Auto-Generating Databases about Batteries and Solar-Cells for Data-driven Materials Discovery

This talk will introduce the ‘chemistry aware’ natural processing tool, ChemDataExtractor,[1,2] and illustrate its ability to auto-generate databases on battery and photovoltaic materials and device information. The battery database contains a total of ¼ million data records and includes chemical names of anodes, cathodes and electrolytes as well as five device properties.[3,4] The photovoltaics database contains ½ million data records and includes chemical names and device properties of perovskite-based and dye-sensitized solar cells.[5] The database auto-generation methods are discussed as is the applicability of these large databases to machine-learning pipelines which enable a design-to-device approach to data-driven materials discovery.[6]<br/><br/><b>References</b><br/>[1] Swain and Cole, <i>J. Chem. Inf. Model</i>. 2016, 56, 10, 1894–1904 www.chemdataextractor.org<br/>[2] Mavracic, Court, Isazawa, Elliott, Cole, <i>J. Chem. Inf. Model</i>. 2021, 61, 9, 4280–4289 www.chemdataextractor2.org<br/>[3] Huang, Cole, <i>Sci. Data</i> (Springer Nature), 2020, 7, 260 https://doi.org/10.1038/s41597-020-00602-2<br/>[4] Huang, Cole, J. Chem. Inf. Model. (2022) DOI: https://doi.org/10.1021/acs.jcim.2c00035<br/>[5] Beard, Cole, <i>Sci. Data</i> (Springer Nature), 2022, DOI: 10.1038/s41597-022-01355-w<br/>[6] Cole, <i>Acc. Chem. Res</i>. 2020, 53, 599-610.