Using Genetic Algorithm Study On The Structural Properties Of The Metal Surface Nanoclusters

As the most popular functional unit of nanodevices, supported nanoclusters plays a significant role in the nanotechnology. As the player of many physical and chemical processes on surfaces, such as growth, self-diffusion, nucleation, and surface catalysis, supported nanoclusters is always one of the hottest topics in surface science. Among the many properties of supported nanoclusters, the structural information is one of the most important aspects that people interest in. However, to our knowledge, the researches reported so far are either on small clusters containing only several atoms or islands with hundreds to thousands of atoms. For the supported cluster with medium size, which is believed to be the most promising materials for the next generation of microelectronics and ultra-high-density recording, it is rarely discussed. In this thesis, we just focus on the structure of medium size of clusters.In Chapter one, we take a brief review on the history of supported nanoclusters, and introduce some popular experimental and theoretical methods in the study of supported nanoclusters. We also take a view on the key results about the supported nanoclusters gained in the past years, and demonstrate some problems needed to be solved. In Chapter two, we describe the calculation model and method we employed, including the genetic algorithm and embedded-atom method.In Chapter three, the lower-energy structures (LESs) of nanoclusters supported on fcc(111) and fcc(110) surfaces are systematically studied. The structural feature, energy distribution, and numbers of LESs are investigated with the changing of surfaces and cluster sizes. A novel theoretical model is developed to explain the properties of LESs by considering atomic interactions. We found the lower-energy structures of clusters supported on fcc(111) surfaces are mainly determined by the nearest-neighbor (NN) adatom-adatom interactions and the adatom-substrate interactions. The next nearest-neighbor (NNN) adatom-adatom interactions only play a slight influence on the energy distribution of LESs. If the adatom-substrate interaction is not sensitive to the surrounding of edge atoms in clusters, there is only one type of structures in the LES group, i.e., the structures with the maximum number of NN bonds, and all of this type of structures are included in the group. Otherwise, i.e., the adatom-substrate interaction is not sensitive to the surrounding of edge atoms, the type of structures in the LES group will change with the increasing of cluster size: When the cluster is small, only the structures maximizing the number of NN bonds can be included; When the cluster size increases, types of structures with less NN bonds are energetically preferred by the LES group. Meanwhile, the clusters always keep a sharp shape, i.e., close to be triangular. On the later kinds of surfaces, i.e., the surface where the adatom-substrate interaction is not sensitive to the surrounding of edge atoms, we found an interesting structural refreshment which means the structure with less NN bonds replaces the one with more NN bonds and then enters into the LES group, and the later kind of structures is gradually excluded from the LES group with cluster size increase.On fcc(110) surfaces, however, we found the lower-energy structure (LES) is very sensitive to the next nearest-neighbor (NNN) adatom-adatom interaction. If the effective NNN adatom-adatom interaction is repulsive, only one type of structures can be included in the LES group, i.e., the linear chain structure. Otherwise, the LES will change its structural type when cluster size increases, through the structural replacement similar to the one on fcc(111) surfaces.On the other hand, based on our discussion on the lower-energy structures, we obtain the equilibrium shape of two-dimensional islands on both fcc(111) and fcc(110) surfaces. Our result is in a good agreement with the experimental observation. We also present our explanation on the natural reconstruction on fcc(110) surfaces by considering atomic interactions.In Chapter four, we investigate the magic number behavior in nanoclusters supported on fcc(001) surfaces. We found that on Au(001) surface, the magic number sequence is always kept while that on Cu(001) surfaces is gradually disappeared with cluster size increase. Based on the novel theoretical model we developed, we surprisingly found that such different behaviors are triggered by the weak next nearest-neighbor (NNN) adatom-adatom interactions. If the NNN adatom-adatom interaction is attractive, e.g., on Cu(001) surfaces, the closed square or quasi-square shell, which is the structure of magic number cluster and is enhanced in the much stronger nearest-neighbor interactions, will be gradually destabilized and eventually broken, thus the magic number sequence finally disappears at large size cluster limit. Interestingly, the magic number sequence is disappeared before the breaking of closed square or quasi-square shell. If the NNN adatom-adatom interaction is repulsive, e.g., on Au(001) surfaces, the closed square or quasi-square shell is always persisted and thus the magic number sequence is always kept. Besides, based on the theoretical model we developed, we obtain the equilibrium shape of two-dimensional islands on fcc(001) surfaces, and it is consistent with the experimental observation, which indicates our model is reliable and our simulation and analysis are reasonable.In the last Chapter, the conclusion of this thesis is presented and some prospects for the ongoing researches are also proposed.