Using light, or an applied voltage, to drive a chemical reaction at a surface – most often splitting water to make hydrogen. Thin materials are natural candidates because nearly every atom is a surface atom, and because their edges are usually far more active than their flat faces.
Going deeper
In MoS2, hydrogen evolution happens at the edges, and activity scales with the number of edge sites rather than with area. A photocatalyst for water splitting needs band edges that straddle both redox levels, 1.23 eV apart, with extra margin for the overpotentials of real reactions.
Two ways to drive a reaction at a surface
In electrocatalysis an electrode potential supplies the energy for a reaction such as hydrogen evolution, oxygen evolution or CO2 reduction. A good catalyst reaches a useful current at a small extra voltage beyond the thermodynamic value, the overpotential; commonly quoted figures of merit are the overpotential needed for 10 mA per cm2, the Tafel slope, which describes how steeply current rises with voltage, and stability over many hours.
In photocatalysis a absorbs light instead. Electrons in its reduce, holes in its valence band oxidise. Splitting water requires 1.23 eV thermodynamically, so the band gap must be larger, band edges must straddle both redox levels, and in practice the gap needs to be substantially wider to cover overpotentials and losses.
Edges, not faces
In 2007, MoS2 nanoparticles were grown on gold, their edge sites counted atom by atom with , and their hydrogen evolution measured in solution: activity rose linearly with the number of edge sites, not with area. Calculations had already predicted that hydrogen binds to the metal edges with almost zero free energy, like platinum, placing them near the top of the volcano plot, where catalysts bind the intermediate neither too strongly nor too weakly.
The result set the strategy for layered catalysts: expose more edges with small or vertically aligned , activate the flat basal plane with and , or convert it to the phase. Related ideas extend to for electrocatalysis and to graphitic carbon nitride as a photocatalyst.
Comparing catalysts fairly
Current per geometric area rewards rough, high-surface-area electrodes. Meaningful comparisons normalise to the electrochemically active area or, better, report turnover per active site, and state the loading, the electrolyte, the reference electrode and any correction for resistance. Faradaic efficiency confirms that the current produced the claimed product. Trace platinum dissolving from a counter electrode can deposit on the working electrode and fake high activity.
Photocatalysis has its own traps. Sacrificial reagents that consume holes make hydrogen evolution look far better than overall water splitting, and rates per gram of catalyst depend on the lamp, the reactor and the light intensity. Apparent quantum yield at a stated wavelength and long-term stability make results comparable.
For specialists
Surface reactions driven by (photocatalysis) or by electrode potential (electrocatalysis), judged by overpotential, Tafel slope and turnover per site rather than by raw current. In MoS2 the basal plane is close to inert while the edge sites carry the hydrogen-evolution activity – the result that made layered catalyst candidates – so activity scales with edge length, defect density and phase, and any comparison needs an active-site count rather than a geometric area.