Search NASA⌕ Search

SEARCH · Search NASA

Results for “INTERATOMIC FORCES”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 307 records · Page 17

Development of a Modified Embedded Atom Force Field for Zirconium Nitride Using Multi-Objective Evolutionary Optimization

Zirconium nitride (ZrN) exhibits exceptional mechanical, chemical, and electrical properties, which make it attractive for a wide range of technological applications, including wear-resistant coatings, protection from corrosion, cutting/shaping tools, and nuclear breeder reactors. Despite its broad usability, an atomic scale understanding of the superior performance of ZrN, and its response to external stimuli, for example, temperature, applied strain, and so on, is not well understood. This is mainly due to the lack of interatomic potential models that accurately describe the interactions between Zr and N atoms. To address this challenge, we develop a modified embedded atom method (MEAM) interatomic potential for the Zr–N binary system by training against formation enthalpies, lattice parameters, elastic properties, and surface energies of ZrN (and, in some cases, also Zr3N4) obtained from density functional theory (DFT) calculations. The best set of MEAM parameters are determined by employing a multiobjective global optimization scheme driven by genetic algorithms. Our newly developed MEAM potential accurately reproduces structure, thermodynamics, energetic ordering of polymorphs, as well as elastic and surface properties of Zr–N compounds, in excellent agreement with DFT calculations and experiments. As a representative application, we employed molecular dynamics simulations based on this MEAM potential to investigate the atomic scale mechanisms underlying fracture of bulk and nanopillar ZrN under applied uniaxial strains, as well as the impact of strain rate on their mechanical behavior. These simulations indicate that bulk ZrN undergoes brittle fracture irrespective of the strain rate, while ZrN nanopillars show quasi-plasticity owing to amorphization at the crack front. The MEAM potential for Zr–N developed in this work is an invaluable tool to investigate atomic-scale mechanisms underlying the response of ZrN to external stimuli (e.g, temperature, pressure etc.), as well as other interesting phenomena such as precipitation.

36 MATERIALS SCIENCE↗

Pressure-tuned Frustration of Magnetic Coupling in Elemental Europium

Applying linear response and the magnetic force theorem in correlated density functional theory, the intersublattice exchange constants of antiferromagnetic Eu are calculated and found to vanish near the pressure of P c = 82 GPa, just where magnetic order is validated experimentally to be lost. The Eu 4f 7 moment remains unchanged at high pressure, again in agreement with spectroscopic measurements, leaving the picture of perfect frustration of interatomic Ruderman-Kittel-Kasuya-Yoshida couplings in a broad metallic background, leaving a state of electrons strongly exchange coupled to arbitrarily oriented, possibly quasistatic local moments. This strongly frustrated state gives way to superconductivity at T c = 1.7 K, observed experimentally. Furthermore, these phenomena, and free energy considerations related to correlations, suggest an unusual phase of matter that is discussed within the scenarios of the Doniach Kondo lattice phase diagram, the metallic spin glass class, and itinerant spin liquid or spin gas systems.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Datasets for Custom-trained Machine-learning Interatomic Potentials: Nitric Acid Aqueous Solution

This dataset was generated using an iterative active learning strategy with the ArcaNN software package (https://github.com/arcann-chem/arcann_training) to train machine-learning interatomic potentials (MLIPs) for aqueous nitric acid. Each active-learning cycle consisted of three stages: (1) training, (2) exploration, and (3) labeling. The initial training set comprised approximately 800 randomly selected configurations from a previous study by Lewis et al. (https://doi.org/10.1021/jp205510q), which investigated nitric acid solutions at 2, 3, 4, and 5 mol/L. For all configurations, single-point calculations of atomic forces and total energies were performed at the quantum density functional theory BLYP-D2 and PBE-D3 levels of theory using the CP2K Quickstep module. Valence electrons were treated explicitly, while core electrons on all atoms were represented by norm-conserving Goedecker–Teter–Hutter (GTH) pseudopotentials. Long-range dispersion interactions were accounted for using Grimme dispersion corrections. Wave functions were expanded in a mixed Gaussian-and-plane-wave scheme using TZV2P-MOLOPT basis sets for all elements and an 800 Ry auxiliary plane-wave cutoff for the electron density. Self-consistent field convergence was accelerated using orbital transformation and Direct Inversion in the Iterative Subspace, with a convergence threshold of 10^{-6}. All single-point calculations were carried out in periodic orthorhombic cells whose dimensions match those of the molecular configurations sampled from earlier trajectories. The CELL_REF keyword in CP2K was used to define a fixed reference cell, ensuring consistency in the reference data used for MLIP training, particularly when cell fluctuations are present in NpT simulations. The resulting high-fidelity energies and forces constitute the ground-truth labels used to train the MLIPs contained in this dataset.

