Machine Learning Potential Data for PFAS Diffusion in Water and Adsorption on Graphene

A 5.2 MB ZIP file contains data for a machine learning potential (MLP) trained on ab initio data for a representative PFAS molecule in aqueous solution. The dataset, authored by Romain Dupuis and last updated on 2026-05-11, compares the MLP against established nonreactive and reactive (ReaxFF) force fields for structural, vibrational, and dynamical properties. It also includes a preliminary extension illustrating the MLP's applicability to PFAS adsorption on graphene.

Use Cases

• Benchmarking machine learning potentials against traditional force fields based on the comparative analysis described.
• Simulating PFAS diffusion dynamics in water based on the MLP's reported agreement with experimental data.
• Studying PFAS adsorption on carbon-based nanomaterials like graphene based on the preliminary extension mentioned.
• Investigating the trade-offs between accuracy and computational efficiency in molecular simulations as highlighted in the description.

Strengths

• Dataset size is 5.2 MB, indicating a focused and manageable download.
• The MLP is reported to yield diffusion coefficients in good agreement with experimental data.
• The study critically compares the MLP against established ReaxFF and nonreactive force fields.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset's primary focus is on a single representative PFAS molecule, though transferability is noted.

Provenance

Source

figshare

Collection Method

Data generated from ab initio calculations and machine learning potential training.

Freshness

Last updated 2026-05-11 19:33:26; freshness should be verified.