Machine Learning Interatomic Potentials for Energy Materials: Architectures, Training Strategies, and Applications
Source abstract
Abstract
ABSTRACT Machine learning interatomic potentials (MLIPs) provide a practical route to bridge first‐principles accuracy with the spatiotemporal scales required for modeling complex energy materials. By learning potential‐energy surfaces from density functional theory (DFT) data, MLIPs enable atomistic simulations up to 10 4 –10 6 atoms over nanosecond timescales, thereby extending access to phenomena inaccessible to ab initio molecular dynamics. This review presents a phenomenon‐oriented synthesis of MLIP development, spanning descriptor‐based models, Gaussian process frameworks with uncertainty quantification, equivariant graph neural networks, and foundation models trained on large‐scale datasets. We assess how architectural choices, training strategies (including active learning and multi‐fidelity approaches), and dataset design influence accuracy, transferability, and computational efficiency across representative energy systems. Applications in solid‐state electrolytes, battery electrodes, electrocatalysts, perovskite photovoltaics, and high‐entropy materials are discussed with both performance metrics and methodological rationale. While MLIPs capture qualitative trends and relative property variations, absolute predictions remain sensitive to DFT functional choices, dataset representativeness, and treatment of long‐range interactions. We address a structured validation framework linking model fidelity to experimentally relevant observables and system complexity. Also, we outline emerging directions including generative inverse design, autonomous discovery workflows, and foundation‐model fine‐tuning, highlighting opportunities and limitations for predictive deployment in energy materials research.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Machine Learning in Materials Science · Inorganic Chemistry and Materials · Electrocatalysts for Energy Conversion
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Department of Chemistry Ulsan National Institute of Science and Technology (UNIST) Ulsan Republic of Korea
Department of Energy Science Sungkyunkwan University Suwon Republic of Korea
Department of Energy Sungkyunkwan University Suwon Republic of Korea
Center For 2D Quantum Heterostructures Institute For Basic Science (IBS) Suwon Republic of Korea
Funding
National Research Foundation of Korea
References (100+)
Show source-provided references
1. 10.1021/acs.chemrev.1c0002210.1021/acs.chemrev.1c00022 ↗
2. 10.1021/acs.chemrev.0c0111110.1021/acs.chemrev.0c01111 ↗
3. 10.1038/s41563‐020‐0777‐610.1038/s41563‐020‐0777‐6 ↗
4. 10.1146/annurev‐physchem‐082720‐03425410.1146/annurev‐physchem‐082720‐034254 ↗
5. 10.1063/5.016089810.1063/5.0160898 ↗
6. 10.1002/aenm.20190324210.1002/aenm.201903242 ↗
7. 10.1063/1.496619210.1063/1.4966192 ↗
8. 10.1002/adma.20190276510.1002/adma.201902765 ↗
9. 10.1103/PhysRevX.8.04104810.1103/PhysRevX.8.041048 ↗
10. 10.1038/ncomms1389010.1038/ncomms13890 ↗
11. 10.1103/PhysRevLett.98.14640110.1103/PhysRevLett.98.146401 ↗
12. 10.1103/PhysRevB.87.18411510.1103/PhysRevB.87.184115 ↗
13. 10.1103/PhysRevLett.104.13640310.1103/PhysRevLett.104.136403 ↗
14. 10.1126/sciadv.170181610.1126/sciadv.1701816 ↗
15. 10.1103/PhysRevB.95.09420310.1103/PhysRevB.95.094203 ↗
16. SchNet: a Continuous‐filter Convolutional Neural Network for Modeling Quantum Interactions
17. Neural Message Passing for Quantum Chemistry
18. 10.1021/acs.jctc.9b0018110.1021/acs.jctc.9b00181 ↗
19. 10.1038/s41467‐022‐29939‐510.1038/s41467‐022‐29939‐5 ↗
20. 10.1038/s42256‐024‐00956‐x10.1038/s42256‐024‐00956‐x ↗
21. 10.52202/068431‐083010.52202/068431‐0830 ↗
22. 10.1038/s41467‐023‐36329‐y10.1038/s41467‐023‐36329‐y ↗
23. N.Thomas T.Smidt S.Kearnes et al. “Tensor Field Networks: Rotation‐ and Translation‐equivariant Neural Networks for 3D Point Clouds ”arXiv:1802.08219(2018).
