Pith. sign in

REVIEW 3 major objections 4 minor 68 references

Modeling phase transformations in Mn-rich disordered rocksalt cathodes with machine learning interatomic potentials

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Mn3+ migration initiates the rocksalt-to-spinel-like transformation in a Mn-rich cathode, with tetrahedral Mn2+ following the ordering, and the resulting δ-phase shows solid-solution voltage behavior.

desk verdict A solid MLIP-MD study of the DRX-to-spinel transformation with a plausible but proxy-dependent mechanistic claim; referee it, but push for DFT validation of the Mn valence assignments. read the letter →

arxiv 2506.20605 v2 pith:CHYP5SS4 submitted 2025-06-25 cond-mat.mtrl-sci

Peichen Zhong , Bowen Deng , Shashwat Anand , Tara Mishra , Gerbrand Ceder This is my paper
classification cond-mat.mtrl-sci
keywords Mn-richdisorderedrocksaltcathodedelta-phasetransformationmanganesemigrationmachinelearninginteratomicpotentialCHGNetcharge-informedmoleculardynamicsintercalationvoltageprofile0-TMtetrahedralchannels
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims to identify the trigger of the phase transformation that makes Mn-rich disordered rocksalt cathodes perform well in lithium-ion batteries: in Li$_{0.6}$Mn$_{0.8}$Ti$_{0.1}$O$_{1.9}$F$_{0.1}$, the migration of Mn$^{3+}$ ions into tetrahedral sites initiates the rearrangement toward a partially disordered spinel-like $\delta$-phase. Using a machine learning interatomic potential fine-tuned to r2SCAN density functional theory, the authors ran nanosecond, charge-informed molecular dynamics at 1273 K and watched the transformation happen in silico. They find that tetrahedral Mn$^{2+}$ appears only after about 0.6 ns, once long-range spinel-like ordering has begun, so Mn$^{2+}$ is a consequence of the ordering rather than a prerequisite for cation hops. If correct, this overturns the earlier picture that Mn$^{3+}$ disproportionation must produce mobile Mn$^{2+}$ first, and it explains why the transformed $\delta$-phase shows solid-solution low-voltage behavior and higher capacity than either the disordered rocksalt or the ordered spinel.

What carries the argument

The load-bearing machinery is a fine-tuned CHGNet graph neural network potential, trained on 88,852 r2SCAN-DFT structures covering stable and partially disordered Li–Mn–Ti–O–F configurations and TM migration pathways. CHGNet outputs per-atom magnetic moments, which the paper bins into Mn$^{2+}$/Mn$^{3+}$/Mn$^{4+}$ valence labels (4.1–5.0, 3.25–4.1, and 2.5–3.25 $\mu_B$, respectively), making the molecular dynamics 'charge-informed' even though the model does not compute long-range Coulombic interactions explicitly. A cluster expansion supplies the initial disordered structures, and the 0-TM-site enumeration scheme converts the relaxed lithiation states into intercalation voltage profiles.

What would settle it

Run the same transformation with an interatomic potential that includes explicit long-range electrostatics (or with DFT-based molecular dynamics on a smaller cell) and check whether tetrahedral Mn$^{3+}$ still dominates before spinel ordering peaks appear; alternatively, operando X-ray absorption spectroscopy during the first cycles could determine whether tetrahedral Mn$^{2+}$ appears before or after spinel ordering is observed.

Watch

Extended reading notes

Core claim

The central discovery is that in the delithiated structure Li$_{0.6}$Mn$_{0.8}$Ti$_{0.1}$O$_{1.9}$F$_{0.1}$ at 1273 K, transition-metal migration proceeds by Mn$^{3+}$ moving through the octahedral–tetrahedral–octahedral pathway during the first ~0.6 ns, before any substantial tetrahedral Mn$^{2+}$ population appears. The concentration of tetrahedral Mn$^{2+}$ rises only as spinel characteristic XRD peaks and 16$c$/16$d$ site ordering develop, which the paper interprets as Mn$^{2+}$ being stabilized by the emerging spinel-like local environment rather than enabling the migration. The transformed $\delta$-phase contains more 0-TM tetrahedral channels than the initial disordered structure, and its computed voltage profile shows a nearly solid-solution intercalation in the high-Li region, with no sharp two-phase step, giving a higher accessible Li capacity from the 0-TM-to-tetrahedral-Li conversion than the DRX parent.

Load-bearing premise

The entire mechanistic conclusion rests on the assumption that the magnetic moments predicted by the machine learning potential are a faithful proxy for manganese valence states in hot, partially ordered, delithiated structures; if that proxy is biased, the observed order of Mn$^{3+}$ and Mn$^{2+}$ migration could be a model artifact rather than the physical mechanism.