Dinpajooh, Mohammadhasan [Pacific Northwest Nation↗

When more data hurts: Optimizing data coverage while mitigating diversity-induced underfitting in an ultrafast machine-learned potential

Machine-learned interatomic potentials (MLIPs) are becoming an essential tool in materials modeling. However, optimizing the generation of training data used to parametrize the MLIPs remains a significant challenge. This is because MLIPs can fail when encountering local environments too different from those present in the training data. The difficulty of determining a priori the environments that will be encountered during molecular dynamics simulation necessitates diverse, high-quality training data. Here, this study investigates how training data diversity affects the performance of MLIPs using the Ultra-Fast force field (UF 3 ) to model amorphous silicon nitride. We employ expert and autonomously generated data to create the training data and fit four force field variants to subsets of the data. Our findings reveal a critical balance in training data diversity: insufficient diversity hinders generalization, while excessive diversity can exceed the MLIP's learning capacity, reducing simulation accuracy. Specifically, we found that the UF 3 variant trained on a subset of the training data, in which nitrogen-rich structures were removed, offered vastly better prediction and simulation accuracy than any other variant. By comparing these UF 3 variants, we highlight the nuanced requirements for creating accurate MLIPs, emphasizing the importance of application-specific training data to achieve optimal performance in modeling complex material behaviors.

ab initio molecular dynamics↗

Modification of a force field for molecular dynamics simulations of silicon etching by chlorine atoms

Here, a modified classical molecular dynamics (MD) force field is presented for the interaction between silicon (Si) and chlorine (Cl). The original version of the force field is shown to significantly overestimate the probability of Si etching by thermal Cl atoms. However, the modified force field corrects this problem and results in generally good agreement with experimental data. Further, it is shown that while the modification of the force field improves the prediction of Si spontaneous etching with Cl atoms, it does not degrade predictions of atomic-layer etching of Si with Cl[Formula: see text] molecules.

74 ATOMIC AND MOLECULAR PHYSICS↗

HydraGNN v5.0

HydraGNN v5.0 expands the code base into a more portable, scalable, and flexible framework for scientific graph learning, with particular strength in atomistic machine-learning interatomic potentials and large-scale distributed training. The release adds Fully Sharded Data Parallel (FSDP) support alongside existing DDP and DeepSpeed paths, including FSDP-aware checkpointing and optimizer integration, and introduces a configurable multi-precision training workflow supporting FP32, BF16, and FP64 across GPUs and Intel XPUs. For atomistic modeling, HydraGNN v5.0 strengthens its MLIP capabilities through dynamic graph construction at every forward pass, energy-conserving force prediction via automatic differentiation, and per-atom energy loss formulations, while extending EGNN models to properly handle periodic boundary conditions. The release also broadens model expressiveness through graph-level attribute conditioning, adds new multi-task and model-parallel extensions such as MACE support and encoder/decoder branch optimization, and expands application coverage with integrated examples for datasets including OC25, Nabla2-DFT, QCML, Open Polymers 2026, and OPF. In parallel, HydraGNN v5.0 improves production readiness through performance optimizations for large-scale runs, stratified sampling and linear-regression preprocessing utilities, and tested installation scripts for DOE supercomputers including Frontier, Aurora, Perlmutter, and Andes. Overall, the release advances HydraGNN as a robust software platform for scalable graph neural networks across materials science, chemistry, and scientific machine learning workflows

Lupo Pasini, Massimiliano [Oak Ridge National Labo↗

Modeling Helium Segregation to the Surfaces of Plasma-Exposed Tungsten as a Function of Temperature and Surface Orientation

