Engineering Papers⌕ Search

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.

At least 1,657 records · Page 92

Ensemble Simulations on Leadership Computing Systems

Scientific productivity can be enhanced through workflow management tools, relieving large High Performance Computing (HPC) system users from the tedious tasks of scheduling and designing the complex computational execution of scientific applications. This paper presents a study on the usage of ensemble workflow tools to accelerate science using the Summit and Frontier supercomputing systems. The research aims to connect science domain simulations using Oak Ridge Leadership Computing Facility (OLCF) supercomputing platforms with ensemble workflow methods in order to accelerate HPC-enabled discovery and boost scientific impact. We present the coupling, porting and optimization of Radical-Cybertools on three applications: Chroma, NAMD and LAMMPS. The tools augment traditional HPC monolithic runs with a pilot scheduler. Lessons-learned are discussed for physics, biology and materials science applications. We discuss intrinsic limitations of coupling and porting ensemble workflow tools to applications that run on large HPC systems. The origins of technical challenges and their solutions developed during the implementation process are discussed. Data management strategies, OLCF’s policies for ensembles, and natively supported workflow tools are also summarized.

Georgiadou, Antigoni [ORNL] (ORCID:000000020977631↗

Superatomic chemistry

Superatoms are atomic clusters with tailored size and composition that mimic the chemistry of atoms in the periodic table. However, unlike the atoms whose chemistry is governed by their valence electron orbitals, the chemistry of superatoms is governed by their highest occupied molecular orbitals. In addition, due to their large size and non-spherical geometry, superatoms can promote unusual reactions and serve as the building blocks of cluster assembled materials with properties very different from conventional materials. This perspective highlights the unique role of superatoms in chemical and material sciences by focusing on superhalogens, which not only possess electron affinities larger than those of halogens but also can be stable when multiply charged. We discuss how these unique features of superhalogens enable noble gas atoms like argon to form chemical bonds at room temperature and zinc to exhibit an oxidation state of +3. Here, the advantages of using superhalogens in the synthesis of water-resistant materials for solar cells, halogen-free electrolytes for solid-state batteries, and multiferroic materials are also discussed.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Geranium Photodiodes for Hard Xray Detection

This report summarizes the investigations engaged at UC Davis under subcontract B632083, as related to the LLNL-LDRD Project “High-Speed X-Ray Imager Arrays.” The first section entails work done within the Electrical and Computer Engineering (ECE) Department. The purpose of the work done so far is to understand photodiode design and which parameters are conducive to the design goals, which are: I: Maximize the detector quantum efficiency. The device must have a quantum efficiency of roughly 5% at 80 keV. II: Minimize temporal response. Collection time is limited to 1ns. All charge must be collected in that time. III: Minimize dark current. In this document, various photodiode parameters and designs are explored, and their impact on these goals are tabulated. In addition to these, various instruments that have been constructed for either testing these parameters or helping teammates with their experiments. The second section entails work done with the Materials Science Engineering (MSE) Department. In accordance with the proposed development of the germanium ROIC backside imagers, a critical step in the micromachining of these devices was the removal of the substrate on which the PIN junctions were epitaxially grown. A common technique for the micromachining of silicon wafers is the use of chemical etching alongside an “etch-stop” layer. In this technique, a chemical etchant will remove material until it reaches the etch-stop layer, at which the etch rate will drop significantly. This high selectivity (ratio of etch rates) allows any unevenness in the initial etching to be effectively smoothed away in the event that part of the wafer surface reaches the etch-stop layer sooner than another. An initial literature review found a paper by Divan et al. in which the authors used ion implantation to dope germanium wafers and achieve selectivity on the order of 100, sufficient for a reliable etch-stop.

42 ENGINEERING↗

Orbital Engineering Band Degeneracy in a Dual-Square Carbon-Oxide Framework

Electron band degeneracies in momentum space give rise to exotic quantum phenomena that have sparked intense interest in condensed matter physics and materials science. Nodal-lines─isolines in k-space formed by the incidental touching of two bands that share the same energy but belong to discrete eigenstates─arise in the presence of symmetries that preclude effective hybridization. Despite recent advances in the design, bottom-up assembly, and engineering of exotic electronic states in graphene nanomaterials, the extension of this approach to access synthetic two-dimensional (2D) quantum materials derived from metal- or covalent-organic frameworks (COFs) has lagged behind. Here we present a molecular orbital engineering approach for designing and fabricating an edge-centered dual square lattice within a π-conjugated 2D-tetraoxa[8]circulene (2D-TOC) COF. First-principles calculations and scanning tunnelling spectroscopy reveal the emergence of Frontier states at the center of a 3 × 3 lattice that give rise to Dirac nodal-lines in 2D-TOC. Our findings not only provide a general guide for the design of conjugated COFs with custom tailored electronic properties from molecular fragments but enable the exploration of emergent topological phenomena in synthetic 2D materials with potential application for high-speed, low-power data processing, transmission, and storage.

