Skip to main content

BenchmarkSet1500: High-Accuracy Excited-State Reference Benchmark Dataset for Organic Semiconductors

The Benchmark1500 dataset is the first curated collection of organic molecules with high-level excited-state calculations at this scale, developed to enable reproducible computational chemistry and data-driven discovery. It integrates multiple levels of electronic structure theory (TD-DFT, CASSCF, NEVPT2), allowing direct, systematic comparison between methods.

Each entry includes molecular identifiers alongside consistently computed electronic properties. The dataset is provided both as a searchable database and as a structured CSV file, enabling immediate integration into machine learning pipelines and computational workflows.

Applications​

  • Machine learning: High-quality predictive and generative models can be trained using this large-scale organic semiconductor dataset computed at high accuracy
  • Optoelectronic screening: The dataset can be screened for a range of applications, including thermally activated delayed fluorescence and inverted singlet–triplet gaps relevant to organic light-emitting diodes
  • Structure–property relationships: Structure–property relationships can be systematically derived from the dataset

What to do next​

Related links:

About this page

If you would like to contribute content to the PSDI Knowledge Base or have feedback you would like to give on this guidance, please contact us.