Effects of Self-Compression On Metal Clusters

Traditionally, periodic unit cells are used to model the surface properties of metal catalysts, even those that exist as nanoparticles. We are interested in finding more realistic ways of modeling these particles, in order to determine whether the traditional approach is valid at this length scale. We have used density functional theory to study the effects of relaxation on gold and platinum clusters, ranging from 13 to 923 atoms, as well as on their island analogs. Calculations of this size have been made possible by leadership computing resources, containing up to 40,960 nodes, at Argonne National Laboratory. We find that self-compression leads freestanding clusters to converge to weaker oxygen adsorption energies than the values calculated from traditional surface science slabs. The island systems proposed were able to capture this compression effect and gave better predictions of adsorption energies on clusters. However, the agreement between the two systems appeared only after quantum-size-effects ended. We find that below the critical size of 561 atoms on gold, and 147 on platinum, electronic features on freestanding clusters cannot be represented by continuous systems.