Interatomic Potential Research Articles

Geometry-enhanced pretraining on interatomic potentials

Geometry-enhanced pretraining on interatomic potentials

Accurate machine learning force fields via experimental and simulation data fusion

Machine Learning (ML)-based force fields are attracting ever-increasing interest due to their capacity to span spatiotemporal scales of classical interatomic potentials at quantum-level accuracy. They can be trained based on high-fidelity simulations or experiments, the former being the common case. However, both approaches are impaired by scarce and erroneous data resulting in models that either do not agree with well-known experimental observations or are under-constrained and only reproduce some properties. Here we leverage both Density Functional Theory (DFT) calculations and experimentally measured mechanical properties and lattice parameters to train an ML potential of titanium. We demonstrate that the fused data learning strategy can concurrently satisfy all target objectives, thus resulting in a molecular model of higher accuracy compared to the models trained with a single data source. The inaccuracies of DFT functionals at target experimental properties were corrected, while the investigated off-target properties were affected only mildly and mostly positively. Our approach is applicable to any material and can serve as a general strategy to obtain highly accurate ML potentials.

Collaboration on Machine-Learned Potentials with IPSuite: A Modular Framework for Learning-on-the-Fly.

The field of machine learning potentials has experienced a rapid surge in progress, thanks to advances in machine learning theory, algorithms, and hardware capabilities. While the underlying methods are continuously evolving, the infrastructure for their deployment has lagged. The community, due to these rapid developments, frequently finds itself split into groups built around different implementations of machine-learned potentials. In this work, we introduce IPSuite, a Python-driven software package designed to connect different methods and algorithms from the comprehensive field of machine-learned potentials into a single platform while also providing a collaborative infrastructure, helping ensure reproducibility. Furthermore, the data management infrastructure of the IPSuite code enables simple model sharing and deployment in simulations. Currently, IPSuite supports six state-of-the-art machine learning approaches for the fitting of interatomic potentials as well as a variety of methods for the selection of training data, running of ab initio calculations, learning-on-the-fly strategies, model evaluation, and simulation deployment.

Atomistic investigation of phase transformations in NiTiCu shape memory alloys

NiTiCu shape memory alloys have attracted much attention for their ability to reduce hysteresis, which is required fast responses in actuator and sensor applications, but the correlation between the properties and specific process of phase transformation is unclear. In this work, an atomic-level understanding of the detailed phase transformation behavior of NiTiCu shape-memory alloys is achieved based on a newly developed interatomic potential for the Ni–Ti–Cu ternary system. Free energy calculations based on the thermodynamic integration method faithfully reproduce reported experimental trends, such as the compositional dependency of different phase transformation paths. Further, large-scale molecular dynamics simulations performed on nanocrystalline alloys provide visual evidence of the detailed phase transformation process involving different martensitic transformation paths. Our results suggest that although both the composition and grain size can be adjusted to realize slim hysteresis, the effect of increasing the Cu amount is only significant if changes in the phase transformation modes are involved at relatively low Cu contents. When the material exhibits only a single phase transformation mode at relatively high Cu contents, the additional Cu is not highly effective in reducing the hysteresis.

Dual-channel phonon transport leads to low thermal conductivity in pyrochlore La2Hf2O7

The pyrochlore oxide La2Hf2O7 (LHO) has attracted extensive attention in thermal management applications. Accurate determination and comprehensive understanding of thermal conductivity (κ) stand as imperative prerequisites for the effective utilization and manipulation of LHO in various applications. In this study, with the assistance of machine learning interatomic potential, we investigate the microscopic mechanisms of phonon thermal transport in LHO crystals. Our findings show that the thermal transport in LHO defies a precise explication through the conventional framework of the Boltzmann transport equation (BTE) due to the emergence of diffusive phonon modes, engendered by strong anharmonicity. Using the state-of-the-art dual-phonon model incorporating both propagating and diffusive modes, we successfully reproduce the experimentally measured results in both magnitude and temperature dependence. It is also shown that the thermal conduction in LHO crystals is dominated by the propagating modes at low temperatures while by the diffusive modes at high temperatures, which synergistically lead to a low κ for LHO crystals. The novel insights into phonon thermal transport provide a stepping stone for regulating the thermal transport properties of LHO crystals.

