Search NASA⌕ Search

SEARCH · Search NASA

Results for “activity descriptors”

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 199 records · Page 11
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

The effects of warming and soil chemistry on bacterial community structure in Arctic tundra soils

Soil microbial communities in the Arctic play a critical role in regulating the global carbon (C) cycle. Vast amounts of C are stored in northern latitude soils, and rising temperatures in the Arctic threaten to thaw permafrost, making relatively inaccessible C sources more available for mineralization by soil microbes. Few studies have characterized how microbial community structure responds to thawing permafrost in the context of varying soil chemistries associated with contrasting tundra landscapes. Here, we subjected active layer and permafrost soils from upland and lowland tundra sites on the North Slope of Alaska to a soil-warming incubation experiment and compared soil bacterial community profiles (obtained by 16S rRNA amplicon sequencing) before and after incubation. The influence of soil composition (characterized by mid-infrared [MIR] spectroscopy) on bacterial community structure and class abundance was analyzed using redundancy and correlation analyses. We found increased abundances of Alphaproteobacteria, Gammaproteobacteria, and Bacteroidetes [Sphingobacteriia] post incubation, particularly in permafrost soils. The categorical descriptors site and soil layer had the most explanatory power in our predictive models of bacterial community structure, highlighting the close relationship between soil bacteria and the soil environment. Specific soil chemical attributes characterizing the soil environments that were found to be the best predictors include MIR spectral bands associated with inorganic C, silicates, amide II (C=N stretch), and carboxylics (C-O stretch), and MIR peak ratios representing C substrate quality. Overall, these results further characterize soil bacterial community shifts that may occur as a frozen environment with limited access to C sources, such as is found in undisturbed permafrost, transitions to warmer and more C-available environments, such as is predicted in thawing permafrost due to climate change.

16S rRNA sequencing↗
NASA NTRS2004

Active Flow Control on a Boundary-Layer-Ingesting Inlet

Boundary layer ingestion (BLI) is explored as means to improve overall system performance for Blended Wing Body configuration. The benefits of BLI for vehicle system performance benefit are assessed with a process derived from first principles suitable for highly-integrated propulsion systems. This performance evaluation process provides framework within which to assess the benefits of an integrated BLI inlet and lays the groundwork for higher-fidelity systems studies. The results of the system study show that BLI provides a significant improvement in vehicle performance if the inlet distortion can be controlled, thus encouraging the pursuit of active flow control (AFC) as a BLI enabling technology. The effectiveness of active flow control in reducing engine inlet distortion was assessed using a 6% scale model of a 30% BLI offset, diffusing inlet. The experiment was conducted in the NASA Langley Basic Aerodynamics Research Tunnel with a model inlet designed specifically for this type of testing. High mass flow pulsing actuators provided the active flow control. Measurements were made of the onset boundary layer, the duct surface static pressures, and the mass flow through the duct and the actuators. The distortion was determined by 120 total pressure measurements located at the aerodynamic interface plane. The test matrix was limited to a maximum freestream Mach number of 0.15 with scaled mass flows through the inlet for that condition. The data show that the pulsed actuation can reduce distortion from 29% to 4.6% as measured by the circumferential distortion descriptor DC60 using less than 1% of inlet mass flow. Closed loop control of the actuation was also demonstrated using a sidewall surface static pressure as the response sensor.

Gorton, Susan Althoff↗
DOE OSTI2022

Connecting Material Properties and Redox Flow Cell Cycling Performance through Zero-Dimensional Models

Improvements in redox flow battery (RFB) performance and durability can be achieved through the development of new active materials, electrolytes, and membranes. While a rich design space exists for emerging materials, complex tradeoffs challenge the articulation of unambiguous target criteria, as the relationships between component selection and cycling performance are multifaceted. Here, we derive zero-dimensional, analytical expressions for mass balances and cell voltages under galvanostatic cycling, enabling direct connections between material/electrolyte properties, cell operating conditions, and resulting performance metrics (e.g., energy efficiency, capacity fade). To demonstrate the utility of this modeling framework, we highlight several considerations for RFB design, including upper bound estimation, active species decay, and membrane/separator conductivity-selectivity tradeoffs. Furthermore, we also discuss modalities for extending this framework to incorporate kinetic losses, distributed ohmic losses, and multiple spatial domains. Importantly, because the mass balances are solved analytically, hundreds of cycles can be simulated in seconds, potentially facilitating detailed parametric sweeps, system optimization, and parameter estimation from cycling experiments. More broadly, this approach provides a means for assessing the impact of cell components that simultaneously influence multiple performance-defining processes, aiding in the elucidation of key descriptors and the identification of favorable materials combinations for specific applications.

