Dec 3, 2024
8:00pm - 10:00pm
Hynes, Level 1, Hall A
Michael Häfner1,2,Matteo Bianchini1,2
Universität Bayreuth1,Bayerisches Zentrum für Batterietechnik2
NaAlCl
4 is an established solid electrolyte in high-temperature Na-based battery systems,[1] but its ionic conductivity is not sufficient for room-temperature applications.[2] To address this shortcoming, we explored the substitution of various elements into the structure of NaAlCl
4 with density functional theory and machine learning methods to identify and evaluate stable substitution options.
The computationally demanding vibrational contributions to the most promising substitutions were obtained with the assistance of on-the-fly machine-learned potentials, which expedite the phonon calculations by at least one order of magnitude at a minor error of 0.7±1 meV/atom at room temperature.
Isovalent substitutions were found to be most favorable, with potassium and silver being promising choices as substitutes for sodium and gallium for aluminum, yielding a stability below 4 meV/atom compared to their respective ternary chlorides. Evaluations of the configurational space of these substitutions indicate large disorder, with the exchange of Al with Ga resulting in a full solid solution.
Substitutions with ions of differing charges usually did not result in stable structures, with one notable exception for the ion pair Al
3+ and Zn
2+. Based on the lattice of Na
2ZnCl
4, a stable structure type with separate layers for the differently-charged cations and a significant amount of Na vacancies was found.
The ionic conductivity of the substituted structure was evaluated using MD simulations assisted by the general-purpose machine learning model MACE-MP-0,[3] and the substitution of Al
3+ and vacancies into Na
2ZnCl
4 was found to significantly enhance the mobility of the Na
+ ions in the structure.
In conclusion, our investigation may assist the fast, reliable discovery and evaluation of novel fast Na conductors and other materials by inclusion of thermodynamic properties and machine learning.
References:
[1] S. Hikari, “ZEBRA Batteries”, Springer New York, New York (2014), pp. 2165-2169
[2] J. Park, J. P. Son, et al.
ACS Energy Lett., 7, pp. 3293-3301 (2022)
[3] I. Batatia, P. Benner, et al.
arXiv:2401.00096