Predicting dynamics from structure in a sodium silicate glass DOI Creative Commons
Rasmus Christensen, Morten M. Smedskjær

MRS Bulletin, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 31, 2024

Abstract Understanding the dynamics of atoms in glasses is crucial for unraveling origin relaxation and glass transition as well predicting transport properties. However, identifying structural features controlling atom remains challenging. Recently, machine learning models based on graph neural networks (GNNs) have successfully been used to predict future dynamics, but these prior studies focused primarily model systems such Kob–Andersen-type Lennard–Jones mixtures. This study investigates use local descriptors, GNN models, molecular simulations clarify atomics a realistic system (sodium silicate) across varying time scales. By harnessing capabilities different representations, we develop effective sodium ions within glassy silicate network, solely initial positions. We further demonstrate viability our approach through comparison previously proposed methods. Our findings pave way designing new formulations with tailored dynamical properties (e.g., electrolytes batteries). Impact statement Glass science has long grappled understanding fundamental nature dynamics. The governing principles atomic remain elusive it not obvious what look structure. While previous simplified systems, first that can be accurately multi-time scale complex oxide from static comparing architectures, establish outperform conventional descriptors prediction, representations being able effectively capture multibody correlations govern show scales up nanoseconds are at least partially encoded configuration itself, showing completely stochastic process. capability structure major implications could provide tools rational design materials functionalities, possibly accelerating development advanced applications areas solid-state batteries nuclear waste immobilization. Graphical abstract

Language: Английский

Applications of machine‐learning interatomic potentials for modeling ceramics, glass, and electrolytes: A review DOI
Shingo Urata, Marco Bertani, Alfonso Pedone

et al.

Journal of the American Ceramic Society, Journal Year: 2024, Volume and Issue: unknown

Published: June 9, 2024

Abstract The emergence of artificial intelligence has provided efficient methodologies to pursue innovative findings in material science. Over the past two decades, machine‐learning potential (MLP) emerged as an alternative technology density functional theory (DFT) and classical molecular dynamics (CMD) simulations for computational modeling materials estimation their properties. MLP offers more computation compared DFT, while providing higher accuracy CMD. This enables us conduct realistic using models with atoms longer simulation times. Indeed, number research studies utilizing MLPs significantly increased since 2015, covering a broad range structures, ranging from simple complex, well various chemical physical phenomena. As result, there are high expectations further applications field science industrial development. review aims summarize applications, particularly ceramics glass science, fundamental theories facilitate future progress utilization. Finally, we provide summary discuss perspectives on next challenges development application MLPs.

Language: Английский

Citations

6
Slowly quenched, high pressure glassy B2O3 at DFT accuracy DOI
Debendra Meher, Nikhil V. S. Avula, Sundaram Balasubramanian

et al.

The Journal of Chemical Physics, Journal Year: 2025, Volume and Issue: 162(4)

Published: Jan. 24, 2025

Modeling inorganic glasses requires an accurate representation of interatomic interactions, large system sizes to allow for intermediate-range structural order, and slow quenching rates eliminate kinetically trapped motifs. Neither first principles-based nor force field-based molecular dynamics (MD) simulations satisfy these three criteria unequivocally. Herein, we report the development a machine learning potential (MLP) classic glass, B2O3, which meets goals well. The MLP is trained on condensed phase configurations whose energies forces atoms are obtained using periodic quantum density functional theory. Deep MD based this accurately predict equation state densification glass with slower from melt. At ambient conditions, larger than 1011 K/s shown lead artifacts in structure. Pressure-dependent x-ray neutron structure factors compare excellently experimental data. High-pressure show varied coordination geometries boron oxygen, concur observations.

Language: Английский

Citations

0
Revisiting Machine Learning Potentials for Silicate Glasses: The Missing Role of Dispersion Interactions DOI Creative Commons
Alfonso Pedone, Marco Bertani,

Matilde Benassi

et al.

Journal of Chemical Theory and Computation, Journal Year: 2025, Volume and Issue: unknown

Published: April 24, 2025

Machine learning interatomic potentials (MLIPs) offer a promising alternative to traditional force fields and ab initio methods for simulating complex materials such as oxide glasses. In this work, we present the first evaluation of pretrained MACE (Multi-ACE) model [D.P. Kovács et al., J. Chem. Phys. 159(2023), 044118] silicate glasses, using sodium silicates test case. We compare its performance with DeePMD-based MLIP specifically trained on compositions [M. Bertani Theory Comput. 20(2024), 1358-1370] assess their accuracy in reproducing structural dynamical properties. Additionally, investigate role dispersion interactions by incorporating D3(BJ) correction both models. Our results show that while accurately reproduces neutron structure factors, pair distribution functions, Si[Qn] speciation, it performs slightly worst elastic properties calculations. However, is suitable simulations The inclusion significantly improves reproduction density MLIPs, highlighting critical glass modeling. These findings provide insight into transferability general MLIPs disordered systems emphasize need dispersion-aware training data sets developing accurate

Language: Английский

Citations

0
Revealing the reconstruction mechanism of AgPd nanoalloys under fluorination based on a multiscale deep learning potential DOI Creative Commons
Longfei Guo,

Shuang Shan,

Xiaoqing Liu

et al.

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 160(17)

Published: May 3, 2024

The design of heterogeneous catalysts generally involves optimizing the reactivity descriptor adsorption energy, which is inevitably governed by structure surface-active sites. A prerequisite for understanding structure–properties relationship precise identification real site structures, rather than relying on conceived structures derived from bulk alloy properties. However, it remains a formidable challenge due to dynamic nature nanoalloys during catalytic reactions and lack accurate efficient interatomic potentials simulations. Herein, generalizable deep-learning potential Ag–Pd–F system developed based dataset encompassing bulk, surface, nanocluster, amorphous, point defected configurations with diverse compositions achieve comprehensive description interactions, facilitating prediction surface formation diffusion energy barrier utilized investigate structural evolutions AgPd fluorination. involve inward F, outward Ag in Ag@Pd nanoalloys, AgFx species mixed Janus shape deformation cuboctahedron sphere Pd@Ag nanoalloys. Moreover, effects atomic dislocation migration reconstructing pathway are highlighted. It demonstrated that stress relaxation upon F serves as intrinsic driving factor governing reconstruction

Language: Английский

Citations

3
Molecular dynamics simulations of thermomechanical properties of silicone-modified phenolic polymer DOI
Jie Xiao, Guodong Fang, Bing Wang

et al.

Composites Science and Technology, Journal Year: 2024, Volume and Issue: unknown, P. 110878 - 110878

Published: Sept. 1, 2024

Language: Английский

Citations

2
Development and Validation of Neural Network Potentials for Multicomponent Oxide Glasses DOI

Ryuki Kayano,

Yaohiro Inagaki,

Ryuta Matsubara

et al.

The Journal of Physical Chemistry C, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 2, 2024

Language: Английский

Citations

2
Characterizing medium‐range order structure of binary silicate glasses using ring analysis and persistent homology DOI Creative Commons

Amirhossein Fadavi Firooz,

Rasmus Christensen, Christophe A. N. Biscio

et al.

Journal of the American Ceramic Society, Journal Year: 2024, Volume and Issue: 107(12), P. 7739 - 7750

Published: May 22, 2024

Abstract Several fundamental questions about the medium‐range order (MRO) structure of oxide glasses remain unanswered. How do we define MRO in glass? Should only consider covalently bonded rings or also repeating patterns non‐chemically atom clusters? Is first sharp diffraction peak (FSDP) factor constituted by those rings? In this study, focusing on binary silicate glasses, compare as determined using persistent homology and classical ring analysis. While latter identifies chemically rings, former captures both ring/loop structures. Our analyses are based atomic configurations established through molecular dynamics simulations three series alkali with varying modifier content. First, characterize size shape study how they contribute to FSDP. We show that loops can be directly extracted ignoring modifiers from analysis setting initial radii for Si O atoms zero. Then, demonstrate although FSDP, especially at low content, nonbonded features need considered fully explain

Language: Английский

Citations

1
First-principles NMR of oxide glasses boosted by machine learning DOI Creative Commons
Thibault Charpentier

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

Published: June 26, 2024

Machine-learning prediction of NMR tensors allows simulation experiments at finite temperature for models thousands atoms.

Language: Английский

Citations

1
Development of a machine learning interatomic potential for exploring pressure-dependent kinetics of phase transitions in germanium DOI
Andrea Fantasia, Fabrizio Rovaris, Omar Abou El Kheir

et al.

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 161(1)

Published: July 2, 2024

We introduce a data-driven potential aimed at the investigation of pressure-dependent phase transitions in bulk germanium, including estimate kinetic barriers. This is achieved by suitably building database several configurations along minimum energy paths, as computed using solid-state nudged elastic band method. After training model based on density functional theory (DFT)-computed energies, forces, and stresses, we provide validation rigorously test unexplored paths. The resulting agreement with DFT calculations remarkable wide range pressures. exploited large-scale isothermal-isobaric simulations, displaying local nucleation R8 to β-Sn pressure-induced transformation, taken here an illustrative example.

Language: Английский

Citations

1
Exciting DeePMD: Learning excited-state energies, forces, and non-adiabatic couplings DOI
Lucien Dupuy, Neepa T. Maitra

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 161(13)

Published: Oct. 1, 2024

We extend the DeePMD neural network architecture to predict electronic structure properties necessary perform non-adiabatic dynamics simulations. While learning excited state energies and forces follows a straightforward extension of approach for ground-state forces, how learn map between coupling vectors (NACV) local chemical environment descriptors is less trivial. Most implementations machine-learning-based inherently approximate NACVs, with an underlying assumption that energy-difference-scaled NACVs are conservative fields. overcome this approximation, implementing method recently introduced by Richardson [J. Chem. Phys. 158, 011102 (2023)], which learns symmetric dyad NACV. The efficiency accuracy our demonstrated through example methaniminium cation CH2NH2+.

Language: Английский

Citations

1