pnnl/HypoGen
An innovative workflow for automatic knowledge discovery in material science, combining experimental automation with advanced machine learning and artificial intelligence techniques
Yin, Tim [Pacific Northwest National Laboratory (P↗
SEARCH · Engineering Papers
Results for “material science”
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.
Calculating topological properties of artificial graphene in B-field
SAND2025-03258O Calculating topological properties of artificial graphene in B-field is a user-friendly tool designed to analyze artificial graphene systems influenced by magnetic fields. It helps researchers understand the unique properties of these materials by calculating the local Chern marker, a key indicator of their behavior. With just one file, this software runs easily on any device with Matlab, making it accessible for scientists and engineers. It provides valuable insights into the electronic characteristics of artificial graphene, which supports advancements in material science and technology, paving the way for innovative applications in electronics and beyond. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.
Spataru, Dan [Sandia National Lab. (SNL-CA), Liver↗
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↗
REDI – Readiness Engine for Data Integration
The Readiness Engine for Data Integration (REDI) is an open-source framework for automating, standardizing, and assessing the process of preparing scientific data for AI training. REDI implements a five-stage pipeline (ingest, preprocess, transform, structure, output) with per-stage provenance instrumentation via Flowcept, domain-aware transformation logic (PII anonymization, regridding, graph encoding, and more), and built-in readiness assessment and validation modes. REDI has been evaluated across climate, proteomics, materials science, and nuclear fusion datasets, demonstrating near-ideal parallel scaling to 100 nodes on OLCF's Frontier system. REDI is deployable as an agent-callable skill in coding environments such as Claude Code and OpenAI Codex, and is complemented by SetGo for FAIR compliance and catalog publication.
Brewer, Wesley [Oak Ridge National Laboratory (ORN↗
Protocols for isolating and characterizing polysaccharides from plant cell walls: a case study using rhamnogalacturonan-II
Abstract Background In plants, a large diversity of polysaccharides comprise the cell wall. Each major type of plant cell wall polysaccharide, including cellulose, hemicellulose, and pectin, has distinct structures and functions that contribute to wall mechanics and influence plant morphogenesis. In recent years, pectin valorization has attracted much attention due to its expanding roles in biomass deconstruction, food and material science, and environmental remediation. However, pectin utilization has been limited by our incomplete knowledge of its structure. Herein, we present a workflow of principles relevant for the characterization of polysaccharide primary structure using nature’s most complex polysaccharide, rhamnogalacturonan-II (RG-II), as a model. Results We outline how to isolate RG-II from celery and duckweed cell walls and from red wine using chemical or enzymatic treatments coupled with size-exclusion chromatography. From there, we applied mass spectrometry (MS)-based techniques to determine the glycosyl residue and linkage compositions of the intact RG-II and derived oligosaccharides including special considerations for labile monosaccharides. In doing so, we demonstrated that in the duckweed Wolffiella repanda the arabinopyranosyl (Ara p ) residue of side chain B is substituted at O- 2 with rhamnose. We used electrospray-MS techniques to identify non-glycosyl modifications including methyl-ethers, methyl-esters, and acetyl-esters on RG-II-derived oligosaccharides. We then showed the utility of proton nuclear magnetic resonance spectroscopy ( 1 H-NMR) to investigate the structure of intact RG-II and to complement the RG-II dimerization studies performed using size-exclusion chromatography. Conclusions The complexity of pectic polysaccharide structures has hampered efforts aimed at their valorization. In this work, we used RG-II as a model to demonstrate the steps necessary to isolate and characterize polysaccharides using chromatographic, MS, and NMR techniques. The principles can be applied to the characterization of other saccharide structures and will help inform researchers on how saccharide structure relates to functional properties in the future.