Editorial extensions

If this is right

  • If Mn$^{3+}$ migration initiates the transformation, increasing the population of mobile Mn$^{3+}$ (for example by the right delithiation depth) should accelerate δ-phase formation, and Mn$^{2+}$ need not be introduced beforehand.
  • The δ-phase's higher 0-TM channel concentration directly supports the experimentally observed rate improvement after cycling, because percolating 0-TM channels are the known route for facile Li transport in the FCC anion framework.
  • Partial cation disorder is sufficient to remove the two-phase reaction that plagues ordered spinel LiMn$_2$O$_4$ at high Li content, predicting smoother voltage and less mechanical degradation.
  • The 0-TM-to-Li$_{tet}$ conversion delivers more capacity in the δ-phase than in either the DRX or the ordered spinel, identifying the δ-phase as the active phase responsible for the high capacity of cycled Mn-rich cathodes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A testable corollary the paper leaves implicit: the transformation should require a threshold delithiation level to create enough Mn$^{3+}$ and vacancies; cycling to shallow depths might suppress δ-phase formation, which could be checked by comparing capacity evolution for different voltage windows.
  • The magnetic-moment-to-valence assignment could be checked directly by recomputing the same trajectories with an interatomic potential that adds explicit long-range electrostatics or by comparing time-resolved Mn valence from XANES on an operating cell; a systematic bias would change the ordering of the Mn$^{3+}$ and Mn$^{2+}$ events.
  • Because the paper keeps the transition-metal and anion sublattices fixed when computing voltage profiles, the predicted solid-solution behavior could change if further TM rearrangement or anion motion occurs during slow (de)lithiation; that is a modeling choice, not a result of the paper.
  • The same sampling pipeline could be applied to other Mn-rich compositions, such as the Ti/F variants listed in the methods, to map the composition window over which the δ-phase transformation occurs.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The manuscript reports a machine-learning-potential molecular dynamics study of the transformation from a disordered rocksalt to a partially disordered spinel-like (δ) phase in Li0.6Mn0.8Ti0.1O1.9F0.1. The authors fine-tune the CHGNet interatomic potential on a large r2SCAN-DFT dataset (88,852 structures), run NVT MD at 1273 K from a delithiated disordered configuration, and observe growth of spinel XRD peaks, an energy drop of about 60 meV per anion, and an increase in the occurrence of 0-TM tetrahedra. From CHGNet-predicted magnetic moments, they classify Mn valence states and infer that tetrahedral Mn3+ appears before tetrahedral Mn2+, concluding that Mn3+ migration initiates the ordering while Mn2+ is a consequence of spinel-like ordering. They then compute topotactic Li-vacancy intercalation voltage profiles for the MD-derived δ-phase and compare them with DRX and ordered spinel structures, reporting solid-solution low-voltage behavior and a higher accessible capacity in the δ-phase.

Significance. If the conclusions hold, the paper provides an atomic-scale mechanistic explanation of the DRX-to-δ phase transformation and demonstrates that a fine-tuned, charge-informed MLIP can capture coupled cation migration and cation ordering in a complex oxide. The work has notable strengths: a large and composition-aware training set with reported test errors of 2 meV/atom in energy, 68 meV/Å in force, and 0.019 μB in magnetic moment; a forward simulation that starts from a disordered configuration and produces spinel-like order without fitting the transformation itself; clear definition of the 0-TM to Li-tet conversion processes used to interpret voltage profiles; and publicly available data and code. The central temporal-ordering claim, however, rests entirely on magnetic-moment-based valence assignments whose transferability across octahedral-to-tetrahedral coordination changes is not validated. The electrochemical conclusions also depend on a restricted Li-vacancy configurational search. These issues are fixable within the scope of the manuscript, so the significance is conditional on the additional validation and analysis requested below.

