About
I am a PhD candidate in computational materials science at Queen's University, working with Laurent Karim Béland in the Department of Mechanical and Materials Engineering. I use first-principles simulation, machine-learning interatomic potentials, and kinetic modelling to understand how atomic-scale defects govern the performance of structural alloys and catalytic surfaces — connecting vacancy chemistry in nuclear materials to product selectivity in CO₂ conversion.
Currently working on
- Ni–Cr–Al k-ART production runs. MACE-driven kinetic trajectories are now resolving vacancy–solute clustering and Al segregation to grain boundaries on second-to-hour timescales — the regime where oxide-scale chemistry is actually decided.
- CO₂ on stepped & defected Cu. Mapping how step-edge vacancies and Zn / Ag dopants reshape CO₂ binding energetics and the *CO – *COH branching point that gates C₂⁺ selectivity.
- Zr–Nb radiation damage in hcp. Hybrid ML / EAM benchmarking against DFT to establish the accuracy floor needed before pushing k-ART into the hcp regime relevant to CANDU fuel cladding.
- Foundation-model screening. Stress-testing MACE-MP-0 as a zero-shot baseline for new alloy chemistries, with light fine-tuning where it matters for saddle-point energetics.
Recent updates
May 2026
Ni–Cr–Al k-ART production trajectories with MACE potentials are now resolving multi-vacancy clustering and Al grain-boundary enrichment over experimentally meaningful timescales — the temporal window where conventional MD runs out. CO₂ binding screening has expanded to stepped and defected Cu facets, applying the same vacancy-mediated segregation framework developed for the Ni-alloy work to predict the surface compositions actually accessible under reaction conditions.
Winter 2026
Zr–Nb radiation-damage modelling underway with hybrid ML/EAM benchmarking against DFT references — pushing the long-timescale kinetic framework from fcc Ni alloys into the hcp regime relevant to reactor-grade Zr. Co-instructed and TA'd MECH 479 (Nanomaterials) for the second consecutive year, with a new computational module on reproducible MD workflows.
2025
First-author manuscript on Cr–vacancy interactions submitted to Scripta Materialia (preprint, SSRN 5759303). Completed k-ART parameterisation with MACE potentials for fcc Ni alloys. Began CO₂ catalysis DFT screening on alloyed Cu and Zr-alloy point-defect calculations.
Research
My work centres on a single question: how do point defects interact with alloying elements to control composition, structure, and properties at interfaces? Vacancy–solute coupling is the common thread — governing oxidation resistance in Ni-based superalloys, radiation tolerance in Zr fuel cladding, and product selectivity on catalytic surfaces for CO₂ conversion. Machine-learning interatomic potentials (MACE, ACE) and the kinetic Activation-Relaxation Technique (k-ART) bridge electronic-structure calculations to experimentally relevant timescales. Four interconnected projects share a core DFT → ML potential → kinetic modelling pipeline:
Vacancy–solute coupling in Ni–Cr–Al — How Cr raises vacancy concentrations and drives Al to grain boundaries, connecting electronic structure to oxidation resistance.
k-ART for long-timescale defect kinetics — Parameterising k-ART with MACE and ACE potentials to bridge DFT energetics to observable microstructural evolution.
Radiation damage in Zr alloys — Extending the vacancy–solute framework to Zr-based fuel cladding under irradiation, with hybrid ML/EAM potentials for hcp systems.
CO₂ catalysis on metal surfaces — Surface-defect energetics on alloyed Cu controlling binding, selectivity, and reaction pathways for electrochemical CO₂ conversion.
Shared multi-scale pipeline
Electronic
DFT
Å · ps
Surrogate
ML potential
MACE · ACE
Long timescale
k-ART
nm · s–h
Observable
Microstructure
segregation · barriers
All four projects feed into the same multi-scale workflow: DFT reference data trains ML potentials, which enable k-ART kinetics at experimentally relevant timescales. Universal foundation models (MACE-MP-0, CHGNet) are increasingly enabling zero-shot screening across new chemistries before committing to full active-learning cycles. See
Research for project details and emerging work on autonomous simulation workflows.
Methods
Multi-scale workflow: Quantum ESPRESSO and NWChem for DFT, MACE / ACE machine-learning potentials, LAMMPS for MD, and k-ART for long-timescale kinetics — orchestrated with Python (ASE, pymatgen, NumPy, SciPy, matplotlib) and bash on SLURM-managed HPC clusters. Universal foundation models (MACE-MP-0, CHGNet) provide zero-shot baselines for new chemistries, and LLM-driven agents are beginning to automate convergence testing and adaptive sampling around the loop. See
Research for the full toolkit.
Teaching
I teach undergraduate courses in materials science and chemistry at Queen's University, with prior experience coordinating laboratory sections at Ontario Tech. My teaching emphasises reproducible computation, hands-on problem-solving, and responsible integration of AI tools in scientific workflows. See
Teaching for course details.
Background
MSc in Materials Science (2023) and BSc in Chemistry (2021) from Ontario Tech University. Earlier research applied quantum chemistry methods to ion-pair trapping mechanisms in functionalised organic systems, developing the computational surface-chemistry intuition that now informs my work on alloy grain boundaries and catalytic surfaces.