Friday 28 February 2025
A team of researchers has developed a machine learning-based approach that can quickly and accurately predict excitonic effects in two-dimensional (2D) materials. Excitons are essentially bound pairs of electrons and holes, which play a crucial role in determining the optical properties of these materials.
Traditionally, predicting excitonic effects requires complex calculations using methods such as many-body perturbation theory or the Bethe-Salpeter equation. However, these approaches can be computationally expensive and time-consuming, limiting their applicability for large-scale material screening.
The new approach, developed by a team of scientists from Monash University in Australia, uses machine learning to predict exciton binding energies (EBEs) in 2D materials. The researchers trained their model using data from the Computational 2D Materials Database (C2DB), which contains information on over 4,000 monolayers.
The machine learning algorithm used by the team is a random forest regressor, which combines multiple decision trees to improve predictive accuracy. By training this model on C2DB data, the researchers were able to develop a robust predictor of EBEs in 2D materials.
One of the key advantages of this approach is its ability to quickly and accurately predict EBEs for large numbers of materials. This could be particularly useful for identifying 2D materials with desirable excitonic properties, such as high binding energies or tunable optical absorption.
The researchers also used a Bayesian optimization framework to identify the top-performing materials in their dataset. This approach allowed them to efficiently search for materials with high EBEs and narrow down the list of potential candidates.
The team’s results show that their machine learning-based approach can accurately predict EBEs in 2D materials, often outperforming traditional methods. They also identified several new candidate materials with promising excitonic properties, including some that have not been extensively studied before.
Overall, this research demonstrates the power of machine learning for predicting complex material properties and could have significant implications for the development of new 2D materials with exciting optical and electronic properties.
Cite this article: “Machine Learning Accelerates Excitonic Effects Prediction in 2D Materials”, The Science Archive, 2025.
Machine Learning, Excitons, Two-Dimensional Materials, Computational Database, Random Forest Regressor, Binding Energies, Optical Properties, Material Screening, Bayesian Optimization, Nanomaterials
