Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Exploring the energy landscape of aluminas through machine learning interatomic potential

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A universal machine-learned interatomic potential for aluminas, trained on 3,335 DFT-labelled structures, reproduces known polymorphs and predicts high-pressure phases absent from the training set.

desk verdict A credible and unusually comprehensive alumina NEP with open data and code; the high-pressure extrapolation and 4000 K claim are oversold and need revision. read the letter →

arxiv 2412.02191 v2 pith:37BQOG5O submitted 2024-12-03 cond-mat.mtrl-sci

Lei Zhang , Wenhao Luo , Renxi Liu , Mohan Chen , Zhongbo Yan , Kun Cao This is my paper · ORCID
classification cond-mat.mtrl-sci
keywords machinelearninginteratomicpotentialneuroevolutionaluminapolymorphstransitionalaluminasphasediagramactiveγ-Al2O3structurehigh-pressurephases
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

Alumina appears in many structures, but the atomic arrangements of its transitional forms remain contested. The paper proposes a single machine-learned interatomic potential, built with the neuroevolution potential (NEP) approach on a dataset of 3,335 DFT-labelled structures selected by active learning, that is accurate across crystalline, amorphous, liquid, and non-stoichiometric aluminas. If the central claim holds, this potential makes near-DFT-quality simulation of aluminas feasible on large systems and long timescales, including the defective transitional phases that experiments cannot fully resolve. It also allows the paper to construct phase diagrams and to evaluate competing structural models of γ-Al2O3, which is the payoff a sympathetic reader would care about.

What carries the argument

The load-bearing object is the neuroevolution potential (NEP), a machine-learned interatomic potential in which the total energy is a sum of atom-centred site energies depending on radial and angular descriptors of the local environment, here built from Chebyshev polynomials and the atomic cluster expansion. Its transferability is manufactured by an active-learning loop that uses farthest point sampling in descriptor space to select the most representative new configurations from molecular dynamics runs, labels them with DFT, and iterates until the potential stops improving. Free-energy phase boundaries are then computed with nonequilibrium thermodynamic integration, and the γ-Al2O3 model assessment is carried by a differential-evolution structure search that varies cation occupancies over the allowed Wyckoff sites and uses NEP energies as the stability criterion.

What would settle it

A direct test would be to add Rh2O3(II)- and CaIrO3-type configurations to the published training set, retrain, and check whether the extrapolated phase boundaries in Fig. 7(b) move outside the stated uncertainties; if they do, the claim of comprehensive structural-space coverage fails. A shorter check is to compute the free-energy difference between Rh2O3(II)- and CaIrO3-type Al2O3 at high temperature with an independent DFT-based method, since the paper itself reports decreasing accuracy above about 2000 K for that boundary.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that the NEP captures the energy landscape of alumina well enough to be a universal potential: energy, force, and stress root-mean-square errors of 18.11 meV/atom, 250.88 meV/Å, and 81.01 MPa on the training set, with agreement to DFT equations of state, elastic constants, phonons (except LO-TO splitting from the lack of non-analytical corrections), thermal expansion, thermal conductivity, and the radial distribution functions of liquid and amorphous alumina. In the phase-diagram part, the NEP yields anharmonic-free-energy phase boundaries among α-, θ-, δ-, κ-, and γ-Al2O3 and, despite containing no high-pressure phases, correctly predicts negative Clapeyron slopes for the α→Rh2O3(II) and Rh2O3(II)→CaIrO3 transitions, with the former boundary agreeing with experiments near 90 GPa and the latter becoming less accurate above 2000 K. In the structure-search part, the potential favors a cation distribution with a roughly 97:3 ratio of spinel to non-spinel sites under the Smrčok model, slightly above the 94:6 ratio in the original model, and disfavors the nearest-neighbour cation pairs present in the Luo model.

Load-bearing premise

The load-bearing premise is that the training set, which contains no high-pressure Rh2O3(II)- or CaIrO3-type alumina, nevertheless covers the local atomic environments needed for the potential to make trustworthy predictions about those phases; if a high-pressure coordination environment was under-represented, the extrapolated phase boundaries would be unreliable rather than revealing.

Editorial extensions

If this is right

  • Alumina simulations that previously required empirical potentials can now be done at near-DFT accuracy over nanosecond timescales and tens of thousands of atoms, covering crystalline, amorphous, liquid, and non-stoichiometric AlxOy.
  • The NEP phase diagram gives concrete, anharmonic free-energy-based phase boundaries among α-, θ-, δ-, and κ-Al2O3, including near-parallel δ/θ boundaries that explain observed phase coexistence.
  • The moderate extrapolation to 4000 K and 200 GPa predicts α→Rh2O3(II) and Rh2O3(II)→CaIrO3 boundaries with correct negative Clapeyron slopes, supporting use of the potential outside the training regime.
  • The differential-evolution workflow supplies a NEP-based criterion for judging γ-Al2O3 models: spinel-site occupancy ratio of about 97:3 is energetically favored under the Smrčok model, and low-energy structures avoid nearest-neighbor cation pairs.
  • The potential, dataset, and structure search code are released, enabling other groups to extend coverage to χ- and η-alumina and to refine high-pressure structures.

