Empirical Interatomic Potentials Research Articles

Effect of surface reconstructions on misfit dislocation formation in InAs/GaAs(001)

The effect of surface reconstructions on misfit dislocation (MD) formation in InAs/GaAs(001) is theoretically investigated using empirical interatomic potentials. According to the calculated cohesive energy using empirical interatomic potentials, in the case of InAs/GaAs(001) system applied surface reconstruction, the system with MD of a 5/7-atom ring core structure is stable compared with a coherently grown system at 3.0 monolayer, which is larger than that of the ideal surface. This result reveals that the energy gain due to MD formation on the reconstructed surface is smaller than that of the ideal surface. The analysis of atomic configurations of the reconstructed surface with MD confirms that the strains caused by MD formation destabilize the reconstructed surface. Furthermore, we estimate the growth behavior of InAs/GaAs(001) using macroscopic theory and find that MD formation is suppressed by surface reconstruction. These results suggest that surface reconstruction is crucial for MD formation and the resultant growth behavior.

Effect of sub-surface hydrogen on intrinsic crack tip plasticity in aluminium

ABSTRACTThe effects of sub-surface hydrogen and mixed mode loading on dislocation emission in aluminium are studied using a combination of techniques including crack simulations with an empirical interatomic potential, generalised stacking fault energy (GSF) calculations, with empirical interactions and Density Functional Theory, and the model by Rice which links the critical stress intensity factor to the unstable stacking energy. The crack orientation is and the loading is composed of a moderate traction along and a shear along , such that Shockley partials are emitted along the crack plane. The role of the relaxations around the H atoms and of the concentration of H in the glide plane, in the GSF calculation, is revealed by comparing Rice's model to the results of brute force simulations. The enhanced GSF is then calculated ab initio. The conclusion is a large decrease of the critical load to emit a dislocation, due to the displacement transverse to the glide direction. The effect of sub-surface hydrogen is negligible with respect to the mechanical one.

Theoretical investigations on the structural stability and miscibility in BAlN and BGaN alloys: bond-order interatomic potential calculations

Structural stability and miscibility of BxAl1−xN and BxGa1−xN alloys including 3-fold coordinated hexagonal (Hex), 4-fold coordinated wurtzite (WZ) and zinc blende (ZB) structures are systematically investigated based on empirical bond-order potential (BOP), which incorporates electrostatic energies due to bond and ionic charges. We find that the Hex structure is stabilized for free-standing BxAl1−xN and BxGa1−xN alloys with boron composition while the ZB (WZ) structure is stabilized for () both in BxAl1−xN and BxGa1−xN alloys due to the electrostatic energy. Furthermore, the miscibility of BxAl1−xN and BxGa1−xN alloys with 4-fold coordinated structures are found to be higher than that with Hex structure owing to the chemical energy contribution to the excess energy. These results suggest that our empirical interatomic potentials approach is feasible to clarify structural stability not only for the 3-fold coordinated Hex structure but also 4-fold coordinated WZ and ZB structures in BxAl1−xN and BxGa1−xN alloys.

SIMPLE-NN: An efficient package for training and executing neural-network interatomic potentials