We provide a description of the dependence on surface crystallographic orientation and temperature of the segregation of helium implanted with energies consistent with low-energy plasma exposure to tungsten surfaces. Here, we describe multiscale modeling results based on a hierarchical approach to scale bridging that incorporates atomistic studies based on a reliable interatomic potential to parameterize a spatially dependent drift-diffusion-reaction cluster-dynamics code. An extensive set of molecular dynamics (MD) simulations has been performed at 933 K and/or 1200 K to determine the probabilities of desorption and modified trap mutation that occurs as small, mobile Hen (1 ≤ n ≤ 7) clusters diffuse from the near-surface region toward surfaces of varying crystallographic orientation due to an elastic interaction force that provides the thermodynamic driving force for surface segregation. These near-surface cluster dynamics have significant effects on the surface morphology, the near-surface defect structures, and the amount of helium retained in the material upon plasma exposure, for which we have developed an extensive MD database of cumulative evolution during high-flux helium implantation at 933 K, which we compare to our properly parameterized cluster-dynamics model. This validated model is then used to evaluate the effects of temperature on helium retention and subsurface helium clustering.

Blondel, Sophie↗

van der Waals interaction as the starting point for an effective field theory

We consider the system of three 4 He atoms to assess whether a pure van der Waals potential can be used as a starting point for an effective field theory to describe three-body processes in ultracold atomic systems. Using a long-range van der Waals interaction in combination with short-distance two-body counterterms, we analyze the dependence of two- and three-body observables on the short-distance regulator that is required due to the singular nature of the van der Waals interaction. We benchmark our approach with results obtained with the realistic 4 He- 4 He LM2M2 potential and find good agreement. We furthermore show that in this effective field theory approach no three-body force is required at leading order and that universal van der Waals physics leads to a universal correlation between three-body observables in the absence of an Efimov three-body parameter.

74 ATOMIC AND MOLECULAR PHYSICS↗

Development of Multicomponent EAM Potential for Ni-Based Superalloy

We initiated the development of multi-component EAM potentials for Aluminides and Carbides, key phases in Ni-based Superalloys. The goal is to utilize the MD simulation to understand the deformation dynamics that contribute to the formation of voids and creep initiation. For this purpose, we constructed the raw data from ab-initio (molecular dynamics) MD simulations fed into the potential development code and used Nickel as the base metal with the addition of a number of various elements including Aluminum, Chromium, Tungsten. We then developed the EAM potentials for the aluminide and carbide phases using the force-fitting code MEAMfit. Our generated potential reproduces the fundamental properties of the Ni3Al and Mo23C6 phases. We verified further the EAM potential through the thermal stability test at different temperatures and by reproducing the elastic constants consistent with the experimental values.

36 MATERIALS SCIENCE↗

Helium segregation on surfaces of plasma-exposed tungsten

Here we report a hierarchical multi-scale modeling study of implanted helium segregation on surfaces of tungsten, considered as a plasma facing component in nuclear fusion reactors. We employ a hierarchy of atomic-scale simulations based on a reliable interatomic interaction potential, including molecular-statics simulations to understand the origin of helium surface segregation, targeted molecular-dynamics (MD) simulations of near-surface cluster reactions, and large-scale MD simulations of implanted helium evolution in plasma-exposed tungsten. We find that small, mobile He-n (1 <= n <= 7) clusters in the near-surface region are attracted to the surface due to an elastic interaction force that provides the thermodynamic driving force for surface segregation. Elastic interaction force induces drift fluxes of these mobile Hen clusters, which increase substantially as the migrating clusters approach the surface, facilitating helium segregation on the surface. Moreover, the clusters' drift toward the surface enables cluster reactions, most importantly trap mutation, in the near-surface region at rates much higher than in the bulk material. Moreover, these near-surface cluster dynamics have significant effects on the surface morphology, near-surface defect structures, and the amount of helium retained in the material upon plasma exposure. We integrate the findings of such atomic-scale simulations into a properly parameterized and validated spatially dependent, continuum-scale reaction-diffusion cluster dynamics model, capable of predicting implanted helium evolution, surface segregation, and its near-surface effects in tungsten. This cluster-dynamics model sets the stage for development of fully atomistically informed coarse-grained models for computationally efficient simulation predictions of helium surface segregation, as well as helium retention and surface morphological evolution, toward optimal design of plasma facing components.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Development of an interatomic potential for the Ta–Li system

