A Systematic Literature Review on the Application of Artificial Neural Networks for Predicting Sn-Cu Lead-Free Solder Joint Strength

Abstract

Artificial neural networks (ANNs) are increasingly used to predict the mechanical strength and reliability of Sn-Cu lead-free solder joints in modern electronic packaging. However, current research shows varying modelling practices and limited comparison across methods. This paper reviews ANN applications in predicting Sn-Cu solder joint strength using a structured search on Scopus and Web of Science, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyse (PRISMA) guidelines. A total of 21 relevant studies were analysed. The review identifies three key research directions: (i) ANN-based prediction of solder joint strength and reliability with improved accuracy over empirical and finite-element models; (ii) fatigue, creep and life prediction models for lead-free solders under diverse thermomechanical loading; and (iii) hybrid AI techniques such as genetic algorithms and physics-guided networks that enhance performance and model robustness. The review reveals that ANN models consistently outperform traditional predictive techniques; however, challenges persist in data scarcity, experimental complexity and the integration of micro-mechanistic knowledge into learning architectures. This evidence underscores the necessity for hybrid physics-guided ANN models and standardised benchmarking protocols to advance reliable prediction ecosystems for Sn–Cu solder interconnects in mission-critical applications.