Machine-learning-driven quantum chemically accurate calculations of the catalyst itself: which sites are active, how adsorbates bind, and how the surface reorganises under reaction conditions. The questions that set everything downstream are decided here — activity and selectivity descriptors, the reaction mechanism, and whether a material stays intact at operating potential.

• Adsorption and reaction energetics from chemically accurate quantum chemistry
• Implicit solvation and the reaction environment of the double layer
• Transient oxidation states and sub-surface oxygen
• Grain boundaries and surface reconstruction
• High-throughput screening for activity and selectivity descriptors

The region between catalyst and electrolyte, where the electric double layer, the local pH and the ion distribution together decide the rate. We model the interface by combining high-level quantum chemistry with multi-scale modeling techniques (many body expansion and machine learning potentials), deriving a predictive, atomistic picture of the interface and reaction processes.

• Electric double layer structure and surface charging
• Cation and field effects on reaction kinetics
• Local pH, chemical reactions and buffer equilibria
• Ion diffusion, migration and steric repulsion
• Proton-coupled electron transfer

Coupling the interface description to transport across a whole device: reactions, diffusion, migration and convection through a porous gas diffusion electrode. This is where a catalyst that looks good on paper meets the cell it has to work in, and where cell design turns out to govern which products come out.

• Porous gas diffusion electrodes and multi-phase transport
• Poisson–Nernst–Planck and electroneutrality models
• Kinetics, porosity and surface roughness
• Digital twins of reported experimental electrolysers
• Product selectivity across the catalyst layer