Neural network molecular dynamics study of LiGe2(PO4)3: Investigation of structure

Machine learning interatomic potential for LiGe2(PO4)3 in a form of a set of neural networks (DeePMD-model) was trained on DFT data. DFT simulations in GGA PBE approximation were performed for ordered and disordered LiGe2(PO4)3 supercell with 432 atoms which allowed the developed potential to describe crystalline, molten and glassy states. The developed DeePMD-potential was verified using literature and our experimental data demonstrating good agreement with it. Neural network molecular dynamic simulations indicate that P atoms have tetrahedral oxygen environment in both crystalline and glassy state. Coordination environment of Ge atoms is more complex. They have octahedral oxygen environment in crystalline state and mixed environment in glassy state: 4-, 5- and 6-coordinated Ge atoms were found. It was shown by calculation of orientational order parameters that in glassy LiGe2(PO4)3 4-coordinated Ge atoms have tetrahedral oxygen environment, while 6-coordinated Ge atoms have octahedral oxygen environment. Analysis of O-Ge-O angle distribution for 5-coordinated Ge atoms in glassy LiGe2(PO4)3 demonstrates that the environment of Ge atoms in this case is represented by various structures: square pyramids, trigonal bipyramids and transitional structures from 5-coordinated Ge atoms to tetrahedrons.

Exploring Challenging Properties of Liquid Metallic Systems through Machine Learning: Liquid La and Li4Pb Systems.

In this machine learning (ML) study, we delved into the unique properties of liquid lanthanum and the Li4Pb alloy, revealing some unexpected features and also firmly establishing some of the debated characteristics. Leveraging interatomic potentials derived from ab initio calculations, our investigation achieved a level of precision comparable to first-principles methods while at the same time entering the hydrodynamic regime. We compared the structure factors and pair distribution functions to experimental data and unearthed distinctive collective excitations with intriguing features. Liquid lanthanum unveiled two transverse collective excitation branches, each closely tied to specific peaks in the velocity autocorrelation function spectrum. Furthermore, the analysis of the generalized specific heat ratio in the hydrodynamic regime investigated with the ML molecular dynamics simulations uncovered a peculiar behavior, impossible to discern with only ab initio simulations. Liquid Li4Pb, on the other hand, challenged existing claims by showcasing a rich array of branches in its longitudinal dispersion relation, including a high-frequency LiLi mode with a nonhydrodynamic optical character that maintains a finite value as q → 0. Additionally, we conducted an in-depth analysis of various transport coefficients, expanding our understanding of these liquid metallic systems. In summary, our ML approach yielded precise results, offering new and captivating insights into the structural and dynamic aspects of these materials.

Melting point of iron at high pressure: An assessment of uncertainties and effect of electronic temperature

An accurate calculation of the melting point of iron at various pressures in the Earth's core is important for understanding the core structure, geodynamo, and the Earth's history. Previous studies have assessed the melt line of iron at these extreme conditions using various experimental measurement techniques as well as both ab initio and classic molecular dynamics simulations. However, experimental measurements have uncertainties up to several hundred Kelvin, and inconsistencies remain among simulation results. In this work, we propose an iterative framework that couples density functional theory (DFT) calculations and molecular dynamics simulations performed using an ensemble of interatomic potentials to assess the effect of electronic temperature on the melting point. We systematically validate the potentials by comparing lattice constants and phonon dispersion curves at 0 K and enthalpy differences between liquid and HCP, FCC, BCC phases of iron close to the melt line at 300 GPa with DFT. Our results show that HCP iron melts at 6144 K (at 300 GPa), BCC phase is thermodynamically unstable, and FCC is metastable at this temperature. The melting points of FCC and BCC phases at 300 GPa are 5858 and 5647 K, respectively.

Thermoelectric properties of lead-free anti-perovskites X3BN (B = Bi, Sb, X = Mg, Ca, Sr): A theoretical study based on first-principles calculations and machine learning interatomic potential

The search for non-toxic and low-cost perovskites with high thermoelectric performance is still a challenge despite low thermal conductivity. The thermoelectric properties of nitride anti-perovskites X3BN (B = Bi, Sb, X = Mg, Ca, Sr) with a cubic structure were investigated using density functional theory and machine learning interatomic potential. The low Debye temperature and thermal conductivity were obtained due to strong lattice anharmonicity, and the phonon vibration modes were also analyzed. The high band degeneracy and suitable bandgap lead to a large power factor. The maximum power factor is 7.54 mW/mk2 for Mg3BiN at 900 K. We obtained a maximum ZT of 1.49 for p-type Sr3BiN, and it is 1.22 for n-type doping at 900 K. The ZT for Mg3BiN is 1.18 and 1.19 for p-type and n-type doping, respectively. Our calculations reveal that these anti-perovskites are excellent materials for non-toxic and low-cost thermoelectric applications.

