Thermal transport of Li$_3$PS$_4$ solid electrolytes with ab initio accuracy DOI Creative Commons
Davide Tisi, Federico Grasselli, Lorenzo Gigli

и другие.

arXiv (Cornell University), Год журнала: 2024, Номер unknown

Опубликована: Янв. 1, 2024

The vast amount of computational studies on electrical conduction in solid-state electrolytes is not mirrored by comparable efforts addressing thermal conduction, which has been scarcely investigated despite its relevance to management and (over)heating batteries. reason for this lies the complexity calculations: one hand, diffusion ionic charge carriers makes lattice methods formally unsuitable, due lack equilibrium atomic positions needed normal-mode expansion. On other prohibitive cost large-scale molecular dynamics (MD) simulations heat transport large systems at ab initio levels hindered use MD-based methods. In paper, we leverage recently developed machine-learning potentials targeting different functionals (PBEsol, r$^2$SCAN, PBE0) a state-of-the-art formulation Green-Kubo theory multicomponent compute conductivity promising electrolyte, Li$_3$PS$_4$, all polymorphs ($\alpha$, $\beta$, $\gamma$). By comparing MD estimates with low-temperature, nondiffusive $\gamma$-Li$_3$PS$_4$, highlight strong anharmonicities negligible nuclear quantum effects, hence further justifying even phases. Finally, ion-conducting $\alpha$ $\beta$ phases, where approach mandatory, our indicate weak temperature dependence conductivity, glass-like behavior effective local disorder characterizing these Li-diffusing

Язык: Английский

Mechanism of Charge Transport in Lithium Thiophosphate DOI Creative Commons

Lorenzo Gigli,

Davide Tisi, Federico Grasselli

и другие.

Chemistry of Materials, Год журнала: 2024, Номер 36(3), С. 1482 - 1496

Опубликована: Фев. 5, 2024

Lithium ortho-thiophosphate (Li3PS4) has emerged as a promising candidate for solid-state electrolyte batteries, thanks to its highly conductive phases, cheap components, and large electrochemical stability range. Nonetheless, the microscopic mechanisms of Li-ion transport in Li3PS4 are far from being fully understood, role PS4 dynamics charge still controversial. In this work, we build machine learning potentials targeting state-of-the-art DFT references (PBEsol, r2SCAN, PBE0) tackle problem all known phases (α, β, γ), system sizes time scales. We discuss physical origin observed superionic behavior Li3PS4: activation flipping drives structural transition phase, characterized by an increase Li-site availability drastic reduction energy diffusion. also rule out any paddle-wheel effects tetrahedra phases─previously claimed enhance diffusion─due orders-of-magnitude difference between rate flips hops at temperatures below melting. finally elucidate interionic dynamical correlations transport, highlighting failure Nernst–Einstein approximation estimate electrical conductivity. Our results show strong dependence on target reference, with PBE0 yielding best quantitative agreement experimental measurements not only electronic band gap but conductivity β- α-Li3PS4.

Язык: Английский

Процитировано

15
Transferability and Accuracy of Ionic Liquid Simulations with Equivariant Machine Learning Interatomic Potentials DOI
Zachary A. H. Goodwin, Malia B. Wenny, Julia H. Yang

и другие.

The Journal of Physical Chemistry Letters, Год журнала: 2024, Номер 15(30), С. 7539 - 7547

Опубликована: Июль 18, 2024

Ionic liquids (ILs) are an exciting class of electrolytes finding applications in many areas from energy storage to solvents, where they have been touted as "designer solvents" can be mixed precisely tailor the physiochemical properties. As using machine learning interatomic potentials (MLIPs) simulate ILs is still relatively unexplored, several questions need answered see if MLIPs transformative for ILs. Since often not pure, but either together or contain additives, we first demonstrate that a MLIP trained compositionally transferable; i.e., applied mixtures ions directly on, while only being on few same ions. We also investigated accuracy novel IL, which experimentally synthesize and characterize. Our ∼200 DFT frames reasonable agreement with our experiments DFT.

Язык: Английский

Процитировано

14
Thermal transport of glasses via machine learning driven simulations DOI Creative Commons
Paolo Pegolo, Federico Grasselli

Frontiers in Materials, Год журнала: 2024, Номер 11

Опубликована: Март 6, 2024

Accessing the thermal transport properties of glasses is a major issue for design production strategies glass industry, as well plethora applications and devices where are employed. From computational standpoint, chemical morphological complexity calls atomistic simulations interatomic potentials able to capture variety local environments, composition, (dis)order that typically characterize glassy phases. Machine-learning (MLPs) emerging valid alternative computationally expensive ab initio simulations, inevitably run on very small samples which cannot account disorder at different scales, empirical force fields, fast but often reliable only in narrow portion thermodynamic composition phase diagrams. In this article, we make point use MLPs compute conductivity glasses, through review recent theoretical tools series numerical vitreous silica silicon, both pure intercalated with lithium.

