Light-driven chemical reactions on semiconductor surfaces have potential for
addressing energy and pollution needs through efficient chemical synthesis; however, little is known about the time evolution of excited states that determine reaction pathways. Here, we study the photo-oxidation of methoxy (CH₃O) derived from methanol on the rutile TiO₂(110) surface using ab initio simulations to create a molecular movie of the process. The movie sequence reveals a wealth of information on the reaction intermediates, time scales, and energetics. The reaction is broken in three stages, described by Lewis structures directly derived from the “hole” wave functions that lead to the concept of “photoinduced C−H acidity”. The insights gained from this work can be generalized to a set of simple rules that can predict the efficiency of photo-oxidation reactions in reactant−catalyst pairs.

Hydrogen production in photoelectrochemical cells constitutes an important avenue toward carbon-free fuel. The most convenient process for hydrogen production is the splitting of water molecules, which necessitates a catalytic
reaction involving a semiconductor. Here, we introduce a framework for the study of photocatalyzed reactions on semiconductor surfaces based on time-dependent density functional theory that explicitly accounts for the evolution of electronically excited states. Within this framework, we investigate the possibility of hole-mediated splitting of molecularly adsorbed water on a representative metal oxide surface the rutile TiO₂(110). We find that oxidative dehydrogenation of water is possible in synergy with thermal effects at temperatures between 60 and 100 K only when defects like Ti interstitials are present in the subsurface region. This study presents a general computational strategy for describing photoexcited semiconductor/adsorbate interfaces and also demonstrates that the occurrence of water dissociation on the rutile TiO₂(110) surface depends sensitively on the local atomic environment and external parameters such as temperature.