Comparison of Interatomic Potentials for Modeling Defects in Graphene Using Molecular Dynamics

In this work, we tested the ability of classical interatomic potentials to describe the energy characteristics of defects of various dimensionality in graphene crystals. Brenner's Reactive Empirical Bond Order potentials (second generation REBO, AIREBO, AIREBO-M), Tersoff potentials, as well as BOP and LCBOP potentials were considered. The data obtained in this work using the molecular dynamics method was compared with literature data obtained using the density functional theory. It is noted that when modeling point and linear defects, the potentials of the REBO family and the LCBOP potential demonstrate the best agreement with the literature data. For modeling pseudo-graphene crystals, the best fit is demonstrated by the Tersoff B-N-C potential, which shows slightly overestimated energy values for linear and point defects, but most accurately describes the geometry of the crystal lattice. The potential of BOP demonstrates its inability to correctly model defect configurations with high densities of eight-member defect rings. When simulating four-member carbon defect rings, most potentials exhibit distortions in the crystal lattice that are not observed in the density functional theory calculations.

Line shapes of isotropic and depolarized CILS spectra of neon: Interaction potential and pair polarizability

The present work proposes three interaction potentials between two gaseous neon atoms and applies them to the study of models representing the anisotropy and trace of the induced pair polarizability tensor. The proposed empirical interatomic potentials for the gaseous Ne-Ne interaction are developed by fitting the Improved Lennard-Jones (ILJ), Modified Tang-Toennies (MTT) and Barker et al. (BFW) potentials simultaneously to the spectroscopic and transport properties over a wide temperature range. The validity of the proposed interatomic potentials is verified by comparing the calculated and experimental second pressure virial coefficient over a wide temperature range. In addition to this, the results show satisfactory accordance between the experimental and calculated vibrational energy levels using the deduced pair potential parameters compared to those reported for neon gas. These comparisons exhibit acceptable accuracy when the present work potential parameters are inserted to reproduce the transport and thermo-physical properties, besides the spectral moments and the spectral line-shape intensities for neon gas. The isotropic and depolarized collision-induced light scattering (CILS) spectra are produced and analyzed as functions of the potential considered. They exhibit sensitivity to both the attractive part of the potential and the short-range values of the polarizability anisotropy or mean at intermediate and high frequencies. Mean and anisotropy are deduced directly from spectral line shapes using a new inversion method.

Transferable Water Potentials Using Equivariant Neural Networks.

Machine learning interatomic potentials (MLIPs) have emerged as a technique that promises quantum theory accuracy for reduced cost. It has been proposed [J. Chem. Phys. 2023, 158, 084111] that MLIPs trained on solely liquid water data cannot accurately transfer to the vapor-liquid equilibrium while recovering the many-body decomposition (MBD) analysis of gas-phase water clusters. This suggests that MLIPs do not directly learn the physically correct interactions of water molecules, limiting transferability. In this work, we show that MLIPs using equivariant architecture and trained on 3200 liquid water structures reproduces liquid-phase water properties (e.g., density within 0.003 g/cm3 between 230 and 365 K), vapor-liquid equilibrium properties up to 550 K, the MBD analysis of gas-phase water cluster up to six-body interactions, and the relative energy and the vibrational density of states of ice phases. We show that potentials developed using equivariant MLIPs allow transferability for arbitrary phases of water that remain stable in nanosecond long simulations.

