典型纯金属凝固过程中的Ostwald分步形核机制及结构遗传性

One of the significant problems in the process of designing and manufacturing materials (e.g. Materials Genome Initiative, MGI) is how to control solidification process. In the past, we learned the description of solidification and nucleation process from the classical nucleation theory (CNT) and recognized that the melt nucleates directly to the thermodynamically most stable phase through only one step. Recently, some research results imply that the solidification process, especially the solidification of deeply supercooled liquid, in some cases, is a multi-step process in which several metastable phases exist rather than one step process. This multi-step phenomenon questions the CNT greatly. Actually, this multi-step nucleation phenomenon was observed by Ostwald more than 100 years ago and consequently named by his as "Ostwald's step rule". Therefore, it is necessary to modify the CNT immediately. This project is proposed for this purpose. In this project, we plan to use computer simulation, spectral analysis and theoretical modeling to investigate the solidification (nucleation) process of three elemental metals: bcc-Fe, hcp-Mg and fcc-Al. Our aim is to shed new light of the Ostwald's step rule on the mechanism of CNT. Our works are divided into four parts. The first part is to design three efficient algorithms to accurately characterize micro-structures on atom scale, including the Cluster-Type Index Method (CTIM), the Bond-Orientational Order Parameter (BOP), and the Voronoi polyhedron (VP) method. The second part is to build the core-shell model of five-fold twinning cluster in solely and uniform field according to the first principle calculations. The third part is the core of the project. We will systematically analyze the nucleation and growth process of three elemental metals with different cooling speeds by molecular dynamics simulations and spectral analyses and clarify all the details of the multi-step nucleation process of both isolated cluster and periodical bulk. The final part is to build a multi-step nucleation model on basis of the deduction of CNT in the aspect of thermodynamics and to generalize the mechanism of structural heredity of metastable twinning crystal during solidification. Based on the research of this project, a new theoretical system about the non-equilibrium solidification process will be built and will enhance the understanding of this process and development of new materials.

凝固过程控制是材料制备设计过程（如"材料基因组"计划）中至关重要的一个环节。最新研究结果显示：许多凝固过程，如深过冷条件下的凝固，不是"一步到位"，而是分步行核，中间会出现大量亚稳相。这种现象严重冲击了目前仍然通用的经典行核理论（Classical Nucleation Theory, CNT），急需对该理论进行修正。本项目拟通过对典型的bcc-Fe、hcp-Mg、fcc-Al三种纯金属的凝固过程（形核过程）进行计算机模拟、光谱分析以及理论模型研究，从中制定完整的Ostwald分步行核机制。具体内容包括：原子标定技术的构建；通过第一性原理计算，构建孤立和均匀场条件下的团簇壳层结构模型；通过分子动力学模拟和光谱分析三种纯金属在不同冷却速度下的形核及晶核长大情况，总结形核过程中的系列关键参数；按照CNT的推导思路，从热力学角度构建完整的分步形核理论，并从中总结作为亚稳相孪晶结构的遗传特性。

我们对三种典型纯金属（BCC-Fe, FCC-Al, HCP-Mg）在凝固过程中的结构的演变进行了系统的研究。首先，我们改善了目前已有的原子标定的技术，将VP方法BOP方法相互结构，用VP法进行近邻原子判断，同时采用VP法标定BCC原子，然后用BOP方法中的Frenkel指数标定区分固液相原子，最后用BOP的四重指数q4来区分FCC和HCP原子。事实证明我们的方法比前人的方法更精确且适应能力更强。.在此基础上，我们对不同金属的不同势函数在描述相变的适应性验证方面提出了一个系统思路，这是让我们的模拟研究能够获得信服结果的前提基础。然后，我们对三类单金属体系的凝固过程做了系统的MD模拟研究，在给出系统热力学分析的基础上，仔细分析了体系的形核长大过程中的结构演化，明确提出了有序化领先致密化的演化过程。.我们还专门针对固液界面的微观结构提出了我们自己的认识——共轭双膜结构，这是国际上首次提出如此精确的结构模型。.在本项目支持下，我们针对多种透明和不透明单晶的结构及水和甘油的混合液的粘度进行了系统的布里渊散射光谱的测试分析，获得了多种弹性性能数值，包括国际上首次的光弹常数以及混合液的粘度值。.我们还针对Al做了系统的嵌入固相团簇诱导体系凝固的模拟研究，发现：这是目前验证经典形核理论最有效的方法，可以有效获得与晶核尺寸对应的过冷度，同时我们发现了超过一定过冷度之后的spinodal凝固现象，及不需要形核而直接凝固长大的过程。在改变晶核形状的研究过程中，我们发现形状因子对过冷度的影响效果，从而更精确获得了固液界面能。.最新的研究结果还告诉我们无论是在结晶还是玻璃化的过程中，体系的固态性质的表现在剪切试验中非常明显。.在这些研究的基础上，我们将提出液相统一理论。.所有的这些研究结果，都有可能对目前国际上对诸如液相结构、固液界面结构、凝固过程等基础问题产生一些新的认识和看法。