Dirac nodal line semimetal↗

First Post-Flight Status Report for the Microgravity Science Glovebox

The Microgravity Science Glovebox (MSG) was launched to the International Space Station (ISS) this year on the second Utilization Flight (UF2). After successful on-orbit activation, the facility began supporting an active microgravity research program. The inaugural NASA experiments operated in the unit were the Solidification Using a Baffle in Sealed Ampoules (SUBSA, A. Ostrogorski, PI), and the Pore Formation and Mobility (PFMI, R. Grugel, PI) experiments. Both of these materials science investigations demonstrated the versatility of the facility through extensive use of telescience. The facility afforded the investigators with the capability of monitoring and operating the experiments in real-time and provided several instances in which the unique combination of scientists and flight crew were able to salvage situations which would have otherwise led to the loss of a science experiment in an unmanned, or automated, environment. The European Space Agency (ESA) also made use of the facility to perform a series of four experiments that were carried to the ISS via a Russian Soyuz and subsequently operated by a Belgium astronaut during a ten day Station visit. This imaginative approach demonstrated the ability of the MSG integration team to handle a rapid integration schedule (approximately seven months) and an intensive operations interval. Interestingly, and thanks to aggressive attention from the crew, the primary limitation to experiment thru-put in these early operational phases is proving to be the restrictions on the up-mass to the Station, rather than the availability of science operations.

Baugher, Charles R., III↗

The Reconstruction and Failure Analysis of the Space Shuttle Columbia

Several days following the Columbia accident a team formed and began planning for the reconstruction of Columbia. A hangar at the Kennedy Space Center was selected for this effort due to it's size, available technical workforce and materials science laboratories and access to the vehicle ground processing infrastructure. The Reconstruction team established processes for receiving, handling, decontamination, tracking, identifying, cleaning and assessment of the debris. Initially, a 2-dimensional reconstruction of the Orbiter outer mold line was developed. As the investigation progressed fixtures which allowed a 3-dimensional reconstruction of the forward portions of the left wing's leading edge was developed. To support the reconstructions and forensic analyses a Materials and Processes (M&P) 'team was formed. This M&P team established processes for recording factual observations, debris cleaning, and engineering analysis. Fracture surfaces and thermal effects of selected airframe debris were assessed, and process flows for both nondestructive and destructive sampling and evaluation of debris were developed. The Team also assessed left hand airframe components that were believed to be associated with a structural breach of Columbia. A major portion of this analysis was evaluation of metallic deposits were prevalent on left wing leading edge components. Extensive evaluation of the visual, metallurgical and chemical nature of the deposits provided conclusions that were consistent with the visual assessments and interpretations of the NASA lead teams and the findings of the Columbia Accident Investigation Board. Analytical data collected by the M&P Team showed that a significant thermal event occurred at the left wing leading edge in the proximity of LH RCC Panels 8-9, and a correlation was formed between the deposits and overheating in these areas to the wing leading edge components. The analysis of deposits also showed exposure to temperatures in excess of 1649 C (3200 F), which would severely degrade support structure, tiles, and RCC panel materials. The integrated failure analysis of wing leading edge debris and deposits strongly supported the hypothesis that a breach occurred at LH RCC Panel 8.

Russell, Richard↗

Quantum computing hardware for HEP algorithms and sensing

Quantum information science harnesses the principles of quantum mechanics to realize computational algorithms with complexities vastly intractable by current computer platforms. Typical applications range from quantum chemistry to optimization problems and also include simulations for high energy physics. The recent maturing of quantum hardware has triggered preliminary explorations by several institutions (including Fermilab) of quantum hardware capable of demonstrating quantum advantage in multiple domains, from quantum computing to communications, to sensing. The Superconducting Quantum Materials and Systems (SQMS) Center, led by Fermilab, is dedicated to providing breakthroughs in quantum computing and sensing, mediating quantum engineering and HEP based material science. The main goal of the Center is to deploy quantum systems with superior performance tailored to the algorithms used in high energy physics. In this Snowmass paper, we discuss the two most promising superconducting quantum architectures for HEP algorithms, i.e. three-level systems (qutrits) supported by transmon devices coupled to planar devices and multi-level systems (qudits with arbitrary N energy levels) supported by superconducting 3D cavities. For each architecture, we demonstrate exemplary HEP algorithms and identify the current challenges, ongoing work and future opportunities. Furthermore, we discuss the prospects and complexities of interconnecting the different architectures and individual computational nodes. Finally, we review several different strategies of error protection and correction and discuss their potential to improve the performance of the two architectures. This whitepaper seeks to reach out to the HEP community and drive progress in both HEP research and QIS hardware.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Sequential optical response suppression for chemical mixture characterization