Reading between the lines

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

  • The success of extrapolation here implies a general principle for building machine-learned potentials: explicit coverage of local atomic environments, rather than the list of phases in the training set, determines transferability to unseen structures.
  • Because the NEP lacks non-analytical corrections for long-range Coulomb interactions, its vibrational and thermal properties are most trustworthy where polarization effects are not decisive; the paper's thermal-conductivity agreement suggests this limitation is minor for α-Al2O3, but an extension with explicit charge terms could close the gap in ionic phonon dispersions.
  • If the 97:3 spinel-to-non-spinel ratio is robust, it provides an energetically anchored prediction for γ-Al2O3 that could be tested by revisiting electron-diffraction refinements with the NEP-optimized occupancy distribution, since experimental samples may retain local constraints that prevent full relaxation.
  • The same active-learning-plus-search workflow could be applied to other vacancy-structured oxides, such as spinel ferrites or defective catalytically active oxides, where cation-distribution combinatorics block direct DFT enumeration.
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 / 5 minor

Summary. The paper presents a neuroevolution potential (NEP) for alumina trained on a dataset of 3,335 DFT(PBEsol) configurations covering crystalline polymorphs (α, θ, κ, δ, γ models), amorphous/liquid structures, non-stoichiometric AlxOy, and clusters. Validation includes EOS, elastic constants, phonon dispersions (without LO-TO splitting), thermal conductivity, thermal expansion, melting point, liquid RDF, amorphous quenching, and phase diagrams computed using nonequilibrium thermodynamic integration (NETI). The authors claim the potential enables accurate simulations of various aluminas at larger scales and longer timescales, and that it successfully extrapolates the high-pressure phase diagram of α-Al2O3 to Rh2O3(II)- and CaIrO3-type phases not present in the training set. They also introduce a differential-evolution structure search workflow and apply it to evaluate the Smrčok and Luo models of γ-Al2O3.

Significance. If fully supported, the NEP would be a valuable open resource for the alumina community, and the combination of active learning, NETI-based free-energy calculations, and a structure search workflow is a useful methodological template. The paper reports a broad and mostly careful validation effort (EOS, elastic constants, thermal transport, disordered phases), and the open data and code are clear strengths that increase reproducibility. The main significance hinges on the high-pressure extrapolation claim; the current evidence for that claim is partial, so the paper's impact would be strengthened by direct tests on the high-pressure phases and by aligning the temperature range of the advertised claims with the validated range.

major comments (3)
  1. [§III.C, Fig. 7(b), and abstract] The claim of successful extrapolation over [0, 4000] K is not supported by the evidence presented. The paper itself states that for the Rh2O3(II)-to-CaIrO3 boundary the accuracy 'diminishes above 2000 K,' and at 1000 K the computed Clapeyron slope (-4.04 MPa/K) differs from the reference value (-9.4 MPa/K) by more than a factor of two. Since the abstract advertises a phase diagram up to 4000 K, the authors should either restrict the claim to lower temperatures or provide additional validation at state points above 2000 K, for example direct DFT free-energy calculations at selected thermodynamic conditions along the boundary.
  2. [§III.C, last paragraph] The inference that reasonable agreement of one phase boundary with experiment implies that 'the structural space is comprehensively explored' is not a direct test of local-environment coverage. The training set contains no Rh2O3(II)- or CaIrO3-type structures, and a low training RMSE does not certify extrapolation to unseen coordination environments. To support the extrapolation claim, the authors should directly compare NEP and DFT energies, equations of state, or phonons for the high-pressure phases, or provide a descriptor-based distance analysis demonstrating that the active-learning dataset already sampled the relevant local environments.
  3. [§III.B, Fig. 5(a)] The computed melting point of 2530 K is roughly 9% above the experimental value of 2327 K. The proposed explanation (defect-free crystal versus real samples with defects, surfaces, and container interfaces) is plausible but not quantitatively supported. Because the paper claims high accuracy at high temperatures, the authors should either provide convergence checks of the two-phase method (e.g., system size, interface orientation, and heating protocol) or temper the claim that the melting point is captured with high accuracy.
minor comments (5)
  1. [§III.B, first paragraph] There is a typo: 'we first calculat the melting point' should be 'we first calculate the melting point.'
  2. [Fig. 6 caption] The sentence 'The inset, which enlarges the region marked by the red dashed lines, is reflected the relative height of the first peak' is grammatically incorrect; 'is reflected' should be 'reflects.'
  3. [§II.A, dataset description] The variant labels jp50s1, jp50s2, jp8bs1, and jp8bs2 are not defined in the main text; a brief explanation or a pointer to the supplementary material would help readers.
  4. [TABLE II] The reported 2σ ranges for the 16c and 48f Wyckoff positions include negative occupancies (e.g., -0.8–3.3), which are unphysical for site-occupation numbers. If these ranges come from a normal approximation, this should be stated explicitly so that readers do not interpret them as literal allowed occupation counts.
  5. [Reference list] Reference [71] contains a formatting issue ('Physical review B52' should be 'Physical Review B 52'); several other references also lack consistent journal-name styling.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the NEP fit and its downstream property predictions are standard interpolation/extrapolation, with the high-pressure phases absent from the training set.

full rationale

The paper's central derivation is a standard supervised MLIP construction: NEP is regressed to DFT energies, forces, and virials over an actively learned dataset, and subsequent property calculations (EOS, phonons, thermal expansion, phase boundaries, and the gamma-alumina structure search) evaluate that fitted potential. None of these properties is itself an input to the fit, and the highest-risk claim, namely the Rh2O3(II)- and CaIrO3-type high-pressure phase boundaries, is explicitly computed from phases absent from the training set, so it is an extrapolation checked against independent experimental and DFT references rather than a fitted quantity. Agreement of EOS and phonons for polymorphs included in the dataset is interpolation, but the paper presents it as fitting assessment rather than as an independent prediction, and it does not use that agreement to ground the extrapolation claim. No load-bearing self-citation, imported uniqueness theorem, or ansatz-smuggling step was identified; the main weakness is empirical accuracy diminishing above 2000 K, which is an external-validation concern, not circularity.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central results rest on the quality of the DFT labels, the locality assumption of the NEP, and the coverage of the active-learning dataset. No new physical entities are introduced. The main fitted quantity is the neural potential itself, whose weights and hyperparameters are trained on the DFT data.