09 BIOMASS FUELS↗
Benchmarking the performance of uncertainty quantification methods for neural network-based interatomic potentials
Machine-learned interatomic potentials (ML-IAPs) continue to gain popularity as accurate, computationally efficient replacements for traditional, physics-based interatomic potentials and expensive ab initio methods. Uncertainty quantification (UQ) of ML-IAPs is a growing area of research as UQ is critical in many applications of IAPs, such as developing curated datasets, active learning-based data augmentation, self-improving models, and estimating the uncertainty of molecular dynamics simulations. In this paper, we construct and benchmark a series of different neural network potentials (NNPs) with varying network architectures to determine the performance of these models with respect to both the mean and uncertainty calibration error. Each NNP method is specifically designed to predict either epistemic or aleatoric uncertainty with particular focus on the differences in behavior between the epistemic and aleatoric uncertainty estimates. We benchmark these methods using multiple datasets common in the ML-IAP literature. The results show that the aleatoric uncertainty from single-shot model architectures is a competitive alternative to ensemble-based epistemic uncertainty predictions in regions of sufficient data-density. However, in regions where the representative data is sparse, aleatoric uncertainty models tend to overpredict and epistemic methods tend to underpredict the actual model error. We conclude that the type of UQ is crucial when discussing performance of probabilistic model results as different methods have different performance characteristics depending on the regime in which they are evaluated. Therefore, the type of UQ method should be carefully evaluated against both the data characteristics and requirements for the intended application.
97 MATHEMATICS AND COMPUTING↗
EvoDiffMol: evolutionary diffusion framework for 3D molecular design with optimized properties
Designing molecules with specific target properties remains a fundamental challenge in computational chemistry. While existing approaches show promise, most rely on simplified representations like SMILES strings or 2D graphs that lack essential three-dimensional geometric information. We present EvoDiffMol, a computational framework that integrates evolutionary algorithms with three-dimensional diffusion models for property-driven molecular generation. The method operates through adaptive evolutionary optimization, where population-based selection guides the generation process toward desired property landscapes. EvoDiffMol supports both unconstrained molecular design and scaffold-constrained generation that preserves fixed substructures while optimizing complementary regions. Comprehensive evaluation demonstrates exceptional performance, achieving the highest drug-likeness score (0.94) among all compared state-of-the-art methods while maintaining excellent validity, uniqueness, and novelty. Beyond single property optimization, the framework demonstrates flexible multi-property optimization capabilities, simultaneously controlling multiple molecular descriptors including synthetic accessibility, lipophilicity, topological polar surface area, and clinically relevant ADMET properties such as cardiotoxicity (hERG) and intestinal permeability (Caco-2). This adaptability spans from simple descriptors to practical pharmaceutical endpoints without requiring complete model retraining. The framework achieves precise control over target property values, generating molecules with properties closely matching specified targets for both single and multiple descriptors. Scaffold-constrained experiments preserve fixed molecular cores while maintaining effective property optimization. The three-dimensional representation offers advantages in maintaining structural validity during iterative optimization, with potential for geometry-aware applications in materials science and drug discovery.
3D molecular generation↗
Hydrogen transport in yttrium hydride under asymmetric heat
Metal hydrides are a promising moderator material for high temperature fission reactors. Yttrium hydride can be loaded to a high hydrogen density with relatively high hydrogen stability at temperatures up to 800°C. This makes yttrium hydride a potential moderator material for microreactors as foreseen in the fission surface power program. However, during the operation of such advanced reactors temperature gradients are expected which can change the local hydrogen density in the moderator. Hydrogen diffusion in metals is driven by a concentration gradient (Fick’s law) and thermal diffusion (Soret diffusion). Thermal diffusion is the transport of hydrogen, or other species, due to a temperature gradient. For example, hydrogen might migrate from the hot side of a sample to the cold side of a sample. Measuring Fickian diffusion is achieved through various permeation or absorption experiments, however measuring thermal diffusion is challenging and has rarely been performed. The Hydrogen Experimental Apparatus for Thermal Diffusion (HEATD) experiment is designed to induce thermal diffusion in samples and quench those samples so that the hydrogen distribution can be analyzed using hot vacuum extraction (HVE). One side of the sample was heated to a high temperature e.g., 800°C, while the other side of the sample is at a lower temperature. The sample was held under the applied temperature gradient for a given time until the anticipated hydrogen diffusion has occurred. The actual time depends depend on the sample composition and hydrogen concentration. The results from HVE showed that thermal diffusion took place in the specimen and the Soret coefficient was calculated.
