ReaxFF力场方法及其在含能材料中应用的研究进展

(1.淮阴工学院 化学工程学院,江苏 淮安 223003; 2.湖南云箭集团有限公司 制导武器装备研究院,湖南 长沙 410100; 3.南京理工大学 化学与化工学院,江苏 南京 210094)

量子化学; 含能材料; ReaxFF力场优化; 分子模拟; 理论计算

Overview of ReaxFF Force Field Method and It's Application in Energetic Materials
SONG Liang1,MEI Zheng2,ZHANG Tian-cheng1,ZHOU Su-qin1,JU Xue-hai3

(1.Faculty of Chemical Engineering, Huaiyin Institute of Technology, Huaian Jiangsu 223003, China; 2.Research Institute of Guidance Weapon Equipment, Hunan Vanguard Group Co. Ltd., Changsha 410100, China; 3.School of Chemistry and Chemical Engineering, Nanjing University of Science and Technology, Nanjing 210094, China)

quantum chemistry; energetic materials; ReaxFF force field optimization; molecular simulation; theoretical calculation

DOI: 10.14077/j.issn.1007-7812.202211009

备注

从ReaxFF分子动力学在含能材料领域的研究出发,聚焦于研究ReaxFF力场的构成和发展、力场优化方法和应用等方面。在力场开发方法方面,对局域优化方法和全局优化方法进行了重点介绍,其中多目标进化策略、增强的粒子群优化算法和基于神经网络等开发策略均有所涵盖。ReaxFF力场在含能材料模拟中的应用主要集中于探索微观结构反应性、动态演化、初始化学分解路径以及中间和稳定产物分布。根据外部条件不同而分为加热和加压、冲击加载、激光和电场等应用场景。研究主体涉及单质/共混/共晶炸药、含有缺陷炸药以及金属复合炸药。最后,针对ReaxFF的准确性和迁移性进行了展望,认为在人工智能框架下的ReaxFF反应力场训练可以大幅提高其在复杂反应中的准确性。在连续体仿真模型中,若将其局域反应度等关键指标的唯象规律替换为基于ReaxFF的反应规律,将是实现一个包含力学和化学动力学响应耦合的多尺度全流程仿真计算的重要途径。附
The review starts from the research of ReaxFF molecular dynamics in the field of energetic materials and focuses on the composition and development of the ReaxFF force field, optimization methods, and applications. Among the development methods, local and global optimization methods are highlighted, including multiple objective evolutionary strategies, enhanced particle swarm optimization algorithm, and neural network reactive force field. The application of the ReaxFF force field in energetic material simulations has mainly focused on exploring microstructural reactivity, dynamic evolution, initial chemical decomposition pathways, and intermediate and stable product distributions. Depending on the different external conditions, the application fields are heating and pressurization, shock loading, laser, and electric field. The main research concerns single component/mixture/co-crystal explosives, defective explosives, and metal composite explosives. Finally, there are some suggestions for the accuracy and transferability of ReaxFF. ReaxFF reactive force field trained with the artificial intelligence framework can significantly increase its accuracy in complex reactions. If reaction laws based on ReaxFF replace phenomenological laws for critical indicators such as local reactivity in continuum simulation models, this would be a crucial step towards realizing the multi-scale full-process simulations with mechanical and chemical kinetic response couplings. The 134 references are attached.