The microstructures of materials are characterized by defects, which determine the materials performance. To evaluate them, researchers working on atomistic simulations often perform calculations with different codes (e.g., in scale-bridging approaches). In such cases, it is beneficial to reuse existing atomic structures (samples). To aid the reuse of these computational samples, well-described and harmonized metadata and data is crucial, however, most standardization approaches are focused only on perfect crystal structures, and not on defects. The description of defects and application level information of atomic samples is needed. Another challenge is that file formats used for crystallographic structures (which are often specifically tailored to a certain code) have different representations, which can lead to difficulties transitioning between codes and information being dropped. Finally, another issue preventing the reusability of atomic structure samples is the lack of information on the creation workflow or provenance.

We develop an application-level ontology for material science computational samples, CMSO, which initially describes samples on the atomistic level. The use of this ontology is aided by a software tool for automated annotation of structures using available atomic structural codes. Pyscal-rdf provides a way for users of common atomistic structure codes to implement RDF store and SPARQL querying. The use of this controlled vocabulary in a linked open data form ensures interoperability between different structural file formats and software, while also offering the possibility of making data findable and reusable.