Experimental vs Computational Data

[Sorahawk]

Hi, I’m starting work on a project to populate an internal database with existing published material property data retrieved from various open sources online.
I’m currently trying to understand Materials Project better so I can start to utilise the API.

My question is, how can I determine whether data for a certain material is experimentally obtained, vs those that are computationally predicted by MP?
I understand that there is some kind of theoretical tag which applies to materials which do not have a corresponding experimental entry from ICSD, but does this apply for all materials in MP?
In other words, is it possible to have an experimental material entry that does not come from ICSD?

Thanks for the help!

[peterschindler]

How is the 'theoretical' tag determined?

Hi @KyuJung_Jun The theoretical tag comes from the ICSD database. If any of the matching ICSD structures is tagged as experimental, then we tag our material as theoretical: False. Right now, we don’t match to partial occupancy ICSD, although that is in the works. We also plan on adding more databases such as TCOD and PCOD, as well as the Pauling Files. For ordered structures, we use the structure matcher on the final MP structure as well as the initial structures to determine if they match t…

Does this previous post answer your question?

[dwinston]

My question is, how can I determine whether data for a certain material is experimentally obtained, vs those that are computationally predicted by MP?

Most of the data served by MP’s API is computationally predicted by MP. Tagging a material document with a theoretical value of False means only that the structure* labeled as representative of that material is the “same” structure – to within a set of tolerances – as an experimentally obtained structure.

Examples of experimentally obtained data on MP are: the thermodynamical data available from the “Thermo” app (and https://www.materialsproject.org/rest/v2/materials/{formula}/exp API endpoint), the ion energies used to construct Pourbaix diagrams, and reference enthalpies of formation used to compare to computed data in the Reaction Calculator. Some of the curated datasets at https://portal.mpcontribs.org/ (parts of which may appear on material detail pages) were experimentally obtained.

*this (final) structure was computationally predicted (“relaxed”) using an initial structure – perhaps one obtained from ICSD – as input.