Atomistic Modeling Method Development and Its Application on the Study of the Nucleation Process of Crystalline Materials

Haiyang Niu*

School of Materials Science and Engineering, Northwestern Polytechnical University,Xi'an 710072

ABSTRACT: The study of the nucleation process of crystalline materials is one of the most fundamental research areas in material sciences, condensed matter physics, pharmaceutics and geosciences and many others. Nucleation is the starting point of the solidification process of materials, which is not only the prerequisite of the growing process, but also has decisive influence on the structure, morphology and properties of the materials, therefore, investigating the nucleation process thoroughly at atomic-scale is of great importance. However, limited by the spatial and temporal resolution, it remains a great challenge for the current experimental methods to observe directly the nucleation process of material. Computational material science, such as molecular dynamics, has played crucial role in the area discussed above, which has greatly advanced the development of material science. Yet despite remarkable process, the timescale of nucleation is still much longer than what can be reached in a direct molecular dynamic simulation. Advanced molecular dynamics method has shown to be a promising solution to crack this conundrum. In this talk, we will introduce a new approach, namely X-ray piloted Metadynamics (PNAS, 115, 5348, 2018), which is achieved by introducing the X-ray diffraction information of material into advanced enhanced sampling method Metadynamics. We have successfully applied this approach into simulating the nucleation process of many materials, such as SiO₂ and water (PRL, 122, 245501, 2019). The study of ice freezing shows that the nucleation is preceded by a large increase in tetrahedrally coordinated water molecules, and ice embryo is formed in such area. Based on the concept of material genome engineering, we have proposed a new approach by combing the advanced molecular dynamics methods and deep learning methods to build atomic potential of material with ab initio accuracy. Such approach has been successfully applied in the system of gallium (Nature Communications,11, 2654, 2020). In this work, we have uncovered the mechanism of deep quenching property of gallium and the competition puzzling between α and β phases during nucleation. We believe our study thus offers a path to calculate phase diagrams and study the nucleation of complex materials with ab initio accuracy at an affordable cost.

Keywords: advanced molecules dynamics methods; nucleation; collective variables; water; gallium.