Disorder-Dependent Li Diffusion in Li₆PS₅Cl Investigated by Machine-Learning Potential

ACS Applied Materials & Interfaces, Journal Year: 2024, Volume and Issue: 16(35), P. 46442 - 46453

Published: Aug. 26, 2024

Solid-state electrolytes with argyrodite structures, such as Li

Language: Английский

---

Atomistic Simulation of HF Etching Process of Amorphous Si₃N₄ Using Machine Learning Potential

ACS Applied Materials & Interfaces, Journal Year: 2024, Volume and Issue: 16(36), P. 48457 - 48469

Published: Aug. 28, 2024

An atomistic understanding of dry-etching processes with reactive molecules is crucial for achieving geometric integrity in highly scaled semiconductor devices. Molecular dynamics (MD) simulations are instrumental, but the lack reliable force fields hinders widespread use MD etching simulations. In this work, we develop an accurate neural network potential (NNP) simulating process amorphous Si3N4 HF molecules. The surface reactions diverse local environments considered by incorporating several types training sets: baseline structures, reaction-specific data, and general-purpose sets. Furthermore, NNP refined through iterative comparisons density functional theory results. Using trained NNP, carry out simulations, which allow detailed observation analysis key such as preferential sputtering, modification, yield, threshold energy, distribution products. Additionally, a simple continuum model, built from simulation results, effectively reproduces composition obtained By establishing computational framework scale bridging, work will pave way more efficient design industry, enhancing device performance manufacturing precision.

Language: Английский

---

Local Structures of Ex-Solved Nanoparticles Identified by Machine-Learned Potentials

Nano Letters, Journal Year: 2024, Volume and Issue: 24(14), P. 4224 - 4232

Published: April 1, 2024

In this study, we identify the local structures of ex-solved nanoparticles using machine-learned potentials (MLPs). We develop a method for training by sampling heterointerface configurations as set with its efficacy tested on Ni/MgO system, illustrating that error in interface energy is only 0.004 eV/Å2. Using developed scheme, train an MLP Ni/La0.5Ca0.5TiO3 ex-solution system and both exo- endo-type particles. The established model aligns well experimental observations, accurately predicting nucleation size 0.45 nm. Lastly, density functional theory calculations atomistic verify kinetic barrier dry reforming methane are substantially reduced 0.49 eV catalysts compared to impregnated catalysts. Our findings offer insights into structures, growth mechanisms, underlying origin catalytic properties nanoparticles.

Language: Английский

---

Applications of machine learning in surfaces and interfaces

Chemical Physics Reviews, Journal Year: 2025, Volume and Issue: 6(1)

Published: March 1, 2025

Surfaces and interfaces play key roles in chemical material science. Understanding physical processes at complex surfaces is a challenging task. Machine learning provides powerful tool to help analyze accelerate simulations. This comprehensive review affords an overview of the applications machine study systems materials. We categorize into following broad categories: solid–solid interface, solid–liquid liquid–liquid surface solid, liquid, three-phase interfaces. High-throughput screening, combined first-principles calculations, force field accelerated molecular dynamics simulations are used rational design such as all-solid-state batteries, solar cells, heterogeneous catalysis. detailed information on for

Language: Английский

---

Fine-Tuned Global Neural Network Potentials for Global Potential Energy Surface Exploration at High Accuracy

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

Published: March 19, 2025

Machine learning potential (MLP), by global energy surfaces (PES), has demonstrated its great value in finding unknown structures and reactions via PES exploration. Due to the diversity complexity of data set, an outstanding challenge emerges achieving high accuracy (e.g., error <1 meV/atom), which is essential determine thermodynamics kinetics properties. Here, we develop a lightweight fine-tuning MLP architecture, namely, AtomFT, that can explore globally simultaneously describe target system accurately. The AtomFT takes pretrained many-body function corrected neural network (MBNN) as basis potential, exploits iteratively updates atomic features from MBNN model, finally generates contribution. By implementing architecture on commonly available CPU platform, show efficiency both training inference demonstrate performance challenging problems, including oxides with low defect content, molecular reactions, crystals─in all systems, potentials enhance significantly prediction 1 meV/atom.

Language: Английский

---

Data-Efficient Multifidelity Training for High-Fidelity Machine Learning Interatomic Potentials

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

Published: Dec. 17, 2024

Machine learning interatomic potentials (MLIPs) are used to estimate potential energy surfaces (PES) from

Language: Английский

---

Fluoride-Ion Conduction by Synergic Rotation of the Anion Sublattice for Tl_4.5SnF_8.5 Analogues

Chemistry of Materials, Journal Year: 2024, Volume and Issue: 36(17), P. 8488 - 8495

Published: Aug. 28, 2024

Fluoride-ion conductors have attracted great attention as solid electrolytes for all-solid-state fluoride-ion batteries with high energy densities surpassing those of conventional lithium-ion batteries. Well-known examples include fluorite-type PbSnF4 intrinsic fluoride vacancies and tysonite-type La0.9Ba0.1F2.9 (LBF) extrinsic introduced by aliovalent substitution. In contrast to the dynamics ions through vacancies, an isolated anion sublattice could provide a unique means interstitial diffusion because its rotational flexibility. this study, we employed Tl4.5SnF8.5, which contains located between SnF6 octahedra, investigated relationship cell volume conductivity upon varying ionic radius tin site substituent fixed carrier amount. Tl4.5Sn0.9Y0.1F8.4 exhibited maximum minimum activation energy. Ball milling material led room-temperature comparable that LBF. Neural-network potential molecular was also used elucidate mechanism. The octahedra were found undergo motion, mediated hopping ions. These new design strategy complement previous approach based on introduction vacancies.

Language: Английский

---

Modified Activation-Relaxation Technique (ARTn) Method Tuned for Efficient Identification of Transition States in Surface Reactions

Journal of Chemical Theory and Computation, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 6, 2024

Exploring potential energy surfaces (PES) is essential for unraveling the underlying mechanisms of chemical reactions and material properties. While activation-relaxation technique (ARTn) a state-of-the-art method identifying saddle points on PES, it often faces challenges in complex landscapes, especially surfaces. In this study, we introduce iso-ARTn, an enhanced ARTn that incorporates constraints orthogonal hyperplane employs adaptive active volume. By leveraging neural network (NNP) to conduct exhaustive point search Pt(111) surface with 0.3 monolayers oxygen coverage, iso-ARTn achieves success rate 8.2% higher than original ARTn, 40% fewer force calls. Moreover, effectively finds various without compromising rate. Combined kinetic Monte Carlo simulations event table construction, NNP demonstrates capability reveal structures consistent experimental observations. This work signifies substantial advancement investigation enhancing both efficiency breadth searches.

Language: Английский

---