MC3D: The Materials Cloud FAIR and full-provenance first-principles database of experimentally known inorganics

Highly curated databases of materials and their properties have become an increasingly precious resource for several applications, from property prediction with machine learning methods to material screening through state-of-the-art first-principles simulations. Here, we present the MC3D database of first-principles properties for experimentally known inorganics, obtained starting from structures extracted from three major (and mostly, but not only, experimental) databases: the Pauling File, the Inorganic Crystal Structure Database, and the Crystallography Open Database. Following a curation process, the resulting 72,609 unique stoichiometric compounds are refined using density-functional theory calculations at the PBEsol level, as implemented in Quantum ESPRESSO complemented with the SIRIUS library for efficient use of GPU hardware. The AiiDA materials informatics infrastructure (http://aiida.net) manages all workflow stages, ensuring complete traceability and preserving the full provenance of simulations, with the results openly available on the Materials Cloud (https://mc3d.materialscloud.org). We discuss the protocols, verification steps, and AiiDA's automation and error-handling features, which lead to validated and robust workflows for electronic structure simulations, along with current applications of the database.