major comments (3)
  1. [III.B, Fig. 5; Methods II.A; Discussion, final paragraph] The conclusion that tetrahedral Mn3+ migration initiates the transformation and that tetrahedral Mn2+ appears only after spinel-like ordering is based entirely on classifying Mn valence from CHGNet-predicted magnetic moments using fixed bins (Mn2+: 4.1–5.0 μB, Mn3+: 3.25–4.1 μB). The bins are calibrated for equilibrium local environments, but the quantity under study is precisely the octahedral-to-tetrahedral coordination change during migration in high-temperature, delithiated, partially ordered structures. A coordination-dependent shift in the predicted local moment—for example, a tetrahedral Mn3+ with a moment above 4.1 μB or a tetrahedral Mn2+ with a moment below 4.1 μB—would make the observed sequence a proxy artifact rather than a physical mechanism. The paper's own Discussion acknowledges that the model does not include explicit long-range electrostatics and uses magnetic moments as a charge proxy, and no DFT recomputation of moments or oxidation states on nonequilibrium MD snapshots is reported. I request a validation step: recompute r2SCAN (or r2SCAN+U) magnetic moments and, if possible, oxidation-state indicators on representative tetrahedral and octahedral Mn environments from the early and late MD trajectories, and show that the bin assignments are stable across coordination. Without this, the central mechanistic claim that Mn3+ migrates before Mn2+ is not established.
  2. [III.A–III.B, Figs. 4–5] The temporal-ordering narrative appears to be drawn from a single 2 ns MD trajectory at 1273 K starting from one randomly delithiated configuration. The onset of spinel ordering and the delayed rise of tetrahedral Mn2+ at approximately 0.6 ns are presented without replicate simulations or any statistical measure, so the sequence could reflect the particular initial configuration rather than a robust mechanistic ordering. I ask the authors to state explicitly how many independent trajectories were used and, if only one, to add several independent starting configurations or otherwise quantify run-to-run variability in the onset times of Mn3+_tet and Mn2+_tet. This is load-bearing because the paper's central claim is a temporal-ordering claim.
  3. [III.C and Appendix C] The claim that the δ-phase exhibits solid-solution intercalation with no two-phase reaction is based on a convex hull built from a greedy sequential enumeration of only 20 Li-vacancy configurations per delithiation step, with the TM and anion sublattices fixed to a single MD-derived structure. Given the large configurational space of a partially disordered phase, this restricted sampling may miss low-energy Li-vacancy orderings, and the absence of a flat voltage plateau could be an artifact of incomplete enumeration rather than a physical property of the δ-phase. Please provide a convergence check with respect to the number of configurations per step (or a more systematic enumeration) for at least one composition, so the solid-solution conclusion can be separated from sampling limitations.
minor comments (4)
  1. [Abstract vs. Section III] The abstract states the simulated composition as Li0.6Mn0.8Ti0.1O1.9F0.1, while the Results section describes the delithiated structure as Li0.5□0.6Mn0.8Ti0.1O1.9F0.1; please reconcile the Li content and the vacancy notation.
  2. [Methods II.B] In the migration-pathway sampling description, the intermediate tetrahedral site is called the 'T h site' several times; this should be 'Td site' to match the earlier definition and standard notation.
  3. [Fig. 5b] The caption and text refer to 'Mn3+tet' and 'Mn2+tet' but the plot axis is described only as occupancy per formula unit; please label the ordinate explicitly so the reader can distinguish tetrahedral occupancy from total Mn content.
  4. [Introduction, first use of 'charge-informed MD'] Because the model uses magnetic moments as a proxy for charge and does not explicitly include long-range Coulombic interactions, a brief clarification at the first use of 'charge-informed' would help avoid overstating what the potential computes; the Discussion already provides this caveat.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the MD simulation is forward, and valence assignments are post-hoc outputs, not fitted constraints.

full rationale

The central mechanistic claim—that Mn3+ migration initiates the disordered-to-delta transformation and tetrahedral Mn2+ emerges only as spinel-like ordering forms—is read off from a 2 ns NVT MD trajectory started from a disordered rocksalt configuration. This trajectory was not constrained to produce spinel-like ordering or any particular Mn valence sequence; the structural evolution and the time-resolved magnetic moments are outputs of the fine-tuned Product-CHGNet, which was trained on r2SCAN-DFT energies, forces, stresses, and magnetic moments. The valence bins (Mn2+: 4.1–5.0 mu_B, Mn3+: 3.25–4.1 mu_B, Mn4+: 2.5–3.25 mu_B) are fixed post-hoc thresholds applied to predicted moments, not parameters fitted to the observed temporal sequence. Therefore, the observed ordering (Mn3+_tet before Mn2+_tet) is an emergent simulation result rather than a restatement of training labels or a fitted constraint. The paper's use of CHGNet [Ref 21] and its consistency check against the earlier CHGNet-based simulation are self-citations, but they are not load-bearing: the new fine-tuned model, the new Li-Mn-Ti-O-F composition, and the new production MD are presented, and the mechanism is inferred from the new trajectory. The acknowledged limitation that magnetic moments are only a proxy for charge and that long-range electrostatics are not explicitly computed is a model-accuracy concern, not a circularity; it does not make the prediction equivalent to an input by construction. No equation in the paper reduces the claimed result to a fitted parameter or to a definition.

Assumptions & free parameters 3 free parameters · 4 assumptions · 2 invented entities

The central simulation relies on fitted or assumed ingredients: the magnetic-moment-to-valence mapping, the CE Hamiltonian for initial structures, the fine-tuned MLIP itself, and the 0-TM conversion framework for voltage. None are invented physical entities; they are modeling choices with stated accuracy claims. The most fragile is the magnetic moment proxy for charge, which directly underpins the mechanistic conclusion about Mn3+ vs Mn2+ migration. The MLIP finetuning is data-driven but relies on a fixed functional form and a specific training set; the test error is good but does not guarantee accuracy for high-energy migration barriers. The voltage profile enumeration is a simplified model.

