Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

A predictive machine learning force-field framework for liquid electrolyte development

Nature Machine Intelligence volume 7, pages 543–552 (2025)Cite this article

Subjects

A preprint version of the article is available at arXiv.

Abstract

Despite the widespread applications of machine learning force fields (MLFFs) in solids and small molecules, there is a notable gap in applying MLFFs to simulate liquid electrolytes—a critical component of current commercial lithium-ion batteries. Here we introduce ByteDance Artificial intelligence Molecular simulation Booster (BAMBOO), a predictive framework for molecular dynamics simulations, with a demonstration of its capability in the context of liquid electrolytes for lithium batteries. We design a physics-inspired graph equivariant transformer architecture as the backbone of BAMBOO to learn from quantum mechanical simulations. Additionally, we introduce an ensemble knowledge distillation approach and apply it to MLFFs to reduce the fluctuation of observations from molecular dynamics simulations. Finally, we propose a density alignment algorithm to align BAMBOO with experimental measurements. BAMBOO demonstrates state-of-the-art accuracy in predicting key electrolyte properties such as density, viscosity and ionic conductivity across various solvents and salt combinations. The current model, trained on more than 15 chemical species, achieves an average density error of 0.01 g cm−3 on various compositions compared with experiment.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on SpringerLink
  • Instant access to the full article PDF.

USD 39.95

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Overview of BAMBOO.
Fig. 2: Effects of GET layers, ensemble knowledge distillation and density alignment.
Fig. 3: Atomic charge distributions and solvation structure fractions of three simulated LiFSI electrolytes using BAMBOO.

Similar content being viewed by others

Data availability

DFT datasets of clusters are available at https://huggingface.co/datasets/mzl/bamboo (ref. 72). The input parameters, template of atomic structures for LAMMPS MD simulations and the final trained, ensemble knowledge distilled and density-aligned model of BAMBOO to reproduce the results in the paper are available via Zenodo at https://doi.org/10.5281/zenodo.14603020 (ref. 73).

Code availability

The source codes, including the GET model, the training module and the LAMMPS interface for the MD simulations, are available via GitHub at https://github.com/bytedance/bamboo and via Zenodo at https://zenodo.org/records/14603020 (ref. 73).

References

  1. Xu, K. Nonaqueous liquid electrolytes for lithium-based rechargeable batteries. Chem. Rev. 104, 4303–4418 (2004).

    Article  Google Scholar 

  2. Xu, K. Electrolytes, Interfaces and Interphases: Fundamentals and Applications in Batteries (Royal Society of Chemistry, 2023).

  3. Meng, Y. S., Srinivasan, V. & Xu, K. Designing better electrolytes. Science 378, eabq3750 (2022).

    Article  Google Scholar 

  4. Unke, O. T. et al. Machine learning force fields. Chem. Rev. 121, 10142–10186 (2021).

    Article  Google Scholar 

  5. Frenkel, D. & Smit, B. Understanding Molecular Simulation: From Algorithms To Applications (Elsevier, 2023).

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

    Article  Google Scholar 

  7. Zhang, L., Han, J., Wang, H., Car, R. & E, W. Deep potential molecular dynamics: a scalable model with the accuracy of quantum mechanics. Phys. Rev. Lett. 120, 143001 (2018).

    Article  Google Scholar 

  8. Lysogorskiy, Y. et al. Performant implementation of the atomic cluster expansion (PACE) and application to copper and silicon. npj Comput. Mater. 7, 97 (2021).

    Article  Google Scholar 

  9. Schütt, K. T., Sauceda, H. E., Kindermans, P.-J., Tkatchenko, A. & Müller, K.-R. SchNet—a deep learning architecture for molecules and materials. J. Chem. Phys. 148, 241722 (2018).

    Article  Google Scholar 

  10. Batzner, S. et al. E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials. Nat. Commun. 13, 2453 (2022).

    Article  Google Scholar 

  11. Liao, Y.-L. & Smidt, T. Equiformer: equivariant graph attention transformer for 3D atomistic graphs. In Proc. Eleventh International Conference on Learning Representations https://openreview.net/forum?id=KwmPfARgOTD (ICLR, 2023).

