Direct Simulation of LiNi0.8Mn0.1Co0.1O2 Transport Properties Using an Efficient and Accurate Machine Learning Potential

Constantijn H. J. A. van de Wetering; Jian He; Nongnuch Artrith; Rolande W. Nolsen

arxiv: 2605.19747 · v1 · pith:5M6PNOTYnew · submitted 2026-05-19 · ❄️ cond-mat.mtrl-sci

Direct Simulation of LiNi0.8Mn0.1Co0.1O2 Transport Properties Using an Efficient and Accurate Machine Learning Potential

Jian He , Constantijn H. J. A. van de Wetering , Rolande W. Nolsen , Nongnuch Artrith This is my paper

Pith reviewed 2026-05-20 04:25 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci

keywords machine learning potentialNMC811lithium diffusionmolecular dynamicscathode transportactive learningbattery materials

0 comments

The pith

A machine learning potential built from limited DFT data enables direct large-scale simulation of lithium self-diffusion in NMC811.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper aims to establish that an efficient machine learning potential can reproduce the accuracy of density functional theory for lithium ion movement in NMC811 while running on far larger systems and longer times than direct DFT allows. A sympathetic reader would care because understanding the actual rates and mechanisms of lithium transport is necessary to improve fast-charging performance in high-nickel cathodes. The work shows that active learning on a near-ground-state dataset starting from a fine-tuned foundation model produces a usable potential with only modest reference data.

Core claim

The authors show that a machine learning potential for LiNi0.8Mn0.1Co0.1O2, constructed by fine-tuning a MACE foundation model and refining it through active learning on a near-ground-state dataset with a limited number of DFT calculations, supports molecular dynamics trajectories that yield lithium diffusion coefficients while preserving DFT-level accuracy at time and length scales inaccessible to direct density functional theory.

What carries the argument

The machine learning potential for NMC811 trained via active learning on DFT data, which serves as a fast surrogate for the potential energy surface in molecular dynamics of lithium ion transport.

If this is right

Lithium self-diffusion coefficients in NMC811 become directly computable from long, large molecular dynamics trajectories.
Transport mechanisms can be examined at scales where collective effects and realistic defect concentrations appear.
The same potential can be reused for repeated simulations under varied temperatures or compositions without new DFT runs.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The method could be transferred to related layered oxides to map how transition-metal ratios change diffusion rates.
Longer trajectories made possible by the potential might reveal rare events such as lithium trapping near defects.
Direct access to diffusion data at experimental length scales could tighten the link between atomistic models and measured rate capability.

Load-bearing premise

The errors remaining in the trained machine learning potential do not meaningfully alter the lithium diffusion pathways or energy barriers that appear in the molecular dynamics runs.

What would settle it

Running short DFT-based molecular dynamics on small cells and finding that the lithium diffusion coefficients or hop barriers differ substantially from those obtained with the machine learning potential on the same cells.

Figures

Figures reproduced from arXiv: 2605.19747 by Constantijn H. J. A. van de Wetering, Jian He, Nongnuch Artrith, Rolande W. Nolsen.

**Figure 2.** Figure 2: Fine-tuning of the MACE mp-0 model on randomly generated [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗

**Figure 4.** Figure 4: (a) Energy profiles predicted by DFT (solid lines) and the MLP (dashed lines) for [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗

**Figure 5.** Figure 5: (a) Schematic illustration of an NEB calculation on a Li [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗

**Figure 6.** Figure 6: MLP-predicted lithium self-diffusion coefficients from large-scale ænet–LAMMPS [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗

**Figure 7.** Figure 7: (a)Diffusion coefficients obtained at elevated temperatures were extrapolated to [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗

read the original abstract

The rate capability of layered lithium nickel manganese cobalt oxide (NMC) cathode materials plays a decisive role in high-power applications such as fast charging, necessitating a detailed understanding of lithium-ion diffusion. However, the mechanisms governing lithium-ion transport in NMC remain insufficiently understood, both experimentally and computationally. In this study, we employ an advanced and efficient machine learning potential (MLP) to simulate lithium self-diffusion in LiNi0.8Mn0.1Co0.1O2 (NMC811), enabling direct large-scale molecular dynamics (MD) simulations. The workflow integrates a fine-tuned MACE (Message Passing Atomic Cluster Expansion) foundation model as a structural generator and leverages an active learning strategy applied to a near-ground-state dataset. This approach enables the construction of a reliable MLP for NMC811 in a data-efficient manner using a limited number of density functional theory (DFT) reference calculations. Based on this potential, we performed MD simulations to predict lithium diffusion coefficients. The MLP-based simulations preserve the accuracy of DFT while overcoming its time and length scale limitations, thereby allowing direct simulation of lithium self-diffusion in NMC811.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This applies a fine-tuned MACE model with active learning to run larger MD simulations of Li diffusion in NMC811, but the near-ground-state dataset leaves the barrier accuracy open to question.

