Research

Connecting AI to the real physics of materials.

Our ability to generate and screen new materials is accelerating — but our theory still struggles to connect that capability to the underlying physics of materials under real operating conditions. My research works to close that gap, using first-principles methods, statistical mechanics, and machine learning to model how energy materials actually behave when they're working.

Computational Materials & Catalysis

Iridium oxide (IrO₂) is the state-of-the-art catalyst for the oxygen evolution reaction in acidic water electrolyzers — central to clean hydrogen — yet it dissolves and restructures under operation. My work follows this single problem across three connected efforts: seeing how IrO₂ degrades, predicting those pathways cheaply, and discovering the surface structures that actually appear under electrochemical conditions.

Seeing degradation: collective dissolution in IrO₂

Published · JACS 2026 first-principles + in situ LP-TEM
rutile {110} surface point defect monolayer delamination
Atomic-scale dissolution pathways observed on IrO₂ {110}: point defects, reconstruction, and monolayer delamination.

This study combined first-principles modeling with in situ liquid-phase transmission electron microscopy to resolve, atom by atom, how IrO₂ nanocrystals dissolve under oxidizing conditions. Computational Wulff constructions incorporating high-index facets predicted how nanoparticle shape evolves with electrochemical potential, while atomically resolved imaging revealed several distinct collective dissolution pathways — high-index facet formation, monolayer reconstruction, step-edge formation, and monolayer delamination — with device-scale electrolyzer tests confirming the same restructuring. I contributed to the computational side of this multiscale effort.

Predicting degradation: ML-accelerated sampling

In preparation machine-learned interatomic potentials
A machine-learned potential evaluates candidate structures cheaply; sampling explores the energy landscape — settling in minima, fluctuating, and hopping barriers — to find the thermodynamically favored end state.

Predicting how a nanoparticle restructures means searching an enormous configurational space — far too large for first-principles methods alone. This work develops a machine-learned interatomic potential with near-DFT accuracy and couples it to a physically informed structure-generation scheme, prioritizing the undercoordinated sites where degradation initiates. The result is a framework that efficiently maps size- and morphology-dependent degradation pathways and consistently recovers the thermodynamically favored end states — generalizable beyond degradation to growth and restructuring.

Discovering surface structures: grand-canonical Monte Carlo

Master's thesis · expected Dec 2026 GCMC + MLIP
ordered termination GCMC reconstructed · hydroxylated ensemble
Grand-canonical Monte Carlo samples surface composition under fixed electrochemical conditions, recovering reconstructed and hydroxylated motifs without human-selected starting guesses.

My master's thesis brings the previous two threads together. Most IrO₂ surface modeling still starts from a handful of human-chosen terminations — clean, O-terminated, OH-terminated — which bakes in selection bias. Instead, I'm building a grand-canonical Monte Carlo workflow coupled to a machine-learned interatomic potential that samples surface composition under fixed chemical potentials, producing an ensemble of reconstructed, hydroxylated, and dissolution-prone surface structures as a function of potential and pH. The goal is a grand-canonical surface phase map of rutile IrO₂ that moves beyond single lowest-energy terminations toward a realistic, ensemble-based picture of reactive surfaces.

Working title: Grand Canonical Monte Carlo on Rutile IrO₂ Surfaces under Electrochemical Conditions using Machine-Learned Interatomic Potentials. Advised by Dr. Aleksandra Vojvodic.

Applied Project · Water & Energy

Recovering water from industrial cooling towers

Separate from my materials research, I spent a year on a team tackling evaporative water loss in mechanical-draft cooling towers — a major drain on freshwater in arid regions. We designed a dual-retrofit strategy: upstream sensible cooling to reduce how much water must evaporate, plus downstream vapor recovery to condense and reclaim part of the humid exhaust, all compatible with existing plant infrastructure.

14.3% condensation efficiency from the best hybrid slab–mesh recovery surface — validated at bench scale, with recovery rising as inlet temperature dropped.

Crucially, this was a research and techno-economic project, not just a proof of concept. We ran feasibility, sensitivity, and market analyses, then built a staged, economically grounded deployment plan for El Paso Electric's Newman Station — a pilot-first, cell-by-cell retrofit designed to fit a real outage window and scale across inland, water-stressed power plants in the Southwest, including Las Cruces Utilities sites.

PHOTO
mock cooling tower / bench setup
PHOTO
WERC team · 1st place, New Mexico

1st place overall at the WERC Environmental Design Contest (New Mexico State University, April 2025), $3,000.

Looking Ahead

Where I want to go

In a PhD, I want to help develop the theory that connects AI to the physics of materials discovery — especially where the material landscape is complex but the technological need is urgent: longer-lasting batteries, stable solar materials, better materials for wind and grid infrastructure, and structural and plasma-facing materials for nuclear fusion. These problems look different but share one bottleneck — configurational spaces too vast to explore by intuition or experiment alone — and none will be solved by a single field working in isolation.

Talks & Conferences
Awards & Grants
John M. Vohs Director's Award2026$3,500
WERC Environmental Design Contest — 1st place overallNew Mexico State University · team award · 2025$3,000
Water Center Student Innovation Awardteam award · 2025$2,000
Kleinman Center for Energy Policy Research Grant2025$900
AIChE Undergraduate Research — 1st place (Boston '25), 2nd place (San Diego '24)Catalysis & Reaction Engineering
Publications
Direct Observation of Collective Dissolution Mechanisms in Iridium Oxide Nanocrystals
S. A. Vigil, R. Thatcher, J. Nicolas, Z. Lin, D. Intriago, M. Fratarcangeli, M. C. Huang, A. I. Kankanamge, A. Vojvodic, I. A. Moreno-Hernandez
Journal of the American Chemical Society, 2026, 148, 7102–7112
Predicting Degradation and Morphological Changes in Rutile Iridium Oxide Nanoparticles via Machine-Learning Accelerated Sampling
J. Nicolas, R. Thatcher, M. Huang, A. Vojvodic
In preparation

ORCID 0009-0000-9635-6826  ·  Google Scholar ↗