Abstract
Developing neural network potentials (NNPs) accurate under non-equilibrium dynamics is challenging, as such systems require extensive sampling beyond equilibrium phases. Here we construct high-fidelity NNPs for zinc oxide (ZnO), a polymorphic ionic solid, using density functional theory (DFT) reference data. To efficiently capture transitional configurations, we combine enhanced-sampling molecular dynamics with empirical potentials, data distillation, and pretraining on short-range atomic energies (A-Train), followed by transfer learning with DFT-relabeled datasets. This hierarchical approach improves transferability across polymorphs and stress states. We further introduce effective charge separation, treating long-range Coulombic terms analytically while short-range residual interactions are learned by the NNP. The optimal effective charges fall in the range 0.5–1.0 qe, consistent with dielectric-screened values derived from formal charges but distinct from Bader estimates. Motivated by this observation, we propose a simple data-driven protocol in which effective charges are optimized by comparing DFT reference energies with explicit Coulomb calculations, without additional NNP training. This strategy improves accuracy and transferability in DFT-level predictions of energies, forces, and stress. Together, these results provide a practical charge-selection framework for robust NNP development in ionic solids, enabling reliable simulation of polymorphic phase transformations and non-equilibrium dynamics.
Similar content being viewed by others
Data availability
LAMMPS input files for running molecular dynamics simulations with the trained NNP using the hybrid/overlay pair style are available at: (https://github.com/gsjung0419/Hybrid_NNP). Atomic energy training (A-Train) and transfer learning examples are available at: (https://github.com/gsjung0419/TorchANITutorials). The additional data that support the findings of this study are available from the corresponding author upon reasonable request.
References
Escobar, A., Donado, F., Moctezuma, R. E. & Weeks, E. R. Direct observation of crystal nucleation and growth in a quasi-two-dimensional nonvibrating granular system. Phys. Rev. E 104, 044904. https://doi.org/10.1103/PhysRevE.104.044904 (2021).
Sleutel, M., Lutsko, J., Van Driessche, A. E. S., Durán-Olivencia, M. A. & Maes, D. Observing classical nucleation theory at work by monitoring phase transitions with molecular precision. Nat. Commun. 5, 5598 (2014).
Li, M. et al. Revealing thermally-activated nucleation pathways of diffusionless solid-to-solid transition. Nat. Commun. 12, 4042 (2021).
Nielsen, M. H., Aloni, S. & De Yoreo, J. J. In situ TEM imaging of CaCO3 nucleation reveals coexistence of direct and indirect pathways. Science 345, 1158–1162 (2014).
De Yoreo, J. J. et al. Crystallization by particle attachment in synthetic, biogenic, and geologic environments. Science 349, aaa6760. https://doi.org/10.1126/science.aaa6760 (2015).
Matsumoto, M., Saito, S. & Ohmine, I. Molecular dynamics simulation of the ice nucleation and growth process leading to water freezing. Nature 416, 409–413, https://doi.org/10.1038/416409a (2002).
Wang, J. et al. Molecular dynamics and density functional studies of a body-centered-tetragonal polymorph of ZnO. Phys. Rev. B—Condens. Matter Mater. Phys. 76, 172103 (2007).
Laio, A. & Parrinello, M. Escaping free-energy minima. Proc. Natl. Acad. Sci. 99, 12562, https://doi.org/10.1073/pnas.202427399 (2002).
Okumura, H. & Okamoto, Y. Multibaric–multithermal ensemble molecular dynamics simulations. J. Comput. Chem. 27, 379–395, https://doi.org/10.1002/jcc.20351 (2006).
Yoshimoto, Y. Extended multicanonical method combined with thermodynamically optimized potential: Application to the liquid-crystal transition of silicon. J. Chem. Phys. 125, 184103. https://doi.org/10.1063/1.2363987 (2006).
Sugita, Y. & Okamoto, Y. Replica-exchange molecular dynamics method for protein folding. Chem. Phys. Lett. 314, 141–151, https://doi.org/10.1016/S0009-2614(99)01123-9 (1999).
Behler, J. & Csányi, G. Machine learning potentials for extended systems: a perspective. Eur. Phys. J. B 94, 142. https://doi.org/10.1140/epjb/s10051-021-00156-1 (2021).
