Session L21: Emerging Trends in Molecular Dynamics Simulations and Data Analytics II

Overcoming the Time Limitation in Molecular Dynamics Simulation of Crystal Nucleation: A Persistent-Embryo Approach

The nucleation of a crystalline phase in liquid is a typical rare event and is usually inaccessible within the limited timescales of conventional molecular dynamics (MD) simulations under experimental conditions. In this talk, we present a “persistent embryo” method to facilitate crystal nucleation in MD simulations, in which a small crystal embryo is inserted into the liquid, with spring forces applied to keep the embryo from melting. The springs are gradually weakened as the embryo grows and are completed removed when the nucleus size remains significantly smaller than the critical size. In this way, one can observe the spontaneous fluctuations of a critical nucleus in an unbiased environment, without any assumptions on the shape of the critical nucleus. We applied this method to simulate crystal nucleation under realistic experimental conditions and obtained results that compare favorably with experiments and other simulation methods. We will show this accelerated dynamic approach provides ample statistics for the critical sampling of the nucleation.

  

TBD

tbd   

Cavitation in Water Induced by a SnO2 Nanoparticle and a Strong Electric Field

Molecular dynamic simulations are performed to examine the effect of an electric field on a system consisting of a SnO₂ nanoparticle embedded in water under ambient conditions. Cavitation is observed in the presence of uniform electric fields ranging from 0.042 to 0.25 V/Angstrom in both the SPCE and the Hydrogen-Bonding Polarizable (HBP) force field models for water. Over at least one order of magnitude, the cavity onset time t is related to the electric field through the Kohlrausch-Williams-Watts form E(t) =E₀exp(-(t/τ)^β , with b = d*/(d* + 2) = 3/7 where d* = d/2, d being the dimensionality of the system. The bubbles are found to rapidly collapse upon removal of the electric field. Results for the structure and dynamics of water along with the electric field distributions in the system will be presented.   

Interaction Potential for Faceted Nanoparticles.

There is a growing interest in studying the assembly of faceted nanocrystals due to the potential of such anisotropic particles to yield unconventional structures with unique electrical, optical, and mechanical properties. While the van der Waals (vdW) interactions between spherical particles can be well-described by analytical functions, no such closed-form expressions are available to describe the vdW interactions between faceted particles. This is especially problematic in molecular simulations of particle assembly, where exact descriptions of interparticle energies in terms of interatomic interactions become computationally prohibitive, requiring a polynomial time algorithm. In this work, we formulate analytical expressions for vdW interactions between faceted particle by taking advantage of available expressions for macroscopic body interactions and by converting the energy integration over the particles’ bodies to integration over the particles’ facets. We show that the expressions provide reasonably accurate descriptions of the position and orientation dependence of vdW interactions. Since this method is a constant time algorithm, O(1), it could improve the computational efficiency of simulating faceted particles by several orders of magnitude.   

Atomic Structure of Supported Metal Nano Clusters on MoS2 Monolayer Using Deep Potentials

In nanometer clusters (NCs), the specific arrangement of metal atoms
determines the unique size-dependent functionalities of the NCs and
hence their potential applications. Here we employ a combined approach
utilizing a genetic algorithm and a deep potential (DP)
to determine the lowest energy configurations of the NCs including
metastable structures. The DP is trained using density functional
theory calculations. We discuss the strengths and shortcomings of the
DP approach compared to a standard method utilizing density
functional theory calculations.   

Large-Scale Atomistic Simulations of Materials using SNAP Potentials

Molecular dynamics (MD) is a powerful materials simulation method whose accuracy is limited by the interatomic potential (IAP). SNAP is an automated quantum data-driven approach to IAP generation that balances accuracy and computational cost. The energy is formulated in terms of a very general set of geometric invariants that characterize the local neighborhood of each atom. The SNAP approach has been used to develop potentials for studying plasticity in tantalum, intrinsic defects in indium phosphide, and plasma surface interactions in tungsten and beryllium. In each case, large quantum-mechanical data sets of energy, force, and stress are accurately reproduced and cross-validation on additional test data is performed. The resultant potentials enable high-fidelity MD simulations with thousands to millions of atoms. The relatively large computational cost of SNAP is offset by the LAMMPS implementation, enabling the efficient use of large CPU and GPU clusters.   

