Studying surface degradation of charged Ni-based Li-ion battery cathodes using machine-learning interatomic potentials

Kurzfassung

Nickel-rich cathode materials are industry-leading high-performance materials for Li-ion batteries as they offer high capacity and energy density [1,2]. However, they are known to suffer from structural instability especially at high state of charge [3]. Structural transformations from a layered structure to spinel-/ and or rocksalt-like phases have been observed experimentally in post-mortem studies [4,5]. This process is accompanied by the loss of lattice oxygen [3,5,6]. While overwhelming experimental evidence is available regarding the process, exact atomistic pathways of these transformations are still under debate. While first-principle simulations can help to understand this degradation process on the atomic scale, they are limited by the computational cost of sampling rare reactive events at the appropriate level of theory. To bridge this gap, machine-learning interatomic potentials (MLIP) have become of tremendous interest in materials science [7,8]. In this contribution, we have studied surface degradation pathways of charged NiO2 surfaces using CHGNet as a charge-informed MLIP. We investigated various crystal facets of NiO2 by combining a finetuned MLIP and density-functional-theory (DFT) at the meta-generalized gradient approximation (metaGGA) level. We discuss surface energetics for different terminations and show novel surface degradation pathways obtained by MLIP molecular dynamics. The sampled structures are subsequently verified in terms of energetics using high accuracy DFT. We further discuss oxygen vacancy formation on these new surfaces and how oxygen vacancy defects can trigger follow-up Ni migration as an initial step towards thermodynamically stable end products. Our work showcases how DFT calculations can be efficiently extended with machine learning to study surface degradation in battery materials. Our study reveals novel surface degradation pathways in Nickelbased cathode materials as a first step towards a better understanding of the complex cathode material degradation process. References: [1] Volta Foundation. The Battery Report 2024. 25 January 2025. [2] Li, W., Erickson, E.M. & Manthiram, A. High-nickel layered oxide cathodes for lithium-based automotive batteries. Nat Energy 5, 26–34 (2020) [3] Oswald, S. & Gasteiger, H. A. The Structural Stability Limit of Layered Lithium Transition Metal Oxides Due to Oxygen Release at High State of Charge and Its Dependence on the Nickel Content. J. Electrochem. Soc. 170, 030506 (2023) [4] Lin, F., Markus, I., Nordlund, D. et al. Surface reconstruction and chemical evolution of stoichiometric layered cathode materials for lithium-ion batteries. Nat Commun 5, 3529 (2014) [5] Jiang, M., Danilov, D. L., Eichel, R.-A. et al., A Review of Degradation Mechanisms and Recent Achievements for Ni-Rich Cathode-Based Li-Ion Batteries. Adv. Energy Mater. 11, 2103005 (2021) [6] Jung, R., Metzger, M., Maglia, F. et al. Oxygen Release and Its Effect on the Cycling Stability of LiNixMnyCozO2 (NMC) Cathode Materials for Li-Ion Batteries. J. Electrochem. Soc. 164 (7), A1361- A1377 (2017) [7] Deng, B., Zhong, P., Jun, K. et al. CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling. Nat Mach Intell 5, 1031–1041 (2023) [8] Jacobs, R., Morgan, D., Attarian, S. et al. A practical guide to machine learning interatomic potentials – Status and future. Current Opinion in Solid State and Materials Science 35, 101214 (2025)