The molecular dynamics (MD) simulation is a favored method in materials science for understanding and predicting material properties from atomistic motions. In classical MD simulations, the interaction between atoms is described by an empirical interatomic potential, so the reliability of the simulation hinges on the accuracy of the underlying potential. Recently, machine learning (ML) based interatomic potentials are gaining attention as they can reproduce potential energy surfaces (PES) of abinitio calculations, with a much lower computational cost. Therefore, an efficient code for training ML potentials and inferencing PES in new configurations would widen the application range of MD simulations. Here, we announce an open-source package, SNU Interatomic Machine-learning PotentiaL packagE-version Neural Network (SIMPLE-NN) that generates and utilizes the ML potential based on the artificial neural network with the Behler–Parrinello type symmetry function as descriptors for the chemical environments. SIMPLE-NN uses the Atomic Simulation Environment (ASE) package and Google Tensorflow for high expandability and efficient training, and also supports the in-house code for quasi-Newton method. Notably, the package features a weighting scheme based on the Gaussian density function (GDF), which significantly improves accuracy and reliability of ML potentials by resolving sampling bias that exists in typical training sets. For MD simulations, SIMPLE-NN interfaces with the LAMMPS package. We demonstrate the performance and usage of SIMPLE-NN with examples of SiO2. Program summaryProgram Title: SIMPLE-NNProgram Files doi:http://dx.doi.org/10.17632/pjv2yr7pvr.1Licensing provisions: GPLv3Programming language: Python/C++Nature of problem: Inferencing the potential energy surface for the given system with accuracy comparable to ab initio methods but with much lower computational costs.Solution method: Calculate descriptor vectors that encode local chemical environment. High-dimensional neural network is used to predict the total energy from the descriptor vectors. The trained neural network can be used for molecular dynamics simulations.

Theoretical investigations on the growth mode of GaN thin films on an AlN(0001) substrate

The growth modes of GaN thin films on an AlN(0001) substrate are systematically investigated using our macroscopic theory with the aid of empirical interatomic potential and ab initio calculations. Using empirical interatomic potential calculations, we demonstrate that a GaN/AlN(0001) system with misfit dislocation (MD) is stabilized compared with a coherently grown system when the GaN film thickness exceeds six monolayers. According to the calculated results including the surface energy of GaN, the macroscopic free energy calculations for the growth mode imply that the growth of GaN on AlN(0001) proceeds along the lower-energy path from two-dimensional coherent (2D-coherent) to 2D-MD under Ga-rich conditions but from 2D-coherent to three-dimensional coherent with truncated hexagonal pyramid islands consisting of {} facets under N-rich conditions owing to surface energy anisotropy, which depends strongly on the growth conditions. These results suggest that surface energy anisotropy is crucial for the growth of GaN on an AlN(0001) substrate.

Novel nonlinear coarse-grained potentials of carbon nanotubes

Centimetres-long carbon nanotube (CNT) bundles with tensile strength over 80 GPa have been fabricated and tested recently [Nat. Nanotechnol. 13, 589–595 (2018)], but it is still a tremendous challenge to predict their nonlinear mechanical behaviors by full-atom molecular dynamics (MD) due to the huge computational cost, particularly for carbon nanotube networks. We completely established here the explicit expressions of the chirality-dependent higher-order nonlinear coarse-grained stretching and bending potentials based on the full-atom Reactive Empirical Bond-Order interatomic potential of second generation (REBO potential). In particular, the coarse-grained non-bonded potentials are improved by using the 18–24 Lennard-Jones potential. By comparison with available experimental results and full-atom MD simulations as well as our analytical results, the present nonlinear coarse-grained potentials have high accuracy. The obtained nonlinear coarse-grained potentials can be used to efficiently characterize the nonlinear mechanical behaviors and understand the failure mechanism of the CNT bundles and networks with 2∼5 orders of magnitude reduction in computing time, which should be of great help for designing and assembling CNT-based flexible microdevices.

Adsorption of 5-aminosalicylic acid on kaolinite surfaces at a molecular level

AbstractThe application of clay minerals in therapeutics is becoming important due to their structural and surface physicochemical properties. 5-aminosalicylic acid (5-ASA) is a very common pharmaceutical drug and is used worldwide. The interactions between the 5-ASA molecule and both the aluminol and siloxane surfaces of kaolinite are studied by means of atomistic calculations using force fields based on empirical interatomic potentials and quantum mechanics calculations based on density functional theory. A conformational analysis of 5-ASA has been performed and the anion of 5-ASA was also studied. The calculated adsorption energy values indicate that 5-ASA is likely to be adsorbed on the kaolinite surfaces with greater affinity to the aluminol surface. Hence, kaolinite may be considered as a promising pharmaceutical carrier of 5-ASA.

