Machine Learning Reveals In-Cavity Versus Surface Activity for Selective C─H Borylation by Metal-Organic Framework...

Area of Science:

• Catalysis
• Materials Science
• Computational Chemistry

Background:

• Metal-organic frameworks (MOFs) are versatile platforms for heterogeneous catalysis.
• Distinguishing between pore-confined and surface catalysis in MOFs is challenging.
• Understanding structure-activity relationships is crucial for designing efficient MOF catalysts.

Purpose of the Study:

• To elucidate structure-activity relationships in MOF-supported nickel (Ni) catalysts for selective C-H borylation.
• To identify critical factors governing sp3 versus sp2 C-H borylation selectivity.
• To develop a systematic framework for rational MOF catalyst design.

Main Methods:

• Interpretable machine learning applied to over 470,000 MOF structures.
• Development of 45 chemically meaningful descriptors for MOF structures.
• Analysis of distinct activation mechanisms (HAT vs. CMD) for sp3 and sp2 borylation.

Main Results:

• Identified key structural descriptors influencing MOF catalyst selectivity.
• Revealed distinct mechanisms for sp3 (radical HAT in pores) and sp2 (CMD on surfaces/defects) borylation.
• Achieved high selectivity for sp3 (up to 97.8%) and sp2 (up to 88.7%) borylation using designed Ni catalysts.

Conclusions:

• Established generalizable principles for controlling activity preference in MOF-supported catalysis.
• Demonstrated the power of interpretable machine learning in catalyst design.
• Provided a systematic framework for rational design of MOF catalysts for selective C-H functionalization.