Dataset of theoretical multinary perovskite oxides

Area of Science:

• Materials Science
• Computational Chemistry
• Solid-State Chemistry

Background:

• Perovskite oxides (ABO3) are vital for catalysis, fuel cells, and membranes.
• The vast chemical space of multinary perovskites remains underexplored.
• Computational limitations hinder the discovery of new perovskite compositions.

Purpose of the Study:

• To generate and curate a comprehensive dataset of theoretical multinary oxide and perovskite structures.
• To expand the computed chemical space for perovskite oxide discovery.
• To provide a resource for materials discovery and machine learning model training.

Main Methods:

• Generated 69,407 multinary oxide compositions using the Structure Prediction and Diagnostic Software (SPuDS) program.
• Optimized structures using density functional theory (DFT) with Materials Project (MP) compatible parameters.
• Calculated formation (ΔHf) and decomposition (ΔHd) enthalpies relative to established references.

Main Results:

• Reported a dataset of 66,516 theoretical multinary oxides, with 59,708 identified as perovskites.
• The dataset includes optimized structures and computed thermodynamic properties (enthalpies).
• This work significantly increases the number of computed multinary perovskite compounds.

Conclusions:

• The generated dataset provides a valuable resource for mining novel perovskite materials.
• The dataset facilitates the training of machine learning models for accelerated materials discovery.
• The methodology allows for systematic expansion of the computed perovskite structure database.