free parameters (2)
  • NEP network weights and biases = Not enumerated; trained on 3,335 DFT structures
    The potential is a neural network; its weights are fitted to DFT energies, forces, and virials, so the central results depend on this fit rather than first-principles derivation.
  • NEP hyperparameters (cutoff radius, descriptor size, hidden neurons, regularization coefficients) = Details in supplementary materials
    Chosen by the authors and not stated in the main text; they affect accuracy and transferability of the potential.
assumptions (5)
  • domain assumption PBEsol DFT is sufficiently accurate for alumina energetics and forces.
    All labels are generated with PBEsol; inaccuracies in the functional propagate through training (Section II B).
  • domain assumption Energy locality with a finite cutoff can represent the alumina potential energy surface.
    NEP relies on the Behler-Parrinello locality hypothesis; long-range Coulomb effects are not explicitly included, as acknowledged in the phonon LO-TO discussion (Section III A).
  • domain assumption The iterative active-learning dataset with farthest-point sampling covers the relevant configuration space, including configurations needed for high-pressure extrapolation.
    The high-pressure claim depends on this coverage, since Rh2O3(II) and CaIrO3 structures were not in the training set (Section III C).
  • domain assumption NETI free-energy calculations converge and give accurate phase boundaries for these solids.
    Phase diagram results rely on the nonequilibrium thermodynamic integration method and on the potential being accurate enough for free energies (Section III C).
  • domain assumption Literature structural models for alpha, theta, kappa, delta, and gamma aluminas are valid representatives of the real phases.
    The dataset is built from mined literature structures; errors in those models would bias training and phase comparisons (Section II A).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Exploring the energy landscape of aluminas through machine learning interatomic potential." pith.science (2026). https://pith.science/paper/37BQOG5O

@misc{pith2026241202191,
  author       = {Pith},
  title        = {Pith review of: Exploring the energy landscape of aluminas through machine learning interatomic potential},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/37BQOG5O}},
  note         = {Machine review of arXiv:2412.02191}
}
abstract

Aluminum oxide (alumina, Al$_2$O$_3$) exists in various structures and has broad industrial applications. While the crystal structure of $\alpha$-Al$_2$O$_3$ is well-established, those of transitional aluminas remain highly debated. In this study, we propose a universal machine learning interatomic potential (MLIP) for aluminas, trained using the neuroevolution potential (NEP) approach. The dataset is constructed through iterative training and farthest point sampling, ensuring the generation of the most representative configurations for an exhaustive sampling of the potential energy surface. The accuracy and generality of the potential are validated through simulations under a wide range of conditions, including high temperatures and pressures. A phase diagram is presented that includes both transitional aluminas and $\alpha$-Al$_2$O$_3$ based on the NEP. We also successfully extrapolate the phase diagram of aluminas under extreme conditions ([0, 4000] K and [0, 200] GPa ranges of temperature and pressure, respectively), while maintaining high accuracy in describing their properties under more moderate conditions. Furthermore, combined with our developed structure search workflow, the NEP provides an evaluation of existing $\gamma$-Al$_2$O$_3$ structure models. The NEP developed in this work enables highly accurate dynamic simulations of various aluminas on larger scales and longer timescales, while also offering new insights into the study of transitional aluminas structures.

Figures

Figures reproduced from arXiv: 2412.02191 by the authors.

Figure 1
Figure 1. FIG. 1. Schematic representation of the conventional cell of [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. A flowchart of dataset construction and potential fitting, along with the composition of the final dataset. The [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. (a) Evolution of the various terms in the loss function [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: FIG. 4. (a) The equation of state (EOS) curves for various alumina polymorphs were obtained by fitting the Birch-Murnaghan [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. (a) The melting point of [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. The influence of quenching rates and density on RDF [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7. Phase diagram of aluminas calculated by NEP. (a) The phase diagram of [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Advances in modeling complex materials: The rise of neuroevolution potentials

    cond-mat.mtrl-sci 2025-01 conditional novelty 4.0 of 10

    Neuroevolution potentials, trained with an evolutionary strategy and running on GPUs, match or approach quantum-accurate energies and forces while simulating systems with millions of atoms, at speeds far beyond compet...

Reference graph

Works this paper leans on

110 extracted references · 73 canonical work pages · cited by 1 Pith paper

  1. [1]

    Levin and D

    I. Levin and D. Brandon, Metastable alumina poly- morphs: Crystal structures and transition sequences, Journal of the American Ceramic Society 81, 1995 11 (1998)

    1998

  2. [2]

    Rouquerol, K

    J. Rouquerol, K. S. Sing, and P. Llewellyn, 11 - ad- sorption by metal oxides, in Adsorption by Powders and Porous Solids (Second Edition), edited by F. Rouquerol, J. Rouquerol, K. Sing, P. Llewellyn, and G. Maurin (Academic Press, Oxford, 2014) second edition ed., pp. 393–465

    2014

  3. [3]

    S. D. Jackson and J. S. Hargreaves, Metal oxide cataly- sis, 2 volume set, Vol. 1 (John Wiley & Sons, 2008)

    2008

  4. [4]

