ChemDataExtractor

ChemDataExtractor extracts UV/vis absorption spectroscopy data and associated optical properties from scientific literature for integration with computational electronic-structure predictions.

Key Features:

• Automated Database Generation: Generates a database containing 18,309 records of experimentally determined UV/vis absorption maxima (λmax) and extinction coefficients (ϵ).
• High-throughput Document Mining: Processes 402,034 scientific documents to compile optical-property data at scale.
• Refined Table Extraction Mechanics: Employs refined table extraction mechanics to improve accuracy in retrieving λmax and ϵ from literature tables.
• Integration with Electronic-Structure Calculations: Facilitates high-throughput electronic-structure calculations using the simplified Tamm-Dancoff approach and time-dependent density functional theory to predict λmax and oscillation strengths (f).
• Data Correlation: Produces strong correlations between experimental and computational λmax, f, and ϵ to support validation of in silico predictions.
• Dataset Composition: Includes 8,488 unique compounds, with 5,380 compounds featuring paired experimental and computational data.
• Output Formats: Provides datasets in MongoDB, CSV, and JSON formats.

Scientific Applications:

• Optoelectronic Materials Discovery: Supplies an integrated experimental and computational dataset (8,488 compounds; 5,380 paired) for screening and discovery of optoelectronic materials.
• Computational Validation of Optical Properties: Enables validation of time-dependent density functional theory and simplified Tamm-Dancoff predictions against experimental λmax, f, and ϵ.

Methodology:

Refined table extraction mechanics extract λmax and ϵ from literature and automated processing of 402,034 documents generates the database; high-throughput electronic-structure calculations use the simplified Tamm-Dancoff approach and time-dependent density functional theory to predict λmax and oscillation strengths (f).