Atomistic studies of thin film growth

Abstract

We present here a summary of some recent techniques used for atomistic studies of thin film growth and morphological evolution. Specific attention is given to a new kinetic Monte Carlo technique in which the usage of unique labeling schemes of the environment of the diffusing entity allows the development of a closed data base of 49 single atom diffusion processes for periphery motion. The activation energy barriers and diffusion paths are calculated using reliable manybody interatomic potentials. The application of the technique to the diffusion of 2-dimensional Cu clusters on Cu(111) shows interesting trends in the diffusion rate and in the frequencies of the microscopic mechanisms which are responsible for the motion of the clusters, as a function of cluster size and temperature. The results are compared with those obtained from yet another novel kinetic Monte Carlo technique in which an open data base of the energetics and diffusion paths of microscopic processes is continuously updated as needed. Comparisons are made with experimental data where available.

Self-learning kinetic Monte Carlo method: Application to Cu(111)

“…The recent version of the code allows inclusion of simultaneous displacements for atoms up to the third shell. from our simulations of several types of local environments (straight steps with kinks, compact islands, fractal like islands) we found that in some cases many particle processes play important role in providing atomic transport [28]. They are especially important in the case of low coordination systems, like fractal islands.…”

“…[NEW] Although these results have already appeared in a conference proceeding [28], we present in Table 3 the frequencies of the various contributing atomistic processes. Included in the table are also the frequencies of events that were found in a parallel simulation carried out using the standard KMC simulation with a predetermined catalogue of 294 (49x6) processes involving single-atom peripheral diffusion /citegho03.…”

Diffusion of small two-dimensional Cu islands on Cu(111) studied with a kinetic Monte Carlo method

“…For the larger islands, the diffusion coefficients appear to scale as a function of the size and the scaling exponent is expected to reflect the intervening atomistic processes responsible for the diffusion. 20,26 However, for the smaller islands a consistent knowledge of the variation of their mobility with size and the details of the responsible atomistic processes has not yet been fully established, especially on the ͑111͒ surfaces of fcc metals.…”

The Monte Carlo method in science and engineering

“…In addition, this method can lead to a substantial gain in accuracy because of the inclusion of many-particle processes. Recently, self-learning KMC has been used to study the diffusion of Cu clusters on the Cu(111) surface, 45,46 for which there are many possible types of events and energy barriers, and significant differences between the results obtained using selflearning KMC and ordinary KMC with a restricted rate catalog were observed.…”

“…However, a new "selflearning" KMC method has recently been developed. 45,46 In this method, a precalculated list of diffusion processes and their associated energetics and transition rates isn't needed-rather, at any time during the simulation, the activation energies for all possible single or multi-atom processes are either computed on-the-fly using a saddle-point search procedure, or retrieved from a database in which previously encountered processes are stored. Initially, this method may require an extensive amount of work to calculate the possible events that might occur in the system.…”