DFT-computed Adiabatic Redox Table (DART) Dataset

This repository contains the data release for the DFT-computed Adiabatic Redox Table (DART) dataset. DART is a medium-sized benchmark dataset designed to combine broad chemical diversity with a practical size for method development and comparative evaluation. The dataset was constructed from 44,090 commercially available compounds. Initial structures were generated with RDKit, and the lowest-energy conformer selected for each compound was used for quantum-chemical calculations. Neutral, radical-cation, and radical-anion structures were optimized with ORCA 5.0.1 using the B97-3c composite DFT method. Structures with failed optimizations or imaginary frequencies were discarded, resulting in 26,156 molecules in the final dataset.