Inverse Design of Metal-Organic Frameworks for CH₄/N₂ Separation Enabled by Coupled Machine Learning and Genetic...

Area of Science:

• Materials Science
• Chemical Engineering
• Computational Chemistry

Background:

• Metal-organic frameworks (MOFs) show great potential for gas separation applications.
• Experimental screening of MOFs is limited by time and resource constraints.
• Machine learning (ML) offers a data-driven approach to accelerate materials discovery.

Purpose of the Study:

• To develop a machine learning model integrated with a Tangent Adaptive Genetic Algorithm (TAGA) for the inverse design of MOFs.
• To identify MOF structures with high selectivity for methane/nitrogen (CH₄/N₂) separation.
• To establish a generalizable pathway for designing next-generation separation materials.

Main Methods:

• An accurate ML model was trained on MOF structural features (topology, building units, functional groups).
• The ML model was embedded within the TAGA framework to navigate the chemical space for inverse design.
• High-performance MOF genotypes were identified through evolutionary trajectory analysis.

Main Results:

• MOFs with fsc topology and pyrene, anthracene, or naphthalene ligands demonstrated superior CH₄/N₂ selectivity.
• The top-performing designed MOF achieved an IAST selectivity of 15.92 and a CH₄ uptake of 2.47 mmol g^-1.
• The study successfully predicted high-performance MOF structures for targeted gas separation.

Conclusions:

• This work demonstrates a paradigm shift from trial-and-error to goal-directed materials design using ML.
• The integrated ML-TAGA approach efficiently identifies optimal MOF structures for specific applications.
• This methodology provides a scalable pathway for developing advanced MOFs for gas separation and other fields.