Growth mode in heteroepitaxial system from nano- and macro-theoretical viewpoints

Growth mode in semiconductor heteroepitaxial system such as InAs/GaAs is systematically investigated from nano- and macro-theoretical viewpoints, where ab initio, empirical interatomic potential, and phenomenological macroscopic theory are used. The nanoscopic theories clarify that the misfit dislocation (MD) formation energy depends on orientation such as 1.14 eV/Å for (0 0 1), 0.96 eV/Å for (1 1 0), and 0.68 eV/Å for (1 1 1)A. On the basis of these calculated results, growth mode boundary between two-dimensional with MD formation (2D-MD) and three-dimensional Stranski-Krastanov island (3D-SK) growth modes is successfully determined as functions of surface energy and degree of strain relaxation. The region of 3D-SK growth mode is the largest for (0 0 1) while (1 1 1)A has the largest 2D-MD region and that for (1 1 0) is in between. Using surface energy obtained by ab initio calculations and the degree of strain relaxation estimated by continuum elasticity theory, it is found that (0 0 1) favors 3D-SK growth mode while 2D-MD growth mode appears for (1 1 0) and (1 1 1)A. It should be noted that the (1 1 0) data stay near the growth mode boundary between 3D-SK and 2D-MD growth modes. Reflecting this, growth mode transition from 2D-MD to 3D-SK occurs for (1 1 0) due to In0.25Ga0.75As layer insertion at the interface that decreases surface energy from 50.9 meV/Å2 to 49.1 meV/Å2. Versatility of this approach is discussed by comparing with experimental findings.

Lubrication of dislocation glide in forsterite by Mg vacancies: Insights from Peierls-Nabarro modeling

Dislocation glide is an important contributor to the rheology of olivine under conditions of high stress and low to moderate temperature, such as occur in mantle wedges. Interactions between point defects and dislocation core may alter the Peierls stress, σp, and has been suggested that vacancy-related defects may selectively enhance glide on certain slip systems, changing the olivine deformation fabric. In this study, the Peierls-Nabarro model, parameterized by generalized stacking fault (GSF) energies calculated atomistically using empirical interatomic potentials, is used to determine the effect of bare Mg vacancies on the Peierls stresses of [1 0 0](0 1 0) and [0 0 1](0 1 0) dislocations in forsterite. Mg vacancies considerably reduce GSF energies and, consequently, σp for dislocations gliding on (0 1 0) in olivine. The magnitude of this decrease depends strongly on dislocation and the type of the lattice site, with vacant M2 sites producing the largest reduction of σp. The [0 0 1](0 1 0) slip system is found to be more sensitive than the [1 0 0](0 1 0) slip system to the presence of vacancies. Although, at ambient pressure, σp is lower for [1 0 0](0 1 0) than [0 0 1](0 1 0) edge dislocations, dσp/dP is greater for [1 0 0](0 1 0) dislocations, resulting in a change in the preferred slip system at 1.5 GPa. By preferentially lubricating [0 0 1](0 1 0) glide, Mg vacancies reduce the pressure at which this cross-over occurs. An M2 vacancy concentration at the glide plane of 0.125 defects/site is sufficient to reduce cross-over to 0.7 GPa. This may account for the existence of the B-type olivine deformation fabric in the corners of mantle wedges.

Screw dislocation structure and mobility in body centered cubic Fe predicted by a Gaussian Approximation Potential

