The four most stable structures of ${\mathrm{Ni}}_N$ clusters with $N$ from 2 to 150 have been determined using a combination of the embedded-atom method in the version of Daw, Baskes, and Foiles, the variable metric/quasi-Newton method, and our own Aufbau/Abbau method. A systematic study of energetics, structure, growth, and stability of also larger clusters has been carried through without more or less severe assumptions on the initial geometries in the structure optimization, on the symmetry, or on bond lengths. It is shown that cluster growth is predominantly icosahedral with islands of fcc, tetrahedral and decahedral growth. For the first time in unbiased computations it is found that ${\mathrm{Ni}}_{147}$ is the multilayer (third Mackay) icosahedron. Further, we point to an enhanced ability of fcc clusters to compete with the icosahedral and decahedral structures in the vicinity of $N=79$. In addition, it is shown that conversion from the hcp/anti-Mackay kind of icosahedral growth to the fcc/Mackay one occurs within a transition layer including several cluster sizes. Moreover, we present and apply different analytical tools in studying structural and energetic properties of such a large class of clusters. These include means for identifying the overall shape, the occurrence of atomic shells, the similarity of the clusters with, e.g., fragments of the fcc crystal or of a large icosahedral cluster, and a way of analysing whether the $N$-atom cluster can be considered constructed from the $(N-1)$-atom one by adding an extra atom. In addition, we compare in detail with results from chemical-probe experiment. Maybe the most central result is that first for clusters with $N$ above 80 general trends can be identified.

We use the embedded-atom method (EAM) in the version of Daw, Baskes, and Foiles (DBF) to determine the three most stable isomers of ${\mathrm{Cu}}_N$ clusters with $N$ from 2 to 150. Randomly generated initial configurations are optimized with the variable metric/quasi-Newton method combined with our own Aufbau/Abbau algorithm for searching the global minima. A detailed comparison is made for clusters with up to 60 atoms obtained with the DBF and the Voter-Chen (VC) versions of the EAM, the many-body Gupta, and the Sutton-Chen 9-6 potentials. Although the two EAM potentials have completely different parametrizations, they yield clusters that are structurally and energetically almost identical. On the other hand, the Sutton-Chen potential strongly overestimates the binding energy of the dimer and the small copper clusters with up to 15 atoms, and therefore, yields clusters with shorter bond lengths. For DBF clusters with up to 150 atoms we analyze many structural and energetic properties such as the overall shape, the construction of atomic shells, the similarity of the clusters with fragments of the fcc crystal or of a large icosahedral cluster, and whether the $N$-atom cluster resembles the $(N-1)$-atom one with an extra atom added. The most stable clusters have high symmetry, such as the magic-sized ${\mathrm{Cu}}_{55}$ and ${\mathrm{Cu}}_{147}$ that are the second and third Mackay icosahedra, where the latter was obtained for the first time in a completely unbiased structure optimization. The cluster growth is predominantly icosahedral, with islands of fcc, tetrahedral, and decahedral growth.

We present results of density-functional calculations of structural and electronic properties of naked, stoichiometric zinc-blende-, wurtzite-, and rocksalt-derived CdSe clusters as a function of size of the cluster and of nonstoichiometric CdSe clusters with zinc-blende-derived structure. We have optimized the structure whereby our initial structures were taken as spherical parts of the crystals. The calculations were performed by using a parametrized linear combination of atomic orbitals–density-functional theory–local-density approximation scheme. Our results include the radial distribution of atoms and of Mulliken populations, the electronic energy levels, the band gap, and the stability as a function of size and composition.