Mapping the frontier orbital energies of imidazolium-based cations using machine learning

Area of Science:

• Computational chemistry
• Materials science
• Machine learning

Background:

• Frontier orbital energies (HOMO/LUMO) are crucial for predicting chemical properties like stability and reactivity.
• Density functional theory (DFT) is a common method for calculating these energies, but it is computationally expensive for large datasets.
• Screening numerous isomers of 1-alkyl-3-alkylimidazolium cations ([CnCmim]+) presents a significant computational challenge.

Purpose of the Study:

• To develop a machine learning model for predicting HOMO and LUMO energies of [CnCmim]+ isomers.
• To overcome the computational limitations of DFT for large-scale chemical screening.
• To identify key molecular descriptors influencing frontier orbital energies.

Main Methods:

• Utilized the extreme gradient boosting machine learning method.
• Trained the model on a subset of the [CnCmim]+ isomer space.
• Applied the trained model to predict HOMO and LUMO energies for a vast number of isomers.

Main Results:

• Achieved a mean absolute error (MAE) of 0.4 eV for predicted HOMO energies.
• Achieved a MAE of 0.2 eV for predicted LUMO energies.
• Identified important descriptors for accurate energy predictions.

Conclusions:

• The developed machine learning model offers a computationally efficient alternative to DFT for predicting frontier orbital energies.
• This approach drastically reduces the time and resources required for screening large chemical compound libraries.
• Enables faster exploration of chemical space for applications in materials science and drug discovery.