Structure–Activity Map of Ceria Nanoparticles, Nanocubes, and Mesoporous Architectures

Structure–activity mapping is central to the exploitation and optimization of nanomaterial catalysts in a variety of technologically important heterogeneous reactions, such as automotive catalysis and water gas shift reactions. Here, we present a catalytic activity map for nanoceria, calculated as a function of shape, size, architecture, and defect content, using atom-level models. The activity map reveals that as oxygen is gradually depleted from the nanoceria catalyst, so it becomes energetically more difficult to extract further oxygen. We propose that the oxygen storage capacity (OSC) of ceria corresponds to the level of oxygen depletion, where it becomes thermodynamically prohibitive to extract further oxygen from the material (positive free energy). Moreover, because the reaction enthalpy contributes to the free energy, we predict that the OSC is influenced by the particular reaction being performed. Specifically, the more negative the reaction enthalpy, the higher the potential OSC (notwithstanding entropic contributions). The decrease in catalytic activity during an oxidation reactionemanating from the increase in energy required to extract oxygensuggests that there exists a “window of catalytic operation”, where the activity of the catalyst can be controlled by operating at different points within this window. We show experimentally how the activity can be modified by engineering the oxygen vacancy concentration and hence the oxygen content of the catalyst to facility tunable activity. In addition to the defect content, we find that size (particle diameter, mesoporous wall thickness) and nanostructuring (particle, cube, mesoporous architecture, morphology, and surfaces exposed) are key drivers of catalytic activity. To generate the atom-level models of ceria nanostructures, we use nonequilibrium molecular dynamics to simulate the self-assembly of mesoporous ceria from amorphous nanobuilding blocks, followed by a (simulated) crystallization step; the latter evolves the crystal structure and microstructural features such as grain-boundaries and dislocations. Our simulated crystallizations emanate wholly from a multitude of “random” atom collisions, which result in the spontaneous evolution of a crystalline seed that nucleates crystallization of the whole system. The atomistic models generated by “simulating synthesis” are shown to be in quantitative structural agreement with experiment