read the letter

The paper uses a fine-tuned MACE foundation model and active learning to build a machine learning potential for NMC811, then runs molecular dynamics to compute lithium self-diffusion coefficients directly. This is the main point: getting transport data at system sizes that DFT cannot reach. They keep the number of DFT calculations low by focusing active learning on near-ground-state structures. The resulting MLP is then used for long MD runs to extract diffusion properties. This workflow is straightforward and targets a material that matters for fast-charging batteries. The paper does well at showing a practical route for larger-scale simulations in this specific cathode. It follows known methods for ML potentials but applies them to direct diffusion studies in NMC811, which adds a useful example. The soft spot is around the active learning data. Limiting it to near-ground-state configurations risks missing the transition-state structures that determine Li hop barriers. Because diffusion is exponentially sensitive to those barriers, even small inaccuracies there could change the predicted coefficients. The abstract lacks details on error bars, barrier validations, or comparisons to prior work, so the claim of preserved DFT accuracy is hard to verify without more evidence from the full text. If the manuscript includes those checks, the concern shrinks. Otherwise, it remains a question for reviewers. This is aimed at researchers in battery materials modeling who use or develop ML potentials. A reader wanting to see how these tools perform on lithium transport in layered oxides would find it relevant. It has enough grounding to go through peer review rather than being rejected outright. I recommend sending it for refereeing.

Referee Report

2 major / 1 minor

Summary. The manuscript develops a machine learning potential (MLP) for LiNi0.8Mn0.1Co0.1O2 (NMC811) by fine-tuning a MACE foundation model and applying active learning to a near-ground-state dataset using a limited number of DFT calculations. This MLP is then employed in large-scale molecular dynamics simulations to compute lithium self-diffusion coefficients, with the central claim that the approach preserves DFT accuracy while overcoming the time- and length-scale limitations of direct DFT-based simulations.

Significance. If the MLP accurately reproduces DFT-level barriers and pathways, the work would enable direct, large-scale simulation of activated Li transport in high-Ni layered cathodes, providing mechanistic insight into rate-limiting processes relevant to fast-charging battery applications.

major comments (2)

[Methods section describing the active-learning protocol] The active learning workflow is described as being applied exclusively to a near-ground-state dataset. Because Li diffusion in NMC811 is an activated process whose rate depends exponentially on barrier height, configurations near octahedral-to-tetrahedral saddle points or those involving local transition-metal disorder lie outside this manifold. Without reported validation of force or energy errors on such transition-state structures (or direct comparison of MLP vs. DFT barrier heights), it remains unclear whether the extracted MD diffusion coefficients are free of systematic bias.
[Results section on MD-derived transport properties] The results section reports lithium diffusion coefficients from MLP-MD but provides neither error bars on the coefficients nor quantitative benchmarks against experimental tracer diffusion data or prior ab initio MD studies. Such comparisons are required to substantiate the claim that the MLP 'preserves the accuracy of DFT'.

minor comments (1)

[Abstract] The abstract asserts that the MLP 'preserves the accuracy of DFT' without accompanying quantitative metrics (e.g., force MAE on validation sets or barrier errors); this phrasing should be qualified or supported by explicit numbers in the main text.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review. The comments raise important points about validation of the MLP for activated diffusion processes and the need for quantitative benchmarks. We have addressed both concerns by planning additional analyses and comparisons that will be incorporated into the revised manuscript.

read point-by-point responses

Referee: [Methods section describing the active-learning protocol] The active learning workflow is described as being applied exclusively to a near-ground-state dataset. Because Li diffusion in NMC811 is an activated process whose rate depends exponentially on barrier height, configurations near octahedral-to-tetrahedral saddle points or those involving local transition-metal disorder lie outside this manifold. Without reported validation of force or energy errors on such transition-state structures (or direct comparison of MLP vs. DFT barrier heights), it remains unclear whether the extracted MD diffusion coefficients are free of systematic bias.

Authors: We appreciate the referee's emphasis on the need to validate the MLP specifically for transition-state configurations, given the exponential sensitivity of diffusion rates to barrier heights. The active-learning protocol in the original manuscript was deliberately restricted to near-ground-state structures to achieve data efficiency with a limited DFT budget. To directly address this concern, the revised manuscript will include additional DFT reference calculations on representative saddle-point and disordered configurations (sampled via short exploratory DFT-MD and NEB paths). We will report force and energy errors on these structures and provide a side-by-side comparison of MLP-predicted versus DFT-computed Li diffusion barriers for the dominant octahedral-to-tetrahedral hops. These new results will be presented in an expanded Methods section and a supplementary figure, demonstrating that the MLP reproduces the relevant barriers without introducing systematic bias. revision: yes
Referee: [Results section on MD-derived transport properties] The results section reports lithium diffusion coefficients from MLP-MD but provides neither error bars on the coefficients nor quantitative benchmarks against experimental tracer diffusion data or prior ab initio MD studies. Such comparisons are required to substantiate the claim that the MLP 'preserves the accuracy of DFT'.