25 ENERGY STORAGE↗
DOE OSTI2025

CHEMREASONER: Heuristic Search over a Large Language Model’s Knowledge Space using Quantum-Chemical Feedback

The discovery of new catalysts is essential for the design of new and more efficient chemical processes in order to transition to a sustainable future. We introduce an AI-guided computational screening framework unifying linguistic reasoning with quantum-chemistry based feedback from 3D atomistic representations. Our approach formulates catalyst discovery as an uncertain environment where an agent actively searches for highly effective catalysts via the iterative combination of large language model (LLM)-derived hypotheses and atomistic graph neural network (GNN)-derived feedback. Identified catalysts in intermediate search steps undergo structural evaluation based on spatial orientation, reaction pathways, and stability. Scoring functions based on adsorption energies and barriers steer the exploration in the LLM's knowledge space toward energetically favorable, high-efficiency catalysts. We introduce planning methods that automatically guide the exploration without human input, providing competitive performance against expert-enumerated chemical descriptor-based implementations. By integrating language-guided reasoning with computational chemistry feedback, our work pioneers AI-accelerated, trustworthy catalyst discovery.

artificial intelligence↗
DOE OSTI2026

Ultra-Low Amounts of Transition Metals in Carbon-Based Catalysts Improve Their Alkaline OER Performance: A Systematic Study

The pursuit of green hydrogen production highlights a persistent gap in electrocatalyst research: while academic efforts prioritize cost-efficiency via activity enhancement, industrial viability demands greater emphasis on electrochemical stability. Carbon-based electrocatalysts, particularly those incorporating transition metals, have shown promise in alkaline oxygen evolution (OER) due to their high activity, cost-efficiency, and resource-efficiency. However, these catalysts suffer from insufficient stability under oxidizing conditions compared to pure transition metal catalysts due to erosion of the carbon support resulting from carbon corrosion, among other degradation mechanisms. In this systematic study, the influence of ultra-low amounts (<1 wt %) of iron, cobalt, nickel, and their most common combinations on the stability of a hydrothermally derived, N-doped carbon support and the overall catalyst performance during alkaline OER is systematically explored. By identifying critical stability descriptors and correlating them with synthesis conditions and catalyst properties, primary and secondary corrosion pathways are unraveled. Subsequently, through careful adjustment of carbonization temperature and composition of incorporated transition metals, overall catalyst corrosion can be suppressed immensely. Especially, the inclusion of Ni and Fe is paramount for the formation of stable catalyst materials under laboratory conditions (10 mA/cm 2 in 0.1 M KOH), which is surprisingly unconstrained by the degree of graphitization of the carbon support. Mixing this electrochemically stable material with a graphitic carbon powder results in an excellent stability of over 400 h at 100 mA/cm 2 in 1 M KOH, implying great potential for the future improvement of carbon-based electrodes under oxidizing conditions toward industrial application.

N-doped carbon↗
DOE OSTI2022

Screening binary alloys for electrochemical CO 2 reduction towards multi-carbon products

Electrochemical reduction of CO 2 (eCO 2 R) to high-value chemicals presents an attractive approach for utilizing CO 2 . Copper (Cu) is presently the only electrocatalyst that fulfills this purpose with notable activity, but selectivity remains a problem. To identify catalysts for eCO 2 R with high selectivity towards multicarbon (C 2(+) ) products, we explore binary systems composed of strongly and weakly CO binding metals alloyed with Cu, Fe, Co, Ni, and Pd. A total number of 142 alloys with two commonly studied configurations, L1 2 and L1 0 , are simulated with density functional theory (DFT). We leverage recent progress in the atomistic understanding of the eCO 2 R mechanism and use the binding energies of CO* and C* as descriptors when screening for C 2(+) selectivity. We evaluate the stability of the binary alloys by analyzing the formation energy of the clean alloy surfaces. Our theoretical screening identifies about 16 Cu-based alloys and 18 non-Cu based alloys with optimal C 2(+) selective properties for eCO 2 R. For the non-Cu based binary alloys, the p-block elements play an important role in tuning the C* and CO* adsorption energies. In terms of stability, most of the Cu-based systems alloyed with metals that exhibit strong CO* binding are unstable. Ni-based alloys are more stable than the Co-based alloys followed by the Fe-based alloys, and all the Pd-based alloys are stable. In general, the L1 0 structural Fe, Co, Ni, and Pd-based alloys are more stable than the corresponding L1 2 alloys. Our approach identifies materials known to have good C 2(+) selectivity, but it also proposes several other promising materials that have not previously been tested for eCO 2 R.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

