Search NASA⌕ Search

SEARCH · Search NASA

Results for “ATOMIC MODELS”

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 991 records · Page 55

Graph neural networks predict energetic and mechanical properties for models of solid solution metal alloy phases

Here, we developed a PyTorch-based architecture called HydraGNN that implements graph convolutional neural networks (GCNNs) to predict the formation energy and the bulk modulus for models of solid solution alloys for various atomic crystal structures and relaxed volumes. We trained the GCNN surrogate model on a dataset for nickel–niobium (NiNb) generated by the embedded atom model (EAM) empirical interatomic potential for demonstration purposes. The dataset was generated by calculating the formation energy and the bulk modulus as a prototypical elastic property for optimized geometries starting from initial body-centered cubic (BCC), face-centered cubic (FCC), and hexagonal compact packed (HCP) crystal structures, with configurations spanning the possible compositional range for each of the three types of initial crystal structures. Numerical results show that the GCNN model effectively predicts both the formation energy and the bulk modulus as function of the optimized crystal structure, relaxed volume, and configurational entropy of the model structures for solid solution alloys.

36 MATERIALS SCIENCE↗

Performance Modeling of Vectorized SNAP Inter-Atomic Potentials on CPU Architectures.

SNAP potentials are inter-atomic potentials for molecular dynamics that enable simulations at accuracy levels comparable to density functional theory(DFT) at a fraction of the cost. As such, SNAP scales to on the order of 10 4 — 10 6 atoms. In this work, we explore CPU optimization of potentials computation using SIMD. We note that efficient use of SIMD is non-obvious as the application features an irregular iteration space for various potential terms, necessitating use of SIMD across atoms in a cross matrix, batched fashion. We present a preliminary analytic model to determine the correct batch size for several CPU architectures across several vendors, and show end-to-end speedups between 1.66x and 3.22x compared to the original.

74 ATOMIC AND MOLECULAR PHYSICS↗

Improved microscopic-macroscopic approach incorporating the effects of continuum states

The Woods-Saxon-Strutinsky method (the microscopic-macroscopic method) combined with the Kruppa prescription for positive-energy levels, which is necessary to treat neutron-rich nuclei, is studied to clarify the reason for its success and to propose improvements for its shortcomings. The reason why the plateau condition is met for the Nilsson model but not for the Woods-Saxon model is understood in a new interpretation of the Strutinsky smoothing procedure as a low-pass filter. Essential features of the Kruppa level density is extracted in terms of the Thomas-Fermi approximation modified to describe spectra obtained from diagonalization in truncated oscillator bases. A method is proposed, which weakens the dependence on the smoothing width by applying the Strutinsky smoothing only to the deviations from a reference level density. The BCS equations are modified for the Kruppa spectrum, which is necessary to treat the pairing correlation properly in the presence of a continuum. The potential depth is adjusted for the consistency between the microscopic and macroscopic Fermi energies. It is shown, with these improvements, that the microscopic-macroscopic method is now capable to reliably calculate binding energies of nuclei far from stability.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Electrosphere of macroscopic 'quark nuclei': A source for diffuse MeV emissions from dark matter

Using a Thomas-Fermi model, we calculate the structure of the electrosphere of the quark antimatter nuggets postulated to comprise much of the dark matter. This provides a single self-consistent density profile from ultrarelativistic densities to the nonrelativistic Boltzmann regime that we use to present microscopically justified calculations of several properties of the nuggets, including their net charge, and the ratio of MeV to 511 keV emissions from electron annihilation. We find that the calculated parameters agree with previous phenomenological estimates based on the observational supposition that the nuggets are a source of several unexplained diffuse emissions from the Galaxy. As no phenomenological parameters are required to describe these observations, the calculation provides another nontrivial verification of the dark-matter proposal. The structure of the electrosphere is quite general and will also be valid at the surface of strange-quark stars, should they exist.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Scalable training of graph convolutional neural networks for fast and accurate predictions of HOMO-LUMO gap in molecules