A mean-field algorithm with decoherence and detailed balance for nonadiabatic molecular dynamics

Decoherence and detailed balance are two major issues for mixed quantum/classic nonadiabatic molecular dynamics (NAMD) simulations. In this work we introduce a new mean-field dynamics approach with decoherence and detailed balance (MF-DD) for NAMD. This method is able to explicitly treat the decoherence between different pairs of electronic states. Moreover, the energy-increasing and energy-decreasing electronic transitions are distinguished by dividing the density matrix into two parts. The detailed balance correction is then included by a Boltzmann factor applied to the energy-increasing transitions. The MF-DD is applied to study hot-hole cooling and transfer processes in Si quantum dot (QD) systems. The calculated hot carrier relaxation time is in consistent with experiments. In the QD-pair systems, the cooling time shows weak dependence with the QD spacing. However, the charge transfer rate between QDs is found to decreases exponentially as the QD spacing increases, which is attributed to the decreased state anticrossing strength. When the QD spacing is smaller than 1.1 nm, the hot-carrier transfer between two QDs can be quite efficient. It is also shown that the explicit treatment of decoherence time is important.   

Multi-Resolution Simulations using the Integral Equation Coarse-Graining Method

We use the variable-level coarse-grained (CG) description of polymer melts to obtain the effective CG potentials (ECGPs) for multi-resolution simulations of pure polymer melts represented at various CG resolutions. Starting from the Integral Equation Coarse-Graining approach, we obtain the numerical and analytical ECGPs in the multi-resolution simulations that are different from the ECGPs in single-resolution simulations. Therefore, the composition, temperature, and density dependences of such ECGPs can be investigated. In particular, the ECGPs between the polymer melts, represented by n_bα blobs, decay slowly with a long tail characteristic scaling exponent of N_bα^1/4(φ_α+ (1-φ_α) (n_bα/n_bβ)³))^1/4, where φ_α is the volume mole fraction of species α, N_bα is the number of monomers in a given blob of type α, and β shows the other species. The ECGPs allow one to avoid any hybrid region in multi-resolution simulations while quantitatively producing the structural and thermodynamical properties of the related atomistic systems such as radial distribution function and pressure.   

Development of a universal Electron Force Field

In many cases, the unavailability of an adequate classical force field is a major barrier for molecular dynamics simulations. In these cases, density functional theory (DFT) provides a universal but expensive option. We propose an intermediate approach based on an ‘electron force field’ (eFF) [Su and Goddard, PRL 99, 185003 (2007)]. Inspired by VSEPR theory, we expand the electron density as a sum of overlapping spherical electron ‘balls’ (e-balls). The e-balls interact through electrostatic forces, Pauli repulsion, and exchange-correlation. The electronic total energy is calculated as a universal function of the e-ball positions and widths. As in DFT, it is independent of the nuclei, which interact only electrostatically. We parameterize this universal, many-body function by fitting to DFT calculations of a large number of molecules and solids at equilibrium, distorted, and reaction geometries.   

Modeling of La3+ doping segregation in nanocrystalline yttria-stabilized zirconia using a combintaion of atomistic MD, Monte Carlo and Nudged Elatic Band calculations

The effect of La3+ doping on the structure and ionic conductivity change in nanocrystalline yttriastabilized zirconia (YSZ) was studied using a combination of Monte Carlo and molecular dynamics and Nudge Elastic Band simulations. Simulations of specific grain boundary configurations are developed. Systems with and withoout La doping are studied and equilibrated using a combintaion of techniques and eventually anaylzed using Voronoi tesselation analysis for the density of the dopants.

The simulation revealed the segregation of La3+ at eight tilt grain boundary (GB) structures and predicted an average grain boundary (GB) energy decrease of 0.25 J m-2, which is close to the experimental values reported in the literature. Cation stabilization was found to be the main reason for the GB energy decrease, and energy fluctuations near the grain boundary are smoothed out with La3+ segregation. Both dynamic and energetic analysis on the S13(510)/[001] GB structure revealed La3+ doping hinders O2- diffusion in the GB region, where the diffusion coefficient monotonically decreases with increasing La3+ doping concentration. T   

Ultrafast detonation of hydrazoic acid: a case study of the ChIMES model

Understanding the chemical evolution and states of matter of an energetic material under detonation is challenging due to the short time scales of chemical reactions and risk of experimental work. First-principle molecular dynamics simulations can provide valuable insights into such systems. However, the computational cost associated with those simulations limits their applicability to relatively small systems and short time scales. We have developed the Chebyshev Interaction Model for Efficient Simulation (ChIMES), a reactive force field, using force matching to trajectories from density functional theory (DFT). This force field has been shown to be capable of retaining the accuracy of DFT simulation while increasing orders of magnitude in computational efficiency. In this work, we use ChIMES to study hydrazoic acid, an azide energetic material that exhibits an ultrafast detonation during a shock wave. We find that our models are able to accurately reproduce the structural and dynamic properties computed from DFT at different thermodynamic states. The ability to describe charge transfer and chemical reactions is also discussed.
This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.