Microstructural control of additively manufactured Ti6Al4V via in-situ large-area laser annealing
Not Available
36 MATERIALS SCIENCE↗
SEARCH · Engineering Papers
Results for “Transformer model”
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.
Luminescent Metal–Organic Framework for the Selective Detection of Aldehydes
The detection of toxic, hazardous chemical species is an important task because they pose serious risks to either the environment or human health. Luminescent metal–organic frameworks (LMOFs) as alternative sensors offer rapid and sensitive detection of chemical species. Interactions between chemical species and LMOFs result in changes in the photoluminescence (PL) profile of the LMOFs which can be readily detected using a simple fluorometer. Herein, we report the use of a robust, Zn-based LMOF, [Zn 5 (μ 3 -OH) 2 (adtb) 2 (H 2 O) 5 ·5 DMA] (Zn-adtb, LMOF-341), for the selective detection of benzaldehyde. Upon exposure to benzaldehyde, Zn-adtb experiences significant luminescent quenching, as characterized through PL experiments. Photoluminescent titration experiments reveal that LMOF-341 has a detection limit of 64 ppm and a Ksv value of 179 M –1 for benzaldehyde. Furthermore, we study the guest–host interactions that occur between LMOF-341 and benzaldehyde through in situ Fourier transform infrared and computational modeling employing density functional theory. The results show that benzaldehyde interacts more strongly with LMOF-341 compared to formaldehyde and propionaldehyde. Furthermore, our combined studies also reveal that the mechanism of luminescence quenching originates from an electron-transfer process.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
cm 2 -Scale Synthesis of MoTe 2 Thin Films with Large Grains and Layer Control
Owing to the small energy differences between its polymorphs, MoTe 2 can access a full spectrum of electronic states from the 2H semiconducting state to the 1T' semimetallic state and from the T d Weyl semimetallic state to the superconducting state in the 1T' and T d phase at low temperature. Thus, it is a model system for phase transformation studies as well as quantum phenomena such as the quantum spin Hall effect and topological superconductivity. Careful studies of MoTe 2 and its potential applications require large-area MoTe 2 thin films with high crystallinity and thickness control. Here, we present cm 2 -scale synthesis of 2H-MoTe 2 thin films with layer control and large grains that span several microns. Layer control is achieved by controlling the initial thickness of the precursor MoO x thin films, which are deposited on sapphire substrates by atomic layer deposition and subsequently tellurized. Despite the van der Waals epitaxy, the precursor–substrate interface is found to critically determine the uniformity in thickness and grain size of the resulting MoTe 2 films: MoTe 2 grown on sapphire show uniform films while MoTe 2 grown on amorphous SiO 2 substrates form islands. Furthermore, this synthesis strategy decouples the layer control from the variabilities of growth conditions for robust growth results and is applicable to growing other transition-metal dichalcogenides with layer control.
