Simple pair-wise interactions for hybrid Monte Carlo–molecular dynamics simulations of titania/yttria-doped iron

Abstract

We present pair-wise, charge-neutral model potentials for an iron-titanium-yttrium-oxygen system. These simple models are designed to provide a tractable method of simulating nanostructured ferritic alloys (NFAs) using off-lattice Monte Carlo and molecular dynamics techniques without deviating significantly from the formalism employed in existing Monte Carlo simulations. The model is fitted to diamagnetic density functional theory (DFT) calculations of the various species over a range of densities and concentrations. The resulting model potentials provide reasonable and in some cases even excellent mechanical and thermodynamic properties for the pure metals. The model replicates the qualitative trends in formation energy predicted by DFT, though the energies of formation do not agree as well for dilute systems as they do for more concentrated systems. We find that on-lattice models will consistently favor tetrahedral oxygen interstitial sites over octahedral interstitial sites, while relaxed systems typically favor octahedral sites. This emphasizes the need for the off-lattice simulations for which this potential was designed.

Molecular dynamics simulations of cascade damage near the Y 2 Ti 2 O 7 nanocluster/ferrite interface in nanostructured ferritic alloys

“…However, they are very complicated to be modeled in the cascade simulations. Here, we adopt the simplified interactions, i.e., a charge-neutral Buckingham potential [28] is used to describe Fe-O interactions, while for the case of Fe-Y and Fe-Ti, because their interactions are complex (neither pure ionic nor metallic interactions), here only the ZBL potentials are used to describe Fe-Y and Fe-Ti interactions for displacement cascades. Similar simplified potentials are also adopted for cascade simulations in the ferrite/Y 2 O 3 nanocluster interface in ODS steels, [18] where only the ZBL potential is used to describe Fe-Y interaction.…”

Neural Networks for Modeling Granger Causality

“…The HMC technique is a Markov chain process and consists of dynamic moves which apply the Hamiltonian dynamics and permit sampling states with constant total energy and stochastic moves which permit the technique to sample states with different total energy. Hammond et al (2013) successfully applied the HMC to simulate titania/yttriadoped iron whereas Zhou et al (2013) successfully applied the HMC sampling to identify a support vector machine model and implemented these to simulated data. Zhou et al (2013) successfully applied the HMC to estimate digital planar curves whereas Bretz et al (2012) applied the HMC to study ultra-high energy cosmic rays from their sources to the observer.…”