36 MATERIALS SCIENCE↗
Subsurface mechanical damage of fused silica glass during grinding by various sub-aperture tools with and without ultrasonics
The subsurface mechanical damage (SSD) depth after grinding fused silica glass with a comprehensive set of sub-aperture fixed abrasive grinding tools [cup, wheel, belt, pad, and rotary face mill (with and without ultrasonics)] and process parameters has been statistically measured using the taper wedge technique and evaluated. Consistent with a previously reported grinding model [J. Non-Cryst. Solids 352, 5601–5617 (2006) Crossref , Materials Science and Technology of Optical Fabrication (Wiley & Sons, 2018)], based on the sliding indentation fracture by sliding particles or asperities where the normal load per particle determines the depth of the fracture (and ultimately the overall SSD depth), the dominant factor controlling SSD depth was found to be the abrasive size regardless of the tool type and process conditions. Compared to full aperture grinding methods, the overall SSD depth was higher using the sub-aperture tools, likely due to the higher effective pressure and higher load per particle distribution. Here, in addition to abrasive size, a significant reduction in SSD depth was achieved by: (1) reducing the load distribution on the abrasive particles via increase in contact area and/or decrease in mechanical loading; (2) using a more compliant host tool medium; and (3) in what we believe is a more novel way, using ultrasonics. Combining low abrasive size, larger contact area, and a compliant host, the 6 µm diamond in a resin matrix (Trizact) on a foam pad led to very low SSD depth (~ 4.6 µm), relatively fast grinding rate (186 mm 3 /h), and little or no figure degradation. This grinding tool/process is an attractive choice for final grind, resulting in significantly reduced polish out (i.e., “grey out”) times. With the rotary face mill tool, the use of ultrasonics consistently led to a SSD depth reduction (ranging from 17%–34%). A new fracture mechanics-based model, to the best of our knowledge, where the relevant normal load is parallel to the feed direction, has been developed to explain how ultrasonics leads to lower SSD depth. The key factors, supported by finite element stress analysis and load measurements, are (1) the initiation of fractures at higher z heights during the tool’s ultrasonic vertical oscillations, thus propagating less deep into workpiece; (2) reduction in load (and therefore reduction in fracture propagation distance) due to smaller tool-workpiece feed direction contact area (again caused by higher heights relative to depth of cut); (3) upward movement of the tool during oscillation leads to fracturing toward the surface instead into the depth; and finally (4) at tool’s lowest point of oscillation cycle, there may not be enough time for the fracture to propagate to its full length.
Optics and optical instruments↗
High-precision inversion of dynamic radiography using hydrodynamic features
While radiography is routinely used to probe complex, evolving density fields in research areas ranging from materials science to shock physics to inertial confinement fusion and other national security applications, complications resulting from noise, scatter, complex beam dynamics, etc. prevent current methods of reconstructing density from being accurate enough to identify the underlying physics with sufficient confidence. In this work, we show that using only features that are robustly identifiable in radiographs and combining them with the underlying hydrodynamic equations of motion using a machine learning approach of a conditional generative adversarial network (cGAN) provides a new and effective approach to determine density fields from a dynamic sequence of radiographs. In particular, we demonstrate the ability of this method to outperform a traditional, direct radiograph to density reconstruction in the presence of scatter, even when relatively small amounts of scatter are present. Our experiments on synthetic data show that the approach can produce high quality, robust reconstructions. We also show that the distance (in feature space) between a testing radiograph and the training set can serve as a diagnostic of the accuracy of the reconstruction.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
Physics-assisted generative adversarial network for X-ray tomography
X-ray tomography is capable of imaging the interior of objects in three dimensions non-invasively, with applications in biomedical imaging, materials science, electronic inspection, and other fields. The reconstruction process can be an ill-conditioned inverse problem, requiring regularization to obtain satisfactory results. Recently, deep learning has been adopted for tomographic reconstruction. Unlike iterative algorithms which require a distribution that is known a priori , deep reconstruction networks can learn a prior distribution through sampling the training distributions. In this work, we develop a Physics-assisted Generative Adversarial Network (PGAN), a two-step algorithm for tomographic reconstruction. In contrast to previous efforts, our PGAN utilizes maximum-likelihood estimates derived from the measurements to regularize the reconstruction with both known physics and the learned prior. Compared with methods with less physics assisting in training, PGAN can reduce the photon requirement with limited projection angles to achieve a given error rate. The advantages of using a physics-assisted learned prior in X-ray tomography may further enable low-photon nanoscale imaging.