2D materials↗
Direct Conversion of Methane to Ethylene and Acetylene over an Iron-Based Metal–Organic Framework
Conversion of methane (CH 4 ) to ethylene (C 2 H 4 ) and/or acetylene (C 2 H 2 ) enables routes to a wide range of products directly from natural gas. However, high reaction temperatures and pressures are often required to activate and convert CH 4 controllably, and separating C 2+ products from unreacted CH 4 can be challenging. Here, we report the direct conversion of CH 4 to C 2 H 4 and C 2 H 2 driven by non-thermal plasma under ambient (25 °C and 1 atm) and flow conditions over a metal–organic framework material, MFM-300(Fe). The selectivity for the formation of C 2 H 4 and C 2 H 2 reaches 96% with a high time yield of 334 μmol g cat –1 h –1 . At a conversion of 10%, the selectivity to C 2+ hydrocarbons and time yield exceed 98% and 2056 μmol g cat –1 h –1 , respectively, representing a new benchmark for conversion of CH 4 . In situ neutron powder diffraction, inelastic neutron scattering and solid-state nuclear magnetic resonance, electron paramagnetic resonance (EPR), and diffuse reflectance infrared Fourier transform spectroscopies, coupled with modeling studies, reveal the crucial role of Fe–O(H)–Fe sites in activating CH 4 and stabilizing reaction intermediates via the formation of an Fe–O(CH 3 )–Fe adduct. In addition, a cascade fixed-bed system has been developed to achieve online separation of C 2 H 4 and C 2 H 2 from unreacted CH 4 for direct use. Integrating the processes of CH 4 activation, conversion, and product separation within one system opens a new avenue for natural gas utility, bridging the gap between fundamental studies and practical applications in this area.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Generative Electrolyte Solvent and Formulation Discovery
Molecular mixtures and/or formulations are of great importance in fields ranging from materials science to pharmaceuticals to chemistry. In batteries, electrolytes are complex molecular mixtures consisting of multiple salts and solvents and additives at different concentrations that dictate battery capacity, safety, and cycle life, among others. Unfortunately, due to the complex composition and infinite design space as well as the conflicting property requirements, electrolyte design is the rate-determining step in the design of next generation battery chemistries. In this work, we develop a transformer-based generative AI model − ElectrolyteGPT − capable of generating solvents and electrolyte formulations to satisfy a wide range of desired property requirements. First, we curate an electrolyte-relevant database and develop a new line notation for formulations. Then, we show that ElectrolyteGPT can generate solvents and formulations conditioned on a wide range of important electrolyte properties such as ionic conductivity, oxidative stability, Coulombic efficiency, viscosity, and more. Finally, we experimentally synthesize the generated solvents and fabricate the electrolyte formulations and show that they can meet the desired property requirements and enable longterm cycling in energy-dense anode-free lithium metal batteries. Our work showcases the ability of generative models to address challenges in molecular mixture design for next generation batteries.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Health co-benefits of achieving sustainable net-zero greenhouse gas emissions in California
The achievement of net-zero greenhouse gas (GHG) emissions by 2100 is required to limit global temperature rise below 2 °C above preindustrial levels. Earlier accomplishments of net-zero GHG emissions in developed regions support this global target. In this study, we have developed a road map for California to achieve net-zero GHG emissions sustainably in 2050 by using detailed modelling of energy system transformation, cross-sectoral connectivity and technology penetration, as well as quantify the associated health co-benefits from reduced co-emitted air pollutants. We find that approximately 14,000 premature deaths can be avoided in California in 2050 and that these health co-benefits are disproportionately higher in disadvantaged communities (that is, 35% of avoided deaths will come from 25% of the state’s population). The annualized monetary benefits (US$215 billion) exceed the GHG abatement cost (US$106 billion) by US$109 billion. This road map requires the use of bioenergy with carbon capture and sequestration technology to offset some GHG emissions. However, this technology comes at a price as it would emit a considerable amount of air pollutants and reduce health co-benefits by US$4 billion. Nevertheless, our analysis shows that ambitious GHG reduction efforts can provide substantial health co-benefits, especially for residents of disadvantaged communities.
