Quantify band-gap disagreement among structurally identical records

Determine the magnitude of band-gap disagreement between records from different materials databases that are known to describe the same crystal structure.

Background

The report audits whether records linked across the Materials Project, OQMD, AFLOW, JARVIS-DFT, NOMAD, and COD describe the same material. Composition-based matching is shown to be unreliable, and only 115 of 300 sampled record pairs could be resolved after retrieving supplementary structures. Among the pairs judged to represent the same structure, only two had two band-gap values available, leaving unresolved the extent to which band gaps differ when structural identity—not merely composition or space group—is established. Resolving this issue is necessary for scientifically valid cross-database comparisons.

References

Only two pairs judged to be the same structure had two band-gap values, so the audit cannot yet say how large the disagreement is between records known to describe the same crystal.

— MatQuery: Execution-Grounded Natural-Language Querying over Heterogeneous Materials Databases  (2610.13573 - Zhang et al., 9 Oct 2026) in Section 5.5, “Three-hundred-pair structure audit”