MOF Synthesis Prediction Enabled by Automatic Data Mining and Machine Learning.

Area of Science:

• Materials Science
• Computational Chemistry
• Crystallography

Background:

• Metal-organic frameworks (MOFs) are advanced porous materials with diverse applications.
• Synthesizing MOFs often requires extensive experimentation and optimization of parameters.
• Predictive models for MOF synthesis are crucial for efficient materials discovery.

Purpose of the Study:

• To develop and validate machine learning (ML) models for predicting MOF synthesis conditions.
• To accelerate the discovery and design of novel MOFs through computational prediction.
• To establish a comprehensive MOF synthesis database from scientific literature.

Main Methods:

• Automated extraction of synthesis parameters from published MOF literature.
• Creation of the first large-scale database of MOF synthesis conditions.
• Training and optimization of ML models using the curated MOF synthesis database.
• Prediction of synthesis conditions for new MOF crystal structures.

Main Results:

• The developed ML models demonstrate significant predictive accuracy for MOF synthesis parameters.
• The ML approach outperforms predictions made by human experts in a comparative study.
• The models provide a foundation for rationalizing and accelerating MOF synthesis.

Conclusions:

• Machine learning offers a powerful tool for predicting MOF synthesis, streamlining materials discovery.
• Automated data extraction and ML model training are effective for building predictive synthesis tools.
• The developed web-tool provides accessible automated synthesis prediction for researchers.