network in fatigue lifetime estimation of solder joint in electronic devices under vibration loading

Comprehensive rapid reliability analysis of electronic devices in the design level while simultaneously considering all the crucial parameters is undoubtedly of paramount importance. This paper demonstrates the feasibility of using deep neural network as a comprehensive tool in reliability analysis of the solder joint under mechanical loading. Application of artificial intelligence in lifetime estimation of the solder joint hand in hand with the adequate finite element simulations provides the basis of penetrating reliability analysis into the electronic design level owing to its less computational time consumption. The proposed prediction model is used to capture the effects of electronic chip location, solder joint chemical composition, and vibration loading direction on the solder joint useful lifetime. The results reveal that the predicted solder joint lifetime is sufficiently accurate in comparison with finite element simulations. This capability introduces this approach as a powerful tool in the rapid reliability analysis of the electronic devices under mechanical loading. Furthermore, the accuracy of the finite element simulations is validated through experimental tests.