pith. sign in

arxiv: 2605.25485 · v1 · pith:MYC6FRMAnew · submitted 2026-05-25 · ❄️ cond-mat.mtrl-sci

Transition from Homogeneous to Domain-Wall-Mediated Polarization Switching in BaTiO3: A Machine-Learning Molecular Dynamics Study

Po-Yen Chen , Teruyasu Mizoguchi This is my paper

Pith reviewed 2026-06-29 21:49 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci
keywords BaTiO3polarization switchingdomain wallsmachine learning potentialsmolecular dynamicsferroelectricscoercive fieldsupercell size
0
0 comments X

The pith

BaTiO3 polarization switching transitions from homogeneous to domain-wall-mediated with increasing supercell size, raising the coercive field by over 50%.

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

The paper shows that the pathway for reversing electric polarization in BaTiO3 depends on the size of the simulated crystal. Small supercells switch uniformly without forming domains. Larger supercells allow local polarization fluctuations that nucleate 180 degree domain walls, shifting the mechanism. This change raises the coercive field needed for switching by more than 50 percent. Shannon entropy quantifies how size-dependent disorder controls the macroscopic hysteresis, and the pathway also varies with cell shape and the angle between stress and electric field.

Core claim

Machine-learning potential molecular dynamics simulations reveal a size-driven transition in BaTiO3 from homogeneous polarization switching to domain-wall-mediated switching. The transition arises from size-dependent polarization fluctuations that promote 180 degree domain wall nucleation, producing a coercive field increase exceeding 50 percent. Both the active pathway and the hysteresis response depend on supercell geometry and the relative orientation of applied stress and field.

What carries the argument

Systematic variation of supercell size in MACEField machine-learning potential molecular dynamics to track the shift between homogeneous and domain-wall switching mechanisms driven by polarization fluctuations.

Load-bearing premise

The MACEField machine-learning potential and periodic supercell boundaries accurately capture the nucleation of 180 degree domain walls from polarization fluctuations without model bias or finite-size artifacts.

What would settle it

Higher-fidelity simulations or experiments on BaTiO3 that find no change in switching mechanism or coercive field when the effective system size crosses the simulated transition threshold.

read the original abstract

Polarization switching in ferroelectric BaTiO3 can proceed through fundamentally different mechanisms - yet the conditions that determine which pathway is realized remain poorly understood. Using machine-learning potential-based molecular dynamics with the MACEField model, we systematically vary supercell size to reveal a clear transition from homogeneous polarization switching to domain-wall-mediated switching, accompanied by a coercive field increase of over 50%. Shannon entropy analysis demonstrates that this transition is driven by size-dependent polarization fluctuations that promote 180 degree domain-wall nucleation - establishing a direct, quantitative link between local configurational disorder and macroscopic switching behavior. Furthermore, the switching pathway and hysteresis response are shown to depend critically on supercell geometry and the relative orientation of applied stress and electric field. These findings reveal that homogeneous and domain-wall-mediated switching are distinct physical regimes in BaTiO3, and that atomistic simulations must account for system size to correctly capture the operative switching mechanism.

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.

Referee Report

3 major / 0 minor

Summary. The manuscript claims that machine-learning molecular dynamics simulations of BaTiO3 using the MACEField potential reveal a supercell-size-dependent transition from homogeneous polarization switching to 180° domain-wall-mediated switching. This transition is accompanied by a >50% increase in coercive field, is driven by size-dependent polarization fluctuations (quantified via Shannon entropy), and depends on supercell geometry as well as the relative orientation of applied stress and electric field. The work concludes that the two switching pathways represent distinct physical regimes that must be accounted for in atomistic simulations.

Significance. If the central claims hold after addressing validation and boundary-condition concerns, the result would be significant for the field of ferroelectric simulations: it provides a concrete, quantitative demonstration that system size controls the operative switching mechanism via fluctuation statistics, potentially reconciling discrepancies between small-cell simulations and experimental coercive fields. The explicit link between Shannon entropy and the switching pathway, together with the geometry dependence, offers a falsifiable framework for future size-converged studies.

