Exploring Hydrogen Storage Capacity in Metal-Organic Frameworks: A Bayesian Optimization Approach.

Area of Science:

• Materials Science
• Computational Chemistry
• Artificial Intelligence

Background:

• Metal-organic Frameworks (MOFs) are crucial for gas storage, capture, and sensing.
• High-throughput screening of large MOF databases for optimal adsorption properties is computationally expensive.
• Efficient methods are needed to identify promising MOFs for gas separations and storage applications.

Purpose of the Study:

• To demonstrate the effectiveness of Bayesian optimization (BO) for estimating H₂ uptake in MOFs.
• To significantly reduce the computational cost of screening large MOF databases.
• To provide a framework for optimizing MOF properties and material design insights.

Main Methods:

• Utilized an existing dataset of 98,000 real and hypothetical MOFs.
• Applied Bayesian Optimization (BO) to predict H₂ uptake capability.
• Compared BO with Particle Swarm Optimization (PSO) and a novel Evolutionary-PSO (EPSO) variant.

Main Results:

• BO identified top candidate MOFs by screening less than 0.027% of the database.
• The BO method significantly reduces the experimental effort required for MOF screening.
• PSO and EPSO achieved comparable results to BO in estimating H₂ uptake potential.

Conclusions:

• Bayesian optimization is a highly efficient tool for accelerating the discovery of MOFs for H₂ storage.
• The developed framework offers a transferable approach for optimizing various material properties.
• AI-driven methods like BO and PSO offer powerful alternatives to traditional computational screening for materials science.