    L. K. Hudson, C. Misra, A. J. Perrotta, K. Wefers, and F. Williams, Aluminum oxide, Ullmann’s encyclopedia of industrial chemistry (2000)

    2000

  5. [5]

    Abou-Ziyan, D

    H. Abou-Ziyan, D. Abd El-Raheim, O. Mahmoud, and M. Fatouh, Performance characteristics of thin- multilayer activated alumina bed, Applied Energy 190, 29 (2017)

    2017

  6. [6]

    Halvarsson, V

    M. Halvarsson, V. Langer, and S. Vuorinen, X-ray pow- der diffraction data for κ-al2o3, Powder Diffraction 14, 61 (1999)

    1999

  7. [7]

    J. Kohn, G. Katz, and J. Broder, Characterization of β-ga2o3 and its alumina isomorph, θ-al2o3, American Mineralogist: Journal of Earth and Planetary Materials 42, 398 (1957)

    1957

  8. [9]

    H. C. Stumpf, A. S. Russell, J. Newsome, and C. Tucker, Thermal transformations of aluminas and alumina hydrates-reaction with 44% technical acid., Industrial & Engineering Chemistry 42, 1398 (1950)

    1950

Show all 110 references
  1. [10]

    Verwey, Electrolytic conduction of a solid insulator at high fields the formation of the anodic oxide film on aluminium, Physica 2, 1059 (1935)

    E. Verwey, Electrolytic conduction of a solid insulator at high fields the formation of the anodic oxide film on aluminium, Physica 2, 1059 (1935)

    1935

  2. [11]

    Zhou and R

    R.-S. Zhou and R. L. Snyder, Structures and transfor- mation mechanisms of the η, γ and θ transition alumi- nas, Acta Crystallographica Section B: Structural Sci- ence 47, 617 (1991)

    1991

  3. [12]

    Prins, On the structure of γ-al2o3, Journal of Catal- ysis 392, 336 (2020)

    R. Prins, On the structure of γ-al2o3, Journal of Catal- ysis 392, 336 (2020)

    2020

  4. [13]

    H. O. Ayoola, S. D. House, C. S. Bonifacio, K. Kisslinger, W. A. Saidi, and J. C. Yang, Evaluat- ing the accuracy of common γ-al2o3 structure models by selected area electron diffraction from high-quality crystalline γ-al2o3, Acta Materialia 182, 257 (2020)

    2020

  5. [14]

    Kovarik, M

    L. Kovarik, M. Bowden, K. Khivantsev, J. H. Kwak, and J. Szanyi, Structural complexity of γ-al2o3: The nature of vacancy ordering and the structure of com- plex antiphase boundaries, Acta Materialia 266, 119639 (2024)

    2024

  6. [15]

    Kohn and L

    W. Kohn and L. J. Sham, Self-consistent equations in- cluding exchange and correlation effects, Physical re- view 140, A1133 (1965)

    1965

  7. [16]

    Behler and M

    J. Behler and M. Parrinello, Generalized neural-network representation of high-dimensional potential-energy sur- faces, Phys. Rev. Lett. 98, 146401 (2007)

    2007

  8. [17]

    Behler, Atom-centered symmetry functions for con- structing high-dimensional neural network potentials, The Journal of chemical physics 134 (2011)

    J. Behler, Atom-centered symmetry functions for con- structing high-dimensional neural network potentials, The Journal of chemical physics 134 (2011)

    2011

  9. [18]

    Behler, Neural network potential-energy surfaces in chemistry: a tool for large-scale simulations, Physical Chemistry Chemical Physics 13, 17930 (2011)

    J. Behler, Neural network potential-energy surfaces in chemistry: a tool for large-scale simulations, Physical Chemistry Chemical Physics 13, 17930 (2011)

    2011

  10. [19]

    Bonati and M

    L. Bonati and M. Parrinello, Silicon liquid structure and crystal nucleation from ab initio deep metadynamics, Physical review letters 121, 265701 (2018)

    2018

  11. [20]

    A. P. Bart´ ok, J. Kermode, N. Bernstein, and G. Cs´ anyi, Machine learning a general-purpose interatomic poten- tial for silicon, Physical Review X 8, 041048 (2018)

    2018

  12. [21]

    P. Rowe, V. L. Deringer, P. Gasparotto, G. Cs´ anyi, and A. Michaelides, An accurate and transferable machine learning potential for carbon, The Journal of Chemical Physics 153 (2020)

    2020

  13. [22]

    V. L. Deringer, M. A. Caro, and G. Cs´ anyi, A general- purpose machine-learning force field for bulk and nanos- tructured phosphorus, Nature communications 11, 5461 (2020)

    2020

  14. [23]

    Chen and S

    C. Chen and S. P. Ong, A universal graph deep learn- ing interatomic potential for the periodic table, Nature Computational Science 2, 718 (2022)

    2022

  15. [24]

    K. Song, R. Zhao, J. Liu, Y. Wang, E. Lindgren, Y. Wang, S. Chen, K. Xu, T. Liang, P. Ying, et al., General-purpose machine-learned potential for 16 elemental metals and their alloys, arXiv preprint arXiv:2311.04732 (2023)

    2023 arXiv

  16. [25]

    Batatia, P

    I. Batatia, P. Benner, Y. Chiang, A. M. Elena, D. P. Kov´ acs, J. Riebesell, X. R. Advincula, M. Asta, W. J. Baldwin, N. Bernstein, et al., A foundation model for atomistic materials chemistry, arXiv preprint arXiv:2401.00096 (2023)

    2023 arXiv

  17. [26]

    Cheng, G

    B. Cheng, G. Mazzola, C. J. Pickard, and M. Ceriotti, Evidence for supercritical behaviour of high-pressure liquid hydrogen, Nature 585, 217 (2020)

    2020

  18. [27]

    Wengert, G

    S. Wengert, G. Cs´ anyi, K. Reuter, and J. T. Margraf, Data-efficient machine learning for molecular crystal structure prediction, Chemical science 12, 4536 (2021)

    2021

  19. [28]

    H. Dong, Y. Shi, P. Ying, K. Xu, T. Liang, Y. Wang, Z. Zeng, X. Wu, W. Zhou, S. Xiong, et al., Molecular dynamics simulations of heat transport using machine- learned potentials: A mini-review and tutorial on gpumd with neuroevolution potentials, Journal of Ap- plied Physics 1...