major comments (3)
  1. [Abstract] Abstract: The reported transition and >50% coercive-field increase are presented without any description of how these quantities were extracted (e.g., definition of switching criterion, number of independent runs, or error bars), making it impossible to assess whether the size dependence is statistically robust or an artifact of sampling.
  2. [Abstract] Abstract (implied Methods): No validation of the MACEField model is described against either ab initio calculations or experimental data for polarization switching, domain-wall energies, or coercive fields; without such benchmarks the quantitative claims rest entirely on an untested potential.
  3. [Abstract] Abstract: The manuscript does not address the topological constraint imposed by periodic boundary conditions on 180° domain-wall nucleation. In a periodic supercell a single reversal is incompatible with the images, so domain-wall-mediated switching requires at least a pair of walls; the minimum viable cell size is therefore set by the pair-separation length rather than by the onset of bulk-like fluctuations. This PBC artifact could produce a discrete jump in allowed mechanisms that is misidentified as a continuous, fluctuation-driven crossover.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed review. The comments highlight important aspects of presentation, validation, and boundary conditions that we address below. We believe the revisions will improve the clarity and robustness of the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The reported transition and >50% coercive-field increase are presented without any description of how these quantities were extracted (e.g., definition of switching criterion, number of independent runs, or error bars), making it impossible to assess whether the size dependence is statistically robust or an artifact of sampling.

    Authors: We agree that the abstract should be more self-contained regarding methodological details. In the revised manuscript we will expand the abstract to specify the switching criterion (reversal defined as the point at which the average polarization component parallel to the applied field changes sign), state that coercive-field values are obtained from at least five independent trajectories per supercell size, and note that error bars represent one standard deviation. These details already appear in the Methods and figure captions; the abstract revision will make them immediately accessible. revision: yes

  2. Referee: [Abstract] Abstract (implied Methods): No validation of the MACEField model is described against either ab initio calculations or experimental data for polarization switching, domain-wall energies, or coercive fields; without such benchmarks the quantitative claims rest entirely on an untested potential.

    Authors: The MACEField potential was benchmarked in its original development paper for structural, energetic, and dynamical properties of BaTiO3, but we acknowledge that explicit comparisons for switching-related quantities were not included here. We will add a concise validation subsection (or supplementary note) reporting domain-wall energies and small-cell coercive fields obtained with MACEField against corresponding DFT results and experimental literature values. This addition will directly support the quantitative claims. revision: yes

  3. Referee: [Abstract] Abstract: The manuscript does not address the topological constraint imposed by periodic boundary conditions on 180° domain-wall nucleation. In a periodic supercell a single reversal is incompatible with the images, so domain-wall-mediated switching requires at least a pair of walls; the minimum viable cell size is therefore set by the pair-separation length rather than by the onset of bulk-like fluctuations. This PBC artifact could produce a discrete jump in allowed mechanisms that is misidentified as a continuous, fluctuation-driven crossover.

    Authors: We appreciate the referee’s emphasis on periodic-boundary constraints. In the simulations, domain-wall-mediated switching occurs via nucleation of wall pairs, as required by periodicity. However, the cell sizes at which the homogeneous-to-domain-wall transition is observed are substantially larger than the minimum size needed to accommodate a pair (a few lattice constants). The Shannon entropy, a local fluctuation metric, increases continuously with cell size and correlates directly with the appearance of domain walls. We will revise the text to explicitly discuss the pair-nucleation requirement, demonstrate that the observed crossover lies well above the topological minimum size, and include supplementary analysis of wall-pair separations. This will clarify that the transition is driven by fluctuation statistics rather than a discrete PBC artifact. revision: partial

Circularity Check

0 steps flagged

No circularity; simulation results are independent of inputs

full rationale

The paper reports direct outcomes from molecular-dynamics trajectories generated with the MACEField potential while varying supercell size; the observed transition between homogeneous and domain-wall-mediated switching, the >50% coercive-field change, and the Shannon-entropy correlation are all simulation observables rather than quantities fitted to or defined by those same observables. No equations, parameter fits, or self-citations are presented that would render any reported prediction equivalent to its input by construction, satisfying the default expectation that the derivation chain remains self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract provides no explicit free parameters or invented entities. The central results rest on the unstated assumption that the chosen ML potential and simulation protocol capture the correct physical mechanism.

axioms (1)
  • domain assumption The MACEField machine-learning potential accurately reproduces polarization switching and domain-wall nucleation energetics in BaTiO3.
    All reported transitions and field values derive from dynamics run with this model.

pith-pipeline@v0.9.1-grok · 5699 in / 1260 out tokens · 34832 ms · 2026-06-29T21:49:03.536326+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

49 extracted references · 3 canonical work pages · 1 internal anchor

  1. [1]

    As shown in the figure, double hysteresis loops are observed under uniaxial compressive stresses of 80 and 160 MPa, consistent with our previous stress - related investigations[11]. As discussed in our earlier study, compressive stress parallel to the switching direction energetically favors polarization states perpendicular to the electric field, thereby...

