Multi-Timescale Accelerated Dynamics and Trajectory Analysis of Alloy Surface Transformations Using Novel Interatomic PotentialsGarza, Richard B (2023) Multi-Timescale Accelerated Dynamics and Trajectory Analysis of Alloy Surface Transformations Using Novel Interatomic Potentials. Master's Thesis, University of Pittsburgh. (Unpublished)
AbstractAlthough the equilibrium composition of many alloy surfaces is understood, the rate of transient surface segregation during annealing or oxidation is not known despite their crucial effects on alloy corrosion and catalytic reactions which occur on overlapping timescales. This thesis focuses on computational studies of the surface segregation of CuNi alloys, with methods and analyses generally applicable to all other bimetals. The text summarizes results for the CuNi(100) surface facet primarily in vacuum, with new interatomic potentials introduced to model interactions with oxygen optimized via evolutionary algorithm and deep learning. We first introduce the CuNi alloy, including its thermodynamically expected behavior in vacuum and oxygen environments. Molecular dynamics (MD) are used to simulate the dominant mechanisms which facilitate surface segregation of the binary alloy system. These preliminary findings are validated using first-principles and Monte Carlo annealing with the embedded-atom method (EAM) to find the equilibrium composition profile versus surface depth, a measurement that provides a defined endpoint for later transient simulations. Then, three accelerated methods are utilized to transiently evolve the system towards equilibrium: parallel trajectory splicing (ParSplice), adaptive kinetic Monte Carlo (AKMC), and kinetic Monte Carlo (KMC) from cluster expansion. From nanosecond to second timescales, this hierarchy of multiscale approaches can observe stochastic events not typically seen with standard MD, closing the gap between computational and experimental timescales for surface segregation and providing a timescale for vacuum segregation to occur. The final chapter presents two novel interatomic potentials for the Cu-Ni-O system which can be used within our hierarchy of multi-timescale methods to furnish realistic simulations of alloy oxidation: one reactive forcefield (ReaxFF) developed utilizing an evolutionary algorithm, and a larger set of parameters trained with deep learning tools available in the DeepMD kit. We assess the performance of both forcefields through comparison to DFT results for the Cu, Ni, CuO, NiO, and CuNiO bulk and surface-terminated systems. Share
Details
MetricsMonthly Views for the past 3 yearsPlum AnalyticsActions (login required)
|
![[feed]](/images/feed-icon-32x32.png){kind=icon}