    2024

  20. [29]

    Richard, A

    P. Richard, A. Castellano, R. B´ ejaud, L. Baguet, J. Bouchet, G. Geneste, and F. Bottin, Ab initio phase diagram of gold in extreme conditions, Physical Review Letters 131, 206101 (2023)

    2023

  21. [30]

    Artrith, T

    N. Artrith, T. Morawietz, and J. Behler, High- dimensional neural-network potentials for multicompo- nent systems: Applications to zinc oxide, Physical Re- view B—Condensed Matter and Materials Physics 83, 153101 (2011)

    2011

  22. [31]

    M. F. Calegari Andrade and A. Selloni, Structure of disordered tio 2 phases from ab initio based deep neu- ral network simulations, Physical Review Materials 4, 113803 (2020)

    2020

  23. [32]

    Timmermann, F

    J. Timmermann, F. Kraushofer, N. Resch, P. Li, Y. Wang, Z. Mao, M. Riva, Y. Lee, C. Staacke, M. Schmid, et al., Iro 2 surface complexions identified through machine learning and surface investigations, Physical review letters 125, 206101 (2020)

    2020

  24. [33]

    J. Zhao, J. Byggm¨ astar, H. He, K. Nordlund, F. Djurabekova, and M. Hua, Complex ga 2o3 poly- morphs explored by accurate and general-purpose machine-learning interatomic potentials, npj Computa- tional Materials 9, 159 (2023)

    2023

  25. [34]

    Sivaraman, L

    G. Sivaraman, L. Gallington, A. N. Krishnamoorthy, M. Stan, G. Cs´ anyi,´A. V´ azquez-Mayagoitia, and C. J. Benmore, Experimentally driven automated machine- learned interatomic potential for a refractory oxide, 12 Physical Review Letters 126, 156002 (2021)

    2021

  26. [35]

    L. Ma, J. Wu, T. Zhu, Y. Huang, Q. Lu, and S. Liu, Ultrahigh oxygen ion mobility in ferroelectric hafnia, Physical Review Letters 131, 256801 (2023)

    2023

  27. [36]

    L. C. Erhard, J. Rohrer, K. Albe, and V. L. Deringer, A machine-learned interatomic potential for silica and its relation to empirical models, npj Computational Mate- rials 8, 90 (2022)

    2022

  28. [37]

    L. C. Erhard, J. Rohrer, K. Albe, and V. L. Deringer, Modelling atomic and nanoscale structure in the silicon– oxygen system through active machine learning, Nature Communications 15, 1927 (2024)

    2024

  29. [38]

    Tiwari and T

    J. Tiwari and T. Feng, Accurate prediction of thermal conductivity of al 2o3 at ultrahigh temperatures, Physi- cal Review B 109, 075201 (2024)

    2024

  30. [39]

    Rodr ´ ıguez-Mart ´ ınez, T

    C. Rodr ´ ıguez-Mart ´ ınez, T. Schwedek, E. Salazar, and X. Bokhimi, Utilizing wyckoff sites to construct machine-learning-driven interatomic potentials for crys- talline materials: A case study on α-alumina, The Jour- nal of Physical Chemistry C 128, 1746 (2024)

    2024

  31. [40]

    W. Li, Y. Ando, and S. Watanabe, Effects of density and composition on the properties of amorphous alumina: A high-dimensional neural network potential study, The Journal of Chemical Physics 153 (2020)

    2020

  32. [41]

    T. Du, H. Liu, L. Tang, S. S. Sørensen, M. Bauchy, and M. M. Smedskjaer, Predicting fracture propensity in amorphous alumina from its static structure using machine learning, ACS nano 15, 17705 (2021)

    2021

  33. [42]

    Z. Fan, Z. Zeng, C. Zhang, Y. Wang, K. Song, H. Dong, Y. Chen, and T. Ala-Nissila, Neuroevolution machine learning potentials: Combining high accuracy and low cost in atomistic simulations and application to heat transport, Physical Review B 104, 104309 (2021)

    2021

  34. [43]

    Fan, Improving the accuracy of the neuroevolution machine learning potential for multi-component sys- tems, Journal of Physics: Condensed Matter 34, 125902 (2022)

    Z. Fan, Improving the accuracy of the neuroevolution machine learning potential for multi-component sys- tems, Journal of Physics: Condensed Matter 34, 125902 (2022)

    2022

  35. [44]

    Z. Fan, Y. Wang, P. Ying, K. Song, J. Wang, Y. Wang, Z. Zeng, K. Xu, E. Lindgren, J. M. Rahm, et al., Gpumd: A package for constructing accurate machine- learned potentials and performing highly efficient atom- istic simulations, The Journal of Chemical Physics 157 (2022)