Kocer, E., Ko, T. W. & Behler, J. Neural network potentials: a concise overview of methods. Annu. Rev. Phys. Chem. 73, 163–186, https://doi.org/10.1146/annurev-physchem-082720-034254 (2022).
Pinheiro, M., Ge, F., Ferré, N., Dral, P. O. & Barbatti, M. Choosing the right molecular machine learning potential. Chem. Sci. 12, 14396–14413, https://doi.org/10.1039/D1SC03564A (2021).
Unke, O. T. et al. Machine Learning Force Fields. Chem. Rev. 121, 10142–10186, https://doi.org/10.1021/acs.chemrev.0c01111 (2021).
Lorenz, S., Groß, A. & Scheffler, M. Representing high-dimensional potential-energy surfaces for reactions at surfaces by neural networks. Chem. Phys. Lett. 395, 210–215 (2004).
Behler, J. & Parrinello, M. Generalized neural-network representation of high-dimensional potential-energy surfaces. Phys. Rev. Lett. 98, 146401. https://doi.org/10.1103/PhysRevLett.98.146401 (2007).
Jung, G. S., Choi, J. Y. & Lee, S. M. Active learning of neural network potentials for rare events. Digit. Discov. https://doi.org/10.1039/D3DD00216K (2024).
Mehdi, S., Smith, Z., Herron, L., Zou, Z. & Tiwary, P. Enhanced sampling with machine learning. Annu. Rev. Phys. Chem. 75, 347–370, https://doi.org/10.1146/annurev-physchem-083122-125941 (2024).
Jung, G. S. Enhancing high-fidelity neural network potentials through low-fidelity sampling. APL Mach. Learn. 2, 046102. https://doi.org/10.1063/5.0222779 (2024).
Jung, G. S., Lee, S. & Choi, J. Y. Data Distillation for Neural Network Potentials toward Foundational Dataset. OpenReview.net https://openreview.net/forum?id=mE6ldawl0n (2024).
Jung, G. S. Atomic energy accuracy of neural network potentials: harnessing pretraining and transfer learning. J. Chem. Inf. Model. 65, 4797–4807, https://doi.org/10.1021/acs.jcim.5c00079 (2025).
Özgür, Ü. et al. A comprehensive review of ZnO materials and devices. J. Appl. Phys. 98, https://doi.org/10.1063/1.1992666 (2005).
Ashrafi, A. & Jagadish, C. Review of zincblende ZnO: Stability of metastable ZnO phases. J. Appl. Phys. 102, https://doi.org/10.1063/1.2787957 (2007).
Rasoulkhani, R. et al. Energy landscape of ZnO clusters and low-density polymorphs. Phys. Rev. B 96, 064108 (2017).
Santiso, E. E. & Trout, B. L. A general set of order parameters for molecular crystals. J. Chem. Phys. 134, https://doi.org/10.1063/1.3548889 (2011).
Tong, H. & Tanaka, H. Structural order as a genuine control parameter of dynamics in simple glass formers. Nat. Commun. 10, 5596 (2019).
Steinhardt, P. J., Nelson, D. R. & Ronchetti, M. Bond-orientational order in liquids and glasses. Phys. Rev. B 28, 784–805, https://doi.org/10.1103/PhysRevB.28.784 (1983).
Murata, K. -i & Tanaka, H. Microscopic identification of the order parameter governing liquid–liquid transition in a molecular liquid. Proc. Natl. Acad. Sci. 112, 5956–5961 (2015).
Yoshimoto, Y. Melting of MgO studied using a multicanonical ensemble method combined with a first-principles calculation. J. Phys. Soc. Jpn. 79, 034602 (2010).
Invernizzi, M., Piaggi, P. M. & Parrinello, M. Unified approach to enhanced sampling. Phys. Rev. X 10, 041034. https://doi.org/10.1103/PhysRevX.10.041034 (2020).
Jung, G. S., Yoshimoto, Y., Oh, K. J. & Tsuneyuki, S. Extended ensemble molecular dynamics for thermodynamics of phases. arXiv preprint arXiv:2308.08098 (2023).
Ghasemi, S. A., Hofstetter, A., Saha, S. & Goedecker, S. Interatomic potentials for ionic systems with density functional accuracy based on charge densities obtained by a neural network. Phys. Rev. B 92, 045131 (2015).
