QUAM-AFM: Largest Simulated AFM Image Dataset for Molecular Identification

QUAM-AFM is a dataset of 165 million simulated Atomic Force Microscopy images generated from 685,513 organic molecules. The dataset was created by Quasar Science Resources S.L. and the SPMTH Research Group at Universidad Autónoma de Madrid, funded by the Comunidad de Madrid, and was last updated in May 2024. It includes 24 3D image stacks per molecule, each with 10 tip-sample distances, plus molecular depictions, IUPAC names, and atomic coordinates.

Use Cases

• Training deep learning models for chemical identification based on simulated AFM image stacks.
• Benchmarking image analysis algorithms on constant-height AFM images with varied operational parameters.
• Studying the effect of cantilever oscillation amplitude and tip elastic constant on image features.
• Correlating molecular structures (via ball-and-stick depictions and atomic coordinates) with their simulated AFM signatures.

Strengths

• Contains 165 million images with 256x256 pixel resolution, derived from 685,513 molecules.
• Includes 24 different combinations of AFM operational parameters (amplitude and elastic constant) for systematic analysis.
• Provides auxiliary data per molecule: IUPAC name, chemical formula, atomic coordinates, and atom height maps.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Data is simulated, not experimental, which may limit direct applicability to real-world AFM setups.
• Row count is unknown, which may limit suitability assessment for specific modeling tasks.

Provenance

Source

Quasar Science Resources S.L. and the Scanning Probe Microscopy Theory & Nanomechanics Research Group at Universidad Autónoma de Madrid.

Collection Method

Simulated from a selection of 685,513 molecules spanning relevant bonding structures and chemical species.

Time Range

null

Freshness

Last updated 2024-05-05 07:15:29; freshness should be verified.

Geography

null