A new interatomic potential for the Ta–Li system is introduced to facilitate the study of phase stability, mechanical properties and non-equilibrium dynamics after Li implantation in Ta. Here, this potential is based on a generalization of the embedded atom method (GEAM) and includes contributions from embedding energy, explicit two- and three-body interactions, and nonlocal many-body interaction terms. The parameters of the potential are optimized using energies and atomic forces for a wide range of configurations obtained from ab initio density functional theory (DFT) calculations. The potential is rigorously validated across a range of physical properties, including elastic constants, equations of state, phonon dispersion curves, point defect properties, and melting temperatures for different compositions. Although our potential is trained on a small dataset, its accuracy is comparable to that of available machine learning potentials for Li and Ta. Our simulations show that at temperatures below 500 K, Li atoms in Ta–Li alloys form clusters separated by Ta-rich domains, and we find no evidence of ordered phase formation. For Li concentrations below a few percent, Li atoms preferentially segregate to surfaces and grain boundaries. However, in alloys containing more than ~10% Li, the accumulation of Li in symmetric-tilt grain boundaries can lead to one of the following effects: formation of amorphous-like regions, changes in grain boundary structural units, or lateral movement of the grain boundary.

GEAM potential↗

Foundation models for atomistic simulation of chemistry and materials

Conventional computational methods for modeling chemical and materials systems are limited by system size and timescale, forcing a trade-off between quantum-mechanical accuracy and the sampling needed for realistic observables. Large language and vision foundation models — pre-trained on massive datasets using transformer architectures — have revolutionized many fields. It is thus interesting to ask whether a foundation model — subject to suitable data, parameter scaling and training — could enable learned simulations of chemistry and materials. Here, in this study, we review the field of machine-learned interatomic potentials (MLIPs) and posit that scaling up large and diverse chemical and materials datasets and highly expressive architectures using advanced training strategies should result in models that are: more efficient, transferable, robust to out-of-distribution scenarios, and easier to fine-tune to a variety of downstream physical observables than models trained from scratch on small datasets corresponding to specific, targeted atomistic simulation tasks. We provide specific criteria for creating such large-scale MLIP foundation models, coordinated strategies for their development, evaluation and deployment, and highlight potential emergent capabilities that could transform predictive simulations in chemistry and materials science and accelerate discovery across multiple technological domains.

Yuan, Eric C.-Y. [University of California, Berkel↗

Implementation of a Fe-He Three-body Interatomic Potential

The implementation of a recently-developed interatomic potential describing the interactions of helium in bcc iron is described. This He-Fe potential was based on an empirical fit to the results of ab initio calculations of both the formation and relaxation energies of small helium defect structures in iron. The fitting database included substitutional and interstitial helium, as well as small helium and helium-vacancy clusters. In contrast to previous He-Fe pair potentials, fitting the ab initio forces and energies required the use of a three-body interaction term. The implementation of this potential for atomistic simulations using molecular dynamics presented certain challenges which are discussed here to facilitate its further use in materials research, particularly to investigate the behavior of iron-based alloys that may be employed in fusion energy systems.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Multiobjective Hyperparameter Optimization for Deep Learning Interatomic Potential Training Using NSGA-II

Deep neural network (DNN) potentials are an emerging tool for simulation of dynamical atomistic systems, with the promise of quantum mechanical accuracy at speedups of 10000$\times$. As with other DNN methods, hyperparameters used during training can make a substantial difference in model accuracy, and optimal settings vary with dataset. To enable rapid tuning of hyperparameters for DNN potential training, we developed a scalable multiobjective optimization evolutionary algorithm for supercomputers and tested it on the Summit system at the Oak Ridge Leadership Computing Facility (OLCF). The multiobjective approach is required due to the coupling of two learned values defining the potential: the energy and force. Using a large-scale implementation of the NSGA-II algorithm adapted for training DNN potentials, we discovered several optimal multiobjective combinations, including best choices of activation functions, learning rate scaling scheme, and pairing of the two radial cutoffs used in the three dimensional descriptor function.