Electrocatalysts Development for Hydrogen Oxidation Reaction in Alkaline Media: from Mechanism Understanding to Materials Design

Anion exchange member (AEM) fuel cells have gained great attention recently mainly owing to the advantages of non-precious metal catalysts application. Despite of the similar oxygen reduction reaction (ORR) kinetics in acid and base, the reaction kinetics of hydrogen oxidation reaction (HOR) is two orders of magnitude slower in alkaline systems than in acid. To understand the slower kinetics of HOR in base, two major interpretations have been proposed, including 1) hydrogen binding energy as a major descriptor for HOR in both acid and base; and 2) bifunctional theory based on the contributions from both hydrogen and hydroxide adsorption for HOR in alkaline electrolyte. Here, we discuss the possible HOR mechanisms in alkaline electrolytes with the corresponding Tafel behavior change. Besides the discussion of traditional Tafel-Volmer and Heyrovsky-Volmer HOR mechanisms, the recently proposed hydroxide adsorption step is discussed to illustrate the HOR mechanism difference in acid and base. We further correlate the hydroxide adsorption step involved mechanism with HOR catalyst design, and summarize its role in HOR and HER activities improvement in alkaline electrolyte over the same catalyst. The representative works of alkaline HOR catalyst development were briefly introduced, and the strategy of strengthening interaction between H2 desorption and H2O recombination was finally proposed for future HOR catalyst design in alkaline environment.

Hydrogen oxidation reaction, Hydrogen evolution re↗
DOE OSTI2022

SOMAS: a platform for data-driven material discovery in redox flow battery development

Abstract Aqueous organic redox flow batteries offer an environmentally benign, tunable, and safe route to large-scale energy storage. The energy density is one of the key performance parameters of organic redox flow batteries, which critically depends on the solubility of the redox-active molecule in water. Prediction of aqueous solubility remains a challenge in chemistry. Recently, machine learning models have been developed for molecular properties prediction in chemistry and material science. The fidelity of a machine learning model critically depends on the diversity, accuracy, and abundancy of the training datasets. We build a comprehensive open access organic molecular database “Solubility of Organic Molecules in Aqueous Solution” (SOMAS) containing about 12,000 molecules that covers wider chemical and solubility regimes suitable for aqueous organic redox flow battery development efforts. In addition to experimental solubility, we also provide eight distinctive quantum descriptors including optimized geometry derived from high-throughput density functional theory calculations along with six molecular descriptors for each molecule. SOMAS builds a critical foundation for future efforts in artificial intelligence-based solubility prediction models.

25 ENERGY STORAGE↗
NASA NTRS2021

Anion-Assisted Delivery of Multivalent Cations to Inert Electrodes

To understand and control key electrochemical processes - metal plating, corrosion, intercalation, etc. requires molecular-scale details of the active species at electrochemical interfaces and their mechanisms for de-solvation from the electrolyte. Using free energy sampling techniques we reveal the interfacial speciation of divalent cations in ether-based electrolytes and mechanisms for their delivery to an inert graphene electrode interface. Surprisingly, we find that anion solvophobicity drives a high population of anion-containing species to the interface that facilitate the delivery of divalent cations, even to negatively charged electrodes. Our simulations indicate that cation desolvation is greatly facilitated by cation-anion coupling. We propose anion solvophobicity as a molecular-level descriptor for rational design of electrolytes with increased efficiency for electrochemical processes limited by multivalent cation desolvation.

