Data-driven Design of High Pressure Hydride Superconductors using DFT and Deep Learning DOI Creative Commons
Daniel Wines, Kamal Choudhary

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

The observation of superconductivity in hydride-based materials under ultrahigh pressures (for example, H$_3$S and LaH$_{10}$) has fueled the interest a more data-driven approach to discovering new high-pressure hydride superconductors. In this work, we performed density functional theory (DFT) calculations predict critical temperature ($T_c$) over 900 pressure range (0 500) GPa, where found 122 dynamically stable structures with $T_c$ above MgB$_2$ (39 K). To accelerate screening, trained graph neural network (GNN) model demonstrated that universal machine learned force-field can be used relax arbitrary pressures, significantly reduced cost. By combining DFT GNNs, establish complete map hydrides pressure.

Language: Английский

Improving machine-learning models in materials science through large datasets DOI Creative Commons
Jonathan Schmidt, Tiago F. T. Cerqueira, A. Romero

et al.

Materials Today Physics, Journal Year: 2024, Volume and Issue: 48, P. 101560 - 101560

Published: Sept. 25, 2024

Language: Английский

Citations

12
Data-driven design of high pressure hydride superconductors using DFT and deep learning DOI Creative Commons
Daniel Wines, Kamal Choudhary

Materials Futures, Journal Year: 2024, Volume and Issue: 3(2), P. 025602 - 025602

Published: May 13, 2024

Abstract The observation of superconductivity in hydride-based materials under ultrahigh pressures (for example, H 3 S and LaH 10 ) has fueled the interest a more data-driven approach to discovering new high-pressure hydride superconductors. In this work, we performed density functional theory (DFT) calculations predict critical temperature ( Tc over 900 pressure range (0–500) GPa, where found 122 dynamically stable structures with above MgB 2 (39 K). To accelerate screening, trained graph neural network (GNN) model demonstrated that universal machine learned force-field can be used relax arbitrary pressures, significantly reduced cost. By combining DFT GNNs, establish complete map hydrides pressure.

Language: Английский

Citations

5
Where artificial intelligence stands in the development of electrochemical sensors for healthcare applications-A review DOI Creative Commons
Andreea Cernat, Adrian Groza, Mihaela Tertiş

et al.

TrAC Trends in Analytical Chemistry, Journal Year: 2024, Volume and Issue: 181, P. 117999 - 117999

Published: Oct. 5, 2024

Language: Английский

Citations

3
Modeling the impact of structure and coverage on the reactivity of realistic heterogeneous catalysts DOI Creative Commons
Benjamin W. J. Chen, Manos Mavrikakis

Nature Chemical Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 17, 2025

Language: Английский

Citations

0
Cross-disciplinary perspectives on the potential for artificial intelligence across chemistry DOI Creative Commons
Austin M. Mroz, Annabel R. Basford, Friedrich Hastedt

et al.

Chemical Society Reviews, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

We offer ten diverse perspectives exploring the transformative potential of artificial intelligence (AI) in chemistry, highlighting many challenges we face, and offering strategies to address them.

Language: Английский

Citations

0
Efficient structure-informed featurization and property prediction of ordered, dilute, and random atomic structures DOI
Adam M. Krajewski, Jonathan W. Siegel, Zi‐Kui Liu

et al.

Computational Materials Science, Journal Year: 2024, Volume and Issue: 247, P. 113495 - 113495

Published: Nov. 7, 2024

Language: Английский

Citations

1
Machine learning prediction of materials properties from chemical composition: Status and prospects DOI Open Access
Mohammed Alghadeer, Nyimas Aisyah, Mahmoud Hezam

et al.

Chemical Physics Reviews, Journal Year: 2024, Volume and Issue: 5(4)

Published: Dec. 1, 2024

In materials science, machine learning (ML) has become an essential and indispensable tool. ML emerged as a powerful tool in particularly for predicting material properties based on chemical composition. This review provides comprehensive overview of the current status future prospects using this domain, with special focus physics-guided (PGML). By integrating physical principles into models, PGML ensures that predictions are not only accurate but also interpretable, addressing critical need sciences. We discuss foundational concepts statistical PGML, outline general framework informatics, explore key aspects such data analysis, feature reduction, composition representation. Additionally, we survey latest advancements prediction geometric structures, electronic properties, other characteristics from formulas. The resource tables listing databases, tools, predictors, offering valuable reference researchers. As field rapidly expands, aims to guide efforts harnessing discovery development.

Language: Английский

Citations

1
Optical materials discovery and design with federated databases and machine learning DOI
Victor Trinquet, Matthew L. Evans, Cameron J. Hargreaves

et al.

Faraday Discussions, Journal Year: 2024, Volume and Issue: unknown

Published: July 10, 2024

Combinatorial and guided screening of materials space with density-functional theory related approaches has provided a wealth hypothetical inorganic materials, which are increasingly tabulated in open databases. The OPTIMADE API is standardised format for representing crystal structures, their measured computed properties, the methods querying filtering them from remote resources. Currently, federation spans over 20 data providers, rendering 30 million structures accessible this way, many novel have only recently been suggested by machine learning-based approaches. In work, we outline our approach to non-exhaustively screen dynamic trove next-generation optical materials. By applying MODNet, neural network-based model property prediction, within combined active learning high-throughput computation framework, isolate particular chemistries that should be most fruitful further theoretical calculations experimental study as high-refractive-index making explicit use automated calculations, federated dataset curation learning, releasing these publicly, workflows presented here can periodically re-assessed new databases implement OPTIMADE, suggested.

Language: Английский

Citations

0
Jupyter widgets and extensions for education and research in computational physics and chemistry DOI Creative Commons
Dou Du, Taylor J. Baird, Kristjan Eimre

et al.

Computer Physics Communications, Journal Year: 2024, Volume and Issue: 305, P. 109353 - 109353

Published: Aug. 22, 2024

Language: Английский

Citations

0
Data-driven Design of High Pressure Hydride Superconductors using DFT and Deep Learning DOI Creative Commons
Daniel Wines, Kamal Choudhary

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

The observation of superconductivity in hydride-based materials under ultrahigh pressures (for example, H$_3$S and LaH$_{10}$) has fueled the interest a more data-driven approach to discovering new high-pressure hydride superconductors. In this work, we performed density functional theory (DFT) calculations predict critical temperature ($T_c$) over 900 pressure range (0 500) GPa, where found 122 dynamically stable structures with $T_c$ above MgB$_2$ (39 K). To accelerate screening, trained graph neural network (GNN) model demonstrated that universal machine learned force-field can be used relax arbitrary pressures, significantly reduced cost. By combining DFT GNNs, establish complete map hydrides pressure.

Language: Английский

Citations

0