Machine Learning-Based Quantitative Structure-Property Relationships for the Electronic Properties of Cyano...

Area of Science:

• Computational Chemistry
• Materials Science
• Quantum Chemistry

Background:

• Polycyclic aromatic hydrocarbons (PAHs) are crucial in various fields.
• Cyano-substituted PAHs (CN-PAHs) exhibit unique electronic properties.
• Accurate prediction of electronic properties is vital for designing new materials.

Purpose of the Study:

• To develop a quantitative structure-property relationship (QSPR) model for predicting electronic properties of CN-PAHs.
• To utilize a machine learning (ML) approach combined with the degree of π-orbital overlap (DPO) descriptor.
• To enhance prediction accuracy by incorporating substituent effects.

Main Methods:

• Machine learning methodology applied to QSPR.
• Calculation of 926 data points using Density Functional Theory (DFT) at the B3LYP/6-31+G(d) level.
• Development of a modified DPO descriptor to account for substituent effects.

Main Results:

• The developed ML-DPO model achieved excellent linear correlations for predicting electronic properties.
• High prediction accuracy was obtained: within 0.2 eV for multi-CN-substituted PAHs and 0.1 eV for mono-CN-substituted PAHs.
• The model effectively predicts bandgaps, electron affinities, and ionization potentials.

Conclusions:

• The ML-DPO model provides a highly accurate and efficient method for predicting electronic properties of CN-PAHs.
• This approach facilitates the design and discovery of novel CN-PAH materials with desired electronic characteristics.
• The study demonstrates the power of ML-based QSPR in computational materials science.