The plastic flow behavior of bcc transition metals up to moderate temperatures is dominated by the thermally activated glide of screw dislocations, which in turn is determined by the atomic-scale screw dislocation core structure and the associated kink-pair nucleation mechanism for glide. Modeling complex plasticity phenomena requires the simulation of many atoms and interacting dislocations and defects. These sizes are beyond the scope of first-principles methods and thus require empirical interatomic potentials. Especially for the technological important case of bcc Fe, existing empirical interatomic potentials yield spurious behavior. Here, the structure and motion of the screw dislocations in Fe are studied using a new Gaussian Approximation Potential (GAP) for bcc Fe, which has been shown to reproduce the potential energy surface predicted by density-functional theory (DFT) and many associated properties. The Fe GAP predicts a compact, non-degenerate core structure, a single-hump Peierls potential, and glide on {110}, consistent with DFT results. The thermally activated motion at finite temperatures occurs by the expected kink-pair nucleation and propagation mechanism. The stress-dependent enthalpy barrier for screw motion, computed using the nudged-elastic-band method, follows closely a form predicted by standard theories with a zero-stress barrier of ~1 eV, close to the experimental value of 0.84 eV, and a Peierls stress of ~2 GPa consistent with DFT predictions of the Peierls potential.

Molecular-dynamics simulations of solid phase epitaxy in silicon: Effects of system size, simulation time, and ensemble

Solid phase epitaxial (SPE) recrystallization of amorphous Si on a Si(001) substrate was examined by large-scale (6144–129024 Si atoms), long-time (up to 2000 ns) molecular-dynamics (MD) simulations using the empirical Tersoff interatomic potential. We particularly focused on the effects of the MD cell size, simulation time, and ensemble on the SPE growth rate. We found that the simulations under the isothermal–isochoric conditions (NVT ensemble) show a higher crystallization rate than those under the isothermal–isobaric conditions (NPT ensemble). The system size dealt with in the present MD simulation, i.e., >6144 Si atoms, was enough to estimate the SPE growth rate. The Arrhenius plot of the growth rate between 1300 and 1600 K exhibited a single activation energy, ∼2.4 eV, which is in agreement with the experimental value (∼2.7 eV). However, the growth rate at temperatures below 1300 K deviated from the extrapolated ones from 1300 to 1600 K, which is because recrystallization does not reach a steady state: long-time MD simulations are required to estimate the growth rate at low temperature. The structure analysis of amorphous/crystalline interfaces suggested that the braking of atomic bonds parallel to the interface becomes a rate-limiting step of the SPE growth.

Structural and vibrational properties of SnxGe1-x: Modeling and experiments

The effects of the composition and macroscopic strain on the structural properties and lattice vibrations of SnxGe1-x solid solutions (SSs) are investigated numerically, employing Tersoff empirical inter-atomic potentials, and experimentally. The calculations provide statistical distributions of bond lengths, pair correlation function, and vibrational Raman spectra of the SSs. Using this approach, we are able to evaluate the tin-content-dependent shifts due to the local environment (i.e., changes in the atomic mass and bond stiffness) and strain effects in the calculated Raman spectra and compare them to experimental data. The relative importance of the composition dependent effects of the local environment and strain for epitaxial layers of GeSn solid solutions is analysed.

Inversion Calculation of the Interatomic Potentials for Ni0.75AlxMo0.25–x Alloy Employing Microscopic Phase-Field Model

Based on microscopic phase field theory, the interatomic potentials of Ni3Al (L1(2) structure) and Ni3Mo (DO22 structure) was calculated employing the inversion calculation method, and their dependence on temperature and concentration are also studied. The inversion calculation results demonstrate that the first nearest neighbor interatomic potentials (WNi-Al and WNi-Mo) increase linearly with the temperature increasing continuously. WNi-Al increases but WNi-Mo decreases linearly with the continuous increase of Al concentration, and vice versa. Substituting the inversion calculated WNi-Al and WNi-Mo at 973 K into the microscopic phase-field kinetics model, the atomic temporal evolution pictures of phase transformation process are similar to the results obtained by empirical interatomic potentials. In addition, the precipitated gamma' phase has a higher ordered degree, which is more consistent with the actual microstructure evolution of phase transformation process than the calculated results reported in previous studies using empirical interatomic potentials.

