Big-Data Science in Porous Materials: Materials Genomics and Machine Learning.

Area of Science:

• Materials Science
• Computational Chemistry
• Data Science

Background:

• Millions of potential metal-organic frameworks (MOFs) can be synthesized by combining metal nodes with organic linkers.
• The sheer volume of possible MOFs presents challenges for conventional experimental screening methods.
• Big-data approaches are emerging as a powerful tool for materials discovery and analysis.

Purpose of the Study:

• To review the principles of big-data science and its application to metal-organic frameworks.
• To survey machine learning approaches for representing MOFs in feature space and evaluating their properties.
• To highlight the application of machine learning in MOF research, including gas storage, separation, stability, electronic properties, and synthesis.

Main Methods:

• Introduction to big-data science principles, including training set selection, feature representation, learning architectures, and evaluation strategies.
• Review of machine learning applications in porous materials research.
• Discussion of specific applications in gas storage and separation, material stability, electronic properties, and synthesis.

Main Results:

• Big-data methods enable the study of vast material spaces, facilitating the discovery of complex correlations.
• Machine learning techniques provide effective strategies for representing and analyzing MOFs.
• Machine learning has been successfully applied to various MOF properties and applications, demonstrating its utility in accelerating materials discovery.

Conclusions:

• The vast number of potential MOFs necessitates advanced computational approaches like big-data and machine learning.
• Machine learning offers powerful tools for understanding and predicting MOF behavior, accelerating research and development.
• The application of machine learning in MOF science is rapidly expanding, promising significant future advancements.