Electrochemical interfaces↗
DOE OSTI2022

Towards automating structural discovery in scanning transmission electron microscopy *

Abstract Scanning transmission electron microscopy is now the primary tool for exploring functional materials on the atomic level. Often, features of interest are highly localized in specific regions in the material, such as ferroelectric domain walls, extended defects, or second phase inclusions. Selecting regions to image for structural and chemical discovery via atomically resolved imaging has traditionally proceeded via human operators making semi-informed judgements on sampling locations and parameters. Recent efforts at automation for structural and physical discovery have pointed towards the use of ‘active learning’ methods that utilize Bayesian optimization with surrogate models to quickly find relevant regions of interest. Yet despite the potential importance of this direction, there is a general lack of certainty in selecting relevant control algorithms and how to balance a priori knowledge of the material system with knowledge derived during experimentation. Here we address this gap by developing the automated experiment workflows with several combinations to both illustrate the effects of these choices and demonstrate the tradeoffs associated with each in terms of accuracy, robustness, and susceptibility to hyperparameters for structural discovery. We discuss possible methods to build descriptors using the raw image data and deep learning based semantic segmentation, as well as the implementation of variational autoencoder based representation. Furthermore, each workflow is applied to a range of feature sizes including NiO pillars within a La:SrMnO 3 matrix, ferroelectric domains in BiFeO 3 , and topological defects in graphene. The code developed in this manuscript is open sourced and will be released at github.com/nccreang/AE_Workflows .

47 OTHER INSTRUMENTATION↗
DOE OSTI2022

High-throughput screening to predict highly active dual-atom catalysts for electrocatalytic reduction of nitrate to ammonia

Ammonia is an essential chemical owing to its importance in fertilizer production and other industrial applications. Electrocatalytic nitrate reduction to ammonia (NO 3 RR) holds great promise for low-temperature ammonia production while simultaneously addressing nitrate-based environmental concerns. To provide the mechanistic understanding needed to design an effective electrocatalyst, we systematically investigated the catalytic performance of metal-based dual-atom catalysts (DACs) anchored on two-dimensional (2D) expanded phthalocyanine (Pc) for NO 3 RR. We found that NO 3 RR can efficiently produce ammonia on Cr 2 -Pc, V 2 -Pc, Ti 2 -Pc, and Mn 2 -Pc surfaces with low limiting potentials of – 0.02, – 0.25, – 0.34, and – 0.41 V RHE , respectively. Moreover, using the free energy difference of *NO 3 - and *H as a descriptor, we found that the hydrogen evolution reaction is significantly suppressed on the DAC surface due to an ensemble effect in which the two metal atoms cooperate to selectively form ammonia. We performed high-throughput screening to develop an efficient metal-based DAC for NO 3 - reduction, followed by a mechanistic study to elucidate the NO3RR pathway on the DAC. Finally, this work provides design information for advancing sustainable ammonia synthesis under ambient conditions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Computationally Predicted High-Throughput Free-Energy Phase Diagrams for the Discovery of Solid-State Hydrogen Storage Reactions

The design of multinary solid-state material systems that undergo reversible phase changes via changes in temperature and pressure provides a potential means of safely storing hydrogen. However, fully mapping the stabilities of known or newly targeted compounds relative to competing phases at reaction conditions has previously required many stringent experiments or computationally demanding calculations of each compound’s change in Gibbs energy with respect to temperature, G(T). Here, we have extended the approach of constructing chemical potential phase diagrams based on ΔG f (T) to enable the analysis of phase stability at non-zero temperatures. We first performed density functional theory calculations to compute the formation enthalpies of binary, ternary, and quaternary compounds within several compositional spaces of current interest for solid-state hydrogen storage. Temperature effects on solid compound stability were then accounted for using our recently introduced machine learned descriptor for the temperature-dependent contribution G δ (T) to the Gibbs energy G(T). From these Gibbs energies, we evaluated each compound’s stability relative to competing compounds over a wide range of conditions and show using chemical potential and composition phase diagrams that the predicted stable phases and H2 release reactions are consistent with experimental observations. This demonstrates that our approach rapidly computes the thermochemistry of hydrogen release reactions for compounds at sufficiently high accuracy relative to experiment to provide a powerful framework for analyzing hydrogen storage materials. This framework based on G(T) enables the accelerated discovery of active materials for a variety of technologies that rely on solid-state reactions involving these materials.

