机器学习原子间势分子动力学模拟在电化学储能材料研究中的应用进展-中国材料进展

文章信息/Info

Title:: Application Progress of Machine Learning Interatomic Potential Molecular Dynamics Simulations in the Research of Electrochemical Energy Storage Materials

作者:: 林奕希; 蒋雨桥; 冯相民; 要腾宇; 夏颖慧; 刘振辉; 郑明波; 申来法; 许真铭; 南京航空航天大学材料科学与技术学院江苏省高效电化学储能技术重点实验室，江苏南京 210016

Author(s):: LIN Yixi; JIANG Yuqiao; FENG Xiangmin; YAO Tengyu; XIA Yinghui; LIU Zhenhui; ZHENG Mingbo; SHEN Laifa; XU Zhenming; Jiangsu Key Laboratory of Materials and Technologies for Energy Storage, College of Materials Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China

Keywords:: molecular dynamics simulation; firstprinciples calculation; machine learning; molecular force field; electrochemical energy storage materials

摘要:: 电化学储能材料研究领域对分子模拟有着切实的需求，而经典分子动力学和从头算分子动力学模拟因无法兼顾精度和效率的问题限制了分子模拟的广泛应用。近年来，基于机器学习方法构建原子间势模型得到了快速的发展，机器学习原子间势分子动力学模拟可以兼顾经典分子动力学模拟的计算效率和从头算分子动力学模拟的准确性。为了更好地呈现机器学习原子间势分子动力学模拟技术在电化学储能材料研究领域的应用进展和前景，重点介绍了其在固体电解质、电解液、电极/电解质（液）界面等研究领域的应用，并总结了材料领域机器学习原子间势及其分子动力学模拟所存在的挑战和机遇。

Abstract:: There is a growing demand for molecular simulations in the field of electrochemical energy storage materials research. However, the widespread application of molecular simulations has been limited by the inability of the classical molecular dynamics and ab-initio molecular dynamics to balance the accuracy and efficiency. In recent years, the machine learning-based models for interatomic potentials have developed rapidly, offering the potential for the machine learning interatomic potential molecular dynamics (MLMD) simulations to achieve both the computational efficiency of the classical molecular dynamics and the accuracy of the ab-initio molecular dynamics. To better present the advancements and prospects of the MLMD simulation technology in the research of the electrochemical energy storage materials, this work focuses on its applications in solid electrolytes, electrolytes, and electrode/electrolyte interfaces, and summarizes the challenges and opportunities for the machine learning interatomic potentials and their molecular dynamics simulations in the materials field.