Ensemble-Learning-Guided Optimization Design for Metal-Organic Framework Adsorbents toward CO Adsorption.

Area of Science:

• Materials Science
• Chemical Engineering
• Computational Chemistry

Background:

• Metal-organic frameworks (MOFs) show promise for carbon monoxide (CO) adsorption due to their tunable structures and high surface areas.
• Developing efficient MOF adsorbents typically requires extensive experimental screening, which is time-consuming and resource-intensive.

Purpose of the Study:

• To develop a machine learning strategy for designing high-performance MOFs for CO adsorption.
• To identify key structural and synthesis parameters influencing CO adsorption capacity in MOFs.

Main Methods:

• An ensemble-learning strategy integrating multidimensional feature analysis was employed.
• An extreme gradient boosting model was utilized for predictive modeling, achieving high accuracy (R² > 0.95) with limited data.
• Analysis focused on pore geometry, structural properties, and synthesis conditions of MOFs.

Main Results:

• Porous characteristics were identified as the dominant factor for CO adsorption in pristine MOFs.
• Optimal MOFs for CO adsorption possess one-dimensional, two-dimensional, microporous, or isolated pores with a total pore volume of 0.4-0.6 cm³/g.
• Key structural parameters influencing adsorption were ranked: space groups > geometry > topology, with R3m space group, binuclear paddle wheel geometry, and scorpionate-like topology being optimal.
• For transition metal-modified MOFs, Cu(I) exhibited the strongest CO binding affinity, while Fe(II) and Ni(II) also served as effective binding sites.

Conclusions:

• The study provides a theoretical framework for designing efficient MOF-based adsorbents for CO capture.
• Machine learning accelerates the discovery of optimal MOF structures, reducing experimental efforts.
• Tailoring pore characteristics and incorporating specific transition metals are crucial for enhancing CO adsorption performance in MOFs.