Atomistic Modeling of Defects and Phase Transformations in Energy Materials

Patel, Shane

doi:https://doi.org/10.21985/n2-4ncv-jy20

Work

Atomistic Modeling of Defects and Phase Transformations in Energy Materials

Public

Download PDF

Download All Files (.zip)

Atomistic methods offer a powerful set of tools in the study of materials systems, as they allow materials scientists to ask questions with a high degree of specificity. They are well suited for studying and designing energy materials, critical due to the climate crisis, in part due to their ability to probe defect properties. In this document, we present projects that extend the reach of these tools, and use them to study both long established materials systems, such as Zr cladding in nuclear reactors, as well as cutting edge materials such as disordered layered Li-ion battery cathodes. We begin with a brief overview of how defects are studied using atomistic methods, and how this can apply to different energy material systems, in Chapter 2. In Chapter 3, we adapt the minima hopping method (MHM) to interfacial structure prediction and apply it to study a canonical problem, the non-stoichiometric grain boundaries in SrTiO3. Our method employs a hybrid approach by first exploring the potential energy surface with an empirical force field to generate candidate structures, which are then refined using ab-initio Density Functional Theory (DFT) calculations. Using this approach, we find stable interfacial structures for SrTiO3 (111) and (112) grain boundaries that are lower in energy compared to those reported in the literature for given system size. Our method allows the prediction of interfacial structure at the atomic scale to improve our understanding of grain boundaries and heterointerfaces. In Chapter 4, we study the role of cation disorder in Li3IrO4 cathodes, which have alternating cation planes of pure Li layers and a disordered Ir-Li plane, in facilitating anionic redox, using DFT. We calculate a cluster expansion to explore structural stability in the fully lithiated phase, and subsequently calculate the behavior upon delithiation of both a stable ordered structure and a model disordered structure. We then perform a high-throughput screening of Li3MO4 structures, uncovering novel phases and identifying Li3OsO4, Li3PtO4, and Li3RhO4 as potential candidates for further study as a battery cathode. In Chapter 5, we have developed a novel Moment Tensor Potential (MTP), a class of machine learning interatomic potentials, to flexibly treat dissolved hydrogen in hexagonalα-Zirconium in tetrahedral and octahedral interstitial sites, as well as the variety of relevant hydride phases: γ-ZrH, the dominant δ-ZrH2−x, and ε-ZrH2. Our approach is to train MTP using an active learning scheme based on NPT classical molecular dynamics (MD) simulations at varying temperatures. Our trained MTP is capable of modeling the phase transformation at the solvus boundary between metallic α-Zr and the hydride phases in excellent agreement with DFT, while also capturing the temperature and compositional dependence of the ε-ZrH2 to δ-ZrH2−x transformation. Finally, we validate the capability of the MTP to capture relevant phenomena such as H diffusion, H-vacancy ordering, and point and planar defect behavior.

Creator