free parameters (3)
  • Magnetic moment cutoffs for Mn valence bins (Mn2+: 4.1-5.0 mu_B, Mn3+: 3.25-4.1, Mn4+: 2.5-3.25)
    These bins are chosen by hand to categorize Mn valence states from CHGNet magnetic moment predictions. The choice of cutoffs affects the conclusion about which Mn species migrates first. The paper does not justify these cutoffs with DFT magnetic moment calculations on the specific structures.
  • Selection criteria for delithiated structures in voltage profile (minimum/maximum voltage step)
    The voltage profile enumeration uses a greedy selection criterion based on the minimum or maximum voltage at each step. This is a modeling choice that effectively constructs a convex hull-like path from a limited set of 20 structures per step, which may bias the predicted voltage profile toward smooth solid-solution behavior.
  • Number of structures sampled per MD trajectory (200) and delithiation levels (0.2/0.4/0.6 Li per f.u.)
    These are training data sampling hyperparameters that affect MLIP accuracy but are not central physical parameters. They are chosen to provide coverage of the potential energy surface.
assumptions (4)
  • domain assumption CHGNet magnetic moments are a valid proxy for transition metal valence states.
    Used throughout Sec. III.B to infer Mn2+/Mn3+/Mn4+ populations from the MD trajectory. The paper cites prior CHGNet works [21] but does not independently validate the mapping on the simulated structures. The authors note the model does not explicitly compute long-range electrostatics, which could affect charge accuracy.
  • domain assumption The cluster expansion Hamiltonian (83 ECIs fitted to 322 structures) accurately represents configurational thermodynamics of the Li-Mn-Ti-O-F rocksalt system at 1273 K.
    Used to generate the initial disordered structures for MD (Appendix B). The CE is fitted with <8 meV/atom error, reasonable, but transferability to delithiated compositions is assumed.
  • domain assumption The fine-tuned MLIP accurately predicts energies, forces, and magnetic moments for nonequilibrium, partially ordered, delithiated configurations at 1273 K within the 2 ns timescale.
    The central tooling assumption. Training set includes 88,852 structures, but the test set is from a similar composition at 1000 K, and the production 1273 K trajectory is not explicitly validated against DFT. Systematic softening in universal MLIPs is acknowledged [25].
  • domain assumption The 0-TM to Li_tet conversion reaction (Eq. 3) is the dominant low-voltage intercalation mechanism in the delta-phase.
    Defines the framework for voltage profiles. Justified by spinel-like local structure, but in a partially disordered phase other Li insertion sites could contribute. The paper explicitly restricts analysis to this conversion for high Li content.
invented entities (2)
  • The delta-phase (partially disordered spinel-like phase) in Li-Mn-Ti-O-F independent evidence
    purpose: The central object of study; the paper simulates its formation and electrochemical properties.
    The delta-phase is experimentally observed (Cai et al. [9]), not invented. It has been characterized by XRD and electrochemistry. The paper provides simulated structures representing this phase, but does not introduce a new physical entity.
  • Product-CHGNet, the fine-tuned CHGNet MLIP
    purpose: A surrogate potential energy surface model used to run MD and compute energies for voltage profiles.
    This is a computational model, not a physical entity. Its accuracy is assessed by test errors but it is not an independent physical observation. It is a tool, not an invented physical entity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Modeling phase transformations in Mn-rich disordered rocksalt cathodes with machine learning interatomic potentials." pith.science (2026). https://pith.science/paper/CHYP5SS4

@misc{pith2026250620605,
  author       = {Pith},
  title        = {Pith review of: Modeling phase transformations in Mn-rich disordered rocksalt cathodes with machine learning interatomic potentials},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CHYP5SS4}},
  note         = {Machine review of arXiv:2506.20605}
}
abstract

Mn-rich disordered rocksalt (DRX) cathode materials exhibit a phase transformation from a disordered to a partially disordered spinel-like structure ($\delta$-phase) during electrochemical cycling. In this computational study, we used charge-informed molecular dynamics with a fine-tuned CHGNet foundation potential to investigate the phase transformation in Li$_{x}$Mn$_{0.8}$Ti$_{0.1}$O$_{1.9}$F$_{0.1}$. Our results indicate that transition metal migration occurs and reorders to form the spinel-like ordering in an FCC anion framework. The transformed structure contains a higher concentration of non-transition metal (0-TM) face-sharing channels, which are known to improve Li transport kinetics. Analysis of the Mn valence distribution suggests that the appearance of tetrahedral Mn$^{2+}$ is a consequence of spinel-like ordering, rather than the trigger for cation migration as previously suggested. Calculated equilibrium intercalation voltage profiles demonstrate that the $\delta$-phase, unlike the ordered spinel, exhibits solid-solution signatures at low voltage. A higher Li capacity is obtained than in the DRX phase. This study provides atomic insights into solid-state phase transformation and its relation to experimental electrochemistry, highlighting the potential of machine learning interatomic potentials for understanding complex oxide materials.

Figures

Figures reproduced from arXiv: 2506.20605 by the authors.