08 HYDROGEN↗
DOE OSTI2022

Machine learning reveals factors that control ion mobility in anti-perovskite solid electrolytes

Solid-state batteries are projected to exhibit improved energy densities and safety compared to liquid-electrolyte-based systems. Consequently, the development of solid electrolytes that can conduct cations at rates comparable to that of liquid electrolytes is an active area of research. To facilitate this search it is desirable to understand the chemical and structural features that control ion transport through a crystalline lattice. Here, machine learning (ML) is used to identify features that influence ion mobility in solids, quantify the relative importance of those features, and demonstrate how the variation of individual features alters ion mobility. Adopting the anti-perovskite lattice as a model system, ML algorithms were trained to predict migration barriers from a data set containing hundreds of barriers calculated using density functional theory. In this study, more than 10 6 feature sets of non-redundant descriptors were evaluated for their ability to predict migration barriers. The present analysis correctly reiterates the importance of several features that are known to influence mobility – for example, anion polarizability – but also identifies others whose importance is not widely recognized. Lattice properties such as hopping distance and channel width exhibit the greatest influence on cation mobility. Individual conditional expectation analyses performed on these features shows that barriers decrease as hopping distance decreases and as channel width increases. The defect formation energy – which is commonly associated with the concentration of defects, but not with their mobility – is identified as an important feature for interstitial migration. In sum, this analysis aids in the design of optimal solid electrolytes by simplifying the multi-dimensional design space to a sub-set of properties that are the most important.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Acidity of surface-infiltrated binary oxides as a sensitive descriptor of oxygen exchange kinetics in mixed conducting oxides

Improving the kinetics of O 2 reduction on oxide surfaces is critical in many energy and fuel conversion technologies. Here we show that the acidity scale for binary oxides is a powerful descriptor for tuning and predicting oxygen surface exchange kinetics on mixed conducting oxides. By infiltrating a selection of binary oxides from strongly basic (Li 2 O) to strongly acidic (SiO 2 ) onto the surface of Pr 0.1 Ce 0.9 O 2-δ samples, it was possible to vary the chemical surface exchange coefficient k chem by 6 orders of magnitude, with basic oxides such as Li 2 O increasing k chem by nearly 1,000 times, with surface concentrations as low as 50 ppm impacting k chem . Strikingly, although the pre-exponential of k chem scales linearly with the acidity of the infiltrated binary oxide, there is nearly no change in the activation energy. Here, the origin of these dramatic changes is proposed to arise from the systematic increase in electron concentration at the Pr 0.1 Ce 0.9 O 2-δ surface with the decreasing acidity of the infiltrated binary oxide.

30 DIRECT ENERGY CONVERSION↗
DOE OSTI2025

Automation-Accelerated Electrolyte Design Mitigates Solubility Competition between Redox-Active Molecules and Supporting Salts

In nonaqueous redox-flow batteries (NRFBs), redox-active organic molecules (ROMs) and supporting salts compete for solvation sites, limiting achievable energy density. We combine automated high-throughput experimentation (HTE) with camera-based saturation monitoring and quantitative NMR to measure paired (ROM, salt) solubilities across single and mixed organic solvents. Using 2,1,3-benzothiadiazole (BTZ) with lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) as a model system, we find that a binary m-xylene/acetonitrile mixture dissolves ≈3 M of both BTZ and LiTFSI─surpassing the previously reported 2 M ceiling for neat acetonitrile─by leveraging complementary solvation (MX is BTZ-philic and salt-phobic; ACN stabilizes LiTFSI). A random-forest model (RMSE ≈ 0.24) trained on solvent descriptors highlights log P and salt concentration as dominant predictors and predicts MX/ACN ≈0.3/0.7 (v/v) to be near-optimal. These formulations retain practical viscosity and ∼5 mS·cm –1 conductivity at high loading. In conclusion, the workflow provides a reproducible, data-centric route to NRFB electrolyte design and motivates an open, standardized dual-solute solubility resource for accelerated electrolyte discovery.

Electrolytes↗
DOE OSTI2022

Unraveling Electronic Trends in O* and OH* Surface Adsorption in the MO 2 Transition-Metal Oxide Series

Understanding the bond strength of O* and OH* intermediates to metal-oxide surfaces is key to predicting the catalytic activity in oxygen-based electrochemistry. Here, we uncover highly non-linear trends in O* and OH* adsorption energies across the 3d, 4d, and 5d series of MO 2 transition-metal (TM) oxide surfaces computed within Hubbard- U corrected density functional theory (DFT + U ). Investigating the electronic structure with crystal orbital Hamiltonian populations (COHP) of the relevant metal–oxygen bonds reveals that the spin-dependent coupling strength between metal-d and oxygen-2p atomic orbitals together with the extent of filling of bonding and anti-bonding orbitals are the primary contributors to the adsorption energy. Importantly, we show that the integrated COHP obtained purely from bulk calculations is a highly accurate descriptor for surface adsorption energetics that captures trends across the group 5–12 TM oxide series within 0.19–0.36 eV. Our results suggest a pathway to prediction of adsorption energies for an arbitrary metal–ligand catalyst system.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