  12. Thölke, P. & De Fabritiis, G. TorchMD-NET: equivariant transformers for neural network based molecular potentials. In Proc. Tenth International Conference on Learning Representations https://openreview.net/forum?id=zNHzqZ9wrRB (ICLR, 2022).

  13. 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 (2021).

    Article  Google Scholar 

  14. Zhang, L. et al. A deep potential model with long-range electrostatic interactions. J. Chem. Phys. 156, 124107 (2022).

    Article  Google Scholar 

  15. Anstine, D., Zubatyuk, R. & Isayev, O. AIMNet2: a neural network potential to meet your neutral, charged, organic, and elemental-organic needs. Preprint at ChemRxiv https://doi.org/10.26434/chemrxiv-2023-296ch-v3 (2023).

  16. Unke, O. & Meuwly, M. PhysNet: a neural network for predicting energies, forces, dipole moments and partial charges. J. Chem. Theory Comput. 15, 3678–3693 (2019).

    Article  Google Scholar 

  17. Unke, O. et al. SpookyNet: learning force fields with electronic degrees of freedom and nonlocal effects. Nat. Commun. 12, 12 (2021).

    Article  Google Scholar 

  18. Yu, H. et al. Spin-dependent graph neural network potential for magnetic materials. Phys. Rev. B 109, 144426 (2023).

  19. Chen, C. & Ong, S. A universal graph deep learning interatomic potential for the periodic table. Nat. Comput. Sci. 2, 718–728 (2022).

    Article  Google Scholar 

  20. Deng, B. et al. CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling. Nat. Mach. Intell. 5, 1031–1041 (2023).

    Article  Google Scholar 

  21. Merchant, A. et al. Scaling deep learning for materials discovery. Nature 624, 80–85 (2023).

    Article  Google Scholar 

  22. Zhang, D. et al. DPA-2: towards a universal large atomic model for molecular and material simulation. npj Comput. Mater. https://doi.org/10.1038/s41524-024-01493-2 (2024).

  23. Batatia, I. et al. A foundation model for atomistic materials chemistry. Preprint at https://arxiv.org/abs/2401.00096 (2024).

  24. Kovács, D. P. et al. MACE-OFF23: transferable machine learning force fields for organic molecules. Preprint at https://arxiv.org/abs/2312.15211v1 (2023).

  25. Wang, H. & Yang, W. Force field for water based on neural network. J. Phys. Chem. Lett. 9, 3232–3240 (2018).

    Article  Google Scholar 

  26. Zhang, J., Pagotto, J., Gould, T. & Duignan, T. T. Accurate, fast and generalisable first principles simulation of aqueous lithium chloride. Preprint at https://arxiv.org/abs/2310.12535v1 (2023).

  27. Magdău, I. B. et al. Machine learning force fields for molecular liquids: ethylene carbonate/ethyl methyl carbonate binary solvent. npj Comput. Mater. 9, 146 (2023).

    Article  Google Scholar 

  28. Montes-Campos, H., Carrete, J., Bichelmaier, S., Varela, L. M. & Madsen, G. K. H. A differentiable neural-network force field for ionic liquids. J. Chem. Inf. Model 62, 88–101 (2022).

    Article  Google Scholar 

  29. Wang, F. & Cheng, J. Understanding the solvation structures of glyme-based electrolytes by machine learning molecular dynamics. Chinese J. Struct. Chem. 42, 100061 (2023).

    Google Scholar 

  30. Dajnowicz, S. et al. High-dimensional neural network potential for liquid electrolyte simulations. J. Phys. Chem. B 126, 08 (2022).

    Article  Google Scholar 

  31. Jacobson, L. et al. Transferable neural network potential energy surfaces for closed-shell organic molecules: extension to ions. J. Chem. Theory Comput. 18, 03 (2022).

    Article  Google Scholar 

  32. Fu, X. et al. Forces are not enough: benchmark and critical evaluation for machine learning force fields with molecular simulations. Trans. Mach. Learn. Res. https://openreview.net/pdf?id=A8pqQipwkt (2023).

