UC San Diego

Most natural and engineered crystalline materials are polycrystalline, and grain boundaries (GBs) are the central crystal imperfections for describing these polycrystalline materials. Solute or impurity segregation at GB is a critical interfacial phenomenon because it can induce GB structural transformation, dramatically change the microstructural evolution, and cause catastrophic GB embrittlement. In a broader context, GB segregation (a.k.a adsorption) can control a broad range of kinetic, electronic, thermal, magnetic, and other materials properties. Thus, understanding the underlying physical mechanism of GB segregation is of both fundamental and practical interest to materials science community.

It has been proposed that GB can adopt thermodynamic equilibria like a three-dimensional (3D) bulk phase. Such a GB state can be considered as a 2-D stabilized interfacial phase, which also named as “complexion” to differentiate from abutting bulk phases. Since the properties of GB can be as important as those of bulk phases, it is also useful to develop GB counterparts to “bulk” phase diagrams by constructing GB “complexion” diagrams as a function of temperature and bulk composition. It is interesting to note that, in some systems such as Mo-Ni and Si-Au binary systems, the computed GB diagrams exhibit first-order transformation lines with critical points, which unequivocally suggests the phase-like GB transition behavior. Meanwhile, GB transition can be continuous in many other systems, e.g., Cu-Ag, Cu-Zn, Al-Ga, etc. In both cases, developing GB diagrams as a function of temperature and bulk composition is meaningful and useful.

However, constructing GB diagrams is difficult for both experiments and modeling. On one hand, the advanced electron microscopy and atom probe tomography are common experimental tools used to characterize atomic-level GB structures, but the sample preparation and experimental procedure are complicate for characterizing even one GB. On the other hands, the calculated GB diagrams were mainly based on phenomenological or Lattice-type models, but these models are too simplified and cannot reflect atomistic details of GB. First-principles calculation was used to calculate interfacial energetic diagram of WC, but only few configurations were investigated. Later, large-scale molecular dynamic simulations were adopted to compute GB diagrams, but most of studies were limited to symmetric tilt and twist GBs. The more general (asymmetric) GBs, which are more ubiquitous in polycrystalline materials, are still scarcely studied. Thus, one motivation of this dissertation is to construct GB diagrams for “general” GBs.

Furthermore, a GB has five macroscopic degrees of freedoms (DOFs), thus developing GB diagrams as a function of five DOFs associated with temperature and bulk composition in 7-D space remains a grand challenge even for a simple binary system. Recently, a new class of high-entropy alloys (HEAs), which generally contains five or more consecutive elements, has created extensive research interests. The large compositional space of HEA also makes it impossible to develop GB diagrams as a function of four compositional DOFs and temperature in 5-D space even for one specific GB. Therefore, we applied advanced computational techniques by combining high-throughput simulations and machine learning to predict GB diagrams in high dimension.

Finally, it is also interesting to decipher the fundamental mechanism of GB segregation. Notably, first-principles calculations were used to explain the formation mechanism of a highly asymmetric interfacial structure of WC. Further, ab initio molecular dynamic simulations were performed to verify the enhanced grain growth by electrochemical reduction in Bi2O3-doped ZnO. The understanding of physical mechanism of GB segregation not only broadens our knowledge of GB segregation, but also enriches the segregation theories.