Authors: We agree that error estimates and external benchmarks are necessary to strengthen the accuracy claim. In the revised manuscript we will add statistical error bars to all reported lithium self-diffusion coefficients, derived from the standard error across at least five independent, long MD trajectories. We will also insert a new comparison table (or figure) that quantitatively benchmarks the MLP-MD diffusivities against available experimental tracer-diffusion data for NMC811 and against prior ab initio MD results for related NMC compositions. These additions will be placed in the Results section and will allow readers to directly assess how well the MLP-MD values align with both experiment and direct DFT simulations. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected in derivation chain

full rationale

The paper's central workflow begins with external DFT reference calculations, applies active learning to train an MLP on a near-ground-state dataset, and then runs MD simulations to extract lithium self-diffusion coefficients. This chain depends on independent quantum-mechanical data and standard dynamical propagation rather than any self-referential fitting, parameter renaming, or load-bearing self-citation. No equation or step reduces by construction to a quantity already determined inside the paper; the reported transport properties are genuine predictions from the trained potential and the dynamics.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the premise that a data-efficient active-learning MLP trained on near-ground-state configurations can faithfully reproduce lithium hopping barriers and collective diffusion at finite temperature without systematic bias from the training distribution.

axioms (1)

domain assumption The MACE foundation model can be fine-tuned to NMC811 chemistry with a small number of DFT calculations while retaining transferability to diffusion events.
Invoked in the workflow description as the basis for using the pre-trained model as structural generator.

pith-pipeline@v0.9.0 · 5768 in / 1204 out tokens · 30362 ms · 2026-05-20T04:25:21.970525+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

The workflow integrates a fine-tuned MACE foundation model as a structural generator and leverages an active learning strategy applied to a near-ground-state dataset... NEB calculations... MD simulations... Einstein relation for diffusion coefficient
IndisputableMonolith/Foundation/ArrowOfTime.lean entropy_from_berry unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

lithium self-diffusion coefficients... Arrhenius extrapolation

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages

[1]

S.; Sun, Y.-K

Noh, H.-J.; Youn, S.; Yoon, C. S.; Sun, Y.-K. Comparison of the structural and electro- chemical properties of layered Li[NixCoyMnz]O2 (x = 1/3, 0.5, 0.6, 0.7, 0.8 and 0.85) 17 cathode material for lithium-ion batteries.J. Power Sources2013,233, 121–130

work page
[2]

Hierarchical nickel valence gradient stabilizes high-nickel content layered cathode materials.Nat

Lin, R.; Bak, S.-M.; Shin, Y.; Zhang, R.; Wang, C.; Kisslinger, K.; Ge, M.; Huang, X.; Shadike, Z.; Pattammattel, A. Hierarchical nickel valence gradient stabilizes high-nickel content layered cathode materials.Nat. Commun.2021,12, 2350

work page 2021
[3]

M.; Manthiram, A

Li, W.; Erickson, E. M.; Manthiram, A. High-nickel layered oxide cathodes for lithium- based automotive batteries.Nat. Energy2020,5, 26–34

work page
[4]

Learning from Electrochemical Data: Simple Evaluation and Classification of LiMO2-type-based Positive Electrodes for Li-Ion Batteries.Energy Technol.2017,5, 1670–1679

Kasnatscheew, J.; Evertz, M.; Kloepsch, R.; Streipert, B.; Wagner, R.; Cekic Laskovic, I.; Winter, M. Learning from Electrochemical Data: Simple Evaluation and Classification of LiMO2-type-based Positive Electrodes for Li-Ion Batteries.Energy Technol.2017,5, 1670–1679

work page 2017
[5]

G.; Anthonisamy, N

McClelland, I.; Booth, S. G.; Anthonisamy, N. N.; Middlemiss, L. A.; P´ erez, G. E.; Cussen, E. J.; Baker, P. J.; Cussen, S. A. Direct Observation of Dynamic Lithium Diffusion Behavior in Nickel-Rich, LiNi 0.8 Mn0.1 Co0.1 O2 (NMC811) Cathodes Using OperandoMuon Spectroscopy.Chem. Mater.2023,35, 4149–4158

work page 2023
[6]

