Materials Informatics
Micro-structural informatics: unfolding complexities with a bottom-up approach
Correlations between metals’ micro/nano hardness and the underlying microstructure has a well-established notion. Polycrystals, for instance, consist of multitudes of disoriented grains within a complex polycrystalline network that dictate the mechanical response (i.e. hardness) across nano and micro scales. Nevertheless, the nature of such inherent microstructure-property relations remains elusive and debated to this date. Conventional physics-based frameworks empirically describe grain boundary strengthening effects by a limited set of structural parameters ignoring inherent grain scale hierarchies and intricate topology of the grain boundary network at micro/nano-structural levels. Using a bottom-up approach in Karimi et al. (2023), I adopted a data-driven framework based on the state-of-the-art Graph Neural Net (GNN) model to infer grain-scale hardness from the complex grain boundary microstructural information. The GNN model was trained using Electron backscatter diffraction (EBSD) surface polycrystallinity maps, containing local lattice orientation information, which was supplemented by a nano-mechanical dataset obtained from nanoindentation experiments on a polycrystalline steel sample. Despite inherent complexities, the trained ML model was able to make robust predictions of the load-depth curves over a broad range of grain scales. Further investigations demonstrated that the model performance strongly depends on a certain set of grain-level (topological) attributes such as individual grain size, number of (nearest) neighbors, and grain-grain misorientation angles. As an ongoing project, our model aims to go beyond inferring bulk properties (such as hardness) by accurately forecasting intermittent displacement bursts (i.e. dislocation avalanches or pop-ins) and associated size and statistical distributions solely based on microstructural metrics.
Transport properties in High-entropy alloys (HEAs): sluggish diffusion of defects
Atomic-scale transport properties in HEAs have long been hypothesized to exhibit comparatively slow kinetics, as opposed to pure metals and conventional alloys, hence the term “sluggish diffusion”. This core effect is commonly identified as a principal source of exceptional HEA’s thermo-mechanical properties (e.g. single-phase thermodynamic stability, creep resistance, and high-temperature strength). Sluggishness of diffusion dynamics, in particular, connects to apparent compositional and underlying atomic-level complexities, but in an intricate way. Using a kinetic Monte Carlo (kMC) framework in Karimi et al. (2023), I demonstrated the sluggishness of vacancy dynamics in chemically complex alloys under thermal activation, and further modeled it in terms of subdiffusive fractional Brownian dynamics. I further established direct links between the subdiffusive vacancy dynamics and underlying crystal lattice distortions. The developed model can be further extended to incorporate high deformation rates and irradiation effects which allow for studying transport properties and their sluggish nature in multicomponent alloys under extreme conditions. As an ongoing research project, I am focusing on studying defect-dislocation interactions, in particular hydrogen-induced embrittlement in metals. This project will involve the development of machine-learned neural-net interatomic potentials in the presence of defects and model training based on ab initio calculations. such trained potentials will enable large-scale simulations with ab-initio precision, therefore facilitating accelerated composition search and the discovery of novel materials for energy storage and environmental applications.
