AQ-GNoME: Computed Properties for Materials Discovery

A collection of computed properties for materials from the GNoME database, released in 2026. It includes disorder probability predictions, DFT validation calculations, and Pourbaix electrochemical stability data. The dataset is provided by the Technical University of Denmark and requires citation of specific source publications.

Use Cases

• Training ML models for predicting disorder probability in crystal structures.
• Validating DFT calculations against known materials from the GNoME and Materials Project databases.
• Screening materials for electrochemical stability using computed Pourbaix data.
• Assessing material scarcity using Herfindahl-Hirschman Index (HHI) scores.

Strengths

• Integrates data from two major sources: the GNoME database and the Materials Project.
• Includes multiple computed property types: disorder probabilities, DFT validation, and electrochemical stability.
• Provides clear licensing and citation requirements for each component.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• The primary data file is restricted to non-commercial use (CC BY-NC 4.0).

Provenance

Source

Technical University of Denmark, derived from the GNoME database (Google DeepMind) and the Materials Project.

Collection Method

Computational predictions and calculations (e.g., DFT, disorder probability models) applied to existing crystal structure databases.

Freshness

Last updated 2026-05-19 10:12:05; freshness should be verified.