Machine learning guided tuning charge distribution by composition in MOFs for oxygen evolution reaction.

Area of Science:

• Materials Science
• Chemical Engineering
• Computational Chemistry

Background:

• Traditional synthesis of metal-organic frameworks (MOFs) is often inefficient, requiring extensive trial-and-error.
• Accelerating MOF discovery is crucial for developing advanced materials for various applications.

Purpose of the Study:

• To leverage machine learning (ML) for rapid and efficient synthesis of novel MOFs.
• To establish a predictive model for MOF design based on experimental parameters.

Main Methods:

• A comprehensive library of over 900 MOFs was created, varying metal salts, solvent ratios, reaction times, and temperatures.
• Zeta potentials were used as target variables for training four distinct ML models.
• Random Forest Regression (RFR) and Gradient Boosting Regression (GBR) models were evaluated for their predictive accuracy.

Main Results:

• RFR and GBR models demonstrated strong correlations and accurate predictions for MOF properties.
• Experimentally synthesized MOFs based on ML predictions closely matched the designed parameters.
• The synthesized MOFs exhibited superior electrocatalytic activity for oxygen evolution reactions.

Conclusions:

• Machine learning significantly accelerates the design and synthesis of MOFs.
• This ML-driven approach enables the discovery of MOFs with enhanced performance for applications like electrocatalysis.
• The methodology has broad implications for accelerating materials discovery across diverse scientific fields.