47 OTHER INSTRUMENTATION↗
Predicting ptychography probe positions using single-shot phase retrieval neural network
Ptychography is a powerful imaging technique that is used in a variety of fields, including materials science, biology, and nanotechnology. However, the accuracy of the reconstructed ptychography image is highly dependent on the accuracy of the recorded probe positions which often contain errors. These errors are typically corrected jointly with phase retrieval through numerical optimization approaches. When the error accumulates along the scan path or when the error magnitude is large, these approaches may not converge with satisfactory result. We propose a fundamentally new approach for ptychography probe position prediction for data with large position errors, where a neural network is used to make single-shot phase retrieval on individual diffraction patterns, yielding the object image at each scan point. The pairwise offsets among these images are then found using a robust image registration method, and the results are combined to yield the complete scan path by constructing and solving a linear equation. We show that our method can achieve good position prediction accuracy for data with large and accumulating errors on the order of 10 2 pixels, a magnitude that often makes optimization-based algorithms fail to converge. For ptychography instruments without sophisticated position control equipment such as interferometers, our method is of significant practical potential.
47 OTHER INSTRUMENTATION↗
Reconstructing Richtmyer–Meshkov instabilities from noisy radiographs using low dimensional features and attention-based neural networks
We develop an ML-based approach for density reconstruction based on transformer neural networks. This approach is demonstrated in the setting of ICF-like double shell hydrodynamic simulations wherein the parameters related to material properties and initial conditions are varied. The new method can robustly recover the complex topologies given by the Richtmyer-Meshkoff instability (RMI) from a sequence of hydrodynamic features derived from radiographic images corrupted with blur, scatter, and noise. A noise model is developed to characterize errors in extracting features from synthetic radiographs of the simulated density field. The key component of the network is a transformer encoder that acts on a sequence of features extracted from noisy radiographs. This encoder includes numerous self-attention layers that act to learn temporal dependencies in the input sequences and increase the expressiveness of the model. This approach is shown to exhibit an excellent ability to accurately recover the RMI growth rates, despite the gas-metal interface being greatly obscured by radiographic noise. Our approach can be applied in a broad array of fields involving shock physics and material science.
47 OTHER INSTRUMENTATION↗
Specificity and tunability of efflux pumps: A new role for the proton gradient?
Efflux pumps that transport antibacterial drugs out of bacterial cells have broad specificity, commonly leading to broad spectrum resistance and limiting treatment strategies for infections. It remains unclear how efflux pumps can maintain this broad spectrum specificity to diverse drug molecules while limiting the efflux of other cytoplasmic content. We have investigated the origins of this broad specificity using theoretical models informed by the experimentally determined structural and kinetic properties of efflux pumps. We developed a set of mathematical models describing operation of efflux pumps as a discrete cyclic stochastic process across a network of states characterizing pump conformations and the presence/absence of bound ligands and protons. These include a minimal three-state model that lends itself to clear analytic calculations as well as a five-state model that relaxes some of the simpler model’s most strict assumptions. We found that the pump specificity is determined not solely by the drug affinity to the pump–as is commonly assumed–but it is also directly affected by the periplasmic pH and the transmembrane potential. Therefore, changes to the proton concentration gradient and voltage drop across the membrane can influence how effective the pump is at extruding a particular drug molecule. Furthermore, we found that while both the proton concentration gradient across the membrane and the transmembrane potential contribute to the thermodynamic force driving the pump, their effects on the efflux enter not strictly in a combined proton motive force. Rather, they have two distinguishable effects on the overall throughput. These results highlight the unexpected effects of thermodynamic driving forces out of equilibrium and illustrate how efflux pump structure and function are conducive to the emergence of multidrug resistance.