Figure 1
Figure 1. FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Flowchart illustrating the development process of fine-tuned CHGNet machine learning interatomic potential (MLIP) [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3 [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6 [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: a and 7b illustrate the formation energy con￾vex hull and the voltage profile of the DRX phase, with the δ-phase plotted as a red line for comparison. As there are few 0-TM sites in the DRX structure, a rela￾tively limited Li capacity from the 0-TM conversion is achiev…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

68 extracted references · 56 canonical work pages

  1. [1]

    E. A. Olivetti, G. Ceder, G. G. Gaustad, and X. Fu, Lithium-Ion Battery Supply Chain Considera- tions: Analysis of Potential Bottlenecks in Critical Met- als, Joule1, 229 (2017)

    work page 2017

  2. [2]

    L. Xie, C. Singh, S. K. Mitter, M. A. Dahleh, and S. S. Oren, Toward carbon-neutral electricity and mobility: Is 12 the grid infrastructure ready?, Joule5, 1908 (2021)

    work page 2021

  3. [3]

    Beyond Li-ion

    Y. Tian, G. Zeng, A. Rutt, T. Shi, H. Kim, J. Wang, J. Koettgen, Y. Sun, B. Ouyang, T. Chen, Z. Lun, Z. Rong, K. Persson, and G. Ceder, Promises and Chal- lenges of Next-Generation “Beyond Li-ion” Batteries for Electric Vehicles and Grid Decarbonization, Chemical Reviews121, 1623 (2021)

    work page 2021

  4. [4]

    R. J. Cl´ ement, Z. Lun, and G. Ceder, Cation-disordered rocksalt transition metal oxides and oxyfluorides for high energy lithium-ion cathodes, Energy & Environmental Science13, 345 (2020)

    work page 2020

  5. [5]

    D.-H. Kwon, J. Lee, N. Artrith, H. Kim, L. Wu, Z. Lun, Y. Tian, Y. Zhu, and G. Ceder, The Impact of Sur- face Structure Transformations on the Performance of Li-Excess Cation-Disordered Rocksalt Cathodes, Cell Re- ports Physical Science1, 100187 (2020)

    work page 2020

  6. [6]

    L. Li, Z. Lun, D. Chen, Y. Yue, W. Tong, G. Chen, G. Ceder, and C. Wang, Fluorination-Enhanced Sur- face Stability of Cation-Disordered Rocksalt Cathodes for Li-Ion Batteries, Advanced Functional Materials31, 2101888 (2021)

    work page 2021

  7. [7]

    K. Zhou, Y. Li, S. Zheng, M. Zhang, C. Zhang, C. Battaglia, H. Liu, K. Wang, P. Yan, J. Liu, and Y. Yang, Tailoring the redox-active transition metal con- tent to enhance cycling stability in cation-disordered rock-salt oxides, Energy Storage Materials43, 275 (2021)

    work page 2021

  8. [8]

    J. Ahn, R. Giovine, V. C. Wu, K. P. Koirala, C. Wang, R. J. Cl´ ement, and G. Chen, Ultrahigh-Capacity Rock- salt Cathodes Enabled by Cycling-Activated Structural Changes, Advanced Energy Materials13, 1 (2023)

    work page 2023

Show all 68 references
  1. [9]

    Z. Cai, B. Ouyang, H.-M. Hau, T. Chen, R. Giovine, K. P. Koirala, L. Li, H. Ji, Y. Ha, Y. Sun, J. Huang, Y. Chen, V. Wu, W. Yang, C. Wang, R. J. Cl´ ement, Z. Lun, and G. Ceder, In situ formed partially disordered phases as earth-abundant Mn-rich cathode materials, Na- ture En...

    2023

  2. [10]

    Urban, J

    A. Urban, J. Lee, and G. Ceder, The configurational space of rocksalt-type oxides for high-capacity lithium battery electrodes, Advanced Energy Materials4, 1 (2014)

    2014

  3. [11]

    J. Reed, G. Ceder, and A. Van Der Ven, Layered-to- Spinel Phase Transition in Li xMnO2, Electrochemical and Solid-State Letters4, A78 (2001)

    2001

  4. [12]

    Y.-I. Jang, F. Chou, B. Huang, D. R. Sadoway, and Y.-M. Chiang, Magnetic characterization of orthorhom- bic LiMnO 2 and electrochemically transformed spinel LixMnO2 (x <1), Journal of Physics and Chemistry of Solids64, 2525–2533 (2003)

    2003

  5. [13]

    M. R. Jo, Y. Kim, J. Yang, M. Jeong, K. Song, Y.-I. Kim, J.-M. Lim, M. Cho, J.-H. Shim, Y.-M. Kim, W.-S. Yoon, and Y.-M. Kang, Triggered reversible phase transforma- tion between layered and spinel structure in manganese- based layered compounds, Nature Communications10, 3385 (2019)

    2019

  6. [14]

    M. D. Radin, J. Vinckeviciute, R. Seshadri, and A. Van der Ven, Manganese oxidation as the origin of the anoma- lous capacity of Mn-containing Li-excess cathode mate- rials, Nature Energy4, 639 (2019)

    2019

  7. [15]

    A. G. Squires and D. O. Scanlon, Understanding the lim- its to short-range order suppression in many-component disordered rock salt lithium-ion cathode materials, Jour- nal of Materials Chemistry A11, 13765 (2023)

    2023

  8. [16]

    Puchala, J

    B. Puchala, J. C. Thomas, A. R. Natarajan, J. G. Goiri, S. S. Behara, J. L. Kaufman, and A. Van der Ven, CASM – A software package for first-principles based study of multicomponent crystalline solids, Computational Mate- rials Science217, 111897 (2023)

