A database of experimental Li ion conductivities with an accurate classification model Matthew S Dyer, 1,3 Cameron J. Hargreaves, 1 Michael W. Gaultois, 1,3 Luke M. Daniels, 1 E. J. Watts, 1,3 V. A. Kurlin, 2, 3 M. Moran, 1,3 Y. Dang, 1 R. Morris, 1 A. Morscher, 1 K. Thompson, 1 M. A. Wright, 1 B.-E. Prasad, 1 F. Blanc, 1,3,4 C. M. Collins, 1 C. A. Crawford, 1 B. B. Duff, 1,4 J. Evans, 1 J. Gamon, 1 G. Han, 1 B. T. Leube, 1 H. Niu, 1 A. J. Perez, 1 A. Robinson, 1 O. Rogan, 1,3 P. M. Sharp, 1 E. Shoko, 1 M. Sonni, 1 W. J. Thomas 1 A. Vasylenko 1 L. Wang 1 M. J. Rosseinsky 1,3 1 Department of Chemistry, University of Liverpool, Liverpool, L69 7ZD, U. K, 2 Department of Computer Science, University of Liverpool, Liverpool, L69 3BX, U. K, 3 Leverhulme Research Centre for Functional Materials Design, Materials Innovation Factory, University of Liverpool, Liverpool, L7 3NY, U. K, 4 Stephenson Institute for Renewable Energy, University of Liverpool, Liverpool, L69 7ZF, U. K. An expertly validated database of the Li ion conductivity in solids has been compiled from literature. The database contains 820 entries of theionic conductivity measured by AC impedance spectroscopy, the temperature of the measurement, the composition of the sample and the primary literature source. 1 The database covers 403 unique chemical compositions and is freely available for non-commercial use at http://pcwww.liv.ac.uk/~msd30/lmds/ LiIonDatabase.html. Each material in the database has been assigned to a structural family,enabling the space of Li ion conductors to be mapped out and clearly identifying clusters of structurally related materials. All entries in the database have been mapped onto the full chemical space of all reported solid-state inorganic compounds. The database can be used as the source of data for supervised machine learning problems, and was used to train a reliable classifier into high or low ionic conductivity based solely on composition with an accuracy of over 80 %. This dataset and the models derived from it represent a practical resource for those working in the solid-state Li ion battery community. More generally, the experience gained in constructing and validating this dataset will be of benefit to others seeking to construct their own dataset of material properties from experimental data reported in the scientific literature. References 1. C. J. Hargreaves et al., npj Comput. Mater. 9 (2023) 9
P159E
Made with FlippingBook Learn more on our blog