54 ENVIRONMENTAL SCIENCES↗
Artificial neural network prediction of self-diffusion in pure compounds over multiple phase regimes
Artificial neural networks (ANNs) were developed to accurately predict the self-diffusion constants for pure components in liquid, gas and super critical phases. The ANNs were tested on an experimental database of 6625 self-diffusion constants for 118 different chemical compounds. The presence of multiple phases results in a heavy skew in the distribution of diffusion constants and multiple approaches were used to address this challenge. First, an ANN was developed with the raw diffusion values to assess what the main drawbacks of this direct method were. The first approach for improving the predictions involved taking the log 10 of diffusion to provide a more uniform distribution and reduce the range of target output values used to develop the ANN. The second approach involved developing individual ANNs for each phase using the raw diffusion values. Results show that the log transformation leads to a model with the best self-diffusion constant predictions and an overall average absolute deviation (AAD) of 6.56%. The resultant ANN is a generalized model that can be used to predict diffusion across all three phases and over a diverse group of compounds. The importance of each input feature was ranked using a feature addition method revealing that the density of the compound has the largest impact on the ANN prediction of self-diffusion constants in pure compounds.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Restoring permutational invariance in the Jordan–Wigner transformation
The Jordan–Wigner transformation is a powerful tool for converting systems of spins into systems of fermions, or vice versa. While this mapping is exact, the transformation itself depends on the labelling of the spins. One consequence of this dependence is that approximate solutions of a Jordan–Wigner-transformed Hamiltonian may depend on the (physically inconsequential) labelling of the spins. In this work, we turn to an extended Jordan–Wigner transformation which remedies this problem and which may also introduce some correlation atop the Hartree–Fock solution of a transformed spin Hamiltonian. We demonstrate that this extended Jordan–Wigner transformation can be thought of as arising from a unitary version of the Lie algebraic similarity transformation (LAST) theory. Here, we show how these ideas, particularly in combination with the standard (non-unitary) version of LAST, can provide a potentially powerful tool for the treatment of the XXZ and J 1 –J 2 Heisenberg Hamiltonians.
74 ATOMIC AND MOLECULAR PHYSICS↗
Combining protein sequences and structures with transformers and equivariant graph neural networks to predict protein function
Abstract Motivation Millions of protein sequences have been generated by numerous genome and transcriptome sequencing projects. However, experimentally determining the function of the proteins is still a time consuming, low-throughput, and expensive process, leading to a large protein sequence-function gap. Therefore, it is important to develop computational methods to accurately predict protein function to fill the gap. Even though many methods have been developed to use protein sequences as input to predict function, much fewer methods leverage protein structures in protein function prediction because there was lack of accurate protein structures for most proteins until recently. Results We developed TransFun—a method using a transformer-based protein language model and 3D-equivariant graph neural networks to distill information from both protein sequences and structures to predict protein function. It extracts feature embeddings from protein sequences using a pre-trained protein language model (ESM) via transfer learning and combines them with 3D structures of proteins predicted by AlphaFold2 through equivariant graph neural networks. Benchmarked on the CAFA3 test dataset and a new test dataset, TransFun outperforms several state-of-the-art methods, indicating that the language model and 3D-equivariant graph neural networks are effective methods to leverage protein sequences and structures to improve protein function prediction. Combining TransFun predictions and sequence similarity-based predictions can further increase prediction accuracy. Availability and implementation The source code of TransFun is available at https://github.com/jianlin-cheng/TransFun.
59 BASIC BIOLOGICAL SCIENCES↗
Deep image reconstruction for background subtraction in heavy-ion collisions
Jet reconstruction in an ultrarelativistic heavy-ion collision suffers from a notoriously large, fluctuating thermal background. Traditional background‐subtraction methods struggle to remove this soft background while preserving the jet's hard substructure. In this Letter, we present DeepSub, a machine-learning-based approach for full-event background‐subtraction. DeepSub utilizes a model based on Swin Transformer layers to denoise jet images and disentangle hard jets from the heavy-ion background. DeepSub significantly outperforms existing subtraction techniques by reproducing key jet observables such as jet 𝑝 T and mass, and substructure observables such as girth and the energy correlation function, at the subpercent to percent level. As such, DeepSub paves the way for precision heavy-ion measurements in hitherto inaccessible kinematic regimes.
Qureshi, Umar Sohail [Stanford University, CA (Uni↗
Far-infrared absorption of undoped and Br-doped carbon nanofiber powder in stacked-cup cone configuration
In this study, we performed room-temperature far-infrared (40–650cm –1 ) transmission measurements on undoped and bromine-doped powder samples of carbon nanofibers in stacked-cup cone geometry. The transmission spectra show enhanced transmittance after bromine doping and all spectra were fit to a Drude-Lorentz (DL) model. A decreased metallic conductivity along with a redshift of the lowest semiconducting gap was found with doped samples. A significant decrease in the scattering rate upon heavy doping has been qualitatively explained as partial ordering of intercalated dopant ions. Absorption spectra were derived from the transmission spectra under the assumption of nondispersive reflectance and subsequently compared with DL-model-derived spectra. The free-carrier density of the n -type powder and the electronic mean free path were estimated and compared with reported values for single-walled nanotubes and pyrolytic graphite.