59 BASIC BIOLOGICAL SCIENCES↗
Spatiotemporal imaging and manipulation of surface plasmons
Abstract Surface plasmon polaritons (SPPs) are travelling surface waves that have shown promise for applications in nanophotonics as they provide a direct route toward photon-mediated electrical signal generation – a central paradigm for speeding up and scaling down photonic elements. SPP waves have also generated fundamental interest due to their high-field strength and sub-wavelength confinement, properties that have enabled the surface-enhanced Raman effect. Over the last decade, photoemission electron microscopy (PEEM) has emerged as a pioneering technique for imaging surface electric fields through ultrashort laser pulse mediated electron emission and has therefore become an indispensable tool for characterizing plasmonic phenomena at interfaces in a variety of materials. PEEM offers nanometer spatial resolution and femtosecond temporal resolution, allowing SPPs to be prepared, monitored, and manipulated on the nanometer-femtosecond scale. Through a brief review of recent reports, we aim to introduce PEEM-based SPP imaging and manipulation modalities and highlight their utility in the context of emerging nanoscale and quantum materials science advancements.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
Are We on Track for 2050? A Materials Research & Sustainability Perspective
In commemoration of the Materials Research Society (MRS)'s 50th anniversary, the 2050 panel hosted a discussion to reflect on the past, present, and future of sustainability and the role of materials research and development. Three panelists discussed their views, based on their expertise, about future challenges and lessons from the past. Sustainable development is a broad topic; therefore, the discussion centered on their experience as material researchers and their efforts for a better and greener future. This work is developed in collaboration with the co-authors team, highlighting the need for accelerating research and development efforts, especially in materials science and applications, fostering interdisciplinary partnerships, and mobilizing collective action to address the complex and interconnected sustainability challenges that humanity is currently facing.
ENERGY PLANNING, POLICY, AND ECONOMY,ENVIRONMENTAL↗
Machine Learning-Based Identification of the Interface Regions for Coupling Local and Nonlocal Models
Local-nonlocal coupling approaches provide a means to combine the computational efficiency of local models and the accuracy of nonlocal models. However, the coupling process can be challenging, requiring expertise to identify the interface between local and nonlocal regions. Here, this study introduces a machine learning-based approach to automatically detect the regions in which the local and nonlocal models should be used. The method uses loading functions evaluated at grid points to decide the model selection at those points. Training of the networks is based on datasets provided by classes of loading functions for which reference coupling configurations are computed using accurate coupled solutions, where accuracy is measured in terms of the relative error between the solution to the coupling approach and the solution to the nonlocal model. We study two approaches that vary in data structure. The first, the full-domain input data approach, uses the entire load vector and outputs a complete label vector, performing a global classification. The second, a window-based approach, processes loads into windows and addresses the problem as a node-wise classification where each window's central point is classified individually. The classification problems are solved via deep learning algorithms based on convolutional neural networks. The performance of these approaches is studied on one-dimensional numerical examples using F1-scores and accuracy metrics. Notably, the windowing approach achieves an accuracy of 0.96 and an F1-score of 0.97, highlighting its potential to automate coupling processes effectively and enhance computational efficiency in material science applications.
97 MATHEMATICS AND COMPUTING↗
A 15-Mev/nucleon Iso-Cyclotron for Security and Radioisotope Production
Cargo inspection systems exploit the broad bremsstrahlung spectrum from a 6-10 MeV, low-duty cycle electron accelerator which in the presence of significant backgrounds presents challenges in image and material identification. An alternative approach is to use ions which can excite nuclear states either directly, or through generation of secondary high-energy signature gammas produced from nuclear interactions in a target. RadiaBeam is designing a compact sector isocyclotron 1.25 m in radius, with high-gradient cavities to accelerate multi-ion species up to 15-20 MeV/u with large turn-to turn, centimeter-level separation for low-loss extraction without lossy foil stripping. A strong-focusing radial field profile will be optimized in a separated-sector format for control over machine tune simultaneous with isochronous orbist requirements for high-current (~0.5 milliamp) operation. Innovation in injection will be introduced to replace the high-loss central region. Non-security applications of the cyclotron include medical isotope production, ion radiobiology, as well as material science research and ion instrumentation development.
43 PARTICLE ACCELERATORS↗