Determination of bonding radii from small-molecule crystal structures

Area of Science:

• Crystallography
• Materials Science
• Computational Chemistry

Background:

• X-ray crystallography often does not directly visualize chemical bonds.
• Interatomic distances are commonly used to infer bonding, relying on predefined bonding radii.
• Existing bonding radii sets may not be universally applicable across all crystal structures.

Purpose of the Study:

• To develop an unsupervised computational workflow for deriving accurate bonding radii.
• To create a novel bonding radii set using data from the Crystallography Open Database.
• To improve the reliability of bond identification in crystallographic analyses.

Main Methods:

• An unsupervised machine learning approach was employed.
• Crystal structure data from the Crystallography Open Database was utilized.
• Algorithms were developed to analyze interatomic distances and infer bonding radii.

Main Results:

• A new set of bonding radii was successfully derived.
• The workflow demonstrated effectiveness in identifying bonded atom pairs.
• The derived radii show potential for broader application in crystallographic studies.

Conclusions:

• The unsupervised workflow provides a data-driven method for determining bonding radii.
• This approach can enhance the accuracy of chemical bond identification in crystal structures.
• The derived bonding radii set offers a valuable resource for materials characterization.