When BaZrO(3) is doped with Y in 12.5% of Zr sites, density functional theory with the PBE functional predicts octahedral distortions within a cubic phase yielding a greater variety of proton binding sites than undoped BaZrO(3). Proton binding sites, transition states, and normal modes are found and used to calculate transition state theory rate constants. The binding sites are used to represent vertices in a graph. The rate constants connecting binding sites are used to provide weights for graph edges. Vertex and color coding are used to find proton conduction pathways in BaZr(0.875)Y(0.125)O(3). Many similarly probable proton conduction pathways which can be periodically replicated to yield long range proton conduction are found. The average limiting barriers at 600 K for seven step and eight step periodic pathways are 0.29 and 0.30 eV, respectively. Inclusion of a lattice reorganization barrier raises these to 0.42 and 0.33 eV, respectively. The majority of the seven step pathways have an interoctahedral rate limiting step while the majority of the eight step pathways have an intraoctahedral rate limiting step. While the average limiting barrier of the seven step periodic pathway including a lattice reorganization barrier is closer to experiment, how to appropriately weight different length periodic pathways is not clear. Likely, conduction is influenced by combinations of different length pathways. Vertex and color coding provide useful ways of finding the wide variety of long range proton conduction pathways that contribute to long range proton conduction. They complement more traditional serial methods such as molecular dynamics and kinetic Monte Carlo.

Machine learning, trained on quantum mechanics (QM) calculations, is a powerful tool for modeling potential energy surfaces. A critical factor is the quality and diversity of the training dataset. Here we present a highly automated approach to dataset construction and demonstrate the method by building a potential for elemental aluminum (ANI-Al). In our active learning scheme, the ML potential under development is used to drive non-equilibrium molecular dynamics simulations with time-varying applied temperatures. Whenever a configuration is reached for which the ML uncertainty is large, new QM data is collected. The ML model is periodically retrained on all available QM data. The final ANI-Al potential makes very accurate predictions of radial distribution function in melt, liquid-solid coexistence curve, and crystal properties such as defect energies and barriers. We perform a 1.3M atom shock simulation and show that ANI-Al force predictions shine in their agreement with new reference DFT calculations.

No abstract

We describe a molecular dynamics framework for the direct calculation of the short-ranged structural forces underlying grain-boundary premelting and grain-coalescence in solidification. The method is applied in a comparative study of (i) a Σ9 115 120 o twist and (ii) a Σ9 110 {411} symmetric tilt boundary in a classical embedded-atom model of elemental Ni. Although both boundaries feature highly disordered structures near the melting point, the nature of the temperature dependence of the width of the disordered regions in these boundaries is qualitatively different.The former boundary displays behavior consistent with a logarithmically diverging premelted layer thickness as the melting temperature is approached from below, while the latter displays behavior featuring a finite grain-boundary width at the melting point. It is demonstrated that both types of behavior can be quantitatively described within a sharp-interface thermodynamic formalism involving a width-dependent interfacial free energy, referred to as the disjoining potential. The disjoining potential for boundary (i) is calculated to display a monotonic exponential dependence on width, while that of boundary (ii) features a weak attractive minimum. The results of this work are discussed in relation to recent simulation and theoretical studies of the thermodynamic forces underlying grain-boundary premelting.

Understanding the effect of grain boundaries (GBs) on the deformation and spall behavior is critical to designing materials with tailored failure responses under dynamic loading. This understanding is hampered by the lack of in situ imaging capability with the optimum spatial and temporal resolution during dynamic experiments, as well as by the scarcity of a systematic data set that correlates boundary structure to failure, especially in BCC metals. To fill in this gap in the current understanding, molecular dynamics simulations are performed on a set of 74 bi-crystals in Ta with a [110] symmetric tilt axis. Our results show a correlation between GB misorientation angle and spall strength and also highlight the importance of GB structure itself in determining the spall strength. Specifically, we find a direct correlation between the ability of the GB to plasticity deform through slip/twinning and its spall strength. Additionally, a change in the deformation mechanism from dislocation-meditated to twinning-dominated plasticity is observed as a function of misorientation angles, which results in lowered spall strengths for high-angle GBs.