  33. Ormeño, F. & General, I. Convergence and equilibrium in molecular dynamics simulations. Commun. Chem. 7, 02 (2024).

    Article  Google Scholar 

  34. Wang, X. et al. DMFF: an open-source automatic differentiable platform for molecular force field development and molecular dynamics simulation. J. Chem. Theory Comput. 19, 5897–5909 (2023).

    Article  Google Scholar 

  35. Greener, J. G. & Jones, D. T. Differentiable molecular simulation can learn all the parameters in a coarse-grained force field for proteins. PLoS ONE 16, e0256990 (2021).

    Article  Google Scholar 

  36. Asif, U., Tang, J. & Harrer, S. Ensemble knowledge distillation for learning improved and efficient network. In 24th European Conference on Artificial Intelligence http://ecai2020.eu/papers/521_paper.pdf (ECAI, 2020).

  37. Chandler, D., Weeks, J. D. & Andersen, H. C. Van der Waals picture of liquids, solids, and phase transformations. Science 220, 787–794 (1983).

    Article  Google Scholar 

  38. Kontogeorgis, G. M., Maribo-Mogensen, B. & Thomsen, K. The Debye-Hückel theory and its importance in modeling electrolyte solutions. Fluid Ph. Equilib. 462, 130–152 (2018).

    Article  Google Scholar 

  39. Poier, P., Lagardère, L., Piquemal, J.-P. & Jensen, F. Molecular dynamics using nonvariational polarizable force fields: theory, periodic boundary conditions implementation, and application to the bond capacity model. J. Chem. Theory Comput. 2019, 09 (2019).

    Google Scholar 

  40. Schröder, H., Creon, A. & Schwabe, T. Reformulation of the D3(Becke–Johnson) dispersion correction without resorting to higher than C6 dispersion coefficients. J. Chem. Theory Comput. 11, 3163–3170 (2015).

    Article  Google Scholar 

  41. Torres-Sánchez, A., Vanegas, J. M. & Arroyo, M. Geometric derivation of the microscopic stress: a covariant central force decomposition. J. Mech. Phys. Solids 93, 224–239 (2016).

    Article  MathSciNet  Google Scholar 

  42. Vaswani, A. et al. Attention is all you need. In Advances in Neural Information Processing Systems Vol. 30 (eds Guyon, I. et al.) https://proceedings.neurips.cc/paper_files/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf (Curran, 2017).

  43. Gong, S. et al. Examining graph neural networks for crystal structures: limitations and opportunities for capturing periodicity. Sci. Adv. 9, eadi3245 (2023).

    Article  Google Scholar 

  44. Thompson, A. P. et al. LAMMPS—a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales. Comp. Phys. Comm. 271, 108171 (2022).

    Article  Google Scholar 

  45. Musaelian, A. et al. Learning local equivariant representations for large-scale atomistic dynamics. Nat. Commun. 14, 579 (2023).

    Article  Google Scholar 

  46. Wang, Y. et al. Enhancing geometric representations for molecules with equivariant vector-scalar interactive message passing. Nat. Commun. https://doi.org/10.1038/s41467-023-43720-2 (2024).

  47. Batatia, I., Kovacs, D. P., Simm, G., Ortner, C. & Csányi, G. MACE: higher order equivariant message passing neural networks for fast and accurate force fields. Adv. Neural Inf. Process. Syst. 35, 11423–11436 (2022).

    Google Scholar 

  48. Pelaez, R. P. et al. TorchMD-Net 2.0: fast neural network potentials for molecular simulations. Preprint at https://arxiv.org/abs/2402.17660 (2024).

  49. Martius, G. & Lampert, C. H. Extrapolation and learning equations. In Proc. 35th International Conference on Machine Learning https://proceedings.mlr.press/v80/sahoo18a.html (PMLR, 2018).

  50. Dave, A. R. Automated Design and Discovery of Liquid Electrolytes for Lithium-Ion Batteries. PhD thesis, Carnegie Mellon Univ. (2023).

  51. Hagiyama, K. et al. Physical properties of substituted 1,3-dioxolan-2-ones. Chem. Lett. 37, 210–211 (2008).

    Article  Google Scholar 

  52. Sasaki, Y. in Fluorinated Materials for Energy Conversion (eds Nakajima, T. & Groult, H.) 285–304 (Elsevier Science, 2005).

