从单分子到凝聚相水体系的理论描述:势能面和分子动力学最新进展

Theoretical Description of Water from Single-Molecule to Condensed Phase:Recent Progress on Potential Energy Surfaces and Molecular Dynamics

本文回顾了近十年来水体系的势能面与分子动力学理论研究的最新进展,包括水分子参与的气相反应,固体表面上的吸附与解离动力学,以及从团簇到凝聚相水的结构、振动光谱与统计力学模拟.近年来再次发展起来的机器学习技术,例如结合置换不变多项式的神经网络,或结合基本不变量的神经网络,已被成熟应用于气相与固体表面体系的高精度势能面构造中.对于团簇甚至凝聚相水体系,原子中心神经网络方法或基于核的高斯过程方法应用更为广泛.此外,在多体展开框架下,在气相体系中发展起来的的方法也组成了高维度体系势函数构造的高精度方案.当前凝聚相水体系面临的主要问题是高精度从头算数据集的积累,兼顾计算精度与效率的双杂化密度泛函是一种可能的解决方案.在动力学理论方面,无论是化学反应截面计算还是振动光谱模拟,往往需要合理描述水分子中氢原子的量子效应,才能得到较为可靠的理论计算结果.量子波包动力学方法已经在气相反应机理研究方面有深入的应用,也在包含数个水分子的团簇振动分析中有初步应用.基于路径积分的分子动力学方法正在较大水团簇以及凝聚相水的结构与谱学模拟方面发挥重要作用.

In this work,we review recent progress on the view of potential energy surfaces and molecular dynamics study of water and its related reactions in the last decade or so.Some important gas-phase reactions of water with radicals,chemisorbed dissociative dynamics of water on solid surfaces,and statistical mechanics and vibrational spectrum simulations of water from clusters to the condensed phase have been introduced.The recently developed machine learning techniques,such as the neural networks in a combination of permutational invariant polynomials or fundamental invariants,the atomic neural networks framework,the gaussian approximation potentials with the smooth overlap of atomic position kernel,as well as the many-body expansion framework for the construction of highly accurate potential energy surfaces,have also been discussed.Finally,some suggestions have been provided for further improvement of the potential energy surfaces and dynamics methods of water-related systems.