The characterization of mixtures of non-interacting, spectroscopically similar quantum components has important applications in chemistry, biology, and materials science. We introduce an approach based on quantum tracking control that allows for determining the relative concentrations of constituents in a quantum mixture, using a single pulse which enhances the distinguishability of components of the mixture and has a length that scales linearly with the number of mixture constituents. To illustrate the method, we consider two very distinct model systems: mixtures of diatomic molecules in the gas phase, as well as solid-state materials composed of a mixture of components. A set of numerical analyses are presented, showing strong performance in both settings.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Computationally Guided Development of Components for High Energy Density Solid-State Lithium-Sulfur Batteries

All electric vertical take-off and landing vehicles (eVTOL) for urban air mobility (UAM) concepts face numerous challenging technical barriers before their introduction into the consumer marketplace. The most challenging of these technical barriers to overcome is developing an energy storage system capable of meeting the rigorous aerospace safety and performance criteria1. The performance metrics for eVTOL craft, such as specific energy, specific power, and safety, exceed those of electric automobiles by a factor of two to four. Current state-of-the-art (SOA) lithium-ion batteries are incapable of meeting the key performance criteria of energy and safety for eVTOL. Therefore, next generation advanced chemistries and designs must be developed to meet required performance metrics for electric aviation2. Beyond lithium-ion chemistries, such as lithium-sulfur, show promise in their high energy, while limitations exist in their power and cyclability due to low electrical conductivity and high intermediate solubility in organic liquid electrolytes. Several strategies to overcome the low electrical conductivity involve the use of selenium as a dopant in the active sulfur material, along with the incorporation of 2-dimensional electron-conducting holey-graphene to improve the composite cathodes electronic conductivity. Furthermore, combining this chemistry with a solid-electrolyte avoids the components’ dissolution issues3. However, the development of composite solid-state cathodes is non-trivial as several components must be intimately mixed so that the active component has sufficient access to both electrons and lithium ions to undergo full electrochemical conversion. Mathematical modeling of battery components can assist experimental design through a robust and rigorous combination of computational modeling techniques covering multiple length scales. The objective is to leverage modern computational materials methods combined with battery multiphysics tools to develop radically advanced compatible cathode and electrolyte materials, build and test solid state lithium-sulfur cells and packs. A NASA-based cross-organizational team of high-powered experts combined integrated computational predictive modeling, fundamental chemistry analysis, advanced material science, and battery cell development to tackle this very challenging, multidisciplinary problem. This presentation will show a multiscale computational modeling approach that has produced a novel particle dynamics method called Solid Electrolyte Sphere Approximation Model (SESAM). SESAM modeling targets the 1-10 µm scale structures and provides electromechanical and grain interactions for predictive design guidelines for the manufacturing of solid-state components. Parameters such as particle size and volume fraction of the constituent materials were modeled and experimentally fabricated to optimize electrochemical performance through improved microstructure design. Experimental feedback was provided through ionic and electronic conductivity assessment and structural analysis of developed materials and cell components.

battery↗

Imaging velocity interferometer system for any reflector (VISAR) diagnostics for high energy density sciences

Two variants of optical imaging velocimetry, specifically the one-dimensional streaked line-imaging and the two-dimensional time-resolved area-imaging versions of the Velocity Interferometer System for Any Reflector (VISAR), have become important diagnostics in high energy density sciences, including inertial confinement fusion and dynamic compression of condensed matter. Here, we give a brief review of the historical development of these techniques, then describe the current implementations at major high energy density (HED) facilities worldwide, including the OMEGA Laser Facility and the National Ignition Facility. We illustrate the versatility and power of these techniques by reviewing diverse applications of imaging VISARs for gas-gun and laser-driven dynamic compression experiments for materials science, shock physics, condensed matter physics, chemical physics, plasma physics, planetary science and astronomy, as well as a broad range of HED experiments and laser-driven inertial confinement fusion research.

