CsPb(Cl/Br)3 DFT Dataset: 10,000 Algorithmically Generated Perovskite Structures

10,000 algorithmically generated CsPb(Cl/Br)3 perovskite alloy structures with single-point DFT calculations across four space groups. The data was created by Jarno Laakso of Aalto University to train a machine learning model for predicting material stability. It includes four subsets covering single-point energies, extended concentration tails, relaxation snapshots, and pure endpoint structures.

Use Cases

• Train machine learning models to predict formation energy based on structural features and composition.
• Study the stability landscape of mixed-halide perovskites across different crystal symmetries.
• Benchmark DFT relaxation algorithms using the provided snapshots with atomic forces.
• Analyze the effect of chlorine/bromine concentration and octahedral tilting on material properties.

Strengths

• Contains at least 100 atomic structures per Cl/Br concentration level for all four lattice types in the combined single-point datasets.
• Includes 200 DFT relaxation snapshots with atomic forces for each structure.
• Covers four distinct space groups (Pm-3m, P4/mbm, I4/mcm, Pnma) providing structural diversity.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the full collection is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

Aalto University

Collection Method

Density functional theory (DFT) calculations on algorithmically generated structures.