自适应加速分子动力学的研究及其在α-Fe中间隙原子扩散机制时间多尺度分析中的应用

Multiscale analysis of materials, with significant scientific and application values, is now a challenging topic and frontier in pushing the development of materials science. At present, one of the main bottleneck in materials multiscale research is the modeling and algorithm realization of multi-time-scale analysis method. This project aims to develop a new kind of multi-time-scale analysis method which is called adaptive accelerated molecular dynamics (AAMD). In AAMD, together with the features and advantages of hyperdynamics and metadynamics, newly developed density estimator methods are applied to establish the smooth distribution of atomic configuration with respect to collective variables, based on which a kind of adaptive bias potential is constructed and then mixed with classical molecular dynamics. AAMD is then used to launch the simulations of diffusion of interstitial atoms C and N in α-Fe lattice and areas nearby dislocations at different temperatures, in which atomic scale and experimental time scale can be achieved..This project will not only push the development and application of multi-time-scale analysis method, but also help to deeply understand the diffusion mechanism of two different interstitial atoms in α-Fe, and then gain deep insights of the effect of interstitial atoms on properties and characteristics of steels.

材料多尺度分析是推进材料学科发展的前沿与难点，具有重大的科学与应用价值。目前材料多尺度研究领域的主要瓶颈之一是时间多尺度分析方法的建模和算法实现。本项目拟综合超动力学和巨动力学的特点和优势，采用最新发展的密度场估计方法，建立集合参量关于原子构型的光滑分布模型，并构建一种自适应偏移势，进而与经典分子动力学进行融合，发展一种新型的时间多尺度分析方法- - 自适应加速分子动力学。在此基础上，利用发展的自适应加速分子动力学对不同温度下间隙原子C、N在α-Fe晶格中及位错近邻区域的扩散过程开展空间尺度具有原子级别、时间尺度具有实验室（以上）级别的分子动力学模拟。.本项目的研究将推动时间多尺度分析方法的发展和应用，并有助于深入了解α-Fe中两种间隙原子的扩散机制及其异同，从而为充分认识间隙原子对钢特性的影响机理提供有价值的参考。

本项目完成了集合参量关于原子构型的分布模型的建立，在此基础上设计了自适应偏移势的构建算法，并将其引入至经典动力学方法中，完成了自适应偏移势构建的算法实现。应用密度泛函理论工具和过渡态搜寻方法对N原子在α-Fe晶格的扩散行为展开了深入研究，获得了α-Fe中N原子的表面和体扩散路径和扩散激活能，阐明了N、C间隙原子在α-Fe中扩散机制的异同。进一步研究了gamma-Fe中N原子与空位（v）的相互作用，获得了N-v点缺陷团簇的分布与N原子浓度和温度的关系，揭示了空位浓度随N化学势变化的机理，验证了Kirchheim的热力学理论。