A new transferable interatomic potential for molecular dynamics simulations of borosilicate glasses

Borosilicate glasses are traditionally challenging to model using atomic scale simulations due to the composition and thermal history dependence of the coordination state of B atoms. Here, we report a new empirical interatomic potential that shows a good transferability over a wide range of borosilicate glasses—ranging from pure silicate to pure borate end members—while relying on a simple formulation and a constant set of energy parameters. In particular, we show that our new potential accurately predicts the compositional dependence of the average coordination number of boron atoms, glass density, overall short-range and medium-range order structure, and shear viscosity values for several borosilicate glasses and liquids. This suggests that our new potential could be used to gain new insights into the structure of a variety of advanced borosilicate glasses to help elucidate composition-structure-property relationships—including in complex nuclear waste immobilization glasses.

Adsorption of the tallow amine ethoxylate surfactant Ethomeen T/15 on montmorillonite

Abstract The use of biocompatible surfactants is interesting to increase the properties of clay minerals in an environment-friendly process of charge capacity of organics. The adsorption of the surfactant Ethomeen T/15 in the interlayer space of montmorillonite was studied experimentally and computationally. Different proportions of surfactant, water and phyllosilicate were explored. Calculations at molecular level were performed using forcefields based on empirical interatomic potentials. The surfactant is likely to be adsorbed within the montmorillonite interlayer space forming hydrogen bonds between the H atoms of surfactant and the basal tetrahedral O atoms of the montmorillonite interlayer surface. These hydrogen bonds and the electrostatic interactions between cations and phyllosilicate surface are the main driving forces of the adsorption. Molecular dynamics simulations have indicated several possible conformations of the surfactant molecules in the interlayer space. The combination of molecular modeling, thermal analysis, Fourier transformed infrared spectroscopy (FT-IR) and X-ray diffraction (XRD) has allowed the interpretation of this adsorption process.

Vitrification, crystallization, and atomic structure of deformed and quenched Ni60Nb40 metallic glass

A Ni60Nb40 alloy was vitrified from fcc Ni and bcc Nb by cold rolling and by rapid solidification of an alloy ingot. The Ni60Nb40 that was vitrified by rolling has a lower crystallization onset temperature with single crystallization reaction compared to the two-stage crystallization at higher temperature for quenched Ni60Nb40. Fluctuation electron microscopy and hybrid reverse Monte Carlo modeling show that the as-rolled Ni60Nb40 contains fewer icosahedral-like Voronoi clusters and more crystal-like and bcc-like Voronoi clusters, which are inherited from the bcc Nb. The crystal-like and bcc-like medium range order clusters may be the structural origin for lower crystallization temperature in the as-rolled glass. It is demonstrated that hybrid reverse Monte Carlo simulations incorporating fluctuation microscopy and an empirical interatomic potential are insensitive to the starting structure.

Thermal conductivity of ternary III-V semiconductor alloys: The role of mass difference and long-range order

Thermal transport in bulk ternary III-V arsenide (III-As) semiconductor alloys was investigated using equilibrium molecular dynamics with optimized Albe-Tersoff empirical interatomic potentials. Existing potentials for binary AlAs, GaAs, and InAs were optimized to match experimentally obtained acoustic-phonon dispersions and temperature-dependent thermal conductivity. Calculations of thermal transport in ternary III-Vs commonly employ the virtual-crystal approximation (VCA), where the structure is assumed to be a random alloy and all group-III atoms (cations) are treated as if they have an effective weighted-average mass. Here, we showed that it is critical to treat atomic masses explicitly and that the thermal conductivity obtained with explicit atomic masses differs considerably from the value obtained with the average VCA cation mass. The larger the difference between the cation masses, the poorer the VCA prediction for thermal conductivity. The random-alloy assumption in the VCA is also challenged because X-ray diffraction and transmission electron microscopy show order in InGaAs, InAlAs, and GaAlAs epilayers. We calculated thermal conductivity for three common types of order (CuPt-B, CuAu-I, and triple-period-A) and showed that the experimental results for In0.53Ga0.47As and In0.52Al0.48As, which are lattice matched to the InP substrate, can be reproduced in molecular dynamics simulation with 2% and 8% of random disorder, respectively. Based on our results, thermal transport in ternary III-As alloys appears to be governed by the competition between mass-difference scattering, which is much more pronounced than the VCA suggests, and the long-range order that these alloys support.

