Machine learning-driven alloy digital design for hydrogen storage: a review
Tianqi Liu, Lufeng Xue, Bo Cheng, Yu Zhao, Bang Dou, Marcelo Paredes, Lu-Kai Song
Abstract
Hydrogen has emerged as a promising renewable energy source, garnering significant attention from both academia and industry in recent years. The development of safe, efficient, and economical storage systems is critical for the widespread industrial application of hydrogen. For this purpose, the gaseous, liquid, and solid-state hydrogen storage systems are particularly promising for this purpose owing to their high hydrogen content, moderate operating conditions, and enhanced safety features. Subsequently, the design and fabrication of suitable alloy materials for hydrogen storage systems have emerged as a key research focus. Although conventional design approaches for such alloy materials mainly rely on intensive simulations and experiments, machine learning (ML)-based digital approaches offer greater design efficiency and cost-effectiveness, leading to increased attention in this area. Given this background, the present work aims to review the recent research progress and provide digital insights for future developments in the field. Specifically, this paper introduces the ML models and digital methods employed in alloy design, offers a comprehensive overview of hydrogen storage alloys, and provides a detailed discussion of ML models with significant potential for the design of various types of hydrogen storage alloys.