Abstract Graph Convolutional Neural Network (GCNN) is a popular class of deep learning (DL) models in material science to predict material properties from the graph representation of molecular structures. Training an accurate and comprehensive GCNN surrogate for molecular design requires large-scale graph datasets and is usually a time-consuming process. Recent advances in GPUs and distributed computing open a path to reduce the computational cost for GCNN training effectively. However, efficient utilization of high performance computing (HPC) resources for training requires simultaneously optimizing large-scale data management and scalable stochastic batched optimization techniques. In this work, we focus on building GCNN models on HPC systems to predict material properties of millions of molecules. We use HydraGNN, our in-house library for large-scale GCNN training, leveraging distributed data parallelism in PyTorch. We use ADIOS, a high-performance data management framework for efficient storage and reading of large molecular graph data. We perform parallel training on two open-source large-scale graph datasets to build a GCNN predictor for an important quantum property known as the HOMO-LUMO gap. We measure the scalability, accuracy, and convergence of our approach on two DOE supercomputers: the Summit supercomputer at the Oak Ridge Leadership Computing Facility (OLCF) and the Perlmutter system at the National Energy Research Scientific Computing Center (NERSC). We present our experimental results with HydraGNN showing (i) reduction of data loading time up to 4.2 times compared with a conventional method and (ii) linear scaling performance for training up to 1024 GPUs on both Summit and Perlmutter.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Machine learning for analysis of atomic spectral data

Physics based forward models are the basis on which many experimental diagnostics are interpreted. For some diagnostics, models can be computationally expensive which precludes their use in real time analysis. Reduced models have the potential to capture sufficient physics thereby enabling the desired real time analysis. Using statistical inference and machine learning techniques the application of reduced models for inversion of atomic spectral data used to diagnose magnetic fields in a plasma will be examined. Two approaches are considered, (a) a reduction of the forward model where traditional inversion can be performed on the proxy model, and (b) a reduction of the direct inverse where parameters are a function of measured signal. The resulting inversion is sufficiently fast to be utilized in an online context for digital twinning, and ultimately real-time prediction, design, and control of plasma systems, such as tokamaks. Furthermore, these methods will be demonstrated on both simulated and experimentally measured data.

74 ATOMIC AND MOLECULAR PHYSICS↗

Unraveling Deterministic Mesoscopic Polarization Switching Mechanisms: Spatially Resolved Studies of a Tilt Grain Boundary in Bismuth Ferrite

The deterministic mesoscopic mechanism of ferroelectric domain nucleation is probed at a single atomically-defined model defect: an artificially fabricated bicrystal grain boundary (GB) in an epitaxial bismuth ferrite film. Switching spectroscopy piezoresponse force microscopy (SS-PFM) is used to map the variation of local hysteresis loops at the GB and in its immediate vicinity. It is found that the the influence of the GB on nucleation results in a slight shift of the negative nucleation bias to larger voltages. The mesoscopic mechanisms of domain nucleation in the bulk and at the GB are studied in detail using phase-field modeling, elucidating the complex mechanisms governed by the interplay between ferroelectric and ferroelastic wall energies, depolarization fields, and interface charge. The combination of phase-field modeling and SS-PFM allows quantitative analysis of the mesoscopic mechanisms for polarization switching, and hence suggests a route for unraveling the mechanisms of polarization switching at a single defect level and ultimately optimizing materials properties through microstructure engineering.

Rodriguez, Brian↗

Multiscale and Multimodal Characterization of 2D Titanium Carbonitride MXene

Here, a comprehensive study on the prototype solid solution phase carbonitride MXene Ti 3 CN is conducted using nuclear magnetic resonance, electron spin resonance, total and quasi-elastic neutron scattering, combined with density functional theory-based electronic structure and molecular dynamic calculations. The combination of experiment and theory lead toward rational atomic structural models of Ti 3 CN. The remnant Al ions from the etching process significantly tune the interlayer spacing, distinct from the more typical MXene, Ti 3 C 2 , prepared similarly. Neutron scattering indicates the surface terminations of Ti 3 CN display high oxygen and fluorine concentrations and rather low hydroxyl and hydrogen concentrations. Calculations show that the structure including both the residual Al ions and mixed surface terminations give the best agreement with the measurements. The water molecules in Ti 3 CN are highly immobile, in strong contrast to those in Ti 3 C 2 . The analysis of the electronic structure suggests that the nitride MXene displays higher conductivity than the carbides. Finally, the absence of hydroxyl groups in terminations, the solid-solution in the anion sites, the remnants within layers, and immobile water altogether make the carbonitrides a unique series in the MXene family, implying a further exploration of their exotic properties and applications in energy storage.

