Scientific Challenge

A good catalyst sits near a narrow operating point: the hydrogen adsorption free energy must fall close to thermoneutral, neither binding too strongly nor too weakly. On an ordered surface this is a single number, and the design problem is to tune it. But real catalysts — doped surfaces, high-entropy carbides and nitrides — are not ordered. No two adsorption sites share the same local environment, and the adsorption energy is not a single value but a distribution spread across inequivalent sites. The question that organises our work here is the one this distribution forces: in a compositionally disordered catalyst, what controls where the distribution of adsorption energies is centred, and how wide it is? Activity is no longer set by a representative site but by the statistics of all of them — and a composition is promising only if disorder places enough of that distribution near the thermoneutral window.

Our Approach

The foundation is a clean-descriptor analysis: identifying, on well-defined surfaces, which electronic-structure quantity actually controls adsorption, so that compositional trends can be read as mechanism rather than correlation. Factorial screening across metal pairs and anions isolates the descriptor doing the work, and locates the regime where a simple local descriptor stops being predictive.

From there the direction is statistical. Treating disorder as the object of study — through special quasirandom structures and ensemble calculations across many inequivalent sites — turns the single adsorption energy of the ordered picture into the distribution that a real catalyst actually presents. The aim is to predict not one site’s behaviour but the shape and position of the whole distribution, and to ask which compositions push it toward the catalytic optimum. In this framing, configurational disorder is not noise to be averaged away; it is the design parameter.

Key Insights & Achievements

• Site-resolved descriptors for catalysis. In bimetallic Janus MXenes, factorial screening across metal pairs and anions shows that site-resolved d-band asymmetry — not a sublattice-averaged descriptor — controls hydrogen adsorption near thermoneutrality, and identifies the regime, magnetic exchange splitting on one sublattice, where the local descriptor breaks down.
• Doping pathways to noble-metal-free catalysis. In layered transition-metal carbides and nitrides, targeted electronic-structure engineering through early-transition-metal doping brings hydrogen adsorption into the thermoneutral window without requiring platinum-group metals.

These results establish which electronic-structure quantity governs adsorption on a controlled surface, and where a clean local descriptor ceases to predict it. That boundary — the point at which a single-site picture fails — is exactly where configurational disorder takes over, and where the current direction of the work begins: describing catalytic activity as a statement about the statistics of a disordered surface rather than the properties of one representative site.

• ← Go back to Research