    2023

  9. [17]

    Zhong, F

    P. Zhong, F. Xie, L. Barroso-Luque, L. Huang, and G. Ceder, Modeling intercalation chemistry with multi- redox reactions by sparse lattice models in disordered rocksalt cathodes, PRX Energy2, 043005 (2023)

    2023

  10. [18]

    T. Chen, J. Yang, L. Barroso-Luque, and G. Ceder, Re- moving the Two-Phase Transition in Spinel LiMn 2O4 through Cation Disorder, ACS Energy Letters8, 314 (2023)

    2023

  11. [19]

    Batzner, A

    S. Batzner, A. Musaelian, L. Sun, M. Geiger, J. P. Mailoa, M. Kornbluth, N. Molinari, T. E. Smidt, and B. Kozinsky, E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials, Nature Communications13, 2453 (2022)

    2022

  12. [20]

    Chen and S

    C. Chen and S. P. Ong, A universal graph deep learn- ing interatomic potential for the periodic table, Nature Computational Science2, 718 (2022)

    2022

  13. [21]

    using the GGA+U-based MLIP. C. Intercalation voltage profile ofδ-phase To evaluate the effect of partial disorder on electro- chemical performance, we computed the intercalation voltage profiles of the transformedδ-phase structure us- ing Product-CHGNet. Specifically, we focus...

  14. [22]

    B. Deng, P. Zhong, K. Jun, J. Riebesell, K. Han, C. J. Bartel, and G. Ceder, CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling, Nature Machine Intelligence5, 1031 (2023)

    2023

  15. [23]

    Gilmer, S

    J. Gilmer, S. S. Schoenholz, P. F. Riley, O. Vinyals, and G. E. Dahl, Neural message passing for quantum chem- istry, inInternational conference on machine learning (PMLR, 2017) pp. 1263–1272

    2017

  16. [24]

    F. Zhou, T. Maxisch, and G. Ceder, Configurational Elec- tronic Entropy and the Phase Diagram of Mixed-Valence Oxides: The Case of Li xFePO4, Physical Review Letters 97, 155704 (2006)

    2006

  17. [25]

    Riebesell, R

    J. Riebesell, R. E. A. Goodall, P. Benner, Y. Chiang, B. Deng, G. Ceder, M. Asta, A. A. Lee, A. Jain, and K. A. Persson, A framework to evaluate machine learning crystal stability predictions, Nature Machine Intelligence 7, 836–847 (2025)

    2025

  18. [26]

    B. Deng, Y. Choi, P. Zhong, J. Riebesell, S. Anand, Z. Li, K. Jun, K. A. Persson, and G. Ceder, Systematic soften- ing in universal machine learning interatomic potentials, npj Computational Materials11, 9 (2025)

    2025

  19. [27]

    A. D. Kaplan, R. Liu, J. Qi, T. W. Ko, B. Deng, J. Riebe- sell, G. Ceder, K. A. Persson, and S. P. Ong, A foun- dational potential energy surface dataset for materials, arXiv preprint arXiv:2503.04070 (2025)

    2025

  20. [28]

    M. L. H. Chandrappa, J. Qi, C. Chen, S. Banerjee, and S. P. Ong, Thermodynamics and kinetics of the cath- ode–electrolyte interface in all-solid-state li–s batteries, Journal of the American Chemical Society144, 18009 (2022)

    2022

  21. [29]

    A. Jain, S. P. Ong, G. Hautier, W. Chen, W. D. Richards, S. Dacek, S. Cholia, D. Gunter, D. Skinner, G. Ceder, and K. A. Persson, Commentary: The Materials Project: A materials genome approach to accelerating materials innovation, APL Materials1, 011002 (2013)

    2013

  22. [30]

    H. Ji, A. Urban, D. A. Kitchaev, D.-H. Kwon, N. Ar- trith, C. Ophus, W. Huang, Z. Cai, T. Shi, J. C. Kim, H. Kim, and G. Ceder, Hidden structural and chemical order controls lithium transport in cation-disordered ox- ides for rechargeable batteries, Nature Communications 10, 5...

    2019

  23. [31]

    Ouyang, N

    B. Ouyang, N. Artrith, Z. Lun, Z. Jadidi, D. A. Kitchaev, H. Ji, A. Urban, and G. Ceder, Effect of Flu- orination on Lithium Transport and Short-Range Order 13 in Disordered-Rocksalt-Type Lithium-Ion Battery Cath- odes, Advanced Energy Materials10, 1903240 (2020)

    2020

  24. [32]

    Van der Ven and G

    A. Van der Ven and G. Ceder, Lithium diffusion mech- anisms in layered intercalation compounds, Journal of Power Sources97-98, 529 (2001)

    2001

  25. [33]

    Ceder and A

    G. Ceder and A. Van der Ven, Phase diagrams of lithium transition metal oxides: investigations from first princi- ples, Electrochimica Acta45, 131 (1999)