36 MATERIALS SCIENCE↗

Crystal structure representations for machine learning models of formation energies

Here we introduce and evaluate a set of feature vector representations of crystal structures for machine learning (ML) models of formation energies of solids. ML models of atomization energies of organic molecules have been successful using a Coulomb matrix representation of the molecule. We consider three ways to generalize such representations to periodic systems: (i) a matrix where each element is related to the Ewald sum of the electrostatic interaction between two different atoms in the unit cell repeated over the lattice; (ii) an extended Coulomb-like matrix that takes into account a number of neighboring unit cells; and (iii) an ansatz that mimics the periodicity and the basic features of the elements in the Ewald sum matrix using a sine function of the crystal coordinates of the atoms. The representations are compared for a Laplacian kernel with Manhattan norm, trained to reproduce formation energies using a dataset of 3938 crystal structures obtained from the Materials Project. For training sets consisting of 3000 crystals, the generalization error in predicting formation energies of new structures corresponds to (i) 0.49, (ii) 0.64, and (iii) urn:x-wiley:00207608:media:qua24917:qua24917-math-0001 for the respective representations.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Direct Monte Carlo Simulations of Chemical Equilibrium Composition of Molecular Fluid Mixtures Under Shock Conditions

A new Monte Carlo simulation method is implemented which holds the number and kind of atoms constant rather than the number and kind of molecules for a molecular fluid mixture. Chemical reactions are allowed as a correlated move of atoms such as an interchange of atoms between molecules. Equilibrium chemical composition is determined as an average over the simulation along with equation of state quantities. Results are presented for N 2 + O 2 → 2NO at 30GPa and 3000K. The equilibrium composition is shown to be sensitive to the potentials between unlike species.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Role of local stresses on co-zone twin-twin junction formation in HCP magnesium

Twin-twin interactions control the formation of twin-twin junctions (TTJ) in hexagonal metals when multiple twin variants are activated in a grain. In this work, we employ a combination of two computational techniques, a 3D full-field crystal plasticity model (CP) and large-scale molecular dynamics (MD), to study the TTJ formation associated with two non-parallel {101¯2} twins in Mg. The local intra-granular stresses generated by discrete twins are computed using a spatially resolved CP model. Atomic-scale knowledge regarding formation processes and local stresses is revealed by MD. The combined analyses suggest that the twin junction forms by the migration of the boundaries of both, the impinging and impinged twin, taking place in the immediate vicinity of the contact point. It is further shown that local stress fields that are generated after initial contact promote thickening of the impinging twin, and may facilitate nucleation of a new twin on the opposite boundary of the recipient twin. In conclusion, calculations of the strain energies suggest that formation of the co-zone twin-twin junction is energetically favorable but detwinning of the TTJ upon load reversal or under cyclic loading is not.

36 MATERIALS SCIENCE↗

Novel allelic mutations in murine Serca2 induce differential development of squamous cell tumors

Dominant mutations in the Serca2 gene, which encodes sarco(endo)plasmic reticulum calcium-ATPase, predispose mice to gastrointestinal epithelial carcinoma [1–4] and humans to Darier disease (DD) [14–17]. In this study, we generated mice harboring N-ethyl-N-nitrosourea (ENU)-induced allelic mutations in Serca2: three missense mutations and one nonsense mutation. Mice harboring these Serca2 mutations developed tumors that were categorized as either early onset squamous cell tumors (SCT), with development similar to null-type knockout mice [2,4] (aggressive form; M682, M814), or late onset tumors (mild form; M1049, M1162). Molecular analysis showed no aberration in Serca2 mRNA or protein expression levels in normal esophageal cells of any of the four mutant heterozygotes. There was no loss of heterozygosity at the Serca2 locus in the squamous cell carcinomas in any of the four lines. The effect of each mutation on Ca{sup 2+}-ATPase activity was predicted using atomic-structure models and accumulated mutated protein studies, suggesting that putative complete loss of Serca2 enzymatic activity may lead to early tumor onset, whereas mutations in which Serca2 retains residual enzymatic activity result in late onset. We propose that impaired Serca2 gene product activity has a long-term effect on squamous cell carcinogenesis from onset to the final carcinoma stage through an as-yet unrecognized but common regulatory pathway. -- Highlights: •Novel mutations in murine Serca2 caused early onset or late onset of tumorigenesis. •They also caused higher or lower incidence of Darier Disease phenotype. •3D structure model suggested the former mutations led to severer defect on ATPase. •Driver gene mutations via long-range effect on Ca2+ distributions are suggested.