75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
Predictive Direct DC-Link Control for 7.2 kV Three-Port Low-Inertia Solid-State Transformer With Active Power Decoupling
Promising for applications including renewable energy and electric vehicle fast charging, a medium-voltage (MV) solid-state transformer (SST) typically has multiple modules series stacked, which requires the modules to be single phase. One critical issue of the single-phase SST is the double-line-frequency power ripple. Traditionally, large passives are used to buffer the ripple, resulting in significantly increased volume and cost. This article for the first time proposes active power decoupling (APD) for the SST and a predictive direct DC-link control method, using a current-source single-stage SST as an example. An electrolytic capacitor-less buffer port is used to absorb the ripple, which tolerates up to a 30% voltage ripple for small capacitance. The APD SST control is challenging. First, the current-source SST realizes isolation, DC link, and multiport in a single stage. Second, different from conventional SSTs or APD converters, the low-inertia DC link in this SST has a 40% switching ripple, which is difficult to stabilize. To address the control challenge, a direct DC-link control architecture and a predictive control method are proposed. Simulation and experimental results verify the proposed control method on an MV SiC modular soft-switching solid-state transformer (M-S4T) prototype. Here, the proposed APD reduces the SST volume by 53.8%. Importantly, the proposed concept is not limited to the M-S4T but is generic to other SSTs or APD converters.
14 SOLAR ENERGY↗
ABLE Workflow Copier
A copier template for generating a snakemake workflow with an associated python package for implementing dataset transformation, feature extraction, and modeling.
Pathak, Maharshi [Northeastern Univ., Boston, MA (↗
Vistransformers Explained
The Vistransformers Explained library is a collection of python notebooks that demonstrate the internal mechanics and uses of visual-transformer (ViT) machine learning models. The code implements, with mild modifications, ViT models that have been made publicly available through publication and GitHub code. The value added by this code is in-depth explanations of the mathematics behind the sub-modules of the ViT models, including original figures. Additionally, the library contains the code necessary to implement and train the ViT models. The library does not include example training data for the models; instead, it would rely on users generating their own datasets. The code is based on the PyTorch python library. It does not include any files other than python scripts, modules, or notebooks.
Callis, Skylar↗
Hyperparameter Studies for Vision Transformers Trained on High-Fidelity Simulations
This library is a collection of python modules that define, train, and analyze vision-transformer (ViT) machine learning models. The code implements, with mild modifications, ViT models that have been made publicly available through publication and GitHub code. The training data for these models is hydrodynamic simulation output in the form of numpy arrays. This library contains code to train these ViT models on the hydrodynamic simulation output with a variety of hyperparameters, and to compare the results of such models. Furthermore, the library contains definitions of simple convolutional neural network (CNN) machine learning architectures which can be trained on the same hydrodynamic simulation output. These are included as a reference point to compare the ViT models to. Additionally, the library includes trained ViT and CNN models and example input data for demonstration purposes. The code is based on the PyTorch python library.
Callis, Skylar↗
pnnl/SuperSAM
Our SuperSAM approach transforms the Segment Anything Model (SAM) into an `elastic' supernetwork, enabling the derivation of subnetworks with varying architectures tailored to a wide range of resource constraints.
Tallent, Nathan↗
pnnl/battery-degradation-prognosis
This is a tool for long term prognosis of redox flow battery parameter degradation which uses Transformer-based deep learning models for time series forecasting which was developed as part of the ROVI project.
Saldanha, Emily [Pacific Northwest National Labora↗
TRANSFORM – A Vision for Modern Advanced Reactor System-Level Modeling and Simulation Using Modelica
Explore the source record for details and available documents.
Greenwood, Scott↗