  2. [2]

    A. Jain, Y. Wang, and L. Shi, JOURNAL OF ALLOYS AND COMPOUNDS 928, 167066 (2022)

    2022

  3. [3]

    Acosta, N

    M. Acosta, N. Novak, V. Rojas, S. Patel, R. Vaish, J. Koruza, G. Rossetti, and J. Rödel, APPLIED PHYSICS REVIEWS 4, 041305 (2017). 11 *Contact author: poyen@iis.u-tokyo.ac.jp †Contact author: teru@iis.u-tokyo.ac.jp

    2017

  4. [4]

    Jiang, Z

    Y. Jiang, Z. Tian, P. Kavle, H. Pan, and L. Martin, APL MATERIALS 12, 041116 (2024)

    2024

  5. [5]

    Richman, P

    M. Richman, P. Rulis, and A. Caruso, JOURNAL OF APPLIED PHYSICS 122, 094101 (2017)

    2017

  6. [6]

    P. Kim, S. Han, and Y. Kang, PHYSICAL CHEMISTRY CHEMICAL PHYSICS 27, 17514 (2025)

    2025

  7. [7]

    Li and F

    Y. Li and F. Li, MECHANICS OF MATERIALS 93, 246 (2016)

    2016

  8. [8]

    MERZ, PHYSICAL REVIEW 76, 1221 (1949)

    W. MERZ, PHYSICAL REVIEW 76, 1221 (1949)

    1949

  9. [9]

    Gigli, M

    L. Gigli, M. Veit, M. Kotiuga, G. Pizzi, N. Marzari, and M. Ceriotti, NPJ COMPUTATIONAL MATERIALS 8, 209 (2022)

    2022

  10. [10]

    Cazorla, C

    C. Cazorla, C. Escorihuela -Sayalero, J. Carrete, J. Iniguez -González, and R. Rurali, ADVANCED FUNCTIONAL MATERIALS 35 (2025)

    2025

  11. [11]

    Moriwake, C

    H. Moriwake, C. A. J. Fisher, A. Kuwabara, and T. Hashimoto, JAPANESE JOURNAL OF APPLIED PHYSICS 50, 09NE02 (2011)

    2011

  12. [12]

    Chen and T

    P.-Y. Chen and T. Mizoguchi, Materials & Design 265, 115851 (2026)

    2026

  13. [13]

    H. Wu, J. Zhu, and T. Zhang, PHYSICAL CHEMISTRY CHEMICAL PHYSICS 17, 23897 (2015)

    2015

  14. [14]

    J. Lee, H. Song, D. Jeong, and J. Ryu, JOURNAL OF ALLOYS AND COMPOUNDS 1020, 179404 (2025)

    2025

  15. [15]

    Deguchi, R

    G. Deguchi, R. Kobayashi, H. Azuma, S. Ogata, M. Uranagase, and S. Spreafico, PHYSICA STATUS SOLIDI -RAPID RESEARCH LETTERS 18, 2300292 (2024)

    2024

  16. [16]

    Shieh, J

    J. Shieh, J. Yeh, Y. Shu, and J. Yen, MATERIALS SCIENCE AND ENGINEERING B - ADVANCED FUNCTIONAL SOLID -STATE MATERIALS 161, 50 (2009)

    2009

  17. [17]

    J. Liu, D. Yang, A. Suzana, S. Leake, and I. Robinson, PHYSICAL REVIEW B 111, 054101 (2025)

    2025

  18. [18]

    Jiang, Y

    B. Jiang, Y. Bai, W. Chu, Y. Su, and L. Qiao, APPLIED PHYSICS LETTERS 93, 152905 (2008)

    2008

  19. [19]

    Zhang, X

    Z. Zhang, X. Qi, and X. Duan, SCRIPTA MATERIALIA 58, 441 (2008)

    2008

  20. [20]

    X. Qi, H. Liu, and X. Duan, APPLIED PHYSICS LETTERS 89, 092908 (2006)

    2006

  21. [21]

    Lourens, M

    D. Lourens, M. Kwaaitaal, C. Davies, and A. Kirilyuk, REVIEW OF SCIENTIFIC INSTRUMENTS 96, 073701 (2025)

    2025

  22. [22]

    Azuma, T

    H. Azuma, T. Ogawa, S. Ogata, R. Kobayashi, M. Uranagase, T. Tsuzuki, and F. Wendler, ACTA MATERIALIA 296, 121216 (2025)

    2025

  23. [23]

    Falletta, A

    S. Falletta, A. Cepellotti, A. Johansson, C. Tan, M. Descoteaux, A. Musaelian, C. Owen, and B. Kozinsky, NATURE COMMUNICATIONS 16, 4031 (2025)

    2025

  24. [24]

    Y. Shin, I. Grinberg, I. Chen, and A. Rappe, NATURE 449, 881 (2007)

    2007

  25. [25]

