1. Experiments on low-temperature thin-film growth carried out by Stoldt et al [PRL 2000] indicate that the surface roughness exhibits a complex temperature dependence. While this behavior may be partially explained by the existence of “restricted downward funneling” (RDF) for depositing atoms, the low- temperature behavior was not understood. Accordingly, we have carried out parallel temperature-accelerated dynamics (parTAD) simulations in order to determine the barriers for a variety of intralayer and interlayer diffusion processes. Based on these results we then carried out hybrid molecular dynamics (MD) - kinetic Monte Carlo (KMC) simulations over the temperature range 55 - 180 K. Our results indicate that the complex experimental behavior is due to a competition between a variety of low-barrier processes including edge-zipping, RDF, edge-diffusion, and low-barrier processes for interlayer diffusion, while the short-range attraction of depositing atoms to the substrate also plays a role. Simulating Non-equilibrium Processes over Extended Time- and Length-Scales using Parallel Accelerated Dynamics Jacques G. Amar, University of Toledo, DMR 0907399 Comparison between experimental results for temperature-dependence of surface roughness for Ag/Ag(100) and hybrid MD-KMC simulations. Red arrows indicate “diffusion mechanisms which “turn-on” with increasing temperature T, including edge-zipping, RDF, interlayer diffusion at kinks, and edge-diffusion. Y. Shim and J.G. Amar, Phys. Rev. B 81, 045416 (2010).

2. Simulating Non-equilibrium Processes over Extended Time- and Length-Scales using Parallel Accelerated Dynamics Jacques G. Amar, University of Toledo, DMR 0907399 temperature in TAD simulations in order to maximize their efficiency, while another approach involves the use of localized saddle- point searches. In a third approach we have developed a method for carrying out parallel bond-boost hyperdynamics simulations. In order to further understand Ag/Ag(100) growth, we have also carried out density functional theory calculations of key relaxation mechanisms, and found good agreement with our previous results obtained using empirical potentials. Education: A postdoctoral research associate, Yunsic Shim, and three graduate students, John Royston (M.S. 2009), Giridhar Nandipati (Ph.D. 2009), and Bradley Hubartt have been involved in this research, along with a part-time temporary research associate, Valery Borovikov. While kinetic Monte Carlo (KMC) is an extremely efficient method to carry out non-equilibrium simulations of dynamical processes when the relevant rates are known, in some cases - such as when the relevant processes have a wide range of activation barriers - much of the simulation time can be wasted on repetitive low-barrier events. To address this problem we have recently developed a first-passage-time (FPT) approach to accelerate KMC simulations of metal(100) epitaxial growth with fast edge diffusion. Using this method, we have obtained a speed-up of more than an order of magnitude over regular KMC simulations. More recently, this method has also been extended in order to calculate the full FPT distribution so that fluctuations may be taken into account. (G. Nandipati, Y. Shim, and J.G. Amar, Phys. Rev. B 81, 235415 (2010)) We have recently also developed several methods to extend the time- and length-scales of accelerated dynamics simulations. One approach involves the use of an adaptive method to optimize the high-