Can Deep Learning Search for Exceptional Chiroptical Properties? The Halogenated [6]Helicene Case.

Area of Science:

• Computational Chemistry
• Molecular Modeling
• Spectroscopy

Background:

• Understanding the link between chemical structure and chiroptical properties is crucial for designing new materials.
• Current methods for developing molecules with specific optical properties often rely on inefficient trial-and-error approaches.

Purpose of the Study:

• To predict the chiroptical properties, specifically rotatory strength (R), of a large dataset of halogenated [6]helicenes.
• To develop a computational model that can rapidly predict these properties with high accuracy.
• To gain insights into the structural factors influencing chiroptical behavior.

Main Methods:

• Density Functional Theory (DFT) calculations were employed to generate a diverse dataset of halogenated [6]helicene derivatives.
• Machine learning, specifically deep neural networks, was utilized to train predictive models based on the DFT data.
• Experimental synthesis and characterization were performed for selected high-performance derivatives.

Main Results:

• Developed deep neural network models capable of predicting the maximum rotatory strength (Rmax) for halogenated [6]helicenes with high accuracy and low computational cost.
• Identified key structural features and substitution patterns that significantly influence chiroptical properties.
• Achieved excellent correlation between computationally predicted and experimentally measured rotatory strengths for synthesized compounds.

Conclusions:

• Deep learning models provide a powerful and efficient tool for predicting chiroptical properties of complex organic molecules.
• This data-driven approach can significantly accelerate the discovery and design of novel molecules with tailored optical characteristics.
• The findings pave the way for more rational design strategies in materials science and organic electronics.