Our
ability to predict nanocatalyst reactivity has been hindered
by our lack of atomic-scale understanding of nanocatalyst surface
structure. Do nanocatalyst surfaces adopt a bulk-terminated structure
or do they reconstruct to minimize their free energy, thereby lowering
their reactivity as often observed in vacuum? Similarly, do nanocatalysts
processed at high temperatures maintain their low reactivity, reconstructed
surfaces when used at low temperatures? Using a new technique for
the preparation of anatase nanocatalysts suitable for atomic-scale
imaging and surface spectroscopy, we show that solution-prepared anatase
is terminated by a monolayer of fluorine, which acts as an atomic-scale
oleophobic coating, preventing the accumulation of adventitious carbon.
We further show that the most common TiO2 functionalization
chemistry, a carboxylic acid solution, causes the spontaneous reorganization
of a reconstructed anatase nanocatalyst, leading to a five-fold increase
in reactive sites. This reorganization is not observed when carboxylic
acids are deposited from the gas phase, suggesting that model experiments
in vacuum environments can lead to a nonequilibrium, kinetically trapped
state that may not be catalytically relevant. Aqueous carboxylic acid
solutions produce densely packed carboxylate monolayers with richer
adsorption geometries than previously predicted. Ab initio simulations
show that although the carboxylate termination is somewhat less effective
at removing surface stress than the reconstruction, it is more effective
in lowering the surface energy. This observation suggests that bulk-terminated
metal-oxide nanocrystals may be common in reactive environments, even
if high temperatures are used to process the nanocatalyst or if the
reactant is later rinsed off. As such, the assumption of a bulk-terminated
surface may be a reasonable starting point for “materials-by-design”
approaches to computationally engineered nanocatalysts.