Date of Award
2025-12-01
Degree Name
Doctor of Philosophy
Department
Computational Science
Advisor(s)
Jorge Munoz
Abstract
This dissertation investigates the stability and physical properties of iron-based materials under extreme conditions, utilizing a combination of machine learning techniques and first principles calculations. In the first study, we employ Gaussian Process Regression (GPR) with a Marginalized Graph Kernel to analyze the mechanical stability of non-magnetic body-centered cubic (BCC) iron at high temperatures and pressures, similar to those found in Earth’s core. Our model, trained on data derived from ab initio molecular dynamics (AIMD) simulations, predicts the energy landscape, atomic forces, and potential pathways for instability in BCC iron. The findings suggest that BCC iron remains stable in extreme environments, providing valuable insights into Earth’s deep interior and advancing computational methods for analyzing material stability. The second study uses density functional theory (DFT) with the Vienna Ab Initio Simulation Package (VASP) to assess the effects of magnetic ordering on the structural, electronic, and mechanical properties of FeV alloys. We compute the equation of state, electronic density of states (DOS), partial DOS, and band structures for both ferromagnetic and antiferromagnetic configurations, along with an analysis of the elastic constants. The results reveal significant variations in electronic and mechanical behaviors based on the magnetic states, highlighting the potential applications of FeV alloys in magnetic devices. By integrating machine learning-driven stability predictions with first-principles electronic structure calculations, this dissertation provides a comprehensive understanding of iron-based materials across diverse physical environments. The interaction between these approaches enhances our ability to model complex material behaviors, bridging the gap between computational predictions and experimental challenges in geophysics and materials science.
Language
en
Provenance
Received from ProQuest
Copyright Date
2025-12
File Size
108 p.
File Format
application/pdf
Rights Holder
Blaise Awola Ayirizia
Recommended Citation
Ayirizia, Blaise Awola, "Machine Learning And First-Principle Approach To The Stability And Properties Of Iron-Based Materials Under Extreme Conditions." (2025). Open Access Theses & Dissertations. 4516.
https://scholarworks.utep.edu/open_etd/4516