    1999

  26. [34]

    L. Wang, T. Maxisch, and G. Ceder, A first-principles ap- proach to studying the thermal stability of oxide cathode materials, Chemistry of materials19, 543 (2007)

    2007

  27. [35]

    J. Reed, G. Ceder, and A. V. D. Ven, Layered-to- Spinel Phase Transition in Li xMnO2, Electrochemical and Solid-State Letters4, A78 (2001)

    2001

  28. [36]

    F. Zhou, M. Cococcioni, C. A. Marianetti, D. Morgan, and G. Ceder, First-principles prediction of redox poten- tials in transition-metal compounds with LDA+U, Phys- ical Review B70, 235121 (2004), 0406382

    2004

  29. [37]

    Swathilakshmi, R

    S. Swathilakshmi, R. Devi, and G. Sai Gautam, Perfor- mance of the r2SCAN Functional in Transition Metal Ox- ides, Journal of Chemical Theory and Computation19, 4202 (2023)

    2023

  30. [38]

    J. Sun, A. Ruzsinszky, and J. P. Perdew, Strongly Con- strained and Appropriately Normed Semilocal Density Functional, Physical Review Letters115, 036402 (2015)

    2015

  31. [39]

    Zhong, Z

    P. Zhong, Z. Cai, Y. Zhang, R. Giovine, B. Ouyang, G. Zeng, Y. Chen, R. Cl´ ement, Z. Lun, and G. Ceder, In- creasing capacity in disordered rocksalt cathodes by mg doping, Chemistry of Materials32, 10728 (2020)

    2020

  32. [40]

    Holstun, T

    T. Holstun, T. P. Mishra, L. Huang, H. Hau, S. Anand, X. Yang, C. Ophus, K. Bustillo, L. Ma, S. Ehrlich, and G. Ceder, Accelerating the Electrochemical Formation of theδPhase in Manganese-Rich Rocksalt Cathodes, Advanced Materials37, 1 (2025)

    2025

  33. [41]

    H.-M. Hau, T. Mishra, C. Ophus, T.-Y. Huang, K. Bustilo, Y. Sun, X. Yang, T. Holstun, X. Zhao, S. Wang, Y. Ha, G.-H. Lee, C. Song, J. Turner, J. Bai, L. Ma, K. Chen, F. Wang, W. Yang, B. D. McCloskey, Z. Cai, and G. Ceder, Earth-abundant Li-ion cathode materials with nanoengin...

    2024

  34. [42]

    T. Li, T. S. Geraci, K. P. Koirala, A. Zohar, E. N. Bassey, P. A. Chater, C. Wang, A. Navrotsky, and R. J. Cl´ ement, Structural Evolution in Disordered Rock Salt Cathodes, Journal of the American Chemical Society146, 24296 (2024)

    2024

  35. [43]

    P. Xiao, T. Shi, W. Huang, and G. Ceder, Understand- ing surface densified phases in ni-rich layered compounds, ACS Energy Letters4, 811 (2019)

    2019

  36. [44]

    Z. Deng, T. P. Mishra, W. Xie, D. A. Saeed, G. S. Gau- tam, and P. Canepa, kmcpy: A python package to sim- ulate transport properties in solids with kinetic monte carlo, Computational Materials Science229, 112394 (2023)

    2023

  37. [45]

    J. H. Yang, T. Chen, L. Barroso-Luque, Z. Jadidi, and G. Ceder, Approaches for handling high-dimensional cluster expansions of ionic systems, npj Computational Materials8, 133 (2022)

    2022

  38. [46]

    K. Han, B. Deng, A. B. Farimani, and G. Ceder, Distm- lip: A distributed inference platform for machine learning interatomic potentials, arXiv preprint arXiv:2506.02023 (2025)

    2025

  39. [47]

    H. Zhao, H. D. Deng, A. E. Cohen, J. Lim, Y. Li, D. Fraggedakis, B. Jiang, B. D. Storey, W. C. Chueh, R. D. Braatz, and M. Z. Bazant, Learning heterogeneous reaction kinetics from x-ray videos pixel by pixel, Nature 621, 289 (2023)

    2023

  40. [48]

    X. Fu, T. Xie, N. J. Rebello, B. D. Olsen, and T. Jaakkola, Simulate time-integrated coarse-grained molecular dynamics with geometric machine learning, Transactions on Machine Learning Research (TMLR) (2022)

    2022

  41. [49]

    T. Hsu, B. Sadigh, V. Bulatov, and F. Zhou, Score Dy- namics: Scaling Molecular Dynamics with Picoseconds Time Steps via Conditional Diffusion Model, Journal of Chemical Theory and Computation20, 2335 (2024)

    2024

  42. [50]

    J. Nam, S. Liu, G. Winter, K. Jun, S. Yang, and R. G´ omez-Bombarelli, Flow matching for accelerated simulation of atomic transport in materials, arXiv preprint arXiv:2410.01464 (2024)

    2024

  43. [51]