60 APPLIED LIFE SCIENCES↗

Experimental evidence for ‘buckle, ruck and tuck’ in neutron irradiated graphite

The current mainstream theory for radiation-induced dimensional change of nuclear graphite is based on the notion that interstitially displaced carbon atoms will coalesce into dislocation loops (i.e., additional basal planes). This standard atomic-displacement model has been challenged by theories based on first principles calculations. Furthermore, the so-called ‘ruck and tuck’ of basal planes has been proposed as an alternative mechanism to explain the observed c-axis expansion under irradiation; however, no such defects have been observed experimentally so far. In this study, the first experimental evidence for the presence of a ‘ruck and tuck’ defect in high-temperature neutron-irradiated nuclear graphite is introduced.

36 MATERIALS SCIENCE↗

How do energetic ions damage metallic surfaces?

Surface modification under bombardment by energetic ions observed under different conditions in structural and functional materials and can be either unavoidable effect of the conditions or targeted modification to enhance materials properties. Understanding basic mechanisms is necessary for predicting properties changes. The mechanisms activated during ion irradiation are of atomic scale and atomic scale modeling is the most suitable tool to study these processes. In this paper we present results of an extensive simulation program aimed at developing an understanding of primary surface damage in iron by energetic particles. We simulated 25 keV self-ion bombardment of Fe thin films with (100) and (110) surfaces at room temperature. A large number of simulations, ~400, were carried out allow a statistically significant treatment of the results. The particular mechanism of surface damage depends on how the destructive supersonic shock wave generated by the displacement cascade interacts with the free surface. Three basic scenarios were observed, with the limiting cases being damage created far below the surface with little or no impact on the surface itself, and extensive direct surface damage on the timescale of a few picoseconds. In some instances, formation of large <100> vacancy loops beneath the free surface was observed, which may explain some earlier experimental observations.

36 MATERIALS SCIENCE↗

Nanocrystal programmable assembly beyond hard spheres (or shapes) and other (simple) potentials

Ligands are the key to almost any strategy in the assembly of programmable nanocrystals (or nanoparticles) and must be accurately considered in any predictive model. Hard Spheres (or Shapes) provide the simplest and yet quite successful approach to assembly, with remarkable sophisticated predictions verified in experiments. There are, however, many situations where hard spheres/shapes predictions fail. This prompts three important questions: In what situations should hard spheres/shapes models be expected to work? and when they do not work, Is there a general model that successfully corrects hard sphere/shape predictions? and given other successful models where ligands are included explicitly, and of course, numerical simulations, can we unify hard sphere/shape models, explicit ligand models and all atom simulations? . Further, the Orbifold Topological Model (OTM) provides a positive answer to these three questions. In this paper, I give a detailed review of OTM, describing the concept of ligand vortices and how it leads to spontaneous valence and nanoparticle “eigenshapes” while providing a prediction of the lattice structure, without fitting parameters, which accounts for many body effects not captured by (two-body) potentials. I present a thorough survey of experiments and simulations and show that, to this date, they are in full agreement with the OTM predictions. I conclude with a discussion on whether NC superlattices are equilibrium structures and some significant challenges in structure prediction.

36 MATERIALS SCIENCE↗

Comparison of electron transport calculations in warm dense matter using the Ziman formula

The Ziman formulation of electrical conductivity is tested in warm and hot dense matter using the pseudo-atom molecular dynamics method. Several implementation options that have been widely used in the literature are systematically tested through a comparison to the accurate, but expensive Kohn–Sham density functional theory molecular dynamics (KS-DFT-MD) calculations. As a result, the comparison is made for several elements and mixtures and for a wide range of temperatures and densities, and reveals a preferred method that generally gives very good agreement with the KS-DFT-MD results, but at a fraction of the computational cost.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