Theme 1: Improve the multi-scale modeling paradigm through potential energy landscape (PEL)-based method
Background & Challenge:
The mechanical properties of structural materials are largely controlled by dislocations and their interactions with obstacles. Dislocation-mediated microstructural evolution is an immensely complicated phenomenon involving multiple physics over many orders of magnitude in both spatial and temporal scales. Therefore, the inherently multi-scale processes require a reliable multi-scale mechanistic modeling framework, to ensure that the fundamental mechanisms at atomic level can be accurately captured and then transferred to higher level models at meso- and macro-scales.
Figure. (a) represents a widely-accepted paradigm for the multi-scale modeling on dislocation-related problems. It is expected that the fundamental mechanisms revealed by ab initio or MD method at atomic-scale can be transferred to meso-scale models such as DD or VPSC, and eventually to macro-scale model such as FEM. However, some studies show that conventional MD simulation might provide misleading microstructural evolution mechanisms. An example that supports this is a remarkable controversy seen in Figure. (b): in situ TEM experiments clearly show that a perfect SFT (a vacancy-type defect cluster named stacking fault tetrahedra) is fully absorbed and removed by a single moving dislocation. However, parallel MD simulations show that the same type of obstacle remains intact after cutting by a moving dislocation. The reason for such discrepancy is due to the huge gap in timescale between the in situ TEM (~10-2 s-1) and MD simulations (~107 s-1).
Our Methodology:
The reason why MD simulations cannot reach long timescale is that sometimes the system is trapped into deep energy basin surrounded by high activation barriers. Then according to transition state theory (TST) the kinetics becomes very slow and microstructural evolutions would take much longer time way beyond MD's accessibilities. To enable the long timescale modeling, a new school of thought is to proactively explore the materials underlying potential energy landscape (PEL), identify the transition pathways and barriers, and further incorporate that information into TST calculations. The PEL-based modeling has already been applied in dislocation-mediated mechanics (Fan et al, PNAS 110 (2013) 17756; Fan et al, PRL 109 (2012) 135503), and shows great potential in understanding and predicting the deformation mechanisms under complex environments spanning many orders of magnitude in timescale. Given these successful applications, we therefore propose a “detour” on the current multi-scale modeling roadmap. As illustrated by the green arrows in Figure (a), instead of directly connecting MD with DD/VPSC, we will first employ the PEL-based modeling to gain a more comprehensive knowledge at longer timescale and then pass the hereby obtained atomistic insights to meso-scale models. This would enhance the predictive power of the existing multi-scale modeling framework.
Theme 2: Mechanics of Disordered Solid Materials
Background & Challenge:
Due to the lack of lattice periodicity and consequently the absence of conventional defects, e.g. dislocations or grain boundaries, disordered materials such as glasses can reach very high mechanical strength that is much closer to materials ideal strength, the Frenkel limit, than their crystalline counterparts do. The high strength and high elastic limit of metallic glasses have led to novel applications, e.g. in the biomedical area, sporting, and modern electronics. However, glassy materials generally suffer from lack of macroscopic ductility due to work softening, which results in strain localization and catastrophic failure. This has been the main hurdle for structural application of metallic glasses. While the deformation mechanisms in disordered systems have been extensively studied, the understanding of their plasticity at a fundamental level still faces key challenges in two aspects: (i) First, glasses represent complex non-equilibrium states of matter that lacks long-range order, posing great theoretical challenges; (ii) Secondly, while computational modeling have yielded notable advances in our knowledge of complex materials at the atomistic levels, traditional techniques, e.g. molecular dynamics (MD), can only tackle very short timescales and therefore face formidable challenges in probing materials behavior far from equilibrium.
Our Methodology:
Our recent studies (Fan et al, Nat. Commun, 8 (2017) 15417; Fan et al, PRL 115 (2015) 045501; Fan et al, Nat. Commun, 5 (2014) 5083) suggest that, if we abandon the conventional “topological structure” and adopt the perspective of potential energy landscape (PEL), then it is possible to find some hidden orders in the disordered glassy materials. This PEL-based model thus allows to understand the kinetics and energy dissipations in glassy system and predict the critical phenomena concerning dynamic heterogeneity (e.g. shear bands formation) and memory effect (e.g. thermo-mechanical hysteresis).