ReSolved Dataset (Reduction in Solvents)

The ReSolved dataset is a curated benchmark designed to support machine learning and computational chemistry efforts focused on predicting redox behavior of organic molecules in solution. It provides DFT-computed electron affinities (EA), absolute one-electron reduction potentials (E_red) and DFT-optimised geometries for nearly 20k molecules across five solvents.

Dataset Overview

• Total molecules: 19,785
• Solvents:
  • Acetonitrile (ACN)
  • Water (H₂O)
  • Tetrahydrofuran (THF)
  • Dimethyl sulfoxide (DMSO)
  • Dimethylformamide (DMF)
• Properties included:
  • Electron affinity (EA), eV
  • Absolute one-electron reduction potential (E_red) in a given solvent at 25°C, eV

Source and Construction

Molecules in ReSolved were selected from:

• OMEAD (Organic Materials for Energy Applications Database) (doi.org/10.1016/j.ensm.2021.10.029)
• REDOX dataset(doi.org/10.1063/5.0151122), comprising:
  • Organic radicals (e.g., nitroxyl, phenoxyl, galvinoxyl)
  • Carbonyl compounds (quinones, carboxylates, phenazine-derived radicals)
  • Cyanides

All properties were computed at the DFT at the M06-2X/def2-TZVPD level of theory for the optimised geometries, frequencies at 298K, and electronic energies, and SMD/M06-2X/cc-pVTZ for the Gibbs free energies of solvation (van der Waals model of atomic radii).

Citation

If you use this dataset in your research, please cite the associated publication: TBA