24. 10.1063/5.023126510.1063/5.0231265 ↗
25. 10.1063/5.022674510.1063/5.0226745 ↗
26. 10.1039/D4CP01801J10.1039/D4CP01801J ↗
27. 10.1021/acs.jpclett.1c0160510.1021/acs.jpclett.1c01605 ↗
28. 10.1103/PhysRevB.103.21410210.1103/PhysRevB.103.214102 ↗
29. A.HajibabaeiandK. S.Kim “Towards Universal Sparse Gaussian Process Potentials: Application to Lithium Diffusivity in Superionic Conducting Solid Electrolytes ”arXiv:2009.13179v1103(2020):214102 https://doi.org/10.48550/arXiv.2009.13179.
30. 10.1063/5.029700610.1063/5.0297006 ↗
31. B. M.Wood M.Dzamba X.Fu et al. “UMA: a Family of Universal Models for Atoms ”arXiv:2506.23971(2025).
32. 10.1038/s42256‐023‐00716‐310.1038/s42256‐023‐00716‐3 ↗
33. Learning Smooth and Expressive Interatomic Potentials for Physical Property Prediction
34. 10.1038/s41467‐025‐65662‐710.1038/s41467‐025‐65662‐7 ↗
35. 10.1038/s43588‐022‐00349‐310.1038/s43588‐022‐00349‐3 ↗
36. B.Rhodes S.Vandenhaute V.Šimkus et al. “Orb‐v3: Atomistic Simulation at Scale ”arXiv:2504.06231v2(2025) https://doi.org/10.48550/arXiv.2504.06231.
37. 10.1038/s41586‐023‐06735‐910.1038/s41586‐023‐06735‐9 ↗
38. 10.1038/s41467‐026‐70195‐810.1038/s41467‐026‐70195‐8 ↗
39. 10.1002/aenm.20240387610.1002/aenm.202403876 ↗
40. 10.1021/acs.chemrev.4c0057210.1021/acs.chemrev.4c00572 ↗
41. 10.1039/D5CS00104H10.1039/D5CS00104H ↗
42. 10.1021/acs.jctc.4c0019010.1021/acs.jctc.4c00190 ↗
43. S. Y.Willow T. H.Park G. B.Sim et al. “(3)‐equivariant Interatomic Potential with Iterative Restratification of Many‐body Message Passing ”arXiv:2510.03046(2025) https://doi.org/10.48550/arXiv.2510.03046.
44. 10.1038/s41557‐021‐00716‐z10.1038/s41557‐021‐00716‐z ↗
45. 10.1126/sciadv.160301510.1126/sciadv.1603015 ↗
46. 10.1038/s41524‐024‐01295‐610.1038/s41524‐024‐01295‐6 ↗
47. 10.1016/j.commatsci.2021.11031410.1016/j.commatsci.2021.110314 ↗
48. 10.1038/s41597‐020‐0473‐z10.1038/s41597‐020‐0473‐z ↗
49. 10.5281/zenodo.1010894210.5281/zenodo.10108942 ↗
50. 10.1016/j.mtphys.2024.10156010.1016/j.mtphys.2024.101560 ↗
51. L.Barroso‐Luque M.Shuaibi X.Fu et al. “Open Materials 2024 (OMat24) Inorganic Materials Dataset and Models ”arXiv:2410.12771(2024) https://doi.org/10.48550/arXiv.2410.12771.
52. 10.1021/acscatal.0c0452510.1021/acscatal.0c04525 ↗
53. 10.1021/acscatal.2c0542610.1021/acscatal.2c05426 ↗
54. 10.1038/s41524‐021‐00636‐z10.1038/s41524‐021‐00636‐z ↗
55. 10.1038/s41524‐022‐00696‐910.1038/s41524‐022‐00696‐9 ↗
56. 10.1038/s41598‐023‐46951‐x10.1038/s41598‐023‐46951‐x ↗
57. Fitting to Magnetic Forces Improves the Reliability of Magnetic Moment Tensor Potentials
58. 10.1103/PhysRevB.109.14442610.1103/PhysRevB.109.144426 ↗
59. 10.1103/PhysRevB.110.10442710.1103/PhysRevB.110.104427 ↗
60. K.Ueno S.Ohuchi K.Ichikawa K.Amii andK.Wakasugi “SpinMultiNet: Neural Network Potential Incorporating Spin Degrees of Freedom with Multi‐task Learning ”arXiv:2409.03253(2024).