Coletti, Mark↗

Structural evolution in Au- and Pd-based metallic glass forming liquids and the case for improved molecular dynamics force fields

The results of a combined experimental and computational investigation of the structural evolution of Au 81 Si 19 , Pd 82 Si 18 , and Pd 77 Cu 6 Si 17 metallic glass forming liquids are presented. Electrostatically levitated metallic liquids are prepared, and synchrotron x-ray scattering studies are combined with embedded atom method molecular dynamics simulations to probe the distribution of relevant structural units. Metal–metalloid based metallic glass forming systems are an extremely important class of materials with varied glass forming ability and mechanical processibility. High quality experimental x-ray scattering data are in poor agreement with the data from the molecular dynamics simulations, demonstrating the need for improved interatomic potentials. The first peak in the x-ray static structure factor in Pd 77 Cu 6 Si 17 displays evidence for a Curie–Weiss type behavior but also a peak in the effective Curie temperature. In conclusion, a proposed order parameter distinguishing glass forming ability, 1/(S(T,q 1 )−1, shows a peak in the effective Curie temperature near a crossover temperature established by the behavior of the viscosity, T A .

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Thermomechanics of Deformations in an Anharmonic Solid Body

The macroscopic laws determining the temperature and deformations of an anharmonic solid body in a given external temperature force field have been stated in the form of the first thermodynamics law supplemented by equations of state of the body. The internal and free energies necessary for it are found from the statistical sum in which some of degrees of freedom determining the body shape are discharged from statistical averaging. These functions of state have been calculated up to the first order of the perturbation theory in the anharmonicity for the microscopic dynamic model of the body with the interatomic interaction potential energy given as a series in powers of atom coordinates. The classical region of high temperatures is considered.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

A deep learning interatomic potential developed for atomistic simulation of carbon materials

Interatomic potentials based on neural-network machine learning method have attracted considerable attention in recent years owing to their outstanding ability to balance the accuracy and efficiency in atomistic simulations. In this work, a neural-network potential (NNP) for carbon is generated to simulate the structural properties of various carbon structures. The potential is trained using a database consisting of crystalline and liquid structures obtained by the first-principles density functional theory (DFT) calculations. The developed potential accurately predicts the energies and forces in crystalline and liquid carbon structures, the energetic stability of defected graphene, and the structures of amorphous carbon as the function of density. As a result, the excellent accuracy and transferability of the NNP provide a promising tool for accurate atomistic simulations of various carbon materials with faster speed and much lower cost.

36 MATERIALS SCIENCE↗

Generalizable machine learning potentials for quantum-accurate predictions of non-equilibrium behavior in 2D materials

Machine learning interatomic potentials (ML-IAPs) are emerging as transformative tools in materials modeling, promising quantum-level accuracy at a fraction of the computational cost. However, their ability to generalize beyond equilibrium configurations and to reliably capture defect- and temperature-driven behavior remains underexplored. Here, we develop and benchmark two state-of-the-art ML-IAPs, Spectral Neighbor Analysis Potential (SNAP) and Allegro, on a comprehensive dataset for monolayer MoSe₂. Using density functional theory (DFT) as the reference, we evaluate their performance in capturing stress–strain behavior, phase transition energetics, defect evolution, edge stability, and fracture toughness. Allegro, a deep equivariant neural network potential, surpasses both SNAP and the classical Tersoff potential in accuracy, efficiency, and transferability. Importantly, both ML potentials accurately reproduce experimental fracture measurements and ab initio predictions of inversion domain formation—phenomena well beyond their training sets. Our findings establish ML-IAPs as viable replacements for traditional force fields in the study of non-equilibrium mechanical phenomena, enabling large-scale, high-fidelity simulations in 2D materials and beyond. In conclusion, this work provides a broadly applicable framework for data-driven modeling of structural and functional transformations under extreme conditions.

2D materials↗