  53. Jänes, A., Thomberg, T., Eskusson, J. & Lust, E. Fluoroethylene carbonate and propylene carbonate mixtures based electrolytes for supercapacitors. ECS Trans. 58, 71 (2014).

    Article  Google Scholar 

  54. Gores, H. J. et al. in Handbook of Battery Materials 525–626 (John Wiley & Sons, 2011).

  55. Dave, A. et al. Autonomous optimization of non-aqueous Li-ion battery electrolytes via robotic experimentation and machine learning coupling. Nat. Commun. 13, 5454 (2022).

    Article  Google Scholar 

  56. Logan, E. R. et al. A study of the transport properties of ethylene carbonate-free Li electrolytes. J. Electrochem. Soc. 165, A705–A716 (2018).

    Article  Google Scholar 

  57. Zhu, S. et al. Differentiable modeling and optimization of non-aqueous Li-based battery electrolyte solutions using geometric deep learning. Nat. Commun. 15, 8649 (2024).

    Article  Google Scholar 

  58. Jorgensen, W. L., Maxwell, D. S. & Tirado-Rives, J. Development and testing of the OPLS all-atom force field on conformational energetics and properties of organic liquids. J. Am. Chem. Soc. 118, 11225–11236 (1996).

    Article  Google Scholar 

  59. Zheng, T. et al. Data-driven parametrization of molecular mechanics force fields for expansive chemical space coverage. Chem. Sci. 16, 2730–2740 (2025).

    Article  Google Scholar 

  60. Niblett, S. P., Kourtis, P., Magdău, I. B., Grey, C. P. & Csányi, G. Transferability of datasets between machine-learning interaction potentials. Preprint at https://arxiv.org/abs/2409.05590 (2024).

  61. Zhang, H., Juraskova, V. & Duarte, F. Modelling chemical processes in explicit solvents with machine learning potentials. Nat. Commun. 15, 6114 (2024).

    Article  Google Scholar 

  62. Lei Ba, J., Kiros, J. R. & Hinton, G. E. Layer normalization. Preprint at https://arxiv.org/abs/1607.06450 (2016).

  63. Breneman, C. M. & Wiberg, K. B. Determining atom-centered monopoles from molecular electrostatic potentials. The need for high sampling density in formamide conformational analysis. J. Comput. Chem. 11, 361–373 (1990).

    Article  Google Scholar 

  64. Paszke, A. et al. PyTorch: an imperative style, high-performance deep learning library. Adv. Neural Inf. Process. Syst. 32 (2019).

  65. Kingma, D. P. & Ba, J. Adam: a method for stochastic optimization. Preprint at https://arxiv.org/abs/1412.6980 (2017).

  66. Becke, A. D. Density functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 98, 5648–5652 (1993).

    Article  Google Scholar 

  67. Hellweg, A. & Rappoport, D. Development of new auxiliary basis functions of the Karlsruhe segmented contracted basis sets including diffuse basis functions (def2-SVPD, def2-TZVPPD, and def2-QVPPD) for RI-MP2 and RI-CC calculations. Phys. Chem. Chem. Phys. 17, 1010–1017 (2015).

    Article  Google Scholar 

  68. Weigend, F. Hartree–Fock exchange fitting basis sets for H to Rn. J. Comput. Chem. 29, 167–175 (2008).

    Article  Google Scholar 

  69. Wu, X. et al. Python-based quantum chemistry calculations with GPU acceleration. Preprint at https://arxiv.org/abs/2404.09452v1 (2024).

  70. Shao, Y. et al. Advances in molecular quantum chemistry contained in the Q-Chem 4 program package. Mol. Phys. 113, 184–215 (2015).

    Article  Google Scholar 

  71. Mistry, A., Yu, Z., Cheng, L. & Srinivasan, V. On relative importance of vehicular and structural motions in defining electrolyte transport. J. Electrochem. Soc. 170, 110536 (2023).

    Article  Google Scholar 

  72. Mu, Z. mzl/bamboo. Hugging Face https://doi.org/10.57967/hf/3971 (2025).

  73. Mu, Z. muzhenliang/bamboo: v0.1. Zenodo https://doi.org/10.5281/zenodo.14603020 (2025).

  74. Simeon, G. & De Fabritiis, G. TensorNet: Cartesian tensor representations for efficient learning of molecular potentials. Adv. Neural Inf. Process. Syst. 36, 37334–37353 (2024).

