Highly Efficient Prediction of Dielectric and Electrochemical Properties of Organic Compounds Assisted by Artificial Intelligence

Understanding the dielectric and electrochemical properties
of
organic compounds, pivotal across applications from the development
of OLED screens to energy storage devices, is essential for strategic
material design. This study integrates AI models with DFT modeling
to reliably predict the dielectric strength, dielectric constant,
and electron affinity across a diverse range of organic compounds.
Robust protocols for accurate predictions depend on the selection
of highly correlated features with dielectric properties notably influenced
by the number of electrons and electron affinity. The molecular weight,
number of non-hydrogen atoms, and number of heteroatoms emerge as
crucial features for electron affinity prediction. Ensemble techniques
highlight the effectiveness of a strategic combination of multiple
AI models. These findings illuminate efforts to predict the key properties
of organic compounds comprehensively.