    Y. Xie, J. Vandermause, S. Ramakers, N. H. Pro- tik, A. Johansson, and B. Kozinsky, Uncertainty-aware molecular dynamics from Bayesian active learning for phase transformations and thermal transport in SiC, npj Computational Materials9, 36 (2023)

    2023

  44. [52]

    J. Qi, T. W. Ko, B. C. Wood, T. A. Pham, and S. P. Ong, Robust training of machine learning interatomic potentials with dimensionality reduction and stratified sampling, npj Computational Materials10, 43 (2024)

    2024

  45. [53]

    C. Duan, Y. Du, H. Jia, and H. J. Kulik, Accurate tran- sition state generation with an object-aware equivariant elementary reaction diffusion model, Nature Computa- tional Science3, 1045 (2023)

    2023

  46. [54]

    Q. Zhao, Y. Han, D. Zhang, J. Wang, P. Zhong, T. Cui, B. Yin, Y. Cao, H. Jia, and C. Duan, Harnessing Ma- chine Learning to Enhance Transition State Search with Interatomic Potentials and Generative Models, Advanced Science06240, 1 (2025)

    2025

  47. [55]

    Cheng, Latent Ewald summation for machine learning of long-range interactions, npj Computational Materials 11, 80 (2025)

    B. Cheng, Latent Ewald summation for machine learning of long-range interactions, npj Computational Materials 11, 80 (2025)

    2025

  48. [56]

    Zhong, D

    P. Zhong, D. Kim, D. S. King, and B. Cheng, Ma- chine learning interatomic potential can infer electrical response, arXiv preprint arXiv:2504.05169 (2025)

    2025 arXiv

  49. [57]

    D. Kim, X. Wang, P. Zhong, D. S. King, T. J. Inizan, and B. Cheng, A universal augmentation framework for long-range electrostatics in machine learning interatomic potentials, arXiv preprint arXiv:2507.14302 (2025)

    2025 arXiv

  50. [58]

    Kresse and J

    G. Kresse and J. Furthm¨ uller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set, Computational Materials Science 6, 15 (1996)

    1996

  51. [59]

    Kresse and D

    G. Kresse and D. Joubert, From ultrasoft pseudopoten- tials to the projector augmented-wave method, Physical Review B59, 1758 (1999)

    1999

  52. [60]

    J. W. Furness, A. D. Kaplan, J. Ning, J. P. Perdew, and J. Sun, Accurate and Numerically Efficient r 2SCAN Meta-Generalized Gradient Approximation, The Journal of Physical Chemistry Letters11, 8208 (2020)

    2020

  53. [61]

    Kingsbury, A

    R. Kingsbury, A. S. Gupta, C. J. Bartel, J. M. Munro, S. Dwaraknath, M. Horton, and K. A. Persson, Perfor- mance comparison of r2SCAN and SCAN metaGGA den- sity functionals for solid materials via an automated, high-throughput computational workflow, Physical Re- view Material...

    2022

  54. [62]

    Sanchez, F

    J. Sanchez, F. Ducastelle, and D. Gratias, Generalized cluster description of multicomponent systems, Physica A: Statistical Mechanics and its Applications128, 334 14 (1984)

    1984

  55. [63]

    Wolverton and A

    C. Wolverton and A. Zunger, First-principles prediction of vacancy order-disorder and intercalation battery volt- ages in LixCoO2, Physical Review Letters81, 606 (1998)

    1998

  56. [64]

    van de Walle, Multicomponent multisublattice alloys, nonconfigurational entropy and other additions to the Alloy Theoretic Automated Toolkit, Calphad33, 266 (2009)

    A. van de Walle, Multicomponent multisublattice alloys, nonconfigurational entropy and other additions to the Alloy Theoretic Automated Toolkit, Calphad33, 266 (2009)

    2009

  57. [65]

    Zhong, T

    P. Zhong, T. Chen, L. Barroso-Luque, F. Xie, and G. Ceder, Anℓ 0ℓ2-norm regularized regression model for construction of robust cluster expansions in multicompo- nent systems, Physical Review B106, 024203 (2022)

    2022

  58. [66]

    Barroso-Luque, P

    L. Barroso-Luque, P. Zhong, J. H. Yang, F. Xie, T. Chen, B. Ouyang, and G. Ceder, Cluster expansions of multi- component ionic materials: Formalism and methodology, Physical Review B106, 144202 (2022)

    2022

  59. [67]

    Barroso-Luque, J

    L. Barroso-Luque, J. H. Yang, F. Xie, T. Chen, R. L. Kam, Z. Jadidi, P. Zhong, and G. Ceder, smol: A Python package for cluster expansions and beyond, Journal of Open Source Software7, 4504 (2022)

    2022

  60. [68]

    S. P. Ong, W. D. Richards, A. Jain, G. Hautier, M. Kocher, S. Cholia, D. Gunter, V. L. Chevrier, K. A. Persson, and G. Ceder, Python Materials Genomics (py- matgen): A robust, open-source python library for mate- rials analysis, Computational Materials Science68, 314 (2013)

    2013

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.