Hydrogenation and C-S bond activation pathways in thiophene and tetrahydrothiophene reactions on sulfur-passivated surfaces of Ru, Pt, and Re nanoparticles

Thiophene-H2 reactions proceed via sulfur removal and hydrogenation routes on dispersed metal nanoparticles that become decorated by refractory S-adlayers during catalysis. The identity and kinetic relevance of the required elementary steps are described here based on rates measured at S-chemical potentials set by H2S/H2 ratios similar to those prevalent during practical catalysis on Re, ReSx, Ru, and Pt catalysts. Free energies for S adatom formation (from H2S decomposition and H2 evolution) are strongly exothermic (< -50 kJ mol-1 on Pt(111) and < -150 kJ mol-1 on Re and Ru(0001)), but strong repulsions between S adatoms cause adsorption free energies to increase significantly with coverage on all three surfaces, preventing complete monolayer formation. These adlayers, composed of unreactive S-atoms (S') that cover 1/3–2/3 ML leave residual interstitial spaces (*) that bind S-atoms (S*), intermediates, and transition states reversibly, as required for catalytic turnovers. The number and binding properties of these interstices depend on the identity and chemical state of the nanoparticle bulk phase, which influences S'-binding and coverages and cause large differences in direct desulfurization and hydrogenation turnover rates (per exposed metal atom) on dispersed Re, ReSx, Ru, and Pt. The identity and kinetic relevance of elementary steps for desulfurization (to C4¬ hydrocarbons) and hydrogenation (to tetrahydrothiophene; THT) are similar among these catalysts; they involve the kinetically-relevant formation of a thiophene-derived intermediate (monohydrothiophene on Re and ReSx; dihydrothiophene on Ru and Pt) that either cleaves its C-S bond or “over-hydrogenates” to THT in one surface sojourn. THT then undergoes C-S bond cleavage in secondary reactions that correct such over-hydrogenation to form the more unsaturated species that cleave C-S bonds. THT/C4 product ratios are insensitive to H2S/H2 ratios and thiophene pressure, even though active interstitial spaces are covered by kinetically-detectable coverages of S* and thiophene; therefore, primary and secondary reactions must involve the same active surfaces. The observed increase in THT/C4 ratios with H2 pressure shows that THT formation transition states involve a larger number of H-atoms than for C-S cleavage. The requirement for bound species with intermediate unsaturation (between THT and thiophene) for C-S bond cleavage is reminiscent of the H-shuttling required in C-C and C-O hydrogenolysis, reactions that involve the partial dehydrogenation of alkanes and alkanols, respectively, to weaken such bonds and to increase the formation entropy of the relevant transition states via the evolution of H2(g). These mechanistic details challenge prevalent paradigms about different site requirements for hydrogenation and desulfurization pathways and about how metal-sulfur bond energies act as descriptors of reactivity; in fact, such binding energies merely act to define the refractory S-adlayers that enable the formation of weakly-binding interstices that reversibly bind intermediates and transition states, thus allowing catalytic turnovers.

Yik, Edwin↗
DOE OSTI2025

Overcoming sparse datasets with multi-task learning as applied to high entropy alloys

Abstract The design of novel High Entropy Alloys for use in high-temperature applications is an area of active interest due to their potential to provide exceptional properties compared to conventional alloys. Since the increased popularity of machine learning, an important cog in the design process has been training surrogate models on alloy properties. However, these Single-Task models are trained on individual mechanical properties and do not take advantage of the relatedness between properties. Multi-Task models can capture the interdependencies between tasks, leading to potentially more accurate predictions for all tasks. In this paper, we investigate if Multi-Task models can show improvement over Single-Task models when used for predicting the mechanical properties of these alloys. To ensure fair evaluation between the models, we apply L 0 regularization and skip connections to the models, which allows them to adjust the number of model parameters and depth for optimal performance. We find that the Multi-Task models can leverage task relationships to perform better than Single-Task models, especially for high amounts of missing data in the tasks. Furthermore, adding simple auxiliary targets can boost Multi-Task performance even further despite not being effective as input descriptors to single-task models themselves. We anticipate that the proposed strategies can achieve more accurate predictions and consequently enable better design capabilities for such data-constrained domains without incurring much additional computational cost.

Debnath, Arindam (ORCID:0000000194274499)↗