Kocer, E. et al. Iterative charge equilibration for fourth-generation high-dimensional neural network potentials. J. Chem. Phys. 162, 124106 (2025).
Ko, T. W., Finkler, J. A., Goedecker, S. & Behler, J. A fourth-generation high-dimensional neural network potential with accurate electrostatics including non-local charge transfer. Nat. Commun. 12, 398. https://doi.org/10.1038/s41467-020-20427-2 (2021).
Cheng, B. Latent Ewald summation for machine learning of long-range interactions. npj Comput. Mater. 11, 80. https://doi.org/10.1038/s41524-025-01577-7 (2025).
Wang, Y. et al. Neural P $ 3$ M: a long-range interaction modeling enhancer for geometric GNNs. Adv. Neural Inf. Process. Syst. 37, 120336–120365 (2024).
Tang, W., Sanville, E. & Henkelman, G. A grid-based Bader analysis algorithm without lattice bias. J. Phys.: Condens. Matter 21, 084204 (2009).
Li, Q.-J. et al. Development of robust neural-network interatomic potential for molten salt. Cell Rep. Phys. Sci. 2, 2021 (2021).
Shen, C. et al. SuperSalt: equivariant neural network force fields for multicomponent molten salts system. Nat. Commun. 16, 7280. https://doi.org/10.1038/s41467-025-62450-1 (2025).
Chen, X. et al. TensorAlloy: An automatic atomistic neural network program for alloys. Comput. Phys. Commun. 250, 107057 (2020).
Chen, C. & Ong, S. P. A universal graph deep learning interatomic potential for the periodic table. Nat. Comput. Sci. 2, 718–728 (2022).
Deng, B. et al. CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling. Nat. Mach. Intell. 5, 1031–1041 (2023).
Gubler, M., Finkler, J. A., Schäfer, M. R., Behler, J. & Goedecker, S. Accelerating fourth-generation machine learning potentials using quasi-linear scaling particle mesh charge equilibration. J. Chem. Theory Comput. 20, 7264–7271, https://doi.org/10.1021/acs.jctc.4c00334 (2024).
Schleife, A., Fuchs, F., Furthmüller, J. & Bechstedt, F. First-principles study of ground-and excited-state properties of MgO, ZnO, and CdO polymorphs. Phys. Rev. B—Condens. Matter Mater. Phys. 73, 245212 (2006).
Calzolari, A. & Nardelli, M. B. Dielectric properties and Raman spectra of ZnO from a first principles finite-differences/finite-fields approach. Sci. Rep. 3, 2999. https://doi.org/10.1038/srep02999 (2013).
Morkoç, H. & Özgür, Ü. Zinc oxide: fundamentals, materials and device technology. (John Wiley & Sons, 2008).
Segura, A., Sans, J. A., Manjón, F. J., Munoz, A. & Herrera-Cabrera, M. J. Optical properties and electronic structure of rock-salt ZnO under pressure. Appl. Phys. Lett. 83, 278–280 (2003).
Agrawal, R., Peng, B. & Espinosa, H. D. Experimental-computational investigation of ZnO nanowires strength and fracture. Nano Lett. 9, 4177–4183 (2009).
Plimpton, S. Fast parallel algorithms for short-range molecular dynamics. J. Comput. Phys. 117, 1–19, https://doi.org/10.1006/jcph.1995.1039 (1995).
Lewis, G. V. & Catlow, C. R. A. Potential models for ionic oxides. J. Phys. C: Solid State Phys. 18, 1149 (1985).
Crain, J. et al. Reversible pressure-induced structural transitions between metastable phases of silicon. Phys. Rev. B 50, 13043–13046, https://doi.org/10.1103/PhysRevB.50.13043 (1994).
Beniash, E. et al. The hidden structure of human enamel. Nat. Commun. 10, 4383. https://doi.org/10.1038/s41467-019-12185-7 (2019).
Jung, G. S. & Buehler, M. J. Multiscale modeling of muscular-skeletal systems. Annu. Rev. Biomed. Eng. 19, 435–457, https://doi.org/10.1146/annurev-bioeng-071516-044555 (2017).
