Integrating Theory and Experiment with Graph Neural Networks to Classify Molecular Self-Assembly on Metal Surfaces

Key Points

• The research aims to develop a machine learning framework for classifying molecular self-assembly on metal surfaces using graph neural networks.
• Utilized graph neural networks to predict adsorption energy of molecular layers.
• Classified self-assembled layers of previously unseen molecules to test model generalizability.
• Integrated computational models with experimental data for improved analysis.
• Achieved an R-squared value of 0.985 for adsorption energy regression.
• Effectively classified self-assembled layers of three new molecules with high accuracy.
• Established framework supports the rational design of surface-supported functional nanostructures.