    2022

  36. [45]

    de Koning, A

    M. de Koning, A. Antonelli, and S. Yip, Single- simulation determination of phase boundaries: A dynamic clausius–clapeyron integration method, The Journal of Chemical Physics 115, 11025 (2001)

    2001

  37. [46]

    Freitas, M

    R. Freitas, M. Asta, and M. De Koning, Nonequilibrium free-energy calculation of solids using lammps, Compu- tational Materials Science 112, 333 (2016)

    2016

  38. [47]

    Cajahuaringa and A

    S. Cajahuaringa and A. Antonelli, Non-equilibrium free- energy calculation of phase-boundaries using lammps, Computational Materials Science 207, 111275 (2022)

    2022

  39. [48]

    Smrˇ cok, V

    L. Smrˇ cok, V. Langer, and J. Kˇ rest’an,γ-alumina: a single-crystal x-ray diffraction study, Acta Crystallo- graphica Section C: Crystal Structure Communications 62, i83 (2006)

    2006

  40. [49]

    Z. Luo, Structure of boehmite-derived γ-alumina and its transformation mechanism revealed by electron crys- tallography, Acta Crystallographica Section B: Struc- tural Science, Crystal Engineering and Materials 77, 772 (2021)

    2021

  41. [50]

    Liu and J

    P. Liu and J. Skogsmo, Space-group determination and structure model for κ-al2o3 by convergent-beam elec- tron diffraction (cbed), Acta Crystallographica Section B: Structural Science 47, 425 (1991)

    1991

  42. [51]

    Kovarik, M

    L. Kovarik, M. Bowden, A. Genc, J. Szanyi, C. H. Peden, and J. H. Kwak, Structure of δ-alumina: to- ward the atomic level understanding of transition alu- mina phases, The Journal of Physical Chemistry C 118, 18051 (2014)

    2014

  43. [52]

    Kovarik, M

    L. Kovarik, M. Bowden, D. Shi, J. Szanyi, and C. H. Peden, Structural intergrowth in δ-al2o3, The Journal of Physical Chemistry C 123, 9454 (2019)

    2019

  44. [53]

    C. V. Chandran, C. E. Kirschhock, S. Radhakrishnan, F. Taulelle, J. A. Martens, and E. Breynaert, Alumina: discriminative analysis using 3d correlation of solid- state nmr parameters, Chemical Society Reviews 48, 134 (2019)

    2019

  45. [54]

    Lee, C.-F

    M.-H. Lee, C.-F. Cheng, V. Heine, and J. Klinowski, Distribution of tetrahedral and octahedral a1 sites in gamma alumina, Chemical Physics Letters 265, 673 (1997)

    1997

  46. [55]

    Ealet, M

    B. Ealet, M. Elyakhloufi, E. Gillet, and M. Ricci, Elec- tronic and crystallographic structure of γ-alumina thin films, Thin solid films 250, 92 (1994)

    1994

  47. [56]

    H. P. Pinto, R. M. Nieminen, and S. D. Elliott, Ab initio study of γ- al 2 o 3 surfaces, Physical review B 70, 125402 (2004)

    2004

  48. [57]

    Digne, P

    M. Digne, P. Sautet, P. Raybaud, P. Euzen, and H. Toulhoat, Use of dft to achieve a rational under- standing of acid–basic properties of γ-alumina surfaces, Journal of Catalysis 226, 54 (2004)

    2004

  49. [58]

    Paglia, C

    G. Paglia, C. E. Buckley, A. L. Rohl, B. A. Hunter, R. D. Hart, J. V. Hanna, and L. T. Byrne, Tetragonal structure model for boehmite-derived γ-alumina, Phys. Rev. B 68, 144110 (2003)

    2003

  50. [59]

    F. D. Cortes-Vega, W. Yang, J. Zarate-Medina, S. R. Brankovic, J. M. Herrera Ram ´ ırez, and F. C. Rob- les Hernandez, Room-temperature synthesis of χ-al2o3 and ruby (α-cr:al2o3), CrystEngComm 20, 3505 (2018)

    2018

  51. [60]

    Busca, Structural, surface, and catalytic properties of aluminas, in Advances in catalysis, Vol

    G. Busca, Structural, surface, and catalytic properties of aluminas, in Advances in catalysis, Vol. 57 (Elsevier,

  52. [61]

    C. Y. Ouyang, i. c. v. ˇSljivanˇ canin, and A. Baldereschi, First-principles study of γ-al2o3 (100) surface, Phys. Rev. B 79, 235410 (2009)

    2009

  53. [62]

    Paglia, E

    G. Paglia, E. S. Boˇ zin, and S. J. Billinge, Fine-scale nanostructure in γ-al2o3, Chemistry of Materials 18, 3242 (2006)

    2006

  54. [63]

    Drautz, Atomic cluster expansion for accurate and transferable interatomic potentials, Phys

    R. Drautz, Atomic cluster expansion for accurate and transferable interatomic potentials, Phys. Rev. B 99, 014104 (2019)

    2019

  55. [64]

    S. De, A. P. Bart´ ok, G. Cs´ anyi, and M. Ceriotti, Com- paring molecules and solids across structural and al- chemical space, Physical Chemistry Chemical Physics 18, 13754 (2016)

    2016

  56. [65]

    A. P. Bart´ ok, S. De, C. Poelking, N. Bernstein, J. R. Kermode, G. Cs´ anyi, and M. Ceriotti, Machine learning unifies the modeling of materials and molecules, Science advances 3, e1701816 (2017)

    2017

  57. [66]

    Kresse and J

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

    1996

  58. [67]

    Kresse and J

    G. Kresse and J. Furthm¨ uller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Physical Review B 54, 11169 (1996). 13