    Google Scholar 

Download references

Acknowledgements

We acknowledge insightful discussion on ion transport theory with A. Mistry, an Assistant Professor at the Colorado School of Mines. We also acknowledge the experimental data points provided by A. Dave, a former PhD student at Carnegie Mellon University, upon request. H.W. and M.C. worked as interns at ByteDance Research during this study.

Author information

Author notes

  1. These authors contributed equally: Sheng Gong, Yumin Zhang, Zhenliang Mu.

Authors and Affiliations

  1. ByteDance Research, Bellevue, WA, USA

    Sheng Gong, Yumin Zhang, Zhiao Yu, Zhi Wang, Zhenze Yang, Xiaojie Wu & Wen Yan

  2. ByteDance Research, Beijing, China

    Zhenliang Mu, Zhichen Pu, Hongyi Wang, Xu Han, Mengyi Chen, Tianze Zheng, Lifei Chen, Shaochen Shi, Weihao Gao & Liang Xiang

Authors
  1. Sheng Gong
    View author publications

    Search author on:PubMed Google Scholar

  2. Yumin Zhang
    View author publications

    Search author on:PubMed Google Scholar

  3. Zhenliang Mu
    View author publications

    Search author on:PubMed Google Scholar

  4. Zhichen Pu
    View author publications

    Search author on:PubMed Google Scholar

  5. Hongyi Wang
    View author publications

    Search author on:PubMed Google Scholar

  6. Xu Han
    View author publications

    Search author on:PubMed Google Scholar

  7. Zhiao Yu
    View author publications

    Search author on:PubMed Google Scholar

  8. Mengyi Chen
    View author publications

    Search author on:PubMed Google Scholar

  9. Tianze Zheng
    View author publications

    Search author on:PubMed Google Scholar

  10. Zhi Wang
    View author publications

    Search author on:PubMed Google Scholar

  11. Lifei Chen
    View author publications

    Search author on:PubMed Google Scholar

  12. Zhenze Yang
    View author publications

    Search author on:PubMed Google Scholar

  13. Xiaojie Wu
    View author publications

    Search author on:PubMed Google Scholar

  14. Shaochen Shi
    View author publications

    Search author on:PubMed Google Scholar

  15. Weihao Gao
    View author publications

    Search author on:PubMed Google Scholar

  16. Wen Yan
    View author publications

    Search author on:PubMed Google Scholar

  17. Liang Xiang
    View author publications

    Search author on:PubMed Google Scholar

Contributions

Conceptualization: Y.Z., W.Y., W.G., Z.M., Z. Yu, S.G., T.Z., X.H., Z. Yang, Z.W., L.C., X.W., S.S. and L.X. Methodology: S.G., Y.Z., Z.M., Z.P., H.W., M.C., W.Y. and W.G. Investigation: S.G., Y.Z., Z.M., Z.P., H.W., M.C., X.H., Z. Yu, W.Y. and W.G. Supervision: W.G., W.Y. and L.X. Writing: S.G., Y.Z., Z.M., Z.P., H.W., W.Y. and W.G.

Corresponding authors

Correspondence to Weihao Gao or Wen Yan.

Ethics declarations

Competing interests

All authors were employees of ByteDance when conducting this research project. ByteDance holds intellectual property rights pertinent to the research presented here. Furthermore, the innovations described here have resulted in the filing of a patent application in China (application no. 202311322469.2), which is currently pending.

Peer review

Peer review information

Nature Machine Intelligence thanks Jiayu Peng and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information (download PDF )

Supplementary Notes 1–11, Figs. 1–16, Tables 1–16 and Equations (1)–(68).

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Gong, S., Zhang, Y., Mu, Z. et al. A predictive machine learning force-field framework for liquid electrolyte development. Nat Mach Intell 7, 543–552 (2025). https://doi.org/10.1038/s42256-025-01009-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • Version of record:

  • Issue date:

  • DOI: https://doi.org/10.1038/s42256-025-01009-7

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing