(M.Sc./PhD Spring/Summer 2026) : Multiscale Modeling of Lithium Diffusion in Graphite Anodes
The project focuses on advancing the understanding of lithium-ion diffusion in synthetic graphite through an interdisciplinary approach that integrates advanced material characterization with computational modeling. By linking graphite’s structural and defect features to its lithium-ion transport properties, the research aims to inform the design of next-generation materials for faster charging lithium-ion batteries. The team will develop and calibrate diffusion models informed by experimental data, correlate computational predictions with electrochemical performance metrics such as cycling and impedance results, and interpret detailed structural insights from XRD and electron microscopy analyses. Collaborative multiscale modeling will connect atomistic, mesoscale, and continuum perspectives, enabling a comprehensive understanding of transport phenomena, while outcomes will be disseminated through high-impact publications and presentations for academic and industrial audiences.
(M.Sc./PhD Spring/Summer 2026) : AI-Driven Discovery of Lightweight Structural Alloys through Rapid Synthesis and Data Feedback Loops
This project aims to speed up the discovery of lightweight, high-strength structural alloys (especially aluminum-based) using Artificial Intelligence (AI) and Machine Learning (ML). The goal is to design alloys that help reduce fuel consumption and greenhouse gas emissions in transportation by providing better strength-to-weight ratios. Working with the National Research Council (NRC) and the University of Waterloo, the project combines computational alloy design with rapid experimental synthesis using Hot Isostatic Pressing (HIP). The process integrates three stages.
(M.Sc./PhD Spring/Summer 2026) : Development of JAX-Based Classical Density Functional Theory Tools for Investigating Crystallization and Phase Stability
This project aims to develop JAX-based computational tools to investigate crystallization phenomena within the framework of classical density functional theory (cDFT). Although cDFT provides a rigorous theoretical foundation for studying the formation and stability of crystalline phases, its practical application has been limited by the numerical instabilities of advanced tensor-based Fundamental Measure Theory (FMT) functionals. The proposed work will implement and extend heuristic, numerically stable cDFT models in JAX to enable robust simulation of fluid-to-solid transitions and interfacial crystallization processes. Using these tools, the project will explore the crystallization behavior of Lennard-Jones fluids and related systems, capturing structural transitions from liquid droplets to amorphous and crystalline phases (HCP and FCC). This open-source JAX package will provide a flexible platform for investigating free-energy landscapes, stability relations, and nucleation pathways, bridging theory and computation in the study of condensed matter systems.
(M.Sc./PhD Spring/Summer 2026) : QRNN Potential for Predictive Corrosion Modeling in Molten Salt Reactors
This project develops a GPU-accelerated machine learning potential using the Charge Recursive Neural Network (QRNN) framework to model impurity-driven corrosion in structural alloys within molten FLiNaK salts. By combining large-scale quantum mechanical training data with active learning, the model captures polarization, dispersion, and charge transfer effects critical to alloy degradation. Integrated into TorchSim and LAMMPS, the QRNN will enable reactive molecular dynamics simulations that predict thermophysical properties and corrosion behavior under reactor conditions, advancing materials design and safety for next-generation molten salt and small modular reactor technologies.
(M.Sc./PhD Spring/Summer 2026) : Physics-Informed Neural Networks for Dual Oxide Layer Growth and Degradation in Nuclear Alloys
This project develops a high-performance Physics-Informed Neural Network (PINN) leveraging advanced GPU computing to predict the growth, stability, and degradation of dual oxide layers in nickel and iron-based alloys exposed to halide-rich environments in advanced nuclear reactors. Building on established oxide growth frameworks, the GPU-optimized PINN captures coupled dynamics of electron transport, ionic diffusion, and halide interactions within the layered structure of stainless steel passive films. By integrating the Point Defect Model with GPU-parallelized PDE solvers, the framework enables large-scale simulations of halide ingress and oxide evolution under diverse reactor conditions. Training combines physics-based simulations with experimental data from electrochemical impedance spectroscopy, oxide thickness measurements, and redox mapping, providing predictive insights into corrosion mechanisms critical for nuclear materials design and reactor safety.
Qualifications:
- Strong programming skills (e.g., Python, C++, or Julia).
- Strong background in computational materials science or electrochemical modeling
- Experience with DFT or MD simulations (VASP, LAMMPS, or similar)
- Knowledge of machine learning interatomic potentials (e.g., SNAP, GAP, or NequIP)
- Familiarity with phase-field modeling or continuum-scale transport equations (Fick, Cahn–Hilliard)
- Proficiency in Python, JAX, PyTorch, for simulation and data analysis
If interested, please submit your CV, a brief statement of research interests, and references to cgtetsas@uwaterloo.ca
Application Instructions
Applicants should submit the following documents as a single PDF to cgtetsas@uwaterloo.ca
- Cover letter outlining research interests and fit for the position
- Curriculum vitae (CV) including publications and code repositories (if applicable)
- Contact information for two to three referees