Publications
Atomically engineering Cu/Ta interfaces
This report summarizes the major research and development accomplishments for the late start LDRD project (investment area: Enable Predictive Simulation) entitled 'Atomically Engineering Cu/Ta Interfaces'. Two ultimate goals of the project are: (a) use atomistic simulation to explore important atomistic assembly mechanisms during growth of Cu/Ta multilayers; and (b) develop a non-continuum model that has sufficient fidelity and computational efficiency for use as a design tool. Chapters 2 and 3 are essentially two papers that address respectively these two goals. In chapter 2, molecular dynamics simulations were used to study the growth of Cu films on (010) bcc Ta and Cu{sub x}Ta{sub 1-x} alloy films on (111) fcc Cu. The results indicated that fcc crystalline Cu films with a (111) texture are always formed when Cu is grown on Ta. The Cu films are always polycrystalline even when the Ta substrate is single crystalline. These polycrystalline films are composed of grains with only two different orientations, which are separated by either orientational grain boundaries or misfit dislocations. Periodic misfit dislocations and stacking fault bands are observed. The Cu film surface roughness was found to decrease with increasing adatom energy. Due to a Cu surface segregation effect, the Cu{sub x}Ta{sub 1-x} films deposited on Cu always have a higher Cu composition than that used in the vapor mixture. When Cu and Ta compositions in the films are comparable, amorphous structures may form. The fundamental origins for all these phenomena have been studied in terms of crystallography and interatomic interactions. In chapter 3, a simplified computational method, diffusional Monte Carlo (dMC) method, was developed to address long time kinetic processes of materials. Long time kinetic processes usually involve material transport by diffusion. The corresponding microstructural evolution of materials can be analyzed by kinetic Monte Carlo simulation methods, which essentially simulate structural evolution by tracing each atomic jump. However, if the simulation is carried out at a high temperature, or a jump mechanism with a very low energy barrier is encountered, the jump frequency may approach the atom vibration frequency, and the computational efficiency of the kinetic Monte Carlo method rapidly decreases to that of a molecular dynamics simulation. The diffusional Monte Carlo method addresses the net effects of many atom jumps over a finite duration, kinetically controlled process. First, atom migration due to both random and non-random jumps is discussed. The concept of dMC is then introduced for random jump diffusion. The validity of the method is demonstrated using several diffusion cases in one-, two- and three-dimensional spaces, including the dissolution of spinodal structures. The application of the non-random diffusion theory to spinodal decomposition is also demonstrated.