RMG Database for Chemical Property Prediction.

Area of Science:

• Chemical Engineering
• Computational Chemistry
• Physical Chemistry

Background:

• Accurate prediction of chemical properties is crucial for developing kinetic models.
• Existing methods often lack comprehensive datasets and robust estimation tools for diverse chemical systems.
• The Reaction Mechanism Generator (RMG) database aims to address these limitations.

Purpose of the Study:

• To present the Reaction Mechanism Generator (RMG) database, a comprehensive resource for chemical property prediction.
• To detail the curated datasets and estimation methods for thermodynamics, kinetics, solvation, and transport properties.
• To facilitate the construction and analysis of chemical kinetic mechanisms.

Main Methods:

• The RMG database integrates curated datasets and predictive models for thermochemistry, kinetics, and transport properties.
• Thermochemistry prediction utilizes group additivity schemes, corrections for various effects (e.g., radical, polycyclic), and a graph convolutional neural network trained on DFT calculations.
• Kinetics estimation employs rate rule schemes trained on extensive curated reaction data.

Main Results:

• The database includes extensive libraries for thermochemical parameters (4564 entries) and kinetic parameters (21,000 reactions).
• It incorporates advanced correction schemes for solvent-solute effects and group additivity models for property prediction.
• A graph convolutional neural network enhances thermochemistry prediction accuracy.

Conclusions:

• The RMG database provides kineticists with readily accessible estimates for critical parameters.
• It significantly speeds up kinetic analysis by enabling hypothesis testing and model construction.
• The database serves as a valuable tool for validating kinetic parameters from other sources.