    1996

  59. [68]

    P. E. Bl¨ ochl, Projector augmented-wave method, Phys- ical Review B 50, 17953 (1994)

    1994

  60. [69]

    Kresse and D

    G. Kresse and D. Joubert, From ultrasoft pseudopoten- tials to the projector augmented-wave method, Physical Review B 59, 1758 (1999)

    1999

  61. [70]

    J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vy- drov, G. E. Scuseria, L. A. Constantin, X. Zhou, and K. Burke, Restoring the density-gradient expansion for exchange in solids and surfaces, Physical Review Letters 100, 136406 (2008)

    2008

  62. [71]

    D. G. Cahill, S.-M. Lee, and T. I. Selinder, Thermal conductivity of κ-al 2 o 3 and α-al 2 o 3 wear-resistant coatings, Journal of applied physics 83, 5783 (1998)

    1998

  63. [72]

    Powell, C

    R. Powell, C. Y. Ho, and P. E. Liley, Thermal conduc- tivity of selected materials, Vol. 8 (US Department of Commerce, National Bureau of Standards Washington, DC, 1966)

    1966

  64. [73]

    Togo and I

    A. Togo and I. Tanaka, First principles phonon calcu- lations in materials science, Scripta Materialia 108, 1 (2015)

    2015

  65. [74]

    R. M. Pick, M. H. Cohen, and R. M. Martin, Micro- scopic theory of force constants in the adiabatic approx- imation, Physical Review B 1, 910 (1970)

    1970

  66. [75]

    Z. Fan, H. Dong, A. Harju, and T. Ala-Nissila, Homo- geneous nonequilibrium molecular dynamics method for heat transport and spectral decomposition with many- body potentials, Physical Review B 99, 064308 (2019)

    2019

  67. [76]

    Y. S. Touloukian, Thermal expansion, Nonmetallic solids, Thermophysical properties of matter 13, 244 (1977)

    1977

  68. [77]

    Schneider, Cooperative determination of the melting point of alumina, Pure and Applied Chemistry 21, 115 (1970)

    S. Schneider, Cooperative determination of the melting point of alumina, Pure and Applied Chemistry 21, 115 (1970)

    1970

  69. [78]

    Ansell, S

    S. Ansell, S. Krishnan, J. R. Weber, J. J. Felten, P. C. Nordine, M. A. Beno, D. L. Price, and M.-L. Saboungi, Structure of liquid aluminum oxide, Physical Review Letters 78, 464 (1997)

    1997

  70. [79]

    Vashishta, R

    P. Vashishta, R. K. Kalia, A. Nakano, and J. P. Rino, Interaction potentials for alumina and molecular dy- namics simulations of amorphous and liquid alumina, Journal of Applied Physics 103 (2008)

    2008

  71. [80]

    Vashishta, R

    P. Vashishta, R. K. Kalia, A. Nakano, and J. P. Rino, Erratum:” molecular dynamics simulation stud- ies of amorphous and liquid alumina”(j. appl. phys. 103, 083504 (2008), Journal of Applied Physics 105, 59901 (2009)

    2008

  72. [81]

    L. J. Alvarez, L. E. Leon, J. F. Sanz, M. J. Capitan, and J. A. Odriozola, Surface structure of cubic aluminum oxide, Physical Review B 50, 2561 (1994)

    1994

  73. [82]

    L. J. Alvarez, L. E. Leon, J. F. Sanz, M. J. Capitan, and J. A. Odriozola, Computer simulation of. gamma.- al2o3 microcrystal, The Journal of Physical Chemistry 99, 17872 (1995)

    1995

  74. [83]

    Matsui, A transferable interatomic potential model for crystals and melts in the system cao-mgo-al2o3-sio2, Mineralogical Magazine 58, 571 (1994)

    M. Matsui, A transferable interatomic potential model for crystals and melts in the system cao-mgo-al2o3-sio2, Mineralogical Magazine 58, 571 (1994)

    1994

  75. [84]

    F. H. Streitz and J. W. Mintmire, Electrostatic poten- tials for metal-oxide surfaces and interfaces, Phys. Rev. B 50, 11996 (1994)

    1994

  76. [85]

    D. R. Neuville, D. de Ligny, L. Cormier, G. S. Hen- derson, J. Roux, A.-M. Flank, and P. Lagarde, The crystal and melt structure of spinel and alumina at high temperature: An in-situ xanes study at the al and mg k-edge, Geochimica et Cosmochimica Acta 73, 3410 (2009)

    2009

  77. [86]

    L. B. Skinner, A. C. Barnes, P. S. Salmon, L. Hennet, H. E. Fischer, C. J. Benmore, S. Kohara, J. R. Weber, A. Bytchkov, M. C. Wilding, et al., Joint diffraction and modeling approach to the structure of liquid alumina, Physical Review B—Condensed Matter and Materials Physics ...

