Executive Summary : | The preparation of a database for cathodes in metal ion batteries will involve searching for battery-related descriptors, screening ion conducting materials based on their ionic conductivity, thermodynamic phase stability, high electronic conduction, and electrochemical performance. Composition-based features, such as atomic mass, number of valence electrons, and bond-valence sums, will be used as a first step in fingerprinting. The least absolute shrinkage and selection operator (LAssO) will be used to create top features and non-redundant features. After successful construction of predictive models, unknown combinations will be tested individually to obtain AV and GC. The testing datasets will include rare earth elements, mobile ions, oxygen, H-free, and free from toxic elements. Polycrystalline samples of predicted candidate materials will be synthesized in the laboratory, and X-ray diffraction techniques will be used to identify crystallographic phases. Cindle type cells will be fabricated in an Ar filled glove box, and cell performance will be measured in constant-current-constant-voltage mode using a battery tester. Electrochemical measurements will be performed using a potentiostat. The average voltage and gravimetric capacity will be verified using experimental data as predicted by the machine learning model. In addition to discovering new cathodes, structural, electronic, and electrochemical properties will also be explored. |