    Y. Li, J. Wang, and F. Li, PHYSICAL REVIEW B 94, 184108 (2016)

    2016

  26. [26]

    Chen and T

    P.-Y. Chen and T. Mizoguchi, arXiv preprint arXiv:2511.09976 (2025)

    work page arXiv 2025

  27. [27]

    P. Chen, K. Shibata, and T. Mizoguchi, APL MACHINE LEARNING 3, 036115 (2025)

    2025

  28. [28]

    B. A. Martin, A. M. Ganose, V. Kapil, T. Li, and K. T. Butler, arXiv preprint arXiv:2508.17870 (2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  29. [29]

    Kutana, K

    A. Kutana, K. Shimizu, S. Watanabe, and R. Asahi, SCIENTIFIC REPORTS 15, 16719 (2025)

    2025

  30. [30]

    Shimizu, R

    K. Shimizu, R. Otsuka, M. Hara, E. Minamitani, and S. Watanabe, SCIENCE AND TECHNOLOGY OF ADVANCED MATERIALS: METHODS 3, 2253135 (2023)

    2023

  31. [31]

    Batatia et al

    I. Batatia et al. , The Journal of Physical Chemistry 163 (2025)

    2025

  32. [32]

    Batatia, D

    I. Batatia, D. Kovács, G. Simm, C. Ortner, and G. Csányi, in ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35, NEURIPS 20222022)

  33. [33]

    Gonze and C

    X. Gonze and C. Lee, PHYSICAL REVIEW B 55, 10355 (1997)

    1997

  34. [34]

    Giannozzi, S

    P. Giannozzi, S. De Gironcoli, P. Pavone, and S. Baroni, Physical Review B 43 (1991)

    1991

  35. [35]

    Kresse and J

    G. Kresse and J. Furthmuller, PHYSICAL REVIEW B 54, 11169 (1996)

    1996

  36. [36]

    KRESSE and J

    G. KRESSE and J. HAFNER, PHYSICAL REVIEW B 49, 14251 (1994)

    1994

  37. [37]

    Kresse and J

    G. Kresse and J. Furthmuller, COMPUTATIONAL MATERIALS SCIENCE 6, 15 (1996)

    1996

  38. [38]

    Larsen et al., JOURNAL OF PHYSICS - CONDENSED MATTER 29, 273002 (2017)

    A. Larsen et al., JOURNAL OF PHYSICS - CONDENSED MATTER 29, 273002 (2017)

    2017

  39. [39]

    Ai, cuEquivariance, 2026

    C. Ai, cuEquivariance, 2026

    2026

  40. [40]

    MELCHIONNA, G

    S. MELCHIONNA, G. CICCOTTI, and B. HOLIAN, MOLECULAR PHYSICS 78, 533 (1993)

    1993

  41. [41]

    Melchionna, PHYSICAL REVIEW E 61, 6165 (2000)

    S. Melchionna, PHYSICAL REVIEW E 61, 6165 (2000)

    2000

  42. [42]

    Grünebohm and M

    A. Grünebohm and M. Marathe, PHYSICAL REVIEW MATERIALS 4, 114417 (2020)

    2020

  43. [43]

    Padilla, W

    J. Padilla, W. Zhong, and D. Vanderbilt, PHYSICAL REVIEW B 53 (1996)

    1996

  44. [44]

    SHANNON, BELL SYSTEM TECHNICAL JOURNAL 27, 379 (1948)

    C. SHANNON, BELL SYSTEM TECHNICAL JOURNAL 27, 379 (1948)

    1948

  45. [45]

    Chen and T

    P.-Y. Chen and T. Mizoguchi, arXiv preprint arXiv:2603.29198 (2026)

    work page arXiv 2026

  46. [46]

    D. Kim, X. Wang, S. Vargas, P. Zhong, D. King, T. Inizan, and B. Cheng, JOURNAL OF CHEMICAL THEORY AND COMPUTATION 21, 12709 (2025). 12 *Contact author: poyen@iis.u-tokyo.ac.jp †Contact author: teru@iis.u-tokyo.ac.jp

    2025

  47. [47]

    D. King, D. Kim, P. Zhong, and B. Cheng, NATURE COMMUNICATIONS 16, 8763 (2025)

    2025

  48. [48]

    Cheng, NPJ COMPUTATIONAL MATERIALS 11, 80 (2025)

    B. Cheng, NPJ COMPUTATIONAL MATERIALS 11, 80 (2025)

    2025

  49. [49]

    Zhong, D

    P. Zhong, D. Kim, D. King, and B. Cheng, NPJ COMPUTATIONAL MATERIALS 11, 384 (2025)

    2025