Correlation between the Stability of Substituted Cobaltocenium and Molecular Descriptors

Area of Science:

• Materials Science
• Computational Chemistry
• Electrochemistry

Background:

• Metallocenium cations are crucial for anion exchange membranes in fuel cells, offering thermal and alkaline stability.
• Chemical modification of cyclopentadienyl rings enhances cation stability, but predicting these effects is complex.

Purpose of the Study:

• To establish the relationship between bond dissociation energy (BDE) and chemistry-informed descriptors for cobaltocenium derivatives.
• To develop a predictive model for cation stability using electronic structural calculations and machine learning.

Main Methods:

• Electronic structural calculations were performed for 118 cobaltocenium derivatives.
• 12 molecular descriptors were analyzed to understand their correlation with BDE.
• A chemistry-informed feed-forward neural network was trained using k-fold cross-validation.

Main Results:

• A nonlinear relationship was found between BDE and the electronic character/frontier orbital energies of substituent groups.
• The neural network accurately predicted BDE with a mean absolute error of ~1 kcal/mol.
• Promising cobaltocenium modifications for experimental validation were identified.

Conclusions:

• Incorporating chemical knowledge into neural networks enhances predictive power for molecular properties, even with limited data.
• This approach provides valuable insights for designing stable metallocenium cations for fuel cell applications.