47 OTHER INSTRUMENTATION↗

Digital Platform Informed Certification of Components Derived from Advanced Manufacturing Technologies

The Transformational Challenge Reactor is being designed at Oak Ridge National Laboratory to demonstrate the feasibility of constructing a reactor core using advanced manufacturing technology. This technology includes additive manufacturing combined with machine learning, materials science, and data science technologies in an effort to facilitate the expansion of additive manufacturing into advanced nuclear energy systems and other applications requiring a high level of quality assurance. The Transformational Challenge Reactor is employing additive manufacturing and artificial intelligence to deliver a new approach. Beginning in FY21, the focus of the program has shifted away from demonstrating a reactor, and instead, towards delivering on four key thrust areas: (1) artificial intelligence-informed design, (2) advanced materials, (3) integrated sensing and control, and (4) the digital platform. Of these four thrust areas, the most pertinent to this report is the digital platform. The digital platform has the potential to be a key enabler for a paradigm shift in how components, those derived from advanced manufacturing technologies, are certified for use in nuclear applications. This is achieved primarily using machine learning to discover correlations from the abundance of data produced through additive manufacturing and those physical properties critical to the performance of the component.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Distributed quantum approximate optimization algorithm on a quantum-centric supercomputing architecture

Quantum approximate optimization algorithm (QAOA) has shown promise in solving combinatorial optimization problems by providing quantum speedup on near-term gate-based quantum computing systems. However, QAOA faces challenges for high-dimensional problems due to the large number of qubits required and the complexity of deep circuits, limiting its scalability for real-world applications. In this study, we present a distributed QAOA (DQAOA), which leverages distributed computing strategies to decompose a large computational workload into smaller tasks that require fewer qubits and shallower circuits than are necessary to solve the original problem. These sub-problems are processed using a combination of high-performance and quantum computing resources. The global solution is iteratively updated by aggregating sub-solutions, allowing convergence toward the optimal solution. We demonstrate that DQAOA can handle considerably large-scale optimization problems (e.g., 1000-bit problem), achieving a high solution quality and short time-to-solution, outperforming existing strategies. Furthermore, we realize DQAOA on a quantum-centric supercomputing architecture, paving the way for practical applications of gate-based quantum computers in real-world optimization tasks. To extend DQAOA’s applicability to materials science, we further develop an active learning algorithm integrated with our DQAOA (AL-DQAOA), which involves machine learning, DQAOA, and active data production in an iterative loop. We successfully optimize photonic structures using AL-DQAOA, indicating that solving real-world optimization problems using gate-based quantum computing is feasible. We expect the proposed DQAOA to be applicable to a wide range of optimization problems and AL-DQAOA to find broader applications in material design.

Kim, Seongmin [ORNL] (ORCID:0000000159063004)↗

Neutron-Resonance Transmission Analysis with a Compact Deuterium-Tritium Neutron Generator

Neutron Resonance Transmission Analysis (NRTA) is a spectroscopic technique which uses the resonant absorption of neutrons in the epithermal range to infer the isotopic composition of an object. This spectroscopic technique has relevance in many traditional fields of science and nuclear security. NRTA in the past made use of large, expensive accelerator facilities to achieve precise neutron beams, significantly limiting its applicability. Here, we describe a series of NRTA experiments where we use a compact, low-cost deuterium-tritium (DT) neutron generator to produce short neutron beams (2.6 m) along with a 6 Li-glass neutron detector. The time-of-flight spectral data from five elements – silver, cadmium, tungsten, indium, and 238 U – clearly show the corresponding absorption lines in the 1-30 eV range. The experiments show the applicability of NRTA in this simplified configuration, and prove the feasibility of this compact and low-cost approach. This could significantly broaden the applicability of NRTA, and make it practical and applicable in many fields, such as material science, nuclear engineering, and arms control.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

"Electrostructural Phase Changes" In Charged Particulate Clouds: Planetary and Astrophysical Implications

