Understanding the superionic phase transition of lithium nitride from machine learning force fields Gabriel Krenzer 1 , Johan Klarbring 1 , Kasper Tolborg 1 , Hugo Rossignol 2 , Andrew R. McCluskey 3 , Benjamin J. Morgan 4 , Aron Walsh 1 1 Department of Materials, Imperial College London, Exhibition Road, London SW7 2AZ, UK, 2 Trinity College Dublin, Dublin, Ireland, 3 European Spallation Source ERIC, P.O. Box 176, SE-221 00, Lund, Sweden, 4 Department of Chemistry, University of Bath, Claverton Down, Bath BA2 7AY, UK Superionic conductors have promising applications as solid-state electrolytes. Atomistic mechanisms underlying ion transportin the superionic regime, as well as the physical parameters leading some materials to exhibit a Faraday-type II superionic transition before melting, are, however, poorly understood. A better understanding of the fundamental physics of both the superionic regime and the superionic phase transition is necessary for developing superionic solid electrolytes for practical applications. Several mechanisms have been proposed to explain superionic phase transitions: order-disorder transitions [1, 2, 3, 4] , competition between local atomic preferences usually referred to as frustration [5] , and overpopulation of existing lattice sites sometimes referred to as jamming [6, 7]. In addition, previous experimental studies have shown dramatic changes in phonon behaviour across the superionic transition of specific compounds. In CuCrSe2 [3] , γ-LiAlO2 [4] , and AgCrSe2 [8] , the superionic transition is accompanied by a breakdown of specific phonon modes. Here, we use machine learning molecular dynamics simulations to study the superionic phase transition of Li 3 N, an archetypal superionic conductor. The simulations allow to compute diffusive properties across a wide range of temperatures and provide quantitative characterisation of the superionic phase transition. The simulations provide insights into the relationship between anharmonic phononsand ion transport. The simulations also reveal the diffusion mechanism, how non-Arrhenius diffusion occurs synchronously with non-Arrhenius defect formation, and how defect concentration lead to distinct diffusion correlations. The transport mechanisms studied across the superionic phase transition of Li3N and the direct link between diffusion activation energy and defect formation energy proposed here should be relevant to a broad range of superionic materials. References 1. Eapen, J. & Anamareddy, A. Entropic crossoversin superionic fluorites from specific heat. Ionics 23,1043–1047 (2017). 2. Wang, J., Ding, J., Delaire, O. & Arya, G. Atomisticmechanisms underlying non-arrhenius ion transport 3. in superionic conductor AgCrSe 2 . ACS Applied Energy Materials (2021). 4. Niedziela, J. L. et al. Selective breakdown ofphonon quasiparticles across superionic transition inCuCrSe 2 . Nature Physics 15, 73–78 (2019). 5. Hu, Q. et al. Observation of specific optical phononmodes dominating Li ion diffusion in γ-LiAlO 2 ceramic. Ceramics International 47, 17980–17985(2021). 6. Wood, B. C. et al. Paradigms of frustration in superionic solid electrolytes. Philosophical Transactions 7. of the Royal Society A 379, 20190467 (2021). 8. He, X., Zhu, Y. & Mo, Y. Origin of fast ion diffusion in super-ionic conductors. Nature Communications 8, 1–7 (2017). 9. Annamareddy, A. & Eapen, J. Low dimensionalstring-like relaxation underpins superionic conduction in fluorites and related structures. Scientific reports 7, 1–12 (2017). 10. Ding, J. et al. Anharmonic lattice dynamics andsuperionic transition in AgCrSe 2 . Proceedings ofthe National Academy of Sciences 117, 3930–3937(2020).
P185E
Made with FlippingBook Learn more on our blog