Язык: Английский

Процитировано

10
Uncertainty quantification by direct propagation of shallow ensembles DOI Creative Commons
Matthias Kellner, Michele Ceriotti

Machine Learning Science and Technology, Год журнала: 2024, Номер 5(3), С. 035006 - 035006

Опубликована: Июнь 17, 2024

Abstract Statistical learning algorithms provide a generally-applicable framework to sidestep time-consuming experiments, or accurate physics-based modeling, but they introduce further source of error on top the intrinsic limitations experimental theoretical setup. Uncertainty estimation is essential quantify this error, and make application data-centric approaches more trustworthy. To ensure that uncertainty quantification used widely, one should aim for are accurate, also easy implement apply. In particular, including an existing architecture be straightforward, add minimal computational overhead. Furthermore, it manipulate combine multiple machine-learning predictions, propagating over modeling steps. We compare several well-established frameworks against these requirements, propose practical approach, which we dub direct propagation shallow ensembles, provides good compromise between ease use accuracy. present benchmarks generic datasets, in-depth study applications field atomistic machine chemistry materials. These examples underscore importance using formulation allows errors without making strong assumptions correlations different predictions model.

Язык: Английский

Процитировано

9
Analyzing Atomic Interactions in Molecules as Learned by Neural Networks DOI Creative Commons
Malte Esders,

Thomas Schnake,

Jonas Lederer

и другие.

Journal of Chemical Theory and Computation, Год журнала: 2025, Номер unknown

Опубликована: Янв. 10, 2025

While machine learning (ML) models have been able to achieve unprecedented accuracies across various prediction tasks in quantum chemistry, it is now apparent that accuracy on a test set alone not guarantee for robust chemical modeling such as stable molecular dynamics (MD). To go beyond accuracy, we use explainable artificial intelligence (XAI) techniques develop general analysis framework atomic interactions and apply the SchNet PaiNN neural network models. We compare these with of fundamental principles understand how well learned underlying physicochemical concepts from data. focus strength different species, predictions intensive extensive properties are made, analyze decay many-body nature interatomic distance. Models deviate too far known physical produce unstable MD trajectories, even when they very high energy force accuracy. also suggest further improvements ML architectures better account polynomial interactions.

Язык: Английский

Процитировано

1
Understanding Defects in Amorphous Silicon with Million‐Atom Simulations and Machine Learning DOI Creative Commons
Joe D. Morrow, C. Ugwumadu, D. A. Drabold

и другие.

Angewandte Chemie International Edition, Год журнала: 2024, Номер 63(22)

Опубликована: Март 22, 2024

The structure of amorphous silicon (a-Si) is widely thought as a fourfold-connected random network, and yet it defective atoms, with fewer or more than four bonds, that make particularly interesting. Despite many attempts to explain such "dangling-bond" "floating-bond" defects, respectively, unified understanding still missing. Here, we use advanced computational chemistry methods reveal the complex structural energetic landscape defects in a-Si. We study an ultra-large-scale, quantum-accurate model containing million thousands individual allowing reliable defect-related statistics be obtained. combine descriptors machine-learned atomic energies develop classification different types results suggest revision established floating-bond by showing fivefold-bonded atoms a-Si exhibit wide range local environments-analogous fivefold centers coordination chemistry. Furthermore, shown (but not threefold) tend cluster together. Our provides new insights into one most studied solids, has general implications for disordered materials beyond alone.

Язык: Английский

Процитировано

6
Adaptive energy reference for machine-learning models of the electronic density of states DOI
Wei Bin How, Sanggyu Chong, Federico Grasselli

и другие.

Physical Review Materials, Год журнала: 2025, Номер 9(1)

Опубликована: Янв. 22, 2025

Язык: Английский

Процитировано

0
Interpolating numerically exact many-body wave functions for accelerated molecular dynamics DOI Creative Commons
Yannic Rath, George H. Booth

Nature Communications, Год журнала: 2025, Номер 16(1)

Опубликована: Фев. 26, 2025

Abstract While there have been many developments in computational probes of both strongly-correlated molecular systems and machine-learning accelerated dynamics, remains a significant gap capabilities simulating accurate non-local electronic structure over timescales on which atoms move. We develop an approach to bridge these fields with practical interpolation scheme for the correlated many-electron state through space atomic configurations, whilst avoiding exponential complexity underlying states. With small number wave functions as training set, we demonstrate provable convergence near-exact potential energy surfaces subsequent dynamics propagation valid many-body function inference its variational retaining mean-field scaling. This represents profoundly different paradigm direct established approaches. combine this modern approaches systematically resolve trajectories converge thermodynamic quantities high-throughput several million interpolated explicit validation their accuracy from only few numerically exact quantum chemical calculations. also highlight comparison traditional machine-learned potentials or surfaces.

Язык: Английский

Процитировано

0
Shifting sands of hardware and software in exascale quantum mechanical simulations DOI
Ravindra Shinde, Claudia Filippi, Anthony Scemama

и другие.

Nature Reviews Physics, Год журнала: 2025, Номер unknown

Опубликована: Апрель 25, 2025

Язык: Английский

Процитировано

0
Machine Learning and Statistical Mechanics: Shared Synergies for Next Generation of Chemical Theory and Computation DOI
Rose K. Cersonsky,

Bingqing Cheng,

Marco De Vivo

и другие.

Journal of Chemical Theory and Computation, Год журнала: 2025, Номер unknown

Опубликована: Май 9, 2025

Язык: Английский

Процитировано

0