61. 10.1002/qua.2489010.1002/qua.24890 ↗
62. 10.1063/1.501977910.1063/1.5019779 ↗
63. 10.1103/PhysRevB.105.11520210.1103/PhysRevB.105.115202 ↗
64. 10.1021/acs.accounts.0c0068910.1021/acs.accounts.0c00689 ↗
65. 10.1038/s41524‐025‐01623‐410.1038/s41524‐025‐01623‐4 ↗
66. 10.1038/s41563‐025‐02272‐010.1038/s41563‐025‐02272‐0 ↗
67. 10.1038/sdata.2014.2210.1038/sdata.2014.22 ↗
68. 10.1063/5.008920010.1063/5.0089200 ↗
69. D. S.Levine M.Shuaibi E. W. C.Spotte‐Smith et al. “The Open Molecules 2025 (OMol25) Dataset Evaluations and Models ”arXiv:2505.08762(2025) https://doi.org/10.48550/arXiv.2505.08762.
70. 10.1038/s41597‐026‐06628‐210.1038/s41597‐026‐06628‐2 ↗
71. 10.1021/acs.jcim.7b0008310.1021/acs.jcim.7b00083 ↗
72. 10.1093/nar/gkae105910.1093/nar/gkae1059 ↗
73. 10.1093/nar/gkv95110.1093/nar/gkv951 ↗
74. 10.1063/1.481232310.1063/1.4812323 ↗
75. 10.1021/acs.jcim.5c0126210.1021/acs.jcim.5c01262 ↗
76. C.Gabellini N.Shenoy S.Thaler et al. “OpenQDC: Open Quantum Data Commons ”arXiv:2411.19629(2024) https://doi.org/10.48550/arXiv.2411.19629.
77. D. S.Levine N.Liesen L.Chua et al. “The Open Polymers 2026 (OPoly26) Dataset and Evaluations ”arXiv:2512.23117(2025).
78. A.Sriram L. M.Brabson X.Yu et al. “The Open DAC 2025 Dataset for Sorbent Discovery in Direct Air Capture ”arXiv:2508.03162(2025).
79. 10.1016/j.jcp.2014.12.01810.1016/j.jcp.2014.12.018 ↗
80. 10.1137/15M105418310.1137/15M1054183 ↗
81. 10.1103/PhysRevB.99.01410410.1103/PhysRevB.99.014104 ↗
82. 10.1063/1.355371710.1063/1.3553717 ↗
83. 10.1103/PhysRevLett.120.14300110.1103/PhysRevLett.120.143001 ↗
84. 10.1021/acs.chemmater.9b0129410.1021/acs.chemmater.9b01294 ↗
85. M.Neumann J.Gin B.Rhodes et al. “Orb: a Fast and Scalable Neural Network Potential ”arXiv:2410.22570(2024) https://doi.org/10.48550/arXiv.2410.22570.
86. Rasmussen C. E. · 2006
87. 10.1007/978-3-540-28650-9_410.1007/978-3-540-28650-9_4 ↗
88. 10.1063/5.000508410.1063/5.0005084 ↗
89. 10.1038/s41586‐020‐03072‐z10.1038/s41586‐020‐03072‐z ↗
90. 10.1103/PhysRevB.90.10410810.1103/PhysRevB.90.104108 ↗
91. Distributed Gaussian Processes, in Proc. 32nd Int. Conf. Mach
92. 10.1162/08997660030001490810.1162/089976600300014908 ↗
93. 10.1088/2632‐2153/adfa6810.1088/2632‐2153/adfa68 ↗
94. 10.1038/s42256‐025‐01055‐110.1038/s42256‐025‐01055‐1 ↗
95. 10.1038/s41524‐024‐01500‐610.1038/s41524‐024‐01500‐6 ↗
96. 10.1038/s41524‐025‐01727‐x10.1038/s41524‐025‐01727‐x ↗
97. 10.1038/s41524‐025‐01796‐y10.1038/s41524‐025‐01796‐y ↗
98. 10.1038/s41524‐025‐01895‐w10.1038/s41524‐025‐01895‐w ↗
99. J.Gasteiger J.Groß andS.Günnemann “Directional Message Passing for Molecular Graphs in Int. Conf. Learn. Represent ”arXiv:2003.03123(2020) https://doi.org/10.48550/arXiv.2003.03123.
100. GemNet: Universal Directional Graph Neural Networks for Molecules