McNulty, R.; Hampson, E.; N

C. McNulty, R.; Hampson, E.; N. Cutler, L.; P. Grey, C.; M. Dose, W.; R. Johnson, L. Understanding the limits of Li-NMC811 half-cells.J. Mater. Chem. A2023,11, 18302– 18312

work page
[7]

Tubtimkuna, S.; Phattharasupakun, N.; Bunyanidhi, P.; Sawangphruk, M. Diffusion of Zirconium (IV) Ions from Coated Thick Zirconium Oxide Shell to the Bulk Structure of Ni-Rich NMC811 Cathode Leading to High-Performance 18650 Cylindrical Li-Ion Batteries.Adv. Mater. Technol.2022,7, 2200436

work page 2022
[8]

J.; Xu, C.; Griffith, K

M¨ arker, K.; Reeves, P. J.; Xu, C.; Griffith, K. J.; Grey, C. P. Evolution of Structure and Lithium Dynamics in LiNi0.8 Mn0.1 Co0.1 O2 (NMC811) Cathodes during Electrochemical Cycling.Chem. Mater.2019,31, 2545–2554. 18

work page 2019
[9]

S.; Brant, W

Chien, Y.-C.; Liu, H.; Menon, A. S.; Brant, W. R.; Brandell, D.; Lacey, M. J. Rapid determination of solid-state diffusion coefficients in Li-based batteries via intermittent current interruption method.Nat. Commun.2023,14, 2289

work page 2023
[10]

Van der Ven, A.; Ceder, G.; Asta, M.; Tepesch, P. D. First-principles theory of ionic diffusion with nondilute carriers.Phys. Rev. B2001,64, 184307

work page
[11]

Ven, A. V. d.; Ceder, G. Lithium Diffusion in Layered LixCoO2.Electrochem. Solid-State Lett.2000,3, 301

work page 2000
[12]

Lithium diffusion mechanisms in layered intercalation com- pounds.J

Van der Ven, A.; Ceder, G. Lithium diffusion mechanisms in layered intercalation com- pounds.J. Power Sources2001,97-98, 529–531

work page
[13]

Defect Physics and Chemistry in Layered Mixed Transition Metal Oxide Cathode Materials: (Ni,Co,Mn) vs (Ni,Co,Al).Chem

Hoang, K.; Johannes, M. Defect Physics and Chemistry in Layered Mixed Transition Metal Oxide Cathode Materials: (Ni,Co,Mn) vs (Ni,Co,Al).Chem. Mater.2016,28, 1325–1334

work page 2016
[14]

L.; Song, J.; Gauvin, R

Jaberi, A.; Trudeau, M. L.; Song, J.; Gauvin, R. Study of Lithium Transport in NMC Layered Oxide Cathode Material Using Multiscale Computational Approach.ACS Appl. Energy Mater.2024,7, 7724–7736

work page 2024
[15]

S.; Markovsky, B.; Aurbach, D.; Major, D

Dixit, M.; Kosa, M.; Lavi, O. S.; Markovsky, B.; Aurbach, D.; Major, D. T. Thermody- namic and kinetic studies of LiNi 0.5 Co 0.2 Mn 0.3 O 2 as a positive electrode material for Li-ion batteries using first principles.Phys. Chem. Chem. Phys.2016,18, 6799–6812

work page 2016
[16]

Accelerated Atomistic Modeling of Solid-State Battery Materials With Machine Learning.Front

Guo, H.; Wang, Q.; Stuke, A.; Urban, A.; Artrith, N. Accelerated Atomistic Modeling of Solid-State Battery Materials With Machine Learning.Front. Energy Res.2021,9

work page 2021
[17]

Generalized Neural-Network Representation of High- Dimensional Potential-Energy Surfaces.Phys

Behler, J.; Parrinello, M. Generalized Neural-Network Representation of High- Dimensional Potential-Energy Surfaces.Phys. Rev. Lett.2007,98, 146401

work page 2007
[18]

P.; Kondor, R.; Cs´ anyi, G

Bart´ ok, A. P.; Kondor, R.; Cs´ anyi, G. On representing chemical environments.Phys. Rev. B2013,87, 184115. 19

work page
[19]

An implementation of artificial neural-network potentials for atomistic materials simulations: Performance for TiO2.Comput

Artrith, N.; Urban, A. An implementation of artificial neural-network potentials for atomistic materials simulations: Performance for TiO2.Comput. Mater. Sci.2016,114, 135–150

work page 2016
[20]

Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics.Phys

Zhang, L.; Han, J.; Wang, H.; Car, R.; E, W. Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics.Phys. Rev. Lett.2018,120, 143001

work page 2018
[21]

The design space of e (3)-equivariant atom-centered interatomic potentials.arXiv preprint arXiv:2205.06643, 2022

Batatia, I.; Batzner, S.; Kov´ acs, D. P.; Musaelian, A.; Simm, G. N. C.; Drautz, R.; Ort- ner, C.; Kozinsky, B.; Cs´ anyi, G. The Design Space of E(3)-Equivariant Atom-Centered Interatomic Potentials. 2022;http://arxiv.org/abs/2205.06643

work page arXiv 2022
[22]

P.; Simm, G.; Ortner, C.; Csanyi, G

Batatia, I.; Kovacs, D. P.; Simm, G.; Ortner, C.; Csanyi, G. MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fields.Adv. Neu. Inf. Pro. Sys.2022,35, 11423–11436

work page 2022
[23]

Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set.Comput

Kresse, G.; Furthm¨ uller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set.Comput. Mater. Sci.1996,6, 15–50

work page 1996
[24]

Efficient iterative schemes for ab initio total-energy calcula- tions using a plane-wave basis set.Phys

Kresse, G.; Furthm¨ uller, J. Efficient iterative schemes for ab initio total-energy calcula- tions using a plane-wave basis set.Phys. Rev. B1996,54, 11169–11186

work page
[25]

P.; Burke, K.; Ernzerhof, M

Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple.Phys. Rev. Lett.1996,77, 3865–3868

work page 1996
[26]

A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu.J

Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu.J. Chem. Phys.2010,132, 154104

work page 2010
[27]

Hubbard-corrected DFT energy functionals: The LDA+U description of correlated systems.Int

Himmetoglu, B.; Floris, A.; De Gironcoli, S.; Cococcioni, M. Hubbard-corrected DFT energy functionals: The LDA+U description of correlated systems.Int. J. Qua. Chem. 2014,114, 14–49. 20

work page 2014
[28]

A.; Morgan, D.; Ceder, G

Zhou, F.; Cococcioni, M.; Marianetti, C. A.; Morgan, D.; Ceder, G. First-principles prediction of redox potentials in transition-metal compounds with LDA + U.Phys. Rev. B2004,70, 235121

work page
[29]

C.; Bare˜ no, J.; Yan, J.; Chen, G.; Hauser, A.; Croy, J

Garcia, J. C.; Bare˜ no, J.; Yan, J.; Chen, G.; Hauser, A.; Croy, J. R.; Iddir, H. Surface Structure, Morphology, and Stability of Li(Ni 1/3 Mn1/3 Co1/3 )O2 Cathode Material.J. Phys. Chem. C2017,121, 8290–8299

work page
[30]

M.; Ghanty, C.; Major, D

Schipper, F.; Dixit, M.; Kovacheva, D.; Talianker, M.; Haik, O.; Grinblat, J.; Erick- son, E. M.; Ghanty, C.; Major, D. T.; Markovsky, B. Stabilizing nickel-rich layered cathode materials by a high-charge cation doping strategy: zirconium-doped LiNi 0.6 Co 0.2 Mn 0.2 O 2.J. Mater. Chem. A2016,4, 16073–16084

work page
[31]

Constructing first-principles phase diagrams of amor- phous LixSi using machine-learning-assisted sampling with an evolutionary algorithm

Artrith, N.; Urban, A.; Ceder, G. Constructing first-principles phase diagrams of amor- phous LixSi using machine-learning-assisted sampling with an evolutionary algorithm. J. Chem. Phys.2018,148

work page 2018
[33]

Kong, L.; Li, J.; Sun, L.; Yang, H.; Hao, H.; Chen, C.; Artrith, N.; Torres, J. A. G.; Lu, Z.; Zhou, Y. Overcoming the Size Limit of First Principles Molecular Dynamics Simulations with an In-Distribution Substructure Embedding Active Learner.2023,

work page 2023
[34]

DP-GEN: A concurrent learning platform for the generation of reliable deep learning based potential energy models.Com

Zhang, Y.; Wang, H.; Chen, W.; Zeng, J.; Zhang, L.; Wang, H.; E, W. DP-GEN: A concurrent learning platform for the generation of reliable deep learning based potential energy models.Com. Phys. Commun.2020,253, 107206

work page 2020
[35]

Andrade, M. F. C.; Ko, H.-Y.; Zhang, L.; Car, R.; Selloni, A. Free energy of proton transfer at the water–TiO 2 interface from ab initio deep potential molecular dynamics. Chem. Sci.2020,11, 2335–2341. 21

work page 2020
[36]

Insights into the Atomic Mechanism of Lithium-Ion Diffusion in Li 6PS5Cl via a Machine Learning Potential.Chem

Chen, J.; Fang, M.; Wu, Q.; Tang, S.; Zheng, J.; Wei, C.; Cao, X.; Shi, Y.; Xu, N.; He, Y. Insights into the Atomic Mechanism of Lithium-Ion Diffusion in Li 6PS5Cl via a Machine Learning Potential.Chem. Mater.2025,37, 591–599

work page 2025
[37]

Pushing the Limit of Molecular Dynamics with Ab Initio Accuracy to 100 Million Atoms with Machine Learning

Jia, W.; Wang, H.; Chen, M.; Lu, D.; Lin, L.; Car, R.; Weinan, E.; Zhang, L. Pushing the Limit of Molecular Dynamics with Ab Initio Accuracy to 100 Million Atoms with Machine Learning. Proc. Int. Conf. High Perform. Comput. Netw. Storage Anal. 2020; pp 1–14

work page 2020
[38]

Efficient and accurate machine-learning interpolation of atomic energies in compositions with many species.Phys

Artrith, N.; Urban, A.; Ceder, G. Efficient and accurate machine-learning interpolation of atomic energies in compositions with many species.Phys. Rev. B2017,96, 014112. 22 Supporting Information Direct Simulation of LiNi 0.8Mn0.1Co0.1O2 Transport Properties Using an Efficient and Accurate Machine Learning Potential Jian He, Constantijn H. J. A. van de We...

work page 2000

[1] [1]

S.; Sun, Y.-K

Noh, H.-J.; Youn, S.; Yoon, C. S.; Sun, Y.-K. Comparison of the structural and electro- chemical properties of layered Li[NixCoyMnz]O2 (x = 1/3, 0.5, 0.6, 0.7, 0.8 and 0.85) 17 cathode material for lithium-ion batteries.J. Power Sources2013,233, 121–130

work page

[2] [2]

Hierarchical nickel valence gradient stabilizes high-nickel content layered cathode materials.Nat

Lin, R.; Bak, S.-M.; Shin, Y.; Zhang, R.; Wang, C.; Kisslinger, K.; Ge, M.; Huang, X.; Shadike, Z.; Pattammattel, A. Hierarchical nickel valence gradient stabilizes high-nickel content layered cathode materials.Nat. Commun.2021,12, 2350

work page 2021

[3] [3]

M.; Manthiram, A

Li, W.; Erickson, E. M.; Manthiram, A. High-nickel layered oxide cathodes for lithium- based automotive batteries.Nat. Energy2020,5, 26–34

work page

[4] [4]

Learning from Electrochemical Data: Simple Evaluation and Classification of LiMO2-type-based Positive Electrodes for Li-Ion Batteries.Energy Technol.2017,5, 1670–1679

Kasnatscheew, J.; Evertz, M.; Kloepsch, R.; Streipert, B.; Wagner, R.; Cekic Laskovic, I.; Winter, M. Learning from Electrochemical Data: Simple Evaluation and Classification of LiMO2-type-based Positive Electrodes for Li-Ion Batteries.Energy Technol.2017,5, 1670–1679

work page 2017

[5] [5]

G.; Anthonisamy, N

McClelland, I.; Booth, S. G.; Anthonisamy, N. N.; Middlemiss, L. A.; P´ erez, G. E.; Cussen, E. J.; Baker, P. J.; Cussen, S. A. Direct Observation of Dynamic Lithium Diffusion Behavior in Nickel-Rich, LiNi 0.8 Mn0.1 Co0.1 O2 (NMC811) Cathodes Using OperandoMuon Spectroscopy.Chem. Mater.2023,35, 4149–4158

work page 2023

[6] [6]

McNulty, R.; Hampson, E.; N

C. McNulty, R.; Hampson, E.; N. Cutler, L.; P. Grey, C.; M. Dose, W.; R. Johnson, L. Understanding the limits of Li-NMC811 half-cells.J. Mater. Chem. A2023,11, 18302– 18312

work page

[7] [7]

Tubtimkuna, S.; Phattharasupakun, N.; Bunyanidhi, P.; Sawangphruk, M. Diffusion of Zirconium (IV) Ions from Coated Thick Zirconium Oxide Shell to the Bulk Structure of Ni-Rich NMC811 Cathode Leading to High-Performance 18650 Cylindrical Li-Ion Batteries.Adv. Mater. Technol.2022,7, 2200436

work page 2022

[8] [8]

J.; Xu, C.; Griffith, K

M¨ arker, K.; Reeves, P. J.; Xu, C.; Griffith, K. J.; Grey, C. P. Evolution of Structure and Lithium Dynamics in LiNi0.8 Mn0.1 Co0.1 O2 (NMC811) Cathodes during Electrochemical Cycling.Chem. Mater.2019,31, 2545–2554. 18

work page 2019

[9] [9]

S.; Brant, W

Chien, Y.-C.; Liu, H.; Menon, A. S.; Brant, W. R.; Brandell, D.; Lacey, M. J. Rapid determination of solid-state diffusion coefficients in Li-based batteries via intermittent current interruption method.Nat. Commun.2023,14, 2289

work page 2023

[10] [10]

Van der Ven, A.; Ceder, G.; Asta, M.; Tepesch, P. D. First-principles theory of ionic diffusion with nondilute carriers.Phys. Rev. B2001,64, 184307

work page

[11] [11]

Ven, A. V. d.; Ceder, G. Lithium Diffusion in Layered LixCoO2.Electrochem. Solid-State Lett.2000,3, 301

work page 2000

[12] [12]

Lithium diffusion mechanisms in layered intercalation com- pounds.J

Van der Ven, A.; Ceder, G. Lithium diffusion mechanisms in layered intercalation com- pounds.J. Power Sources2001,97-98, 529–531

work page

[13] [13]

Defect Physics and Chemistry in Layered Mixed Transition Metal Oxide Cathode Materials: (Ni,Co,Mn) vs (Ni,Co,Al).Chem

Hoang, K.; Johannes, M. Defect Physics and Chemistry in Layered Mixed Transition Metal Oxide Cathode Materials: (Ni,Co,Mn) vs (Ni,Co,Al).Chem. Mater.2016,28, 1325–1334

work page 2016

[14] [14]

L.; Song, J.; Gauvin, R

Jaberi, A.; Trudeau, M. L.; Song, J.; Gauvin, R. Study of Lithium Transport in NMC Layered Oxide Cathode Material Using Multiscale Computational Approach.ACS Appl. Energy Mater.2024,7, 7724–7736

work page 2024

[15] [15]

S.; Markovsky, B.; Aurbach, D.; Major, D

Dixit, M.; Kosa, M.; Lavi, O. S.; Markovsky, B.; Aurbach, D.; Major, D. T. Thermody- namic and kinetic studies of LiNi 0.5 Co 0.2 Mn 0.3 O 2 as a positive electrode material for Li-ion batteries using first principles.Phys. Chem. Chem. Phys.2016,18, 6799–6812

work page 2016

[16] [16]

Accelerated Atomistic Modeling of Solid-State Battery Materials With Machine Learning.Front

Guo, H.; Wang, Q.; Stuke, A.; Urban, A.; Artrith, N. Accelerated Atomistic Modeling of Solid-State Battery Materials With Machine Learning.Front. Energy Res.2021,9

work page 2021

[17] [17]

Generalized Neural-Network Representation of High- Dimensional Potential-Energy Surfaces.Phys

Behler, J.; Parrinello, M. Generalized Neural-Network Representation of High- Dimensional Potential-Energy Surfaces.Phys. Rev. Lett.2007,98, 146401

work page 2007

[18] [18]

P.; Kondor, R.; Cs´ anyi, G

Bart´ ok, A. P.; Kondor, R.; Cs´ anyi, G. On representing chemical environments.Phys. Rev. B2013,87, 184115. 19

work page

[19] [19]

An implementation of artificial neural-network potentials for atomistic materials simulations: Performance for TiO2.Comput

Artrith, N.; Urban, A. An implementation of artificial neural-network potentials for atomistic materials simulations: Performance for TiO2.Comput. Mater. Sci.2016,114, 135–150

work page 2016

[20] [20]

Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics.Phys

Zhang, L.; Han, J.; Wang, H.; Car, R.; E, W. Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics.Phys. Rev. Lett.2018,120, 143001

work page 2018

[21] [21]

The design space of e (3)-equivariant atom-centered interatomic potentials.arXiv preprint arXiv:2205.06643, 2022

Batatia, I.; Batzner, S.; Kov´ acs, D. P.; Musaelian, A.; Simm, G. N. C.; Drautz, R.; Ort- ner, C.; Kozinsky, B.; Cs´ anyi, G. The Design Space of E(3)-Equivariant Atom-Centered Interatomic Potentials. 2022;http://arxiv.org/abs/2205.06643

work page arXiv 2022

[22] [22]

P.; Simm, G.; Ortner, C.; Csanyi, G

Batatia, I.; Kovacs, D. P.; Simm, G.; Ortner, C.; Csanyi, G. MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fields.Adv. Neu. Inf. Pro. Sys.2022,35, 11423–11436

work page 2022

[23] [23]

Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set.Comput

Kresse, G.; Furthm¨ uller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set.Comput. Mater. Sci.1996,6, 15–50

work page 1996

[24] [24]

Efficient iterative schemes for ab initio total-energy calcula- tions using a plane-wave basis set.Phys

Kresse, G.; Furthm¨ uller, J. Efficient iterative schemes for ab initio total-energy calcula- tions using a plane-wave basis set.Phys. Rev. B1996,54, 11169–11186

work page

[25] [25]

P.; Burke, K.; Ernzerhof, M

Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple.Phys. Rev. Lett.1996,77, 3865–3868

work page 1996

[26] [26]

A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu.J

Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu.J. Chem. Phys.2010,132, 154104

work page 2010

[27] [27]

Hubbard-corrected DFT energy functionals: The LDA+U description of correlated systems.Int

Himmetoglu, B.; Floris, A.; De Gironcoli, S.; Cococcioni, M. Hubbard-corrected DFT energy functionals: The LDA+U description of correlated systems.Int. J. Qua. Chem. 2014,114, 14–49. 20

work page 2014

[28] [28]

A.; Morgan, D.; Ceder, G

Zhou, F.; Cococcioni, M.; Marianetti, C. A.; Morgan, D.; Ceder, G. First-principles prediction of redox potentials in transition-metal compounds with LDA + U.Phys. Rev. B2004,70, 235121

work page

[29] [29]

C.; Bare˜ no, J.; Yan, J.; Chen, G.; Hauser, A.; Croy, J

Garcia, J. C.; Bare˜ no, J.; Yan, J.; Chen, G.; Hauser, A.; Croy, J. R.; Iddir, H. Surface Structure, Morphology, and Stability of Li(Ni 1/3 Mn1/3 Co1/3 )O2 Cathode Material.J. Phys. Chem. C2017,121, 8290–8299

work page

[30] [30]

M.; Ghanty, C.; Major, D

Schipper, F.; Dixit, M.; Kovacheva, D.; Talianker, M.; Haik, O.; Grinblat, J.; Erick- son, E. M.; Ghanty, C.; Major, D. T.; Markovsky, B. Stabilizing nickel-rich layered cathode materials by a high-charge cation doping strategy: zirconium-doped LiNi 0.6 Co 0.2 Mn 0.2 O 2.J. Mater. Chem. A2016,4, 16073–16084

work page

[31] [31]

Constructing first-principles phase diagrams of amor- phous LixSi using machine-learning-assisted sampling with an evolutionary algorithm

Artrith, N.; Urban, A.; Ceder, G. Constructing first-principles phase diagrams of amor- phous LixSi using machine-learning-assisted sampling with an evolutionary algorithm. J. Chem. Phys.2018,148

work page 2018

[32] [33]

Kong, L.; Li, J.; Sun, L.; Yang, H.; Hao, H.; Chen, C.; Artrith, N.; Torres, J. A. G.; Lu, Z.; Zhou, Y. Overcoming the Size Limit of First Principles Molecular Dynamics Simulations with an In-Distribution Substructure Embedding Active Learner.2023,

work page 2023

[33] [34]

DP-GEN: A concurrent learning platform for the generation of reliable deep learning based potential energy models.Com

Zhang, Y.; Wang, H.; Chen, W.; Zeng, J.; Zhang, L.; Wang, H.; E, W. DP-GEN: A concurrent learning platform for the generation of reliable deep learning based potential energy models.Com. Phys. Commun.2020,253, 107206

work page 2020

[34] [35]

Andrade, M. F. C.; Ko, H.-Y.; Zhang, L.; Car, R.; Selloni, A. Free energy of proton transfer at the water–TiO 2 interface from ab initio deep potential molecular dynamics. Chem. Sci.2020,11, 2335–2341. 21

work page 2020

[35] [36]

Insights into the Atomic Mechanism of Lithium-Ion Diffusion in Li 6PS5Cl via a Machine Learning Potential.Chem

Chen, J.; Fang, M.; Wu, Q.; Tang, S.; Zheng, J.; Wei, C.; Cao, X.; Shi, Y.; Xu, N.; He, Y. Insights into the Atomic Mechanism of Lithium-Ion Diffusion in Li 6PS5Cl via a Machine Learning Potential.Chem. Mater.2025,37, 591–599

work page 2025

[36] [37]

Pushing the Limit of Molecular Dynamics with Ab Initio Accuracy to 100 Million Atoms with Machine Learning

Jia, W.; Wang, H.; Chen, M.; Lu, D.; Lin, L.; Car, R.; Weinan, E.; Zhang, L. Pushing the Limit of Molecular Dynamics with Ab Initio Accuracy to 100 Million Atoms with Machine Learning. Proc. Int. Conf. High Perform. Comput. Netw. Storage Anal. 2020; pp 1–14

work page 2020

[37] [38]

Efficient and accurate machine-learning interpolation of atomic energies in compositions with many species.Phys

Artrith, N.; Urban, A.; Ceder, G. Efficient and accurate machine-learning interpolation of atomic energies in compositions with many species.Phys. Rev. B2017,96, 014112. 22 Supporting Information Direct Simulation of LiNi 0.8Mn0.1Co0.1O2 Transport Properties Using an Efficient and Accurate Machine Learning Potential Jian He, Constantijn H. J. A. van de We...

work page 2000