Mechanical Properties of Small Quasi-Square Graphene Nanoflakes

The rise of straintronics—the possibility of fine-tuning the electronic properties of nanosystems by applying strain to them—has enhanced the interest in characterizing the mechanical properties of these systems when they are subjected to tensile (or compressive), shear and torsion strains. Four parameters are customarily used to describe the mechanical behavior of a macroscopic solid within the elastic regime: Young’s and shear moduli, the torsion constant and Poisson’s ratio. There are some relations among these quantities valid for elastic continuous isotropic systems that are being used for 2D nanocrystals without taking into account the non-continuous anisotropic nature of these systems. We present in this work computational results on the mechanical properties of six small quasi-square (aspect ratio between 0.9 and 1.1) graphene nanocrystals using the PM7 semiempirical method. We use the results obtained to test the validity of two relations derived for macroscopic homogeneous isotropic systems and sometimes applied to 2D systems. We show they are not suitable for these nanostructures and pinpoint the origin of some discrepancies in the elastic properties and effective thicknesses reported in the literature. In an attempt to recover one of these formulas, we introduce an effective torsional thickness for graphene analogous to the effective bending thickness found in the literature. Our results could be useful for fitting interatomic potentials in molecular mechanics or molecular dynamics models for finite carbon nanostructures, especially near their edges and for twisted systems.

Non-singular straight dislocations in anisotropic crystals

A non-singular dislocation theory of straight dislocations in anisotropic crystals is derived using simplified anisotropic incompatible first strain gradient elasticity theory. Based on the non-singular theory of dislocations for anisotropic crystals, all dislocation key-formulas of straight dislocations are derived in generalized plane strain, for the first time. In this model, the singularity of the dislocation fields at the dislocation core is regularized owing to the nonlocal nature of strain gradient elasticity. The non-singular dislocation fields of straight dislocations are obtained in terms of two-dimensional anisotropic Green functions of simplified anisotropic strain gradient elasticity. All necessary Green functions, including the two-dimensional Green tensor of the twofold anisotropic Helmholtz-Navier operator and the two-dimensional F\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\varvec{F}$$\\end{document}-tensor of generalized plane strain, are derived as sum of the classical part and a gradient part in terms of Meijer G-functions. Among others, we calculate the regularization of the Barnett solution for the elastic distortion of straight dislocations in an anisotropic crystal. In the framework of simplified anisotropic first strain gradient elasticity, the necessary material parameters are computed for cubic materials including aluminum (Al), copper (Cu), iron (Fe) and tungsten (W) using a second nearest-neighbour modified embedded-atom-method interatomic potential. The elastic distortion and stress fields of screw and edge dislocations of 12⟨111⟩\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\frac{1}{2} \\langle 111\\rangle$$\\end{document} Burgers vector in bcc iron and bcc tungsten and screw and edge dislocations of 12⟨110⟩\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\frac{1}{2} \\langle 110\\rangle$$\\end{document} Burgers vector in fcc copper and fcc aluminum have been computed and presented in contour plots.

Deep-potential enabled multiscale simulation of gallium nitride devices on boron arsenide cooling substrates

High-efficient heat dissipation plays critical role for high-power-density electronics. Experimental synthesis of ultrahigh thermal conductivity boron arsenide (BAs, 1300 W m−1K−1) cooling substrates into the wide-bandgap semiconductor of gallium nitride (GaN) devices has been realized. However, the lack of systematic analysis on the heat transfer across the GaN-BAs interface hampers the practical applications. In this study, by constructing the accurate and high-efficient machine learning interatomic potentials, we perform multiscale simulations of the GaN-BAs heterostructures. Ultrahigh interfacial thermal conductance of 260 MW m−2K−1 is achieved, which lies in the well-matched lattice vibrations of BAs and GaN. The strong temperature dependence of interfacial thermal conductance is found between 300 to 450 K. Moreover, the competition between grain size and boundary resistance is revealed with size increasing from 1 nm to 1000 μm. Such deep-potential equipped multiscale simulations not only promote the practical applications of BAs cooling substrates in electronics, but also offer approach for designing advanced thermal management systems.

Mechanical Properties of Single and Polycrystalline Solids from Machine Learning

AbstractCalculating the elastic and mechanical characteristics of non‐crystalline solids can be challenging due to the high computational cost of ab initio methods and the low accuracy of empirical potentials. This paper proposes a computational technique for efficient calculations of mechanical properties of polycrystals, composites, and multi‐phase systems from atomistic simulations with high accuracy and reasonable computational cost. The calculated elastic moduli of polycrystalline diamond and their dependence on grain size are determined using a developed approach based on actively learned machine learning interatomic potentials (MLIPs). These potentials are trained on local fragments of the polycrystalline system, and ab initio calculations are used to compute forces, stresses, and energies. This technique allows researchers to perform extensive calculations of the mechanical properties of complex solids with different compositions and structures, achieving high accuracy and facilitating the transition from ideal (single crystal) systems to more realistic ones.

Enhancing the Quality and Reliability of Machine Learning Interatomic Potentials through Better Reporting Practices

Enhancing the Quality and Reliability of Machine Learning Interatomic Potentials through Better Reporting Practices

A new interatomic potential of W-Ni-Fe systems for point defects and mechanical property studies

W-Ni-Fe tungsten heavy alloys (WHAs) are a kind of ductile phase toughened composites that are becoming increasingly interesting as promising alternatives to monolithic tungsten for fusion reactor plasma-facing materials, because of their outstanding combination of strength, ductility and toughness. Understanding the radiation-induced microstructural features of WHAs associated with their mechanical property changes is therefore a topic of active research. The theoretical study of radiation-induced collision cascades by means of atomistic simulations such as classical molecular dynamics (MD) is highly desirable to provide a detailed analysis for the atom-level primary damage production. Here, we develop a Finnis–Sinclair type interatomic potential for W-Ni-Fe ternary systems based on previously proposed unary W and Fe potentials, and use it to study the point defects and mechanical properties of WHAs. The developed potential yields the bulk and defect properties of pure Ni that are in good agreement with experimental data and density functional theory calculations. In addition, it is also capable of reproducing alloy properties with a satisfactory accuracy, especially regarding the solute migration energies, as well as the ground-state interstitial defect configurations. The constructed potential further shows reasonable transferability to other mechanically relevant properties that are not contained in the fitting database. The present potential here will enable large-scale MD simulations of cascade, deformation and fracture of WHAs.

Development of a genetic algorithm based interatomic potential and application in thermal conductivity study of ThO2 grain boundaries

ThO2 is a promising fuel for next-generation reactors due to its superior thermo-physical properties. In this study, we present a Coulomb-Buckingham-Morse interatomic potential for ThO2, developed using a homemade genetic algorithm (GA) method. The potential accurately captures fundamental properties, such as lattice constants, elastic constants, point defect energies, phonon spectra, and thermal conductivity. We apply this potential to perform a systematic study of thermal transport of twenty-one tilt grain boundaries (GBs) in ThO2, which have a significant effect on the macroscopic properties of polycrystalline materials. We demonstrate that the microscopic structure, GB excess energy, and GB disorder index are well correlated with the Kapitza conductance for a wide variety of GBs. Additionally, we discuss the effect of high temperature on the Kapitza conductance, where the GB structure undergoes disordering when approaching the melting temperature, resulting in an increase of thermal resistance. Finally, a detailed analysis of the phonon density of states is presented to reveal the mechanisms of thermal conductance in ThO2.

Studies of Ni-Cr complexation in FLiBe molten salt using machine learning interatomic potentials

In nuclear and/or solar applications that involve molten salts, impurities frequently enter the salt as either fission products or via corrosion. Impurities can interact and make complexes, but the impact of such complexation on the properties of the salts and corrosion rates has not been understood. Common impurities in molten salts, such as FLiBe, include Cr, Ni, and Fe. Here, we investigate the complexation of Cr and Ni in FLiBe using molecular dynamics based on a machine learning interatomic potential (MLIP) fitted using the atomic cluster expansion (ACE) method. The MLIP allows us to overcome the challenges of simultaneously needing accurate energetics and long time scale to study complexation. We demonstrate that impurity behavior is more difficult to capture than that of concentrated elements with MLIPs due to less sampling in training data, but that this can be overcome by using active learning strategies to obtain a robust fit. Our findings suggest that there is a weak but potentially significant binding free energy between CrF2 and NiF2 in eutectic FLiBe of -0.112eV. Under certain conditions this binding creates a significant short-range order between the species and lowers the redox potential of NiF2 in the presence of CrF2 in FLiBe, making Ni dissolution more favorable in the presence of Cr as compared to its dissolution in pure FLiBe. However, we find little impact of this complexation on the diffusivity of Ni and Cr. Overall, the methodology presented here suggests an approach to modeling complexation with MLIPs and suggests that interactions between dissolved cations could be playing a significant role in some salt thermophysical properties.