Abstract

While a variety of methods have been developed to carry out atomistic simulations of thinfilm growth at small deposition angles with respect to the substrate normal, due to the complex morphology as well as the existence of multiple scattering of depositing atoms by the growing thin-film, realistically modeling the deposition process for large deposition angles can be quite challenging. Accordingly, we have developed a computationally efficient method based on the use of a single graphical processing unit (GPU) to carry out molecular dynamics (MD) simulations of the deposition and growth of thin-films via glancing angle deposition. Using this method we have carried out large-scale MD simulations, based on an embedded-atom-method potential, of Cu/Cu(100) growth up to 20 monolayers (ML) for deposition angles ranging from 50 • to 85 • and for both random and fixed azimuthal angles. A variety of quantities including the porosity, roughness, lateral correlation length, average grain size, strain, and defect concentration are used to characterize the thin-film morphology. For large deposition angles (θ ≥ 80 o ) we find well-defined columnar growth while for smaller angles, columnar growth has not yet set in. In addition, for θ = 70 o − 85 o the thinfilm porosity and columnar tilt angles (for fixed azimuthal angle φ) are in reasonable agreement with experiments. We also find that for both random and fixed φ the average strain is initially compressive but becomes tensile after the onset of columnar growth, in good qualitative agreement with recent experimental observations. Our results also indicate that for large deposition angles a variety of complex dynamical processes including coalescence, large-scale collective events, and budding play a key role in determining the evolution of the surface morphology and microstructure.