There is empirical evidence that freely-suspended triboelectrostatically charged particulate clouds of dielectric materials undergo rapid conversion from (nominally) monodispersed "aerosols" to a system of well-defined grain aggregates after grain motion or fluid turbulence ceases within the cloud. In United States Microgravity Laboratory Space Shuttle experiments USML-1 and USML-2, it was found that ballistically-energized grain dispersions would rapidly convert into populations of filamentary aggregates after natural fluid (air) damping of grain motion. Unless continuously disrupted mechanically, it was impossible to maintain a non-aggregated state for the grain clouds of sand-size materials. Similarly, ground- based experiments with very fine dust-size material produced the same results: rapid, impulsive "collapse" of the dispersed grains into well-defined filamentary structures. In both ground-based and microgravity experiments, the chains or filaments were created by long-range dipole electrostatic forces and dipole-induced dielectric interactions, not by monopole interactions. Maintenance of the structures was assisted by short-range static boundary adhesion forces and van der Waals interactions. When the aggregate containers in the USML experiments were disturbed after aggregate formation, the quiescently disposed filaments would rearrange themselves into fractal bundles and tighter clusters as a result of enforced encounters with one another. The long-range dipole interactions that bring the grains together into aggregates are a product of randomly-distributed monopole charges on the grain surfaces. In computer simulations, it has been shown that when the force vectors of all the random charges (of both sign) on a grain are resolved mathematically by assuming Coulombic interaction between them, the net result is a dipole moment on individual grains, even though the grains are electrically neutral insofar as there is no predominance, on their surface, of one charge sign over another. The random charges of both sign derive from natural grain-to-grain interactions that produce triboelectrification via charge exchange every time grain surfaces make contact with one another. The conversion from a random distribution of grains (upon which there are randomly distributed charges) into an organization of electrostatically-ordered aggregates, can be regarded (within the framework of granular-material science) as an "electrical or Coulombic phase change" of the particulate cloud. It is not totally dissimilar from the more normal phase-change concept in which, for example, a gas with long free-path-molecules suddenly becomes a solid as a result of structural ordering of the molecules (notably, also the result of electronic forces, albeit at a different scale). In both the gas-to-solid case, and the aerosol-to-aggregate case, the same materials and charges are present before and after the phase change, but their arrangement now has a higher degree of order and a lower-energy configuration. An input of energy into the system is required to reverse the situation. The aggregates in the USML experiments were observed to undergo at least two phase changes as noted above. The point about phase changes, and by implication, the "electrostructural" reorganizations in particulate clouds, is the following: (a) they can occur very rapidly, almost spontaneously, above a critical cloud density, (b) in going from a higher energy state to a lower energy state, they convert to a denser system, (c) energy must be required to reverse the situation, implying that energy is released during the high-to-low energy phase change. In applying this information to natural particulate clouds, some inferences can be made (it is stressed that reference is still to dielectric materials attracted by dipole forces). There are several natural settings to which the USML observations apply, and to which the phase-change implications likewise apply. Dense clouds of triboelectrically-charged, kinetically-energized grains are to be found in volcanic eruptions (particularly on earth), aeolian dust storms (particularly on Mars), meteorite impact ejecta curtains (on all planets), in "immature" debris rings around planets (e.g., that from which our own Moon may have condensed), and in gravitationally collapsing protoplanetary dust/planetesimal debris disks where dielectric granules are being increasingly brought into collisional relationships with one another (increasing both charge exchange and physical proximity). It is noted that in many of these cases, the degree of electrical charging on the grains is likely to be much higher than that in the USML experiments where charging was not enhanced above the "normal", naturally encountered level for the particular materials (quartz, glass, and various silicate minerals). Application of the phase-change concept suggests that volcanic, aeolian, and impact debris clouds may, under certain circumstances, undergo rapid, impulsive, or even catastrophic collapse into a denser state that will lead to rapid precipitation or fall-out of suspended particulates. Although this idea has been suggested previously by the author , the phase change concept possibly permits some new insights into cloud-system behavior. For example, in a protoplanetary debris disk, the work of gravity may suddenly be enhanced by electrostatically driven collapse of the system when materials reach a critical intergranular spacing or grain number density. This might reduce the rate of planet formation by orders of magnitude (considering the ratio of g-forces to electrostatic ones for very small grains in close proximity), and indeed, it might drive the collapse into a system configuration that would perhaps not be created by gravity alone. Additional information is contained in the original.

