Materials Informatics Framework for Accelerated Discovery of High-Refractive-Index 2D Materials

Area of Science:

• Materials Science
• Condensed Matter Physics
• Computational Materials Science

Background:

• Predicting optical properties of 2D materials is vital for photonics but challenging due to theory-experiment discrepancies.
• Existing methods struggle with accurate and efficient screening of novel 2D materials for specific applications.

Purpose of the Study:

• To develop a physics-guided machine learning (ML) framework for accelerated screening of 2D materials.
• To enable accurate prediction of optical properties for designing next-generation photonic materials.

Main Methods:

• Combined first-principles density functional theory (DFT) calculations with graph neural network (GNN) models.
• Integrated experimental spectroscopic validation and Cauchy-model for optical property correction.
• Developed a universal method for defining 2D material thickness and corrected DFT-derived optical properties.

Main Results:

• Created a database of over 1000 transition metal dichalcogenides (TMDs) monolayers and their optical properties.
• Developed a ML model to accurately calculate refractive indices in the near-infrared (near-IR) region (755-1064 nm).
• Validated the ML model against independent 2D materials databases, confirming structure-property correlations.

Conclusions:

• The ML framework significantly accelerates the screening of 2D materials with tailored optical functionalities.
• High-refractive-index materials like Bi2Te2Se show promise for enhanced field confinement in monolayer waveguides.
• This approach facilitates the discovery and design of novel 2D materials for integrated photonics.