A Self-Evolutionary Methodology for Reverse Design of Novel MOFs

Area of Science:

• Materials Science
• Computational Chemistry
• Machine Learning

Background:

• High-throughput simulations and machine learning (ML) are used to find metal-organic frameworks (MOFs) for gas storage.
• Current methods often focus on existing MOF databases, limiting the discovery of novel, high-performance structures.
• Intelligent prediction of new MOF structures with improved properties is needed.

Purpose of the Study:

• To develop an efficient self-evolutionary methodology for discovering novel MOFs with superior energy gas carrier performance.
• To go beyond extracting top performers from existing databases and intelligently predict new MOF structures.
• To showcase the method's capability using methane (CH4) storage in MOFs at room temperature.

Main Methods:

• Proposed an efficient self-evolutionary methodology for searching high-performance MOFs.
• Introduced a Tangent Adaptive Genetic Algorithm (TAGA) for structural evolution.
• Employed an eXtreme Gradient Boosting (XGBoost) machine learning model as the fitness function.
• Utilized a database of 51,163 hypothetical MOFs (hMOFs).

Main Results:

• The TAGA-XGBoost strategy rapidly identified new MOF structures with improved gravimetric and volumetric capacities for CH4 storage.
• Predicted MOFs demonstrated higher storage capacities than the best materials in the original database.
• The best predicted materials achieved 580 cm³(STP)/g at 35 bar and a working capacity of 218 cm³(STP)/cm³ between 65 and 5.8 bar.
• Analysis revealed organic linkers systematically influence MOF storage performance.

Conclusions:

• The proposed self-evolutionary methodology effectively accelerates the discovery of novel MOFs for energy gas storage.
• This approach enables the prediction of unprecedented MOF materials with enhanced performance.
• The methodology offers a pathway for discovering new materials for various practical applications beyond gas storage.