Marshall, J. R.↗

Growth of CdZnTe Crystals for Radiation Detector Applications by Directional Solidification

Advances in Cadmium Zinc Telluride (Cd(sub 1-x)Zn(sub x)Te) growth techniques are needed for the production of large-scale arrays of gamma and x-ray astronomy. The research objective is to develop crystal growth recipes and techniques to obtain large, high quality CdZnTe single crystal with reduced defects, such as charge trapping, twinning, and tellurium precipitates, which degrade the performance of CdZnTe and, at the same time, to increase the yield of usable material from the CdZnTe ingot. A low gravity material experiment, "Crystal Growth of Ternary Compound Semiconductors in Low Gravity Environment", will be performed in the Material Science Research Rack (MSRR) on International Space Station (ISS). One section of the flight experiment is the melt growth of CdZnTe ternary compounds. This talk will focus on the ground-based studies on the growth of Cd(sub 0.80)Zn(sub 0.20)Te crystals for radiation detector applications by directional solidification. In this investigation, we have improved the properties that are most critical for the detector applications (electrical properties and crystalline quality): a) Electrical resistivity: use high purity starting materials (with reproducible impurity levels) and controlled Cd over pressure during growth to reproducibly balance the impurity levels and Cd vacancy concentration b) Crystalline quality: use ultra-clean growth ampoule (no wetting after growth), optimized thermal profile and ampoule design, as well as a technique for supercool reduction to growth large single crystal with high crystalline quality

Su, Ching-Hua↗

Field Programmable Gate Arrays for Enhancing the Speed and Energy Efficiency of Quantum Dynamics Simulations

We present the first application of field programmable gate arrays (FPGAs) as new, customizable hardware architectures for carrying out fast and energy-efficient quantum dynamics simulations of large chemical/material systems. Instead of tailoring the software to fixed hardware, which is the typical case for writing quantum chemistry code for central processing units (CPUs) and graphics processing units (GPUs), FPGAs allow us to directly customize the underlying hardware (even at the level of specific electrical signals in the circuit) to give a truly optimized computational performance for quantum dynamics calculations. By offloading the most intensive and repetitive calculations onto an FPGA, we show that the computational performance of our real-time electron dynamics calculations can even exceed that of optimized commercial mathematical libraries running on high-performance GPUs. In addition to this impressive computational speedup, we show that FPGAs are immensely energy-efficient and consume 4 times less energy than modern GPU or CPU architectures. These energy savings are a practical and important metric for supercomputing centers (many of which exceed over $1 million in power costs alone), as exascale computing capabilities become more widespread and commonplace. Taken together, the implementation techniques and performance metrics of our study demonstrate that FPGAs could play a promising role in upcoming quantum chemistry and materials science applications, particularly for the acceleration and energy-efficient execution of quantum dynamics calculations.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Bayesian Optimization of Catalysis with In-Context Learning

Large language models (LLMs) can perform accurate classification with zero or few examples through in-context learning (ICL), allowing the model to observe query-relevant examples at inference time and eliminating the need for additional weight updates to generalize beyond its original training data. We extend this capability to regression with uncertainty estimation using frozen LLMs (e.g., GPT-4o, Gemini), enabling Bayesian optimization (BO) in natural language without explicit model training or feature engineering. We apply this to materials discovery by representing materials as synthesis and testing procedures for use in natural language prompts. This Bayesian, design-first approach prioritizes optimization toward target material properties before detailed characterization, in contrast to conventional experimental workflows that often emphasize characterization of suboptimal materials. On benchmarks like aqueous solubility and oxidative coupling of methane (OCM), BO-ICL matches or outperforms Gaussian processes. In live experiments on the reverse water–gas shift (RWGS) reaction, BO-ICL identifies multimetallic catalysts that approach equilibrium CO yield within 6 and 10 iterations from a pool of 3,700 and 360,000 candidates, respectively. Our method redefines materials representation and accelerates discovery, with broad applications across catalysis, materials science, and AI.

Calibration↗

AtomSets as a hierarchical transfer learning framework for small and large materials datasets

Abstract Predicting properties from a material’s composition or structure is of great interest for materials design. Deep learning has recently garnered considerable interest in materials predictive tasks with low model errors when dealing with large materials data. However, deep learning models suffer in the small data regime that is common in materials science. Here we develop the AtomSets framework, which utilizes universal compositional and structural descriptors extracted from pre-trained graph network deep learning models with standard multi-layer perceptrons to achieve consistently high model accuracy for both small compositional data (<400) and large structural data (>130,000). The AtomSets models show lower errors than the graph network models at small data limits and other non-deep-learning models at large data limits. They also transfer better in a simulated materials discovery process where the targeted materials have property values out of the training data limits. The models require minimal domain knowledge inputs and are free from feature engineering. The presented AtomSets model framework can potentially accelerate machine learning-assisted materials design and discovery with less data restriction.

Chen, Chi (ORCID:0000000180087043)↗