Prediction of the structures of free and oxide-supported nanoparticles by means of atomistic approaches: the benchmark case of nickel clusters

The structures of Ni/MgO nanoparticles are studied by means of global optimization searches. The results from four different model potentials, sharing the same functional forms but different parametrizations, are reported and compared. Two parametrizations over four give qualitatively correct results, and one of them is also quantitatively satisfactory. The other models fail to explain some qualitative features observed in the experiments, such as the formation of hcp nanodots at small sizes or the transition to fcc structures at large sizes. The important features that an atomistic potential must present for the correct prediction of Ni cluster structures are discussed and generalized.