    2013

  78. [87]

    C. Levi, V. Jayaram, J. Valencia, and R. Mehrabian, Phase selection in electrohydrodynamic atomization of alumina, Journal of Materials Research 3, 969 (1988)

    1988

  79. [88]

    J. R. Weber, C. D. Anderson, D. R. Merkley, and P. C. Nordine, Solidification behavior of undercooled liquid aluminum oxide, Journal of the American Ceramic So- ciety 78, 577 (1995)

    1995

  80. [89]

    S.-M. Lee, D. G. Cahill, and T. H. Allen, Thermal con- ductivity of sputtered oxide films, Physical review B52, 253 (1995)

    1995

  81. [90]

    C. Shi, O. L. Alderman, D. Berman, J. Du, J. Neue- feind, A. Tamalonis, J. R. Weber, J. You, and C. J. Benmore, The structure of amorphous and deeply su- percooled liquid alumina, Frontiers in Materials 6, 38 (2019)

    2019

  82. [91]

    critical size

    R. Bagwell, G. Messing, and P. Howell, The formation of α-al2o3 from θ-al2o3: The relevance of a “critical size” and: Diffusional nucleation or “synchro-shear”?, Journal of Materials Science 36, 1833 (2001)

    2001

  83. [92]

    Levin, L

    I. Levin, L. Bendersky, D. Brandon, and M. R¨ uhle, Cubic to monoclinic phase transformations in alumina, Acta materialia 45, 3659 (1997)

    1997

  84. [93]

    S.-H. Cai, S. N. Rashkeev, S. T. Pantelides, and K. Sohlberg, Phase transformation mechanism between γ-and θ-alumina, Physical review B 67, 224104 (2003)

    2003

  85. [94]

    Kachi, K

    S. Kachi, K. Momiyama, and S. Shimizu, An electron diffraction study and a theory of the transformation from γ-fe2o3 to α-fe2o3, Journal of the Physical Society of Japan 18, 106 (1963)

    1963

  86. [95]

    Huang, X

    Y. Huang, X. Peng, and X.-Q. Chen, The mechanism of θ-to α-al2o3 phase transformation, Journal of Alloys and Compounds 863, 158666 (2021)

    2021

  87. [96]

    J.-F. Lin, O. Degtyareva, C. T. Prewitt, P. Dera, N. Sata, E. Gregoryanz, H.-K. Mao, and R. J. Hemley, Crystal structure of a high-pressure/high-temperature phase of alumina by in situ x-ray diffraction, Nature Materials 3, 389 (2004)

    2004

  88. [97]

    Tsuchiya, T

    J. Tsuchiya, T. Tsuchiya, and R. M. Wentzcovitch, Transition from the rh 2 o 3 (ii)-to-ca ir o 3 struc- ture and the high-pressure-temperature phase diagram of alumina, Physical Review B—Condensed Matter and Materials Physics 72, 020103 (2005)

    2005

  89. [98]

    Caracas and R

    R. Caracas and R. Cohen, Prediction of a new phase transition in al2o3 at high pressures, Geophysical re- search letters 32 (2005)

    2005

  90. [99]

    Umemoto and R

    K. Umemoto and R. M. Wentzcovitch, Prediction of an u2s3-type polymorph of al2o3 at 3.7 mbar, Proceedings of the National Academy of Sciences 105, 6526 (2008)

    2008

  91. [100]

    J. Kato, K. Hirose, H. Ozawa, and Y. Ohishi, High- pressure experiments on phase transition boundaries be- tween corundum, rh2o3 (ii)-and cairo3-type structures in al2o3, American Mineralogist 98, 335 (2013)

    2013

  92. [101]

    H. Zhou, Y. Ji, Y. Wang, K. Feng, B. Luan, X. Zhang, and L.-Q. Chen, First-principles lattice dynamics and thermodynamic properties of α-, θ-, κ-and γ-al2o3 and solid state temperature-pressure phase diagram, Acta Materialia 263, 119513 (2024). 14

    2024

  93. [102]

    J. G. Kirkwood, Statistical mechanics of fluid mixtures, The Journal of chemical physics 3, 300 (1935)

    1935

  94. [103]

    Jarzynski, Nonequilibrium equality for free energy differences, Physical Review Letters 78, 2690 (1997)

    C. Jarzynski, Nonequilibrium equality for free energy differences, Physical Review Letters 78, 2690 (1997)

    1997

  95. [104]

    Levin, L

    I. Levin, L. Bendersky, D. Brandon, and M. R¨ uhle, Cubic to monoclinic phase transformations in alumina, Acta Materialia 45, 3659 (1997)

    1997

  96. [105]

    Lippens and J

    B. Lippens and J. De Boer, Study of phase transfor- mations during calcination of aluminum hydroxides by selected area electron diffraction, Acta Crystallograph- ica 17, 1312 (1964)

    1964

  97. [106]

    Jayaram and C

    V. Jayaram and C. Levi, The structure of δ-alumina evolved from the melt and the γ→ δ transformation, Acta metallurgica 37, 569 (1989)

    1989

  98. [107]

    Wilson, The dehydration of boehmite, γ-alooh, to γ- al2o3, Journal of Solid State Chemistry 30, 247 (1979)

    S. Wilson, The dehydration of boehmite, γ-alooh, to γ- al2o3, Journal of Solid State Chemistry 30, 247 (1979)

    1979

  99. [108]

    Funamori and R

    N. Funamori and R. Jeanloz, High-pressure transforma- tion of al2o3, Science 278, 1109 (1997)

    1997

  100. [109]

    Paglia, A

    G. Paglia, A. L. Rohl, C. E. Buckley, and J. D. Gale, Determination of the structure of γ-alumina from inter- atomic potential and first-principles calculations: The requirement of significant numbers of nonspinel posi- tions to achieve an accurate structural model, Phys. Rev....

    2005

  101. [110]

    Storn and K

    R. Storn and K. Price, Differential evolution–a simple and efficient heuristic for global optimization over con- tinuous spaces, Journal of global optimization 11, 341 (1997)

    1997

  102. [111]

    The equivalent spinel sites in a tetragonal lattice with I41amd symmetry are the 4a position (tetrahedrally co- ordinated) and the 8d position (octahedrally coordi- nated) in the Wyckoff notation

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.