Development of a machine learning potential for graphene

We present an accurate interatomic potential for graphene, constructed using the Gaussian Approximation Potential (GAP) machine learning methodology. This GAP model obtains a faithful representation of a density functional theory (DFT) potential energy surface, facilitating highly accurate (approaching the accuracy of ab initio methods) molecular dynamics simulations. This is achieved at a computational cost which is orders of magnitude lower than that of comparable calculations which directly invoke electronic structure methods. We evaluate the accuracy of our machine learning model alongside that of a number of popular empirical and bond-order potentials, using both experimental and ab initio data as references. We find that whilst significant discrepancies exist between the empirical interatomic potentials and the reference data - and amongst the empirical potentials themselves - the machine learning model introduced here provides exemplary performance in all of the tested areas. The calculated properties include: graphene phonon dispersion curves at 0 K (which we predict with sub-meV accuracy), phonon spectra at finite temperature, in-plane thermal expansion up to 2500 K as compared to NPT ab initio molecular dynamics simulations and a comparison of the thermally induced dispersion of graphene Raman bands to experimental observations. We have made our potential freely available online at [http://www.libatoms.org].

Compositional and strain analysis of In(Ga)N/GaN short period superlattices

Extensive high resolution transmission and scanning transmission electron microscopy observations were performed in In(Ga)N/GaN multi-quantum well short period superlattices comprising two-dimensional quantum wells (QWs) of nominal thicknesses 1, 2, and 4 monolayers (MLs) in order to obtain a correlation between their average composition, geometry, and strain. The high angle annular dark field Z-contrast observations were quantified for such layers, regarding the indium content of the QWs, and were correlated to their strain state using peak finding and geometrical phase analysis. Image simulations taking into thorough account the experimental imaging conditions were employed in order to associate the observed Z-contrast to the indium content. Energetically relaxed supercells calculated with a Tersoff empirical interatomic potential were used as the input for such simulations. We found a deviation from the tetragonal distortion prescribed by continuum elasticity for thin films, i.e., the strain in the relaxed cells was lower than expected for the case of 1 ML QWs. In all samples, the QW thickness and strain were confined in up to 2 ML with possible indium enrichment of the immediately abutting MLs. The average composition of the QWs was quantified in the form of alloy content.

Phonon optimized interatomic potential for aluminum

We address the problem of generating a phonon optimized interatomic potential (POP) for aluminum. The POP methodology, which has already been shown to work for semiconductors such as silicon and germanium, uses an evolutionary strategy based on a genetic algorithm (GA) to optimize the free parameters in an empirical interatomic potential (EIP). For aluminum, we used the Vashishta functional form. The training data set was generated ab initio, consisting of forces, energy vs. volume, stresses, and harmonic and cubic force constants obtained from density functional theory (DFT) calculations. Existing potentials for aluminum, such as the embedded atom method (EAM) and charge-optimized many-body (COMB3) potential, show larger errors when the EIP forces are compared with those predicted by DFT, and thus they are not particularly well suited for reproducing phonon properties. Using a comprehensive Vashishta functional form, which involves short and long-ranged interactions, as well as three-body terms, we were able to better capture interactions that reproduce phonon properties accurately. Furthermore, the Vashishta potential is flexible enough to be extended to Al2O3 and the interface between Al-Al2O3, which is technologically important for combustion of solid Al nano powders. The POP developed here is tested for accuracy by comparing phonon thermal conductivity accumulation plots, density of states, and dispersion relations with DFT results. It is shown to perform well in molecular dynamics (MD) simulations as well, where the phonon thermal conductivity is calculated via the Green-Kubo relation. The results are within 10% of the values obtained by solving the Boltzmann transport equation (BTE), employing Fermi’s Golden Rule to predict the phonon-phonon relaxation times.