Elemental phosphorus has a striking variety of allotropes, which we analyze by looking at stable phosphorus clusters. We determine the ground-state structures of Pn clusters in a wide range of compositions (n = 2–50) using density functional calculations and global optimization techniques. We explain why the high-energy white phosphorus is so easily formed, compared to the much more stable allotropes – the tetrahedral P4 cluster is so much more stable than nearby compositions that only by increasing the size to P10 one can get a more stable non-P4-based structure. Starting from 17 atoms, phosphorus clusters have a single-stranded structure, consisting of a set of well-resolved structural units connected by P2 linking fragments. The investigation of relative stability has revealed even-odd alternations and structural magic numbers. The former are caused by the higher stability of clusters with even numbers of atoms due to closed electronic shells. The structural magic numbers are associated with the presence of particular stable structural units and lead to enhanced stability of P18+12k (k = 0, 1, 2) clusters. We also compare the energies of the obtained ground-state structures with clusters of different phosphorus allotropes. Clusters of fibrous phosphorus are energetically the closest to the ground states, white phosphorus clusters are found to be less stable, and the least stable allotrope at the nanocluster scale is black phosphorene.
Systematic structure prediction of CunAum nanoclusters was carried out for a wide compositional area (n + m ≤ 15) using the evolutionary algorithm USPEX and DFT calculations. The obtained structural data allowed us to assess the local stability of clusters and their suitability for catalysis of CO oxidation. Using these two criteria, we selected several most promising clusters for an accurate study of their catalytic properties. The adsorption energies of reagents, reaction paths, and activation energies were calculated. We found several cases with low activation energies and explained these cases using the patterns of structural change at the moment of CO2 desorption. The unique case is the Cu7Au6 cluster, which has extremely low activation energies for all transition states (below 0.05 eV). We thus showed that higher flexibility due to the binary nature of nanoclusters makes it possible to achieve the maximum catalytic activity. Considering the lower price of copper, Cu–Au nanoparticles are a promising new family of catalysts.