Wang, F. & Landau, D. P. Efficient, multiple-range random walk algorithm to calculate the density of states. Phys. Rev. Lett. 86, 2050–2053, https://doi.org/10.1103/PhysRevLett.86.2050 (2001).
Jung, G. S., Irle, S. & Sumpter, B. G. Dynamic aspects of graphene deformation and fracture from approximate density functional theory. Carbon 190, 183–193, https://doi.org/10.1016/j.carbon.2022.01.002 (2022).
Jung, G. S., Myung, H. & Irle, S. Artificial neural network potentials for mechanics and fracture dynamics of two-dimensional crystals*. Mach. Learn.: Sci. Technol. 4, 035001, https://doi.org/10.1088/2632-2153/accd45 (2023).
Jung, G. S., Choi, J. Y. & Lee, S. M. AL-ASMR: Active Learning of Atomistic Surrogate Models for Rare Events. (Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States), 2023).
Giannozzi, P. et al. QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials. J. Phys.: Condens. Matter 21, 395502, https://doi.org/10.1088/0953-8984/21/39/395502 (2009).
Giannozzi, P. et al. Advanced capabilities for materials modelling with Quantum ESPRESSO. J. Phys.: Condens. Matter 29, 465901, https://doi.org/10.1088/1361-648x/aa8f79 (2017).
Dal Corso, A. Pseudopotentials periodic table: From H to Pu. Comput. Mater. Sci. 95, 337–350, https://doi.org/10.1016/j.commatsci.2014.07.043 (2014).
Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953–17979, https://doi.org/10.1103/PhysRevB.50.17953 (1994).
Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865–3868, https://doi.org/10.1103/PhysRevLett.77.3865 (1996).
Grimme, S., Antony, J., Ehrlich, S. & Krieg, H. A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 132, 154104. https://doi.org/10.1063/1.3382344 (2010).
Hjorth Larsen, A. et al. The atomic simulation environment—a Python library for working with atoms. J. Phys.: Condens. Matter 29, 273002, https://doi.org/10.1088/1361-648X/aa680e (2017).
Loche, P. et al. Fast and flexible long-range models for atomistic machine learning. J. Chem. Phys. 162, 142501 (2025).
Gao, X., Ramezanghorbani, F., Isayev, O., Smith, J. S. & Roitberg, A. E. TorchANI: a free and open source PyTorch-based deep learning implementation of the ANI neural network potentials. J. Chem. Inf. Model. 60, 3408–3415, https://doi.org/10.1021/acs.jcim.0c00451 (2020).
Hendrycks, D. & Gimpel, K. Gaussian Error Linear Units (GELUs). arXiv [cs.LG] (2020).
Kingma, D. P. & Ba, J. Adam: A Method for Stochastic Optimization. arXiv [cs.LG] (2017).
Loshchilov, I. & Hutter, F. Decoupled Weight Decay Regularization. arXiv [cs.LG] (2019).
Raymand, D., Van Duin, A. C. T., Baudin, M. & Hermansson, K. A reactive force field (ReaxFF) for zinc oxide. Surf. Sci. 602, 1020–1031 (2008).
Acknowledgements
This research used resources from the Compute and Data Environment for Science (CADES) at the Oak Ridge National Laboratory and National Energy Research Scientific Computing Center (NERSC), a DOE Office of Science User Facility for access to additional supercomputing resources. This work is also supported as a part of a user project at the Center for Nanophase Materials Sciences (CNMS), a US Department of Energy, Office of Science User Facility at Oak Ridge National Laboratory. This work was supported by the Laboratory Directed Research and Development Program (LDRD) of Oak Ridge National Laboratory (NEAT), managed by UT-Battelle, LLC, for the US Department of Energy under contract DEAC05-00OR22725.
Author information
Authors and Affiliations
Contributions
G.S.J. conceived the idea, developed codes, performed simulations, training, and evaluations, and wrote and edited the draft. L.C. contributed to the discussion and conceptual development and provided revisions to improve the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Jung, G.S., Cheng, L. Neural network potentials with effective charge separation for non-equilibrium dynamics of ionic solids: a ZnO case study. npj Comput Mater (2026). https://doi.org/10.1038/s41524-025-01946-2
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41524-025-01946-2
