Industrial catalysis, separations, and electrochemical technologies rely on functional materials whose performance is governed by elusive atomic-scale phenomena, such as bond rearrangements, ion migration, and framework breathing. Traditional density-functional theory (DFT) with harmonic approximations often fails to capture these events. To resolve them in silico, I combine machine-learning interatomic potentials (MLIPs) with rare-event sampling algorithms. This combined approach delivers quantum-accurate energetics and fully anharmonic kinetics on the time and length scales at which functional materials truly operate. In turn, it transforms qualitative intuition into quantitative design principles for next-generation catalysts, adsorbents, and electrochemical interfaces.
Representative Applications
| Functional material | Mechanistic understanding unlocked with MLIPs + rare-event sampling | Industrial relevance |
|---|---|---|
| SALEM-2 (MOF) |
Reveals transient metal–ligand dissociation that governs solvent-assisted linker exchange (SALE). |
Guides post-synthetic modification of MOF adsorbents and catalysts. |
| Cs-RHO (Zeolite) |
Captures water-driven cation migration and breathing that amplify CO₂ uptake under humid conditions. |
Enables design of robust adsorbents for flue-gas and direct-air capture. |
| CO2-reduction (Molecular Catalysis) | Quantifies ion-specific solvation that retunes activation barriers at electrified interfaces. |
Informs electrolyte and ligand selection for efficient electrochemical reactions. |
| Ag(111) (Surface Catalysis) |
Provides anharmonic desorption barriers and sticking coefficients essential for accurate microkinetic models. |
Improves kinetic modelling for partial methanol oxidation and related oxidation reactions. |
Impact
My modelling toolkit collapses months of trial-and-error into days by supplying reactor-ready energetics, kinetics, and transport coefficients without the prohibitive cost of exhaustive DFT or high-throughput synthesis. Because each model is rooted in first-principles data yet accelerated by machine learning, entire families of catalysts or adsorbents can be virtually screened before a single experiment is run, cutting R&D cycles and material waste. My dual background in experimentation and computation ensures that the simulations pose laboratory-relevant questions and output metrics—activation barriers, sticking coefficients, and uptake capacities—that chemists and engineers can integrate directly into process simulators or digital twins. This data-centric approach is readily transferrable to batteries, membranes, and other emerging functional materials, positioning me to drive AI-enabled research that de-risks scale-up and speeds innovative products to market across the chemicals and energy sectors.