Identifying porous cage subsets in the Cambridge Structural Database using topological data analysis.

Area of Science:

• Materials Science
• Computational Chemistry
• Data Science

Background:

• Metal-organic cages (MOCs) and organic cages (OCs) are rationally designable materials with growing synthesis numbers.
• Identifying and cataloging existing MOCs and OCs within large databases like the Cambridge Structural Database (CSD) is crucial for future research.
• The CSD is a key repository for crystalline structures, but efficient methods for extracting specific molecular architectures like cages are needed.

Purpose of the Study:

• To develop and present a novel cage mining methodology for identifying MOCs and OCs in the CSD.
• To establish comprehensive datasets of MOCs and OCs from experimental data.
• To demonstrate the utility of these datasets through high-throughput screening for specific applications.

Main Methods:

• A cage mining methodology integrating topological data analysis with supervised and unsupervised machine learning techniques was employed.
• The methodology was applied to the Cambridge Structural Database to extract and classify MOC and OC structures.
• The derived datasets were utilized for high-throughput screening of MOCs and OCs.

Main Results:

• The study successfully derived the first comprehensive dataset of 1839 metal-organic cages (MOCs).
• The largest experimental dataset of 7736 organic cages (OCs) was compiled as of March 2022.
• High-throughput screening identified promising MOCs and OCs for xenon/krypton separation.

Conclusions:

• The developed cage mining methodology provides an effective means to identify and catalog MOCs and OCs.
• The created MOC and OC datasets represent significant resources for materials discovery and design.
• These datasets enable rapid screening for applications such as xenon/krypton separation, highlighting their practical value.